Power distribution network fault repairing and self-healing method and system

Abstract

The invention relates to the technical field of line fault repair, and discloses a power distribution network fault repair self-healing method and system, and the method comprises the steps: obtaining the power distribution network fault power failure time predicted by a trained fault power failure time prediction model; obtaining load data after the power distribution network breaks down, and combining a preset load weight coefficient and the predicted power failure time of the power distribution network to construct an optimization objective function of island division; constructing constraint conditions of island operation under island division; establishing an island division mathematical model according to the constraint conditions and the optimization objective function; solving the island division mathematical model by combining mathematical optimization and a heuristic algorithm through the predicted power distribution network fault power failure time to obtain an island division scheme; and obtaining a fault recovery scheme according to the island division scheme. According to the method, island division is carried out when the power distribution network breaks down, the island range is dynamically adjusted, the load recovery amount is increased, and the fault repairing and self-healing capacity of the power distribution network is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of distribution network fault repair, and in particular to a distribution network fault repair self-healing method and system. Background Art

[0002] As the coverage of the distribution network continues to expand, a large number of distributed power sources are connected to the distribution network, and the types of power loads are more diverse and the distribution is more complex. This increases the probability of distribution network failures. The traditional method of relying on traditional manual inspections and manual repair of faults is no longer efficient. There is an urgent need for automated and intelligent fault repair and self-healing methods to quickly and accurately handle various fault situations and ensure the normal operation of the distribution network.

[0003] In view of this, a distribution network fault repair self-healing method and system are needed. Summary of the invention

[0004] In view of the problem in the prior art that it is difficult to repair the fault by manual inspection and manual repair when a complex distribution network fails, the present invention provides a distribution network fault repair and self-healing method and system, which can divide the distribution network into islands when a fault occurs, dynamically adjust the island range to achieve maximum active support for the load, and improve the fault repair and self-healing capability. The specific technical solution is as follows:

[0005] A distribution network fault repair and self-healing method, comprising:

[0006] Obtain the distribution network fault outage time predicted by the trained fault outage time prediction model;

[0007] Obtain the load data after the distribution network fails, and build the optimization objective function of island division by combining the preset load weight coefficient and the predicted distribution network failure outage time;

[0008] Construct the constraints of island operation under island division;

[0009] According to the constraint conditions and the optimization objective function, a mathematical model for island partitioning is established;

[0010] By predicting the power outage time of the distribution network fault, combining mathematical optimization and heuristic algorithm, the island partition mathematical model is solved to obtain the island partition scheme;

[0011] A fault recovery plan is obtained based on the island division plan.

[0012] Preferably, the obtaining of distribution network fault outage time information predicted by a trained fault outage time prediction model includes:

[0013] Obtaining fault characteristic data of the distribution network and the fault power outage time corresponding to the fault characteristic data;

[0014] The fault feature data and the fault power outage time are input as sample data into the LSTM model for learning and training to obtain a trained fault power outage time prediction model;

[0015] The real-time operation data of the distribution network is input into the trained fault outage time prediction model, and the predicted distribution network fault outage time is output.

[0016] Preferably, the fault characteristic data includes the coverage of the distribution automation system, the fault type, the maintenance distance, the climatic conditions, the operating proficiency of the maintenance personnel and the damage degree of the faulty component.

[0017] Preferably, the load data after the distribution network fails is obtained, and the optimization objective function of island division is constructed by combining the preset load weight coefficient and the predicted distribution network failure outage time, including:

[0018] Obtain the number of load nodes in the power outage area after a distribution network failure, the active power demand of each load, and the operating status information of each load;

[0019] According to the number of load nodes in the blackout area, the active power demand of each load, the operating status information of each load, the preset load weight coefficient and the predicted distribution network fault blackout time, the optimization objective function of island division is constructed; the expression of the optimization objective function is:

[0020]

[0021] Where: T is the predicted power outage time of distribution network fault; N is the number of load nodes in the power outage area; ω i is the load weight coefficient of the i-th load; is the active power demand of the i-th load at time t; It is a 0-1 variable, indicating whether the i-th load is restored at time t. If restored, it is set to 1, otherwise it is set to 0.

[0022] Preferably, the constraint conditions for constructing island operation under island division include constructing real-time risk constraints on island power over-limit probability, constructing power supply output constraints, constructing switch operation times constraints and constructing load shortest power supply duration constraints.

[0023] Preferably, the construction of the real-time risk constraint of island power limit-exceeding probability includes:

[0024] Obtain the active power demand of the load in the island after the distribution network failure, the active power output of the slave power source, the maximum active power of the main power source, and the connection status between the island and the main power source, and calculate the probability risk index of the island power over-limit at each moment. The calculation formula is as follows:

[0025]

[0026] Where: R t is the active power demand of the load in the island at time t; is the active power output from the power source at time t during event x; is the maximum active output of the main power supply during event x; x represents the connection status between the island and the main power supply; Pr{·} is the probability of event {x} being true;

[0027] According to the risk index R of the island power exceeding the limit probability at each moment t And the preset confidence level β, the real-time risk constraint of island power limit exceeding probability is constructed as:

[0028] R t <β

[0029] Where: β is the preset confidence level of the given constraint.

[0030] The power output constraint constructed includes the output constraint of the main power source, the output constraint and the capacity constraint of the energy storage system.

[0031] Preferably, solving the island partitioning mathematical model by predicting the power outage time of the distribution network fault in combination with mathematical optimization and heuristic algorithm includes:

[0032] Obtain the preset energy priority in the distribution network and select the root node of the distribution network;

[0033] According to the root node, the model is solved by using the branch and bound method under the constraints of the number of switch operations and the shortest load power supply duration, and the distribution network switch timing state variables are output to obtain the initial island division scheme within the predicted distribution network fault outage time.

[0034] According to the switch timing state variables of the distribution network, the model that meets all constraints is solved by a heuristic algorithm to obtain the final island partitioning scheme.

[0035] A distribution network fault repair and self-healing system, applied to the above-mentioned distribution network fault repair and self-healing method, comprises:

[0036] A power outage prediction acquisition unit, used to acquire the distribution network fault power outage time predicted by the trained fault power outage time prediction model;

[0037] An objective function construction unit is used to obtain load data after a distribution network failure occurs, and to construct an optimization objective function for island division by combining a preset load weight coefficient and a predicted distribution network failure outage time;

[0038] A constraint construction unit, used to construct constraint conditions for island operation under island division;

[0039] A model building unit, used to build an island partitioning mathematical model according to the constraint conditions and the optimization objective function;

[0040] A model solving unit is used to solve the island partitioning mathematical model by combining mathematical optimization and heuristic algorithms through the predicted power outage time of the distribution network fault to obtain an island partitioning scheme;

[0041] The fault repair plan generating unit is used to obtain a distribution network fault repair self-healing plan according to the island division plan.

[0042] Compared with the prior art, the present invention has the following beneficial effects:

[0043] A distribution network fault repair self-healing method of the present invention obtains the distribution network fault outage time predicted by the trained fault outage time prediction model; obtains the load data after the distribution network fails, and combines the preset load weight coefficient and the predicted distribution network fault outage time to construct the optimization objective function of island division; constructs the constraint conditions of island operation under island division; establishes the island division mathematical model according to the constraint conditions and the optimization objective function; solves the island division mathematical model by the predicted distribution network fault outage time, combines mathematical optimization and heuristic algorithm, and obtains the island division scheme; obtains the fault recovery scheme according to the island division scheme. When the distribution network fails, the present invention establishes the optimization target by ensuring the power supply of key loads with island operation under the premise of satisfying the constraint of the model island operation, solves according to the constraint and objective function, obtains the island division scheme, realizes the island division when the distribution network fails, and dynamically adjusts the island range. Compared with the traditional static analysis method, the present invention not only ensures the continuous and stable operation of the island, but also improves the load recovery amount, and significantly improves the self-healing ability of the distribution network fault repair. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following is a brief introduction to the drawings required for the specific embodiments or the description of the prior art. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn according to the actual scale.

[0045] Figure 1 The present invention is a flow chart of a distribution network fault repair and self-healing method.

[0046] Figure 2 The present invention is a flow chart of another embodiment of a method for self-healing a distribution network fault repair according to the present invention.

[0047] Figure 3 This is a schematic diagram of a distribution network fault repair self-healing system of the present invention. DETAILED DESCRIPTION

[0048] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0049] It should be understood that when used in this specification and the appended claims, the terms "include" and "comprises" indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0050] It should also be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0051] It should be further understood that the term "and / or" used in the present description and the appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0052] See the following examples Figure 1 , pictures and Figure 3 .

[0053] The present application provides a method for self-healing a power distribution network fault, including:

[0054] Step S1, obtaining the distribution network fault outage time predicted by the trained fault outage time prediction model;

[0055] According to the characteristics of the distribution network (such as network structure, load type, historical fault data, etc.), select appropriate fault outage time prediction models, such as LSTM model, ARIMA model and CART algorithm model, etc. According to the selected model, use the historical distribution network fault related data (including fault occurrence time, fault type, repair time, operating conditions at the time, etc.) to train the selected model, adjust the model parameters, and evaluate the accuracy and generalization ability of the model through cross-validation and other means to ensure that the trained model can accurately predict the fault outage time.

[0056] During the actual operation of the distribution network, when a fault is detected or a fault alarm signal is received, real-time characteristic data related to the current fault (such as the location of the fault point, the lines involved, the load conditions at the time, etc.) are collected, and these data are sorted in the format required by the model and input into the trained model. The model outputs the corresponding fault outage time prediction value based on its internal algorithm logic.

[0057] Step S2: Obtain load data after a distribution network failure occurs, and construct an optimization objective function for island division by combining a preset load weight coefficient and a predicted distribution network failure outage time; specifically, including:

[0058] Obtain the number of load nodes in the power outage area after a distribution network failure, the active power demand of each load, and the operating status information of each load;

[0059] According to the number of load nodes in the blackout area, the active power demand of each load, the operating status information of each load, the preset load weight coefficient and the predicted distribution network fault blackout time, the optimization objective function of island division is constructed; the expression of the optimization objective function is:

[0060]

[0061] Where: T is the predicted power outage time of distribution network fault; N is the number of load nodes in the power outage area; ω i is the load weight coefficient of the i-th load; is the active power demand of the i-th load at time t; It is a 0-1 variable, indicating whether the i-th load is restored at time t. If restored, it is set to 1, otherwise it is set to 0.

[0062] By adding the consideration of load importance and power demand to the optimization objective function, an optimization objective function is constructed to ensure the power supply of important loads as much as possible while maximizing the load recovery, so that the subsequent island division work can be carried out around the principle of optimization. Through mathematical quantification, it is ensured that important loads are given priority during the island formation process, the impact of faults on key power consumption departments and users is minimized, and the rationality of distribution network fault response is improved.

[0063] Step S3, constructing the constraints for island operation under island division; specifically, constructing the real-time risk constraint of island power over-limit probability, constructing the power output constraint, constructing the switch operation number constraint and constructing the shortest power supply duration constraint for the load.

[0064] (1) Constructing the real-time risk constraint of island power limit exceeding probability includes:

[0065] Obtain the active power demand of the load in the island after the distribution network failure, the active power output of the slave power source, the maximum active power of the main power source, and the connection status between the island and the main power source, and calculate the probability risk index of the island power over-limit at each moment. The calculation formula is as follows:

[0066]

[0067] Where: R t is the active power demand of the load in the island at time t; is the active power output from the power source at time t during event x; is the maximum active output of the main power supply during event x; x represents the connection status between the island and the main power supply; Pr{·} is the probability of event {x} being true.

[0068] According to the risk index R of the island power exceeding the limit probability at each moment t And the preset confidence level β, the real-time risk constraint of island power limit exceeding probability is constructed as:

[0069] R t <β

[0070] Where: β is the preset confidence level of the given constraint.

[0071] (2) Constructing power output constraints includes:

[0072] The output constraints of the main power supply are as follows:

[0073]

[0074] Where: P host.t is the active power output by the main power supply at time t; P host.max The maximum output power of the main power supply; P load.t is the total load power demand in the island at time t; P client.t is the total output power of other slave power supplies in the island at time t.

[0075] The output constraints of the energy storage system as the main power source are as follows:

[0076] P ess.t =P host.t =∑P load.t -∑P client.t

[0077]

[0078] Where: P ess.t is the output power of the energy storage at time t; They are the maximum charging and discharging powers allowed by energy storage. The following capacity constraints exist during energy storage operation:

[0079] S soc.min ≤S soc.t ≤S soc.max

[0080] Where: S soc.t is the state of charge of the energy storage at time t; S soc.min , S soc.max They are the upper and lower limits of the energy storage state of charge respectively.

[0081] It should be noted that, for example, in a distributed energy system, randomness is also included from power sources generally including photovoltaics and wind power, so it is also necessary to calculate photovoltaic output constraints, wind power output constraints, etc.

[0082] (3) The switch operation times constraint is constructed as follows:

[0083]

[0084] Where: S is the set of distribution network switches; T is the predicted power outage time of distribution network fault; S max is the maximum number of switch operations; is the state of switch s at time t, which is set to 1 when the switch is closed and to 0 when it is open.

[0085] Constructing a switch operation number constraint can ensure that the total number of switch operations during the fault outage time does not exceed the maximum allowable value.

[0086] (4) The shortest power supply duration constraint for the load is constructed as follows:

[0087]

[0088] Where: N is the distribution network load set; T min The shortest power supply duration for the load; T n The duration of continuous power supply to load n.

[0089] The continuity of load power supply is ensured by constructing the shortest load power supply duration constraint.

[0090] The load types in the distribution network are complex and diverse and have different timing characteristics. The switch status and network structure are flexible and changeable during the dynamic adjustment process. The model solution involves multi-variable, multi-constraint, and nonlinear mixed integer combination optimization problems.

[0091] By constructing constraint conditions, the actual operating status of the isolated island after its division is regulated from different aspects to ensure that the basic requirements for power system safety and stability are met during the formation and operation of the isolated island, avoiding violations of electrical operation principles and equipment operation specifications.

[0092] Step S4: establishing an island partitioning mathematical model according to the constraint conditions and the optimization objective function;

[0093] The optimization objective function of the constructed island division and the constraints of each island operation are integrated into a complete mathematical model framework. The optimization objective function is used as the goal to be solved, and the constraints are used as restrictions on the range of variable values ​​and the relationship between equality and inequality.

[0094] The established mathematical model of island partitioning provides a mathematical tool for quantitative analysis for solving the island partitioning scheme, transforms the actual distribution network island partitioning problem into a mathematically solvable problem, and finds an island partitioning scheme that meets various operating requirements and meets the optimization objectives.

[0095] Step S5, solving the island partition mathematical model by using the predicted distribution network fault outage time, combined with mathematical optimization and heuristic algorithms, to obtain an island partition scheme;

[0096] According to the characteristics of the island partition mathematical model, a suitable mathematical optimization algorithm is selected for solving. Specifically, by predicting the power outage time of the distribution network fault, combining mathematical optimization and heuristic algorithms, solving the island partition mathematical model includes:

[0097] Obtain the preset energy priority in the distribution network and select the root node of the distribution network;

[0098] The setting of load priority can be classified according to the degree of importance to social production and life. For example, hospitals, firefighting, emergency command centers and other loads related to life safety, emergency support and key social service operations are listed as important loads with the highest priority; communication base stations, traffic lights and other loads that affect information communication and traffic order are classified as the second highest priority; ordinary residents' daily electricity, general commercial electricity and other relatively less important loads are listed as lower priority loads, etc.

[0099] At the same time, priority weights can also be set to assign corresponding energy priority weights to different types of loads.

[0100] Analyze the topological structure of the distribution network line network, and select nodes that are in key contact positions, have a wide power supply range, and are connected to a large number of important loads as root node candidates. You can also give priority to nodes that are connected to a large number of high-priority loads and reliable distributed energy as root nodes, because starting with such nodes for subsequent island division and other operations is more conducive to ensuring the power supply of important loads and the rational use of energy resources.

[0101] According to the root node, the model is solved by using the branch and bound method under the constraints of the number of switch operations and the shortest load power supply duration, and the distribution network switch timing state variables are output to obtain the initial island division scheme within the predicted distribution network fault outage time.

[0102] According to the switch timing state variables of the distribution network, the model that meets all constraints is solved by a heuristic algorithm to obtain the final island partitioning scheme.

[0103] The heuristic method modifies the switch timing state variables of the output distribution network to avoid frequent island adjustments by reducing the load recovery amount; at the same time, the load weight and power demand are considered, so that only the branch with the smallest weight can be cut off during adjustment, thereby improving the power supply reliability of important loads.

[0104] By solving the mathematical model of island partitioning through an effective algorithm, a scientific and reasonable island partitioning scheme can be obtained quickly and accurately under complex distribution network operating conditions and numerous constraints, providing clear operational guidance for subsequent fault recovery work and improving the distribution network's response capability and resource optimization configuration capability under fault conditions.

[0105] Step S6: Obtain a fault recovery plan according to the island division plan, which specifically includes:

[0106] The first step is to remotely control and start various power generation equipment in the island according to the operating requirements of the power generation equipment, and adjust its power output according to the load demand to maintain power balance.

[0107] The second step is to remotely control the closing of corresponding load switches step by step according to the preset load priority and the order of importance of the corresponding loads to restore power supply to different loads.

[0108] Step 3: When the voltage, frequency and other parameters of the islands on both sides meet the grid-connected conditions, the interconnecting switch is closed by remote control to realize the electrical connection between the islands, and the system status is continuously monitored during the recovery process.

[0109] Step 4: Install monitoring equipment at key locations to collect power operation parameters. When abnormal conditions such as voltage, frequency, overload, etc. are found, timely adjust the power generation output, load access or contact switch status to ensure a smooth recovery process.

[0110] It should be noted that by remotely controlling the switches and start / stop buttons of the distribution network and combining them with the island division scheme obtained in this application, remote controlled orderly fault recovery can be achieved, which significantly improves the self-healing ability of line fault repair and realizes intelligent and automated line fault repair.

[0111] When a distribution network fails, the present invention establishes an optimization target by ensuring the power supply of key loads with island operation under the premise of satisfying the constraints of the model island operation, solves the problem according to the constraints and the objective function, and obtains an island division scheme, thereby realizing island division when a distribution network fails and dynamically adjusting the island range. Compared with the traditional static analysis method, the present invention not only ensures the continuous and stable operation of the island, but also increases the load recovery amount, significantly improving the self-healing ability of the distribution network fault repair.

[0112] Specifically, in a preferred implementation of the present application, the obtaining of distribution network fault outage time information predicted by a trained fault outage time prediction model includes:

[0113] Step S11, acquiring fault characteristic data of the distribution network and the fault power outage time corresponding to the fault characteristic data;

[0114] Extract relevant information from the distribution automation system, such as the area covered by the system, the status of each monitoring node, etc., to determine the coverage data of the distribution automation system. Collect the distance information between the fault point and the departure point of the maintenance personnel during each fault maintenance to determine the characteristic data of the maintenance distance; at the same time, record the relevant information of the maintenance personnel, and then analyze the completion of multiple maintenance tasks of different maintenance personnel to evaluate their operational proficiency and make quantitative annotations. Collect the climatic condition data corresponding to the period of fault occurrence, including detailed meteorological information such as temperature, humidity, wind speed, whether there is rain or snow, because these factors may be related to the fault. According to the on-site maintenance report and the equipment damage assessment report, determine the degree of damage of the faulty component. For example, it can be divided into different levels such as slight, moderate, and severe according to the severity of the damage, and make corresponding numerical representations to facilitate subsequent data processing.

[0115] The fault characteristic data collected from different channels, such as distribution automation system coverage, fault type, maintenance distance, climate conditions, maintenance personnel's proficiency and degree of damage to faulty components, will be sorted out and formatted in a unified way to remove duplicate, erroneous or incomplete data records.

[0116] For each fault event, the sorted fault feature data is matched one by one with the corresponding power outage time after the actual occurrence of the fault to form a complete set of sample data records, ensuring that the fault features in each set of data can accurately correspond to the corresponding fault power outage time, providing reliable input data for subsequent model training.

[0117] By comprehensively and meticulously collecting and organizing the fault feature data of the distribution network and the corresponding fault outage time, a sample data set that fits the actual situation can be constructed. This data set covers multiple key factors that affect the fault outage time, providing a data basis for the subsequent training of an accurate and reliable fault outage time prediction model.

[0118] Step S12: input the fault feature data and the fault power outage time as sample data into the LSTM model for learning and training to obtain a trained fault power outage time prediction model;

[0119] The collected sample data are normalized, and the fault feature data with different value ranges (for example, the maintenance distance may be a value of several thousand meters, and the operation proficiency may be a score between 0-1) are uniformly mapped to a specific interval (such as the [0,1] interval) to avoid adverse effects on model training due to large differences in data magnitude. This also helps to improve the convergence speed and training effect of the model.

[0120] The sample data is divided into a training set and a validation set according to a certain ratio (such as 70% for training and 30% for validation). The structural parameters such as the number of nodes in the input layer, hidden layer, and output layer of the LSTM model are determined according to the dimension of the input data (i.e., the number of fault feature data) and the output target (the single-dimensional data of the fault outage time). For example, if there are 10 types of fault feature data as input, the number of input layer nodes can be set to 10, and the number of output layer nodes is 1 to output the predicted outage time.

[0121] At the same time, select a suitable activation function (such as the commonly used sigmoid, tanh, etc.) for the calculation of the hidden layer nodes, and determine the appropriate loss function and optimization algorithm to build a complete LSTM model architecture.

[0122] Input the training set data batch by batch into the built LSTM model, and perform iterative training according to the set training rounds (for example, 50 rounds, which can be adjusted according to the actual training effect). In each round, the model performs forward propagation calculation based on the input fault feature data to obtain the corresponding fault outage time prediction value, and then calculates the error between the predicted value and the actual value through the loss function, and then uses the optimization algorithm to adjust the model's weight parameters according to the error back propagation, and continuously optimizes the model so that it can more accurately predict the outage time based on the input fault features.

[0123] After each round of training, the model is verified using the validation set data, and evaluation indicators such as the loss value on the validation set (such as mean square error, mean absolute error, etc.) are calculated to observe the performance of the model on data that has not participated in the training, and to determine whether the model is overfitting. If overfitting occurs, appropriate measures (such as increasing the amount of data, adding regularization terms, etc.) can be taken to adjust it to ensure that the model has good generalization ability. After multiple rounds of training and verification, when the evaluation indicators of the model on the validation set reach the expected effect (such as the mean square error is reduced to an acceptable range), save the trained LSTM model parameters so that they can be directly loaded and used in the actual prediction scenario later.

[0124] By using the LSTM model to learn and train sample data, the complex nonlinear relationship between fault feature data and fault outage time can be mined, enabling the model to accurately predict fault outage time based on newly input fault features. The LSTM model itself has advantages in processing time series data and data with long-term dependencies, and is very suitable for data prediction scenarios such as distribution network fault outage time, which is affected by multiple sequential and interrelated factors.

[0125] Step S13: input the real-time operation data of the distribution network into the trained fault outage time prediction model, and output the predicted distribution network fault outage time.

[0126] When a distribution network fault occurs, sensors, monitoring equipment and distribution automation systems installed at key locations of the distribution network collect fault-related operating data in real time, including the specific location of the fault, the corresponding fault type, the climate conditions at the time, the approximate distance from the maintenance personnel, the initial damage of the faulty components, etc., to ensure that the collected data covers the fault feature data dimensions used in the previous training model.

[0127] The sorted real-time operation data of the distribution network is input into the trained fault outage time prediction model. The model calculates and outputs the corresponding fault outage time prediction value based on the mapping relationship between the learned features and the outage time.

[0128] In this embodiment, when a fault actually occurs in the distribution network, the trained model can be quickly used to accurately predict the fault outage time based on the fault feature data collected in real time, so that the operation and maintenance personnel know the approximate duration of the power outage in advance, so that they can formulate island division plans in a more targeted and efficient manner, which helps to improve the overall efficiency of distribution network fault handling and power supply reliability.

[0129] The present application also provides a distribution network fault repair and self-healing system, which is applied to the aforementioned distribution network fault repair and self-healing method, comprising:

[0130] A power outage prediction acquisition unit, used to acquire the distribution network fault power outage time predicted by the trained fault power outage time prediction model;

[0131] An objective function construction unit is used to obtain load data after a distribution network failure occurs, and to construct an optimization objective function for island division by combining a preset load weight coefficient and a predicted distribution network failure outage time;

[0132] A constraint construction unit, used to construct constraint conditions for island operation under island division;

[0133] A model building unit, used to build an island partitioning mathematical model according to the constraint conditions and the optimization objective function;

[0134] A model solving unit is used to solve the island partitioning mathematical model by combining mathematical optimization and heuristic algorithms through the predicted power outage time of the distribution network fault to obtain an island partitioning scheme;

[0135] The fault repair plan generating unit is used to obtain a distribution network fault repair self-healing plan according to the island division plan.

[0136] The functions of each unit in a distribution network fault repair self-healing system of this embodiment are the same as those in an embodiment of a distribution network fault repair self-healing method, and the technical effects are also the same, so they will not be repeated.

[0137] Compared with the prior art, the present invention has the following beneficial effects:

[0138] A distribution network fault repair self-healing method of the present invention obtains the distribution network fault outage time predicted by a trained fault outage time prediction model; obtains load data after the distribution network fault occurs, and combines a preset load weight coefficient and the predicted distribution network fault outage time to construct an optimization objective function for island partitioning; constructs constraint conditions for island operation under island partitioning; establishes an island partitioning mathematical model based on the constraint conditions and the optimization objective function; solves the island partitioning mathematical model based on the predicted distribution network fault outage time, combined with mathematical optimization and heuristic algorithms, to obtain an island partitioning scheme; and obtains a fault recovery scheme based on the island partitioning scheme.

[0139] Those of ordinary skill in the art will appreciate that the units of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition of each example has been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0140] In the embodiments provided by the present invention, it should be understood that the division of units is only a logical function division, and there may be other division methods in actual implementation, for example, multiple units can be combined into one unit, one unit can be split into multiple units, or some features can be ignored, etc.

[0141] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0142] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-0nlyMemory), random access memory (RAM, RandomAccessMemory), mobile hard disk, magnetic disk or optical disk, etc., which can store program code.

[0143] 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 aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. These modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and specification of the present invention.

Claims

1. A distribution network fault repair and self-healing method, characterized in that: include: Obtain the distribution network fault outage time predicted by the trained fault outage time prediction model; Obtain the load data after the distribution network fails, and build the optimization objective function of island division by combining the preset load weight coefficient and the predicted distribution network failure outage time; Construct the constraints of island operation under island division; According to the constraint conditions and the optimization objective function, a mathematical model for island partitioning is established; By predicting the power outage time of the distribution network fault, combining mathematical optimization and heuristic algorithm, the island partition mathematical model is solved to obtain the island partition scheme; A fault recovery plan is obtained based on the island division plan.

2. A distribution network fault repair and self-healing method according to claim 1, characterized in that: The method of obtaining the distribution network fault outage time information predicted by the trained fault outage time prediction model includes: Obtaining fault characteristic data of the distribution network and the fault power outage time corresponding to the fault characteristic data; The fault feature data and the fault power outage time are input as sample data into the LSTM model for learning and training to obtain a trained fault power outage time prediction model; The real-time operation data of the distribution network is input into the trained fault outage time prediction model, and the predicted distribution network fault outage time is output.

3. A distribution network fault repair and self-healing method according to claim 1, characterized in that: The fault characteristic data include the coverage of the distribution automation system, the fault type, the maintenance distance, the climatic conditions, the operation proficiency of the maintenance personnel and the damage degree of the faulty component.

4. A distribution network fault repair and self-healing method according to claim 1, characterized in that: The method of acquiring load data after a distribution network failure occurs, combining a preset load weight coefficient and a predicted distribution network failure outage time, and constructing an optimization objective function for island division includes: Obtain the number of load nodes in the blackout area after a distribution network failure, the active power demand of each load, and the operating status information of each load; construct an optimization objective function for island partitioning based on the number of load nodes in the blackout area, the active power demand of each load, the operating status information of each load, the preset load weight coefficient, and the predicted distribution network failure blackout time; the expression of the optimization objective function is: Where: T is the predicted power outage time of distribution network fault; N is the number of load nodes in the power outage area; ω i is the load weight coefficient of the i-th load; is the active power demand of the i-th load at time t; It is a 0-1 variable, indicating whether the i-th load is restored at time t.

5. A distribution network fault repair and self-healing method according to claim 4, characterized in that: The constraint conditions for island operation under island division include constructing a real-time risk constraint on the probability of island power exceeding the limit, constructing a power output constraint, constructing a switch operation number constraint, and constructing a load shortest power supply duration constraint.

6. A distribution network fault repair and self-healing method according to claim 5, characterized in that: The construction of the real-time risk constraint of island power over-limit probability includes: Obtain the active power demand of the load in the island after the distribution network failure, the active power output of the slave power source, the maximum active power output of the main power source, and the connection status between the island and the main power source, and calculate the probability risk index of the island power over-limit at each moment. The calculation formula is as follows: Where: R t is the active power demand of the load in the island at time t; is the active power output from the power source at time t during event x; is the maximum active power output of the main power supply at event x; x represents the connection status between the island and the main power supply; Pr{·} is the probability of event {x} being true; according to the risk index R of the island power exceeding the limit probability at each moment t And the preset confidence level β, the real-time risk constraint of island power limit exceeding probability is constructed as: R t ＜β Where: β is the preset confidence level of the given constraint.

7. A distribution network fault repair and self-healing method according to claim 5, characterized in that: The method of solving the island partitioning mathematical model by predicting the power outage time of the distribution network fault in combination with mathematical optimization and heuristic algorithm includes: Obtain the preset energy priority in the distribution network and select the root node of the distribution network; According to the root node, the model is solved by using the branch and bound method under the constraints of the number of switch operations and the shortest load power supply duration, and the distribution network switch timing state variables are output to obtain the initial island division scheme within the predicted distribution network fault outage time. According to the switch timing state variables of the distribution network, the model that meets all constraints is solved by a heuristic algorithm to obtain the final island partitioning scheme.

8. A distribution network fault repair self-healing system, characterized in that: A distribution network fault repair and self-healing method applied to any one of claims 1 to 7, comprising: A power outage prediction acquisition unit, used to acquire the distribution network fault power outage time predicted by the trained fault power outage time prediction model; An objective function construction unit is used to obtain load data after a distribution network failure occurs, and to construct an optimization objective function for island division by combining a preset load weight coefficient and a predicted distribution network failure outage time; A constraint construction unit, used to construct constraint conditions for island operation under island division; A model building unit, used to build an island partitioning mathematical model according to the constraint conditions and the optimization objective function; A model solving unit is used to solve the island partitioning mathematical model by combining mathematical optimization and heuristic algorithms through the predicted power outage time of the distribution network fault to obtain an island partitioning scheme; The fault repair plan generating unit is used to obtain a distribution network fault repair self-healing plan according to the island division plan.

Images