A power grid fault diagnosis method applying red-billed blue magpie algorithm under multiple dimension reduction

Abstract

The application discloses a power grid fault diagnosis method using a red-billed blue magpie algorithm under multiple dimension reduction, converts a high-voltage power grid into a topological structure, regards three elements of a power transmission line, a transformer and a bus as detection sections, regards circuit breakers at both ends of the elements as nodes at both sides of the detection sections, the nodes are the circuit breakers, step 2, determines node state expected values at both sides of each section under different fault states of each section; based on overcurrent alarm information coding of the nodes at both sides of the section and the node state expected values at both sides of each section under different fault states of each section, optimizes a fault diagnosis target function by using a red-billed blue magpie optimization algorithm, obtains a fault state of each section and corrected overcurrent alarm information coding of nodes at both sides of each section under the condition of minimizing the fault diagnosis target function. The application solves the problems of low fault positioning accuracy and slow speed under the high-voltage power grid.

Description

TECHNICAL FIELD

[0001] The application relates to a power system relay protection field and discloses a power grid fault diagnosis method applying a red-beaked blue crane algorithm under multiple dimension reduction. BACKGROUND

[0002] With the gradual expansion of the scale of the power system, the connection between power grids is closer, the topological structure and operation mode of the high-voltage power transmission grid are increasingly complex, and in the face of the traditional power grid fault diagnosis scheme, the requirements of increasingly complex power grids cannot be met, the fault diagnosis accuracy is not high, the fault tolerance is low, the speed and reliability are difficult to meet, and the fault diagnosis capability is still insufficient when the information is misreported. The application has obvious improvement in the accuracy of fault positioning of high-voltage complex power grids by reducing the dimension of the reported data, simplifying the topological structure of the power grid, introducing the red-beaked blue crane algorithm to optimize the objective function positioning, and the auxiliary positioning of the relay protection online detection system.

[0003] In order to improve the fault diagnosis effect, the information reported by the relay protection device installed on each sectional switch or circuit breaker is widely used for fault component judgment, but the relay protection device is prone to misreporting. In order to realize power grid fault diagnosis, expert systems, fuzzy Petri nets, Bayesian networks, artificial neural networks, rough set theory and other methods are usually used. Although the above methods improve the fault diagnosis capability, the accuracy of the uploaded information is required to be high, and if information loss occurs, the effect of power grid fault diagnosis will be affected. In view of this problem, one method is to classify the fault information and extract appropriate feature quantities, and to determine the fault probability of each component by using the D-S data fusion method (Wang Yu, Xu Changbao, Zhu Jianyang, etc. Multi-source information fusion power grid fault diagnosis method based on improved Petri net and Hilbert transformation [J]. Electrical Measurement and Instrumentation, 2021.), but if the uploaded information is wrong, the diagnosis result obtained will not be consistent with the actual situation. Another method is to reconstruct the fault diagnosis model, and to perform fault positioning through multi-source information auxiliary judgment (Zhang Chunmei, Xu Xingque, Liu Silin. Distribution network fault diagnosis technology based on multi-source data fusion [J]. Journal of Shanghai Jiaotong University, 2024.), but it is only suitable for single fault condition and the application scene is limited. At present, most of the fault diagnosis models have too many variable dimensions, the fault diagnosis model is complex and difficult to construct, the application scene is single, the uploaded information is required to be extremely high, and the model cannot be diagnosed in special misreporting scenarios, which has a certain misreporting risk. The new power grid fault diagnosis method of the present application can simplify the variables in the model and the topological structure of the power grid, and prevent misjudgment in special scenarios by using the relay protection online monitoring system, thereby ensuring the accuracy of the fault diagnosis model. SUMMARY

[0004] The application aims at the above-mentioned problems existing in the prior art, and provides a power grid fault diagnosis method applying the Magpie algorithm under multiple dimension reduction.

[0005] The above-mentioned purpose of the application can be realized by the following technical means:

[0006] A power grid fault diagnosis method applying the Magpie algorithm under multiple dimension reduction comprises the following steps:

[0007] Step 1: converting a high-voltage power grid into a topological structure, regarding three elements of a power transmission line, a transformer and a bus as detection sections, regarding circuit breakers at both ends of the elements as nodes at both sides of the detection sections, and regarding the nodes as the circuit breakers,

[0008] Step 2: determining node state expected values at both sides of each section under different fault states of each section;

[0009] Step 3: based on overcurrent alarm information coding of nodes at both sides of the section and the node state expected values at both sides of each section under different fault states of each section, optimizing a fault diagnosis target function by using the Magpie optimization algorithm, and under the condition of minimizing the fault diagnosis target function, obtaining a fault state of each section and corrected overcurrent alarm information coding of nodes at both sides of each section.

[0010] As described above, the node state expected values and the overcurrent alarm information coding both comprise 1, -1 and 0, which respectively correspond to existence of positive, negative direction or undetected fault current of the nodes.

[0011] As described above, the fault diagnosis target function is based on the following formula:

[0012]

[0013] In the formula, ω1, ω2 and ω3 are weight coefficients, m k is a correction parameter, m k =1 when the node state expected value of the node considering the false alarm is greater than the overcurrent alarm information coding, otherwise m k =0; n k is a correction parameter, n k =1 when the node state expected value of the node considering the false alarm is less than the overcurrent alarm information coding, otherwise n k =0; s i represents whether the i-th section is faulty, s i =0 indicates no fault, s i =1 indicates existence of the fault, m k , n k , s i are all of a Boolean type, I j is overcurrent alarm information coding uploaded at the node j after the fault, I jR is the number of sections, D is the number of switches, and y is the fault hypothesis sequence of the section.

[0014] As described above, the section in step 2 is simplified based on the region similarity principle, that is, when any section in a region fails respectively, all node state quantities calculated outside the region do not change, so all sections in the region can be equivalent to one section.

[0015] As described above, the red-billed blue optimization algorithm is used to optimize the fault diagnosis objective function, including the following steps:

[0016] a) Initialization based on the following formula:

[0017] x p,q = (ub-lb) x Ranfl + lb

[0018] In the formula, p represents the serial number of a single individual, q represents the position dimension serial number of the individual, represents the fault state of the section in the qth position dimension of the pth individual, ub and lb are the upper and lower bounds of the search space respectively, and Ranfl represents a random number for generating a standard normal distribution.

[0019] b) Find food, in the process of finding food, the red-billed blue searches in small groups or crowds,

[0020] Search for food in small groups based on the following formula:

[0021]

[0022] In the formula, t represents the current iteration number, X h (t+1) represents the hth new search agent position of the t+1th iteration, h is the serial number, X h (t) represents the hth new search agent position of the tth iteration, u represents the number of red-billed blues randomly selected from all participating searchers, X w (t) represents the wth individual, X h represents the individual with serial number h, X rs (t) represents the randomly selected search agent in the current iteration, and Ranf2 represents a random number for generating a standard normal distribution.

[0023] Search for food in crowds based on the following formula:

[0024]

[0025] In the formula, v represents the number of search agents in the cluster when exploring food, r represents the total number of red-billed blues participating in the search, and Ranf3 represents a random number for generating a standard normal distribution.

[0026] c) attacking prey,

[0027] The attack mathematical model corresponding to small group action is:

[0028]

[0029] The attack mathematical model corresponding to group action is:

[0030]

[0031] In the formula, X food (t) represents the position of food, the adjustment factor LP=(1-t / T)^(2*(t / T)), T is the total number of iterations, Ranr1 and Ranr2 represent random numbers for generating standard normal distribution,

[0032] d) storing food

[0033] The storage of food is based on the following formula:

[0034]

[0035] In the formula, And Respectively represent the fitness values of the hth red-billed blue magpie before and after position update.

[0036] A computer device comprising a memory and a processor, the memory stores a computer program, and the processor implements the steps of the above fault diagnosis method when executing the computer program.

[0037] A computer readable storage medium, which stores a computer program, the computer program is executed by a processor to implement the steps of the above fault diagnosis method.

[0038] A computer program product, comprising a computer program, which is executed by a processor to implement the steps of the above fault diagnosis method.

[0039] Compared with the prior art, the present application has the following beneficial effects:

[0040] (1) Compared with other methods, the present method has a wider range of use, can adapt to complex high-voltage power grid fault problems, avoids false positives, and meets the system reliability and speed requirements;

[0041] (2) The dimension reduction of the power grid topology structure proposed in the present method simplifies the positioning objective function and the system topology structure, and is most conducive to the use of optimization algorithms;

[0042] (3) The method determines the fault section, and adopts the relay protection online monitoring system to prevent fault diagnosis in a special false alarm scene, thereby ensuring the accuracy of fault diagnosis. BRIEF DESCRIPTION OF DRAWINGS

[0043] Figure 1 Fig. 1 is a high-voltage power grid topology structure diagram, wherein T represents a transformer, B represents a bus, L represents a line, S represents a power source, and the subscript number represents a serial number;

[0044] Figure 2 Fig. 2 is an equivalent topology structure of a power grid, wherein the number without brackets is a node serial number, the number with brackets is a section serial number, S4' is an equivalent power source when a closed-loop network is decomposed into an open-loop network, and has the same meaning as S4;

[0045] Figure 3 Fig. 3 is a signal transmission process diagram;

[0046] Figure 4 Fig. 4 is a power grid topology dimension reduction diagram; wherein the number without brackets is a node serial number after dimension reduction, the number with brackets is a section serial number after dimension reduction, a, b and c are similar sections in the respective frames;

[0047] Figure 5 Fig. 5 is a hierarchical equivalent structure diagram; wherein the number with a superscript is a node serial number after dimension reduction;

[0048] Figure 6 Fig. 6 is a region similar circuit;

[0049] Figure 7 Fig. 7 is a power grid example model diagram; wherein the number without brackets is a node serial number in the example model, the number with brackets is a section serial number in the example model, and the number in the frame indicates the number of intelligent terminal devices equipped in the corresponding section in the model.

[0050] Figure 8 Fig. 8 is a flowchart of the present application. DETAILED DESCRIPTION

[0051] In order to facilitate those skilled in the art to understand and implement the present application, the present application will be further described in detail below in conjunction with examples, and the examples described herein are only used to illustrate and explain the present application, and are not a limitation on the present application.

[0052] A power grid fault diagnosis method applying the Red-billed Blue Magpie algorithm under multiple dimension reductions, comprising the following steps:

[0053] Step 1: converting a high-voltage power grid into a topology structure, regarding a power transmission line, a transformer and a bus as a detected section, regarding circuit breakers at both ends of an element as nodes at both sides of the detected section, and determining node state expected values at both sides of each section under different fault conditions of each section.

[0054] Equivalent high-voltage power grid topology, and information correction and one-dimensional reduction. The high-voltage power grid topology structure is shown in Figure 1 , T represents a transformer, B represents a bus, L represents a line, S represents a power source, and the breakers are connected at both ends of each element. The diagnosis of power grid faults mainly aims at the diagnosis of the faults of the above-mentioned elements in the power grid, and the breakers at both ends of each element act when a fault occurs to realize the main and backup protection. According to Figure 1 , it can be seen from the power grid structure that the variable dimension in the traditional fault diagnosis model is extremely large, and the breakers are jointly affected by multiple protections, so that the protection state quantity is several times more than the breaker state quantity, which is not conducive to the solution of the fitness function based on the protection and breaker state. Moreover, the protection device usually has a redundant system, so that the probability of distortion of the protection quantity is small, and the protection action directly corresponds to the breaker action, and the breaker tripping depends on the fault state of the corresponding element in the power grid, so the tripping information of the breaker uploaded by the relay protection device can be directly used for power grid fault diagnosis. It is fault diagnosis, not line protection. After the fault occurs and the protection acts, the actual fault element is diagnosed according to the node information to prevent misjudgment caused by protection refusal, misoperation, missed report, false report, etc. Thus, the high-voltage topology structure can be equivalent, and the transmission line, transformer and bus three elements are regarded as detection sections, and the breakers at both ends of the elements are regarded as nodes at both sides of the detection sections. The equivalent topology structure diagram of the high-voltage power grid is shown in Figure 2 , S is still the system power source, S4 and S4' are the same power source, the black dots are nodes, i.e. breakers, and the line segments between the dots are detection sections equivalent to the transmission line, transformer and bus three elements.

[0055] The breakers in the power grid are controlled by the relay protection device, and the relay protection device is usually used to monitor the fault current flowing through the breakers when the breakers do not act after a fault occurs and report it. When the relay protection device detects the fault current flowing through a node (breaker), the overcurrent alarm information I j is encoded as 1, -1 and 0 and reported again. When there is a positive or negative direction or no fault current is detected at a node, the fault current information encoded as 1, -1 and 0 is reported. However, due to the installation of the relay protection device outdoors, the uploaded switch information may be missed or false due to environmental influences. Moreover, in a complex power grid, the fault diagnosis mainly relies on the switch quantity, so the complex ring network system can be equivalent to an open-loop structure. Thus, the fault section discrimination problem is converted into an objective function optimization problem, but the variable dimension in the objective function will be greatly increased when considering the missed and false reporting of the switch information uploaded by the relay protection device, which seriously affects the algorithm optimization. Therefore, the switch information reported by the relay protection device is corrected and reduced in dimension.

[0056] The information transmission process of the relay protection acquisition device is that the node expected value is modified to the node actual state quantity, the multiple sets of section fault information assumed when using the step three optimization algorithm, the node expected value 1, 0 or -1 is calculated through the section fault condition of each set, the node actual state quantity 1, 0 or -1 is determined by the information uploaded by the relay protection device, and then the expected value and the actual state quantity are jointly substituted into the objective function of formula (1) for solving. According to the solving result, the new section assumption condition is automatically updated by the optimization algorithm in step three, and the expected value is calculated again according to the assumption. The cycle is repeated until the maximum iteration number is reached. The transmission process is shown in FIG. 3. Figure 3 The right side is the node state quantity actually reported by the relay protection device. The information transmission is divided into three cases through the three kinds of node actual state quantity uploaded by the relay protection device. In each case, the left side is the three possible results of the node expected value calculation: 1, -1 and 0. If the calculation result is different from the actual state quantity, it is a false report. The false report and the omission are unified as a false report during information transmission. According to the actual node state quantity, there will be only one correct reporting case. The correction process of the correct reporting in FIG. 3 is marked as z k The false reporting case is corrected again: the correction process of the larger node expected value calculation result is marked as m k , and the correction process of the smaller expected value is marked as n k When m k or n k = 1, it indicates that the node has information false reporting. Thus, the simplified fault diagnosis objective function minFit(y) is obtained:

[0057]

[0058] In the formula, ω1, ω2, ω3 are weight coefficients, m k is a correction parameter, m k = 1 when the expected value of the node state quantity after considering the false reporting is greater than the overcurrent alarm information code, otherwise m k = 0; n k is a correction parameter, n k = 1 when the expected value of the node state quantity after considering the false reporting is less than the overcurrent alarm information code, otherwise n k = 0; m k or n k = 1 indicates that the node has information false reporting, s i is used to indicate whether the ith section is faulty, s i = 0 indicates no fault, and s i = 1 indicates that there is a fault, m k , n k , s i are all Boolean types, |I j -I' j| is the main criterion for power grid fault diagnosis, I j | is the coding of over-current alarm information uploaded at node j after fault, I j | is the expected value of the state quantity of the mistaken node, I k +n k | is the auxiliary criterion for power grid fault diagnosis, | is the number of fault sections, R is the number of sections, D is the number of switches, y is the sequence of fault hypotheses of the section, j, i, k are serial number parameters.

[0059]

[0060] where "¬" represents logical negation, "+" and "-" represent algebraic addition and subtraction operations, respectively, is the function of the fault hypothesis variable s i .

[0061] Step 2: Determine the expected values of the node states on both sides of each section under different fault states of each section.

[0062] The bus, circuit breaker, and line in the line are equivalent to a section, and the fault state of each section is taken as the object to be processed. Combined with the over-current information reported by each node, the result is the fault condition of each section, and the corresponding section is 1, i.e. fault occurs. Among them, the fault state is taken as the object to be processed, and the regional equivalence principle is used to optimize the topology structure of the power grid.

[0063] Each branch in the equivalent topology diagram of the high-voltage power grid shown in Figure 2 is equivalent to a region (a), (b), or (c), and the topology dimension reduction diagram is shown in Figure 4 Each region contains several sections, and the switch function vector I * (X) calculated by formula (2) can be found that when any section in a region fails, all node state quantities calculated outside the region do not change, so all sections in the region can be equivalent to one section, so that a region only consists of one section and the nodes on both sides of the section, i.e. the "regional similarity" principle greatly simplifies the number of nodes, and the power grid topology structure is optimized to two layers by this principle. The first layer of the equivalent model shown in Figure 5 is optimized by the optimization algorithm, and the fault region is located. Then the topology outside the fault region is equivalent to a power source. Taking the case where the fault occurs in region a as an example, the equivalent circuit of the second layer is shown in Figure 6 , and then the fault section inside region a is located. Finally, the principle is used to simplify other regions, and the overall structure of the complex power grid system is optimized.

[0064] Step 3: Based on the overcurrent alarm information coding of the nodes on both sides of the section and the expected value of the node state on both sides of each section under different fault states of each section, the red-billed blue magpie optimization algorithm is used to optimize the fault diagnosis objective function, and under the condition of minimizing the fault diagnosis objective function, the fault state of each section and the corrected overcurrent alarm information coding of the nodes on both sides of each section are obtained.

[0065] After double optimization of the variable dimension and structure topology of the complex power grid, the variable dimension in the fault diagnosis model is reduced, and the objective function of fault diagnosis is as shown in formula (1) above. In the cycle process proposed in step one, the red-billed blue magpie algorithm updates the section hypothesis so that the objective function Fit(y) gradually decreases. The fault positioning problem can be equivalent to the optimization problem of the minimum value of the objective function, and the section hypothesis condition corresponding to the minimum objective function value is the predicted section fault positioning result.

[0066] The algorithm is as follows:

[0067] a. Initialization. Initialization includes population initialization and parameter initialization. Population initialization is to randomly generate a group of solutions (called population), each solution corresponding to the position of a red-billed blue magpie (i.e. a fault hypothesis sequence composed of fault states of each section). Parameter initialization is to set the parameters of the algorithm, such as population size, maximum number of iterations, etc.:

[0068] x p,q = (ub-lb) x Ranfl + lb (3)

[0069] In the formula, p represents the serial number of a single individual, q represents the position dimension serial number of the individual, x p,q represents the fault state of the section of the qth position dimension of the pth individual, ub and lb are the upper and lower bounds of the search space respectively, the upper bound corresponds to the fault state of the section in the fault diagnosis model (1 for fault), the lower bound corresponds to the fault state of the section in the fault diagnosis model (0 for no fault), and Ranfl represents a random number for generating a standard normal distribution.

[0070] b. Finding food. In the process of finding food, red-billed blue magpies usually search in small groups or flocks to obtain higher search efficiency. And they use various techniques to search for food resources, and their adaptability and flexibility enable them to use various hunting strategies to ensure adequate food supply. That is, in the process of foraging, the position is updated through the formula, and the position update formula usually contains global search and local search.

[0071] 1) When searching for food in small groups, iterate:

[0072]

[0073] where t represents the current iteration number, X h (t+1) represents the hth new search agent position of the t+1th iteration, h is the serial number, X h (t) represents the hth new search agent position of the tth iteration, u represents the number of red-billed blue magpies in the 2-5 small groups randomly selected from all the red-billed blue magpies participating in the search, X w (t) represents the wth individual, X h represents the individual with serial number h, X rs (t) represents the search agent randomly selected in the current iteration, Ranf2 represents a random number for generating a standard normal distribution.

[0074] 2) Iteration when searching for food in groups:

[0075]

[0076] where v represents the number of search agents of the cluster when exploring food, between 10 and r, r represents the total number of red-billed blue magpies participating in the search of the cluster. It is also randomly selected from the entire population, Ranf3 represents a random number for generating a standard normal distribution, and the remaining parameters and functions have the same meaning as formula (4).

[0077] Finding food is the process of changing the position of part of the individual in the red-billed blue magpie algorithm, the purpose is to find the position direction that minimizes the objective function, and the change constitutes is not directly obtained from the optimal position, therefore, it is necessary to repeatedly iterate to find, and 1), 2) represent two ways of changing the position, which is automatically judged by the algorithm.

[0078] c. Attacking prey. Red-billed blue magpies have high hunting proficiency and cooperative spirit when chasing prey, they take different actions for different targets, emphasizing the multiple strategies and skills they possess, small groups of action for small prey or plants, using tactics such as rapid pecking, jumping to catch prey, or flying to catch insects; group action for large prey, successfully obtaining food in various situations.

[0079] 1) The attack mathematical model corresponding to small group action is:

[0080]

[0081] 2) The attack mathematical model corresponding to group action is:

[0082]

[0083] where X food(t) represents the position of the food, the adjustment factor LP = (1-t / T)^(2*(t / T)), T is the total number of iterations, Ranr1 and Ranr2 represent the random numbers used to generate the standard normal distribution (mean 0, standard deviation 1), and the meanings of the remaining parameters and functions are the same as b.

[0084] The attack on prey involves changing the group's position to move the overall position closer to the optimal position. The position change process is divided into parts 1) and 2), which are automatically judged and selected by the algorithm.

[0085] d. Storing food

[0086] Besides searching for and attacking food, the red-billed blue magpie stores surplus food in secluded locations for future food shortages. This process preserves information about the solution, facilitating the finding of the globally optimal solution. The mathematical model is as follows:

[0087]

[0088] In the formula, and represents the fitness value of the h-th red-billed blue magpie before and after the position update, respectively; the meanings of the other parameters and functions are the same as b.

[0089] Storing food is equivalent to judging whether the current position change is reasonable. If the objective function is not optimized after the position change, the position before the change is used to continue judging until the number of iterations reaches the set maximum number. When the algorithm obtains the optimal position, the objective function value will increase in subsequent iterations. Therefore, the optimal position can always be maintained through the content of segment d.

[0090] The undervoltage alarm of the online relay protection monitoring system is used to assist in fault diagnosis. In special cases where the relay protection device issues a false alarm and the false alarm result corresponds to a fault in another section, the faulty section may be misdiagnosed. The online relay protection monitoring system can collect and process power consumption information for each section in real time. Since it is mostly installed indoors, the reliability of the reported information is high. When a component in the power grid fails, the protection device operates, causing a voltage loss in the faulty section. The data acquisition system reports an alarm due to the voltage loss, and the control center receives all alarm information. However, since there are many power sources in the power grid, one power source is designated as the main power source, and the others are considered auxiliary power sources. After a voltage loss alarm in a faulty section, considering that there are faults on all paths connecting a section to all power sources, that section may also trigger a voltage loss alarm due to lack of power supply. Therefore, the alarm information can serve as an auxiliary diagnostic information source to eliminate misdiagnosis of the faulty section caused by special false alarm situations, thus ultimately determining the faulty section.

[0091] Example: Simulation comparison and verification analysis of fault location under different fault types. Utilizing PSCAD to build... Figure 7The model is shown, and fault positioning simulation verification is performed, wherein S is a power supply, and feeder sections are (1) to (16). To verify the accuracy and effectiveness of the new method, the accuracy of fault section positioning under single and double fault types is compared between the new method and the method in another document, which considers relay protection device false negatives and false positives, does not reduce the dimension of variables, and does not use a fault auxiliary judgment method. Under single fault, sections (2) and (4) are set to be faulty respectively, and different nodes are set to be false positives respectively. The test results are shown in Tables 1 and 2. Under double fault, sections (2) and (11) are set to be faulty, and different nodes are set to be false positives respectively. The test results are shown in Table 3.

[0092] Table 1 Fault table of section (2)

[0093]

[0094] Table 2 Fault table of section (4)

[0095]

[0096]

[0097] Under single fault, sections (2) or regions (4) are faulty. Taking region (2) as an example, the overcurrent alarm information collected by the relay protection online monitoring system is coded as [0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]. When there is no false positive node, the system receives current information of each node as [I1, K, I 20 ] = [1, 1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1]. When different nodes have false positives, the power grid is hierarchically optimized according to the region similarity principle of the present application, so that the dimension of the variables is reduced by nearly half. Then, the fault diagnosis positioning model of formula (1) is used to determine that region (1) is faulty. Then, region (1) is calculated equivalently, and the fault section (2) is located according to the red-beaked crane algorithm of the present application. Finally, the fault section (2) is determined by combining the fault positioning auxiliary judgment to exclude false positives. However, as shown in Table 1, when positioning is performed according to the method in another document without reducing the dimension of variables and without using a fault auxiliary judgment method, there is an inaccurate fault positioning, and false positives under special conditions are not considered.

[0098] Table 3 Fault table of regions (2) and (11)

[0099]

[0100] Under double faults, the fault occurs in section (2) and region (11), the overcurrent alarm information collected by the relay protection online monitoring system is coded as [0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], when no node exists false alarm, the system receives the current information of each node as [I1, K, I 20 ] = [1, 1, -1, -1, -1, -1, -1, -1, -1, -1, 0, 0, -1, -1, -1, -1, -1, -1, -1, -1], when different nodes exist false alarm, according to the method steps for determining the fault region provided in the patent, the fault section (2) and section (11) are finally located. As can be seen from Table 3, in the method of no-fault positioning auxiliary judgment, when nodes 11 and 13 exist false alarm, the fault diagnosis model directly judges that sections (10) and (12) are faulty, which leads to misjudgment in special scenarios; but according to the patent provided in the present application, if (10) and (12) are faulty, section (2) should have no loss of pressure alarm signal, and section (12) should have an alarm signal, which is contrary to the actual loss of pressure alarm signal, so it can be judged that the nodes near sections (10) and (12) exist false alarm, and combined with the positioning auxiliary judgment principle and the fault diagnosis result, it can be concluded that the actual fault sections are (2) and (11).

[0101] Tables 1, 2 and 3 reflect that the new power grid fault diagnosis method provided in the patent is more accurate than the diagnosis model without variable dimension reduction, and the applicable scenarios are not limited, which is applicable under single or double faults, the multi-source information auxiliary positioning used can prevent misjudgment in special false alarm scenarios, and plays a good role, and the positioning result shows that the method has good stability and applicability.

[0102] In one embodiment, a computer device is also provided, including a memory and a processor, the memory stores a computer program, and the processor implements the steps in the above method embodiments when executing the computer program.

[0103] In one embodiment, a computer readable storage medium is provided, which stores a computer program, and the computer program is executed by a processor to implement the steps in the above method embodiments.

[0104] In one embodiment, a computer program product is provided, which includes a computer program, and the computer program is executed by a processor to implement the steps in the above method embodiments.

[0105] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by a computer program instructing relevant hardware, and the computer program can be stored in a non-volatile computer readable storage medium. When the computer program is executed, the computer program can include the processes of the above-mentioned embodiment methods.

[0106] It should be noted that the embodiments described in the present application are only examples illustrating the spirit of the present application. Those skilled in the art can make various modifications or supplements to the described embodiments or replace them with similar ways, without departing from the spirit of the present application or exceeding the scope defined by the appended claims.

Claims

1. A power grid fault diagnosis method using the Red-billed Quelea algorithm in multiple dimension reduction, characterized in that, The method comprises the following steps: Step 1, converting a high-voltage power grid into a topological structure, regarding three elements of a power transmission line, a transformer and a bus as a detection section, regarding circuit breakers at both ends of the element as nodes at both sides of the detection section, the node being the circuit breaker, and the information transmission process of a relay protection collection device being a node expected value modified into a node actual state quantity, false positives and false negatives being unified as false reports during information transmission, Step 2, determining node state expected values at both sides of each section under different fault states of each section; Step 3, based on overcurrent alarm information coding of nodes at both sides of the section and node state expected values at both sides of each section under different fault states of each section, optimizing a fault diagnosis target function by using a red-billed blue magpie optimization algorithm, under the condition of minimizing the fault diagnosis target function, obtaining a fault state of each section and modified overcurrent alarm information coding of nodes at both sides of each section, and realizing fault auxiliary judgment in combination with a loss-of-voltage alarm of a relay protection online monitoring system: taking the loss-of-voltage alarm as an auxiliary information source, and finally determining a fault section, The node state expected value and the overcurrent alarm information coding both comprise 1, -1 and 0, which respectively correspond to existence of a positive, negative direction or undetected fault current of the node, The fault diagnosis target function is based on the following formula: where ω1, ω2, ω3 are weight coefficients, m k is a correction parameter, m k = 1 if the expected value of the node state variable after considering the false alarm is greater than the overcurrent alarm information code of the node, otherwise m k = 0; n k is a correction parameter, n k = 1 if the expected value of the node state variable after considering the false alarm is less than the overcurrent alarm information code of the node, otherwise n k = 0; s i is used to represent whether the i-th section is faulty, s i = 0 means no fault, s i = 1 means there is a fault, m k , n k , s i are all of Boolean type, I j is the overcurrent alarm information code uploaded at the node j after the fault, I j ' is the expected value of the node state variable after considering the false alarm, R is the number of sections, D is the number of switches, and y is the fault hypothesis sequence of the section.

2. The power grid fault diagnosis method of claim 1, wherein, The section in step 2 is simplified based on a region similarity principle, that is, when any section in a region respectively occurs a fault, all node state quantities calculated outside the region do not change, so all sections in the region can be equivalent to one section.

3. The power grid fault diagnosis method of claim 1, wherein, Optimizing the fault diagnosis target function by using the red-billed blue magpie optimization algorithm comprises the following steps: a) initializing based on the following formula: x p,q = (ub - lb) x Ranfl + lb In the formula, p represents the serial number of a single individual, q represents the serial number of the position dimension of the individual, x p,q represents the fault state of the section of the qth position dimension of the pth individual, ub and lb are the upper and lower bounds of the search space, respectively, and Ranfl represents a random number for generating a standard normal distribution; b) searching for food, in the process of searching for food, the red-billed blue magpie searches in a small group or in a crowd, searching for food in a small group based on the following formula: where t represents the current iteration number, X h (t+1) represents the hth new search agent position of the t+1th iteration, h is a serial number, X h (t) represents the hth new search agent position of the tth iteration, u represents the number of red-billed blue magpies randomly selected from all the red-billed blue magpies participating in the search, X w (t) represents the wth individual, X h represents the individual with serial number h, X rs (t) represents a search agent randomly selected in the current iteration, Ranf2 represents a random number for generating a standard normal distribution; searching for food in a crowd based on the following formula: In the formula, v represents a search agent number of a cluster when exploring food, r represents a total number of red-billed blue magpies participating in search of the cluster, and Ranf3 represents a random number for generating a standard normal distribution; c) attacking prey, the attack mathematical model corresponding to the small group action is: the attack mathematical model corresponding to the crowd action is: wherein X food (t) indicates the position of the food, the adjustment factor LP = (1 - t / T)^(2*(t / T)), T is the total number of iterations, Ranr1 and Ranr2 indicate random numbers for generating a standard normal distribution, d) storing food storing food is based on the following formula: wherein and respectively represent the fitness values of the hth Red-billed Blue Magpie before and after the position update.

4. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that, The processor implements the steps of the fault diagnosis method in any one of claims 1 to 3 when executing the computer program.

5. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the steps of the fault diagnosis method in any one of claims 1 to 3.

6. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to implement the steps of the fault diagnosis method in any one of claims 1 to 3.

Images