CN107294775B - Communication network optimization method based on analytic hierarchy process and genetic algorithm - Google Patents
Communication network optimization method based on analytic hierarchy process and genetic algorithm Download PDFInfo
- Publication number
- CN107294775B CN107294775B CN201710428743.2A CN201710428743A CN107294775B CN 107294775 B CN107294775 B CN 107294775B CN 201710428743 A CN201710428743 A CN 201710428743A CN 107294775 B CN107294775 B CN 107294775B
- Authority
- CN
- China
- Prior art keywords
- network
- service
- scheme
- index
- matrix
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
- 238000004891 communication Methods 0.000 title claims abstract description 76
- 238000000034 method Methods 0.000 title claims abstract description 76
- 230000002068 genetic effect Effects 0.000 title claims abstract description 42
- 238000005457 optimization Methods 0.000 title claims abstract description 30
- 230000008569 process Effects 0.000 title claims abstract description 23
- 238000009826 distribution Methods 0.000 claims abstract description 24
- 239000011159 matrix material Substances 0.000 claims description 59
- 239000013307 optical fiber Substances 0.000 claims description 49
- 238000011156 evaluation Methods 0.000 claims description 26
- 239000000835 fiber Substances 0.000 claims description 23
- 230000035772 mutation Effects 0.000 claims description 20
- 108090000623 proteins and genes Proteins 0.000 claims description 19
- 238000004364 calculation method Methods 0.000 claims description 14
- 230000013011 mating Effects 0.000 claims description 12
- 238000007726 management method Methods 0.000 claims description 8
- 238000005070 sampling Methods 0.000 claims description 8
- 238000010276 construction Methods 0.000 claims description 6
- 238000011160 research Methods 0.000 claims description 6
- 230000008859 change Effects 0.000 claims description 5
- 230000007614 genetic variation Effects 0.000 claims description 5
- 230000001419 dependent effect Effects 0.000 claims description 3
- 230000004069 differentiation Effects 0.000 claims description 3
- 238000012417 linear regression Methods 0.000 claims description 3
- 238000009827 uniform distribution Methods 0.000 claims description 3
- 230000000694 effects Effects 0.000 claims description 2
- 102000004169 proteins and genes Human genes 0.000 claims description 2
- 230000009467 reduction Effects 0.000 claims description 2
- 239000000126 substance Substances 0.000 claims 1
- 238000004458 analytical method Methods 0.000 abstract description 5
- 238000004445 quantitative analysis Methods 0.000 abstract description 2
- 230000008901 benefit Effects 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 208000001613 Gambling Diseases 0.000 description 1
- 206010064571 Gene mutation Diseases 0.000 description 1
- 206010063385 Intellectualisation Diseases 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000005520 cutting process Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000035800 maturation Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 230000002028 premature Effects 0.000 description 1
- 238000010187 selection method Methods 0.000 description 1
- 230000009897 systematic effect Effects 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L45/00—Routing or path finding of packets in data switching networks
- H04L45/02—Topology update or discovery
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/06—Management of faults, events, alarms or notifications
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Other Investigation Or Analysis Of Materials By Electrical Means (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention belongs to the problem of network optimization, relates to optimization of a service route distribution scheme in a power communication network, and particularly relates to a communication network optimization method based on an analytic hierarchy process and a genetic algorithm. The invention starts from the overall risk degree and the node and business risk balance degree of the power communication network, adopts an analytic hierarchy process and combines a qualitative and quantitative method to stratify a complex decision system, and provides quantitative basis for analysis and decision by comparing the importance of various associated factors layer by layer, thereby providing a standard for evaluating the quality of the communication network. And aiming at the standard, an improved genetic algorithm is adopted, a new coding mode and a variation mode are provided, the problem that the expression capacity of the binary coding of the traditional genetic algorithm to the problem is insufficient is solved, and the communication network is optimized.
Description
Technical Field
The invention belongs to the problem of network optimization, and particularly relates to optimization of a service routing distribution scheme in a power communication network.
Background
The power communication network is an infrastructure for intellectualization and automation of power production, and as an important means for guaranteeing the safety, stability and economic operation of the power grid, the power communication network occupies a great position in a power system.
Meanwhile, the power communication network is used as an infrastructure for bearing interactive information between power systems, once a safety event occurs in the power communication network, the service quality of the power communication system is directly influenced, and the safe and stable operation of the power network is threatened. In recent years, with the increasing scale of power systems, the maintenance and management aspects are greatly improved, but the failure rate and the accident frequency of the power communication network are not reduced correspondingly. Meanwhile, the low resource utilization rate is also a ubiquitous problem in the power communication network. The method mainly shows that various power grid service channels are unreasonable in organization, the channel organizations of different types of services do not have clearly divided management interfaces, and the condition of adopting roundabout switching routes exists in large quantity; the method is lack of predictability of the service, circuit time slots are scattered due to the fact that network service cutting and expanding are carried out for multiple times in the construction process, resources occupied by emergency service rush-through are not released in time, and the like, and therefore cross resources of transmission equipment are wasted.
The communication network optimization is used as an important component of communication network construction, and aims to provide a correct course for the development of power communication, realize maximum saving in the aspects of network structure optimization and improvement of various power grid service carrying capacities and obtain maximum benefits. Specifically, a reasonable planning scheme for a newly-built network and an optimization and modification scheme for the existing network can be provided from multiple aspects such as a network structure, a bearing service, bandwidth management and scheduling by deeply analyzing the current situation and a service model of a communication transmission network, and corresponding research results are obtained in the aspect. For example, great waves et al propose a service-oriented risk balancing route allocation mechanism to reduce the overall risk of a channel segment and a network, Hao Wang et al propose a secondary problem of a power grid and a service center to solve the load balancing problem in the network, tornarore M et al indicate the difference between the statistical availability and the actual availability of the channel, and propose an improved availability route allocation algorithm (3W-availability Aware Routing,3WAR) to effectively reduce the difference between the actual availability and the target availability. Although the method can improve the Availability of the service channel, the Availability-Aware Routing (AAR) algorithm cannot avoid the phenomenon that important power communication services are too concentrated, and further increases the network operation risk. In the aspect of load balancing, the Liu Nian-zu and Chen Xiao utilizes a heuristic algorithm (Compact Tree, CT) to obtain the first k shortest paths between two points, and the load balancing is realized by taking the path with the maximum available bandwidth as a service route. However, the algorithm only performs balanced distribution of the service from the aspect of bandwidth, and does not consider the service importance, so that the algorithm is not suitable for solving the problem of service risk distribution based on the service importance in the power communication network. Furthermore, the inventor of the present application proposed a balance network topology optimization algorithm based on node importance, but the algorithm does not evaluate the reliability of the power communication network from the business level.
In order to comprehensively evaluate the optimized scheme, the invention introduces an analytic hierarchy process. AHP is a systematic analysis method for analyzing qualitative and quantitative combination of multi-target, multi-criterion, multi-factor and complex large systems, which stratifies a complex decision-making system and provides quantitative basis for analysis and decision-making by comparing importance of various associated factors layer by layer. By applying the method, subjective influence is checked and reduced to a certain extent, so that evaluation tends to be more scientific, and basis is provided for selecting the optimal decision.
On the other hand, as communication networks expand, the computational load for finding an optimal routing scheme rapidly increases. Numerical calculation results show that as the network cluster grows, the genetic algorithm can improve topology optimization by two orders of magnitude compared with exhaustive search [7 ]. Aiming at the problem of communication network optimization, the invention improves the genetic algorithm and provides a new coding mode. Compared with the traditional heuristic algorithm, the method can increase the search space and find a better solution.
Disclosure of Invention
As cities develop, communication networks become increasingly large. Node diversity and traffic diversity also make optimization of communication networks a difficult problem. The invention provides a standard for evaluating the quality of a communication network based on the overall risk degree of the network and the risk balance degree of the nodes and the service. And aiming at the standard, the genetic algorithm is improved, a new coding mode and a new variation mode are provided, and the communication network is optimized.
A communication network optimization method based on an analytic hierarchy process and a genetic algorithm is characterized by comprising the following steps:
step 1, calculating a service weight w for each service from the service type and the service scheduling grade; the service types of the power communication network mainly include distribution automation, dispatching automation and relay protectionProtection, network management service, full-automatic device, dispatching telephone, frequency control service, etc.; the service scheduling level can be divided into a first-level network, a second-level network, a third-level network, a fourth-level network and a fifth-level network; the service weight w is the service type weight wTAnd scheduling rank weight wDA weighted sum of (a);
step 2, analyzing the failure rate of the computing equipment and the optical fiber; through data research, the relation function of the equipment failure rate and the number of services borne by the equipment is a piecewise function, when the number of the services is smaller than a certain value, the failure rate is a stable value, and when the number of the services exceeds the certain value, the failure rate and the number of the services borne by the equipment are in a nonlinear relation; the failure rate of the optical fiber is not changed along with the number of the services, is only related to the inherent attributes of the optical fiber and can be obtained by historical data;
step 3, calculating a standard for evaluating the reliability of the communication network, evaluating the reliability of the communication network by using three indexes of network total risk, resource risk balance and service risk balance, and acquiring the network total risk, the resource risk balance and the service risk balance based on the equipment risk, the optical fiber risk and the service risk;
step 4, based on the improved genetic algorithm, adopting a new coding mode and a new variation mode to optimize the network routing scheme;
and 5, optimizing the communication network according to the standard for evaluating the network reliability in the step 3 by using the improved genetic algorithm in the step 4.
In the above communication network optimization method based on the analytic hierarchy process and the genetic algorithm, the specific operation method in step 1 is as follows: respectively scoring service types { distribution automation, scheduling automation, relay protection, network management service, full-automatic devices, scheduling telephone, administrative telephone, integrated data network, scheduling data, video teleconference, protection PCM, communication PCM, video service, frequency control service } and service scheduling grades { primary network, secondary network, tertiary network, quaternary network and quinary network }, wherein the scoring interval is [1,10], and the score is given by relevant experts;
w=0.7×wT+0.3×wD
wherein, wTTo businessType of service, wDRating the traffic scheduling level, wT∈[1,10], wD∈[1,10](ii) a The score of each service, i.e. the weight w, can be calculated according to the service type and the scheduling level of each service.
In the above method for optimizing a communication network based on an analytic hierarchy process and a genetic algorithm, the step 2 specifically includes the following substeps:
step 2.1, obtaining the failure rate of the equipment: the observation and research of the existing data show that when the number n of the services borne by the equipment is less than a certain value, the failure rate of the equipment is a stable value; when the number n of the services is increased again, the relation between the equipment use failure rate and the number of the services is in a nonlinear relation; the nonlinear regression model is expressed as
yi=f(xi,θ)+εi,i=1,2,...,n
Wherein, yiIs a dependent variable; non-random vector xi=(xi1,xi2,...,xik) ' is an independent variable; theta ═ theta0,θ1,...,θp) ' is an unknown parameter vector; epsiloniAre random error terms and satisfy independent identically distributed assumptions;
for non-linear regression models, the parameter θ is estimated using the least squares method, i.e. the method is used to
To a minimumCalled theta as nonlinear least square estimation; when the f function is assumed to be continuously differentiable for the parameter theta, a normal equation set can be established by using a differentiation method to minimize Q (theta)The Q function is applied to the parameter thetajCalculating the partial derivatives and making them be 0 to obtain p +1 equations
defining the device utilization failure rate η (n) function as a piecewise function, i.e.
Where n is the number of services, c is a constant value, α is a parameter, which is equivalent to that in the nonlinear model
Step 2.2, obtaining the fiber fault rate: defining that the failure rate of the optical fiber does not change with the number of the services carried by the optical fiber and is only related to the inherent attributes of the optical fiber; according to the invention, the failure rate of each optical fiber can be obtained according to the ratio of the historical failure days to the working days, and then the average failure rate of each type of optical fiber is calculated according to the difference of the types of the optical fibers, namely the failure rate of the optical fiber.
In the above communication network optimization method based on the analytic hierarchy process and the genetic algorithm, step 3, the network reliability index is obtained based on the following definitions: definition e denotes the total equipment (equipment) of the communication network, f denotes the fiber (fiber), b denotes the traffic (business); n ise,nf,nbRespectively representing the total number of equipment, optical fibers and services in the network; in a network, one device or optical fiber usually carries a plurality of services, and one service is also carried on a plurality of devices and optical fibers; device eiB (e) for carried service seti) Denotes, | B (e)i) I represents the number of services in the set, fiber fiB (f) for carried service seti) Represents; likewise, bearer service biE (b) for device integrationi) F (b) for showing, optical fiber collectioni) Represents; service biW for the weight valueiIndicating that the failure rate of the device, in addition to being related to the intrinsic properties of the device, is related to the amount of traffic carried by the device, whereas the failure rate of the fiber is related to the intrinsic properties of the fiber, respectively η (e)i,|B(ei) I) and η (f)i) Indicating the failure rate of the equipment and the optical fiber;
device eiDegree of risk (ERD)iIndicating the possibility of failure of the device and the effect on the whole network after failure; the calculation formula is
Likewise, the optical fiber fiDegree of risk FRDiIs calculated by the formula
Service biDegree of risk BRDiIndicating the possibility of failure of the service and the impact on the network; the probability of the service failure is the probability of any resource bearing the service failure; the calculation formula is
The total risk (NRD) of the network is the sum of the risks of all resources in the network and is calculated by the formula
The Resource Risk Balance (RRBD) is the reciprocal of the variance sum of the risk of equipment and optical fibers in the network and has the calculation formula of
Also, the Business Risk Balance (BRBD) is calculated as
Optimization of a communication network requires a reduction in the overall risk (NRD) of the network, while improving the Resource Risk Balance (RRBD) and Business Risk Balance (BRBD) of the network.
In the above communication network optimization method based on the analytic hierarchy process and the genetic algorithm, in step 3, the evaluation method for the reliability of the communication network is performed based on the following steps:
step 3.1, construction of an evaluation index system: aiming at the reliability of the communication network route scheme, three layers of evaluation index systems are constructed, namely a scheme layer, a factor layer and an index layer,
step 3.2, constructing a same degree decision matrix, which specifically comprises the following steps:
(1) defining a communication network route plan with N1,N2,...,NmM schemes, each scheme having p1,p2,...,pnObtaining a multi-scheme evaluation decision matrix S by each index, namely
Selecting the optimal value of each category index in m schemes to form an ideal scheme A0,
A0=[p01,p02,...,p0n]T
The total network risk degree takes the minimum value, the resource risk balance degree takes the maximum value, and the service risk balance degree takes the maximum value; note A0The f (f) index has a value of p0f;
(2) Constructing a similarity matrix of the evaluated scheme index and the ideal scheme index: from the evaluated scheme AiEach index p in (i ═ 1, 2.., m)ijAnd the ideal scheme A0Middle correspondence index value p0fOf a same degree aijThe identity matrix T can be obtained as
Wherein, for the index of the maximum value of the corresponding index of each scheme in the ideal scheme, the index value in each scheme is divided by the corresponding index value in the ideal scheme, namely aij=pij/pof(ii) a For the index of the minimum value of the corresponding index of each scheme in the ideal scheme, dividing the index value in the ideal scheme by the corresponding index value in each scheme, namely aij=pof/pij;
Step 3.3, calculating the weight of the evaluation index: calculating the weight of the evaluation index by adopting an analytic hierarchy process, wherein the specific calculation steps are as follows;
3.31, constructing an n multiplied by n index comparison matrix; comparing n index elements pairwise to form an n-order comparison matrix, i.e. P ═ (P)ij)n×n;
Relative comparisons between two factors are made here using the 1-9 and reciprocal scale method, and the scale rules are as follows:
pij1, the element i and the element j have the same importance on the previous level factor;
pijelement i is slightly more important than element j, 3;
pijelement i is more important than element j, 5;
pijelement i is much more important than element j, 7;
pijelement i is more important than element j, 9;
pij2k, k 1,2,3,4, the importance of the elements i and j being between pij2k-1 with pij2k + 1;
otherwise, then there are
Step 3.32, solving the characteristic vector and index weight of the matrix: the columns of the matrix are summed and normalized for each column, as follows:
a new matrix B ═ B is obtainedij)n×n
The sum of each row of the matrix is 1;
summing every row of the matrix B to obtain a characteristic vector X
Normalizing the characteristic vector X
The feature vector X can be obtained, i.e.
X=[x1,x2,...,xn]
Step 3.33, consistency check of the contrast matrix: for verifying the correctness and rationality of the index weight distribution, the consistency of the comparison matrix is checked, and the maximum characteristic root is calculated to obtain
Wherein (PX)iRepresenting the ith element in a column vector obtained by multiplying the matrix P by the characteristic vector X, wherein i is 1, 2. The formula represents that each element in the column vector is divided by the product of the order and the corresponding weight respectively, and then summation is carried out;
calculate the consistency Index (Constant Index) with the formula
Wherein n is the matrix order;
calculating a random consistency ratio expressed by
Wherein, RI is an average random consistency index and can be obtained by table look-up according to the matrix order;
when CR is less than 0.10, the contrast matrix is kept consistent, namely the weight distribution is reasonable; otherwise, the value of the elements of the contrast matrix needs to be readjusted, and the weight is redistributed;
step 3.34, obtaining an evaluation scheme: determining the selected scheme AiAnd the ideal scheme A0With weights of the same matrix R, i.e.
R=TXT
Element R in Rk(k 1, 2.. m) is the sum of the same degrees of the weight of the k-th evaluated scheme and the ideal scheme, namely
According to R in the weighted identity matrix RkThe reliability order of the communication power grid line planning scheme can be obtained according to the size of the data, and the maximum value is the most reliable value.
In the above method for optimizing a communication network based on an analytic hierarchy process and a genetic algorithm, in step 4, the new genetic code is defined based on the following:
for each service biFirstly, the first n shortest paths are calculated by Dijkstra algorithm and are marked as Ai=[ai1,ai2,…,ain],aijRepresentation service biJ (th) total routing mode, and storing each path aijPassing device E (a)ij) And an optical fiberF(aij) (ii) a In practice, the size of n can be selected according to the size of a network and actual requirements, if n is larger, the problem search space is larger, and if n is smaller, the optimization problem can be converged more quickly;
the routing scheme of the whole network is coded intoWherein c isi∈[1,2,3,…,n]If c is aiIf k, it means that the ith service selects its kth routing mode aik。
In the above method for optimizing a communication network based on an analytic hierarchy process and a genetic algorithm, in step 4, a genetic variation mode is defined based on the following:
the genetic variation of the novel genetic coding scheme can be varied from k to any arbitrary j (k ≠ j; k, j ∈ [1,2,3, …, n)]) Because the overall optimal solution is more likely to be composed of the individual optimal solutions, the defined variation probability is proportional to the inherent reliability of each route, and α is the inherent reliability of the jth route of the ith serviceijIs calculated in a manner that
Thus, the traffic path is from aijVariation is aikThe probability of (k ≠ j) is
In the above method for optimizing a communication network based on an analytic hierarchy process and a genetic algorithm, step 5 specifically includes the following substeps:
step 5.1, initializing a group: defining the overall optimal solution to be more formed by the individual optimal solution, and determining the individual optimal solution through the inherent reliability of each routing mode of the service; therefore, uniform sampling from a polynomial distribution proportional to the inherent reliability of the routing mode can often achieve a better population than sampling from a uniform distribution; for the ith code ciSampling clothesForm of the distribution from a plurality of terms as follows
If the population is N in total, the usual practice is to take 2N individuals from the multi-term distribution and then select the top N better individuals from it; better individuals are generated by calculating and comparing the reliability of each scheme in step 3.34; 2N x N is requiredbThe value is taken by secondary random, the length of bit string is the total number of services nbSince the individual genes are complex in the present invention, the population size N should be selected to be at least larger than the coding length N in order to ensure the diversity of the population genesb;
Step 5.2, carrying out population evolution based on the population initialized in the step 5.1: generating a new group through elite selection, rotating table betting selection, crossing and mutation;
step 5.21, selecting elite: selecting a part of the population with higher fitness in the previous generation population to directly enter the next generation; by the method, better individuals can be reserved, so that the condition of backing up cannot occur in the evolution process; usually, the 20% individuals with the highest fitness are selected as elite to enter the next generation;
and 5.22, selecting the rotary table game: calculating the individual selection probability of the remaining individuals after Elite selection according to the fitness of each individual(fiRepresenting the fitness of each individual) and then randomly extracting part of the individuals from the individual according to the selection probability; in the method, individuals with high fitness are high in probability of being selected, and the superiority of the group is guaranteed; meanwhile, individuals with low fitness are also possibly selected, so that the diversity of group genes is ensured; carousel selection also typically selects 20% of the population; selecting Elite and rotating disc as parents of the next generation, entering a mating pool, and generating the next generation through crossing and variation change in the mating pool;
step 5.23, crossing: crossover refers to the inheritance of two individualsThe process of mass exchange to generate new individuals; if the genetic genes of the parents are good, the offspring generated after combination should also have good genes; the practical meaning under the present problem is that combining part of the traffic paths of one excellent scheme with part of the traffic paths of another excellent scheme may result in a more excellent scheme; the specific method is to randomly select two individuals from a mating pool as parents and then randomly generate [1, n ]b]One integer of (a) is taken as a crossed position, and two individuals are generated to enter the next generation;
step 5.24, mutation: variant refers to the mutation of one or more genes of an individual; mutation is an important operation for keeping population diversity and preventing the population from maturing prematurely; the practical meaning of the problem is that an excellent scheme is finely adjusted, and one or more service paths of the excellent scheme are changed to see whether a more excellent scheme can be generated or not; the specific method is to randomly select an individual from a mating pool and then randomly generate [1, n ]b]Is taken as a mutation position, and then according to the mutation method in 4-2, according to the probabilityMutation of gene j at this position to k; the variant individuals enter the next generation;
step 5.3, judging whether the algorithm finds a local optimal solution or not by the fact that the population elite is not changed after multiple iterations; and judging the correctness of the algorithm by comparing the optimal solution found by the algorithm with the score of the actual network routing scheme distributed through expert experience in practice.
Therefore, the invention has the following advantages: 1. the invention provides 3 evaluation indexes of the total network risk degree, the resource risk balance degree and the business risk balance degree, can comprehensively reflect the comprehensive characteristics of the reliability of the power communication network, and is beneficial to disclosing the problems existing in the communication network. 2. The invention provides a comparison and selection method of a communication network reliability scheme based on a hierarchical analysis method, which organically combines a qualitative method and a quantitative method, hierarchically classifies a complex decision making system, provides quantitative basis for analysis and decision making by comparing the importance of various associated factors layer by layer, and simultaneously inspects and reduces subjective influence to a certain extent, so that evaluation is more scientific, and basis is provided for selecting the optimal decision making. 3. The invention provides an improved genetic algorithm aiming at the optimization of a network routing scheme, and the invention is combined with the reality to improve and innovate from a coding mode and a variation mode, thereby solving the problems of insufficient binary coding mode of the traditional genetic algorithm and the like.
Drawings
FIG. 1 is a schematic diagram of the steps of the proposed scheme.
FIG. 2 is a schematic diagram of the evaluation index architecture of the present invention.
Detailed Description
Step 1: calculating a traffic weight
The importance degree of the service is mainly determined by the service type and the service scheduling grade according to related investigation and expert experience, and the ratio of the contribution of the service type and the service scheduling grade to the importance degree of the service is 0.7 and 0.3 respectively.
Service types { distribution automation, scheduling automation, relay protection, network management service, full-automatic equipment, scheduling telephone, administrative telephone, integrated data network, scheduling data, video teleconference, protection PCM, communication PCM, video service, frequency control service } and service scheduling grades { primary network, secondary network, tertiary network, quaternary network and quinary network } are respectively given a score, the score interval is [1,10], and the score is given by relevant experts.
w=0.7×wT+0.3×wD
Wherein, wTRating the traffic type, wDRating the traffic scheduling level, wT∈[1,10],wD∈[1,10]. The score of each service, i.e. the weight w, can be calculated according to the service type and the scheduling level of each service.
Step 2: calculating fiber and equipment failure rates
2.1 failure Rate of the device
The observation and research of the existing data show that when the number n of the services borne by the equipment is less than a certain value, the failure rate of the equipment is a stable value; when the number n of the services is increased, the relation between the equipment use failure rate and the number of the services is in a nonlinear relation.
The nonlinear regression model can be generally expressed as
yi=f(xi,θ)+εi,i=1,2,...,n
Wherein, yiIs a dependent variable; non-random vector xi=(xi1,xi2,...,xik) ' is an independent variable; theta ═ theta0,θ1,...,θp) ' is an unknown parameter vector; epsiloniAre random error terms and satisfy independent identically distributed assumptions.
For non-linear regression models, the parameter θ is estimated using the least squares method, i.e. the method is used to
To a minimumCalled θ a non-linear least squares estimate. When the f function is assumed to be continuously differentiable for the parameter theta, a normal equation set can be established by using a differentiation method to minimize Q (theta)The Q function is applied to the parameter thetajCalculating the partial derivatives and making them be 0 to obtain p +1 equations
j-0, 1, 2.. p, non-linear least squares estimationIs a solution of the above formula.
In the present invention, the device utilization failure rate η (n) function is defined as a piecewise function, i.e.
Where n is the number of services, c is a constant value, α is a parameter, which is equivalent to that in the nonlinear model
2.2 fiber failure Rate
In the invention, the failure rate of the optical fiber is not changed along with the number of the services carried by the optical fiber, and is only related to the inherent property of the optical fiber. According to the invention, the failure rate of each optical fiber can be obtained according to the ratio of the historical failure days to the working days, and then the average failure rate of each type of optical fiber is calculated according to the difference of the types of the optical fibers, namely the failure rate of the optical fiber.
And step 3: communication network reliability evaluation standard
3-1 network reliability index
The reliability of the communication Network is evaluated by three indexes, namely, Network Risk Degrid (NRD), Resource Risk BalanceDegrid (RRBD) and Business Risk BalanceDegrid (BRBD). To calculate these three indexes, the concept of Equipment Risk (ERD), Fiber Risk (FRD), and Business Risk (BRD) is introduced first.
The invention is characterized in that e represents the total equipment (equipment) of the communication network, f represents the fiber (fiber) and b represents the business (business). n ise,nf,nbRepresenting the total number of devices, fibers and traffic in the network, respectively. In a network, one device or fiber typically carries multiple services, as well as one service carried over multiple devices and fibers. Device eiB (e) for carried service seti) Denotes, | B (e)i) I represents the number of services in the set, fiber fiB (f) for carried service seti) And (4) showing. Likewise, bearer service biE (b) for device integrationi) F (b) for showing, optical fiber collectioni) And (4) showing. Service biW for the weight valueiAnd (4) showing. Analyzing failure rate of surface and equipmentIn addition to being related to the intrinsic properties of the device, it is also related to the amount of traffic carried by the device, and the failure rate of the fiber is related only to the intrinsic properties of the fiber, hence η (e) respectivelyi,|B(ei) I) and η (f)i) The calculation methods for indicating the failure rates of the equipment and the optical fiber, w and η, are given in step 1 and step 2.
Device eiDegree of risk (ERD)iIndicating the likelihood of the device failing and the impact on the overall network after failure. The calculation formula is
Likewise, the optical fiber fiDegree of risk FRDiIs calculated by the formula
Service biDegree of risk BRDiIndicating the likelihood of the service failing and the impact on the network. And the probability of the service failure is the probability of any resource bearing the service failure. The calculation formula is
The total risk (NRD) of the network is the sum of the risks of all resources in the network and is calculated by the formula
The Resource Risk Balance (RRBD) is the reciprocal of the variance sum of the risk of equipment and optical fibers in the network and has the calculation formula of
Also, the Business Risk Balance (BRBD) is calculated as
The optimization of the communication network is to reduce the overall risk (NRD) of the network and simultaneously improve the Resource Risk Balance (RRBD) and Business Risk Balance (BRBD) of the network.
3-2 evaluation method for reliability of communication network
3.2.1 construction of evaluation index System
For the reliability of the communication network route scheme, three evaluation index systems, namely a scheme layer, a factor layer and an index layer, are constructed, as shown in fig. 2.
3.2.2 construction of identity decision matrices
(1) It is assumed that the route of the communication network is planned to have N1,N2,...,NmM schemes, each scheme having p1,p2,...,pnObtaining a multi-scheme evaluation decision matrix S by each index, namely
Selecting the optimal value of each category index in m schemes to form an ideal scheme A0,
A0=[p01,p02,...,p0n]T
The total network risk degree is the minimum value, the resource risk balance degree is the maximum value, and the service risk balance degree is the maximum value. Note A0The f (f) index has a value of p0f。
(2) Constructing a same degree matrix of the evaluated scheme index and the ideal scheme index
From the evaluated scheme AiEach index p in (i ═ 1, 2.., m)ijAnd the ideal scheme A0Middle correspondence index value p0fOf a same degree aijThe identity matrix T can be obtained as
Wherein, for the index of the maximum value of the corresponding index of each scheme in the ideal scheme, the index value in each scheme is divided by the corresponding index value in the ideal scheme, namely aij=pij/pof(ii) a For the index of the minimum value of the corresponding index of each scheme in the ideal scheme, dividing the index value in the ideal scheme by the corresponding index value in each scheme, namely aij=pof/pij。
3.2.3 calculating evaluation index weight
The evaluation index weight is calculated by adopting an analytic hierarchy process, and the specific calculation steps are as follows.
(1) And constructing an n x n index comparison matrix. Comparing n index elements pairwise to form an n-order comparison matrix, i.e. P ═ (P)ij)n×n。
Relative comparisons between two factors are made here using the 1-9 and reciprocal scale method, and the scale rules are as follows:
pij1, the element i and the element j have the same importance on the previous level factor;
pijelement i is slightly more important than element j, 3;
pijelement i is more important than element j, 5;
pijelement i is much more important than element j, 7;
pijelement i is more important than element j, 9;
pij2k, k 1,2,3,4, the importance of the elements i and j being between pij2k-1 with pij2k + 1;
otherwise, then there are
(2) Calculating the eigenvectors and index weights of the matrix
The columns of the matrix are summed and normalized for each column, as follows:
a new matrix B ═ B can be obtainedij)n×n
The sum of each column of the matrix is 1.
Summing every row of the matrix B to obtain a characteristic vector X
Normalizing the characteristic vector X
The feature vector X can be obtained, i.e.
X=[x1,x2,...,xn]
(3) Consistency check of contrast matrix
To verify the correctness and rationality of the index weight assignment, the consistency of the comparison matrix needs to be checked.
Computing the maximum feature root, one can obtain
Wherein (PX)iThe i-th element in the column vector obtained by multiplying the matrix P by the feature vector X is represented, i being 1, 2. The formula represents that each element in the column vector is divided by the product of the order and the corresponding weight, and then summed.
Calculate the consistency Index (Constant Index) with the formula
Wherein n is the matrix order.
Calculating a random consistency ratio expressed by
Wherein, RI is an average random consistency index, and can be obtained by looking up a table according to the matrix order.
When CR is less than 0.10, the contrast matrix is kept consistent, namely the weight distribution is reasonable; otherwise, the values of the elements of the contrast matrix need to be readjusted, and the weights are redistributed.
3.2.4 evaluation protocol
Determining the selected scheme AiAnd the ideal scheme A0With weights of the same matrix R, i.e.
R=TXT
Element R in Rk(k 1, 2.. m) is the sum of the same degrees of the weight of the k-th evaluated scheme and the ideal scheme, namely
According to R in the weighted identity matrix RkThe reliability order of the communication power grid line planning scheme can be obtained according to the size of the data, and the maximum value is the most reliable value.
And 4, step 4: genetic algorithm improvement for optimization problem
4-1 genetic algorithm coding mode
Conventional genetic algorithms use binary coding to represent a solution in the problem space. For the optimization problem involved in the present invention, a solution in the problem space needs to represent the routing way of all traffic in the network. The conventional coding scheme is difficult to represent, so a new coding scheme needs to be proposed.
For each service biFirstly, the first n shortest paths are calculated by Dijkstra algorithm and are marked as Ai=[ai1,ai2,…,ain],aijRepresentation service biJ (th) total routing mode, and storing each path aijPassing device E (a)ij) And an optical fiber F (a)ij). In practice, the size of n can be selected according to the size of the network and actual requirements, if n is larger, the problem search space is larger, and if n is smaller, the optimization problem can be converged faster.
The routing scheme of the whole network is coded intoWherein c isi∈[1,2,3,…,n]If c is aiIf k, it means that the ith service selects its kth routing mode aik。
4-2 Gene mutation Pattern
The genetic algorithm traditionally has only mutations from 0 to 1 or 1 to 0, whereas the genetic variation used in the present context for the coding mode can vary from k to any j (k ≠ j; k, j ∈ [1,2,3, …, n)]) Because the overall optimal solution is more likely to consist of the individual optimal solutions, the probability of variability can be considered to be proportional to the intrinsic reliability of each route, the intrinsic reliability of the jth route for the ith traffic αijIs calculated in a manner that
Thus, the traffic path is from aijVariation is aikThe probability of (k ≠ j) is
And 5: communication network optimization using improved genetic algorithms
5-1 initialization population
In general genetic algorithms, individuals in the initial population are usually randomly generated without prior knowledge about the problem solution space. However, in the present problem, it can be considered that the overall optimal solution is more likely to be composed of individual optimal solutions, and the individual optimal solutions can be determined approximately by the inherent reliability of each routing manner of the traffic. A better approach is therefore to sample uniformly from a polynomial distribution that is proportional to the inherent reliability of the routing, often yielding a better population than from a uniform distribution. For the ith code ciThe sampling obeys a polynomial distribution in the form of
If the population is N in total, it is common practice to take 2N individuals from the multi-term distribution and then select the top N better individuals from the distribution. The manner of selection is given in step 3. This method requires 2N x N to be performedbAnd (4) performing secondary random value taking.
The large population size N provides sufficient pattern sampling capacity for genetic algorithms, which can improve the search quality of the algorithms and prevent the algorithms from converging before maturity. However, large clusters increase the computational complexity of the fitness evaluation, making convergence too slow. In order to guarantee the expression ability, usually choose the group size one order of magnitude more than the length of the coding bit string, the bit string length in this question is the total number of services nbTherefore the selected population size should be slightly larger than nb。
5-2 population evolution
And the group evolution is to generate a new group through elite selection, rotating table betting selection, crossing and mutation on the basis of the previous generation. This step is cycled through a number of times.
Selecting elite: and selecting a part of the population with higher fitness in the previous generation population to directly enter the next generation. By the method, better individuals can be reserved, and the situation of backspace can not occur in the evolution process. The 20% individuals with the highest fitness are usually selected as elite for the next generation.
Rotary discAnd (3) betting selection: calculating the individual selection probability of the remaining individuals after Elite selection according to the fitness of each individual(fiRepresenting the fitness of each individual) and then randomly extracting a portion of the individuals from the selection probabilities. In the method, individuals with high fitness are high in probability of being selected, and the superiority of the group is guaranteed; and meanwhile, individuals with low fitness are also possibly selected, so that the diversity of group genes is ensured. Carousel selection also typically selects 20% of the total population. The individuals selected by elite selection and rotary table gambling are used as parents of the next generation, enter a mating pool, and the next generation is generated in the mating pool through crossing and variation.
And (3) crossing: crossover refers to the process by which two individuals exchange genetic material to create a new individual. If the genetic genes of the parents are superior, the offspring generated after combination should also possess the superior genes. The practical significance of this problem is that combining part of the traffic path of one excellent scheme with part of the traffic path of another excellent scheme may result in a more excellent scheme. The specific method is to randomly select two individuals from a mating pool as parents and then randomly generate [1, n ]b]As a position of the intersection, resulting in two individuals going into the next generation.
Mutation: mutation refers to the mutation of one or more genes in an individual. Mutation is an important operation for maintaining population diversity and preventing premature population maturation. The practical significance of this problem is to fine-tune an excellent solution, change the path of one or several services, and see if a more excellent solution can be generated. The specific method is to randomly select an individual from a mating pool and then randomly generate [1, n ]b]Is taken as a mutation position, and then according to the mutation method in 4-2, according to the probabilityThe gene j at this position was mutated to k. The mutated individuals enter the next generation.
5-3 Algorithm termination
The genetic algorithm in the present problem cannot judge whether the evolution converges through the objective function. After multiple iterations, the population elite is not changed to judge whether the algorithm finds the local optimal solution. And judging the correctness of the algorithm by comparing the optimal solution found by the algorithm with the score of the actual network routing scheme distributed through expert experience in practice.
Claims (5)
1. A communication network optimization method based on genetic algorithm is characterized by comprising the following steps:
step 1, calculating a service weight w for each service from the service type and the service scheduling grade; the service types of the power communication network mainly include distribution automation, dispatching automation, relay protection, network management service, full-automatic devices, dispatching telephones and frequency control service; the service scheduling level can be divided into a first-level network, a second-level network, a third-level network, a fourth-level network and a fifth-level network; the service weight w is the service type weight wTAnd scheduling rank weight wDA weighted sum of (a);
step 2, analyzing the failure rate of the computing equipment and the optical fiber; through data research, the relation function of the equipment failure rate and the number of services borne by the equipment is a piecewise function, when the number of the services is smaller than a certain value, the failure rate is a stable value, and when the number of the services exceeds the certain value, the failure rate and the number of the services borne by the equipment are in a nonlinear relation; the failure rate of the optical fiber is not changed along with the number of the services, is only related to the inherent attributes of the optical fiber and can be obtained by historical data;
step 3, calculating a standard for evaluating the reliability of the communication network, evaluating the reliability of the communication network by using three indexes of network total risk, resource risk balance and business risk balance, acquiring the network total risk, the resource risk balance and the business risk balance based on the equipment risk, the optical fiber risk and the business risk, and evaluating the reliability of the communication network based on the following steps:
step 3.1, construction of an evaluation index system: aiming at the reliability of the communication network route scheme, three layers of evaluation index systems are constructed, namely a scheme layer, a factor layer and an index layer,
step 3.2, constructing a same degree decision matrix, which specifically comprises the following steps:
(1) defining a communication network route plan with N1,N2,...,NmM schemes, each scheme having p1,p2,...,pnObtaining a multi-scheme evaluation decision matrix S by each index, namely
Selecting the optimal value of each category index in m schemes to form an ideal scheme A0,
A0=[p01,p02,...,p0n]T
The total network risk degree takes the minimum value, the resource risk balance degree takes the maximum value, and the service risk balance degree takes the maximum value; note A0The f-th index of (1) is p0f,f=1,2,...,n;
(2) Constructing a similarity matrix of the evaluated scheme index and the ideal scheme index: from the evaluated scheme AiMiddle index pijAnd the ideal scheme A0Middle correspondence index value p0fOf a same degree aijAn identity matrix T may be obtained, where i 1,2
Wherein, for the index of the maximum value of the corresponding index of each scheme in the ideal scheme, the index value in each scheme is divided by the corresponding index value in the ideal scheme, namely aij=pij/pof(ii) a For the index of the minimum value of the corresponding index of each scheme in the ideal scheme, dividing the index value in the ideal scheme by the corresponding index value in each scheme, namely aij=pof/pij;
Step 3.3, calculating the weight of the evaluation index: calculating the weight of the evaluation index by adopting an analytic hierarchy process, wherein the specific calculation steps are as follows;
3.31, constructing an n multiplied by n index comparison matrix; comparing n index elements pairwise to form an n-order comparison matrix, i.e. P ═ (P)ij)n×n;
Relative comparisons between two factors are made here using the 1-9 and reciprocal scale method, and the scale rules are as follows:
pij1, the element i and the element j have the same importance on the previous level factor;
pijelement i is slightly more important than element j, 3;
pijelement i is more important than element j, 5;
pijelement i is much more important than element j, 7;
pijelement i is more important than element j, 9;
pij2k, k 1,2,3,4, the importance of the elements i and j being between pij2k-1 with pij2k + 1;
otherwise, then there are
Step 3.32, solving the characteristic vector and index weight of the matrix: the columns of the matrix are summed and normalized for each column, as follows:
a new matrix B ═ B is obtainedij)n×n
The sum of each row of the matrix is 1;
summing every row of the matrix B to obtain a characteristic vector X
Normalizing the characteristic vector X
The feature vector X can be obtained, i.e.
X=[x1,x2,...,xn]
Step 3.33, consistency check of the contrast matrix: for verifying the correctness and rationality of the index weight distribution, the consistency of the comparison matrix is checked, and the maximum characteristic root is calculated to obtain
Wherein (PX)iRepresenting the ith element in a column vector obtained by multiplying the matrix P by the characteristic vector X, wherein i is 1, 2. The formula represents that each element in the column vector is divided by the product of the order and the corresponding weight respectively, and then summation is carried out;
calculating a consistency index, which is expressed by
Wherein n is the matrix order;
calculating a random consistency ratio expressed by
Wherein, RI is an average random consistency index and can be obtained by table look-up according to the matrix order;
when CR is less than 0.10, the contrast matrix is kept consistent, namely the weight distribution is reasonable; otherwise, the value of the elements of the contrast matrix needs to be readjusted, and the weight is redistributed;
step 3.34, obtaining an evaluation scheme: determining the selected scheme AiAnd the ideal scheme A0With weights of the same matrix R, i.e.
R=TXT
Element R in RkI.e. the sum of the identity of the k-th evaluated solution and the ideal solution, k being 1,2
According to R in the weighted identity matrix RkThe reliability order of the communication power grid line planning scheme can be obtained according to the size of the data, and the maximum value is the most reliable value;
step 4, optimizing a network routing scheme by adopting a coding mode and a variation mode based on an improved genetic algorithm; wherein the content of the first and second substances,
genetic code is based on the following definitions:
for each service biFirstly, the first n shortest paths are calculated by Dijkstra algorithm and are marked as Ai=[ai1,ai2,…,ain],aijRepresentation service biJ (th) total routing mode, and storing each path aijPassing device E (a)ij) And an optical fiber F (a)ij) (ii) a In practice, the size of n can be selected according to the size of a network and actual requirements, if n is larger, the problem search space is larger, and if n is smaller, the optimization problem can be converged more quickly;
the routing scheme of the whole network is coded intoWherein c isi∈[1,2,3,…,n]If c is aiIf k, it means that the ith service selects its kth routing mode aik;
The genetic variation pattern is defined based on the following:
genetic variation of genetic coding patterns can vary from k to arbitrary j, k ≠ j; k, j ∈ [1,2,3, …, n ]]Because the overall optimal solution is more likely to be composed of the individual optimal solutions, the defined variation probability is proportional to the inherent reliability of each route, and α is the inherent reliability of the jth route of the ith serviceijIs calculated in a manner that
Thus, the traffic path is from aijVariation is aikThe probability of (k ≠ j) is
And 5, optimizing the communication network according to the standard for evaluating the network reliability in the step 3 by using the improved genetic algorithm in the step 4.
2. The communication network optimization method based on genetic algorithm as claimed in claim 1, wherein the specific operation method in step 1 is: respectively scoring service types { distribution automation, scheduling automation, relay protection, network management service, full-automatic devices, scheduling telephone, administrative telephone, integrated data network, scheduling data, video teleconference, protection PCM, communication PCM, video service, frequency control service } and service scheduling grades { primary network, secondary network, tertiary network, quaternary network and quinary network }, wherein the scoring interval is [1,10], and the score is given by relevant experts;
w=0.7×wT+0.3×wD
wherein, wTRating the traffic type, wDRating the traffic scheduling level, wT∈[1,10],wD∈[1,10](ii) a The score of each service, i.e. the weight w, can be calculated according to the service type and the scheduling level of each service.
3. The method for optimizing a communication network based on genetic algorithm as claimed in claim 1, wherein said step 2 comprises the following sub-steps:
step 2.1, obtaining the failure rate of the equipment: the observation and research of the existing data show that when the number n of the services borne by the equipment is less than a certain value, the failure rate of the equipment is a stable value; when the number n of the services is increased again, the relation between the equipment use failure rate and the number of the services is in a nonlinear relation; the nonlinear regression model is expressed as
yi=f(xi,θ)+εi,i=1,2,...,n
Wherein, yiIs a dependent variable; non-random vector xi=(xi1,xi2,...,xik) ' is an independent variable; theta ═ theta0,θ1,...,θp) ' is an unknown parameter vector; epsiloniAre random error terms and satisfy independent identically distributed assumptions;
for non-linear regression models, the parameter θ is estimated using the least squares method, i.e. the method is used to
To a minimumCalled theta as nonlinear least square estimation; when the f function is assumed to be continuously differentiable for the parameter theta, a normal equation set can be established by using a differentiation method to minimize Q (theta)The Q function is applied to the parameter thetajCalculating the partial derivatives and making them be 0 to obtain p +1 equations
defining the device utilization failure rate η (n) function as a piecewise function, i.e.
Where n is the number of services, c is a constant value, α is a parameter, α is equivalent to that in the nonlinear model
Step 2.2, obtaining the fiber fault rate: defining that the failure rate of the optical fiber does not change with the number of the services carried by the optical fiber and is only related to the inherent attributes of the optical fiber; the method comprises the steps of firstly obtaining the failure rate of each optical fiber according to the ratio of the historical failure days to the working days, and then calculating the average failure rate of each type of optical fiber according to the difference of the types of the optical fibers, namely the failure rate of the optical fiber.
4. The method for optimizing a communication network based on genetic algorithm as claimed in claim 1, wherein in step 3, the network reliability index is obtained based on the following definition: definition e denotes the total equipment (equipment) of the communication network, f denotes the fiber (fiber), b denotes the traffic (business); n ise,nf,nbRespectively representing the total number of equipment, optical fibers and services in the network; in a network, one device or optical fiber usually carries a plurality of services, and one service is also carried on a plurality of devices and optical fibers; device eiB (e) for carried service seti) Denotes, | B (e)i) I represents the number of services in the set, fiber fiB (f) for carried service seti) Represents; likewise, bearer service biE (b) for device integrationi) F (b) for showing, optical fiber collectioni) Represents; service biW for the weight valueiRepresents; the failure rate of a device is related to the amount of traffic carried by the device, in addition to the intrinsic properties of the device, while the lightThe failure rate of the fiber is only related to the inherent properties of the fiber, and therefore η (e) respectivelyi,|B(ei) I) and η (f)i) Indicating the failure rate of the equipment and the optical fiber;
device eiDegree of risk (ERD)iIndicating the possibility of failure of the device and the effect on the whole network after failure; the calculation formula is
Likewise, the optical fiber fiDegree of risk FRDiIs calculated by the formula
Service biDegree of risk BRDiIndicating the possibility of failure of the service and the impact on the network; the probability of the service failure is the probability of any resource bearing the service failure; the calculation formula is
The total risk (NRD) of the network is the sum of the risks of all resources in the network and is calculated by the formula
The Resource Risk Balance (RRBD) is the reciprocal of the variance sum of the risk of equipment and optical fibers in the network and has the calculation formula of
Also, the Business Risk Balance (BRBD) is calculated as
Optimization of a communication network requires a reduction in the overall risk (NRD) of the network, while improving the Resource Risk Balance (RRBD) and Business Risk Balance (BRBD) of the network.
5. The method for optimizing a communication network based on genetic algorithm as claimed in claim 1, wherein the step 5 comprises the following sub-steps:
step 5.1, initializing a group: defining the overall optimal solution to be more formed by the individual optimal solution, and determining the individual optimal solution through the inherent reliability of each routing mode of the service; therefore, uniform sampling from a polynomial distribution proportional to the inherent reliability of the routing mode can often achieve a better population than sampling from a uniform distribution; for the ith code ciThe sampling obeys a polynomial distribution in the form of
If the population is N in total, the usual practice is to take 2N individuals from the multi-term distribution and then select the top N better individuals from it; better individuals are generated by calculating and comparing the reliability of each scheme in step 3.34; 2N x N is requiredbThe value is taken by secondary random, the length of bit string is the total number of services nbSince the individual genes are complex, the population size N should be selected to be at least larger than the coding length N to ensure the diversity of the population genesb;
Step 5.2, carrying out population evolution based on the population initialized in the step 5.1: generating a new group through elite selection, rotating table betting selection, crossing and mutation;
step 5.21, selecting elite: selecting a part of the population with higher fitness in the previous generation population to directly enter the next generation; by the method, better individuals can be reserved, so that the condition of backing up cannot occur in the evolution process; usually, the 20% individuals with the highest fitness are selected as elite to enter the next generation;
and 5.22, selecting the rotary table game: calculating the individual selection probability of the remaining individuals after Elite selection according to the fitness of each individualfiRepresenting the fitness of each individual, and then randomly extracting part of individuals from the fitness according to the selection probability; in the method, individuals with high fitness are high in probability of being selected, and the superiority of the group is guaranteed; meanwhile, individuals with low fitness are also possibly selected, so that the diversity of group genes is ensured; carousel selection also typically selects 20% of the population; selecting Elite and rotating disc as parents of the next generation, entering a mating pool, and generating the next generation through crossing and variation change in the mating pool;
step 5.23, crossing: crossover refers to the process by which two individuals exchange genetic material to create a new individual; if the genetic genes of the parents are good, the offspring generated after combination should also have good genes; the practical meaning under the present problem is that combining part of the traffic paths of one excellent scheme with part of the traffic paths of another excellent scheme may result in a more excellent scheme; the specific method is to randomly select two individuals from a mating pool as parents and then randomly generate [1, n ]b]One integer of (a) is taken as a crossed position, and two individuals are generated to enter the next generation;
step 5.24, mutation: variant refers to the mutation of one or more genes of an individual; mutation is an important operation for keeping population diversity and preventing the population from maturing prematurely; the practical meaning of the problem is that an excellent scheme is finely adjusted, and one or more service paths of the excellent scheme are changed to see whether a more excellent scheme can be generated or not; the specific method is to randomly select an individual from a mating pool and then randomly generate [1, n ]b]Is used as a mutation position, and then according to the mutation method of step 4, according to the probabilityMutation of gene j at this position to k; the variant individuals enter the next generation;
step 5.3, judging whether the algorithm finds a local optimal solution or not by the fact that the population elite is not changed after multiple iterations; and judging the correctness of the algorithm by comparing the optimal solution found by the algorithm with the score of the actual network routing scheme distributed through expert experience in practice.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710428743.2A CN107294775B (en) | 2017-06-08 | 2017-06-08 | Communication network optimization method based on analytic hierarchy process and genetic algorithm |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710428743.2A CN107294775B (en) | 2017-06-08 | 2017-06-08 | Communication network optimization method based on analytic hierarchy process and genetic algorithm |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107294775A CN107294775A (en) | 2017-10-24 |
CN107294775B true CN107294775B (en) | 2020-02-14 |
Family
ID=60096118
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710428743.2A Expired - Fee Related CN107294775B (en) | 2017-06-08 | 2017-06-08 | Communication network optimization method based on analytic hierarchy process and genetic algorithm |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107294775B (en) |
Families Citing this family (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108400935B (en) * | 2018-02-11 | 2021-02-23 | 国家电网公司信息通信分公司 | Genetic algorithm-based service path selection method and device and electronic equipment |
CN109274539B (en) * | 2018-11-14 | 2019-08-13 | 中国人民解放军陆军工程大学 | Network security based on genetic algorithm configures automatic creation system |
CN111324504B (en) * | 2018-12-14 | 2024-02-27 | 国网山西省电力公司信息通信分公司 | Method and device for monitoring running state of power communication network |
CN110197305B (en) * | 2019-05-31 | 2023-11-17 | 国家电网有限公司 | Relay protection data model searching and optimizing method and system based on shortest path algorithm |
CN110796297B (en) * | 2019-10-21 | 2020-11-10 | 浙江大学 | Electric power system structure optimization method based on balance degree variance and reliability |
CN111245722B (en) * | 2020-01-20 | 2022-03-11 | 重庆邮电大学 | SDN data center network flow forwarding method based on genetic algorithm |
CN113315807A (en) * | 2021-04-16 | 2021-08-27 | 山东英信计算机技术有限公司 | Method, system, device and medium for optimizing cluster decision |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101562568A (en) * | 2009-05-26 | 2009-10-21 | 中国科学院计算技术研究所 | Method and device for generating alternate routes of coverage network |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20130223612A1 (en) * | 2012-02-29 | 2013-08-29 | Avaya Inc. | Dynamic adjustment of multi-dimensional routing rule |
-
2017
- 2017-06-08 CN CN201710428743.2A patent/CN107294775B/en not_active Expired - Fee Related
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101562568A (en) * | 2009-05-26 | 2009-10-21 | 中国科学院计算技术研究所 | Method and device for generating alternate routes of coverage network |
Non-Patent Citations (3)
Title |
---|
基于风险均衡的电力通信业务的路由分配机制;曾庆涛等;《电子与信息学报》;20130630;第35卷(第6期);第1319-1320页 * |
电力通信网风险分析与控制研究;樊冰;《中国博士学位论文全文数据库(电子期刊),工程科技II辑》;20160115;第2016年卷(第01期);摘要、第2.2节 * |
电网设备状态检修策略的研究;郇嘉嘉;《中国博士学位论文全文数据库(电子期刊),工程科技II辑》;20121115;第2012年卷(第11期);第1.4、3.2节 * |
Also Published As
Publication number | Publication date |
---|---|
CN107294775A (en) | 2017-10-24 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107294775B (en) | Communication network optimization method based on analytic hierarchy process and genetic algorithm | |
CN110971525B (en) | Service routing and addressing method for service operation of power communication network | |
EP2742648B1 (en) | Network capacity management system and method | |
CN107453928B (en) | A kind of power communication network node different degree evaluation method and device | |
CN111106999A (en) | IP-optical network communication service joint distribution method and device | |
CN105471647B (en) | A kind of power communication network fault positioning method | |
Borhani et al. | A multicriteria optimization for flight route networks in large-scale airlines using intelligent spatial information | |
CN113705085A (en) | Modeling and risk assessment method for multi-level structure of smart power grid | |
CN109523101A (en) | A kind of distribution Running State fuzzy synthetic appraisement method | |
CN109961362A (en) | P2P platform credit risk dynamic evaluation method and system | |
CN105184415A (en) | Power distribution network reconstruction design method | |
CN110996365B (en) | Heterogeneous network vertical switching algorithm and system based on multi-objective optimization model | |
Huang et al. | Bilevel programming approach to optimizing a logistic distribution network with balancing requirements | |
CN113347589B (en) | Virtual network resource allocation method for intelligent power distribution service | |
Li et al. | Network topology optimization via deep reinforcement learning | |
Ghafori et al. | Best cloud provider selection using integrated ANP-DEMATEL and prioritizing SMI attributes | |
CN114186750A (en) | Routing planning method and device for hierarchical requirements of power business of smart power grid | |
CN106506229B (en) | A kind of SBS cloud application adaptive resource optimizes and revises system and method | |
CN113285832B (en) | NSGA-II-based power multi-mode network resource optimization allocation method | |
Xue et al. | Research on escaping the big-data traps in O2O service recommendation strategy | |
CN115086197A (en) | Method and system for evaluating robustness of communication network | |
CN113327047B (en) | Power marketing service channel decision method and system based on fuzzy comprehensive model | |
Goścień et al. | Efficient network traffic prediction after a node failure | |
CN111290853B (en) | Cloud data center scheduling method based on self-adaptive improved genetic algorithm | |
CN111800301A (en) | Network security evaluation method and system in machine type communication |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20200214 Termination date: 20210608 |
|
CF01 | Termination of patent right due to non-payment of annual fee |