Power grid fault diagnosis and accident handling auxiliary system

Abstract

The invention discloses a power grid fault diagnosis and accident processing auxiliary system which comprises a power grid intelligent monitoring unit, an accident area positioning unit, a fault type diagnosis and analysis unit and an accident processing unit. Wherein the power grid intelligent monitoring unit comprises a data acquisition module, a data screening integration module, a data processing module and a data communication module; after receiving the fault data, the accident area positioning unit positions a power grid fault area based on a genetic algorithm; after the fault type diagnosis and analysis unit receives the fault data, the fault type is quickly judged based on a deep belief network, and a result is output; the accident processing unit comprises an accident recovery module and a power grid dispatching operation recovery module; according to the method, fault positioning and troubleshooting can be rapidly carried out after an accident occurs, power recovery is carried out, the influence of the power grid accident is reduced as much as possible, the system crash accident is effectively prevented, and the system reliability is improved.

Description

Technical Field

[0001] The invention belongs to the technical field of power grid accident processing, and in particular relates to a power grid fault diagnosis and accident processing auxiliary system. Background Art

[0002] The power grid is a real-time distribution network. Once an accident occurs, it is required to be able to quickly analyze the cause and location of the accident, and to eliminate the accident on the basis of isolating the faulty equipment in order to restore the stable operation of the power grid.

[0003] When a power system fails, the traditional accident analysis method is a combination of telephone reporting, on-site inspection, and data collection and monitoring system data. After the accident occurs, the substation or dispatcher reports the accident by phone and goes to the site to check the equipment failure. The local dispatch duty officer combines these report results with SCADA data to analyze the cause of the accident and finally handles it based on personal experience or previous accident handling mechanisms.

[0004] This method is not only slow and inefficient, affecting the stable operation of the power grid, but also as more and more power equipment is added to the power network, the operation and dispatching methods of the power grid are becoming more and more complicated. Simply relying on the experience of dispatchers can no longer effectively restore the operation of the power grid after an accident.

[0005] Therefore, in order to solve the above problems, it is necessary to develop a power grid fault diagnosis and accident handling auxiliary system. Summary of the invention

[0006] The purpose of the present invention is to overcome the deficiencies of the prior art and to provide a power grid fault diagnosis and accident handling auxiliary system, which can quickly locate and troubleshoot faults after an accident occurs, and restore power, thereby minimizing the impact of power grid accidents, effectively preventing system collapse accidents, and improving system reliability.

[0007] The object of the present invention is achieved as follows: A power grid fault diagnosis and accident handling auxiliary system includes a power grid intelligent monitoring unit, an accident area positioning unit, a fault type diagnosis and analysis unit and an accident handling unit;

[0008] The power grid intelligent monitoring unit includes a data acquisition module, a data screening and integration module, a data processing module and a data communication module. The data acquisition module collects data from multiple systems in real time to obtain multi-source real-time data as a basis for fault diagnosis; the data screening and integration module extracts and integrates fault data and non-fault data respectively to form a fault data set; the data processing module is used to process fault data, including classification and sorting of fault data, preprocessing of fault data and fusion of fault data; the data communication module is used to connect the processed multi-source real-time fault data to the accident area positioning unit and the fault type diagnosis and analysis unit;

[0009] After receiving the fault data, the accident area positioning unit locates the fault area of ​​the power grid based on the genetic algorithm; specifically, it includes determining a trunk line in the power grid system, and then dividing the entire power grid into a series of independent sub-areas based on the trunk line. After receiving the fault data, the fault is located in a specific area through the genetic algorithm, and the individual with the largest fitness function value is found in the population containing all the section information of a certain area, that is, the location of the fault point in all the feeder sections in the area;

[0010] After receiving the fault data, the fault type diagnosis and analysis unit quickly determines the fault type based on the deep belief network and outputs the result;

[0011] The accident processing unit includes an accident recovery module and a power grid dispatching operation recovery module. The accident recovery module obtains the optimal solution for power grid recovery based on the minimum spanning tree method, and then performs dispatching operation accident processing on accidents of different fault types through the power grid dispatching operation recovery module.

[0012] Furthermore, the data acquisition module of the power grid intelligent monitoring unit collects multi-source real-time data, specifically including real-time collection of power, voltage and current, transformer temperature, system frequency, circuit breaker opening and closing positions, and transformer tap position information in the power grid based on the remote terminal unit and the feeder terminal unit, as well as real-time collection of main data of the power system and dynamic monitoring data of the bus voltage phasor by a synchronous phase angle measurement unit based on the global positioning system, as well as recording the change process of the power system and the action behavior of the protection device under a clear time sequence through a fault recorder, and collecting transient data under fault conditions.

[0013] Furthermore, in the data processing module of the power grid intelligent monitoring unit, the fault data classification and organization includes unified synchronization of the time of multiple system data to ensure that the multiple system data are data at the same time point; the fault data preprocessing includes filtering, converting and combining the data; the fault data fusion includes based on multi-source data fusion technology, detecting, combining, correlation analyzing and evaluating the relevant data from different channels, different media, different modes, different expressions and different times in the order of acquisition time.

[0014] Furthermore, when the accident area positioning unit locates the power grid fault area based on the genetic algorithm, its switch function is defined as follows:

[0015] s(i)=,c j [1-(x i,k (1)∥x i,k (2)∥…∥x i,k (n1))+-

[0016] ·[x i,p (1)∥x i,p (2)∥…∥x i,p (p)]

[0017] -,c h [1-(x i,h (1)∥x i,h (2)∥…∥x i,h (n2))+-

[0018] ·[x i,q (1)∥x i,q (2)∥…∥x i,q (q)]

[0019] , where s(i) is the expected function of switch i in the circuit, ∥ is the logical operation or, c j 、c h is the switching factor of switch i, x i,k and x i,h represent the fault status of the upper and lower section feeders associated with switch i, respectively, and x i,k (n1) represents the fault state of the n1th feeder section in the upper half of switch i, x i,h (n2) represents the fault state of the n2th feeder section in the lower half of switch i, x i,p and x i,q They represent the fault status of the upper and lower section feeders associated with switch i, respectively;

[0020] The fitness function formula is defined as follows:

[0021]

[0022] In the formula, s j represents the actual variable of switch j, X j represents the total number of feeders in the upper half of switch j, Represents the regular term of the fitness function; perform dual transformation on the above fitness function formula:

[0023]

[0024] , where M is a custom integer, which is twice the total number of switches in the power grid system; then the current maximum value of the fitness function is obtained.

[0025] Furthermore, when the accident area locating unit locates the power grid fault area based on the genetic algorithm, the switch and line coding method is to define the flow direction on the switch during normal operation as the positive direction of the switch. When the current passing through the switch is a fault current, the state of the switch is judged according to the direction of the current. If the current passing through the switch is opposite to the specified positive direction, the switch state is -1.

[0026] Furthermore, when the accident area positioning unit locates the power grid fault area based on the genetic algorithm, the crossover probability and mutation probability functions are defined as follows:

[0027]

[0028] , where P cmax , P cmin and P mmax , P mmin are the maximum and minimum crossover and mutation probabilities, respectively, f is the fitness value of individuals in the population, and f max is the maximum fitness of the offspring individuals in the iterative process, f avg is the average fitness of the offspring individuals in the population during the iteration process.

[0029] Furthermore, the fault type diagnosis and analysis unit discriminates the fault type based on the deep belief network and outputs the result, and the specific steps are as follows: ① After normalizing the fault data, divide it into a training set and a test set at a ratio of 7:3, ② establish a deep belief network model with multiple hidden layers, and initialize the network model parameters, ③ use the unlabeled samples in the training set to pre-train the stacked restricted Boltzmann machine at the bottom of the deep belief network model layer by layer, ④ add a Softmax classifier to the top layer of the deep belief network model, encode the fault type state, and determine the number of nodes in the output layer, ⑤ use labeled samples and use the error back propagation algorithm to supervised fine-tune the entire network parameters of the pre-trained deep belief network model so that the network performance approaches the global optimal solution, ⑥ save the trained deep belief network model, and use the data samples in the test set to test its performance, and output the fault type identification result.

[0030] Furthermore, when the accident recovery module in the accident processing unit obtains the optimal solution for power grid recovery based on the minimum spanning tree method, the distribution lines of the power grid are equivalent to load nodes, and the switches are equivalent to the edges connected to the load nodes. After simplification, a power grid model that meets the definition of graph theory is obtained. Based on the power grid model, a topological graph that meets the requirements of graph theory is obtained, and all nodes and switches are uniformly numbered to obtain a data table of nodes and switches. Then, the objective function of the minimum spanning tree recovery is expressed as:

[0031]

[0032] Where i and j are the switch number and node number respectively, ΔP is the network loss, S i is the switch reliability change, G i is the switching loss, α, β, δ are weight coefficients; the constraint condition of the objective function is:

[0033] ① There should be no isolated nodes in the area where the power grid is to be restored. ② The power supply should not be overloaded when it is restored. If P is the limit load of the restored power supply, P i is the load of each node i, P j is the charge already carried, then it is described by the formula:

[0034]

[0035] Calculate the weights of each switch in the grid restoration area subgraph, arrange them from small to large, and select the switch with the smallest weight. The selected switch can connect all nodes, which can ensure that the entire network is radial. Then the selection at this time is the minimum spanning tree for grid restoration, which is the optimal solution of the mathematical model.

[0036] Furthermore, different types of faults in the power grid dispatching operation recovery module of the accident handling unit include voltage anomalies, 220kv tie line accidents, 500kv line high-voltage protection tripping, series ferromagnetic resonance equipment failures, and system out-of-step failures.

[0037] Furthermore, when an abnormal voltage fault occurs, if the voltage at the fault point exceeds the minimum operating voltage and critical voltage value specified by the provincial dispatching department, the dispatcher needs to start and stop the standby units, adjust the reactive power output of the generator, adjust the number of transformers in operation if the load permits, and adjust the grid access method; when a 220kv interconnection line accident occurs, the duty officer checks the equipment of the faulty and tripped line to ensure that power can be transmitted normally. The provincial dispatcher will force power transmission by correctly selecting the forced transmission end. If the forced transmission is successful, the circuit breaker on the opposite side will be paralleled or closed, otherwise the line will be transferred for maintenance; when an accident occurs, the operator will check the equipment of the faulty and tripped line to ensure that power can be transmitted normally. When the 500kv line high-voltage reactor protection trips, handle it as a transformer accident. If the 500kv line protection and high-voltage reactor protection trip at the same time, handle it as a simultaneous high-voltage reactor and line failure. When a series ferromagnetic resonance equipment failure occurs, the operator on duty needs to immediately disconnect / close the corresponding equipment switch and eliminate the resonance. If the resonance occurs because the switch is closed, the switch needs to be disconnected immediately, otherwise it needs to be reported to the dispatcher. When a system out-of-step failure occurs, the operating system is decoupled according to the out-of-step section to eliminate the out-of-step state.

[0038] Due to the adoption of the above technical scheme, the beneficial effects of the present invention are as follows: through the power grid intelligent monitoring unit, multi-source real-time fault data of the power grid is collected, screened, processed and integrated to form fault data that can be used for fault diagnosis and accident area positioning, and the area and fault type where the fault occurs are quickly located through the accident area positioning unit and the fault type diagnosis and analysis unit, which is convenient for staff to promptly discover and understand the fault condition. When a fault occurs, the accident processing unit provides the optimal solution for power grid recovery and dispatch operations for different fault types, which can effectively improve the processing efficiency after the accident, quickly restore power, and improve the stability of system operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a principle block diagram of the present invention.

[0040] Figure 2 It is a flow chart of the fault type diagnosis and analysis unit in the present invention. DETAILED DESCRIPTION

[0041] The technical solution of the present invention is further specifically described below through embodiments and in conjunction with the accompanying drawings.

[0042] like Figure 1 , Figure 2As shown, a power grid fault diagnosis and accident handling auxiliary system includes a power grid intelligent monitoring unit, an accident area positioning unit, a fault type diagnosis and analysis unit and an accident handling unit.

[0043] Among them, the power grid intelligent monitoring unit includes a data acquisition module, a data screening and integration module, a data processing module and a data communication module. The data acquisition module collects data from multiple systems in real time to obtain multi-source real-time data as a basis for fault diagnosis; the data screening and integration module extracts and integrates fault data and non-fault data respectively, and forms a fault data set; the data processing module is used to process fault data, including classification and organization of fault data, preprocessing of fault data and fusion of fault data; the data communication module is used to connect the processed multi-source real-time fault data to the accident area positioning unit and the fault type diagnosis and analysis unit.

[0044] Preferably, the data acquisition module of the power grid intelligent monitoring unit collects multi-source real-time data, specifically including real-time collection of power, voltage and current, transformer temperature, system frequency, circuit breaker opening and closing positions, and transformer tap position information in the power grid based on the remote terminal unit and the feeder terminal unit, as well as real-time collection of main data of the power system and dynamic monitoring data of the bus voltage phasor by a synchronous phase angle measurement unit based on the global positioning system, and collection of transient data under fault conditions by recording the changing process of the power system and the action behavior of the protection device under a clear time sequence through a fault recorder.

[0045] Preferably, in the data processing module of the power grid intelligent monitoring unit, the fault data classification and organization includes unified synchronization of the time of multiple system data to ensure that the multiple system data are data at the same time point; the fault data preprocessing includes filtering, converting and combining the data; the fault data fusion includes based on multi-source data fusion technology, detecting, combining, correlation analyzing and evaluating the relevant data from different channels, different media, different modes, different expressions and different times in the order of acquisition time.

[0046] Among them, after receiving the fault data, the accident area positioning unit locates the fault area of ​​the power grid based on the genetic algorithm; specifically, it includes determining a trunk line in the power grid system, and then dividing the entire power grid into a series of independent sub-areas based on this trunk line. When the fault data is received, the fault is located in a specific area through a genetic algorithm, and the individual with the largest fitness function value is found in the population containing all the section information of a certain area, that is, the location of the fault point in all the feeder sections in the area.

[0047] Preferably, when the accident area positioning unit locates the power grid fault area based on the genetic algorithm, its switch function is defined as follows:

[0048] s(i)=,c j [1-(x i,k (1)∥x i,k (2)∥…∥x i,k (n1))+-

[0049] ·[x i,p (1)∥x i,p (2)∥…∥x i,p (p)]

[0050] -,c h [1-(x i,h (1)∥x i,h (2)∥…∥x i,h (n2))+-

[0051] ·[x i,q (1)∥x i,q (2)∥…∥x i,q (q)]

[0052] , where s(i) is the expected function of switch i in the circuit, ∥ is the logical operation or, c j 、c h is the switching factor of switch i, x i,k and x i,h represent the fault status of the upper and lower section feeders associated with switch i, respectively, and x i,k (n1) represents the fault state of the n1th feeder section in the upper half of switch i, x i,h (n2) represents the fault state of the n2th feeder section in the lower half of switch i, x i,p and x i,q They represent the fault status of the upper and lower section feeders associated with switch i, respectively;

[0053] The fitness function formula is defined as follows:

[0054]

[0055] In the formula, s j represents the actual variable of switch j, X j represents the total number of feeders in the upper half of switch j, Represents the regular term of the fitness function; perform dual transformation on the above fitness function formula:

[0056]

[0057] Where M is a custom integer, which is twice the total number of switches in the power grid system; the current maximum value of the fitness function is obtained.

[0058] Preferably, when the accident area locating unit locates the power grid fault area based on the genetic algorithm, the switch and line coding method is to define the flow direction of the switch on the switch during normal operation as the positive direction of the switch. When the current passing through the switch is a fault current, the state of the switch is judged according to the direction of the current. If the current passing through the switch is opposite to the specified positive direction, the switch state is -1.

[0059] Preferably, when the accident area positioning unit locates the power grid fault area based on the genetic algorithm, the crossover probability and mutation probability functions are defined as follows:

[0060]

[0061]

[0062] Where P cmax , P cmin and P mmax , P mmin are the maximum and minimum crossover and mutation probabilities, respectively, f is the fitness value of individuals in the population, and f max is the maximum fitness of the offspring individuals in the iterative process, f avg is the average fitness of the offspring individuals in the population during the iteration process.

[0063] Among them, after receiving the fault data, the fault type diagnosis and analysis unit quickly distinguishes the fault type based on the deep belief network and outputs the result; preferably, the fault type diagnosis and analysis unit distinguishes the fault type based on the deep belief network and outputs the result, and the specific steps are as follows: ① After normalizing the fault data, divide it into a training set and a test set at a ratio of 7:3, ② establish a deep belief network model with multiple hidden layers, and initialize the network model parameters, ③ use the unlabeled samples in the training set to pre-train the stacked restricted Boltzmann machine at the bottom of the deep belief network model layer by layer, ④ add a Softmax classifier to the top layer of the deep belief network model, encode the fault type state, and determine the number of nodes in the output layer, ⑤ use labeled samples and use the error back propagation algorithm to supervised fine-tune the entire network parameters of the pre-trained deep belief network model, so that the network performance approaches the global optimal solution, ⑥ save the trained deep belief network model, and use the data samples in the test set to test its performance, and output the fault type identification result.

[0064] Among them, the accident processing unit includes an accident recovery module and a power grid dispatching operation recovery module. The accident recovery module obtains the optimal solution for power grid recovery based on the minimum spanning tree method, and then performs dispatching operation accident processing on accidents of different fault types through the power grid dispatching operation recovery module.

[0065] Preferably, when the accident recovery module in the accident processing unit obtains the optimal solution for power grid recovery based on the minimum spanning tree method, the distribution lines of the power grid are equivalent to load nodes, and the switches are equivalent to edges connected to the load nodes. After simplification, a power grid model that meets the definition of graph theory is obtained. Based on the power grid model, a topological graph that meets the requirements of graph theory is obtained, and all nodes and switches are uniformly numbered to obtain a data table of nodes and switches. Then, the objective function of the minimum spanning fault tree recovery is expressed as:

[0066]

[0067] Where i and j are the switch number and node number respectively, ΔP is the network loss, S i is the switch reliability change, G i is the switching loss, α, β, δ are weight coefficients; the constraint condition of the objective function is:

[0068] ① There should be no isolated nodes in the area where the power grid is to be restored. ② The power supply should not be overloaded when it is restored. If P is the limit load of the restored power supply, P i is the load of each node i, P j is the charge already carried, then it is described by the formula:

[0069]

[0070] Calculate the weights of each switch in the grid restoration area subgraph, arrange them from small to large, and select the switch with the smallest weight. The selected switch can connect all nodes, which can ensure that the entire network is radial. Then the selection at this time is the minimum spanning tree for grid restoration, which is the optimal solution of the mathematical model.

[0071] Preferably, different types of faults in the power grid dispatching operation recovery module of the accident handling unit include voltage anomalies, 220kv tie line accidents, 500kv line high-voltage protection tripping, series ferromagnetic resonance equipment failures, and system out-of-step failures.

[0072] When an abnormal voltage fault occurs, the voltage at the fault point exceeds the minimum operating voltage and critical voltage value specified by the provincial dispatching department. The dispatcher needs to start and stop the standby unit, adjust the reactive power output of the generator, adjust the number of transformers in operation if the load allows, and adjust the grid access method; when a 220kv interconnection line accident occurs, the duty officer checks the equipment of the faulty tripped line to ensure that power can be transmitted normally. The provincial dispatcher will force power transmission by correctly selecting the forced transmission end. If the forced transmission is successful, the circuit breaker on the opposite side will be paralleled or closed. Otherwise, the line will be transferred for maintenance; when a 5 When the high-voltage reactor protection of a 100kv line trips, handle it as a transformer accident. If the 500kv line protection and the high-voltage reactor protection trip at the same time, handle it as a simultaneous failure of the high-voltage reactor and the line. When a series ferromagnetic resonance device fails, the operator on duty needs to immediately disconnect / close the corresponding equipment switch and eliminate the resonance. If the resonance occurs because the switch is closed, the switch needs to be disconnected immediately, otherwise it needs to be reported to the dispatcher. When a system out-of-step failure occurs, the operating system should be decoupled according to the out-of-step section to eliminate the out-of-step state.

[0073] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the specific implementation modes of the present invention can still be modified or replaced by equivalents, and any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be included in the scope of the claims of the present invention.

Claims

1. A power grid fault diagnosis and accident handling auxiliary system, characterized in that: It includes a power grid intelligent monitoring unit, an accident area positioning unit, a fault type diagnosis and analysis unit, and an accident handling unit; The power grid intelligent monitoring unit includes a data acquisition module, a data screening and integration module, a data processing module and a data communication module. The data acquisition module collects data from multiple systems in real time to obtain multi-source real-time data as a basis for fault diagnosis; the data screening and integration module extracts and integrates fault data and non-fault data respectively to form a fault data set; the data processing module is used to process fault data, including classification and sorting of fault data, preprocessing of fault data and fusion of fault data; the data communication module is used to connect the processed multi-source real-time fault data to the accident area positioning unit and the fault type diagnosis and analysis unit; After receiving the fault data, the accident area positioning unit locates the fault area of ​​the power grid based on the genetic algorithm; specifically, it includes determining a trunk line in the power grid system, and then dividing the entire power grid into a series of independent sub-areas based on the trunk line. After receiving the fault data, the fault is located in a specific area through the genetic algorithm, and the individual with the largest fitness function value is found in the population containing all the section information of a certain area, that is, the location of the fault point in all the feeder sections in the area; After receiving the fault data, the fault type diagnosis and analysis unit quickly determines the fault type based on the deep belief network and outputs the result; The accident processing unit includes an accident recovery module and a power grid dispatching operation recovery module. The accident recovery module obtains the optimal solution for power grid recovery based on the minimum spanning tree method, and then performs dispatching operation accident processing on accidents of different fault types through the power grid dispatching operation recovery module.

2. A power grid fault diagnosis and accident handling auxiliary system according to claim 1, characterized in that: The data acquisition module of the power grid intelligent monitoring unit collects multi-source real-time data, specifically including real-time collection of power, voltage and current, transformer temperature, system frequency, circuit breaker opening and closing position and transformer tap position information in the power grid based on the remote terminal unit and the feeder terminal unit, as well as real-time collection of main data of the power system and dynamic monitoring data of the bus voltage phasor by the synchronous phase angle measurement unit based on the global positioning system, and collection of transient data under fault conditions by recording the change process of the power system and the action behavior of the protection device under a clear time sequence through a fault recorder.

3. A power grid fault diagnosis and accident handling auxiliary system according to claim 1, characterized in that: In the data processing module of the power grid intelligent monitoring unit, the fault data classification and organization includes unified synchronization of the time of multiple system data to ensure that the multiple system data are data at the same time point; the fault data preprocessing includes filtering, converting and combining the data; the fault data fusion includes based on multi-source data fusion technology, detecting, combining, correlation analyzing and evaluating the relevant data from different channels, different media, different modes, different expressions and different times in the order of acquisition time.

4. A power grid fault diagnosis and accident handling auxiliary system according to claim 1, characterized in that: When the accident area positioning unit locates the power grid fault area based on the genetic algorithm, its switch function is defined as follows: s(i)={c j [1-(x i,k (1)||x i,k (2)||…||x i,k (n1))]}·[x i,p (1)||x i,p (2)||…||x i,p (p)]-{c h [1-(x i,h (1)||x i,h (2)||…|x i,h (n2))]}·[x i,q (1)||x i,q (2)||…||x i,q (q)] Where s(i) is the expected function of switch i in the circuit, || is the logical operation or, c j 、c h is the switching factor of switch i, x i,k and x i,h represent the fault status of the upper and lower section feeders associated with switch i, respectively, and x i,k (n1) represents the fault state of the n1th feeder section in the upper half of switch i, x i,h (n2) represents the fault state of the n2th feeder section in the lower half of switch i, x i,p and x i,q They represent the fault status of the upper and lower section feeders associated with switch i, respectively; The fitness function formula is defined as follows: In the formula, s j represents the actual variable of switch j, X j represents the total number of feeders in the upper half of switch j, Represents the regular term of the fitness function; perform dual transformation on the above fitness function formula: Where M is a custom integer, which is twice the total number of switches in the power grid system; the current maximum value of the fitness function is obtained.

5. A power grid fault diagnosis and accident handling auxiliary system according to claim 1, characterized in that: When the accident area positioning unit locates the power grid fault area based on the genetic algorithm, the switch and line coding method is to define the flow direction of the switch on the switch during normal operation as the positive direction of the switch. When the current passing through the switch is a fault current, the state of the switch is judged according to the direction of the current. If the current passing through the switch is opposite to the specified positive direction, the switch state is -1.

6. A power grid fault diagnosis and accident handling auxiliary system according to claim 1, characterized in that: When the accident area positioning unit locates the power grid fault area based on the genetic algorithm, the crossover probability and mutation probability functions are defined as follows: Where P cmax , P cmin and P mmax , P mmin are the maximum and minimum crossover and mutation probabilities, respectively, f is the fitness value of individuals in the population, and f max is the maximum fitness of the offspring individuals in the iterative process, f avg is the average fitness of the offspring individuals in the population during the iteration process.

7. A power grid fault diagnosis and accident handling auxiliary system according to claim 1, characterized in that: The fault type diagnosis and analysis unit discriminates the fault type based on the deep belief network and outputs the result, and the specific steps are as follows: ① After normalizing the fault data, divide it into a training set and a test set at a ratio of 7:3; ② establish a deep belief network model with multiple hidden layers and initialize the network model parameters; ③ use the unlabeled samples in the training set to pre-train the stacked restricted Boltzmann machine at the bottom of the deep belief network model layer by layer; ④ add a Softmax classifier to the top layer of the deep belief network model to encode the fault type state and determine the number of nodes in the output layer; ⑤ use the labeled samples and use the error back propagation algorithm to supervised fine-tune the entire network parameters of the pre-trained deep belief network model so that the network performance approaches the global optimal solution; ⑥ save the trained deep belief network model, test its performance using the data samples in the test set, and output the fault type identification result.

8. A power grid fault diagnosis and accident handling auxiliary system according to claim 1, characterized in that: When the accident recovery module in the accident processing unit obtains the optimal solution for power grid recovery based on the minimum spanning tree method, the distribution lines of the power grid are equivalent to load nodes, and the switches are equivalent to the edges connected to the load nodes. After simplification, a power grid model that meets the definition of graph theory is obtained. Based on the power grid model, a topological graph that meets the requirements of graph theory is obtained, and all nodes and switches are uniformly numbered to obtain a data table of nodes and switches. Then, the objective function of the minimum spanning fault tree recovery is expressed as: Where i and j are the switch number and node number respectively, ΔP is the network loss, S i is the switch reliability change, G i is the switching loss, α, β, δ are weight coefficients; the constraint condition of the objective function is: ① There should be no isolated nodes in the area where the power grid is to be restored. ② The power supply should not be overloaded when it is restored. If P is the limit load of the restored power supply, P i is the load of each node i, P j is the charge already carried, then it is described by the formula: Calculate the weights of each switch in the grid restoration area subgraph, arrange them from small to large, and select the switch with the smallest weight. The selected switch can connect all nodes, which can ensure that the entire network is radial. Then the selection at this time is the minimum spanning tree for grid restoration, which is the optimal solution of the mathematical model.

9. A power grid fault diagnosis and accident handling auxiliary system according to claim 1, characterized in that: Different types of faults in the power grid dispatching operation recovery module of the accident handling unit include voltage anomalies, 220kv tie line accidents, 500kv line high-voltage protection tripping, series ferromagnetic resonance equipment failures, and system out-of-step failures.

10. A power grid fault diagnosis and accident handling auxiliary system according to claim 9, characterized in that: When an abnormal voltage fault occurs, if the voltage at the fault point exceeds the minimum operating voltage and critical voltage value specified by the provincial dispatching department, the dispatcher needs to start and stop the standby units, adjust the reactive power output of the generator, adjust the number of transformers in operation if the load allows, and adjust the grid access method; when a 220kv interconnection line accident occurs, the duty officer checks the equipment of the faulty tripped line to ensure whether it can transmit power normally. The provincial dispatcher will force power transmission by correctly selecting the forced transmission end. If the forced transmission is successful, the circuit breaker on the opposite side will be paralleled or closed, otherwise the line will be transferred for maintenance; When a 500kv line high-voltage reactor protection trips, it will be handled as a transformer accident. If the 500kv line protection and high-voltage reactor protection trip at the same time, it will be handled as a high-voltage reactor and line failure at the same time. When a series ferromagnetic resonance equipment failure occurs, the operator on duty needs to immediately disconnect / close the corresponding equipment switch and eliminate the resonance. If the resonance occurs because the switch is closed, the switch needs to be disconnected immediately, otherwise it is necessary to report to the dispatcher. When a system out-of-step failure occurs, the operating system should be decoupled according to the out-of-step section to eliminate the out-of-step state.

Images