Power distribution network power failure time prediction method and system considering distributed resource access

Abstract

The invention specifically discloses a power distribution network power failure time prediction method and system considering distributed resource access, and the method comprises the steps: restoring the power supply of an upstream region of a fault point after fault isolation when a certain element breaks down, searching for all power supplies in a power failure region and interconnection switches connected with a power failure line in a downstream region of the fault point, carrying out transfer area division and transfer load analysis; if no interconnection switch or fault line segment exists in the power failure area, power supply is recovered after the fault is repaired, and the power failure time of the area is calculated; if the interconnection switch exists in the power transfer area and a standby power supply can provide electric energy support, power transfer is carried out on the area, and the power failure time of the area is calculated; and if no standby power supply exists, considering the condition that the distributed resources are used as the power supply to provide electric energy, providing a novel power distribution system power failure time prediction method for different areas involved in power failure, and improving the accuracy of power failure time prediction.

Description

Technical Field

[0001] The present invention belongs to the technical field of power distribution network, and in particular relates to a method and system for predicting power outage time of a distribution network considering access of distributed resources. Background Art

[0002] The distribution network is located at the end of the power grid and is directly connected to the user. It is responsible for distributing the electricity sent by the transmission network to the user. With the implementation of the national "dual carbon" strategic goals and the development and construction of new power systems, more and more distributed resources such as distributed power sources, energy storage devices, and electric vehicles are connected to the distribution network, which has a certain impact on the structure and operation of the distribution system. The new distribution system presents new characteristics such as diversified power supply and interactive power consumption.

[0003] Due to the access of distributed resources, users have more diverse ways to obtain electricity. The outage time and impact range of the new distribution network have also changed compared to the traditional distribution network. Therefore, the analysis method of power outage time should also be changed accordingly. The traditional distribution network only relies on whether the power grid can provide electricity to analyze the operation and outage of the distribution network, and does not take into account the situation of distributed resources providing electricity as the power supply side, resulting in a lack of accuracy in the judgment of the power outage time of the distribution network. Summary of the invention

[0004] To solve the above problems, the purpose of the present invention is to provide a distribution network outage time prediction method and system considering the access of distributed resources. Based on the supplement of distributed resources to the traditional centralized power supply mode, the distributed photovoltaic, decentralized wind power, energy storage devices, electric vehicles and other distributed resources are considered as the power supply side to provide electric energy, and various situations in which their coordinated scheduling jointly assists the distribution network to provide electric energy to users are proposed; for different areas involved in the power outage, a power outage time prediction method under a new distribution system is proposed to further improve the accuracy of the power outage time prediction.

[0005] To achieve the above object, the technical solution of the present invention is: a method for predicting power outage time of a distribution network considering the access of distributed resources, comprising the following steps:

[0006] When a component fails, the power supply is restored to the upstream area after fault isolation. The downstream area of ​​the fault point searches for all power sources in the blackout area and the tie switches connected to the blackout line, and the transfer area is divided and the transfer load is analyzed.

[0007] If there is no tie switch or faulty line segment in the blackout area, wait for the fault to be repaired before power is restored, and calculate the blackout time in the area;

[0008] If there is a contact switch in the transfer area and there is a backup power supply to provide power support, the power supply will be transferred to the area and the power outage time of the area will be calculated;

[0009] If there is no backup power supply:

[0010] If there is a tie switch in the transfer area and it is connected to wind, solar and storage distributed resources, the total power generation of wind, solar and storage is predicted based on the historical output curve of the wind, solar and storage distributed resources during the fault power outage time based on the time convolution network prediction model, and the total load demand is predicted based on the historical time series curve of the power outage load;

[0011] If the total power generation of wind, solar and storage distributed resources is greater than or equal to the total load demand, the power supply to the area will be restored through the wind, solar and storage distributed resources, and the power outage time in the area will be calculated;

[0012] If the total power generation of distributed wind, solar and storage resources is less than the total load demand, and there are important users in the transfer area, the distributed wind, solar and storage resources will be used to transfer power to important users according to the preset load shedding plan, and the power outage time of this type of load will be calculated.

[0013] Preferably, the method of training the initialized temporal convolutional network prediction model using the data of the training set based on the improved particle swarm algorithm further includes:

[0014] Initialize the model parameters, including the inertia weights wini and wfin of the improved particle swarm algorithm, learning coefficients c1ini, c1fin, c2ini, c2fin, number of particles, number of iterations, and number of neurons;

[0015] Using an improved particle swarm algorithm to train the temporal convolutional network prediction model, and determining Pbest and Gbest according to an initial fitness value;

[0016] Iterate and update the Pbest of each particle and the Gbest of the group. If the fitness value of the particle in the current iteration is less than Pbest, it will be assigned to Pbestt, otherwise Pbest remains unchanged. From the perspective of the entire particle group, if there is a particle with a fitness value less than Gbest in this iteration, it will be assigned to Gbest, otherwise Gbest remains unchanged.

[0017] Update the speed and position of each particle;

[0018] Determine whether the maximum number of iterations has been reached. If so, return the optimal parameter combination in the entire iteration process; otherwise, continue to use the IPPSO (particle swarm) algorithm to train the temporal convolutional network prediction model;

[0019] The returned parameters are used to construct the temporal convolutional network prediction model.

[0020] Preferably, using the improved particle swarm algorithm to train the temporal convolutional network prediction model further comprises:

[0021] Exponential function and cosine function are introduced to perform nonlinear improvement on w to fine-tune the range of w, so that w reaches the optimal attenuation interval [0.4, 0.9] and maintains dynamic changes during the entire iteration;

[0022] For the maximum number of iterations Tmax, the improvement of w for the tth time is as follows:

[0023]

[0024] Where wini and wfin are the initial value and final value of w, n is the control coefficient for adjusting the smoothness of the exponential function, n=8.

[0025] Preferably, using the improved particle swarm algorithm to train the temporal convolutional network prediction model further comprises:

[0026] Adaptively adjust c1 and c2 in the improved particle swarm algorithm so that they change linearly according to the number of iterations. The improved formula of the learning factor is as follows:

[0027]

[0028] In the formula: c1ini, c1fin, c2ini, c2fin are the initial and final values ​​of c1 and c2.

[0029] Preferably, obtaining historical power generation data of distributed resources within the distribution network further includes:

[0030] Collect historical operating output data and meteorological data of distributed resources to obtain the time series of distributed power generation;

[0031] Analyze the operating characteristics of distributed resources connected to the distribution network, including distributed photovoltaics, distributed wind power, energy storage devices, and electric vehicles;

[0032] Fit the time-output curves of various distributed resources.

[0033] Preferably, the method further includes: integrating the time-output curve of the distributed resources with the primary topology data of the distribution network to construct a new distribution network topology model, wherein the primary topology data of the distribution network refers to the equipment inventory and topology information of the current distribution network, including the topological connection relationship between transformers, lines, and switch facilities, basic equipment information including equipment number, name, and voltage level, equipment outage parameters including fault outage rate and average fault repair time, and load point data including load capacity, number of users, and importance level.

[0034] Preferably, enumerating the temporal convolutional network prediction model after multiple fault input training to predict the power outage time of the distribution network further includes:

[0035] If there are important users in the transfer area, a load shedding plan will be formulated. For important users, electric vehicles will be called as the power source, combined with wind, solar and storage to transfer power, and the power outage time of this type of load will be calculated.

[0036] Based on the same concept, the present invention also provides a power distribution network outage time prediction system considering distributed resource access, comprising:

[0037] The fault collection module is used to restore power to the upstream area of ​​the fault point after fault isolation when a component fails, search for all power sources in the power outage area and the tie switches connected to the power outage line in the downstream area of ​​the fault point, divide the power supply area, and analyze the power supply load;

[0038] The prediction module is used to wait for the fault to be repaired and the power supply to be restored, and calculate the power outage time of the area if there is no contact switch and fault line segment in the power outage area; if there is a contact switch in the transfer area and there is a backup power supply to provide power support, the power supply to the area is transferred and the power outage time of the area is calculated; if there is no backup power supply: if there is a contact switch in the transfer area and there are wind, solar and storage distributed resources connected, the total power generation of wind, solar and storage is predicted based on the time convolution network prediction model according to the historical output curve of the wind, solar and storage distributed resources during the fault power outage time, and the total load demand is predicted according to the historical time series curve of the power outage load; if the total power generation of the wind, solar and storage distributed resources is greater than or equal to the total load demand, the wind, solar and storage distributed resources are used to transfer power to restore the power supply to the area, and the power outage time of the area is calculated; if the total power generation of the wind, solar and storage distributed resources is less than the total load demand, and there are important users in the transfer area, then according to the preset load shedding plan, the wind, solar and storage distributed resources are used to transfer power to important users, and the power outage time of this type of load is calculated.

[0039] Based on the same concept, the present invention also provides a computer device, comprising:

[0040] A memory, the memory being used to store a processing program;

[0041] A processor, wherein when executing the processing program, the processor implements any one of the above-mentioned methods for predicting power outage time in a distribution network considering access of distributed resources.

[0042] Based on the same concept, the present invention also provides a readable storage medium, on which a processing program is stored. When the processing program is executed by a processor, any one of the above-mentioned methods for predicting power outage time in a distribution network considering distributed resource access is implemented.

[0043] After adopting the above technical solution, compared with the prior art, the present invention has the following advantages:

[0044] The present invention takes into account the situation where various distributed resources such as distributed photovoltaic, decentralized wind power, energy storage devices, electric vehicles, etc. provide electrical energy as the power supply side, and proposes various situations in which they coordinate and dispatch to jointly assist the distribution network to provide electrical energy to users; for different areas involved in the power outage, a power outage time prediction method under a new distribution system is proposed to further improve the accuracy of the power outage time prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The specific embodiments of the present invention are further described in detail below with reference to the accompanying drawings, wherein:

[0046] Figure 1 It is an overall flow chart of a method for predicting power outage time of a distribution network considering access of distributed resources according to an embodiment of the present invention;

[0047] Figure 2 It is a w variation curve diagram of the improved particle swarm algorithm of the present invention;

[0048] Figure 3 It is a basic framework flow chart of the IPPSO-TCN prediction model established by the present invention;

[0049] Figure 4 It is a power outage simulation flow chart of the present invention. DETAILED DESCRIPTION

[0050] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. The advantages and features of the present invention will become more apparent from the following description and claims. It should be noted that the drawings are all in a very simplified form and use non-precise ratios, which are only used to conveniently and clearly assist in explaining the purpose of the embodiments of the present invention.

[0051] It should be noted that all directional indications in the embodiments of the present invention (such as up, down, left, right, front, back, etc.) are only used to explain the relative position relationship, movement status, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0052] Example

[0053] like Figure 1 As shown, the contribution made by the inventor of the present application to the power distribution network outage time prediction method considering the access of distributed resources is shown, including the following steps:

[0054] Step 1: Fit the timing characteristics of distributed resources. That is, analyze the operating characteristics of various distributed resources and fit their time-output curves.

[0055] The operating characteristics of distributed resources connected to the distribution network are analyzed. Distributed resources mainly include distributed photovoltaics, decentralized wind power, energy storage devices, electric vehicles, etc.

[0056] The operating characteristics of distributed photovoltaics are as follows: they are mainly affected by light intensity and temperature, and show diurnal and seasonal characteristics. The output is higher during the daytime at noon and basically zero at night; the output is higher in summer and lower in winter.

[0057] The operating characteristics of distributed wind power are as follows: Mainly affected by wind speed. When the wind speed is greater than the cut-in wind speed set for the wind turbine, the wind turbine starts to run; when the wind speed is greater than the set rated wind speed, the wind turbine runs at rated power; when the wind speed is greater than the set cut-out wind speed, the wind turbine stops running.

[0058] The operating characteristics of energy storage devices are as follows: Different energy storage technologies have different characteristics such as charging and discharging power, energy density, and discharge time, but they play the same role in the power system, that is, by adjusting the charging and discharging power to absorb or release electric energy to smooth out fluctuations in new energy output and ensure safe and stable operation of the power grid.

[0059] The operating characteristics of electric vehicles are as follows: mainly determined by the charging behavior of the car owner. Taking private cars as an example, the charging locations are mainly concentrated in the residence and workplace, and the charging time is mainly concentrated in the night rest period and during work.

[0060] Based on the above analysis of the operating characteristics, the time-output curves of various distributed resources are fitted according to the actual conditions in different regions.

[0061] Step 2: Build a graph model. Integrate the time-output curve of distributed resources with the primary topology data of the distribution network to build a new distribution network topology model.

[0062] Collect historical output data, predicted output data and primary topology data of various distributed resources. The primary topology data of the distribution network mainly refers to the equipment ledger and topology information of the current distribution network, including the topological connection relationship between facilities such as transformers, lines, switches, basic equipment information such as equipment number, name, voltage level, equipment outage parameters such as fault outage rate and average fault repair time, as well as load point data such as load capacity, number of users, and importance level.

[0063] For distributed resources that have been connected to the distribution network, the distribution network topology model is updated according to their actual access situation. For distributed resources that are planned to be connected to the distribution network, based on the network topology and real-time load conditions of the existing distribution system, the access location, access capacity, and access method of the distributed resources are considered to build a new distribution system network topology model.

[0064] Step 3: Determine the power generation of distributed resources. Establish an improved particle swarm optimization-temporal convolutional network (IPPSO-TCN) model to predict the power generation of distributed power sources and analyze the total capacity of distributed resources during the fault outage time.

[0065] First, the historical operating output data and meteorological data of distributed power sources are collected to obtain the time series of distributed power generation. The data is preprocessed, including removing outliers, filling missing values, and standardizing data. The preprocessed data is divided into training sets and test sets.

[0066] Then, the IPPSO-TCN model is established. The model consists of two parts: improving the particle swarm algorithm and running the improved particle swarm algorithm to optimize the parameters of the temporal convolutional network model, thereby establishing the IPPSO-TCN model.

[0067] The particle swarm algorithm (PSO) compares the birds in a flock to particles in a fixed space. Each particle updates its individual optimal value Pbest through its own search. Particles exchange information and jointly search for the global optimal value Gbest. They continuously iterate to find the optimal solution in the search area. In the d-dimensional search space, n particles form a population x = [x1,…,xn], the position of the i-th particle xi = [xi1,…,xid], the speed vi = [vi1,…,vid], the optimal value pi = [pi1,…,pid], and the global optimal value of the particle swarm pg = [pg1,…,pgd]. The formula for updating the speed and position of the k-th particle is as follows:

[0068]

[0069]

[0070] Where: w is the inertia weight, which indicates the degree to which the particle maintains the velocity at the previous moment; c1 and c2 are learning factors; r1 and r2 are random numbers between [0,1].

[0071] The methods to improve the PSO algorithm are as follows:

[0072] (1) Improvement of inertia weight

[0073] Exponential function and cosine function are introduced to improve w nonlinearly. According to the algorithm mechanism of exponential function, when the exponent is negative with e as the base, the image shows signs of rapid decay first and then slow decay, resulting in the change range of w approaching 0 after the 8th iteration, which is not conducive to the dynamic change of weight. Therefore, on this basis, the cosine function is added to fine-tune the range of w, so that w reaches the optimal decay range [0.4, 0.9] and maintains dynamic changes during the entire iteration. The change curve is as follows Figure 2 shown.

[0074] For the maximum number of iterations Tmax, the improvement of w for the tth time is as follows:

[0075]

[0076] Where: wini and wfin are the initial value and final value of w; n is the control coefficient for adjusting the smoothness of the exponential function, n=8.

[0077] (2) Improvement of learning factors

[0078] In the original PSO algorithm, c1 and c2 are fixed values ​​set based on experience, which do not match the trajectory of the particle. This paper makes adaptive time-varying adjustments to c1 and c2 so that they change linearly according to the number of iterations. The improved formula of the learning factor is as follows:

[0079]

[0080] In the formula: c1ini, c1fin, c2ini, c2fin are the initial and final values ​​of c1 and c2.

[0081] Finally, the established IPPSO-TCN model is used to predict the power generation of distributed power sources and obtain the total power generation of distributed power sources during the fault outage time.

[0082] The basic framework of the established IPPSO-TCN prediction model is as follows Figure 3 shown.

[0083] The specific steps of building the prediction model are as follows:

[0084] (1) Obtain historical data, preprocess the data, replace outliers that exceed the threshold with the threshold, fill missing values ​​with the mean interpolation method, and divide the preprocessed data set into training set and test set;

[0085] (2) Initialize the model parameters, including the inertia weights wini and wfin of the IPPSO algorithm, the learning coefficients c1ini, c1fin, c2ini, c2fin, the number of particles, the number of iterations, the number of neurons, etc.;

[0086] (3) Use the IPPSO algorithm to train the prediction model and determine Pbest and Gbest based on the initial fitness value;

[0087] (4) Iterate and update the Pbest of each particle and the Gbest of the group. If the fitness value of the particle in the current iteration is less than Pbest, it is assigned to Pbestt, otherwise Pbest remains unchanged; from the perspective of the entire particle group, if there is a particle with a fitness value less than Gbest in this iteration, it is assigned to Gbest, otherwise Gbest remains unchanged;

[0088] (5) Update the speed and position of each particle;

[0089] (6) Determine whether the maximum number of iterations has been reached. If the maximum number of iterations has been reached, return the optimal parameter combination in the entire iteration process; otherwise, return to step (3);

[0090] (7) Use the returned parameters to build a model and make predictions about distributed resources.

[0091] Step 4: Power outage simulation: Analyze the support probability of distributed resources as backup power sources, and use the minimum path method to simulate faulty components, power outage scope, and power outage time.

[0092] Enumerate single component faults, record the process of circuit breaker tripping, fault isolation, load transfer, and power supply restoration after the fault, determine the scope of power outage affected by the fault, and further predict the fault power outage time of the distribution network.

[0093] The specific process is as follows:

[0094] 1. When a component fails, the circuit breaker trips and causes a power outage on the entire line. After locating and isolating the fault, restore power to the upstream area of ​​the fault point. Perform the following analysis on the downstream area of ​​the fault point;

[0095] 2. Search for all power sources in the blackout area and the tie switches connected to the blackout lines, divide the transfer areas, analyze the transfer loads, and rank the loads according to the analysis results;

[0096] 3. If there is a contact switch in the transfer area and there is a backup power supply to provide power support, the power supply will be transferred to the area and the power outage time of the area will be calculated. If there is no backup power supply, the next step of analysis will be carried out;

[0097] 4. If there is a tie switch in the transfer area and it is connected to distributed resources, the IPPSO-TCN model is established to predict the total wind, solar and storage power generation in the area based on the historical output curve of wind, solar and storage during the fault power outage time. The total load demand is predicted based on the historical time series curve of the power outage load;

[0098] 5. If the total power generation of wind, solar and storage is greater than or equal to the total load demand, the power supply will be restored to the area through wind, solar and storage, and the power outage time of the area will be calculated. Otherwise, the next step of analysis will be carried out;

[0099] 6. If there are important users in the transfer area, a load shedding plan will be formulated. For important users, electric vehicles will be used as the power supply side, combined with wind, solar and storage to transfer power, and the power outage time of this type of load will be calculated;

[0100] 7. If there is no interconnecting switch or faulty line segment in the power outage area, wait for the fault to be repaired before restoring power supply and calculate the power outage time in the area.

[0101] The overall flow chart of power outage simulation is as follows: Figure 4 shown.

[0102] Step 5: Predict the power outage time. Simulate the power outage event of the new distribution network multiple times and predict the power outage time in different power outage areas.

[0103] The power outage time in the upstream area of ​​the fault point is:

[0104] t 故障点上游区域 =t 故障定位隔离 +t 故障点上游恢复供电操作

[0105] The power outage time in the backup power supply area is:

[0106] t 备用电源转供区域 =t 故障定位隔离 +t 联络开关切换

[0107] The power outage time in the distributed resource transfer area is:

[0108] t 分布式资源转供区域 =t 故障定位隔离 +t 分布式资源总发电量判断 +t 联络开关切换

[0109] The power outage time in the area without contact switch is:

[0110] t 无联络开关区域 =t 故障定位隔离 +t 故障修复

[0111] The power outage time of the fault section is:

[0112] t 故障段 =t 故障定位隔离 +t 故障修复

[0113] Compared with the prior art, the present invention has the following beneficial effects:

[0114] In view of the current large-scale access of distributed resources to the distribution network, the traditional distribution network outage time analysis method is no longer applicable. The present invention considers the situation where a variety of distributed resources such as distributed photovoltaics, distributed wind power, energy storage devices, electric vehicles, etc. provide electricity as the power supply side, and proposes a variety of situations in which they coordinate and dispatch to jointly assist the distribution network to provide electricity to users; for different areas involved in the power outage, a new power outage time prediction method under the distribution system is proposed to further improve the accuracy of the power outage time prediction.

[0115] The present invention proposes a prediction model based on IPPSO-TCN to predict the power generation of distributed power sources. In view of the problem that the PSO algorithm is prone to fall into the local optimum and the problem that the parameters of the TCN model are difficult to determine, the exponential function, cosine function and time parameter are introduced into the PSO algorithm to improve w, c1 and c2, and the improved IPPSO algorithm is obtained. The IPPSO algorithm is used to optimize the parameters of TCN, thereby improving the prediction performance of the model.

[0116] See also Figure 4 The power outage simulation flow chart shows the actual process during the operation of the distribution system.

[0117] A method for predicting power outage time of a distribution network considering the access of distributed resources comprises the following steps:

[0118] When a component fails, the power supply is restored to the upstream area after fault isolation. The downstream area of ​​the fault point searches for all power sources in the blackout area and the tie switches connected to the blackout line, and the transfer area is divided and the transfer load is analyzed.

[0119] If there is no tie switch or faulty line segment in the blackout area, wait for the fault to be repaired before power is restored, and calculate the blackout time in the area;

[0120] If there is a contact switch in the transfer area and there is a backup power supply to provide power support, the power supply will be transferred to the area and the power outage time of the area will be calculated;

[0121] If there is no backup power supply:

[0122] If there is a tie switch in the transfer area and it is connected to wind, solar and storage distributed resources, the total power generation of wind, solar and storage is predicted based on the historical output curve of the wind, solar and storage distributed resources during the fault power outage time based on the time convolution network prediction model, and the total load demand is predicted based on the historical time series curve of the power outage load;

[0123] If the total power generation of wind, solar and storage distributed resources is greater than or equal to the total load demand, the power supply to the area will be restored through the wind, solar and storage distributed resources, and the power outage time in the area will be calculated;

[0124] If the total power generation of distributed wind, solar and storage resources is less than the total load demand, and there are important users in the transfer area, the distributed wind, solar and storage resources will be used to transfer power to important users according to the preset load shedding plan, and the power outage time of this type of load will be calculated.

[0125] Based on the same concept, the present invention also provides a power distribution network outage time prediction system considering distributed resource access, comprising:

[0126] The fault collection module is used to restore power to the upstream area of ​​the fault point after fault isolation when a component fails, search for all power sources in the power outage area and the tie switches connected to the power outage line in the downstream area of ​​the fault point, divide the power supply area, and analyze the power supply load;

[0127] The prediction module is used to wait for the fault to be repaired and the power supply to be restored, and calculate the power outage time of the area if there is no contact switch and fault line segment in the power outage area; if there is a contact switch in the transfer area and there is a backup power supply to provide power support, the power supply to the area is transferred and the power outage time of the area is calculated; if there is no backup power supply: if there is a contact switch in the transfer area and there are wind, solar and storage distributed resources connected, the total power generation of wind, solar and storage is predicted based on the time convolution network prediction model according to the historical output curve of the wind, solar and storage distributed resources during the fault power outage time, and the total load demand is predicted according to the historical time series curve of the power outage load; if the total power generation of the wind, solar and storage distributed resources is greater than or equal to the total load demand, the wind, solar and storage distributed resources are used to transfer power to restore the power supply to the area, and the power outage time of the area is calculated; if the total power generation of the wind, solar and storage distributed resources is less than the total load demand, and there are important users in the transfer area, then according to the preset load shedding plan, the wind, solar and storage distributed resources are used to transfer power to important users, and the power outage time of this type of load is calculated.

[0128] Based on the same concept, the present invention also provides a computer device, comprising:

[0129] A memory, the memory being used to store a processing program;

[0130] A processor, wherein when executing the processing program, the processor implements any one of the above-mentioned methods for predicting power outage time in a distribution network considering access of distributed resources.

[0131] Based on the same concept, the present invention also provides a readable storage medium, on which a processing program is stored. When the processing program is executed by a processor, any one of the above-mentioned methods for predicting power outage time in a distribution network considering distributed resource access is implemented.

[0132] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0133] 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 is essentially 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. The computer software product is stored in a storage medium, including a number of instructions to enable 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, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program codes.

[0134] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting power outage time in a distribution network considering the access of distributed resources, characterized in that: The following steps are involved: When a component fails, the power supply is restored to the upstream area after fault isolation. The downstream area of ​​the fault point searches for all power sources in the blackout area and the tie switches connected to the blackout line, and the transfer area is divided and the transfer load is analyzed. If there is no tie switch or faulty line segment in the blackout area, wait for the fault to be repaired before power is restored, and calculate the blackout time in the area; If there is a contact switch in the transfer area and there is a backup power supply to provide power support, the power supply will be transferred to the area and the power outage time of the area will be calculated; If there is no backup power supply: If there is a tie switch in the transfer area and it is connected to wind, solar and storage distributed resources, the total power generation of wind, solar and storage is predicted based on the historical output curve of the wind, solar and storage distributed resources during the fault power outage time based on the time convolution network prediction model, and the total load demand is predicted based on the historical time series curve of the power outage load; If the total power generation of wind, solar and storage distributed resources is greater than or equal to the total load demand, the power supply to the area will be restored through the wind, solar and storage distributed resources, and the power outage time in the area will be calculated; If the total power generation of distributed wind, solar and storage resources is less than the total load demand, and there are important users in the transfer area, the distributed wind, solar and storage resources will be used to transfer power to important users according to the preset load shedding plan, and the power outage time of this type of load will be calculated.

2. The method for predicting power outage time of a distribution network considering access of distributed resources according to claim 1, characterized in that: The method of training the initialized temporal convolutional network prediction model based on the improved particle swarm algorithm using the data of the training set further includes: Initialize the model parameters, including the inertia weights wini and wfin of the improved particle swarm algorithm, learning coefficients c1 ini, c1fin, c2 ini, c2fin, number of particles, number of iterations, and number of neurons; Using an improved particle swarm algorithm to train the temporal convolutional network prediction model, and determining Pbest and Gbest according to an initial fitness value; Iterate and update the Pbest of each particle and the Gbest of the group. If the fitness value of the particle in the current iteration is less than Pbest, it will be assigned to Pbestt, otherwise Pbest remains unchanged. From the perspective of the entire particle group, if there is a particle with a fitness value less than Gbest in this iteration, it will be assigned to Gbest, otherwise Gbest remains unchanged. Update the speed and position of each particle; Determine whether the maximum number of iterations has been reached. If so, return the optimal parameter combination in the entire iteration process; otherwise, continue to use the IPPSO algorithm to train the temporal convolutional network prediction model; The returned parameters are used to construct the temporal convolutional network prediction model.

3. The method for predicting power outage time of a distribution network considering access of distributed resources as claimed in claim 2, characterized in that: Using the improved particle swarm algorithm to train the temporal convolutional network prediction model further includes: Exponential function and cosine function are introduced to perform nonlinear improvement on w to fine-tune the range of w, so that w reaches the optimal attenuation interval [0.4, 0.9] and maintains dynamic changes during the entire iteration; For the maximum number of iterations Tmax, the improvement of w for the tth time is as follows: Where wini and wfin are the initial value and final value of w, n is the control coefficient for adjusting the smoothness of the exponential function, n=8.

4. The method for predicting power outage time of a distribution network considering access of distributed resources as claimed in claim 2, characterized in that: Using the improved particle swarm algorithm to train the temporal convolutional network prediction model further includes: Adaptively adjust c1 and c2 in the improved particle swarm algorithm so that they change linearly according to the number of iterations. The improved formula of the learning factor is as follows: In the formula: c1 ini, c1 fin, c2 ini, c2fin are the initial values ​​and final values ​​of c1 and c2.

5. The method for predicting power outage time of a distribution network considering access of distributed resources according to claim 1, characterized in that: Obtaining historical power generation data of distributed resources within the distribution network further includes: Collect historical operating output data and meteorological data of distributed resources to obtain the time series of distributed power generation; Analyze the operating characteristics of distributed resources connected to the distribution network, including distributed photovoltaics, distributed wind power, energy storage devices, and electric vehicles; Fit the time-output curves of various distributed resources.

6. The method for predicting power outage time of a distribution network considering access of distributed resources according to claim 5, characterized in that: It also includes integrating the time-output curve of distributed resources with the primary topology data of the distribution network to construct a new distribution network topology model. The primary topology data of the distribution network refers to the equipment inventory and topology information of the existing distribution network, including the topological connection relationship between transformers, lines, and switch facilities, basic equipment information including equipment number, name, and voltage level, equipment outage parameters including fault outage rate and average fault repair time, and load point data including load capacity, number of users, and importance level.

7. The method for predicting power outage time of a distribution network considering access of distributed resources according to claim 1, characterized in that: If there are important users in the transfer area, a load shedding plan will be formulated. For important users, electric vehicles will be called as the power source, combined with wind, solar and storage to transfer power, and the power outage time of this type of load will be calculated.

8. A power outage time prediction system for a distribution network considering the access of distributed resources, characterized in that: include: The fault collection module is used to restore power to the upstream area of ​​the fault point after fault isolation when a component fails, search for all power sources in the power outage area and the tie switches connected to the power outage line in the downstream area of ​​the fault point, divide the power supply area, and analyze the power supply load; The prediction module is used to wait for the power supply to be restored after the fault is repaired if there is no tie switch and fault line segment in the power outage area, and calculate the power outage time in the area; If there is a contact switch in the transfer area and there is a backup power supply to provide power support, the power supply will be transferred to the area and the power outage time of the area will be calculated; if there is no backup power supply: if there is a contact switch in the transfer area and it is connected to wind, solar and storage distributed resources, the total power generation of wind, solar and storage is predicted based on the historical output curve of the wind, solar and storage distributed resources during the fault power outage time based on the time convolution network prediction model, and the total load demand is predicted based on the historical time series curve of the power outage load; if the total power generation of the wind, solar and storage distributed resources is greater than or equal to the total load demand, the power supply will be transferred through the wind, solar and storage distributed resources to restore the power supply to the area and calculate the power outage time of the area; If the total power generation of distributed wind, solar and storage resources is less than the total load demand, and there are important users in the transfer area, the distributed wind, solar and storage resources will be used to transfer power to important users according to the preset load shedding plan, and the power outage time of this type of load will be calculated.

9. A computer device, characterized in that: include: A memory, the memory being used to store a processing program; A processor, wherein when executing the processing program, the processor implements the method for predicting power outage time in a distribution network taking into account the access of distributed resources as described in any one of claims 1 to 7.

10. A readable storage medium, characterized in that: The readable storage medium stores a processing program, and when the processing program is executed by a processor, the method for predicting power outage time in a distribution network considering access of distributed resources as described in any one of claims 1 to 7 is implemented.

Images