Power grid fault diagnosis method applying Red-beak-Blue-Tapie algorithm under multiple dimensionality reduction

Abstract

The invention discloses a power grid fault diagnosis method applying a Red-beak-blue-magpie algorithm under multiple dimensionality reduction, and the method comprises the steps: converting a high-voltage power grid into a topological structure, taking a power transmission line, a transformer and a bus as detection sections, taking circuit breakers at two ends of each element as nodes at two sides of the detection sections, and enabling the nodes to be the circuit breakers; step 2, node state expected values of two sides of each section under different fault states of each section are determined; and based on the over-current alarm information codes of the nodes at the two sides of each section and the node state expected values at the two sides of each section in different fault states, optimizing the fault diagnosis objective function by using a Red-beak-Baijie optimization algorithm, and obtaining the fault state of each section under the condition of minimizing the fault diagnosis objective function. And the corrected over-current alarm information codes of the nodes on the two sides of each section are obtained. According to the invention, the problems of low fault positioning accuracy and slow speed under the high-voltage power grid are solved.

Description

Technical Field

[0001] The invention relates to a power grid fault diagnosis method using a red-billed blue magpie algorithm under multiple dimensionality reduction in the field of power system relay protection. Background Art

[0002] As the scale of the power system gradually expands, the connection between power grids becomes closer, and the topological structure and operation mode of the high-voltage transmission network become increasingly complex. In the face of the fact that the traditional power grid fault diagnosis scheme cannot meet the requirements of increasingly complex power grids, the fault diagnosis accuracy is not high, the fault tolerance is low, it is difficult to meet the requirements of speed and reliability, and the fault diagnosis ability is still insufficient when the information is misreported. The present invention significantly improves the accuracy of high-voltage complex power grid fault location by reducing the dimension of reported data, simplifying the power grid topological structure, introducing the red-billed blue crane algorithm to optimize the positioning of the objective function, and assisting the positioning of the relay protection online detection system.

[0003] In order to improve the fault diagnosis effect, the information reported by the relay protection devices installed on each section switch or circuit breaker is widely used to judge the faulty components, but the relay protection devices are prone to false alarms. 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 ability, they have high requirements on the accuracy of the uploaded information. If information loss occurs, it will affect the power grid fault diagnosis effect. To address this problem, one method is to classify the fault information and extract appropriate feature quantities, and use the DS data fusion method to determine the fault probability of each component (Wang Yu, Xu Changbao, Zhu Jianyang, et al. Multi-source information fusion power grid fault diagnosis method based on improved Petri net and Hilbert transform [J]. Electrical Measurement and Instrumentation, 2021.), but if there is an error in the uploaded information under this method, the obtained diagnosis result may not be consistent with the actual situation. Another method is to rebuild the fault diagnosis model and locate the fault through multi-source information assisted 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 applicable to single fault conditions and has limited applicable scenarios. At present, most fault diagnosis models have too large variable dimensions, and the fault diagnosis models are complex and difficult to build. The applicable scenarios are single and have extremely high requirements for uploaded information. At the same time, they cannot be diagnosed in special false alarm scenarios, and there is a certain risk of false alarms. For the new method of power grid fault diagnosis studied in the patent of this invention, the variables and power grid topology in the model can be simplified, and the relay protection online monitoring system can be used to prevent misjudgment in special scenarios, thereby ensuring the accuracy of the fault diagnosis model. Summary of the invention

[0004] The purpose of the present invention is to provide a power grid fault diagnosis method using a red-billed blue magpie algorithm under multiple dimensionality reduction in response to the above-mentioned problems existing in the prior art.

[0005] The above-mentioned purpose of the present invention can be achieved by the following technical means:

[0006] A power grid fault diagnosis method using a red-billed blue magpie algorithm under multiple dimensionality reduction, comprising the following steps:

[0007] Step 1: Convert the high-voltage power grid into a topological structure, consider the transmission line, transformer and busbar as the detection section, and consider the circuit breakers at both ends of the components as nodes on both sides of the detection section. The nodes are circuit breakers.

[0008] Step 2: Determine the expected value of the node status on both sides of each section under different fault conditions of each section;

[0009] Step 3: Based on the overcurrent alarm information coding of the nodes on both sides of the section and the expected values ​​of the node states on both sides of each section under different fault conditions of each section, the red-billed blue magpie optimization algorithm is used to optimize the fault diagnosis objective function. 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.

[0010] As mentioned above, the node state expected value and the overcurrent alarm information encoding both include 1, -1 and 0, corresponding to the presence of positive, negative direction or no fault current detected in the node, respectively.

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

[0012]

[0013] In the formula, ω1, ω2, ω3 are weight coefficients, m k To correct the parameters, the expected value of the node state after the node misreporting is greater than the overcurrent alarm information encoding time m k =1, otherwise m k =0;n k To correct the parameters, the expected value of the node state after the node is considered to be less than the overcurrent alarm information encoding n k =1, otherwise n k =0;s i To indicate whether the i-th section is faulty, when s i =0 means no fault, s i =1 indicates a fault exists, m k 、n k 、s i All are Boolean type, I j is the overcurrent alarm information code uploaded at node j after the fault, I j' is the expected value of the node state after considering the misreport, R is the number of sections, D is the number of switches, and y is the fault hypothesis sequence of the section.

[0014] As mentioned above, the segments in step 2 are simplified based on the regional similarity principle, that is, when any segment in a certain area fails, all node state quantities calculated outside this area do not change, so all segments in the area can be equivalent to one segment.

[0015] As described above, the optimization of the fault diagnosis objective function using the red-billed blue magpie optimization algorithm includes the following steps:

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

[0017] x p,q =(ub-lb)×Ranf1+lb

[0018] In the formula, p represents the serial number of a single individual, q represents the serial number of the position dimension of the individual, represents the fault state of the segment of the qth position dimension of the pth individual, ub and lb are the upper and lower bounds of the search space respectively, and Ranf1 represents the random number used to generate the standard normal distribution.

[0019] b) Searching for food. In the process of searching for food, the red-billed blue magpie searches in small groups or flocks.

[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 in the t+1th iteration, h is the sequence number, X h (t) represents the hth new search agent position in 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 the randomly selected search agent in the current iteration, and Ranf2 represents the random number used to generate the standard normal distribution.

[0023] Search for food in groups 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 blue magpies participating in the search, and Ranf3 represents the random number used to generate the standard normal distribution;

[0026] c) attacking prey,

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

[0028]

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

[0030]

[0031] Where, X food (t) represents the location 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,

[0032] d) Food storage

[0033] Storing food is based on the following formula:

[0034]

[0035] In the formula, and They represent the fitness values ​​of the h-th red-billed blue magpie before and after the position update.

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

[0037] A computer-readable storage medium stores a computer program, which implements the steps of the above-mentioned fault diagnosis method when executed by a processor.

[0038] A computer program product comprises a computer program, which implements the steps of the above fault diagnosis method when executed by a processor.

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

[0040] (1) Compared with other methods, this method has a wider range of applications and can adapt to complex high-voltage power grid fault problems while avoiding misjudgment and meeting system reliability and speed requirements;

[0041] (2) The grid topology dimensionality reduction proposed by this method simplifies the positioning objective function and system topology, which is most conducive to the use of the optimization algorithm;

[0042] (3) This method not only determines the fault section, but also uses a relay protection online monitoring system to prevent fault diagnosis in special false alarm scenarios, thereby ensuring the accuracy of fault diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is a topological structure diagram of a high-voltage power grid, where T represents transformer, B represents bus, L represents line, S represents power supply, and the subscript numbers represent serial numbers;

[0044] Figure 2 is the equivalent topological structure of the power grid, where the numbers without brackets are node numbers, the numbers with brackets are segment numbers, and S4' is the equivalent power supply when the closed-loop network is disassembled into an open-loop network, which has the same meaning as S4;

[0045] Figure 3 It is a diagram of the signal transmission process;

[0046] Figure 4 It is a schematic diagram of the dimension reduction of the power grid topology; the numbers without brackets are the node numbers after the dimension reduction, and the numbers with brackets are the segment numbers after the dimension reduction; a, b, and c are segments with similar regions selected respectively;

[0047] Figure 5 is a hierarchical equivalent structure diagram; the numbers with superscripts are the node numbers after dimensionality reduction;

[0048] Figure 6 For regional similarity circuits;

[0049] Figure 7 It is a diagram of a power grid calculation model; the numbers without brackets are the node numbers in the calculation model, the numbers with brackets are the section numbers in the calculation model, and the numbers in the boxes represent the numbers of the smart terminal devices equipped in the corresponding sections in the model.

[0050] Figure 8 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0051] In order to facilitate those skilled in the art to understand and implement the present invention, the present invention is further described in detail below in conjunction with embodiments. The embodiments described herein are only used to illustrate and explain the present invention and are not intended to limit the present invention.

[0052] A power grid fault diagnosis method using a red-billed blue magpie algorithm under multiple dimensionality reduction, comprising the following steps:

[0053] Step 1: Convert the high-voltage power grid into a topological structure. Consider the transmission line, transformer and busbar as the detection section, and the circuit breakers at both ends of the components as the nodes on both sides of the detection section. Determine the expected value of the node state on both sides of each section under different fault conditions.

[0054] Equivalent high-voltage power grid topology, and perform information correction and one-time dimensionality reduction. The high-voltage power grid topology structure is as follows Figure 1 As shown in the figure, T represents transformer, B represents busbar, L represents line, S represents power supply, and both ends of each component are connected by circuit breakers. The diagnosis of power grid faults is mainly aimed at the diagnosis of the above-mentioned component faults in the power grid. The circuit breakers at both ends of each component will be activated when a fault occurs to realize the main and backup protection. Figure 1 From the grid structure shown, it can be seen that the variable dimension in the traditional fault diagnosis model is extremely large, and the circuit breaker will be affected by multiple protections, making the protection state quantity many times larger than the circuit breaker state quantity, which is not conducive to solving the fitness function based on the protection and circuit breaker status. In addition, the protection equipment is usually installed with a redundant system to reduce the probability of distortion of the protection quantity. At the same time, the protection action directly corresponds to the circuit breaker action, and the circuit breaker tripping depends on the fault state of the corresponding component in the power grid. Therefore, the tripping information of the circuit breaker uploaded by the relay protection device can be directly used for power grid fault diagnosis. It is fault diagnosis, not line protection. It is to diagnose according to the information of each node after the fault occurs and the protection is activated, and the actual faulty component is judged to prevent misjudgment caused by protection refusal, false operation, omission, false alarm, etc. In this way, the high-voltage topology structure can be equivalent, and the three components of the transmission line, transformer and busbar are regarded as the detection section, and the circuit breakers at both ends of the components are regarded as nodes on both sides of the detection section. The equivalent topology structure diagram of the high-voltage power grid is as follows Figure 2 As shown, S is still the system power supply, S4 and S4' are the same power supply, the black dots are nodes, i.e., circuit breakers, and the line segments between the dots are equivalent detection sections of the three components of the transmission line, transformer, and busbar.

[0055] Each circuit breaker in the power grid is controlled by a relay protection device. The relay protection device is often used to monitor the current fault information flowing through the circuit breaker when the circuit breaker does not operate after a fault occurs and report it. When the relay protection device detects that a fault current flows through a node (circuit breaker), the overcurrent alarm information I is sent to the relay protection device according to the direction of the fault current. j It is encoded as 1, -1 and 0 and then reported. When a node has a positive or negative direction or no fault current is detected, the reported fault current information is encoded as 1, -1 and 0. However, since the relay protection device is installed outdoors, it is affected by the environment, and the uploaded switch information may be missed or falsely reported. In addition, since fault diagnosis in complex power grids mainly relies on switch quantity, the complex ring network system can be equivalent to an open-loop structure. Therefore, the fault section identification problem is converted into an objective function optimization problem. However, after considering the scenarios of missed or false reports of uploaded switch information, the variable dimension in the objective function will increase significantly, which seriously affects the algorithm optimization. Therefore, it is considered to correct and reduce the switch information reported by the relay protection device.

[0056] The information transmission process of the relay protection acquisition device is to obtain the actual state of the node after the expected value of the node is corrected. The multiple groups of section fault information assumed in the optimization algorithm in step three are used to calculate the expected value of each node 1, 0 or -1 through the section fault situation of each group. The actual state of each node 1, 0 or -1 is determined by the information uploaded by the relay protection device, and then the expected value and the actual state are substituted into the objective function of formula (1) to solve. According to the solution result, the optimization algorithm in step three automatically updates the new section assumption situation, and calculates the expected value again according to the assumption, and repeats this cycle until the maximum number of iterations is reached. The transmission process is as follows Figure 3 As shown in the figure, the right side is the node state quantity actually reported by the relay protection device. The three actual node state quantities uploaded by the relay protection device divide the information transmission into three cases. In each case, the left side is the three possible results of the expected value calculation of each node: 1, -1 and 0. If the calculation result is different from the actual state quantity, it is a false alarm. When transmitting information, false alarms and missed alarms are unified as false alarms. According to the actual node state quantity, there is only one correct reporting situation. As shown in Figure (3), the correction process of the correct report is marked as z k , and then correct the misreporting: the correction process for the node expected value calculation result with the larger value is marked as m k , the correction process with the smaller expected value is marked as n k , when m k or k =1, it means that the node has misreported information. Thus, the reported information is corrected and simplified to obtain the fault diagnosis objective function minFit(y):

[0057]

[0058] In the formula, ω1, ω2, ω3 are weight coefficients, m k To correct the parameters, the expected value of the node state after the node misreporting is greater than the overcurrent alarm information encoding time m k =1, otherwise m k =0;n k To correct the parameters, the expected value of the node state after the node is considered to be less than the overcurrent alarm information encoding n k =1, otherwise n k =0;m k or k =1, indicating that the node has an information error. i To indicate whether the i-th section is faulty, when s i =0 means no fault, s i =1 indicates a fault exists, m k 、n k 、s i All are Boolean types, |I j -I' j| is the main judgment basis for power grid fault diagnosis, I j is the overcurrent alarm information code uploaded at node j after the fault, I j ' is the expected value of the node state after considering the misreporting, |m k +n k |It is an auxiliary criterion for power grid fault diagnosis. is the number of faulty sections, R is the number of sections, D is the number of switches, y is the fault hypothesis sequence of the section, and j, i, k are all serial number parameters.

[0059]

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

[0061] Step 2: Determine the expected value of the node status on both sides of each section under different fault conditions of each section;

[0062] The busbar, circuit breaker and line in the line are equivalent to sections. The fault status of each section is taken as the object to be processed. Combined with the overcurrent information reported by each node, the result is the fault condition of each section. If the corresponding section is 1, a fault occurs. The fault status is taken as the object to be processed, and the regional equivalence principle is used to optimize the power grid topology.

[0063] Will Figure 2 In the high-voltage power grid equivalent topology diagram shown, each branch is equivalent to an area (a), (b) or (c). The topology dimension reduction diagram is shown in Figure 4 As shown, each region contains several segments, and the switching function vector I calculated by formula (2) * (X) It can be found that when a fault occurs in any section of a region, the state quantities of all nodes calculated outside this region do not change. Therefore, all sections in the region can be equivalent to one section, so that one region is composed of only one section and the nodes on both sides of the section. That is, the "regional similarity" principle is adopted to greatly simplify the number of nodes, and the power grid topology structure is optimized into two layers based on this principle. Figure 5 The first layer of the grid equivalent model is optimized to locate the fault area, and then the topology outside the fault area is equivalent to the power supply. Taking the fault in area a as an example, the second layer is obtained after equivalent. Figure 6 The equivalent circuit of the area shown is used to locate the fault section inside area a. Finally, the principle is used to simplify other areas and optimize the complex power grid system structure as a whole.

[0064] Step 3: Based on the overcurrent alarm information coding of the nodes on both sides of the section and the expected values ​​of the node states on both sides of each section under different fault conditions of each section, the red-billed blue magpie optimization algorithm is used to optimize the fault diagnosis objective function. 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 dual optimization of the variable dimension and structural topology of the complex power grid, the variable dimension in the fault diagnosis model is reduced. The objective function of fault diagnosis is shown in the above formula (1). In the loop process mentioned in step 1, the red-billed blue magpie algorithm will update the section hypothesis so that the objective function Fit(y) gradually becomes smaller. The fault location problem can be equivalent to the problem of optimizing the minimum value of the objective function. The section hypothesis corresponding to the minimum objective function value is the predicted section fault location result.

[0066] The algorithm is as follows:

[0067] a. Initialization. Initialization includes population initialization and parameter initialization. Population initialization is to randomly generate a set of solutions (called population), each solution corresponds to the position of a red-billed blue magpie (that is, the fault hypothesis sequence composed of the fault status 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)×Ranf1+lb (3)

[0069] In the formula, p represents the serial number of a single individual, q represents the serial number of the individual's position dimension, and x p,q represents the fault state of the segment in 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 segment in the fault diagnosis model (1 means faulty), and the lower bound corresponds to the fault state of the segment in the fault diagnosis model (0 means no fault). Ranf1 represents the random number used to generate the standard normal distribution.

[0070] b. Searching for food. In the process of searching for food, red-billed blue magpies usually search in small groups or groups to obtain higher search efficiency. And they use various techniques to search for food resources. Their adaptability and flexibility enable them to adopt a variety of hunting strategies to ensure sufficient food supply. That is, simulating the process of red-billed blue magpies foraging, the position is updated through the formula. The position update formula usually includes two parts: global search and local search.

[0071] 1) Iterate as the group searches for food:

[0072]

[0073] In the formula, t represents the current iteration number, X h (t+1) represents the hth new search agent position in the t+1th iteration, h is the sequence number, X h (t) represents the hth new search agent position in the tth iteration, u represents the number of red-billed blue magpies in a small group of 2 to 5 randomly selected from all the red-billed blue magpies participating in the search, and 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 the random number used to generate the standard normal distribution.

[0074] 2) Iterate as the group searches for food:

[0075]

[0076] In the formula, v represents the number of search agents in the cluster when exploring food, which is between 10 and r, and r represents the total number of red-billed blue magpies participating in the search in the cluster. It is also randomly selected from the entire group, Ranf3 represents the random number used to generate the standard normal distribution, and the other parameters and functions have the same meaning as formula (4).

[0077] Searching for food is the process of position change of some individuals in the red-billed blue magpie algorithm. The purpose is to find the position direction that minimizes the objective function. Because the change composition does not directly lead to the optimal position, repeated iterations are required to find the optimal position. Parts 1) and 2) represent two ways of position change, which are automatically determined by the algorithm.

[0078] c. Attacking prey. Red-billed blue magpies have a high degree of hunting proficiency and cooperation when chasing prey. They take different actions for different targets, emphasizing the various strategies and skills they possess. When targeting small prey or plants, they act in small groups, using tactics such as fast pecking, jumping to catch prey, or flying to catch insects; when targeting large prey, they act in groups and successfully obtain food in various situations.

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

[0080]

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

[0082]

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

[0084] The purpose of attacking prey is to change the group position so that the overall position is closer to the optimal position. The position change process is part 1) and part 2, which are automatically judged and selected by the algorithm.

[0085] d. Food storage

[0086] In addition to searching for and attacking food, the red-billed blue magpie will store excess food in a hidden place for future consumption when food is scarce. This process retains the solution information and facilitates finding the global optimal solution. The mathematical model is:

[0087]

[0088] In the formula, and They respectively represent the fitness values ​​of the h-th red-billed blue magpie before and after the position is updated. The other parameters and functions have the same meanings as b.

[0089] Storing food means judging whether the current position change is reasonable. If the objective function is not optimized after the position change, the position before the change will be 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 be always maintained through the content of the dth paragraph.

[0090] Combined with the pressure loss alarm of the relay protection online monitoring system, auxiliary fault judgment is realized. When the relay protection device misreports and the misreport result corresponds to another section fault, the fault section may be misjudged. The relay protection online monitoring system can collect and process the power consumption information of each section in real time, and it is mostly installed indoors, and the reliability of the reported information is high. When a component in the power grid fails, the protection device operates to cause the fault section to lose pressure, and the collection system reports the alarm information due to the pressure loss. At this time, the control center will get all the alarm information. However, there are many power supplies in the power grid, so one of the power supplies is used as the main power supply, and the rest of the power supplies are regarded as auxiliary power supplies. After the fault section loses pressure alarm, considering that there are faults on the path connected to a section and all power supplies, the section will also lose pressure alarm due to no power supply. Therefore, the alarm information can be used as an auxiliary judgment information source to eliminate the misjudgment of the fault section caused by special false alarms, so as to finally determine the fault section.

[0091] Example: Comparison and verification analysis of fault location simulation under different fault types. Figure 7The model shown in the figure is used for fault location simulation verification, where S is the power supply and the feeder sections are (1) to (16). In order to verify the accuracy and effectiveness of the new method proposed in this paper, another document considering the underreporting and false alarm of the relay protection device, without reducing the dimension of the variables and without using the fault auxiliary judgment method and the patent method are used to compare the accuracy of fault section location under single and double fault types. Under a single fault, sections (2) and (4) are set to fail respectively, and different nodes are set to have false alarms respectively. The test results are shown in Tables 1 and 2; under a double fault, sections (2) and (11) are set to fail, and different nodes are set to have false alarms respectively. The test results are shown in Table 3.

[0092] Table 1 Section (2) Faults

[0093]

[0094] Table 2 Faults in section (4)

[0095]

[0096]

[0097] In the case of a single fault, section (2) or area (4) fails. Taking area (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 alarm at any node, the current information received by the system at each node is [I1,K,I 20 ]=[1,1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1];When different nodes are misreported, the grid is hierarchically optimized according to the regional similarity principle proposed in the present invention, so that the variable dimension is reduced by nearly half. Then, the fault diagnosis and location model of formula (1) can be used to determine that a fault has occurred in area (1), and then equivalent calculation is performed on area (1). At the same time, the red-billed blue crane algorithm used in the present invention is used to locate the fault section (2). Finally, the fault location auxiliary judgment is combined to eliminate misjudgment and determine the fault section (2). However, it can be seen from Table 1 that when the non-dimensionality reduction variable is not located using the fault auxiliary judgment method proposed in another document, there is an inaccurate fault location, and misjudgment in special circumstances is not taken into account.

[0098] Table 3 Faults in areas (2) and (11)

[0099]

[0100] Under double fault conditions, section (2) and area (11) fail. The overcurrent alarm information collected by the online monitoring system of relay protection is coded as [0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0]. When there is no false alarm at any node, the current information received by the system at each node is [I1,K,I 20 ]=[1,1,-1,-1,-1,-1,-1,-1,-1,-1,0,0,-1,-1,-1,-1,-1,-1,-1,-1,-1], when false alarms occur at different nodes, the steps of the method for determining the fault area according to the patent of the present invention are similar to those for a single fault, and the faulty sections (2) and (11) are finally located. As can be seen from Table 3, in the method without fault location auxiliary judgment, when nodes 11 and 13 are misreported, the fault diagnosis model directly judges that sections (10) and (12) are faulty, which leads to misjudgment in special scenarios; however, according to the patent of the present invention, combined with the decompression alarm information, if (10) and (12) fail, section (2) should have no decompression alarm signal, while section (12) should have an alarm signal, which is inconsistent with the actual decompression alarm signal. Therefore, it can be judged that the nodes near sections (10) and (12) are misreported. Combined with the location 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 proposed in the patent of this invention is more accurate in positioning than the diagnostic model without variable dimension reduction, and its applicable scenarios are not limited. It is applicable to single or double faults. The multi-source information assisted positioning used can prevent misjudgment in special false alarm scenarios and plays a good role. The positioning results show that the method has good stability and applicability.

[0102] In one embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, 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, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0104] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0105] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing related hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods.

[0106] It should be noted that the embodiments described in the present invention are merely examples of the spirit of the present invention. Those skilled in the art may make various modifications or additions to the described embodiments or replace them in similar ways, but they will not deviate from the spirit of the present invention or exceed the scope defined by the attached claims.

Claims

1. A power grid fault diagnosis method using the red-billed blue magpie algorithm under multiple dimensionality reduction, characterized in that: The following steps are involved: Step 1: Convert the high-voltage power grid into a topological structure, consider the transmission line, transformer and busbar as the detection section, and consider the circuit breakers at both ends of the components as nodes on both sides of the detection section. The nodes are circuit breakers. Step 2: Determine the expected value of the node status on both sides of each section under different fault conditions of each section; Step 3: Based on the overcurrent alarm information coding of the nodes on both sides of the section and the expected values ​​of the node states on both sides of each section under different fault conditions of each section, the red-billed blue magpie optimization algorithm is used to optimize the fault diagnosis objective function. 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.

2. According to the method of power grid fault diagnosis using the red-billed blue magpie algorithm under multiple dimensionality reduction described in claim 1, it is characterized in that: The node state expected value and overcurrent alarm information encoding both include 1, -1 and 0, corresponding to the presence of positive, negative direction or no fault current detected in the node, respectively.

3. According to the method of power grid fault diagnosis using the red-billed blue magpie algorithm under multiple dimensionality reduction described in claim 1, it is characterized in that: The fault diagnosis objective function is based on the following formula: In the formula, ω1, ω2, ω3 are weight coefficients, m k To correct the parameters, the expected value of the node state after the node misreporting is greater than the overcurrent alarm information encoding time m k =1, otherwise m k =0;n k To correct the parameters, the expected value of the node state after the node is considered to be less than the overcurrent alarm information encoding n k =1, otherwise n k =0;s i To indicate whether the i-th section is faulty, when s i =0 means no fault, s i =1 indicates a fault exists, m k 、n k 、s i All are Boolean type, I j is the overcurrent alarm information code uploaded at node j after the fault, I j ' is the expected value of the node state after considering the misreport, R is the number of sections, D is the number of switches, and y is the fault hypothesis sequence of the section.

4. According to the method of power grid fault diagnosis using the red-billed blue magpie algorithm under multiple dimensionality reduction described in claim 1, it is characterized in that: The segments in step 2 are simplified based on the regional similarity principle, that is, when any segment in a certain area fails, all node state quantities calculated outside this area do not change, so all segments in the area can be equivalent to one segment.

5. According to the method of power grid fault diagnosis using the red-billed blue magpie algorithm under multiple dimensionality reduction described in claim 1, it is characterized in that: The method of optimizing the fault diagnosis objective function using the red-billed blue magpie optimization algorithm comprises the following steps: a) Initialize based on the following formula: x p,q =(ub-lb)×Ranf1+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, represents the fault state of the segment of the qth position dimension of the pth individual, ub and lb are the upper and lower bounds of the search space respectively, and Ranf1 represents the random number used to generate the standard normal distribution. b) Searching for food. In the process of searching for food, the red-billed blue magpie searches in small groups or flocks. Search for food in small groups based on the following formula: In the formula, t represents the current iteration number, X h (t+1) represents the hth new search agent position in the t+1th iteration, h is the sequence number, X h (t) represents the hth new search agent position in 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 the randomly selected search agent in the current iteration, and Ranf2 represents the random number used to generate the standard normal distribution. Search for food in groups based on the following formula: In the formula, v represents the number of search agents in the cluster when exploring food, r represents the total number of red-billed blue magpies participating in the search, and Ranf3 represents the random number used to generate the standard normal distribution; c) attacking prey, The attack mathematical model corresponding to the small group action is: The attack mathematical model corresponding to group action is: Where, X food (t) represents the location 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, d) Food storage Storing food is based on the following formula: In the formula, and They respectively represent the fitness value of the h-th red-billed blue magpie before and after the position is updated.

6. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the fault diagnosis method according to any one of claims 1 to 4 are implemented.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the fault diagnosis method according to any one of claims 1 to 4 are implemented.

8. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the fault diagnosis method according to any one of claims 1 to 4 are implemented.

Images