搜档网
当前位置:搜档网 › Aggregated multicast–a comparative study

Aggregated multicast–a comparative study

Aggregated multicast–a comparative study
Aggregated multicast–a comparative study

Aggregated Multicast—A Comparative Study

1

UCLA CSD TR#020011

Jun-Hong Cui,Jinkyu Kim

Dario Maggiorini,Khaled Boussetta and Mario Gerla

Computer Science Department,University of California,Los Angeles,CA90095

Abstract

Multicast state scalability is among the critical issues which delay the deployment of IP multicast.In our previous work,we proposed a scheme,called aggregated multicast to reduce multicast state.The key idea is that multiple groups are forced to share a single delivery tree.We presented some initial results to show that multicast state can be reduced, sometimes at the expense of bandwidth overhead.In this report,we develop a more quantitative assessment of the cost/bene?t trade-offs.We introduce metrics to measure multicast state and tree management overhead for multicast schemes.We then compare aggregated multicast with conventional multicast schemes,such as source speci?c tree scheme and shared tree scheme.Our extensive simulations show that aggregated multicast can achieve signi?cant routing state and tree management overhead reduction while containing the expense of extra resources(bandwidth waste and tunnelling overhead,etc.).We conclude that aggregated multicast is a very cost-effective and promising direction for scalable transit domain multicast provisioning.

I.I NTRODUCTION

IP Multicast has been a very hot area of research,development and testing for more than one decade since Stephen Deering established the IP multicast model in1988[5].However,IP mul-ticast is still far from being widely deployed in the Internet.Among the issues which delay the deployment,state scalability is one of the most critical ones.

IP multicast utilizes a tree delivery structure on which data packets are duplicated only at fork nodes and are forwarded only once over each link.By doing so IP multicast can scale well to support very large multicast groups.However,a tree delivery structure requires all tree nodes to maintain per-group(or even per-group/source)forwarding information,which increases linearly with the number of groups.Growing number of forwarding state entries means more memory re-quirement and slower forwarding process since every packet forwarding action involves an address look-up.Thus,multicast scales well to the number of members within a single multicast group. But,it suffers from scalability problems when the number of simultaneous active multicast groups is very large.

To improve multicast state scalability,we proposed a novel scheme to reduce multicast state, which we call aggregated multicast.In this scheme,multiple multicast groups are forced to share one distribution tree,which we call an aggregated tree.This way,the number of trees in the net-work may be signi?cantly reduced.Consequently,forwarding state is also reduced:core routers only need to keep state per aggregated tree instead of per group.The trade-off is that this ap-proach may waste extra bandwidth to deliver multicast data to non-group-member nodes.In our earlier work[7,8],we introduced the basic concept of aggregated multicast,proposed an algorithm to assign multicast groups to delivery trees with controllable bandwidth overhead and presented some initial results to show that multicast state can be reduced through inter-group tree sharing. However,a thorough performance evaluation of aggregated multicast is needed:what level of the gain does aggregated multicast offer over conventional multicast schemes?In this report,we pro-pose metrics to measure multicast state and tree management overhead for multicast schemes.We then compare aggregated multicast with conventional multicast schemes,such as source speci?c tree scheme and shared tree scheme.Our extensive simulations show that aggregated multicast can achieve signi?cant state and tree management overhead reduction while at reasonable expense (bandwidth waste and tunnelling overhead,etc.).

The rest of this report is organized as follows.Section II gives a classi?cation of multicast schemes.Section III reviews the concept of aggregated multicast and presents a new algorithm for group-tree matching.Section IV then discusses the implementation issues for different multicast schemes and de?nes metrics to measure multicast state and tree management overhead,and Sec-tion V provides an extensive simulation study of different multicast schemes.Finally Section VI summarizes the contributions of our work.

II.A C LASSIFICATION OF M ULTICAST S CHEMES

According to the type of delivery tree,we classify the existing intra-domain multicast routing protocols into two categories(It should be noted that,in this report,we only consider intra-domain multicasting):in the?rst category,protocols construct source speci?c tree,and in the second category,protocols utilize shared tree.For the convenience of discussion,we call the former category as source speci?c tree scheme,and the latter one as shared tree scheme.According to this classi?cation,we can say,DVMRP[11],PIM-DM[4],and MOSPF[10]belong to source speci?c tree scheme category,while CBT[3],PIM-SM[6],and BIDIR-PIM[9]are basically shared tree schemes(of course,PIM-SM can also activate source speci?c tree when needed). Source speci?c tree scheme constructs a separate delivery tree for each https://www.sodocs.net/doc/a14546451.html,ly,each

R2

S2

2

R

1

(a) Source Specific Tree

S

R2

S2

S

2

R

1

(c) Bi-directional Shared Tree

R2

S2

S

2

R

1

(b) Unidirectional Shared Tree

R2

S2

S

2

R

1

(d) Packet of S1 delivering on Bi-directional Shared Tree

Fig.1.Different types of trees for the same group G with sources(S1,S2)and receivers(R1,R2). source of a group utilizes its own tree to deliver data to the receivers in the group.The shared tree scheme instead constructs trees based on per-group and all the sources of a group use the same tree to deliver data to the receivers.In other words,multiple sources of the same group share a single delivery tree.Shared tree can be unidirectional or bi-directional.PIM-SM is a unidirectional shared tree scheme.CBT and BIDIR-PIM are bi-directional shared tree schemes.Fig.1shows the different types of trees for the same group G with sources(S1,S2)and receivers(R1,R2).For source speci?c tree schemes,two trees are set up for group G.For the unidirectional shared tree scheme,one tree is set up.Each source needs to unicast packets to the rendezvous point(RP)or build source speci?c state on all nodes along the path between the source and the RP.For the last scheme,only one bi-directional tree will work.A source can unicast packet to the nearest on-tree node instead of RP.And each on-tree node can deliver packets along the bi-directional tree. Compared with conventional multicast schemes,aggregated multicast raises tree-sharing to an even higher level—inter-group tree sharing,where multiple multicast groups are forced to share one aggregated tree.An aggregated tree can be either a source speci?c tree or a shared tree,while a shared tree can be either unidirectional or bi-directional.We are going to review the basic concept of aggregated multicast and discuss some related issues in the following section.

Customer networks, domain D

Fig.2.Domain peering and a cross-domain multicast tree,tree nodes:D1,A1,Aa,Ab,A2,B1,A3,C1,covering group G0(D1,B1,C1).

III.A GGREGATED M ULTICAST

A.Concept of Aggregated Multicast

Aggregated multicast[7,8]is proposed to reduce multicast state,and it is targeted to intra-domain multicast provisioning.The key idea is that,instead of constructing a tree for each individ-ual multicast group in the core network(backbone),multiple multicast groups are forced to share a single aggregated tree.

Fig.2illustrates a hierarchical inter-domain network peering.Domain A is a regional or national ISP’s backbone network,and domain D,X,and Y are customer networks of domain A at a certain location(say,Los Angeles),and domain E is a customer network of domain A in another location (say,Seattle).Domain B and C can be other customer networks(say,in Boston)or some other ISP’s networks that peer with A.A multicast session originates at domain D and has members in domain B and C.Routers D1,A1,A2,A3,B1and C1form the multicast tree at the inter-domain level while A1,A2,A3,Aa and Ab form an intra-domain sub-tree within domain A(there may be other routers involved in domain B and C).Consider a second multicast session that originates at domain D and also has members in domain B and C.For this session,a sub-tree with exactly the same set of nodes will be established to carry its traf?c within domain A.Now if there is a third multicast session that originates at domain X and it also has members in domain B and C,then router X1instead of D1will be involved,but the sub-tree within domain A still involves the same set of nodes:A1,A2,A3,Aa,and Ab.

To facilitate our discussions,we make the following de?nitions.For a group G,we call terminal nodes the nodes where traf?c enters or leaves a domain,A1,A2,and A3in our example.We call transit nodes the tree nodes that are internal to the domain,such as Aa and Ab in our example.

In conventional IP multicast,all the nodes in the above example that are involved within domain A must maintain separate state for each of the three groups individually though their multicast trees are actually of the same“shape”.Alternatively,in the aggregated multicast,we can setup a pre-de?ned tree(or establish a tree on demand)that covers nodes A1,A2and A3using a single multicast group address(within domain A).This tree is called an aggregated tree(AT)and it is shared by more than one multicast groups(three groups in the above example).We say an aggregated tree T covers a group G if all terminal nodes for G are member nodes of T.Data from a speci?c group is encapsulated at the incoming terminal node using the address of the aggregated tree.It is then distributed over the aggregated tree and decapsulated at exiting terminal nodes to be further distributed to neighboring networks.This way,transit router Aa and Ab only need to maintain a single forwarding entry for the aggregated tree regardless how many groups are sharing it.

Thus,aggregated multicast can reduce the required multicast state.Transit nodes don’t need to maintain state for individual groups;instead,they only maintain forwarding state for a smaller number of aggregated trees.The management overhead for the distribution trees is also reduced. First,there are fewer trees that exchange refresh messages.Second,tree maintenance can be a much less frequent process than in conventional multicast,since an aggregated tree has a longer life span.

B.Group-Tree Matching in Aggregated Multicast

Aggregated multicast achieves state reduction through inter-group tree sharing—multiple groups share a single aggregated tree.When a group is started,an aggregated tree should be assigned to the group following some rules.If a dense set of aggregated trees is pre-de?ned,things will be easy:just choose the tree with minimum cost which can cover the group.While in the dynamic case(aggregated tree are established on demand),a more elaborate group-tree matching algorithm is needed.

When we try to match a group G to an aggregated tree T,we have four cases:

1.T can cover G and all the tree leaves are terminal nodes for G,then this match is called perfect match for G;

2.T can cover G but some of the tree leaves are not terminal nodes for G,then this match is a pure-leaky match(for G);

3.T can not cover G and all the tree leaves are terminal nodes for G,then this match is called a pure-incomplete match;

4.T can not cover G and some of the tree leaves are not terminal nodes for G,we name this match as incomplete leaky match.

Namely,we denote the case when some of the tree leaves are not terminal nodes for the group G as leaky match and the case when the tree can not cover the group G as incomplete match.Clearly, leaky match includes case2and4,and incomplete match includes case3and4.

To give examples,the aggregated tree T0with nodes(A1,A2,A3,Aa,Ab)in Fig.2is a perfect match for our early multicast group G0which has members(D1,B1,C1).However,if the above aggregated tree T0is also used for group G1which only involves member nodes(D1,B1),then it is a pure-leaky match since traf?c for G1will be delivered to node A3(and will be discarded there since A3does not have state for that group).Obviously,the aggregated tree T0is an pure-incomplete match for multicast group G2which has members(D1,B1,C1,E1)and an incomplete leaky match for multicast group G3with members(D1,B1,E1).

We can see that leaky match helps to improve inter-group tree sharing.A disadvantage of leaky match is that some bandwidth is wasted to deliver data to nodes that are not members for the group. Leaky match may be unavoidable since usually it is not possible to establish aggregated trees for all possible group combinations.In the incomplete match case,we have two ways to get a tree for the group.One way is to construct a bigger tree by moving the entire group to a new larger aggregated tree,or,to extend the current aggregated tree to a bigger tree.Extending a tree might involve a lot of overhead,because all the groups which use the extended aggregated tree need to make the corresponding adjustment.An alternative way is to use“tunnelling”.Here we give an example. Suppose member E1in domain E decides to join group G0in Fig.2.Instead of constructing a bigger tree,an extension“tunnel”can be established between edge router A4(connecting domains A and E)and edge router A1.This solution combines features of multicast inter-group tree sharing and tunnelling;it still preserves core router scalability properties by pushing complexity to edge routers.We can see that,if we employ tunnelling instead of tree extension,then an incomplete match only involves tunnelling.An incomplete leaky match will activate tunnelling and will also waste resources because of leaky matching.

C.A New Group-Tree Matching Algorithm

Here we present a new group-tree matching algorithm which is used in our simulation.To avoid the overhead of tree extension,this algorithm uses tunnelling for incomplete match.First,we introduce some notations and de?nitions.

C.1Overhead De?nition

A network is modelled as an undirected graph G(V,E).Each edge(i,j)is assigned a positive cost c ij=c ji,which represents the cost to transport a unit of data from node i to node j(or from j to i).Given a multicast tree T,total cost to distribute a unit of data over that tree is

C(T)=

(i,j)∈T

c ij.(1) If every link is assume

d to hav

e equal cost1,tree cost is simply C(T)=|T|?1,where|T| denotes the number o

f nodes in T.This assumption holds in this paper.Let MT S(Multicast Tree Set)denote the current set of multicast trees established in the network.A“native”multicast tree(constructed by some conventional multicast routin

g algorithm,denoted by A)for a multicast

group G is denoted by T A

G

.

For any aggregated tree T,as mentioned in Section III-B,it is possible that T does not have a perfect match with group G,which means that the match is leaky match or incomplete match.In leaky match case,some of the leaf nodes of T are not the terminal nodes for G,and then packets reach some destinations that are not interested in receiving them.Thus,there is bandwidth over-head in aggregated multicast.We assume each link has the same bandwidth,and each multicast group has the same bandwidth requirement,then it is easy to get that the percentage bandwidth overhead(denoted byδL(G,T))is actually equal to the percentage link cost overhead:

δL(G,T)=C(T)?C(T A

G

))

C(T A

G

)

,(2)

Apparently,δL(G,T)is0for perfect match and pure-incomplete match.

In incomplete match case,T can not cover all the members of group G,and some tunnels need to be set up.Data packets of G exits from the leaf nodes of T,and tunnels to the corresponding terminal nodes of G.Clearly,there is tunnelling overhead caused by unicasting data packets to group terminal nodes.Each tunnel’s cost can be measured by the link cost along the tunnel. Assume there are k G tunnels for group G,and each tunnel is denoted by T t

G,i

,where1≤i≤k G, then we de?ne the percentage tunnelling overhead for this incomplete match as

δI(G,T)= k G

i=1

C(T t

G,i

)

C(T A

G

)

.(3)

It is easy to tell thatδI(G,T)is0for perfect match and pure-leaky match.

C.2Algorithm Description

Our new group-tree matching algorithm is based on bandwidth overhead and tunnelling over-head.Let l t be the given bandwidth overhead threshold for leaky match,and t t be the given tunnelling overhead threshold for incomplete match.When a new group is started,

https://www.sodocs.net/doc/a14546451.html,pute a“native”multicast tree T A

for G based on the multicast group membership;

G

2.for each tree T in MT S,computeδL(G,T)andδI(G,T);ifδL(G,T)

3.among all candidates,choose the one such that f(δL(G,T),δI(G,T))is minimum and denote it as T m,then T m is used to deliver data for G;if T m can not cover G,the corresponding tunnels will be set up;

4.if no candidate found in step2,T A

is used for G and is added to MT S.

G

In step3,f(δL(G,T),δI(G,T))is a function to decide how to choose the?nal tree from a set of candidates.In our simulations,

f(δL(G,T),δI(G,T))=δL(G,T)+δI(G,T).(4) Actually,this function can be chosen according to the need in the real scenarios.For example,we can give more weight to bandwidth overhead if bandwidth is our main concern.

IV.E XPERIMENT M ETHODOLOGY

In an aggregated multicast scheme,sharing a multicast tree among multiple groups may signif-icantly reduce the states at network core routers and correspondingly the tree management over-head.However,what level of gain can aggregated multicast get over other multicast schemes?In this section,we will discuss some implementation issues for different multicast schemes in our simulations,and de?ne metrics to measure multicast state and tree management overhead.Then in Section V,we will compare aggregated multicast with other multicast schemes through simula-tions.

A.Implementation of Multicast Schemes in SENSE

We do our simulations in SENSE(S imulation E nvironment for N etwork S ystem E volution)[2], which is a network simulator developed at the network research laboratory at UCLA to perform wired network simulation experiments.

In SENSE,we can support the source speci?c tree scheme,the shared tree scheme(with unidi-rectional tree and bi-directional tree),and the aggregated multicast scheme(with source speci?c

tree,unidirectional shared tree and bi-directional shared tree).It should be noted that,the multicast schemes we discuss here are not speci?c multicast routing protocols,since the goal of this work is to study the gain of aggregated multicast over conventional multicast schemes.The comparison is between schemes,not protocols.

We implement each multicast scheme with a centralized method.For each scheme,there is a centralized processing entity(called multicast controller),which has the knowledge of network topology and multicast group membership.The multicast controller is responsible for constructing the multicast tree according to different multicast schemes and then distributing the routing tables to the corresponding nodes.In the implementation,we did not model the membership acquisi-tion and management procedures which depend on the speci?c multicast routing protocol.This omission reduces the bias and improves the fairness in comparing different multicast schemes. The multicast controller will read group and member dynamics from a pre-generated(or generated on-the-?y)trace?le.

For shared tree scheme(either unidirectional or bi-directional)and aggregated multicast scheme with shared tree(unidirectional or bi-directional),a core node or a rendezvous point(RP)is needed when a tree is constructed.To achieve better load balancing,the core node should be chosen carefully.In our implementation,for all multicast schemes using shared trees,a set of possible core routers are pre-con?gured.Then,when a group is initialized,the core is chosen so as to minimize the cost of the tree.

In an aggregated multicast scheme,the multicast controller also needs to manage aggregated trees and multicast groups and manipulate group-tree matching algorithm.The multicast controller has the same responsibility as the tree manager(mentioned in[7,8])in aggregated multicast.It collects group join messages and assigns aggregated trees to groups.Once it determines which aggregated tree to use for a group,the tree manager can install corresponding state at the terminal nodes involved.

B.Performance Metrics

The main purpose of tree sharing is to reduce multicast state and tree maintenance overhead. So,multicast state and tree management overhead measures are of most concern here.In our experiments,we introduce the following metrics.

Number of multicast trees(or number of trees for shorthand)is de?ned as|MT S|,where MTS denotes the current set of multicast trees established in the networks.This metric is a direct measurement for the multicast tree maintenance overhead.The more multicast trees,the more

memory required and the more processing overhead involved(though the tree maintenance over-head depends on the speci?c multicast routing protocols).

Forwarding state in transit nodes(or transit state for shorthand).Without losing generality, we assume a router needs one state entry per multicast address in its forwarding table.As we de?ned in Section III,in a multicast tree,there are transit nodes and terminal nodes.We note that forwarding state in terminal nodes can not be reduced in any multicast scheme.Even in aggregated multicast,the terminal nodes need to maintain the state information for individual groups.So,to assess the state reduction,we measure the forwarding state in transit nodes only.

V.S IMULATIONS

In this section,we compare aggregated multicast with conventional multicast schemes through extensive simulation,and quantitatively evaluate the gain of aggregated multicast.

A.Multicast Trace Generation

A.1Multicast Group Models

Given the lack of experimental large scale multicast traces,we have chosen to develop member-ship models that exhibit locality and group correlation preferences.In our simulation,we use the group model previously developed in[8]:the random node-weighted model.For completeness, we provide here a summary description of this model.

The random node-weighted model.This model statistically controls the number of groups a node will participate in based on its weight:for two nodes i and j with weight w(i)and w(j) (0

number of groups that have j as a member,then it is easy to prove that,in average,N(i)

N(j)=w(i)

w(j)

.

Assuming the number of nodes in the network is N and nodes are numbered from1to N.To each node i,1≤i≤N,is assigned a weight w(i),0≤w(i)≤1.Then a group can be generated as the following procedure:

for i=1to N do

generate p,a random number uniformly between0and1,let it be p

if p

add i as a group member

end if

end for

Following this model,the average size of multicast groups is N n

i=1

w(i).

A.2Multicast Membership Dynamics

Generally,there are two methods to control multicast group member dynamics.The?rst one is to create new members(sources and receivers)for a group according to some pre-de?ned statistics (arrive rate and member life time etc.),then decide the termination of a group based on the distri-bution of the group size.This is actually a member-driven dynamic.As to the other method,we call it group-driven dynamics,which means that,group characteristics(group size,group arrival rate,and group life time)are de?ned?rst and then group members are generated according to groups.In our experiment,we use the second method,in which the group statistics are controlled ?rst(using the random node weighted model).Actually,the second method looks more reasonable for many real life multicast applications(such as video conference,tele-education,etc.).In any event,the speci?c method used to control group member dynamics is not expected to affect our simulation results.

In our experiment,given a group life period(t1,t2),and the group member set g,where|g|=n, for any node m i∈g,1≤i≤n,its join time and leave time are denoted by t join(m i)and t leave(m i)separately.Then the member dynamics is controlled as follows:

for i=1to n do

m i∈g

t join(m i)=get rand(t1,t2);(get a random time in(t1,t2))

t leave(m i)=get rand(t join(m i),t2);(get a random time in(t join(m i),t2))

end for

It is not dif?cult to know that the average life time of each member is|t2?t1|/4.

B.Results and Analysis

We now present results from simulation experiments using a real network topology,vBNS back-bone[1].

In vBNS backbone,there are43nodes,among which FORE ASX-1000nodes(16of them)are assumed to be core routers only(i.e.will not be terminal nodes for any multicast group)and are assigned weight0.Any other node is assigned a weight0.05to0.8according to link bandwidth of the original backbone router–the rationale is that,the more the bandwidth on the outgoing(and incoming)links of a node,the more the number of multicast groups it may participate in.So,we assign weight0.8to nodes with OC-12C links(OC-12C-linked nodes for shorthand),0.2to nodes

with OC-3C links (OC-3C-linked nodes),and 0.05to nodes with DS-3links (DS-3-linked nodes).In simulation experiments,multicast session requests arrive as a Poisson process with arrival rate λ.Sessions’life time has an exponential distribution with average μ.At steady state,the

average number of sessions is ˉN

=λ×μ.During the life time of each multicast session,group members are generated dynamically according to group-driven method introduced earlier.Group membership is controlled using the random node-weighted model.Performance data is collected at certain time points (e.g.at T =10μ),when steady state is reached,as “snapshot”.

First,we design experiments to compare unidirectional shared tree scheme (UST scheme for shorthand)vs aggregated multicast scheme with unidirectional shared tree (AM w/UST scheme for short hand).In this set of experiments,each member of a group can be a source and a receiver.Once a multicast session starts up,its core node (or RP)is randomly chosen from the 16core routers in the network.For aggregated multicast scheme with unidirectional shared tree,the algo-rithm speci?ed in Section III-C is used to match a group to a tree.When members join or leave a group,its aggregated tree will be adjusted according to the matching algorithm.Correspondingly,the routing algorithm A is PIM-SM like routing algorithm which uses unidirectional shared tree.

0500

1000

15002000

2500

N u m b e r o f t r e e s Number of groups

02000

4000

6000

800010000

12000

T r a n s i t s t a t e Number of groups (b)Fig.3.Simulation results for UST and AM w/UST when only pure-leaky match (tth=0)is allowed

In our ?rst experiment,for aggregated multicast,we only allow pure-leaky match,which means

that the tunnelling overhead threshold(represented as tth)is0.We vary the bandwidth overhead threshold(represented as lth)from0to0.3.For UST scheme and AM w/UST scheme with differ-ent bandwidth threshold,we run simulations to show how the aggregation of aggregated multicast “scales”with the average number of concurrent groups.The results are plotted in Fig.3.As to the number of trees(see Fig.3(a)),clearly,for UST scheme,it is almost a linear function of the number of groups.For AM w/UST scheme,as the number of groups becomes bigger,the number of trees also increases,but the increase is much less than UST(even for perfect match(lth=0), the number of trees is only1150instead of2500for UST when there are2500groups).Also this “increase”decreases as there are more groups,which means that as more groups are pumped into the network,more groups can share an aggregated tree.Fig.3(b)shows us the change of transit state with the number of concurrent groups.It has similar trend to metric number of trees.Transit state is reduced from12800to7400(above40%reduction)even for perfect match when2500 groups come.A general observation is that,when bandwidth overhead threshold is increased,that is,more bandwidth is wasted,number of trees decreases and transit state falls,which means more aggregation.Therefore,there is a trade-off between state and tree management overhead reduction and bandwidth waste.

In our second experiment,for aggregated multicast,we only allow pure-incomplete match, which means that the bandwidth overhead threshold(represented as lth)is0.We vary the tun-nelling overhead threshold(represented as tth)from0to0.3and want to look at the effect of tunnelling overhead threshold in the aggregation.Fig.4plots the results,which give us curves similar to Fig.3.However,we can see that tunnelling overhead threshold affects the aggregation signi?cantly:when tth=0.3,and group number is2500,almost5groups share one tree,and transit state is reduced about70percentage.When group number increases,we can expect even much more aggregation.The stronger in?uence of tunnelling overhead threshold on aggregation is not a surprise:the higher the tunnelling overhead threshold is,the more chance for a group to use a small tree for data delivery,the more likely for more groups to share a single aggregated tree. Our third experiment considers both bandwidth overhead and tunnelling overhead.And the simulation results are shown in Fig.5.All the results tell what we expect:more aggregation achieved when we sacri?ce more(bandwidth and tunnelling)overhead.

We have shown the results for comparing unidirectional shared tree scheme(UST)vs aggregated multicast scheme with unidirectional shared tree(AM w/UST).Similar results are obtained for source speci?c tree scheme(SST)vs aggregated multicast scheme with source speci?c tree(AM

0500

1000

15002000

2500

N u m b e r o f t r e e s Number of groups

02000

4000

6000

800010000

12000

T r a n s i t s t a t e Number of groups (b)

Fig.4.Simulation results for UST and AM w/UST when only pure-incomplete match (lth=0)is allowed

0500

1000

15002000

2500

N u m b e r o f t r e e s Number of groups

02000

4000

6000

800010000

12000

T r a n s i t s t a t e Number of groups (b)Fig.5.Simulation results for UST and AM w/UST when both leaky match and incomplete match are allowed

w/SST)and bi-directional shared tree scheme (BST)vs aggregated multicast with bi-directional shared tree (AM w/BST).Some representative results are shown below.

In the set of experiments for BST vs AM w/BST,we also assume all the group members can be sources and receivers.In bi-directional shared tree scheme,the core node (or RP)only helps to construct the delivery tree,no unicasting traf?c from sources to the core node,as is different from unidirectional shared tree scheme.Fig.6plots the results for the same simulation scenarios as Fig.5except that the multicast schemes are different.We can see that,the metric of “number of trees”in Fig.6is the same as that in Fig.5.This is because the group-tree mapping procedures in AM w/UST and AM w/BST are exactly the same.However,the transit state is different in the two ?gures,since,in bi-directional shared tree scheme,the core node is a transit node and does not keep group speci?c state.Thus,more state reduction is achieved (when lth=0.3and tth=0.3,the state reduction is around 75%).

0500

1000

15002000

2500

N u m b e r o f t r e e s

02000

4000

6000

8000

10000

12000

14000

16000

T r a n s i t s t a t e Number of groups (b)Fig.6.Simulation results for BST and AM w/BST when both leaky match and incomplete match are allowed In the set of experiments for SST vs AM w/SST,we assume there is only one source for each group.And the source is randomly chosen from group members.In AM w/SST,only the two groups with the same source can share an aggregated tree.Obviously,this will reduce the ability of aggregation,though many other factors,such as multicast group model,network topology,and

member dynamics,etc.will affect the aggregation.Fig.7shows the results for SST vs AM w/SST.

0500

1000

15002000

2500

N u m b e r o f t r e e s Number of groups

02000

4000

6000

8000

10000

T r a n s i t s t a t e Number of groups (b)Fig.7.Simulation results for SST and AM w/SST when both leaky match and incomplete match are allowed From our simulation result and analysis,the bene?ts of aggregated multicast are mainly in the following two areas:(1)tree management overhead reduction by reducing the number of trees needed to be maintained in the network;(2)state reduction at transit nodes.The price to pay is bandwidth waste and tunnelling cost.The above simulation results con?rm our claim while demonstrate the following trends:(1)if we are willing to sacri?ce more bandwidth or tunnelling cost (by lifting the bandwidth overhead threshold and tunnelling overhead threshold correspond-ingly),more or better aggregation is achieved;by “more aggregation”we mean more groups can share an aggregated tree (in average)and correspondingly more state reduction;(2)better aggre-gation is achievable as the number of concurrent groups increases.The later point is especially important since one basic goal of aggregated multicast is scalability in the number of concurrent groups.Furthermore,aggregated multicast can be applied on top of any conventional multicast scheme.The aggregation ability depends on the underlying multicast scheme.AM w/UST and AM w/BST gives similar aggregation,but AM w/UST involves more unicasting overhead (from source to RP).Compared with AM w/UST and AM w/BST,AM w/SST generally yields less ag-gregation.However,as we know,source speci?c tree scheme (SST)is much more simpler than

shared tree scheme(UST and BST).

VI.C ONCLUSIONS AND F UTURE W ORK

In this report,we?rst gave a classi?cation of multicast schemes,then had a short review of aggregated multicast.For aggregated multicast,we proposed a new group-tree dynamic matching algorithm using tunnelling.We implemented different multicast schemes in SENSE.Through extensive simulations,we compared aggregated multicast with conventional multicast schemes and evaluated its gain over other schemes.Our simulations have shown that signi?cant state and tree management overhead reduction(up to75%state reduction in our experiments)can be achieved with reasonable bandwidth and tunnelling overhead(0.1to0.3),etc..Thus aggregated multicast is a very promising scheme for transit domain multicast provisioning.

We are now in the process of developing an actual aggregated multicast routing protocol testbed for real application scenarios.The testbed will allow us to better evaluate the state reduction and control overhead.

R EFERENCES

[1]vBNS backbone network.https://www.sodocs.net/doc/a14546451.html,/.

[2]SENSE:Simulation Environment for Network System Evolution.https://www.sodocs.net/doc/a14546451.html,/NRL/hpi/resources.html,2001.

[3] A.Ballardie.Core Based Trees(CBT version2)multicast routing:protocol speci?cation.IETF RFC2189,September1997.

[4]S.Deering,D.Estrin,D.Farinacci,and V.Jacobson.Protocol Independent Multicast(PIM),Dense Mode Protocol:Speci?-

cation.Internet draft,March1994.

[5]Stephen Deering.Multicast routing in a datagram internetwork.Ph.D thesis,December1991.

[6] D.Estrin,D.Farinacci,A.Helmy,D.Thaler,S.Deering,M.Handley,V.Jacobson,C.Liu,P.Sharma,and L.Wei.Protocol

Independent Multicast-Sparse Mode(PIM-SM):Protocol Speci?cation.IETF RFC2362,June1998.

[7]Aiguo Fei,Jun-Hong Cui,Mario Gerla,and Michalis Faloutsos.Aggregated multicast:an approach to reduce multicast state.

In the proceedings of Sixth Global Internet Symposium(GI2001),November2001.

[8]Aiguo Fei,Jun-Hong Cui,Mario Gerla,and Michalis Faloutsos.Aggregated Multicast with Inter-Group Tree Sharing.In the

proceedings of NGC2001,November2001.

[9]Mark Handley and et al.Bi-directional Protocol Independent Multicast(BIDIR-PIM).Internet draft:draft-ietf-pim-bidir-

03.txt,June2001.

[10]J.Moy.Multicast routing extensions to OSPF.RFC1584,March1994.

[11] C.Partridge,D.Waitzman,and S.Deering.Distance vector multicast routing protocol.RFC1075,1988.

电力电子技术答案第五版(全)

电子电力课后习题答案 第一章电力电子器件 1.1 使晶闸管导通的条件是什么? 答:使晶闸管导通的条件是:晶闸管承受正相阳极电压,并在门极施加触发电流(脉冲)。 或者U AK >0且U GK >0 1.2 维持晶闸管导通的条件是什么?怎样才能使晶闸管由导通变为关断? 答:维持晶闸管导通的条件是使晶闸管的电流大于能保持晶闸管导通的最小电流,即维持电流。 1.3 图1-43中阴影部分为晶闸管处于通态区间的电流波形,各波形的电流最大值均为 I m ,试计算各波形的电流平均值I d1 、I d2 、I d3 与电流有效值I 1 、I 2 、I 3 。 解:a) I d1= Im 2717 .0 )1 2 2 ( 2 Im ) ( sin Im 2 1 4 ≈ + = ?π ω π π π t I 1= Im 4767 .0 2 1 4 3 2 Im ) ( ) sin (Im 2 1 4 2≈ + = ?π ? π π π wt d t b) I d2= Im 5434 .0 )1 2 2 ( 2 Im ) ( sin Im 1 4 = + = ?wt d t π π ? π I 2= Im 6741 .0 2 1 4 3 2 Im 2 ) ( ) sin (Im 1 4 2≈ + = ?π ? π π π wt d t c) I d3= ?= 2 Im 4 1 ) ( Im 2 1π ω π t d I 3= Im 2 1 ) ( Im 2 1 2 2= ?t dω π π 1.4.上题中如果不考虑安全裕量,问100A的晶阐管能送出的平均电流I d1、I d2 、I d3 各为多 少?这时,相应的电流最大值I m1、I m2 、I m3 各为多少? 解:额定电流I T(AV) =100A的晶闸管,允许的电流有效值I=157A,由上题计算结果知 a) I m1 35 . 329 4767 .0 ≈ ≈ I A, I d1 ≈0.2717I m1 ≈89.48A

电力电子技术期末考试试题及答案修订稿

电力电子技术期末考试 试题及答案 Coca-cola standardization office【ZZ5AB-ZZSYT-ZZ2C-ZZ682T-ZZT18】

电力电子技术试题 第1章电力电子器件 1.电力电子器件一般工作在__开关__状态。 2.在通常情况下,电力电子器件功率损耗主要为__通态损耗__,而当器件开关频率较高时,功率损耗主要为__开关损耗__。 3.电力电子器件组成的系统,一般由__控制电路__、_驱动电路_、_主电路_三部分组成,由于电路中存在电压和电流的过冲,往往需添加_保护电路__。 4.按内部电子和空穴两种载流子参与导电的情况,电力电子器件可分为_单极型器件_、_双极型器件_、_复合型器件_三类。 5.电力二极管的工作特性可概括为_承受正向电压导通,承受反相电压截止_。 6.电力二极管的主要类型有_普通二极管_、_快恢复二极管_、_肖特基二极管_。 7.肖特基二极管的开关损耗_小于_快恢复二极管的开关损耗。 8.晶闸管的基本工作特性可概括为__正向电压门极有触发则导通、反向电压则截止__。 9.对同一晶闸管,维持电流IH与擎住电流IL在数值大小上有IL__大于__IH 。 10.晶闸管断态不重复电压UDSM与转折电压Ubo数值大小上应为,UDSM_大于__Ubo。 11.逆导晶闸管是将_二极管_与晶闸管_反并联_(如何连接)在同一管芯上的功率集成器件。 的__多元集成__结构是为了便于实现门极控制关断而设计的。 的漏极伏安特性中的三个区域与GTR共发射极接法时的输出特性中的三个区域有对应关系,其中前者的截止区对应后者的_截止区_、前者的饱和区对应后者的__放大区__、前者的非饱和区对应后者的_饱和区__。 14.电力MOSFET的通态电阻具有__正__温度系数。 的开启电压UGE(th)随温度升高而_略有下降__,开关速度__小于__电力MOSFET 。 16.按照驱动电路加在电力电子器件控制端和公共端之间的性质,可将电力电子器件分为_电压驱动型_和_电流驱动型_两类。 的通态压降在1/2或1/3额定电流以下区段具有__负___温度系数,在1/2或1/3额定电流以上区段具有__正___温度系数。 18.在如下器件:电力二极管(Power Diode)、晶闸管(SCR)、门极可关断晶闸管(GTO)、电力晶体管(GTR)、电力场效应管(电力MOSFET)、绝缘栅双极型晶体管(IGBT)中,属于不可控器件的是_电力二极管__,属于半控型器件的是__晶闸管_,属于全控型器件的是_GTO 、GTR 、电力

电力电子技术试题及答案(B)

电力电子技术答案 2-1与信息电子电路中的二极管相比,电力二极管具有怎样的结构特点才使得其具有耐受高压和大电流的能力? 答:1.电力二极管大都采用垂直导电结构,使得硅片中通过电流的有效面积增大,显著提高了二极管的通流能力。 2.电力二极管在P 区和N 区之间多了一层低掺杂N 区,也称漂移区。低掺杂N 区由于掺杂浓度低而接近于无掺杂的纯半导体材料即本征半导体,由于掺杂浓度低,低掺杂N 区就可以承受很高的电压而不被击穿。 2-2. 使晶闸管导通的条件是什么? 答:使晶闸管导通的条件是:晶闸管承受正向阳极电压,并在门极施加触发电流(脉冲)。或:uAK>0且uGK>0。 2-3. 维持晶闸管导通的条件是什么?怎样才能使晶闸管由导通变为关断? 答:维持晶闸管导通的条件是使晶闸管的电流大于能保持晶闸管导通的最小电流,即维持电流。 要使晶闸管由导通变为关断, 可利用外加电压和外电路的作用使流过晶闸管的电流降 到接近于零的某一数值以下,即降到维持电流以下,便可使导通的晶闸管关断。 2-4图2-27中阴影部分为晶闸管处于通态区间的电流波形,各波形的电流最大值均为I m ,试计算各波形的电流平均值I d1、I d2、I d3与电流有效值I 、I 、I 。 πππ4 π4 π2 5π4a) b)c) 图1-43 图2-27 晶闸管导电波形 解:a) I d1= π21?π πωω4 )(sin t td I m =π2m I (122+)≈0.2717 I m I 1= ?π πωωπ 4 2 )()sin (21 t d t I m =2m I π 2143+≈0.4767 I m b) I d2 = π1?π πωω4)(sin t td I m =π m I ( 12 2 +)≈0.5434 I m I 2 = ? π π ωωπ 4 2) ()sin (1 t d t I m = 2 2m I π 21 43+ ≈0.6741I m c) I d3=π21?2 )(π ωt d I m =41 I m I 3 =? 2 2 ) (21π ωπt d I m = 2 1 I m 2-5上题中如果不考虑安全裕量,问100A 的晶阐管能送出的平均电流I d1、I d2、I d3各为多少?这时,相应的电流最大值I m1、I m2、 I m3各为多少? 解:额定电流I T(AV)=100A 的晶闸管,允许的电流有效值I=157A,由上题计算结果知 a) I m1≈4767.0I ≈329.35, I d1≈0.2717 I m1≈89.48 b) I m2≈ 6741 .0I ≈232.90, I d2≈0.5434 I m2≈126.56 c) I m3=2 I = 314, I d3= 4 1 I m3=78.5 2-6 GTO 和普通晶闸管同为PNPN 结构,为什么GTO 能够自关断,而普通晶闸管不能? 答:GTO 和普通晶阐管同为PNPN 结构,由P1N1P2和N1P2N2构成两个晶体管V1、V2,分别具有共基极电流增益 1α和2α, 由普通晶阐管的分析可得, 121=+αα是器件临界导通的条件。1 21>αα+两个等效晶体管过饱和而导通;

电力电子技术试卷3份答案

《电力电子技术》试卷1答案 一、填空(每空1分,36分) 1、请在正确的空格内标出下面元件的简称: 电力晶体管GTR;可关断晶闸管GTO;功率场效应晶体管MOSFET;绝缘栅双极型晶体管IGBT;IGBT是MOSFET和GTR的复合管。 2、晶闸管对触发脉冲的要求是要有足够的驱动功率、触发脉冲前沿要陡幅值要高和触发脉冲要与晶闸管阳极电压同步。 3、多个晶闸管相并联时必须考虑均流的问题,解决的方法是串专用均流电抗器。 4、在电流型逆变器中,输出电压波形为正弦波,输出电流波形为方波。 5、型号为KS100-8的元件表示双向晶闸管晶闸管、它的额定电压为800V伏、额定有效电流为100A。 6、180°导电型三相桥式逆变电路,晶闸管换相是在同一桥臂上的上、下二个元件之间进行;而120o导电型三相桥式逆变电路,晶闸管换相是在不同桥臂上的元件之间进行的。 7、当温度降低时,晶闸管的触发电流会增加、正反向漏电流会下降;当温度升高时,晶闸管的触发电流会下降、正反向漏电流会增加。 8、在有环流逆变系统中,环流指的是只流经逆变电源、逆变桥而不流经负载的电流。环流可在电路中加电抗器来限制。为了减小环流一般采控用控制角α大于β的工作方式。 9、常用的过电流保护措施有快速熔断器、串进线电抗器、接入直流快速开关、控制快速移相使输出电压下降。(写出四种即可) 10、双向晶闸管的触发方式有Ⅰ+、Ⅰ-、Ⅲ+、Ⅲ-四种。 二、判断题,(每题1分,10分)(对√、错×) 1、在半控桥整流带大电感负载不加续流二极管电路中,电路出故障时会出现失 控现象。(√) 2、在用两组反并联晶闸管的可逆系统,使直流电动机实现四象限运行时,其中 一组逆变器工作在整流状态,那么另一组就工作在逆变状态。(×) 3、晶闸管串联使用时,必须注意均流问题。(×) 4、逆变角太大会造成逆变失败。(×) 5、并联谐振逆变器必须是略呈电容性电路。(√) 6、给晶闸管加上正向阳极电压它就会导通。(×) 7、有源逆变指的是把直流电能转变成交流电能送给负载。(×) 8、在单相全控桥整流电路中,晶闸管的额定电压应取U2。(×) 9、在三相半波可控整流电路中,电路输出电压波形的脉动频率为300Hz。(×) 10、变频调速实际上是改变电动机内旋转磁场的速度达到改变输出转速的目的。 (√) 三、选择题(每题3分,15分)

电力电子技术课后题答案

0-1.什么是电力电子技术? 电力电子技术是应用于电力技术领域中的电子技术;它是以利用大功率电子器件对能量进行变换和控制为主要内容的技术。国际电气和电子工程师协会(IEEE)的电力电子学会对电力电子技术的定义为:“有效地使用电力半导体器件、应用电路和设计理论以及分析开发工具,实现对电能的高效能变换和控制的一门技术,它包括电压、电流、频率和波形等方面的变换。” 0-2.电力电子技术的基础与核心分别是什么? 电力电子器件是基础。电能变换技术是核心. 0-3.请列举电力电子技术的 3 个主要应用领域。 电源装置;电源电网净化设备;电机调速系统;电能传输和电力控制;清洁能源开发和新蓄能系统;照明及其它。 0-4.电能变换电路有哪几种形式?其常用基本控制方式有哪三种类型? AD-DC整流电;DC-AC逆变电路;AC-AC交流变换电路;DC-DC直流变换电路。 常用基本控制方式主要有三类:相控方式、频控方式、斩控方式。 0-5.从发展过程看,电力电子器件可分为哪几个阶段? 简述各阶段的主要标志。可分为:集成电晶闸管及其应用;自关断器件及其应用;功率集成电路和智能功率器件及其应用三个发展阶段。集成电晶闸管及其应用:大功率整流器。自关断器件及其应用:各类节能的全控型器件问世。功率集成电路和智能功率器件及其应用:功率集成电路(PIC),智能功率模块(IPM)器件发展。 0-6.传统电力电子技术与现代电力电子技术各自特征是什么? 传统电力电子技术的特征:电力电子器件以半控型晶闸管为主,变流电路一般 为相控型,控制技术多采用模拟控制方式。 现代电力电子技术特征:电力电子器件以全控型器件为主,变流电路采用脉宽 调制型,控制技术采用PWM数字控制技术。 0-7.电力电子技术的发展方向是什么? 新器件:器件性能优化,新型半导体材料。高频化与高效率。集成化与模块化。数字化。绿色化。 1-1.按可控性分类,电力电子器件分哪几类? 按可控性分类,电力电子器件分为不可控器件、半控器件和全控器件。 1-2.电力二极管有哪些类型?各类型电力二极管的反向恢复时间大约为多少? 电力二极管类型以及反向恢复时间如下: 1)普通二极管,反向恢复时间在5us以上。 2)快恢复二极管,反向恢复时间在5us以下。快恢复极管从性能上可分为快速恢复和超快速恢复二极管。前者反向恢复时间为数百纳秒或更长,后者在100ns 以下,甚至达到20~30ns,多用于高频整流和逆变电路中。 3)肖特基二极管,反向恢复时间为10~40ns。 1-3.在哪些情况下,晶闸管可以从断态转变为通态? 维持晶闸管导通的条件是什么? 1、正向的阳极电压; 2、正向的门极电流。两者缺一不可。阳极电流大于维持电流。

电力电子技术 复习题答案

第二章: 1.晶闸管的动态参数有断态电压临界上升率du/dt和通态电流临界上升率等,若 du/dt过大,就会使晶闸管出现_ 误导通_,若di/dt过大,会导致晶闸管_损坏__。 2.目前常用的具有自关断能力的电力电子元件有电力晶体管、可关断晶闸管、 功率场效应晶体管、绝缘栅双极型晶体管几种。简述晶闸管的正向伏安特性 答: 晶闸管的伏安特性 正向特性当IG=0时,如果在器件两端施加正向电压,则晶闸管处于正向阻断状态,只有很小的正向漏电流流过。 如果正向电压超过临界极限即正向转折电压Ubo,则漏电流急剧增大,器件开通。 随着门极电流幅值的增大,正向转折电压降低,晶闸管本身的压降很小,在1V左右。 如果门极电流为零,并且阳极电流降至接近于零的某一数值IH以下,则晶闸管又回到正向阻断状态,IH称为维持电流。 3.使晶闸管导通的条件是什么 答:使晶闸管导通的条件是:晶闸管承受正向阳极电压,并在门极施加触发电流(脉冲)。或:uAK>0且uGK>0。 4.在如下器件:电力二极管(Power Diode)、晶闸管(SCR)、门极可关断晶闸管 (GTO)、电力晶体管(GTR)、电力场效应管(电力MOSFET)、绝缘栅双极型晶体管(IGBT)中,属于半控型器件的是 SCR 。 5.晶闸管的擎住电流I L 答:晶闸管刚从断态转入通态并移除触发信号后,能维持导通所需的最小电流。 6.晶闸管通态平均电流I T(AV) 答:晶闸管在环境温度为40C和规定的冷却状态下,稳定结温不超过额定结温时所允许流过的最大工频正弦半波电流的平均值。标称其额定电流的参数。 7.晶闸管的控制角α(移相角) 答:从晶闸管开始承受正向阳极电压起到施加触发脉冲止的电角度,用a表示,也称触发角或控制角。

电力电子技术试题及答案(1)

《电力电子技术》试卷 一.填空(共15分,1分/空) 1.电力电子技术通常可分为()技术和()技术两个分支。 2.按驱动电路信号的性质可以将电力电子器件分为()型器件和()型器件两类,晶闸管属于其中的()型器件。 3.晶闸管单相桥式全控整流电路带反电动势负载E时(变压器二次侧电压有效值为U ,忽略主电路 2 各部分的电感),与电阻负载时相比,晶闸管提前了电角度δ停止导电,δ称为()角,数量关系为δ=()。 4.三相桥式全控整流电路的触发方式有()触发和()触发两种,常用的是()触发。 5.三相半波可控整流电路按联接方式可分为()组和()组两种。 6.在特定场合下,同一套整流电路即可工作在()状态,又可工作在()状态,故简称变流电路。 7.控制角α与逆变角β之间的关系为()。 二.单选(共10分,2分/题) 1.采用()是电力电子装置中最有效、应用最广的一种过电流保护措施。 A.直流断路器 B. 快速熔断器 C.过电流继电器 2.晶闸管属于()。 A.不可控器件 B. 全控器件 C.半控器件 3.单相全控桥式整流电路,带阻感负载(L足够大)时的移相范围是()。 A.180O B.90O C.120O 4.对三相全控桥中共阴极组的三个晶闸管来说,正常工作时触发脉冲相位应依次差()度。 A.60 B. 180 C. 120 5.把交流电变成直流电的是()。 A. 逆变电路 B.整流电路 C.斩波电路 三.多选(共10分,2分/题) 1.电力电子器件一般具有的特征有。 A.所能处理电功率的大小是其最重要的参数 B.一般工作在开关状态 C.一般需要信息电子电路来控制 D.不仅讲究散热设计,工作时一般还需接散热器 2.下列电路中,不存在变压器直流磁化问题的有。 A.单相全控桥整流电路 B.单相全波可控整流电路 C.三相全控桥整流电路 D.三相半波可控整流电路 3.使晶闸管关断的方法有。 A.给门极施加反压 B.去掉阳极的正向电压 C.增大回路阻抗 D.给阳极施加反压 4.逆变失败的原因有。 A.触发电路不可靠 B.晶闸管发生故障 C.交流电源发生故障 D.换相裕量角不足 5.变压器漏抗对整流电路的影响有。 A.输出电压平均值降低 B.整流电路的工作状态增多 C.晶闸管的di/dt减小 D.换相时晶闸管电压出现缺口 四.判断(共5分,1分/题) 1.三相全控桥式整流电路带电阻负载时的移相范围是150O。() 2.晶闸管是一种四层三端器件。()

王兆安版电力电子技术试卷及答案

20××-20××学年第一学期期末考试 《电力电子技术》试卷(A) (时间90分钟 满分100分) (适用于 ××学院 ××级 ××专业学生) 一、 填空题(30分,每空1分)。 1.如下器件:电力二极管(Power Diode )、晶闸管(SCR )、门极可关断晶闸管(GTO )、电力晶体管(GTR )、电力场效应管(电力MOSFET )、绝缘栅双极型晶体管(IGBT )中,属于不可控器件的是________,属于半控型器件的是________,属于全控型器件的是________;属于单极型电力电子器件的有________,属于双极型器件的有________,属于复合型电力电子器件得有 ________;在可控的器件中,容量最大的是________,工作频率最高的是________,属于电压驱动的是________,属于电流驱动的是________。(只写简称) 2.单相桥式全控整流电路中,带纯电阻负载时,α角移相范围为 _,单个晶闸管所承受的最大正向电压和反向电压分别为 和 ;带阻感负载时,α角移相范围为 ,单个晶闸管所承受的最大正向电压和反向电压分别为 和 。 3.直流斩波电路中最基本的两种电路是 和 。 4.升降压斩波电路呈现升压状态时,占空比取值范围是__ _。 5.与CuK 斩波电路电压的输入输出关系相同的有 、 和 。 6.当采用6脉波三相桥式电路且电网频率为50Hz 时,单相交交变频电路的输出上限频率约为 。 7.三相交交变频电路主要有两种接线方式,即 _和 。 8.矩阵式变频电路是近年来出现的一种新颖的变频电路。它采用的开关器件是 ;控制方式是 。 9.逆变器按直流侧提供的电源的性质来分,可分为 型逆变器和 型逆变器。 10.把电网频率的交流电直接变换成可调频率的交流电的变流电路称为 。 二、简答题(18分,每题6分)。 1.逆变电路多重化的目的是什么?如何实现?串联多重和并联多重逆变电路各应用于什么场合? 2.交流调压电路和交流调功电路有什么异同? 3.功率因数校正电路的作用是什么?有哪些校正方法?其基本原理是什么? 三、计算题(40分,1题20分,2题10分,3题10分)。 1.一单相交流调压器,电源为工频220V ,阻感串联作为负载,其中R=0.5Ω,L=2mH 。 试求:①开通角α的变化范围;②负载电流的最大有效值;③最大输出功率及此时电源侧的功率因数;④当2πα=时,晶闸管电流有效值,晶闸管导通角和电源侧功率因数。 2..三相桥式电压型逆变电路,工作在180°导电方式,U d =200V 。试求输出相电压的基波幅值U UN1m 和有效值U UN1、输出线电压的基波幅值U UV1m 和有效值U UV1、输出线电压中7次谐波的有效值U UV7。 3 .如图所示降压斩波电路E=100V ,L 值极大,R=0.5Ω,E m =10V ,采用脉宽调制控制方式,T=20μs ,当t on =5μs 时,计算输出电压平均值U o ,输出电流平均值

电力电子技术试卷及答案-第一章

电力电子技术试题(第一章) 一、填空题 1、普通晶闸管内部有PN结,,外部有三个电极,分别是极极和极。 1、三个、阳极A、阴极K、门极G。 2、晶闸管在其阳极与阴极之间加上电压的同时,门极上加上电压,晶闸管就导通。 2、正向、触发。 3、、晶闸管的工作状态有正向状态,正向状态和反向状态。 3、阻断、导通、阻断。 4、某半导体器件的型号为KP50—7的,其中KP表示该器件的名称为,50表示,7表示。 4、普通晶闸管、额定电流50A、额定电压700V。 5、只有当阳极电流小于电流时,晶闸管才会由导通转为截止。 5、维持电流。 6、当增大晶闸管可控整流的控制角α,负载上得到的直流电压平均值会。 6、减小。 7、按负载的性质不同,晶闸管可控整流电路的负载分为性负载,性负载和负载三大类。 7、电阻、电感、反电动势。 8、当晶闸管可控整流的负载为大电感负载时,负载两端的直流电压平均值会,解决的办法就是在负载的两端接一个。 8、减小、并接、续流二极管。 9、工作于反电动势负载的晶闸管在每一个周期中的导通角、电流波形不连续、呈状、电流的平均值。要求管子的额定电流值要些。 9、小、脉冲、小、大。 10、单结晶体管的内部一共有个PN结,外部一共有3个电极,它们分别是极、极和极。 10、一个、发射极E、第一基极B1、第二基极B2。 11、当单结晶体管的发射极电压高于电压时就导通;低于电 压时就截止。 11、峰点、谷点。 12、触发电路送出的触发脉冲信号必须与晶闸管阳极电压,保证在管子阳极电压每个正半周内以相同的被触发,才能得到稳定的直流电压。 12、同步、时刻。 13、晶体管触发电路的同步电压一般有同步电压和电压。 13、正弦波、锯齿波。 14、正弦波触发电路的同步移相一般都是采用与一个或几个的叠加,利用改变的大小,来实现移相控制。 14、正弦波同步电压、控制电压、控制电压。 15、在晶闸管两端并联的RC回路是用来防止损坏晶闸管的。 15、关断过电压。 16、为了防止雷电对晶闸管的损坏,可在整流变压器的一次线圈两端并接一个或。 16、硒堆、压敏电阻。 16、用来保护晶闸管过电流的熔断器叫。 16、快速熔断器。 二、判断题对的用√表示、错的用×表示(每小题1分、共10分) 1、普通晶闸管内部有两个PN结。(×) 2、普通晶闸管外部有三个电极,分别是基极、发射极和集电极。(×) 3、型号为KP50—7的半导体器件,是一个额定电流为50A的普通晶闸管。() 4、只要让加在晶闸管两端的电压减小为零,晶闸管就会关断。(×) 5、只要给门极加上触发电压,晶闸管就导通。(×) 6、晶闸管加上阳极电压后,不给门极加触发电压,晶闸管也会导通。(√) 7、加在晶闸管门极上的触发电压,最高不得超过100V。(×) 8、单向半控桥可控整流电路中,两只晶闸管采用的是“共阳”接法。(×) 9、晶闸管采用“共阴”接法或“共阳”接法都一样。(×) 10、增大晶闸管整流装置的控制角α,输出直流电压的平均值会增大。(×) 11、在触发电路中采用脉冲变压器可保障人员和设备的安全。(√) 12、为防止“关断过电压”损坏晶闸管,可在管子两端并接压敏电阻。(×) 13、雷击过电压可以用RC吸收回路来抑制。(×) 14、硒堆发生过电压击穿后就不能再使用了。(×) 15、晶闸管串联使用须采取“均压措施”。(√)

电力电子技术期末考试试题及答案

电力电子技术试题 第1章 电力电子器件 1.电力电子器件一般工作在__开关__状态。 2.在通常情况下,电力电子器件功率损耗主要为__通态损耗__,而当器件开关频率较高时,功率损耗主要为__开关损耗__。 3.电力电子器件组成的系统,一般由__控制电路__、_驱动电路_、 _主电路_三部分组成,由于电路中存在电压和电流的过冲,往往需添加_保护电路__。 4.按内部电子和空穴两种载流子参与导电的情况,电力电子器件可分为_单极型器件_ 、 _双极型器件_ 、_复合型器件_三类。 5.电力二极管的工作特性可概括为_承受正向电压导通,承受反相电压截止_。 6.电力二极管的主要类型有_普通二极管_、_快恢复二极管_、 _肖特基二极管_。 7.肖特基二极管的开关损耗_小于_快恢复二极管的开关损耗。 8.晶闸管的基本工作特性可概括为 __正向电压门极有触发则导通、反向电压则截止__ 。 9.对同一晶闸管,维持电流IH 与擎住电流I L 在数值大小上有I L __大于__IH 。 10.晶闸管断态不重复电压UDSM 与转折电压Ubo 数值大小上应为,UDSM _大于__Ubo 。 11.逆导晶闸管是将_二极管_与晶闸管_反并联_(如何连接)在同一管芯上的功率集成器件。 12.GTO 的__多元集成__结构是为了便于实现门极控制关断而设计的。 13.MOSFET 的漏极伏安特性中的三个区域与GTR 共发射极接法时的输出特性中的三个区域有对应关系,其中前者的截止区对应后者的_截止区_、前者的饱和区对应后者的__放大区__、前者的非饱和区对应后者的_饱和区__。 14.电力MOSFET 的通态电阻具有__正__温度系数。 15.IGBT 的开启电压UGE (th )随温度升高而_略有下降__,开关速度__小于__电力MOSFET 。 16.按照驱动电路加在电力电子器件控制端和公共端之间的性质,可将电力电子器件分为_电压驱动型_和_电流驱动型_两类。 17.IGBT 的通态压降在1/2或1/3额定电流以下区段具有__负___温度系数, 在1/2或1/3额定电流以上区段具有__正___温度系数。 18.在如下器件:电力二极管(Power Diode )、晶闸管(SCR )、门极可关断晶闸管(GTO )、电力晶体管(GTR )、电力场效应管(电力MOSFET )、绝缘栅双极型晶体管(IGBT )中,属于不可控器件的是_电力二极管__,属于半控型器件的是__晶闸管_,属于全控型器件的是_ GTO 、GTR 、电力MOSFET 、IGBT _;属于单极型电力电子器件的有_电力MOSFET _,属于双极型器件的有_电力二极管、晶闸管、GTO 、GTR _,属于复合型电力电子器件得有 __ IGBT _;在可控的器件中,容量最大的是_晶闸管_,工作频率最高的是_电力MOSFET ,属于电压驱动的是电力MOSFET 、IGBT _,属于电流驱动的是_晶闸管、GTO 、GTR _。 第2章 整流电路 1.电阻负载的特点是_电压和电流成正比且波形相同_,在单相半波可控整流电阻性负载电路中,晶闸管控制角α的最大移相范围是_0-180O _。 2.阻感负载的特点是_流过电感的电流不能突变,在单相半波可控整流带阻感负载并联续流二极管的电路中,晶闸管控制角α的最大移相范围是__0-180O _ , 2__,续流二极管承受的最大反向电压为2_(设U 2为相电压有效值)。 3.单相桥式全控整流电路中,带纯电阻负载时,α角移相范围为__0-180O _,单个晶闸管所承受的最大正向电压和反向电压分别为22 和2;带 阻感负载时,α角移相范围为_0-90O _,单个晶闸管所承受的最大正向电压和反向电压分别为2_和2_;带反电动势负载时,欲使电阻上的电流不出 现断续现象,可在主电路中直流输出侧串联一个_平波电抗器_。 4.单相全控桥反电动势负载电路中,当控制角α大于不导电角δ时,晶闸管的导通角θ =_π-α-δ_; 当控制角α小于不导电角 δ 时,晶闸管的导通角 θ =_π-2δ_。 5.电阻性负载三相半波可控整流电路中,晶闸管所承受的最大正向电压UFm 2_,晶闸管控制角α的最大移相范围是_0-150o _,使负载电流连续的条 件为__o 30≤α__(U2为相电压有效值)。 6.三相半波可控整流电路中的三个晶闸管的触发脉冲相位按相序依次互差_120o _,当它带阻感负载时,α的移相范围为__0-90o _。 7.三相桥式全控整流电路带电阻负载工作中,共阴极组中处于通态的晶闸管对应的是_最高__的相电压,而共阳极组中处于导通的晶闸管对应的是_最低_的相电压;这种电路 α 角的移相范围是_0-120o _,u d 波形连续的条件是_o 60≤α_。 8.对于三相半波可控整流电路,换相重迭角的影响,将使用输出电压平均值__下降_。 9.电容滤波单相不可控整流带电阻负载电路中,空载时,输出电压为2_,随负载加重Ud 逐渐趋近于_0.9 U 2_,通常设计时,应取RC ≥_1.5-2.5_T ,此 时输出电压为Ud ≈__1.2_U 2(U 2为相电压有效值,T 为交流电源的周期)。 10.电容滤波三相不可控整流带电阻负载电路中,电流 id 断续和连续的临界条件是_=RC ω。 11.实际工作中,整流电路输出的电压是周期性的非正弦函数,当 α 从0°~90°变化时,整流输出的电压ud 的谐波幅值随 α 的增大而 _增大_,当 α 从90°~180°变化时,整流输出的电压 ud 的谐波幅值随 α 的增大而_减小_。

电力电子技术试题20套及答案

考试试卷( 1 )卷 一、填空题(本题共8小题,每空1分,共20分) 1、电子技术包括______________和电力电子技术两大分支,通常所说的模拟电子技术和数字电子技术就属 于前者。 2、为减少自身损耗,提高效率,电力电子器件一般都工作在_________状态。当器件的工作频率较高时, _________损耗会成为主要的损耗。 3、在PWM控制电路中,载波频率与调制信号频率之比称为_____________,当它为常数时的调制方式称为 _________调制。在逆变电路的输出频率范围划分成若干频段,每个频段内载波频率与调制信号频率之比为桓定的调制方式称为___分段同步_________调制。 4、面积等效原理指的是,___冲量______相等而__形状_____不同的窄脉冲加在具有惯性的环节上时,其效果 基本相同。 5、在GTR、GTO、IGBT与MOSFET中,开关速度最快的是__MOSFET_,单管输出功率最大 的是__GTO_,应用最为广泛的是__IGBT___。 6、设三相电源的相电压为U2,三相半波可控整流电路接电阻负载时,晶闸管可能承受的最大 反向电压为电源线电压的峰值,即,其承受的最大正向电压为。 7、逆变电路的负载如果接到电源,则称为逆变,如果接到负载,则称为逆变。 8、如下图,指出单相半桥电压型逆变电路工作过程中各时间段电流流经的通路(用V1,VD1,V2,VD2表示)。 (1) 0~t1时间段内,电流的通路为___VD1_____;(2) t1~t2时间段内,电流的通路为___V1____; (3) t2~t3时间段内,电流的通路为__VD2_____;(4) t3~t4时间段内,电流的通路为__V2_____;(5) t4~t5 时间段内,电流的通路为___VD1____;

2010电力电子技术参考答案(A)

………密………封………线………以………内………答………题………无………效…… 电子科技大学二零零九至二零一零学年第二学期期末考试 电力电子技术课程考试题 A 卷(120 分钟)考试形式:闭卷考试日期201 年月日课程成绩构成:平时10 分,期中10 分,实验10 分,期末70 分 一、填空题(本大题共十二小题每空一分,共44分) 1.请在空格内标出下面元件的简称:电力晶体GTR ;可关断晶闸管__SCR___;功率场效应晶体管 MOSFET;绝缘栅双极型晶体管IGBT;IGBT是MOSFET和GTR的复合管。 2.在电流型逆变器中,输出电压波形为正弦波,输出电流波形为___方___波。 3.180°导电型三相桥式逆变电路,晶闸管换相是在同一桥臂上的上、下二个元件之间进行;而120o 导电型三相桥式逆变电路,晶闸管换相是在不同桥臂上的元件之间进行的。 4.直流斩波电路按照输入电压与输出电压的高低变化来分类有_降压斩波电路;升压斩波电路; 升降压斩波电路。 5.为了减小变流电路的开、关损耗,通常让元件工作在软开关状态,软开关电路种类很多,但归纳 起来可分为零电流开关与零电压开关两大类。 6.直流斩波电路在改变负载的直流电压时,常用的控制方式有等频调宽控制;等宽调频控制;脉宽 与频率同时控制三种。 7.通常变流电路实现换流的方式有器件换流,电网换流,负载换流,强迫换流四种。 8.普通晶闸管的图形符号是,三个电极分别是阳极A,阴极K 和门极G晶闸管的导通 条件是阳极加正电压,阴极接负电压,门极接正向电压形成了足够门极电流时晶闸管导通;关断条件是: 当晶闸管阳极电流小于维持电流I H时,导通的晶闸管关断。 9.有源逆变指的是把直流能量转变成交流能量后送给电网的装置。 10.造成逆变失败的原因有逆变桥晶闸管或元件损坏,供电电源缺相,逆变角太小,触发脉冲丢失或 未按时到达等几种。 11.锯齿波触发电路的主要环节是由同步环节;锯齿波形成;脉冲形成;整形放大;强触发及输出环

电力电子技术试卷及答案..

一、填空题(每空1分,34分) 1、实现有源逆变的条件为和。 2、在由两组反并联变流装置组成的直流电机的四象限运行系统中,两组变流装置分别工作在正组状态、状态、反组状态、状态。 3、在有环流反并联可逆系统中,环流指的是只流经而不流经 的电流。为了减小环流,一般采用αβ状态。 4、有源逆变指的是把能量转变成能量后送给装置。 5、给晶闸管阳极加上一定的电压;在门极加上电压,并形成足够的电流,晶闸管才能导通。 6、当负载为大电感负载,如不加续流二极管时,在电路中出现触发脉冲丢失时 与电路会出现失控现象。 7、三相半波可控整流电路,输出到负载的平均电压波形脉动频率为H Z;而三相全控桥整流电路,输出到负载的平均电压波形脉动频率为H Z;这说明电路的纹波系数比电路要小。 8、造成逆变失败的原因有、、、等几种。 9、提高可控整流电路的功率因数的措施有、、、等四种。 10、晶闸管在触发开通过程中,当阳极电流小于电流之前,如去掉脉冲,晶闸管又会关断。 三、选择题(10分) 1、在单相全控桥整流电路中,两对晶闸管的触发脉冲,应依次相差度。 A 、180度;B、60度;C、360度;D、120度; 2、α= 度时,三相半波可控整流电路,在电阻性负载时,输出电压波形处于连续和断续的临界状态。 A、0度; B、60度; C 、30度;D、120度; 3、通常在晶闸管触发电路中,若改变的大小时,输出脉冲相位产生移动,达到移相控制的目的。 A、同步电压; B、控制电压; C、脉冲变压器变比; 4、可实现有源逆变的电路为。

A、单相全控桥可控整流电路 B、三相半控桥可控整流电路 C、单相全控桥接续流二极管电路 D、单相半控桥整流电路 5、由晶闸管构成的可逆调速系统中,逆变角βmin选时系统工作才可靠。 A、300~350 B、100~150 C、00~100 D、00 四、问答题(每题9分,18分) 1、什么是逆变失败?形成的原因是什么? 2、为使晶闸管变流装置正常工作,触发电路必须满足什么要求? 五、分析、计算题:(每题9分,18分) 1、三相半波可控整流电路,整流变压器的联接组别是D/Y—5,锯齿波同步触发电路中的信号综合管是NPN型三极管。试确定同步变压器TS的接法钟点数为几点钟时,触发同步定相才是正确的,并画出矢量图。(8分) 2、如图所示:变压器一次电压有效值为220V,二次各段电压有效值为100V,电阻性负载,R=10Ω,当控制角α=90o时,求输出整流电压的平均值U d,负载电流I d,并绘出晶闸管、整流二极管和变压器一次绕组电流的波形。(10分) 一、填空题 1、要有一个直流逆变电源,它的极性方向与晶闸管的导通方向一致,其幅极应稍大于逆变桥直流侧输出的平均

《电力电子技术》课后答案完整版

王兆安《电力电子技术》(第4版)课后习题解 第1章 电力电子器件 1.1 使晶闸管导通的条件是什么? 答:使晶闸管导通的条件是:晶闸管承受正向阳极电压,并在门极注入正向触发电流。 1.2 维持晶闸管导通的条件是什么?怎样才能使晶闸管由导通变为关断? 答:维持晶闸管导通的条件是使晶闸管的电流大于能保持晶闸管导通的最小电流(即维持电流),即H A I I >。 要使晶闸管由导通变为关断,可通过外加反向阳极电压或减小负载电流的办法,使流过晶闸管的电流降到维持电流值以下,即H A I I <。 1.3 图1-43中阴影部分为晶闸管处于通态区间的电流波形,各波形的电流最大值均为m I 。 试计算各波形的电流平均值1d I ,2d I ,3d I 与电流有效值1I ,2I ,3I 。 解:a ) m m m d I I t d t I I 2717.0)12 2( 2)()(sin 214 1≈+= = ?π ωωπ ππ m m m I I t d t I I 4767.021432)()sin (214 21≈+= =? π ω?πππ b ) m m m d I I t d t I I 5434.0)12 2 ( )()(sin 1 4 2≈+= = ? π ωωπ ππ m m m I I t d t I I 6471.0214322)()sin (1 4 22≈+= = ? π ω?π ππ c ) ? = =20 3 4 1)(21π ωπ m m d I t d I I m m I t d I I 2 1)(21 20 2 3= = ? ωπ π 1.4 上题中如果不考虑安全裕量,问100A 的晶阐管能送出的平均电流1d I 、2d I 、3d I 各为多少?这时,相应的电流最大值1m I 、2m I 、3m I 各为多少? 解:额定电流A I AV T 100)(=的晶闸管,允许的电流有效值A I 157=,由上题计算结果知: a ) A I I m 35.3294767.01≈≈ A I I m d 48.892717.011≈≈ b ) A I I m 90.2326741 .02≈≈ A I I m d 56.1265434.022≈≈ c ) A I I m 31423== A I I m d 5.784 1 33== 1.5 GTO 和普通晶闸管同为PNPN 结构,为什么GTO 能够自关断,而普通晶闸管不能? 答:GTO 和普通晶阐管同为PNPN 结构,由211P N P 和221N P N 构成两个晶体管1V 、2V ,分别具有共基极电 流增益1α和2α,由普通晶阐管的分析可得,121=+αα是器件临界导通的条件。121> αα+两个等效晶体管过饱和而导通;121< αα+不能维持饱和导通而关断。 GTO 之所以能够自行关断,而普通晶闸管不能,是因为GTO 与普通晶闸管在设计和工艺方面有以下几点 图1-43 晶闸管导电波形

电力电子技术期末考试试题及答案

电力电子技术试题 第1章 电力电子器件 1.电力电子器件一般工作在__开关__状态。 5.电力二极管的工作特性可概括为_承受正向电压导通,承受反相电压截止_。 6.电力二极管的主要类型有_普通二极管_、_快恢复二极管_、 _肖特基二极管_。 8.晶闸管的基本工作特性可概括为 __正向电压门极有触发则导通、反向电压则截止__ 。 18.在如下器件:电力二极管(Power Diode )、晶闸管(SCR )、门极可关断晶闸管(GTO )、电力晶体管(GTR )、电力场效应管(电力MOSFET )、绝缘栅双极型晶体管(IGBT )中,属于不可控器件的是_电力二极管__,属于半控型器件的是__晶闸管_,属于全控型器件的是_ GTO 、GTR 、电力MOSFET 、IGBT _;属于单极型电力电子器件的有_电力MOSFET _,属于双极型器件的有_电力二极管、晶闸管、GTO 、GTR _,属于复合型电力电子器件得有 __ IGBT _;在可控的器件中,容量最大的是_晶闸管_,工作频率最高的是_电力MOSFET ,属于电压驱动的是电力MOSFET 、IGBT _,属于电流驱动的是_晶闸管、GTO 、GTR _。 2、可关断晶闸管的图形符号是 ;电力场效应晶体管的图形符号是 绝缘栅双极晶体管的图形符号是 ;电力晶体管的图形符号是 ; 第2章 整流电路 1.电阻负载的特点是_电压和电流成正比且波形相同_,在单相半波可控整流电阻性负载电路中,晶闸管控制角α的最大移相范围是_0-180O _。 2.阻感负载的特点是_流过电感的电流不能突变,在单相半波可控整流带阻感负载并联续流二极管的电路中,晶闸管控制角α的最大移相范围是__0-180O _ ,其承受的最大正反向电压均为_22U __,续流二极管承受的最大反向电压为__22U _(设U 2为相电压有效值)。 3.单相桥式全控整流电路中,带纯电阻负载时,α角移相范围为__0-180O _,单个晶闸管所承受的最大正向电压和反向电压分别为__222U 和_22U ; 带阻感负载时,α角移相范围为_0-90O _,单个晶闸管所承受的最大正向电压和反向电压分别为__22U _和__22U _;带反电动势负载时,欲使电阻上的 电流不出现断续现象,可在主电路中直流输出侧串联一个_平波电抗器_。 5.电阻性负载三相半波可控整流电路中,晶闸管所承受的最大正向电压UFm 等于__22U _,晶闸管控制角α的最大移相范围是_0-150o _,使负载电流连 续的条件为__o 30≤α__(U2为相电压有效值)。 6.三相半波可控整流电路中的三个晶闸管的触发脉冲相位按相序依次互差_120o _,当它带阻感负载时,α的移相范围为__0-90o _。 7.三相桥式全控整流电路带电阻负载工作中,共阴极组中处于通态的晶闸管对应的是_最高__的相电压,而共阳极组中处于导通的晶闸管对应的是_最低_ 的相电压;这种电路 α 角的移相范围是_0-120o _,u d 波形连续的条件是_o 60≤α_。 8.对于三相半波可控整流电路,换相重迭角的影响,将使用输出电压平均值__下降_。 11.实际工作中,整流电路输出的电压是周期性的非正弦函数,当 α 从0°~90°变化时,整流输出的电压ud 的谐波幅值随 α 的增大而 _增大_,当 α 从90°~180°变化时,整流输出的电压 ud 的谐波幅值随 α 的增大而_减小_。 12.逆变电路中,当交流侧和电网连结时,这种电路称为_有源逆变_,欲实现有源逆变,只能采用__全控_电路;对于单相全波电路,当控制角 0< α < π /2 时,电路工作在__整流_状态; π /2< α < π 时,电路工作在__逆变_状态。 13.在整流电路中,能够实现有源逆变的有_单相全波_、_三相桥式整流电路_等(可控整流电路均可),其工作在有源逆变状态的条件是_有直流电动势,其极性和晶闸管导通方向一致,其值大于变流器直流侧平均电压_和__晶闸管的控制角α > 90O ,使输出平均电压U d 为负值_。 第3章 直流斩波电路 1.直流斩波电路完成得是直流到_直流_的变换。 2.直流斩波电路中最基本的两种电路是_降压斩波电路 和_升压斩波电路_。 3.斩波电路有三种控制方式:_脉冲宽度调制(PWM )_、_频率调制_和_(t on 和T 都可调,改变占空比)混合型。 6.CuK 斩波电路电压的输入输出关系相同的有__升压斩波电路___、__Sepic 斩波电路_和__Zeta 斩波电路__。 7.Sepic 斩波电路和Zeta 斩波电路具有相同的输入输出关系,所不同的是:_ Sepic 斩波电路_的电源电流和负载电流均连续,_ Zeta 斩波电路_的输入、输出电流均是断续的,但两种电路输出的电压都为__正_极性的 。 8.斩波电路用于拖动直流电动机时,降压斩波电路能使电动机工作于第__1__象限,升压斩波电路能使电动机工作于第__2__象限,_电流可逆斩波电路能使电动机工作于第1和第2象限。 10.复合斩波电路中,电流可逆斩波电路可看作一个_升压_斩波电路和一个__降压_斩波电路的组合;

相关主题