Background
The rapid development of the intelligent substation brings urgent requirements for positioning the communication link fault of the secondary system, background alarm information caused by the fault of a certain communication link is numerous, and the field operation and maintenance efficiency is greatly reduced.
In the existing intelligent substation, the alarm aiming at the communication link fault stays at a device alarm layer, on one hand, the alarm information represents the original fault information and is displayed on a background operation and maintenance interface, on the other hand, the device alarm information brought by the link fault is numerous, a new step of analysis and processing is not carried out aiming at the alarm information, and the visualized link positioning result is lost, so that the field operation and maintenance are difficult.
The original alarm information of the device actually contains abundant fault information, and analyzing the alarm information of the device and establishing a link related positioning model are the basis of accurate fault positioning.
At present, in the aspects of positioning a communication link of a secondary system of an intelligent substation and improving the operation and maintenance work efficiency, relevant students apply an artificial intelligence technology to carry out relevant attempts, and a new idea (anybo, zheng yongkang, wang yongfu, sheng siqing, li song, zhang hai, zheng super and zheng super) is brought to fault positioning of the communication link of the secondary system of the intelligent substation.
In addition, students and related engineers realize the visualization of the secondary circuit by analyzing the SCD configuration file of the intelligent substation (Xiong Huaqiang, wangyong, gui Xiaozhi, duke, leaf flying. The design and realization of the SCD file visualization management and analysis decision system of the intelligent substation [ J ]. Power automation equipment, 2015,35 (05): 166-171.), and the students also establish a secondary virtual circuit and a secondary physical circuit visualization system (Xuyan, single 32900surpass; the intelligent substation secondary circuit visualization [ J/OL ]. Power system and its automation program based on the SCD file: 1-7 2022-09-21]. DOI:10.19635/J. Cnki.csu-epsa.001. In the aspect of background alarm information analysis, students provide transformer substation alarm information fault knowledge representation methods (Linglingyun, chenqing, jinliu, wang Lei, transformer substation alarm information fault knowledge representation research and application based on the knowledge graph [ J ]. Power system protection and control, 2022,50 (12): 90-99.) to excavate and display the behavior logic among alarm signals, improve the efficiency of background operation and maintenance, and lay a certain foundation for secondary system fault diagnosis and fault location.
The method provides a secondary system communication network numbering rule, communication link fault identification based on alarm information, node link coding rules and node expectation of link fault location are constructed, a fault link location fitness function with multi-information fusion is established, and accurate location of a secondary system communication link is achieved by adopting an adaptive genetic particle swarm algorithm.
Disclosure of Invention
The method is important for improving the field operation and maintenance efficiency of the intelligent substation, simplifying the complex and redundant alarm information judgment process and achieving the visual communication link positioning result. Based on the method, the invention provides an intelligent substation communication link fault positioning method based on multi-warning information fusion. The method establishes a secondary system node link numbering principle, carries out fault identification on a communication link through device alarm information, provides node and link coding rules, establishes node expected mapping under four local connection modes on the basis of typical connection of a secondary system, establishes a multi-information fusion link fault positioning fitness function on the basis, and solves the fitness function positioning fault link by applying an adaptive genetic particle swarm algorithm.
The technical scheme adopted by the invention is as follows: an intelligent substation communication link fault positioning method based on multi-warning information fusion is characterized by comprising the following steps:
step 1: numbering the secondary system communication networks; numbering the secondary system by adopting a node link numbering principle suitable for the whole network of the secondary system, wherein the node link numbering principle suitable for the whole network of the secondary system is a link numbering principle which follows a node numbering principle from top to bottom and follows a link numbering principle of from top to bottom, from left to right, firstly real nodes and then virtual nodes, and firstly the interval and then the interval;
step 2: confirming the state of a relevant node based on the fault identification of the communication link of the alarm information;
and 3, step 3: constructing node link coding principlesEstablishing a node state coding array N j ;
And 4, step 4: four typical local connection structures are constructed, and local node expectations of four link failures are established
Based on
Splicing to form a full network node expectation N
j (l);
And 5: n based on step 4 j (l) Construction of fitness function f it (n);
Step 6: initializing AGA-ABPSO to solve the relevant parameters of the positioning model, and converting N into N j Inputting calculation method, randomly initializing population and establishing link state coding array L i Calculating the expected node State code N j (l) Iterative solution of population optimum f it And (n) outputting the optimal particles to complete link positioning.
The intelligent substation communication link fault positioning method based on multi-warning information fusion is characterized in that virtual node numbering is introduced into nodes which are not only directly connected but also connected through a switch in the node numbering principle; sequentially numbering the spacing layers according to the sequence of protection, measurement and control and wave recording; numbering process layer equipment in sequence according to the sequence of the intelligent terminal and the merging unit; when the device has direct connection and connection through the switch, the virtual node number is adopted when node state coding is carried out on the link connected with the switch; in the principle of link numbering, when a plurality of links are connected to the same device, the links are numbered in sequence from small to large according to the lower nodes of the links.
The method for positioning the fault of the communication link of the intelligent substation based on the fusion of the multiple pieces of alarm information is characterized in that after the background receives the alarm information from the device, the node state is determined according to the following mechanism:
(1) When aiming at a direct link between devices, positioning nodes at two ends of the link through the alarm information of a background receiving device; acquiring node information by adopting a BM field matching algorithm, taking the node information in the alarm information as a key word, and positioning nodes at two ends of a link alarm as actual measurement node state information;
(2) When aiming at a link connected by a switch, adding differentiated analysis of alarm information on the basis of the directly connected link; and combining the differentiated alarm information with a BM field matching algorithm to form node state information containing both ends of the switch link.
The intelligent substation communication link fault positioning method based on multi-warning information fusion is characterized in that the node state coding array N j As shown in formula (1).
Wherein N is j Representing node state array, n j (j =1,2, \8230;, m) is the state encoding of the jth node; m is the total number of nodes including the virtual node;
the intelligent substation communication link fault positioning method based on multi-warning information fusion is characterized in that the node expecting N of the link fault j (l) The following were used:
(1) Background-device connection expectation mapping I, the function is as follows;
wherein
Indicating node expectation under the type I connection structure, and a subscript j indicating an expected state value of a jth node;
representing a loop operation, wherein a is a loop starting point, b is a loop end point, and when the loop is in a loop interval, the operation of the c expression is executed; l
i (i=1,2,L e
1 ) Coding the state of the ith link; e.g. of the type
1 Represents the total number of links; k
I Showing the connection condition of the link lower layer and the station control layer switch SW-A of the switch if the link l
j Has up>A connection to SW-A, then K
I =1, otherwise K
I =0;K
II Showing the connection condition of the upper layer of the link and the station control layer switch SW-A of the switch, if the lower layer of the link has connection with the SW-A, K is
II =1, otherwise K
II =0; "|" represents a logical or operation;
(2) Device-to-terminal misconnection expectation mapping, as in table 1:
TABLE 1 device-terminal Mixed connection Structure mapping (II)
(3) Device-to-terminal network connection expectation mapping, as shown in table 2:
TABLE 2 device-terminal networking map (III)
(4) Desired mapping across interval connections, as in table 3:
TABLE 3 Cross Interval connection Structure mapping (IV)
Calculating local expected node codes on the basis of the local mapping relation
Wherein x represents I-IV, based on the corresponding relationship between the node number of the whole network and the local node number, the local node number is spliced into the expected node number of the whole network by adopting a splicing method of an OR principle, and when a certain node belongs to a plurality of categoriesTaking OR operation from the expected state of the node in the multi-class splicing process to form an expected node coding array N
j (l)。
The intelligent substation communication link fault positioning method based on multi-warning information fusion is characterized in that the fitness function f it (n) is as in formula (4):
wherein, f it (n) is the fitness value of the nth individual; the value of M is taken to account for all the internodes (including dummy nodes)
Dots) 2 times the total; n is a radical of hydrogen
j Representing a node fault coding array; n is a radical of hydrogen
j (l) Representing a desired node encoding array; Σ denotes an accumulated sum; eta is [0, 1]]Positive real numbers in between, called weight coefficients; product of the characterising weight coefficients and the sum of all link state codes, L
i Representing a link state array.
The intelligent substation communication link fault positioning method based on multi-warning information fusion is characterized in that the link state array L i Is a set of parameters, L, that characterizes the link state of the whole network i Is as shown in formula (2);
wherein L is i Representing the link state array,/ i (i=1,2,L e 1 ) For the status coding of the ith link, e 1 Is the total number of links.
The intelligent substation communication link fault positioning method based on multi-warning information fusion is characterized in that in the step 6, the self-adaptive intersection and variation probability is calculated according to the following formula (5) (6):
wherein: p c Representing an adaptive crossover probability; p m Representing an adaptive mutation probability; f. of max Representing a maximum fitness function value in the population; f. of avg Representing the average fitness function value of each generation of population; f represents the larger fitness function value of the 2 individuals subjected to the cross operation; f' represents the fitness function value of the individual subjected to the mutation operation; in addition, k 1 =0.9,k 2 =0.6,k 3 =0.1,k 4 =0.01。
The adaptive inertial weight is as follows (7):
wherein: w is a max And w min Respectively, the maximum and minimum of the inertial weight, usually taken as w max =0.9,w min =0.4;f″、f min 、f avg Respectively representing the current fitness, the minimum fitness and the average fitness of the particles; t and T max Respectively representing the current iteration number and the maximum iteration number.
The adaptive acceleration factor is as follows (8):
wherein: c. C 1max 、c 1min 、c 2max 、c 2min Are acceleration factors c respectively 1 And c 2 Maximum and minimum values of (c); t represents the current iteration number; t is a unit of max Represents the maximum number of iterations; for the most value of the acceleration factor, c is selected 1max =1.3、c 1min =1.1、c 2max =2.0、c 2min =1.2。
Updating the speed and direction of the adaptive particle swarm algorithm as shown in the formulas (9) and (10):
in order to prevent saturation of the sigmiod function in equation (10), the velocity of the particles is generally limited to the range of [ -4,4 ]. sigmiog function as formula (11)
Wherein:
respectively the speed and position of the ith particle at the (k + 1) th iteration; w is the inertial weight;
the speed and the position of the i particles at the k iteration are respectively; c. C
1 、c
2 Self-acceleration factor and social acceleration factor, respectively, typically non-negative constants; r is
1 、r
2 Is in the interval [0,1]A random number in between;
respectively an individual optimal position and an overall optimal position of the particle;
is [0,1 ]]A random number of ranges; exp denotes an exponential function with a natural constant e as the base.
The invention has the beneficial effects that:
(1) According to the method, only the existing communication warning information of the intelligent substation is used for constructing the fitness function, and the node link mapping relation is utilized to realize accurate positioning under the condition of multiple link faults of different levels;
(2) The communication link positioning model based on the adaptive genetic particle swarm algorithm has the advantages of high convergence speed, good adaptability and visual positioning result, and the field operation and maintenance efficiency is improved to a great extent.
Detailed Description
The invention provides an intelligent substation communication link fault positioning method based on multi-warning information fusion, which specifically comprises the following steps of:
step 1: numbering the secondary system communication networks;
and node numbering: the node numbering follows the principle from top to bottom, and virtual node numbering is introduced into the node for the nodes which are directly connected and connected through the switch; numbering the spacing layers in sequence according to the sequence of protection (R), measurement and Control (MC) and wave recording; numbering process layer equipment in sequence according to the sequence of an Intelligent Terminal (IT) and a Merging Unit (MU); in order to determine that each link has a unique corresponding node code, aiming at the condition that partial devices have direct connection and network connection, introducing a virtual node number into the device nodes which have direct connection and network connection through a switch, and adopting the virtual node number when node state coding is carried out on the links related to the network connection through the switch;
and link numbering: the link numbering follows the principle that from top to bottom, from left to right, real nodes are firstly connected with virtual nodes, and then the interval is crossed, and for the condition that the same device is provided with a plurality of links connected with the same device, the links are numbered in sequence from small to large according to the nodes at the lower level of the links.
Based on the typical topology of the secondary system of the intelligent substation as shown in fig. 1, the line interval number formed according to the numbering principle is as shown in fig. 2;
step 2: determining the state of a link-related node based on the communication link fault identification of the alarm information; after receiving the alarm information from the device, the background determines the node state according to the following mechanism;
(1) Determining the state of a link node without being connected by a switch; and aiming at the direct connection link between the devices, the nodes at two ends of the link can be positioned by receiving the alarm information of the device at the background. Acquiring node information by adopting a BM field matching algorithm, taking the node information in the alarm information as a key word, and positioning nodes at two ends of a link alarm as actual measurement node state information;
(2) Determining the state of a link node connected through a switch; the fault link positioning method provided by the invention is characterized in that when a link connected through a switch is dealt with, differential analysis of alarm information is added on the basis of the direct link; as shown in fig. 3, when the lower link of the switch fails, all devices connected to the upper layer of the switch lose the information collected by the MU1, as shown in fig. 4, when the upper link of the switch fails, only the device of the upper layer corresponding to the link sends an alarm; combining BM field matching algorithm according to differentiated alarm information to form node state analysis including both ends of the switch link;
based on the principle of identifying the alarm information link fault, the acquisition of the node state of the whole network is completed to form N j Preparing a node state array;
and 3, step 3: constructing a node link coding rule and establishing a node state coding array N j (ii) a Based on the node state information containing the fault state obtained in the step 2, a node fault coding array N is established according to the formula (1) j ;
Wherein N is j Representing node state array, n j (j =1,2, \8230;, m) is the state encoding of the jth node; m is the total number of nodes including the virtual node;
link state array L i Is a set of parameters, L, that characterizes the link state of the whole network i Is defined as shown in formula (2), and the link status L i As the output result of the method provided by the invention, the formula (2) is a definition rule, and the specific output and use are shown in the following step 6;
wherein L is i Representing the link state array,/ i (i=1,2,L e 1 ) For coding the state of the ith link, e 1 Is the total number of links;
and 4, step 4: establishing secondary system failure link node expectation N j (l) (ii) a According to the connection mode and characteristics of the secondary system in the whole network, four typical connection structures shown in fig. 5 to 8 are established, and four types of node link expectation mappings are established based on fig. 5 to 8, as shown in formula (3) and tables 1 to 3:
(1) Background-device connection expectation mapping I, the function is as follows;
wherein
Indicating node expectation under the type I connection structure, and a subscript j indicating an expected state value of a jth node;
representing a loop operation, wherein a is a loop starting point, b is a loop end point, and when the loop is in a loop interval, the operation of the c expression is executed; e.g. of the type
1 Represents the total number of links; l
i (i=1,2,L e
1 ) Coding the state of the ith link; k
I Showing the connection condition of the station control layer switch SW-A of the switch and the lower layer of the link if the link is l
j Has up>A connection to SW-A, then K
I =1, otherwise K
I =0;K
II Showing the connection condition of the upper layer of the link and the station control layer switch SW-A of the switch, if the lower layer of the link has connection with the SW-A, K is
II =1, otherwise K
II =0; "|" represents a logical or operation;
(2) Device-to-terminal misconnection expectation mapping, as in table 1:
TABLE 1 device-terminal Mixed connection Structure mapping (II)
(3) Device-to-terminal network connection expectation mapping, as shown in table 2:
TABLE 2 device-terminal networking map (III)
(4) Desired mapping across interval connections, as in table 3:
TABLE 3 Cross-granularity connection structure map (IV)
At the base of the above-mentioned local mapping relationCalculating local expected node coding based on
Wherein x represents I-IV, based on the corresponding relation between the node number of the whole network and the local node number, the local node number is spliced into the expected node number of the whole network by adopting the splicing method of the 'OR' principle, when a certain node belongs to a plurality of categories, the expected state of the node is subjected to 'OR' operation in the multi-category splicing process to form an expected node code array N
j (l);
And 5: based on N j (l) Constructing fitness function f of multi-information fusion it (n) of (a). The construction of the fitness function is the basis of solving by applying an intelligent algorithm, and the whole network node expected state code N formed based on the step 4 is j (l) And constructing a formed fitness function as shown in the formula (4):
wherein, f
it (N) is the fitness value of the nth individual, for a total of N individuals; the value of M is taken to be 2 times the total number of interior nodes (including dummy nodes) taking into account all the internodes; n is a radical of
j Representing a node fault coding array; n is a radical of
j (l) Representing a desired node encoding array; Σ denotes an accumulated sum; eta is [0, 1]]Positive real numbers in between, called weight coefficients;
the product of the representation weight coefficient and the sum of all link state codes can avoid one-value multi-solution after the fitness function is added into the term;
and 6: and solving the AGA-ABPSO positioning model and outputting positioning information. The positioning model based on the adaptive genetic particle swarm optimization is shown in FIG. 9, and after the initialization setting of the algorithm-related parameters is completed, the actually measured node fault coding array N is obtained j As algorithm input, randomly initialized population dimension information is used as link state coding L i Forming the node expected state code of the whole network according to the mapping relation in the step 4N j (l) According to the fitness function in the step 5, iterative solution is carried out on the population optimal fitness f it (n), outputting optimal particle information to complete fault link positioning;
calculating the self-adaptive cross and variation probability as shown in formulas (5) and (6):
wherein: p c Representing an adaptive crossover probability; p m Representing an adaptive mutation probability; f. of max Representing a maximum fitness function value in the population; f. of avg Representing the average fitness function value of each generation of population; f represents the larger fitness function value of the 2 individuals subjected to the cross operation; f' represents the fitness function value of the individual performing the mutation operation. In addition, k 1 =0.9,k 2 =0.6,k 3 =0.1,k 4 =0.01。
The adaptive inertial weight is as follows (7):
wherein: w is a max And w min Respectively, the maximum and minimum of the inertial weight, usually taken as w max =0.9,w min =0.4;f″、f min 、f avg Respectively representing the current fitness, the minimum fitness and the average fitness of the particles; t and T max Respectively representing the current iteration number and the maximum iteration number.
The adaptive acceleration factor is as follows (8):
wherein: c. C 1max 、c 1min 、c 2max 、c 2min Are acceleration factors c, respectively 1 And c 2 Maximum and minimum values of; t represents the current iteration number; t is max Representing the maximum number of iterations. For the most value of the acceleration factor, c is selected 1max =1.3、c 1min =1.1、c 2max =2.0、c 2min =1.2。
Updating the speed and direction of the adaptive particle swarm algorithm as shown in the formulas (9) and (10):
in order to prevent saturation of the sigmiod function in equation (10), the velocity of the particles is generally limited to the range of [ -4,4 ]. sigmiog function as formula (11)
Wherein:
respectively the speed and the position of the ith particle in the (k + 1) th iteration; w is the inertial weight;
the speed and the position of the i particles at the k iteration are respectively; c. C
1 、c
2 Self-acceleration factor and social acceleration factor, respectively, typically non-negative constants; r is
1 、r
2 Is in the interval [0,1]A random number in between;
respectively the individual optimum bit of the particleSetting and integrating the optimal positions;
is [0,1 ]]A random number of ranges; exp denotes an exponential function with a natural constant e as the base.
Example analysis: in order to verify the effectiveness of the secondary system communication link fault positioning method provided by the invention, verification is performed based on a typical interval (line, bus and transformer) of a secondary system at 220kV side of an intelligent substation as an example. The schematic diagram of the secondary system multi-interval connection based on the actual system coding is shown in fig. 10. And the nodes and the links are numbered according to the principle in the step one, the node numbers are marked beside the nodes, and the link numbers are marked in the links. The algorithm initializes and sets the number of population N =50, the maximum iteration number T =100, the spatial dimension D =41 of the population, the coding length L =41, and the relevant parameters of the algorithm are as described in the above step 6.
The spliced local expected node code N j x (l) Forming a desired node encoding array N j (l) When the bus bar splicing method is used, the corresponding relation between the whole network node number and the local node number is as follows, the splicing process is carried out according to the following table, 17 NA in the table indicates that 17 8 in the bus bar interval respectively corresponds to 12 3 of the local node number, NA indicates that no correspondence exists with 4 nodes, and mapping is carried out only according to 12 3 corresponding to 17 8:
(1) Station control layer link fault example
Randomly setting relevant link faults of a station control layer, and verifying the accuracy of the positioning method;
assuming that the link 8 is in failure, the node state information obtained by reasoning the failure information is expressed as N j =[1 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0]And inputting the data into an algorithm provided by the text to obtain the individual fitness and the individual dimension information. As can be seen from fig. 11, when the link 8 fails, the fitness value of an individual is 59.9 at most, the corresponding individual number is 19, and as can be seen from fig. 12, the dimension status information of the 19 th individual is: [00 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]The 8 th dimension in the individual is state 1, and the corresponding link 8 fails, with the same result as the expected result. The simulation verification results of the rest links in the station control layer are shown in table 4:
TABLE 4 simulation results of station control layer link failures
(2) Process layer link failure example
And setting the related link failure of the process layer and verifying the effectiveness of the positioning method. In the example of process layer correlation, a link including a virtual node (link 12), a link connected across an interval (link 22), and a process layer switch lower layer link (link 27) are selected for verification. The simulation results are shown in table 5:
(3) Multiple fault location calculation
Setting multiple position link faults including multiple station control layers, links of the station control layers and the process layers connected with the same nodes, cross-interval link of the process layers and verification of the process layers through typical and complex links of the switch links. The simulation results are shown in table 6:
through the 3 example results, the positioning method can accurately position the fault link when dealing with link faults of different levels, particularly faults of direct connection, network connection and cross-interval connection, and the feasibility and the result accuracy of the communication link positioning method based on the background alarm information coding of the intelligent substation are demonstrated;
(4) Adaptive genetic particle swarm algorithm convergence rate comparison
Based on the example (1), the convergence rates of the AGA-ABPSO, AGA and ABPSO algorithms are compared. The convergence rates of the three algorithms are shown in fig. 13. In the aspect of iteration speed, the AGA-ABPSO algorithm shows better performance in dealing with information input and optimization processes based on binary codes. The algorithm sets the maximum iteration for 100 times, and the AGA-ABPSO algorithm adopted by the invention is stabilized at the global maximum fitness and kept stable in 27 iterations, and has higher convergence rate. Although the AGA algorithm can also achieve stable convergence within the maximum iteration number, the AGA algorithm is far behind the AGA-ABPSO algorithm used herein in the aspect of iteration speed, the overall optimal solution is achieved only in the 94 th iteration, and the AGA algorithm also has a redundant iteration phenomenon in the process of the overall optimal solution, so that the iteration speed is low; the ABPSO algorithm iteration speed lies between the two, and reaches global convergence after 59 iterations.
GOOSE in the drawing is a generic object oriented substation event.
SV depicted in the drawings is a sampled value.
The AGA described in the above examples is an adaptive genetic algorithm.
The AGA-ABPSO described in the above examples is an adaptive genetic particle swarm algorithm.
The ABPSO described in the above embodiments is a self-adaptive particle swarm algorithm.
The invention has the beneficial effects that:
(1) According to the method, only the existing communication warning information of the intelligent substation is used for constructing the fitness function, and the node link mapping relation is utilized to realize accurate positioning under the condition of multiple link faults of different levels;
(2) The communication link positioning model based on the adaptive genetic particle swarm algorithm has the advantages of high convergence speed, good adaptability and visual positioning result, and the field operation and maintenance efficiency is improved to a great extent.
The above-described embodiments are merely preferred technical solutions of the present invention, and should not be construed as limiting the present invention. The protection scope of the present invention is defined by the claims, and includes equivalents of technical features of the claims. I.e., equivalent alterations and modifications within the scope hereof, are also intended to be within the scope of the invention.