Power distribution network reliability evaluation method based on load partitioning and improvement control strategy

Abstract

The power distribution network reliability evaluation method based on load partitioning comprises the following steps: S1, establishing an improved power distribution network reliability evaluation model of a switch element-oriented load partitioning technology, and inputting a fault parameter, a load parameter, a system topology parameter and a load branch parameter; s2, forming a system topological structure diagram according to the branch numbers of the line nodes; s3, partitioning the power distribution network according to the system topological structure diagram; s4, introducing fuse factors, and calculating block fault parameters; s5, enumerating fault blocks to form a fault consequence mode table; s6, performing power supply reliability judgment on the areas divided in the step S3, setting fuses in the required areas, and supplementing the influence of fuse branches on the load at the same time; s7, enumerating all fault blocks; s8, adopting a Bayesian network method to obtain weak links of the power distribution network; and S9, calculating a system reliability index. The reliability evaluation precision of the power distribution network is comprehensively improved, and safe and stable operation of a power system is ensured.

Description

Technical Field

[0001] The present invention belongs to the field of power supply safety, and in particular relates to a distribution network reliability assessment method based on load segmentation and an improvement control strategy. Background Art

[0002] With the rapid economic development and the continuous expansion of urban scale, the demand for power supply security is growing. How to ensure the reliability of power supply in urban distribution networks has become a major problem in power grid security and economic operation, and has become a common challenge faced by countries around the world. The major political and social risks brought about by power outages in distribution networks are also increasing. Therefore, it is more urgent to comprehensively improve the reliability of power supply in distribution networks. Since 2000, the annual increase in my country's electricity load has exceeded 10%, and the scale of distribution network capacity ranks first in the world, 1.4 times that of the United States. This has also brought about a surge in the risk of major political and social impacts caused by power outages in distribution networks. The need to comprehensively improve the reliability of power supply in distribution networks is becoming increasingly urgent. Evaluating the reliability of distribution networks can ensure the quality of power supply and is also an important means to achieve reasonable distribution of electric energy. Foreign countries have conducted relatively complete research on the content of distribution network risk assessment. The methods are flexible and varied, and can be applied to more complex network structures, simulate the possible failure process of actual systems, and perform optimal power supply restoration. Some developed countries have begun research on the reliability of distribution networks since the end of the last century, and the reliability has also reached a relatively high level. In the early years, my country did not attach much importance to reliability, so there was a large gap with developed countries. In 1994, after statistics on the distribution systems of 174 major cities across the country, it was found that the highest user power supply reliability rate was 99.85%, that is, the average annual power outage time per household was 13.14 hours; the power outage time in the city with the lowest user power supply reliability rate could reach 373.088 hours / year / household, which was far lower than the power supply reliability level of developed countries abroad during the same period. The power supply reliability of the distribution network plays an important role in the national economy and people's livelihood and in maintaining social public security. It is also an important indicator for measuring the construction and operation level of the power grid. The reliability index of the distribution network reflects the continuous power supply capacity of the power system to users. Research on the reliability of the distribution network can improve the power supply quality of the power grid, which is conducive to social stability and is the basis for stable economic growth. On the other hand, the research on the reliability of the distribution network is also an important reference for urban construction and transformation. Combining the reliability index and the structure of the distribution network, the weak links in the network architecture, operation management, etc. can be analyzed, and corresponding measures can be proposed for the problems that arise, so that the reliability management can develop in a scientific and reasonable way, providing a basis for the transformation and construction of the power grid.

[0003] In the initial stage of the development of my country's power industry, only the transmission system was focused, and the reliability of the distribution network was not well understood, resulting in frequent power outages in the distribution network. With the continuous progress of my country's economy and society, power users are paying more and more attention to the stability and reliability of power supply, and domestic scholars have gradually started to study the reliability assessment of distribution networks. The main technologies currently used to study the reliability of distribution networks are as follows:

[0004] The abductive reasoning network and model prediction method are combined to predict power grid faults and power outages. The abductive reasoning network constructs a power grid fault diagnosis system by processing the complex relationship between the prediction data and the corresponding faults, and uses the model prediction method to predict distribution network faults and power outages based on historical fault data and actual operation data.

[0005] Combining the analytic hierarchy process, fuzzy theory and entropy method, the operating status of the distribution network is evaluated according to the historical data of distribution network faults and power outages, operating conditions and natural environment, and the changing trend of distribution network faults and power outages is analyzed, based on which the operating status of the power grid is predicted.

[0006] Big data is used to analyze the influencing factors of fault power outages, and a random forest fault power outage prediction model is established to predict distribution network fault power outages.

[0007] Due to the late start of research on the reliability of distribution networks in China, my country currently requires that the reliability of distribution networks should not be less than 99.9%, which is still a certain distance from the 99.99% in Europe and the United States. At present, only a few regions in my country have taken the goal of achieving lean and scientific management of distribution networks and actively used advanced calculation and analysis tools to analyze and calculate the reliability indicators of the entire network system and important loads. Most regions are still estimating the reliability of distribution network power supply through the engineering practice experience of technicians, and lack simple, easy-to-operate, and relatively accurate real-time reliability analysis tools for reliability indicator management.

[0008] In recent years, as power supply companies in various regions continue to pay attention to power supply reliability indicators, reliability-related specifications are gradually being formulated, analysis tools are being continuously developed, and reliability algorithm models are being continuously optimized and improved, which has achieved remarkable results in improving the level of reliability analysis. State Grid Corporation of China has built the first batch of high-reliability demonstration areas in typical areas of key cities such as Xiamen and Shanghai to further improve the quantitative calculation and analysis level of system distribution network reliability. At the same time, in order to further optimize the schemes and investments used to improve power supply reliability, in recent years, some power supply companies have conducted cost-benefit economic analysis based on the reliability of distribution network power supply before planning and transforming urban distribution networks, in order to achieve the highest reliability level while minimizing investment costs, and use this as a reference for subsequent engineering construction.

[0009] Although my country has made some progress in the reliability of distribution networks in recent years, there are still the following defects in the technical fields of evaluation algorithms, system reliability identification, switch configuration optimization, and integrated modeling and evaluation integration:

[0010] (1) Technical defects in the research of reliability assessment algorithms: In the research of distribution network reliability assessment algorithms, my country's current research is mainly focused on the theoretical algorithm level, and there is still a certain gap in practical application. Compared with the refined statistical methods based on users abroad, my country's assessment is often based on medium-voltage desktop transformers, lacking comprehensive coverage of users of all voltage levels. This equipment-based statistical method results in insufficiently detailed assessment results and cannot accurately reflect the power supply situation of each user. In addition, my country has not yet effectively distinguished the power outage time caused by major accidents during the assessment process, which further weakens the accuracy and reliability of the assessment results.

[0011] (2) Technical defects in identifying weak links in system reliability and improving strategies: In terms of identifying weak links in system reliability and studying improvement strategies, research on my country's distribution network is relatively limited. The current statistical methods have not fully explored the impact of each component, and there is a lack of effective reward and punishment mechanisms as a driving force, resulting in the protection and replacement strategies for weak link equipment not being systematic and timely. When my country's power supply companies discover and improve weak links in reliability, they lack sufficient incentives and constraints, and improvement measures often rely on experience rather than data-driven precise analysis and decision-making.

[0012] (3) Defects of global optimal switch configuration technology: my country has not yet widely used intelligent computing tools to quickly find the global optimal switch configuration. Most distribution network projects still rely on the engineering practice experience of technicians in the early planning of switch configuration, lacking systematic optimization algorithm support, resulting in low switch configuration efficiency and difficulty in achieving a balance between power supply reliability and economy.

[0013] In general, my country's distribution network reliability assessment has obvious technical deficiencies in algorithm research, identification and improvement of weak links in system reliability, and optimal switch configuration. Summary of the invention

[0014] The purpose of the present invention is to provide a distribution network reliability assessment method based on load segmentation and an improvement control strategy to solve the technical problems existing in the reliability assessment of my country's distribution network in the aspects of algorithm research, identification and improvement of weak links in system reliability, and optimal switch configuration.

[0015] To achieve the above objectives, the specific technical solutions of the load segmentation-based distribution network reliability assessment method and the control strategy improvement are as follows:

[0016] The distribution network reliability assessment method based on load segmentation includes the following steps:

[0017] S1. Establish an improved distribution network reliability assessment model based on load block technology for switch elements, and input fault parameters, load parameters, system topology parameters, and load branch parameters;

[0018] S2. Form a system topology diagram according to the line node branch numbers;

[0019] S3, partitioning the distribution network according to the system topology diagram;

[0020] S4, introduce the fuse factor and calculate the block fault parameters;

[0021] S5, enumerate the fault blocks to form a fault consequence pattern table;

[0022] S6. Determine the power supply reliability of the areas divided in S3, set fuses for the required areas, and supplement the influence of the fuse branches on the load;

[0023] S7, enumerate all fault blocks;

[0024] S8. Use Bayesian network method to find the weak links of distribution network;

[0025] S9. Calculate system reliability indicators.

[0026] As a further improvement of the present invention, the fault parameters in S1 include component failure rate λ, component repair rate μ, fault repair time r=1 / μ, time before failure d=1 / λ, failure frequency f=1 / (d+r) and unavailability rate U=λ / (λ+μ); the load parameters include user type, number of users, load point power peak and load point power average.

[0027] As a further improvement of the present invention, in S3, the switch elements are used as boundaries, and the breadth-first traversal algorithm of the tree is used to visit adjacent nodes step by step starting from the root node of the system topology structure until the entire network is traversed, so as to partition the distribution network into element groups and load areas.

[0028] As a further improvement of the present invention, in S5, the load area types after the distribution network failure are divided into four categories, ABCD, according to the power outage duration of load points in different positions when the component fails. By enumerating the block failures in sequence, the fault type of each load point caused by each component failure can be obtained, that is, the fault consequence pattern table.

[0029] As a further improvement of the present invention, in S8, the Bayesian network method is used to analyze the weak links of the load points, and the BFSIDI index of the distribution network is selected for analysis. The calculation formula of the BFSIDI index is: Where N i is the number of users at the ith load point, U ikis the time it takes for the i-th load point to restore power supply when the k-th component fails, and nload is the number of all load points; the weak link of the distribution network is determined based on the shared amount.

[0030] As a further improvement of the present invention, the system reliability index in S9 includes: system average power outage frequency index System average power outage duration indicator Average power outage duration indicator for users Average power supply availability index Total power loss Total power outage loss LOSS = ENSI * average electricity price.

[0031] The improved control strategy of the distribution network reliability assessment method based on load segmentation includes the following steps:

[0032] S2.1. Set the initial value of the algorithm in the improved distribution network reliability assessment model, input the initialization data, and generate a random initial population;

[0033] S2.2, calculate the fitness function value, that is, calculate the objective function value, and set the fitness to infinity for the scenario that does not meet the constraint conditions, so as to eliminate the individual in the iteration;

[0034] S2.3, use the bidding algorithm to construct the Pareto non-inferior solution set;

[0035] S2.4. Calculate the crowding distance of the particles in the Pareto non-inferior solution set. The crowding distance of a particle is the minimum difference between the particle and other particles in the fitness function value.

[0036] S2.5, trim the Pareto non-inferior solution set and delete particles with small crowding distance;

[0037] S2.6. Select individual and group extreme value particles, and select the best performing individuals and groups from the current solution set;

[0038] S2.7. Calculate the update formula of the random number m generated by each particle, the weight ω and the learning factor c1c2. Set the constant threshold p∈[0,1] as the capture particle x of the three black hole system. i The probability that for each particle x i Each iteration generates a random number m∈[0,1];

[0039] S2.8, particles satisfying m≤p are captured by the three-black hole system, and the rest are updated in the traditional way;

[0040] S2.9, forming new particle positions and velocities;

[0041] S2.10, premature disturbance judgment is performed on the particle, when |Fk -F k-1 |<0.01|F k |, reset the position of the particle so that it can potentially jump out of the local optimum, where F k Represents the objective function value corresponding to the global optimal particle of generation k;

[0042] S2.11, if the iteration termination condition is met, the optimal Pareto non-inferior solution set is output; if not, return to step S2.2 and iterate again to find the optimal solution;

[0043] S2.12. Use the TOPSIS method to sort the individuals in the Pareto archive set and select the optimal solution.

[0044] As a further improvement of the present invention, the initialization data in S2.1 includes the particle position x, the particle search speed v and the current particle position p.

[0045] As a further improvement of the present invention, the bidding algorithm in S2.3 is as follows: in each round, two particles are selected to compare the dominance relationship, and the winner continues to compete with the remaining particles in the group, and a particle that wins the competition with all particles in the group is obtained, which is saved in the Pareto non-inferior solution set and deleted from the group; the competition is restarted until the particles in the group are completely deleted, thereby completing the establishment of the Pareto non-inferior solution set.

[0046] Beneficial effects:

[0047] By dividing the load nodes in the distribution network into multiple independent load blocks, the complex network structure is simplified, making the subsequent analysis more efficient and accurate. This method can more accurately reflect the actual operation status of the distribution network.

[0048] Input fault parameters, load parameters, system topology parameters and load branch parameters to ensure the comprehensiveness and accuracy of the assessment model and improve the reliability of the assessment results.

[0049] Introducing fuse factors and calculating block fault parameters can effectively reduce the fault scope, avoid fault spread, and improve system reliability.

[0050] By enumerating the fault blocks and forming a fault consequence pattern table, we can fully understand the impact of each component failure on each load point and provide basic data for subsequent optimization.

[0051] The Bayesian network method is used to analyze the weak links of the load points, and the BFSIDI indicator is selected for analysis to accurately identify the weak links of the system and improve the overall reliability of the system.

[0052] Taking the switch elements as the boundary, the breadth-first traversal algorithm of the tree is used to partition the distribution network into element groups and load zones, which helps to reasonably allocate resources and improve the operation efficiency of the system.

[0053] The particle swarm optimization (PSO) algorithm is used, combined with the bidding algorithm and the Pareto non-inferior solution set, to achieve multi-objective optimization, ensuring that both economy and practicality are taken into account while improving system reliability.

[0054] Generate a random initial population to ensure that the algorithm can search globally and improve search efficiency.

[0055] The particles are judged for premature disturbances and their positions are reset when the conditions are met, allowing them to potentially jump out of the local optimum, thus avoiding premature convergence of the algorithm and improving computational efficiency and accuracy.

[0056] It is suitable for reliability assessment and improvement of single-power radial distribution networks, and is particularly suitable for improving the reliability of downstream load points.

[0057] The method and strategy of the present invention have good scalability, can be applied to distribution networks of different types and sizes, and adapt to various complex scenarios.

[0058] Calculate the system average outage frequency index (SAIFI), system average outage duration index (SAIDI), user average outage duration index (CAIDI), average power supply availability index (ASAI), total energy loss (ENS), total outage loss (CENS), etc., provide comprehensive evaluation results, and help decision makers formulate effective management strategies.

[0059] The reliability of the distribution network is gradually improved through a multi-stage optimization strategy, including setting fuses, optimizing switch configuration, adding interconnection lines and other measures to ensure the long-term stable operation of the system.

[0060] According to the actual operation of the system, dynamically adjust the optimization strategy to improve the adaptability and flexibility of the system.

[0061] In summary, the present invention provides a distribution network reliability assessment method and improvement control strategy based on load block technology, which comprehensively improves the reliability assessment accuracy and management level of the distribution network through multiple means such as multi-parameter input, load block technology, multi-objective optimization, Bayesian network method, etc. This method can not only accurately identify the weak links of the system, but also provide effective optimization strategies to ensure the safe and stable operation of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 It is a flow chart of a method for evaluating the reliability of a distribution network based on load segmentation according to the present invention;

[0063] Figure 2 It is a flow chart of the control strategy for improving the reliability assessment method of distribution network based on load segmentation;

[0064] Figure 3 It is the wiring diagram of IEEE RTBS ​​Bus6 subsystem;

[0065] Figure 4 This is the network topology diagram of Yangmiao substation 10kV041 Yangmiao 112 line;

[0066] Figure 5 This is the network block topology diagram for Yangmiao to Yangmiao 112 line. DETAILED DESCRIPTION

[0067] In order to better understand the purpose, structure and function of the present invention, the following is a further detailed description of the distribution network reliability assessment method based on load segmentation and the improvement control strategy of the present invention in conjunction with the accompanying drawings.

[0068] Implementation example:

[0069] The load-block-based distribution network reliability assessment method and improvement control strategy of the present invention are intended to solve the problems existing in the algorithm research, identification and improvement of weak links in system reliability, and optimal switch configuration in my country's distribution network reliability assessment.

[0070] According to the characteristics of the distribution network, the number of switches is far less than the number of components. The distribution network is divided into small areas with switches as the boundary. The types of areas are mainly component groups and load areas. Then the improved FMEA method is used to analyze the impact on the reliability of the load area for the component group. According to the power outage duration of load points at different locations when the component fails, the load area types after the distribution network failure are divided into four categories of ABCD. The block faults are enumerated in turn, and the influence of the fuse is considered. The fault type of each load point caused by each component failure can be obtained, that is, the fault consequence mode table. According to the fault consequence mode table and the fault parameters of the system components, the reliability index of the load point with the distribution transformer as the terminal can be calculated, and then the reliability index of the distribution network system can be obtained.

[0071] While calculating the load reliability index, the number of households affected by power outages caused by each component failure is counted and accumulated into the corresponding status statistics of the system component. The conditional probability of diagnostic reasoning is equal to the ratio of the cumulative number of households affected by power outages of the component to the total number of households affected by power outages of the system. With the same investment funds, the improvement effect will be more significant if the weak link equipment of the system is protected or replaced.

[0072] This scheme adopts an improved multi-objective hybrid particle swarm algorithm. This scheme improves the particle swarm algorithm by introducing a three-black hole system capture strategy and a premature perturbation strategy, and then combines the TOPSIS algorithm to obtain the global optimal switch configuration. This method strengthens the local exploration capability and improves the convergence accuracy of the algorithm. The number of switches configured is fixed, the switch configuration position is the particle position, and the number of switches is the particle dimension. Each configuration ensures that the values ​​of the particles in each dimension are different. The improvement of system reliability is reflected in the reduction of power outage losses in economic terms, that is, the total power shortage of the system power outage is minimized. The minimum total power shortage of the system power outage is used as the objective function of the switch configuration to obtain the global optimal switch configuration.

[0073] Reliability in a broad sense refers to the ability or possibility of components, products, and systems to perform specified functions without failure within a certain period of time and under certain conditions. The reliability of urban distribution network power supply refers to the degree of reliability with which users can obtain electric energy from the power system. The reliability of the power system reflects the ability of the power system to supply electricity to users, and the measurement indicators are mainly the quality and uninterrupted degree of the supplied electric energy. From the definition, it can be seen that the reliability of the power supply system is usually evaluated from two aspects, namely safety and adequacy. Safety measures the anti-interference ability of the power system when some components fail; adequacy measures the ability of the power system to continue to supply electric energy to users and meet the power needs of users when components are out of service or load fluctuates. Before conducting a reliability assessment of the distribution network, it is necessary to establish a reliability assessment model and indicators as a basis. The model should be able to reflect the structure, characteristics, operating status, failure mode and impact on users of the distribution network and its equipment.

[0074] The following characteristics should be considered when defining power system reliability indicators:

[0075] A. Based on the observation records of the power system operating status and through computer statistical evaluation, the reliability indicators of various equipment are obtained to evaluate the reliability of the power system.

[0076] B. Existing reliability theory models can be used for evaluation. At present, different countries have defined a variety of reliability indicators according to their own national conditions. According to the research level, the evaluation indicator system can be divided into component level, load level, and system level.

[0077] a. Component-level reliability indicators:

[0078] There are two main states of components in the distribution network, namely normal operation and shutdown. The failure process refers to the process in which the component changes from the normal operation state to the shutdown state. The probability of this process is called the failure rate λ (failure times / year); the opposite is called the repair process. The probability of the repair process is called the repair rate μ (repair times / year). The time when the component is in normal operation is called the component failure-free time, and the time when the component remains in the fault state is called the fault repair time. The length of the above two times is related to the component type and the degree of aging of the component. The degree of freedom of the reliability parameter of the component is 2, that is, if two of the parameters are known, the remaining parameters can be inferred. The above indicators can be used to infer the component fault repair time (year / time), the time before component failure (year / time), the component failure frequency (failure times / year), and the component unavailability rate (1).

[0079] b. Load level reliability index:

[0080] The calculation of load indicators is the main content of reliability assessment. The main indicators for evaluating load reliability include annual power outage rate, annual power outage time and fault repair time.

[0081] Power outage time r of load node i ik The value of depends on the structure of the system:

[0082] (b.1) The i-th load node is on the power supply side

[0083] If there is a switch between the kth component and the i-th load node, the power outage time of the load node i is the switch operation time; if there is no switch between the kth component and the i-th load node, the power outage time of the load node i is the equivalent repair time of the kth component.

[0084] (b.2) The kth element is on the power supply side

[0085] If there is a switch between the kth element and the i-th load node, when the kth element fails, if there is a tie switch behind the i-th load node, then the power outage time of the load node i is the larger value of the two switch operation times; if there is a switch between the kth element and the i-th load node, if there is no tie switch behind the i-th load node, then the power outage time of the load node is the equivalent repair time of the kth element. If there is no switch between the kth element and the i-th load node, then the power outage time of the load node i is the equivalent repair time of the kth element;

[0086] c. Network reliability indicators

[0087] After calculating the reliability index of each load point in the network, we can obtain the reliability index of the entire network, including the system average power outage frequency index, the system average power outage duration index, the user average power outage duration index, the average power supply availability index, the user average power outage power, the total power loss and other reliability indicators.

[0088] (2) Reliability evaluation research methods

[0089] Simulation method and analytical method are two commonly used reliability assessment methods in engineering. The simulation method can calculate the impact of various related events on system reliability. Among them, the more representative one is the Monte Carlo simulation method. Its advantage is that the system scale has little impact on the algorithm, and it has a good effect when calculating the reliability of complex networks; but its disadvantages are also obvious. When the reliability required by the system is high, in order to achieve a higher calculation accuracy, the simulation method takes a long time in the process of calculating the system reliability, resulting in low calculation efficiency. The analytical method is convenient and fast in analyzing the impact of various components on reliability in the system due to its accurate modeling and simple principle. Therefore, it is widely used in distribution network reliability assessment. Commonly used methods in analytical methods include the minimum path method, the minimum cut set method, the network equivalent method, and the fault traversal method. The minimum path method takes the load point as the research object, simplifies the network, first obtains the minimum path between the power supply and each load node, then the components on other paths are equivalent to this path, and finally only calculates the impact of the failure of each component on the minimum path on the load reliability index. The minimum path is oriented to the load point. The disadvantage is that the minimum path between the load point and the power source overlaps, and as the scale of the power system increases, it is very difficult to establish the minimum path. The minimum cut set method is similar to the minimum path algorithm. It uses the depth-first search algorithm to determine the minimum path from the load point to the power source point (if the channel does not pass through the same node or intersection more than twice, then the path between the two nodes is the smallest), and according to the determined minimum path, the minimum connected set (the minimum subset of the set of components that ensures the normal operation of the system is called the minimum connected set) is determined. The relationship between each minimum connected set is parallel, and the relationship between each connected set branch is series. The minimum cut set is formed according to the connection set branch association matrix, and the minimum cut set is processed without intersection. This method only needs to analyze the state that causes power loss, reduces the amount of calculation, and can calculate second-order and above faults. The disadvantage of the minimum cut set method is that for complex distribution systems, due to the large number of components in the system and the large order of the association matrix, it is difficult to obtain the minimum cut set, and the non-intersection processing of the minimum cut set is very complicated. The network equivalence method can transform complex power systems into simple power systems for processing. When the upward equivalent is used, the failure rate calculation formula of the series network is used to calculate the failure rate and fault repair time of the equivalent branch line. When the downward equivalent is used, the upper feeder is used as the series element of the equivalent feeder to obtain the expected value of the reliability index of the node. For the lower feeder, the reliability parameter of the first-end series element is the expected value of the upper node. This method cannot obtain the reliability index of the load point and the entire network at one time. It is necessary to repeat the above process, and then obtain the reliability index of the load point in the lower feeder, and then calculate the system reliability index. The fault traversal algorithm enumerates each fault, determines the fault impact range, and calculates the reliability index.This method can be used to divide the subsystem after the fault into several simple subsystems, and it is easy to observe the impact range of a component failure and the degree to which the negative reliability load is affected by different fault components. However, the location of the fault is different, and its equivalent block subsystem is different, which greatly increases the amount of calculation. From the above, it can be seen that the above traditional analytical methods have corresponding shortcomings. At the same time, as the scale of the system increases, the number of components increases, the fault categories of the system also increase, and the fault state of the system will be many, which greatly increases the amount of calculation. Therefore, it is necessary to study an analytical algorithm that can quickly handle large-scale distribution network systems. This scheme proposes an improved reliability algorithm, which adopts block technology and divides the component area and the load area with the switch element as the boundary to simplify the distribution system structure. The principle is simple, easy to program and implement, and there is no need to perform equivalent processing on the system. A fault traversal algorithm is used to form a fault consequence pattern table. When it is necessary to calculate the reliability of a certain load point, the Bayesian network method is used for diagnostic reasoning to obtain the ranking of components that affect the reliability of the load point, omitting the process of analyzing all component failures. The calculation is simple and the key components of the network are found intuitively and clearly. Compared with the traditional analytical method, this method has a small programming workload and a high program execution efficiency, which is suitable for application in engineering practice.

[0090] 2) Component failure rate prediction research

[0091] The original parameters of components, such as failure rate, are very important in the reliability assessment of distribution networks, and their accuracy will have a direct impact on the calculation results of reliability analysis. The failure rate of components is not a fixed value. It is affected by many factors such as component type, operating environment, and aging degree of components. The reliability parameters of components in other projects in the past can only be used as a reference. If directly quoted, it will bring unnecessary errors to the final results. Therefore, it is necessary to accurately predict and calculate the failure rate of each component in the network, which is an important basis for obtaining accurate system reliability assessment results. This scheme studies the impact of climate factors on the reliability of the power system. Based on gray correlation and fuzzy mathematics theory, fuzzy differences are formed comprehensively, and then known components with smaller fuzzy differences are selected to predict the failure rate of unknown components. The failure rate obtained at this time is a constant value. In reality, it is also necessary to consider the impact of time aging, equipment transformation, weather factors, etc. on the failure rate of components, so a time-varying failure rate model is established.

[0092] (1) Grey correlation principle

[0093] Compared with traditional correlation analysis methods, grey correlation analysis has lower sample requirements and a wider range of applications. At the same time, due to its small amount of calculation, it can be more easily combined with other algorithms. At present, this method has been successfully applied to prediction, comprehensive evaluation, control, pattern recognition and other fields, and is also widely used in the field of power system research. The grey correlation degree is involved in the power system fields such as power consumption correlation analysis in various industries, primary equipment fault diagnosis, and project investment decision-making. The grey correlation principle also plays an important role in the evaluation index system of the construction scale of the distribution network.

[0094] (2) Fuzzy closeness principle

[0095] If any element in the domain U has a membership function μ A The corresponding membership function μ on u A (u) reflects the membership of element u to a fuzzy subset A on the domain U. A The closer the value of (u) is to 1, the higher the degree to which element u belongs to fuzzy subset A, and vice versa. In other words, fuzzy subset A on the domain U refers to all elements with certain properties in U, and these elements have some fuzzy boundaries. To determine the degree of relevance of an element with a certain property in the domain U to subset A, its membership degree can be used to describe the relevance of the element. Pattern recognition is a method used to identify specific object types. When describing or characterizing the characteristics of objective things, there is usually a certain degree of fuzziness. Fuzzy pattern recognition is a recognition pattern performed by fuzzy mathematical methods. In the process of feature recognition, it is necessary to quantify the similarity, so fuzzy distance is introduced to describe the similarity between fuzzy sets. The larger the fuzzy distance, the smaller the similarity between A and B; the fuzzy closeness is introduced to make it positively correlated with the similarity. The larger the value, the greater the similarity between the two.

[0096] (3) Fuzzy difference

[0097] As can be seen from the previous introduction, the similarity between two factors can be expressed by gray correlation or fuzzy proximity, but neither of them can perfectly express the similarity between factors and has its own defects. In order to solve the above problems, the two theories are combined and a new index is constructed. First, the gray correlation is improved and converted into difference. The smaller the value, the greater the similarity between things. Then the difference is added to the fuzzy distance and the average is taken to obtain a new index, which is called fuzzy difference. Suppose there are n subsets on the domain U, and the identified object B is also a set on U. If you want to find which subset B belongs to, first calculate the fuzzy difference between object B and each subset, and sort the calculated series of fuzzy differences, and then find the minimum value among them. Then object B belongs to the subset with the smallest fuzzy difference.

[0098] (4) Line failure rate

[0099] Taking the feeder as an example, the steps of using fuzzy difference to obtain its failure rate are as follows: ① If the reliability parameters of a line are known, the factors that affect its reliability parameters are set as reference objects; ② For lines with unknown reliability parameters, the factors that affect its reliability parameters are set as control objects; ③ Calculate the fuzzy distance and grey correlation between the reference object and each control object respectively, and combine these two indicators to obtain the fuzzy difference; ④ According to the proximity and weighted principles, calculate the line failure rate and other reliability parameters.

[0100] (5) Time-varying failure rate

[0101] The failure rate obtained above is still a constant value, but in reality, the failure rate of components is time-varying. The main factors affecting the change of the failure rate of electrical components are time aging and preventive maintenance. Traditional theory believes that the life of components follows an exponential distribution and its failure rate is constant, which is obviously inconsistent with the actual situation. This solution uses the widely used two-parameter Weibull distribution model to consider the aging factor, while taking into account the influence of preventive maintenance and weather conditions.

[0102] 1) Establishment of reliability assessment model

[0103] (1) Reliability evaluation model

[0104] The traditional FMEA (Failure Effect Mode Analysis) reliability criterion is to determine the impact of all component failures on the reliability of all load points. For complex distribution networks with a large number of distribution components, the calculation efficiency is low. Analyzing the fault judgment criteria of FMEA, the impact of components on load points is closely related to the connection method between the two (circuit breaker or section switch), and combined with the structural characteristics of the distribution network, the number of switches is significantly less than the number of components. Therefore, the distribution network is partitioned by switching components to form component groups and load areas. At the same time, in order to reduce the dimensionality and complexity of the research problem and improve the evaluation efficiency, a component group-oriented analysis method is used to analyze the impact of load area reliability.

[0105] A. Load zoning method for switch elements From the load to the nearest circuit breaker or section switch, there is only one switch element in the area, and different areas cannot contain the same switch element. All elements in the area form an element group, and all load points in the area form a load zone. This method is an extension of the traditional FMEA method, and the reliability criteria are still applicable. Here, the element group is regarded as an independent element, and the reliability parameters of the element group can be obtained from the reliability parameters of each element in the group. For the entire element group, the failure of any element in it has the same impact on the power outage of the load zone.

[0106] B. Reliability Criteria

[0107] The idea of ​​reliability assessment for power systems is to determine whether the load node is connected to the power supply. If there is no connection, determine whether the power supply can be restored and calculate the time to restore the power supply. Due to the existence of protection components, disconnectors and backup power supplies in the system, the duration of power outages at load points in different locations is not the same when components fail. The regional failure consequence types that lead to distribution network failures are classified as follows:

[0108] Class A: When a fault occurs, the load power supply in the area is not affected;

[0109] Class B: When a fault occurs, the load area can be restored to power after the fault is isolated. The power outage time of this load area is the time of isolating the fault.

[0110] Class C: When a fault occurs, after isolating the fault, there is still no path to the power supply. The load area needs to be restored after the tie switch is actuated. The power outage time is the sum of the tie switch operation time and the fault isolation time.

[0111] Class D: When a fault occurs, the load area must wait until the fault is resolved before power can be restored. The power outage time is the fault repair time.

[0112] From the above load zone classification definition, we can see that after the distribution system fails, the type of load fault is determined by the position of the action switch. The classification process of the fault area is as follows:

[0113] a. Locate the faulty component;

[0114] b. Locate the actuated circuit breaker to achieve wide-range fault isolation;

[0115] c. Locate various switches and determine their locations in the smallest fault area, thereby isolating the small fault area;

[0116] The purpose of operating various switches is to isolate the fault, minimize the scope of power outage and fault area, and reduce the impact of the fault on the load. The detailed processing process is as follows:

[0117] Fault handling after a fault occurs is usually divided into two steps: first, the circuit breaker operates to remove the fault. At this time, the area with uninterrupted power supply can be directly determined as a Class A area; if the power supply in this area is interrupted, the type of fault consequence cannot be immediately determined and is classified as an area to be processed; the second step is isolation and maintenance after fault location. When the disconnector operates to isolate the fault, the circuit breaker and the contact switch are operated, and then the main and backup power supplies are supplied to the non-fault area, and the type of fault consequence of the area to be processed is determined by the load power supply recovery situation. Therefore, only through the joint analysis of the two steps can the type of fault consequence of each load point be fully determined. The division process of the above 4-level components is as follows: After a system fault occurs, the search starts from the branch where the faulty component is located and searches backward along the feeder, and stops when the first circuit breaker is found. Then the load point searched before the circuit breaker is Class B, C or D, and the other load points that have not been searched are Class A load points, that is, normal load points. The determination process of class B, C or D load is as follows: disconnect the searched disconnector. If the load area can be connected to the power supply, all nodes in the area are class B; if the load area is not connected to the power supply, but there are nodes connected to the backup power supply, all nodes in the load area are class C; if the area is not connected to the power supply or the backup power supply, all nodes in the area are class D. By enumerating the block faults in sequence, the fault consequence pattern table can be obtained. Since the fault type caused by the switch element failure is the same as the block element failure after the switch, the size of the fault consequence pattern table can be reduced by placing the switch element in the block. Considering the switch failure, the reliability index of the load can be obtained according to the fault consequence pattern table and the fault parameters of the system components, and then various reliability indicators of the network can be obtained. Since there is no equivalence and approximation in the calculation process, the calculation result is accurate.

[0118] (2) Identification of weak links

[0119] Bayesian network is a new type of artificial intelligence algorithm that appeared in the late 1980s. It solves problems such as causal reasoning and diagnostic reasoning through probabilistic reasoning. It uses hypothesis analysis and diagnostic reasoning to find out the weak links in the entire system. The Bayesian network model for power system reliability assessment is composed of a logical "AND" node model and a logical "OR" node model. The series system is a logical "AND" contact model, and the parallel system is a logical "OR" contact model. The Bayesian network model can propose effective improvement measures for weak links and improve the reliability and economic benefits of the distribution network. When an actual fault event occurs, according to the fault type of the load point, when it is known that the average fault repair time of a component is the largest, if a load point in the system has a power outage, the probability caused by the failure of the component is the largest, that is, the probability that the component is the weak link is the largest. However, since the network structure has a great impact on the system reliability, the same component has a different position in the network or a different relative position with each switch, and the impact on the system reliability is also quite different. Therefore, it is difficult to identify the weak links of the entire distribution network by simply analyzing the weak links of the load points. For this purpose, the BFSIDI (number of households during system outage caused by components) indicator of the distribution network is selected for analysis, and the weak links of the distribution network are determined according to the apportioned amount.

[0120] The reliability analysis of the IEEE RTBS ​​Bus6 system is verified. This paper uses the subsystem in the standard example, which consists of feeders F4-F7 as follows Figure 3 shown.

[0121] Through the preliminary reliability assessment of the distribution network and the analysis of the load points and system fault distribution, the reliability can be improved from the following three aspects: first, improve the reliability parameters of the equipment, such as targeted replacement of low-reliability equipment, increase of lightning protection equipment, reinforcement of poles and towers, thickening of lines, etc.; second, shorten the fault time, increase the connection between lines, realize the transfer of loads in the fault area, and shorten the time to restore power supply, but it is necessary to calculate whether the power flow after the transfer can meet the requirements; third, reduce the scope of the fault, such as adding fuses and optimizing switch configuration, which can quickly remove the faulty components and minimize the impact range.

[0122] 1) Basic promotion strategy

[0123] (1) Add a fuse

[0124] In practice, some distribution transformers with smaller capacity are not equipped with fuses for economic reasons. According to the reliability criterion, load point failures can be divided into the influence of the fault of the branch itself and the fault of other branches. If the fault of the distribution transformer branch cannot be isolated in time, it will cause the fault of the load point of other load branches, and then the reliability of the whole system will be reduced. Therefore, it is possible to reasonably equip this part of the distribution transformer with fuses, which can effectively reduce the scope of the fault.

[0125] (2) Quantification of measures to improve component reliability

[0126] Technical means of grid transformation to improve power supply reliability: including cable transformation, increasing line segmentation, increasing branch line protection, distribution network automation transformation and other technical measures, as well as management measures such as reducing equipment downtime rate and downtime. Among them, increasing line segmentation and realizing dual power supply are effective measures and methods to improve reliability. According to the actual line structure, the rectification measures mainly include replacing equipment (transformer), replacing wires (thickening lines), adding fuses (circuit breakers), adding section switches and adding lightning protection equipment (arrester), and strengthening towers.

[0127] (3) Improve the weak links of the system

[0128] By using the proportion of households that experience power outages due to component failures, we can identify the weak links in the system and improve system reliability in a targeted manner. With the same investment funds, we can first protect or replace the equipment in the weak links, and the improvement effect will be more significant. This transformation measure can be achieved by reducing the failure rate of the replaced equipment.

[0129] (4) Adding contact lines

[0130] According to the preliminary reliability evaluation results, for a single-power radial distribution network, the smaller the electrical distance between the power source and the load point, the higher the reliability. This phenomenon can be explained by the reliability criterion. When the upstream component fails, it will definitely cause the downstream component to fail, and the power outage time is the fault recovery time of the component. On the contrary, the downstream component failure will be isolated under the action of the switch, and will not affect the upstream load point. In addition, the reliability of the load points at the head and end of the main feeder is quite different. If the reliability of the downstream load point can be improved, the system reliability will be significantly improved. Therefore, it is proposed to add a tie line at the end of the main feeder to achieve timely transfer of the downstream load point when the upstream fails, which can effectively reduce the fault power outage time of the downstream component. In addition, the improvement effect of adding tie switches at different positions in the system can be explored, and better improvement measures can be proposed.

[0131] (5) Optimize switch configuration

[0132] Switch optimization configuration is a multi-objective optimization problem, which takes reliability and economy as the objective functions, so that the configuration scheme can achieve the lowest cost while improving reliability.

[0133] The following takes the Yangmiao 10kV Yangmiao 112 line as an example. There are 33 transformers in total, 26 public transformers with a total capacity of 2135kVA, and 7 user transformers with a total capacity of 200kVA. This line is a typical radial distribution network with a long main feeder and a complex structure, including multiple branch lines. Therefore, this line is selected for reliability improvement control strategy analysis. The network topology diagram of this line is as follows Figure 4 shown.

[0134] According to the reliability meter evaluation model analysis, this plan proposes several basic transformation measures, which can be combined according to the actual situation. The methods are flexible and varied. The basic measures that can be implemented are as follows:

[0135] Measure 1: For distribution transformers that are not equipped with fuses, add fuses to reduce the fault range;

[0136] Measure 2: Add lightning protection equipment, strengthen towers, thicken lines, and reduce component failure rates;

[0137] Measure 3: Replace equipment based on the weak links of the Bayesian identification system;

[0138] Measure 4: Add a contact switch to realize load transfer in case of fault and shorten the fault time;

[0139] Measure 5: Add a limited number of disconnectors and use the improved particle swarm-TOPSIS combined algorithm to optimize the switch configuration and find the global optimal solution.

[0140] like Figure 1 The distribution network reliability assessment method based on load partitioning is shown in the figure, and the detailed steps are as follows:

[0141] Step 1: Input fault parameters, load parameters, system topology parameters, and load branch parameters.

[0142] Fault parameters include: component failure rate λ, component repair rate μ, fault repair time r=1 / μ, time before failure d=1 / λ, failure frequency f=1 / (d+r) and unavailability rate U=λ / (λ+μ), etc.; load parameters include: user type, number of users, load point power peak, load point power average, etc.; other parameters include: feeder type, length, number, etc. The basic reliability parameters of the components of Yangmiao 112 line of Yangmiao substation, the length parameters of the line, and the basic parameters of the load are shown in Table 1, Table 2, and Table 3.

[0143]

[0144] Note: Line failure rate: ((times / year*km)

[0145] Table 1 Basic reliability parameters of Yangmiao substation Yangmiao 112 line

[0146]

[0147] Table 2 Line length parameters of Yangmiao to Yangmiao 112 line

[0148]

[0149]

[0150] Table 3 Load parameters of Yangmiao substation Yangmiao 112 line

[0151] Step 2: According to the line node branch number, form a system topology diagram. The system topology diagram is as follows: Figure 5 shown.

[0152] Step 3: Using switches as boundaries, plan partitions by traversing the tree in breadth-first order.

[0153] Taking the switch element as the boundary, using the tree breadth-first traversal algorithm, starting from the root node of the system topology structure, visit all the neighbor nodes of the node, then visit all the neighbor nodes of these neighbor nodes, and so on, to partition the distribution network into element groups and load areas. The detailed results of system partitioning can be found in Figure 4 .

[0154] Step 4: Introduce the fuse factor and calculate the block fault parameters.

[0155] Step 5: Enumerate the fault blocks and form a fault consequence pattern table.

[0156] According to the power outage duration of load points at different locations when the component fails, the load area types after the distribution network fault are divided into four categories: ABCD. By enumerating the block faults in sequence, the fault type of each load point caused by each component failure can be obtained, that is, the fault consequence mode table.

[0157] Step 6: Determine whether it is necessary to supplement the fuse branch to determine the impact on the load.

[0158] If necessary, supplement the fuse branches for the effects on other loads.

[0159] If not required, skip to step 7.

[0160] Step 7: Enumerate all faulty blocks.

[0161] If yes, proceed to the next step.

[0162] Step 8: Use Bayesian network method to identify weak links.

[0163] The Bayesian network method is used to analyze the weak links of the load points, and then the BFSIDI (number of households when the system power outage is caused by the component) index of the distribution network is selected for analysis, and the weak links of the distribution network are determined according to the apportionment amount. The calculation formula of this index is: Where N i is the number of users at the i-th load point; U ik is the time it takes for the i-th load point to restore power when the k-th component fails; nload is the number of all load points.

[0164] Step 9: Calculate system reliability index.

[0165] System reliability indicators include: System average power outage frequency index (times / household / year), system average power outage duration index (hours / household / year), average power outage duration indicator for users (hours / power outage household / year), average power supply availability index (%), total power loss (MWh / year / household), total power outage loss LOSS = ENSI*average electricity price (10,000 yuan / year).

[0166] The reliability evaluation model proposed in this scheme is used to evaluate the IEEE RTBS ​​Bus6 subsystem. The load reliability calculation results without considering the switch and busbar faults are consistent with the calculation results of relevant literature and the results of manual calculation, indicating the correctness of the calculation model. At the same time, the calculation results show that the closer the load point is to the power supply end, the higher the reliability. For the load point far away from the power supply, since there is no tie line in the system, when a fault occurs, even if there is a switch isolation fault in the system, the power supply of the load cannot be restored. The reliability indicators of the loads in the same block are roughly the same, and the difference is caused by the different fault parameters of the components on the fuse branch.

[0167]

[0168] Table 4 Load point reliability index

[0169] The above reliability assessment process is calculated without considering bus failure, switch refusal to operate, and standby in the distribution system. When considering switch failure, unreliable fuse blowing, and transformer standby, the comparison of the assessment results of the system reliability index is shown in the following table. Compared with the basic assessment results, it can be seen that: by increasing the transformer standby, the annual average power outage time of users and systems can be greatly shortened, and the power supply availability of the system can be improved. The disadvantage is that it has little effect on the power outage frequency index of the system; on the other hand, adding fuses can effectively improve the power supply reliability of loads and the entire network. Therefore, whether the fuse can ensure reliable operation is very critical. Fuses are mainly installed on load branches, so a fuse can be configured on each load branch of the power system, which provides suggestions for power grid planning and transformation.

[0170] Then, the weak links of typical loads and systems were identified, and the top five weakest links were obtained. The weak links of typical loads are not exactly the same as those of the distribution system, but they have overlapping parts. The weak links of the loads are all transmission lines close to the loads, and the weak links of loads 18 and 23 are the same, which is determined by the fact that they belong to the same load block in the network structure. The component with the longest total repair time in the system is usually the weak link. Since the transformer repair time is much longer than other components, Line 35 is the weakest link in the entire system, and the simulation results are consistent with the actual situation.

[0171] In addition, after improving line 35, the overall reliability parameters of the distribution network have been improved most significantly; on the other hand, since line 48 has the highest failure rate in the distribution network system, improving it effectively increases the average power outage frequency of the system. However, line 48 is located at the end of the distribution system. After it fails, according to the failure consequence model, most of the loads in the system are Class B loads, and power supply can be restored by isolating the fault. Therefore, it is not the weakest link in the distribution system. Through the above analysis, it can be concluded that the simulation results of the example correctly and effectively reflect the weak links of the system.

[0172]

[0173] Table 5 System reliability indicators

[0174]

[0175] Table 6 Weakness identification table

[0176]

[0177] Table 7 System reliability indicators

[0178] This reliability assessment algorithm establishes an improved distribution network reliability assessment model for load block technology for switch elements. According to the power outage duration of load points at different locations when the element fails, the load area types after the distribution network failure are divided into four categories: ABCD. According to the fault consequence mode table and the fault parameters related to the system components, the reliability index of the load point with the distribution transformer as the terminal is calculated. Subsequently, the reliability assessment of the subsystem composed of feeders F4~F7 in IEEE RTBS ​​Bus6 was carried out. The results show that the results of this scheme are consistent with the calculation results of other methods and the results of manual calculation, which verifies the correctness of the calculation model of this scheme.

[0179] like Figure 2 The improvement control strategy of the distribution network reliability assessment method based on load partitioning is shown in the figure, and the detailed steps are as follows:

[0180] Step 1: Initialization: Set the initial value of the algorithm, input the original data, and generate a random initial population. The initialization data includes the particle position x, the particle search speed v and the current particle position p, some parameters of the corresponding power grid, the number of fixed switches and the switch configuration position, etc.

[0181] Step 2: Calculate the fitness function value, that is, calculate the objective function value. For scenarios that do not meet the constraints, set the fitness to infinity so that the individual can be eliminated in the iteration.

[0182] Step 3: Construct a Pareto non-inferior solution set. Using a competition algorithm, two particles are selected in each round to compare the dominance relationship. The winner continues to compete with the remaining particles in the group. If a particle is obtained that wins the competition with all particles in the group, it is saved in the Pareto non-inferior solution set and deleted from the group. The competition is restarted until all particles in the group are completely deleted, thus completing the establishment of the Pareto non-inferior solution set.

[0183] Step 4: Calculate the crowding distance of the particle in the Pareto non-inferior solution set. The particle crowding distance is the minimum difference between the particle and other particles in the fitness function value.

[0184] Step 5: Prune the Pareto non-inferior solution set and delete particles with smaller crowding distance.

[0185] Step 6: Select individual and group extreme value particles, and select the individuals and groups with the best performance from the current solution set.

[0186] Step 7: Each particle generates a random number m. The update formula of weight ω and learning factor c1c2 assumes a constant threshold p∈[0,1] as the three black hole system to capture particle x i The probability that for each particle x i Each iteration generates a random number m∈[0,1].

[0187] Step 8: Determine whether m≤p is satisfied. If so, particle x i Captured by a three-black-hole system, otherwise updated in the traditional way.

[0188] Step 9: Generate new particle position and velocity. Update the position and velocity of the captured particle.

[0189] Step 10: Perform premature disturbance judgment on the particle. When |F k -F k-1 |<0.01|F k |, reset the position of the particle so that it can potentially jump out of the local optimum. k Represents the objective function value corresponding to the global optimal particle of generation k.

[0190] Step 11: Determine the iteration termination condition. If it is satisfied, output the optimal Pareto non-inferior solution set. If it is not satisfied, return to the second step and start the next round of iterative optimization.

[0191] Step 12: Use the TOPSIS method to sort the individuals in the Pareto archive set and select the optimal solution.

[0192] The load reliability index calculation of the four measures for the Yangmiao substation 10kV Yangmiao 112 line is shown in Table 8-9.

[0193] Measure 1: Equip a fuse

[0194] There are 31 load points in this line, of which only 11 are equipped with fuses. The 11 load points connected to them are numbered as follows: 30, 23, 3, 12, 24, 8, 22, 31, 6, 20, and 16.

[0195] Fuses were added to the remaining 20 distribution transformers. The evaluation results are shown in Table 8 below:

[0196]

[0197] Table 8 Optimization of annual fault outage time for typical load points

[0198] Load points were selected in different blocks, and the percentage of reduction in annual fault outage time was calculated for each. The results show that after adding fuses, the outage time of all load points was directly or indirectly reduced, and the impact was significant, with an improvement of 30%.

[0199] Measure 2: Improve component reliability parameters

[0200] Method 1: Strengthen the tower; Method 2: Thicken the line and install lightning protection equipment. The corresponding reliability indicators are shown in Table 9 below.

[0201]

[0202] Table 9 Parameter loads for improving component reliability

[0203] It can be seen from the above calculation results that improving the reliability of components has a significant improvement effect.

[0204] Measure 3: Replace equipment for weak links

[0205] The number of households when a component failure causes a system power outage is used as the evaluation criterion, and the 20 weakest links in the system are numbered accordingly: transformer, 3: circuit breaker, 11: main feeder, 12: branch feeder.

[0206]

[0207] Table 10 System reliability after component replacement

[0208] The system is ranked according to the impact of the number of households caused by component failure when the system is out of service, and the equipment is replaced at the weakest link of the system; 30% and 60% of the weak links are replaced respectively, and the reliability indicators of the obtained system are normalized. The results show that the system reliability indicators are improved, the ratio of power outage time to power outage frequency is significantly reduced, the average availability of the system is improved, and the effect of replacing 60% of the equipment is better than replacing 30% of the equipment.

[0209] Measure 4: Add a contact switch

[0210] The improvement effect of adding tie lines at different locations is explored as follows. Location 1 (95# main feeder) and location 2 (77# main feeder) are taken respectively. The evaluation results are shown in Table 11.

[0211]

[0212] Table 11 adds the reliability index of the tie switch system

[0213] It can be seen from the above results that both schemes have effectively improved the reliability of the load point and the system, and the load point at the end of the main feeder has been improved very well, so the measure is effective. By comparing the tie lines at different locations, the 95# main feeder is closer to the end of the main feeder, so the result of adding tie lines to the 95# main feeder is better than the evaluation result of the 77# tie line. This is related to the line structure. Adding tie lines to the 95# main feeder can achieve a wider range of load transfer in the event of a fault, so the location selection of the tie line also has a great impact.

[0214] Measure 5: Add a limited number of disconnectors and use the improved particle swarm-TOPSIS combined algorithm to optimize the switch configuration and find the global optimal solution.

[0215] Adding isolating switches can quickly isolate faults, reduce the scope of faults, and thus improve system reliability. However, the location of the switch configuration is very important. It is unrealistic to add switch devices to all lines. Therefore, it is very meaningful to study the optimal reliability under the premise of adding a limited number of switch elements. Taking the addition of 6 switch elements as an example, the evaluation results of the global optimal switch configuration achieved by the improved particle swarm algorithm are as follows:

[0216] As shown in Table 12.

[0217]

[0218]

[0219] Table 12 Disconnector Optimization Configuration Load Point Reliability

[0220] The global optimal switch configuration positions are: 52, 21, 13, 100, 36, 123. From the above results, reasonable configuration of isolation switches can effectively reduce the fault recovery time of each load point, thereby improving system reliability.

[0221] The comparison of system reliability indicators of four basic transformation measures and optimized switch configuration is shown in Table 13:

[0222]

[0223] Table 13 System reliability indicators under five measures

[0224] Comparing the original results with the calculation results of other transformation schemes, it can be seen that: all transformation schemes can greatly improve the reliability of system power supply, and all indicators of system reliability have been improved. The four reliability measures of this scheme are effective. At the same time, it can be seen that when the system is configured with fewer fuses, appropriately increasing the number of branch fuses can greatly improve the system reliability, but considering the full process cost of the equipment, although a large-scale increase in fuses can improve reliability, the cost is high. At this time, it is more reasonable to optimize the switch configuration. At the same time, for the distribution network that has been fully equipped with fuses, optimizing the switch configuration is the first choice among various improvement strategies.

[0225] It is to be understood that the present invention is described by some embodiments, and it is known to those skilled in the art that various changes or equivalent substitutions may be made to these features and embodiments without departing from the spirit and scope of the present invention. In addition, under the teachings of the present invention, these features and embodiments may be modified to adapt to specific circumstances and materials without departing from the spirit and scope of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of this application are within the scope of protection of the present invention.

Claims

1. A distribution network reliability assessment method based on load segmentation, characterized in that: The following steps are involved: S1. Establish an improved distribution network reliability assessment model based on load block technology for switch elements, and input fault parameters, load parameters, system topology parameters, and load branch parameters; S2. Form a system topology diagram according to the line node branch numbering; S3, partitioning the distribution network according to the system topology diagram; S4, introduce the fuse factor and calculate the block fault parameters; S5, enumerate the fault blocks to form a fault consequence pattern table; S6. Determine the power supply reliability of the areas divided in S3, set fuses for the required areas, and supplement the influence of the fuse branches on the load; S7, enumerate all fault blocks; S8. Use Bayesian network method to find the weak links of distribution network; S9. Calculate system reliability indicators.

2. The method for evaluating the reliability of a distribution network based on load segmentation according to claim 1 is characterized in that: The fault parameters described in S1 include component failure rate λ, component repair rate μ, fault repair time r=1 / μ, time before failure d=1 / λ, failure frequency f=1 / (d+r) and unavailability rate U=λ / (λ+μ); load parameters include user type, number of users, load point power peak value and load point power average value.

3. The method for evaluating the reliability of a distribution network based on load segmentation according to claim 1 is characterized in that: In S3, the switch elements are used as the boundary, and the breadth-first traversal algorithm of the tree is used to visit the adjacent nodes step by step from the root node of the system topology structure until the entire network is traversed, and the distribution network is partitioned to form element groups and load areas.

4. The method for evaluating the reliability of a distribution network based on load segmentation according to claim 1, characterized in that: In S5, the load area types after the distribution network failure are divided into four categories, ABCD, according to the power outage duration of load points at different locations when the component fails. By enumerating the block faults in turn, the fault type of each load point caused by each component failure can be obtained, that is, the fault consequence mode table.

5. The method for evaluating the reliability of a distribution network based on load segmentation according to claim 2 is characterized in that: In S8, the Bayesian network method is used to analyze the weak links of the load points, and the BFSIDI index of the distribution network is selected for analysis. The calculation formula of the BFSIDI index is: Where N i is the number of users at the ith load point, U ik is the time it takes for the i-th load point to restore power supply when the k-th component fails, and nload is the number of all load points; the weak link of the distribution network is determined based on the shared amount.

6. The method for evaluating the reliability of a distribution network based on load segmentation according to claim 5 is characterized in that: The system reliability indicators in S9 include: System average power outage frequency index System average power outage duration indicator Average power outage duration indicator for users Average power supply availability index Total power loss Total power outage loss LOSS = ENSI * average electricity price.

7. The improved control strategy of the distribution network reliability assessment method based on load segmentation is characterized in that: The following steps are involved: S2.

1. Set the initial value of the algorithm in the improved distribution network reliability assessment model, input the initialization data, and generate a random initial population; S2.2, calculate the fitness function value, that is, calculate the objective function value, and set the fitness to infinity for the scenario that does not meet the constraint conditions, so as to eliminate the individual in the iteration; S2.3, use the bidding algorithm to construct the Pareto non-inferior solution set; S2.

4. Calculate the crowding distance of the particles in the Pareto non-inferior solution set. The crowding distance of a particle is the minimum difference between the particle and other particles in the fitness function value. S2.5, trim the Pareto non-inferior solution set and delete particles with small crowding distance; S2.

6. Select individual and group extreme value particles, and select the best performing individuals and groups from the current solution set; S2.

7. Calculate the update formula of the random number m generated by each particle, the weight ω and the learning factor c1c2. Set the constant threshold p∈[0,1] as the capture particle x of the three black hole system. i The probability that for each particle x i Each iteration generates a random number m∈[0,1]; S2.8, particles satisfying m≤p are captured by the three-black hole system, and the rest are updated in the traditional way; S2.9, forming new particle positions and velocities; S2.10, premature disturbance judgment is performed on the particle, when |F k -F k-1 |<0.01|F k |, reset the position of the particle so that it can potentially jump out of the local optimum, where F k Represents the objective function value corresponding to the global optimal particle of generation k; S2.11, if the iteration termination condition is met, the optimal Pareto non-inferior solution set is output; if not, return to step S2.2 and iterate again to find the optimal solution; S2.

12. Use the TOPSIS method to sort the individuals in the Pareto archive set and select the optimal solution.

8. The improvement control strategy of the load-block-based distribution network reliability assessment method according to claim 7 is characterized in that: The initialization data in S2.1 includes the particle position x, the particle search speed v and the current particle position p.

9. The improved control strategy of the load-block-based distribution network reliability assessment method according to claim 7 is characterized in that: The bidding algorithm in S2.3 is as follows: in each round, two particles are selected to compare the dominance relationship. The winner continues to compete with the remaining particles in the group to obtain a particle that wins the competition with all particles in the group. It is saved in the Pareto non-inferior solution set and deleted from the group; the competition is restarted until all particles in the group are completely deleted, thereby completing the establishment of the Pareto non-inferior solution set.

Images