An active power distribution network risk assessment and optimization control method, system, device and medium

Abstract

The application discloses a kind of active power distribution network risk assessment and optimization control method, system, equipment and medium, it is related to power distribution network technical field.First, the static structure of active power distribution network, photovoltaic access and multi-source heterogeneous operation data are constructed into multi-source data set;The first risk assessment data is obtained by fusing and risk pre-evaluation to multi-source data set;When the first risk assessment data does not exceed the preset over-standard threshold, directly as target data, meet when triggering topology reconstruction, construct optimization model and obtain multiple candidate schemes;Based on the updated network state and multi-source data set of each candidate scheme, the second risk assessment data is obtained by using the coupling mode of polynomial chaos expansion surrogate model and accurate power flow calculation for secondary evaluation;Verify the effectiveness of each candidate scheme, the second risk data corresponding to the scheme that meets the conditions is used as target data and the optimization control scheme is output.The photovoltaic high penetration pain point is solved, the accuracy and efficiency are considered, and the real-time active and accurate prevention and control requirements are met.

Description

Technical Field

[0001] This invention relates to the field of power distribution network technology, and in particular to a method, system, device and medium for risk assessment and optimization control of active power distribution networks. Background Technology

[0002] With the large-scale and decentralized integration of distributed photovoltaic (PV) power into distribution networks, these networks have transformed from traditional passive to active networks, exhibiting bidirectional power flow characteristics. However, PV power output exhibits significant randomness and volatility. Coupled with the heterogeneous nature of multi-source operational data, this leads to multiple risks facing the distribution network, including reverse overload, voltage exceeding limits, and harmonic resonance. These risks are easily propagated and spread across different levels—transformer areas, feeders, and substations—seriously threatening the safe and stable operation of the distribution network. Therefore, improving the accuracy and real-time performance of risk assessment in active distribution networks, and achieving cross-level risk tracing and dynamic optimization control, has become a critical technical problem that urgently needs to be solved.

[0003] Existing risk assessment technologies for distribution networks have significant shortcomings: First, their multi-source data fusion capabilities are weak, making it difficult to effectively resolve the heterogeneous contradictions between photovoltaic (PV) output data and grid operation data in terms of spatiotemporal scale, sampling frequency, and data quality, thus failing to accurately capture high-risk, low-probability events. Second, risk assessment and topology optimization control are disconnected, making it impossible to dynamically adapt to real-time power flow fluctuations caused by rapid changes in PV output. Third, it is difficult to balance assessment efficiency and accuracy; traditional Monte Carlo simulations are computationally complex and time-consuming, while single-proxy models have poor scenario adaptability and cannot maintain high accuracy across all scenarios. Fourth, there is a lack of cross-level risk tracing and closed-loop management mechanisms, making it impossible to track risk transmission paths and pinpoint the source. These deficiencies make it difficult for existing technologies to form a comprehensive risk perception and dynamic collaborative prevention and control capability for active distribution networks with high PV integration. Summary of the Invention

[0004] This invention provides a method, system, device, and medium for risk assessment and optimization control of active power distribution networks, which solves the technical problems of existing technologies in high photovoltaic penetration scenarios, such as insufficient assessment accuracy, low computational efficiency, and lagging optimization control, which make it difficult to meet the real-time and accurate risk prevention and control requirements of active power distribution networks.

[0005] The first aspect of this invention provides a method for risk assessment and optimization control of active power distribution networks, comprising:

[0006] The static structural parameters, photovoltaic access parameters, and multi-source heterogeneous operation data of the active power distribution network are obtained to construct a multi-source dataset.

[0007] The multi-source dataset is fused and risk pre-assessed to obtain the first risk assessment data.

[0008] When the first risk assessment data does not exceed the preset risk exceedance threshold, the first risk assessment data is used as the target risk assessment data, and the control strategy of maintaining the current operation mode is maintained.

[0009] When the first risk assessment data exceeds the preset risk exceedance threshold, topology dynamic reconstruction is triggered based on the first risk assessment data, and then a topology reconstruction optimization model is constructed based on the multi-source dataset and multiple candidate solutions are obtained.

[0010] Based on the updated network state corresponding to each candidate solution and the multi-source dataset, a secondary risk assessment is performed by coupling a multinomial chaotic expansion proxy model with precise power flow calculation to obtain the second risk assessment data corresponding to each candidate solution.

[0011] The validity of the second risk assessment data corresponding to each candidate solution is verified. The second risk assessment data corresponding to the candidate solution that meets the preset validity conditions is taken as the target risk assessment data, and the optimized control scheme corresponding to the candidate solution is output.

[0012] Optionally, the step of performing multi-source data fusion and risk pre-assessment on the multi-source dataset to obtain first risk assessment data includes:

[0013] The multi-source dataset is subjected to hierarchical preprocessing to generate a hierarchical fused dataset.

[0014] The hierarchical fusion dataset is used for probabilistic fuzzy coupled inference to generate probabilistic risk data and target risk severity data at each level;

[0015] The probability risk data and target risk severity data corresponding to each level are converted into risk indicator values ​​for the corresponding level, and then integrated to obtain the first risk assessment data.

[0016] Optionally, the step of performing data hierarchical preprocessing on the multi-source dataset to generate a hierarchical fused dataset includes:

[0017] Probabilistic modeling is performed using real-time photovoltaic power output data from the multi-source dataset to generate photovoltaic power output probability distribution data.

[0018] Using the electrical quantity data in the multi-source dataset, multiple risk level membership degrees are calculated.

[0019] The risk level membership degree is weighted and defuzzified to generate initial risk severity data for each level;

[0020] The environmental feature data and topological state data in the multi-source dataset are used to perform association rule mining to generate high-risk, low-probability factors and corresponding risk labels.

[0021] The photovoltaic output probability distribution data, all the initial risk severity data, the high-risk low-probability factors and the corresponding risk labels are standardized and aligned to generate a hierarchical fusion dataset.

[0022] Optionally, the step of using the hierarchical fusion dataset to perform probabilistic fuzzy coupled inference to generate probabilistic risk data and target risk severity data at each level includes:

[0023] Using the aforementioned hierarchical fusion dataset, the probabilistic flow of non-fault scenarios is calculated through a multinomial chaotic expansion surrogate model, generating probabilistic risk data at each level.

[0024] The probability risk data at each level are compared with the corresponding preset probability thresholds.

[0025] When any probability risk data exceeds the corresponding preset probability threshold, the corresponding level of fuzzy inference unit is invoked to calculate and generate the corresponding target risk severity data.

[0026] When all probability risk data does not exceed the corresponding preset probability threshold, the probability risk data of each level is directly used as the corresponding target risk severity data.

[0027] Optionally, the step of converting the probability risk data and target risk severity data corresponding to each level into risk indicator values ​​for the corresponding level and integrating them to obtain the first risk assessment data includes:

[0028] The probability risk data and target risk severity data corresponding to the transformer substation level are converted into transformer substation level risk index values, which include the severity of reverse heavy overload of distribution transformers and the qualification rate of low-voltage lines.

[0029] The probability risk data and target risk severity data corresponding to the feeder level are converted into feeder level risk index values, which include branch power flow over-limit risk value and feeder reverse load.

[0030] The probabilistic risk data and target risk severity data corresponding to the substation level are converted into substation-level risk index values, which include node voltage deviation risk values ​​and total reverse load of the substation.

[0031] The risk index values ​​at the distribution area level, feeder level, and substation level are integrated to obtain the first risk assessment data.

[0032] Optionally, the step of constructing a topology reconstruction optimization model based on the multi-source dataset and solving it to obtain multiple candidate solutions includes:

[0033] Construct a multi-objective optimization model with the objectives of minimizing network loss, minimizing the amount of sensitive load not restored, and minimizing the risk of reverse heavy overload;

[0034] Under the constraints of radial topology, bidirectional photovoltaic power flow limitation, equipment tolerance and number of switching operations, a commercial integer programming solver was used to obtain multiple candidate topology schemes.

[0035] Based on the network topology corresponding to the candidate topology scheme, photovoltaic power output optimization and load optimization allocation are performed to generate adjustment instructions corresponding to the candidate topology scheme.

[0036] Based on the switch status corresponding to the candidate topology scheme, a topology switch action command is generated, and combined with the high-risk, low-probability factors in the first risk assessment data, equipment maintenance priority suggestion data corresponding to the candidate topology scheme is constructed.

[0037] Multiple candidate solutions are constructed by using the adjustment instructions corresponding to the candidate topology scheme, the equipment operation and maintenance priority suggestion data, and the candidate topology scheme respectively.

[0038] Optionally, the step of performing a secondary risk assessment by coupling a polynomial chaotic expansion surrogate model with precise power flow calculation to obtain the second risk assessment data corresponding to each candidate solution includes:

[0039] Based on the updated network state corresponding to each candidate solution, determine the operating scenario corresponding to the candidate solution;

[0040] When the operating scenario is a non-fault steady-state scenario, the electrical output is approximated by a polynomial chaotic expansion proxy model, and the risk data corresponding to the candidate solution is calculated.

[0041] When the operating scenario is a fault scenario or a scenario after topology reconfiguration, accurate power flow calculation is used to capture risk fluctuations and obtain the risk data corresponding to the candidate solution;

[0042] Convert the risk data in this scenario into risk indicators of the corresponding level;

[0043] Based on the risk indicators and the multi-source dataset, the objective weights and subjective weights of the risk indicators are calculated using the entropy weight method and the analytic hierarchy process, respectively, and the combined weights are obtained by solving the Nash equilibrium.

[0044] The second comprehensive risk value is calculated using the aforementioned risk indicators and their corresponding combined weights.

[0045] The second risk assessment data corresponding to each candidate solution is constructed by using the risk index and the second comprehensive risk value corresponding to each candidate solution.

[0046] The second aspect of this invention provides an active power distribution network risk assessment and optimization control system, comprising:

[0047] The parameter and data acquisition module is used to acquire the static structural parameters, photovoltaic access parameters and multi-source heterogeneous operation data of the active power distribution network, and construct a multi-source dataset.

[0048] The first risk assessment module is used to perform multi-source data fusion and risk pre-assessment on the multi-source dataset to obtain the first risk assessment data.

[0049] The risk assessment module is a control strategy used to take the first risk assessment data as the target risk assessment data and maintain the current operating mode when the first risk assessment data does not exceed the preset risk exceedance threshold.

[0050] The topology reconstruction module is used to trigger dynamic topology reconstruction based on the first risk assessment data when the first risk assessment data exceeds the preset risk exceedance threshold, and then construct a topology reconstruction optimization model based on the multi-source dataset and solve it to obtain multiple candidate solutions.

[0051] The second risk assessment and optimization module is used to perform a second risk assessment based on the updated network state corresponding to each candidate solution and the multi-source dataset, using a multinomial chaotic expansion proxy model coupled with precise power flow calculation, to obtain the second risk assessment data corresponding to each candidate solution.

[0052] The result output module is used to verify the validity of the second risk assessment data corresponding to each candidate solution, take the second risk assessment data corresponding to the candidate solution that meets the preset validity conditions as the target risk assessment data, and output the optimized control scheme corresponding to the candidate solution.

[0053] A third aspect of the present invention provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the active power distribution network risk assessment and optimization control method described above.

[0054] The fourth aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed, implements the active power distribution network risk assessment and optimization control method as described above.

[0055] As can be seen from the above technical solutions, the present invention has the following advantages:

[0056] This invention constructs a multi-source dataset by acquiring static structural parameters, photovoltaic access parameters, and multi-source heterogeneous operation data of an active power distribution network. The multi-source dataset is then fused and pre-assessed to obtain first risk assessment data. When the first risk assessment data does not exceed a preset risk exceedance threshold, it is directly used as the target risk assessment data. When the first risk assessment data exceeds the preset risk exceedance threshold, dynamic topology reconfiguration is triggered based on the first risk assessment data. A topology reconfiguration optimization model is then constructed based on the multi-source dataset, and multiple candidate solutions are obtained. A second risk assessment is performed based on the updated network state and multi-source dataset corresponding to each candidate solution, using a polynomial chaotic expansion surrogate model coupled with precise power flow calculation, to obtain second risk assessment data corresponding to each candidate solution. Finally, the validity of the second risk assessment data corresponding to each candidate solution is verified. The second risk assessment data corresponding to the candidate solution that meets the preset validity conditions is used as the target risk assessment data, and the optimized control scheme corresponding to that candidate solution is output.

[0057] This method effectively addresses the core technical challenges of insufficient accuracy, low computational efficiency, and lagging optimization control in distribution network risk assessment under high photovoltaic penetration scenarios, providing strong support for the safe and stable operation of distribution networks. By integrating static structure, photovoltaic access, and multi-source heterogeneous operation data to construct a multi-source dataset, it comprehensively covers various influencing factors of distribution network operation, providing a complete data foundation for risk assessment. The combination of multi-source data fusion and risk pre-assessment enables early judgment of the distribution network's operating status and timely identification of potential risks. The mechanism of outputting target risk assessment data in different scenarios allows for rapid result output during normal distribution network operation, avoiding unnecessary computational overhead. When risks exceed limits, dynamic topology reconfiguration is triggered, proactively reducing risk levels at the network structure level and improving the lag problem of traditional passive risk management. The secondary risk assessment method, coupled with a polynomial chaotic expansion surrogate model and precise power flow calculation, balances assessment efficiency and accuracy, significantly improving computational speed while ensuring the reliability of risk assessment results, better meeting the needs of real-time risk prevention and control in active distribution networks. By verifying the effectiveness of candidate solutions before outputting the optimized control scheme, we can ensure that the output scheme is truly effective and can effectively reduce the operational risks of the power distribution network. Attached Figure Description

[0058] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0059] Figure 1 A flowchart illustrating the steps of an active power distribution network risk assessment and optimization control method provided in this embodiment of the invention;

[0060] Figure 2 This is a schematic diagram of an active power distribution network provided in an embodiment of the present invention;

[0061] Figure 3 This is a structural block diagram of the active power distribution network risk assessment and optimization control system provided in an embodiment of the present invention;

[0062] Figure 4 This is a structural block diagram of a computer device provided in an embodiment of the present invention. Detailed Implementation

[0063] This invention provides a method, system, device, and medium for risk assessment and optimization control of active power distribution networks, which addresses the technical problem that existing technologies in high photovoltaic penetration scenarios suffer from insufficient assessment accuracy, low computational efficiency, and lagging optimization control, making it difficult to meet the real-time and accurate risk prevention and control requirements of active power distribution networks.

[0064] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0065] Please see Figure 1 , Figure 1 A flowchart illustrating the steps of an active power distribution network risk assessment and optimization control method provided in this embodiment of the invention.

[0066] This invention provides a method for risk assessment and optimization control of active power distribution networks, comprising:

[0067] Step 101: Obtain the static structural parameters, photovoltaic access parameters, and multi-source heterogeneous operation data of the active distribution network to construct a multi-source dataset.

[0068] In embodiments of the present invention, such as Figure 2As shown, the active power distribution network consists of a hierarchical equipment cluster, a photovoltaic access unit cluster, a topology switch component cluster, and a risk assessment and optimization control system. Each photovoltaic access unit cluster is associated with at least one hierarchical equipment cluster. The hierarchical equipment clusters are divided into distribution area equipment groups, feeder equipment groups, and substation equipment groups according to voltage levels, including power equipment such as transformers, lines, and circuit breakers. The photovoltaic access unit cluster includes components such as photovoltaic arrays and inverters. The topology switch component cluster includes components such as tie switches and sectionalizing switches used to adjust the topology structure. The risk assessment and optimization control system communicates with the three types of clusters to complete the acquisition of operating data and the issuance of control commands.

[0069] This step acquires three types of data and integrates them to form a multi-source dataset. Static structural parameters are fixed configuration data of the distribution network, pre-stored in the system memory, including equipment rated capacity (such as transformer rated capacity, line rated current carrying capacity, etc.), line impedance parameters, equipment location information, topology connections, etc., used to construct the basic model of the distribution network. Photovoltaic access parameters include photovoltaic access capacity, inverter control parameters, historical output statistics, etc., collected and uploaded by the monitoring system of the photovoltaic access unit, used to characterize the inherent characteristics of photovoltaic access. Multi-source heterogeneous operation data are real-time or near-real-time dynamic data, collected by various monitoring terminals of the distribution network, specifically including… The data includes real-time photovoltaic output data (including real-time output of photovoltaic arrays, output fluctuations, inverter switching status, etc.), electrical quantity data (including node voltage, branch power flow, distribution transformer load rate, harmonic voltage content, line zero-sequence current, etc.), environmental characteristic data (including irradiance, ambient temperature, etc.), topology status data (including tie switch on / off status, sectionalizing switch operation count, line connection status, etc.), and load data (including real-time load of distribution areas or feeders, load ratio of sensitive users, load time-series fluctuation curves, etc.). This type of data is characterized by its dispersed sources and heterogeneous types, and needs to be preprocessed before it can be used for risk assessment. Integrating the above three types of data will construct a multi-source dataset.

[0070] Step 102: Perform multi-source data fusion and risk pre-assessment on the multi-source dataset to obtain the first risk assessment data.

[0071] In this embodiment of the invention, this step is based on a pre-constructed hierarchical probabilistic fuzzy coupling model and is the core link of the entire risk assessment. This model is designed to address the characteristics of distribution network operation with strong randomness, data heterogeneity, and risk transmission across levels under high photovoltaic penetration. It completes multi-source data fusion and risk quantification through three stages: hierarchical data preprocessing, probabilistic fuzzy coupling inference, and dynamic linkage optimization. Finally, it outputs the first risk assessment data containing three levels of risk indicators: distribution area, feeder, and substation. At the same time, it embeds topology reconfiguration triggering logic to provide a basis for judgment in subsequent optimization control.

[0072] Specifically, the multi-source dataset is first preprocessed with hierarchical layers. Through photovoltaic output probability modeling, risk level membership calculation, weighted average defuzzification, and high-risk low-probability factor mining, heterogeneous operating data is uniformly converted into a standardized hierarchical fusion dataset, solving the problems of inconsistent spatiotemporal scales and difficulty in feature fusion among multi-source data. Next, the probabilistic fuzzy coupled inference stage is entered. A multinomial chaotic expansion surrogate model is used to quickly calculate the probabilistic risk of non-fault scenarios. Combined with a probability threshold triggering mechanism, the fuzzy inference unit is activated to accurately calculate the risk severity when the risk is high, achieving coupling between macro-level probability assessment and micro-level risk tracing. Finally, through dynamic linkage optimization, the probabilistic risks and risk severity at each level are converted into risk indicator values ​​at the transformer area, feeder, and substation levels and integrated into the first risk assessment data. Simultaneously, based on the exceedance of indicators, it is determined whether to trigger dynamic topology reconfiguration, completing the closed-loop processing from data fusion to risk assessment.

[0073] Furthermore, step 102 may include the following sub-steps:

[0074] S11. Perform data layering preprocessing on the multi-source dataset to generate a layered fused dataset.

[0075] It should be noted that data hierarchical preprocessing is the first step in the hierarchical probabilistic fuzzy coupling model. Through specialized algorithms such as photovoltaic output probability modeling, risk membership calculation, weighted average defuzzification, high-risk low-probability factor mining, and data standardization alignment, photovoltaic data, electrical quantity data, environmental feature data, and topological state data are processed in sequence. This fully explores the risk characteristics in various types of data, unifies the units and spatiotemporal formats, and effectively solves the contradictions in spatiotemporal scale, sampling frequency, and data quality of multi-source heterogeneous data. Finally, a standardized hierarchical fusion dataset that can be directly used for risk inference is formed.

[0076] Further, step S11 may include the following sub-steps:

[0077] S111. Probabilistic modeling is performed using real-time photovoltaic power output data from multiple source datasets to generate photovoltaic power output probability distribution data.

[0078] In this embodiment of the invention, kernel density estimation is used to construct a nonparametric probability model for real-time photovoltaic power output data, accurately capturing the multi-peak fluctuation characteristics of photovoltaic power output, and generating photovoltaic power output probability distribution data through this modeling process.

[0079] S112. Using electrical quantity data from multiple source datasets, calculate the membership degree of multiple risk levels.

[0080] In this embodiment of the invention, electrical quantity data is processed by a hierarchical probabilistic fuzzy inference system, which includes a district-level fuzzy inference unit, a feeder-level fuzzy inference unit, and a substation-level fuzzy inference unit.

[0081] The input variables of the fuzzy inference unit at the distribution area level include the reverse load rate of the distribution transformer, the voltage deviation of the low-voltage node, and the photovoltaic power output fluctuation rate; the output variable is the severity of the risk in the distribution area.

[0082] The input variables of the feeder-level fuzzy inference unit include branch power flow load rate, feeder voltage deviation, and feeder reverse power ratio; the output variable is the feeder risk severity.

[0083] The input variables of the substation-level fuzzy inference unit include bus voltage deviation, main transformer reverse load rate, and total reverse power of the substation; the output variable is the substation risk severity.

[0084] For each input variable, three fuzzy subsets are set: low risk (L), medium risk (M), and high risk (H). The membership degree is calculated using a combination of trapezoidal and triangular membership functions.

[0085] The low-risk L membership degree uses a left-half trapezoidal function, assuming that a calculated reverse overload probability is... The low-risk membership function is defined as:

[0086] ;

[0087] in, Reverse overload probability The membership degree to the low-risk level ranges from [0,1], with a larger value indicating a closer approximation to a low-risk state. When the risk level is low, the system is in a low-risk state, and its low-risk membership degree is close to or equal to 1; as the risk level increases... As the membership of low-risk members gradually increases, it gradually decreases until it drops to 0. This is the upper limit threshold of the low-risk range; This serves as a transitional dividing line between low-risk and medium-risk areas.

[0088] The membership degree of medium-risk M uses a triangular function, assuming that a calculated reverse overload probability is... The membership function for medium-risk is defined as follows:

[0089] ;

[0090] In the formula, Indicates the probability of reverse overload. Membership degree to the medium-risk level. When As the risk level gradually increases from low to medium, the degree of membership in the medium-risk area gradually increases; when When the area reaches the center of the medium-risk zone, the degree of membership in the medium-risk zone reaches its maximum; as... It continues to increase and transitions towards high-risk areas, while the membership of medium-risk areas gradually decreases; This represents the left endpoint of the medium-risk range; This represents the peak point of the medium-risk range. It serves as a transitional dividing line between medium and high risk.

[0091] The membership degree of high-risk H uses a right-hand trapezoidal function. Assume that a calculated reverse overload probability is... The high-risk membership function is defined as follows:

[0092] ;

[0093] in, Indicates the probability of reverse overload. Membership degree to high-risk levels. When When the membership is small, the high-risk membership degree is 0; as the membership degree increases... Gradually increasing and entering a high-risk transition zone, the degree of membership in high-risk areas gradually increases; when Once the set threshold is exceeded, the system can consider it to be in a high-risk state, with the high-risk membership degree reaching or approaching 1; This represents the left endpoint of the high-risk zone. This is the lower limit threshold of the high-risk range.

[0094] In the membership function described above, the parameter , , , , , , It can be set based on power grid operation experience, historical operation data, risk tolerance, or dispatch control requirements. The risk levels are not separated by hard, non-overlapping boundaries, but rather through overlapping intervals between adjacent membership functions to achieve a continuous transition of risk states. For example, in the overlapping interval between low and medium risk, as the probability of reverse overload increases, the membership degree of low risk gradually decreases, while the membership degree of medium risk gradually increases; in the overlapping interval between medium and high risk, as the probability of reverse overload further increases, the membership degree of medium risk gradually decreases, while the membership degree of high risk gradually increases.

[0095] S113. Perform weighted average defuzzification on the risk level membership degree to generate initial risk severity data for each level.

[0096] In this embodiment of the invention, after obtaining low-risk, medium-risk, and high-risk membership degrees, a defuzzification method can be used to fuse multiple risk membership degrees according to actual application needs, generating initial risk severity data for each level of distribution area, feeder, and substation. As an optional approach, a weighted average method can be used to calculate the comprehensive risk index.

[0097] ;

[0098] In the formula, This represents the comprehensive risk assessment value, used to characterize the severity of the risk. These represent the quantified risk values ​​corresponding to low, medium, and high risk, respectively. Typically, we can let... This reflects the severity of risks corresponding to different risk levels. Through the above calculations, the membership degrees of multiple fuzzy risk levels can be integrated into a continuously changing risk assessment indicator, providing a quantitative basis for subsequent risk warning, scheduling optimization, or control decisions.

[0099] S114. Use environmental feature data and topological state data from multi-source datasets to perform association rule mining, and generate high-risk, low-probability factors and corresponding risk labels.

[0100] In this embodiment of the invention, an improved association rule mining algorithm is used to separate high-risk, low-probability factors from environmental feature data and topology state data, such as sudden increases in photovoltaic output caused by short-term strong sunlight and abnormal switch operation. The improved association rule mining algorithm is a weighted low-support association rule mining algorithm. The weighted low-support association rule mining algorithm is used to process environmental feature data (light intensity, ambient temperature) and topology state data (connector switch on / off state, number of segment switch operations). Specifically, the weighted low-support association rule mining algorithm includes the following steps:

[0101] 1. Construct a time window transaction set

[0102] According to the preset time window Align the above data and divide the continuously running data into... A time window forms the running state vector:

[0103] ;

[0104] For the k-th time window, environmental characteristics, topological state, photovoltaic output changes, and risk labels are collected to form the original operating state vector:

[0105] ;

[0106] in, This is the running state vector within the k-th time window; This is the original running state vector within the k-th time window; Let be the light intensity within the k-th time window; The ambient temperature during the k-th time window; The load level in the k-th time window; This represents the change in photovoltaic power output during the k-th time window. This represents the on / off status of the communication switch during the k-th time window. This represents the segmented switch operation state under the k-th time window; This represents the line connection or topology status during the k-th time window. This refers to the label of the risk event within the k-th time window.

[0107] For continuous variables such as irradiance, ambient temperature, load level, and changes in photovoltaic output, discretization is performed using preset thresholds or quantiles. For any continuous variable... For example, its discretization result can be expressed as:

[0108] ;

[0109] In the formula, These represent low, medium, and high levels, respectively. The preset discretization threshold is used.

[0110] For each time window, a transaction set is formed. , where each transaction The corresponding operating status within a time window includes at least the light intensity level, ambient temperature level, load level, photovoltaic output change level, interconnection switch on / off status, sectionalizing switch operation status, line connection status, and corresponding risk labels.

[0111] The second step is to generate candidate association rules.

[0112] Extract environment-related transaction items, topology-related transaction items, and runtime status-related transaction items from transaction set D, and construct a candidate antecedent set A. ,in Represents the environment class; Represents topology class, Representing a set of transaction items in the running state, and using risk events as the rule consequent set B, candidate rules are generated:

[0113] ;

[0114] In the formula, A can be a combination of one or more environmental, topological, and operational status factors; B is the risk event label, i.e., the set of rule consequents, including risks such as transformer reverse overload, voltage exceeding limits, branch power flow exceeding limits, harmonic voltage content exceeding standards, or cascading tripping.

[0115] The third step is to calculate the support, confidence, and lift of the candidate rules. First, define the total number of transactions as... For candidate rules The support (sp), confidence (co), and lift (li) are calculated as follows:

[0116] ;

[0117] ;

[0118] ;

[0119] In the formula, N represents the total number of transactions; This indicates the number of transactions containing A; This indicates the number of transactions that simultaneously include A and B; This represents the probability of risk event B occurring in all transactions.

[0120] The fourth step is to introduce risk weight calculation rules for comprehensive scoring.

[0121] To prevent low-probability, high-risk events from being filtered out by the conventional association rule mining process due to low support, this embodiment sets risk weights for different risk consequents. The calculation formula is:

[0122] ;

[0123] In the formula, Indicates the severity of the harm; Indicates the range of affected equipment; Indicates the consequences of historical failures; Indicates the level of attention given to scheduling and operation; To satisfy the requirement that the weight coefficients sum to 1, and For example, the risk weight for cascading tripping is higher than the voltage over-limit weight for a single node.

[0124] Candidate rules Calculate the risk-weighted score:

[0125] ;

[0126] in, These are weighting coefficients used to adjust the impact of confidence, lift, risk weight, and support in rule selection. By retaining low support items in the scoring function and introducing risk weights, rules with low support but high confidence, lift, and risk consequences can still be identified. This is achieved by introducing a scarcity index. This is done by assigning positive weights to rules with extremely low support but high confidence and serious risk consequences, so that these rules can receive higher scores, thereby accurately identifying high-risk, low-probability factors.

[0127] The fifth step is to filter high-risk, low-probability association rules.

[0128] Set minimum confidence threshold Minimum lift threshold Highest support threshold and rule scoring thresholds When the candidate rules are satisfied:

[0129] ;

[0130] This rule is then classified as a high-risk, low-probability association rule. Although such rules appear infrequently in historical samples, once they occur, they are usually strongly correlated with higher-risk events, and therefore can serve as important inputs for subsequent risk pre-assessment and risk tracing.

[0131] Step 6: Output high-risk, low-probability factors and their risk labels.

[0132] The high-risk, low-probability rules obtained through screening are output to a hierarchical probabilistic fuzzy coupling model, serving as environmental topological coupling risk factors in subsequent risk assessments. For example, when the following rules are mined:

[0133] {Short-term strong sunlight, low load, interconnection switch disconnected} → {Transformer reverse heavy overload};

[0134] Furthermore, when the rule meets the confidence level, lift level, and risk-weighted scoring thresholds, "short-term strong sunlight, low load, and disconnection of tie switches" is identified as a high-risk, low-probability factor that could induce reverse heavy overload of distribution transformers. This factor is then input into the substation-level or feeder-level risk assessment unit to improve the sensitivity of risk warnings under conditions of sudden photovoltaic surges, reverse power flow, and abnormal topological constraints.

[0135] S115. Standardize and align the photovoltaic output probability distribution data, all initial risk severity data, high-risk low-probability factors and corresponding risk labels to generate a hierarchical fusion dataset.

[0136] In this embodiment of the invention, the range transformation method is used to standardize, unify the format, and align the spatiotemporal dimensions of the photovoltaic power output probability distribution data, all initial risk severity data, high-risk low-probability factors, and corresponding risk labels, thereby resolving the data heterogeneity contradiction and generating a hierarchical fusion dataset after integration.

[0137] S12. Use hierarchical fusion datasets to perform probabilistic fuzzy coupled inference to generate probabilistic risk data and target risk severity data at each level.

[0138] It should be noted that probabilistic fuzzy coupled inference is the second step of the hierarchical probabilistic fuzzy coupled model. Risk assessment is achieved through a coupling mechanism of multinomial chaotic expansion surrogate model calculation, probability threshold comparison, and scenario-specific fuzzy inference. First, based on the hierarchical fusion dataset, the probabilistic power flow calculation for non-fault scenarios is completed through the multinomial chaotic expansion surrogate model to generate probabilistic risk data at each level. Then, the probabilistic risk data at each level is compared with the corresponding preset probability threshold one by one. When any probabilistic risk data exceeds the corresponding preset threshold, the fuzzy inference unit at that level is called to accurately calculate and generate the target risk severity data. When all probabilistic risk data does not exceed the corresponding threshold, the probabilistic risk data at each level is directly used as the target risk severity data. Finally, the probabilistic risk data and target risk severity data at each level are output.

[0139] Furthermore, step S12 may include the following sub-steps:

[0140] S121. Using a hierarchical fusion dataset, the probabilistic flow of non-fault scenarios is calculated through a multinomial chaotic expansion surrogate model, generating probabilistic risk data at each level.

[0141] In this embodiment of the invention, the complete generation process of probability risk data at each level is as follows:

[0142] 1. Extract three types of random variables—photovoltaic output, load fluctuation, and environmental variables—from the hierarchical fusion dataset, and combine them to form a random input vector. ;

[0143] 2. Define the power flow calculation output vector Y, which includes three types of electrical quantities: node voltage, branch power flow, and reverse power;

[0144] 3. Construct a polynomial chaotic expansion proxy model, with the following formula:

[0145] ;

[0146] In the formula, For orthogonal polynomial basis functions that match the random input distribution; is the expansion factor.

[0147] 4. Select several different sets of random input samples, and call the precise power flow calculation to obtain the node voltage, branch power flow, and reverse power results corresponding to each set of samples;

[0148] 5. Using the least squares method or projection method, obtain all the expansion coefficients of the surrogate model based on the sample input and output data. ;

[0149] 6. Using the fitted polynomial chaotic expansion surrogate model, we can quickly calculate node voltage, branch power flow, and reverse power distribution in massive random scenarios.

[0150] 7. Calculate the probability of each electrical quantity exceeding its limit, including the probability of voltage exceeding its limit, the probability of power flow exceeding its limit, and the probability of reverse overload.

[0151] 8. The above-mentioned over-limit probabilities are divided and collected according to three levels: distribution area, feeder, and substation, and finally the probability risk data of each level is generated.

[0152] S122. Compare the probability risk data of each level with the corresponding preset probability threshold.

[0153] In this embodiment of the invention, a specific preset probability threshold is set for each type of probabilistic risk data (voltage over-limit probability, power flow over-limit probability, reverse heavy overload probability) at each level of distribution area, feeder, and substation. Each probabilistic risk data is compared with the corresponding preset probability threshold one by one, and different evaluation logics of S123 or S124 are triggered according to the comparison result.

[0154] S123. When any probability risk data is greater than the corresponding preset probability threshold, the corresponding level of fuzzy inference unit is called to calculate and generate the corresponding target risk severity data.

[0155] In this embodiment of the invention, the complete execution process of the probability threshold-triggered fuzzy inference mechanism is as follows: when the comparison result of S122 shows that any probability risk data of any level is greater than its corresponding preset probability threshold, the system automatically calls the fuzzy inference unit (transformer area level fuzzy inference unit / feeder level fuzzy inference unit / substation level fuzzy inference unit) of the level to which the risk type belongs; the fuzzy inference unit recalculates the severity of the risk at the level based on the membership function defined in step S112 and the built-in fuzzy rule library, and finally generates the corresponding target risk severity data.

[0156] S124. When all probability risk data do not exceed the corresponding preset probability threshold, the probability risk data of each level are directly used as the corresponding target risk severity data.

[0157] In this embodiment of the invention, when the comparison result of S122 shows that all probability risk data of all levels of distribution area, feeder and substation are less than or equal to their corresponding preset probability threshold, the distribution network is determined to be in a low-risk or normal operation state. There is no need to start any fuzzy inference unit for detailed evaluation. The probability risk data of each level generated by S121 is directly assigned to the corresponding target risk severity data.

[0158] S13. Convert the probability risk data and target risk severity data corresponding to each level into risk indicator values ​​for the corresponding level, and integrate them to obtain the first risk assessment data.

[0159] It should be noted that this step is the core of the third step of dynamic linkage optimization in the hierarchical probabilistic fuzzy coupling model. The input is the probability risk data and target risk severity data of each level generated by S12, and the output is the first risk assessment data. This step also embeds topology reconstruction triggering logic to provide a basis for determining whether to trigger dynamic topology reconstruction in the future.

[0160] The preset risk exceedance conditions include two categories: the first is the original operating quantity exceedance condition, including the transformer reverse load rate ≥80%, feeder voltage exceeding the limit ±5%, branch power flow overload 10%, and harmonic voltage content exceeding the standard; the second is the risk indicator value exceedance condition, that is, any risk indicator value at any level exceeds the corresponding threshold.

[0161] The topology reconfiguration triggering logic includes two types: 1. Single risk trigger: When any risk indicator value at the district level, feeder level, or substation level exceeds the corresponding preset threshold, topology dynamic reconfiguration is triggered; 2. Comprehensive risk trigger: When the first comprehensive risk value exceeds the preset comprehensive risk threshold, topology dynamic reconfiguration is triggered.

[0162] If neither of the above two triggering conditions is met, the current operating status is determined to meet the preset risk normal conditions, and the first risk assessment data is directly output.

[0163] Furthermore, step S13 may include the following sub-steps:

[0164] S131. Convert the probability risk data and target risk severity data corresponding to the transformer substation level into transformer substation level risk indicator values. The transformer substation level risk indicator values ​​include the severity of reverse heavy overload of the distribution transformer and the qualification rate of the low-voltage line.

[0165] In this embodiment of the invention, the probability risk data and target risk severity data corresponding to the transformer substation are extracted. Using the weighted average defuzzification and quantization conversion formula defined in step S113, two standardized values, namely the severeness of reverse overload of the transformer and the qualification rate of the low-voltage line, are calculated respectively. The qualification rate of the low-voltage line is a positive benefit indicator. The reverse standardization process is carried out on this value to generate the low-voltage line voltage non-compliance risk value. Finally, the severeness of reverse overload of the transformer and the low-voltage line voltage non-compliance risk value together constitute the transformer substation risk index value.

[0166] S132. Convert the probability risk data and target risk severity data corresponding to the feeder level into feeder level risk index values. The feeder level risk index values ​​include branch power flow over-limit risk value and feeder reverse load.

[0167] In this embodiment of the invention, the probability risk data and target risk severity data corresponding to the feeder level are extracted. The weighted average defuzzification and quantization conversion formula defined in step S113 is used to calculate the two standardized values ​​of branch power flow over-limit risk value and feeder reverse load, which together constitute the feeder level risk index value.

[0168] S133. Convert the probability risk data and target risk severity data corresponding to the substation level into substation-level risk index values. The substation-level risk index values ​​include node voltage deviation risk values ​​and total reverse load of the substation.

[0169] In this embodiment of the invention, the process of generating substation-level risk index values ​​is as follows: extract the probability risk data and target risk severity data corresponding to the substation level, and calculate the two standardized values ​​of node voltage deviation risk value and total reverse load of the substation respectively through the weighted average defuzzification quantization conversion formula defined in step S113, which together constitute the substation-level risk index value.

[0170] S134. The risk index values ​​at the distribution area level, feeder level, and substation level are integrated to obtain the first risk assessment data.

[0171] In this embodiment of the invention, the risk indicator values ​​at the distribution area level, feeder level, and substation level are directly collected and integrated to obtain the corresponding risk indicator vector:

[0172] ;

[0173] In the formula, These represent the risk values ​​at the transformer substation level, feeder level, and substation level before topology reconfiguration, respectively.

[0174] Based on the combined weights generated in steps S113-S115 Calculate the first comprehensive risk value:

[0175] ;

[0176] The risk indicator vector is integrated with the first comprehensive risk value to generate complete first risk assessment data, which serves as the core basis for triggering subsequent topology reconstruction. When the first comprehensive risk value or any level of risk indicator value exceeds the corresponding threshold, dynamic topology reconstruction is triggered, proceeding to step 104.

[0177] Step 103: When the first risk assessment data does not exceed the preset risk exceedance threshold, the first risk assessment data is used as the target risk assessment data, and the control strategy of maintaining the current operation mode is maintained.

[0178] In this embodiment of the invention, the first risk assessment data is compared with a preset risk exceedance threshold. The preset risk exceedance threshold includes two categories: one is the original operating quantity exceeding the limit (distribution transformer reverse load rate ≥80%, feeder voltage exceeding the limit ±5%, branch power flow overload 10%, harmonic voltage content exceeding the limit), and the other is the risk indicator value exceeding the limit (any risk indicator value at any level exceeds the corresponding threshold). If the first risk assessment data does not exceed any of the above preset risk exceedance thresholds, the current operating state of the distribution network is determined to be safe and stable, and the first risk assessment data is directly assigned as the target risk assessment data, without the need to perform subsequent topology reconstruction and secondary assessment processes.

[0179] Step 104: When the first risk assessment data exceeds the preset risk exceedance threshold, the topology dynamic reconstruction is triggered based on the first risk assessment data, and then a topology reconstruction optimization model is constructed based on the multi-source dataset and multiple candidate solutions are obtained.

[0180] It should be noted that this step is initiated when the first risk assessment data exceeds a preset risk exceedance threshold. The inputs are the first risk assessment data and a multi-source dataset. The core is to reduce the operational risk of the distribution network from the source through dynamic topology reconstruction. This is achieved by constructing a multi-objective optimization model, solving candidate topology schemes under constraints, generating optimized regulation instructions, creating switching instructions and maintenance suggestions, and integrating them to generate candidate solutions. Specifically, firstly, based on the multi-source dataset and the first risk assessment data, a multi-objective optimization model is constructed with the objectives of minimizing network losses, minimizing the amount of sensitive loads not restored, and minimizing the risk of reverse heavy overload. Then, under four types of constraints—radial topology, bidirectional limitation of photovoltaic power flow, equipment tolerance, and number of switching actions—a commercial integer programming solver is used to obtain multiple candidate topology schemes. Next, based on the network structure of each candidate topology scheme, photovoltaic output optimization and load optimization allocation are carried out to generate corresponding regulation instructions. Then, topology switching action instructions are generated according to the switching status of the candidate topology schemes, and equipment maintenance priority suggestion data is constructed by combining the high-risk, low-probability factors in the first risk assessment data. Finally, the regulation instructions, equipment maintenance priority suggestion data, and candidate topology schemes are integrated and packaged to generate multiple complete candidate solutions.

[0181] Furthermore, step 104 may include the following sub-steps:

[0182] S21. Construct a multi-objective optimization model with the objectives of minimizing network loss, minimizing the amount of sensitive load not restored, and minimizing the risk of reverse overload.

[0183] In this embodiment of the invention, based on the network topology parameters, equipment parameters, photovoltaic output data, and load data in the multi-source dataset, and combined with the reverse overload risk information in the first risk assessment data, three objective functions are constructed respectively, and integrated into a weighted multi-objective optimization model, the formula of which is:

[0184] ;

[0185] In the formula, The function to minimize network loss for the system; For sensitive loads that have not been recovered; To minimize the reverse overload risk function; The weights for the three objectives are defined. The three objective functions are then integrated according to their weights to ultimately construct a multi-objective optimization model.

[0186] S22. Under the constraints of radial topology, bidirectional photovoltaic power flow, equipment tolerance, and number of switching operations, a commercial integer programming solver is used to obtain multiple candidate topology schemes.

[0187] In this embodiment of the invention, the process of generating multiple candidate topology schemes is as follows:

[0188] 1. Set four types of constraints for the multi-objective optimization model: (1) Radial topology constraint: ensure that the distribution network always maintains a radial operating structure; (2) Photovoltaic power flow bidirectional restriction constraint: limit the transmission range of photovoltaic reverse power; (3) Equipment tolerance constraint: ensure that all electrical equipment operates within the rated parameter range; (4) Switching operation frequency constraint: control the operation frequency of switching equipment and extend the equipment life;

[0189] 2. Use a commercial integer programming solver (such as CPLEX) to solve the constrained multi-objective optimization model;

[0190] 3. Select the Pareto optimal solution set that satisfies all constraints. Each optimal solution corresponds to a combination of on / off states of tie switches and sectional switches. Finally, generate multiple candidate topology schemes. Optimize the cross-level power flow distribution by adjusting the network structure to reduce the risk of reverse overload and voltage limit exceedance from the source.

[0191] S23. Based on the network topology corresponding to the candidate topology scheme, optimize photovoltaic output and load allocation, and generate adjustment instructions corresponding to the candidate topology scheme.

[0192] In this embodiment of the invention, for each candidate topology scheme, based on its corresponding updated network topology, and combined with real-time photovoltaic power output data and load data from the multi-source dataset, photovoltaic power output curve optimization and optimal load allocation calculation for the entire network are performed respectively, generating adjustment instructions including photovoltaic inverter power output adjustment values ​​and load shedding / transfer instructions.

[0193] S24. Generate topology switch action instructions based on the switch status corresponding to the candidate topology scheme, and construct equipment operation and maintenance priority suggestion data corresponding to the candidate topology scheme by combining the high-risk, low-probability factors in the first risk assessment data.

[0194] In this embodiment of the invention, the on / off status of tie switches and sectionalizing switches in the candidate topology scheme is extracted and compared with the current actual switch status to generate topology switch action instructions containing switch number, action type (closing / opening), and action sequence. Combining the high-risk, low-probability factors mined from the first risk assessment data, all equipment in the distribution network is ranked by risk, and equipment with a high correlation to high-risk factors is prioritized, ultimately constructing equipment operation and maintenance priority suggestion data.

[0195] S25. Using the adjustment instructions corresponding to the candidate topology scheme, the equipment operation and maintenance priority suggestion data, and the candidate topology scheme respectively, multiple candidate solutions are constructed.

[0196] In this embodiment of the invention, for each candidate topology scheme, its corresponding adjustment command, topology switch action command, equipment operation and maintenance priority suggestion data are packaged and integrated with the candidate topology scheme itself to form a complete optimization control scheme; the above integration operation is performed on all candidate topology schemes to finally generate multiple candidate solutions.

[0197] Step 105: Based on the updated network state and multi-source datasets corresponding to each candidate solution, a secondary risk assessment is performed using a multinomial chaotic expansion proxy model coupled with precise power flow calculation to obtain the second risk assessment data corresponding to each candidate solution.

[0198] It should be noted that the scenario-based coupled assessment balances efficiency and accuracy: in non-fault steady-state scenarios, a trained multinomial chaotic expansion surrogate model is used to quickly calculate risk data; in scenarios with faults or topology reconfiguration, precise power flow calculations are used to accurately capture risk fluctuations. After converting the risk data into three levels of risk indicators—transformer area, feeder, and substation—objective weights are calculated using the entropy weight method, subjective weights are calculated using the analytic hierarchy process (AHP), and then combined weights are obtained through Nash equilibrium solution, thereby calculating the second comprehensive risk value. The calculation method for the second comprehensive risk value is completely consistent with that of the first comprehensive risk value, except that the input data is the updated operating state after optimized control. Finally, the three levels of risk indicators and the second comprehensive risk value are integrated to generate the second risk assessment data corresponding to each candidate solution.

[0199] Furthermore, step 105 may include the following sub-steps:

[0200] S31. Based on the updated network status of each candidate solution, determine the operating scenario corresponding to the candidate solution.

[0201] In this embodiment of the invention, for each candidate solution, the updated network state after its execution is analyzed, and the running scenarios are divided into three categories: non-fault steady-state scenario, fault scenario, and topology reconstruction scenario.

[0202] S32. When the operating scenario is a non-fault steady-state scenario, the electrical output is approximated by a polynomial chaotic expansion surrogate model to calculate the risk data corresponding to the candidate solution.

[0203] In this embodiment of the invention, when the operating scenario is a non-fault steady-state scenario, the multinomial chaotic expansion surrogate model trained in step 102 is used. The photovoltaic output, load fluctuation, and environmental variables in the multi-source dataset are used as inputs to quickly approximate electrical outputs such as node voltage, branch power flow, and reverse power. The over-limit probability of each electrical quantity is statistically analyzed, and finally, risk data under this scenario is generated.

[0204] S33. When the operating scenario is a fault scenario or a scenario after topology reconstruction, accurate power flow calculation is used to capture risk fluctuations and obtain risk data corresponding to candidate solutions.

[0205] In this embodiment of the invention, when the operating scenario is a fault scenario or a scenario after topology reconstruction, the proxy model is abandoned and the accurate power flow calculation method is directly called. Based on the updated network topology and operating parameters corresponding to the candidate solution, the accurate values ​​of voltage of each node and power flow of each branch are calculated point by point to accurately capture the risk fluctuations caused by topology changes or faults, and finally generate risk data under the scenario.

[0206] S34. Convert the risk data in this scenario into risk indicators of the corresponding level.

[0207] In this embodiment of the invention, the risk data generated in S32 or S33 is divided into three levels: distribution area, feeder, and substation. Using the weighted average defuzzification and quantization conversion formula defined in step S113, it is converted into risk indicators at the distribution area level, feeder level, and substation level, respectively, forming the risk indicators corresponding to the second risk assessment result, i.e., the risk indicator vector.

[0208] ;

[0209] In the formula, These represent the risk values ​​at the transformer substation level, feeder level, and substation level after topology reconfiguration, respectively.

[0210] S35. Based on risk indicators and multi-source datasets, the objective and subjective weights of risk indicators are calculated using the entropy weight method and the analytic hierarchy process, respectively, and the combined weights are obtained by solving the Nash equilibrium.

[0211] In this embodiment of the invention, a game theory-based combinatorial weighting method is used to generate combinatorial weights. The complete process is as follows:

[0212] 1. Entropy weight method for calculating objective weights: Based on historical operational data from multiple sources, the information entropy of each risk indicator is calculated to obtain objective weights, which reflect the degree of difference in the data itself;

[0213] 2. Calculation of subjective weights using the analytic hierarchy process: Based on the experience of distribution network operation experts, a judgment matrix is ​​constructed and a consistency check is performed to obtain subjective weights, which reflect the importance of different risks.

[0214] 3. Nash Equilibrium Solution for Combined Weights: By using the Nash equilibrium solution method, the differences between objective weights and subjective weights are balanced to obtain the optimal combined weights. These correspond to the weights of risk indicators at the distribution area, feeder, and substation levels, respectively. It should be noted that the second comprehensive risk value is calculated based on the same or redefined combined weights, and by default uses the exact same combined weights as the first comprehensive risk value. It is only redefined when there are significant changes in the power grid operation mode.

[0215] S36. Using risk indicators and corresponding combined weights, calculate the second comprehensive risk value.

[0216] In this embodiment of the invention, the risk indicators at each level generated in S34 are weighted and summed with the corresponding combined weights generated in S35 to finally calculate the second comprehensive risk value corresponding to the candidate solution. The formula is as follows:

[0217] ;

[0218] S37. Using the risk indicators and the second comprehensive risk value corresponding to each candidate solution, construct the second risk assessment data corresponding to the candidate solution.

[0219] In this embodiment of the invention, for each candidate solution, its corresponding risk indicator vector (three-level risk indicators of distribution area, feeder, and substation) is integrated with the second comprehensive risk value to finally construct the second risk assessment data corresponding to the candidate solution, whose structure is completely consistent with the first risk assessment data.

[0220] Step 106: Verify the validity of the second risk assessment data corresponding to each candidate solution. Use the second risk assessment data corresponding to the candidate solution that meets the preset validity conditions as the target risk assessment data, and output the optimized control scheme corresponding to the candidate solution.

[0221] The input for this step is all candidate solutions and their corresponding second risk assessment data, as well as the first comprehensive risk value before topology reconfiguration. The execution process is as follows:

[0222] 1. Define the preset validity condition: the second comprehensive risk value corresponding to the candidate solution is less than the first comprehensive risk value. (Proving the effectiveness of optimization), and the second comprehensive risk value is lower than the preset comprehensive risk threshold, while the values ​​of all individual risk indicators at the three levels of distribution area, feeder and substation are lower than the corresponding individual thresholds;

[0223] 2. Verify one by one whether the second risk assessment data corresponding to each candidate solution meets the preset validity conditions;

[0224] 3. Select candidate solutions that meet the criteria, and use their corresponding second risk assessment data as the target risk assessment data;

[0225] 4. Output the final risk assessment results, risk tracing report and optimization control plan. The system output includes: a three-level risk indicator distribution table, a high-risk area location map, risk transmission path analysis, an optimal topology diagram, optimization control instruction details, and an operation and maintenance priority list.

[0226] 5. The optimized control scheme includes: photovoltaic output adjustment commands, topology switch action commands, load optimization allocation scheme, and equipment operation and maintenance priority suggestions.

[0227] Optionally, this method also includes a risk tracing step: based on the correlation between the three levels of risk indicators, the risk source level and specific equipment are located through causal analysis algorithms; the mapping relationship between photovoltaic access location, penetration ratio and risk indicators is quantified to clarify the risk transmission path and provide a basis for differentiated analysis.

[0228] Please see Figure 3 , Figure 3 This is a structural block diagram of an active power distribution network risk assessment and optimization control system provided in an embodiment of the present invention.

[0229] This invention provides an active power distribution network risk assessment and optimization control system, comprising:

[0230] The parameter and data acquisition module 301 is used to acquire the static structural parameters, photovoltaic access parameters and multi-source heterogeneous operation data of the active distribution network, and construct a multi-source dataset.

[0231] The first risk assessment module 302 is used to perform multi-source data fusion and risk pre-assessment on multi-source datasets to obtain the first risk assessment data.

[0232] The risk assessment module 303 is a control strategy used to take the first risk assessment data as the target risk assessment data and maintain the current operating mode when the first risk assessment data does not exceed the preset risk exceedance threshold.

[0233] The topology reconstruction module 304 is used to trigger dynamic topology reconstruction based on the first risk assessment data when the first risk assessment data exceeds the preset risk exceedance threshold, and then construct a topology reconstruction optimization model based on the multi-source dataset and solve for multiple candidate solutions.

[0234] The second risk assessment and optimization module 305 is used to perform secondary risk assessment based on the updated network state and multi-source dataset corresponding to each candidate solution, using a coupling method of multinomial chaotic expansion proxy model and precise power flow calculation, to obtain the second risk assessment data corresponding to each candidate solution.

[0235] The result output module 306 is used to verify the validity of the second risk assessment data corresponding to each candidate solution, take the second risk assessment data corresponding to the candidate solution that meets the preset validity conditions as the target risk assessment data, and output the optimized control scheme corresponding to the candidate solution.

[0236] The modules interact with each other through a high-speed communication interface. Data transmission uses an encryption protocol to ensure security. The collaborative work between modules follows the process logic of "acquisition-preprocessing-evaluation-output" to ensure work efficiency and reliability.

[0237] Since the above is a system corresponding to an active distribution network risk assessment and optimization control method, and its implementation principle is consistent with that of an active distribution network risk assessment and optimization control method, for the sake of convenience and brevity, those skilled in the art can clearly understand that the specific working process of the system and modules described above can be referred to the corresponding process in the aforementioned method embodiments, and will not be repeated here.

[0238] Please see Figure 4 , Figure 4 This is a structural block diagram of an electronic device provided in an embodiment of the present invention.

[0239] An electronic device according to an embodiment of the present invention includes: a memory 401 and a processor 402. The memory 401 stores a computer program. When the computer program is executed by the processor 402, the processor 402 performs active power distribution network risk assessment and optimization control as described in any of the above embodiments.

[0240] Memory 401 may be an electronic memory such as flash memory, EEPROM (Electrically Erasable Programmable Read-Only Memory), EPROM, hard disk, or ROM. Memory 401 has storage space 403 for program code 413 for performing any of the method steps described above. For example, storage space 403 for program code may include individual program codes 413 for implementing the various steps in the methods described above. This program code may be read from or written to one or more computer program products. These computer program products include program code carriers such as hard disks, CDs, memory cards, or floppy disks. The program code may be compressed, for example, in a suitable form. When run by a computing processing device, this code causes the computing processing device to perform the various steps in the methods described above. This program code may be read from or written to one or more computer program products. These computer program products include program code carriers such as hard disks, CDs, memory cards, or floppy disks. The program code may be compressed, for example, in a suitable form. When these codes are run by a computing processing device, the device causes it to perform the various steps in the active distribution network risk assessment and optimization control method described above.

[0241] This invention also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the active power distribution network risk assessment and optimization control method as described in any of the above embodiments.

[0242] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0243] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, or indirect coupling or communication connection between apparatuses or units, and may be electrical, mechanical, or other forms.

[0244] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0245] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0246] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0247] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for risk assessment and optimization control of active power distribution networks, characterized in that, include: The static structural parameters, photovoltaic access parameters, and multi-source heterogeneous operation data of the active power distribution network are obtained to construct a multi-source dataset. The multi-source dataset is fused and risk pre-assessed to obtain the first risk assessment data. When the first risk assessment data does not exceed the preset risk exceedance threshold, the first risk assessment data is used as the target risk assessment data, and the control strategy of maintaining the current operation mode is maintained. When the first risk assessment data exceeds the preset risk exceedance threshold, topology dynamic reconstruction is triggered based on the first risk assessment data, and then a topology reconstruction optimization model is constructed based on the multi-source dataset and multiple candidate solutions are obtained. Based on the updated network state corresponding to each candidate solution and the multi-source dataset, a second risk assessment is performed by coupling a multinomial chaotic expansion proxy model with precise power flow calculation to obtain the second risk assessment data corresponding to each candidate solution. The validity of the second risk assessment data corresponding to each candidate solution is verified. The second risk assessment data corresponding to the candidate solution that meets the preset validity conditions is taken as the target risk assessment data, and the optimized control scheme corresponding to the candidate solution is output.

2. The active power distribution network risk assessment and optimization control method according to claim 1, characterized in that, The step of performing multi-source data fusion and risk pre-assessment on the multi-source dataset to obtain the first risk assessment data includes: The multi-source dataset is subjected to hierarchical preprocessing to generate a hierarchical fused dataset. The hierarchical fusion dataset is used for probabilistic fuzzy coupled inference to generate probabilistic risk data and target risk severity data at each level; The probability risk data and target risk severity data corresponding to each level are converted into risk indicator values ​​for the corresponding level, and then integrated to obtain the first risk assessment data.

3. The active power distribution network risk assessment and optimization control method according to claim 2, characterized in that, The step of performing data layering preprocessing on the multi-source dataset to generate a layered fused dataset includes: Probabilistic modeling is performed using real-time photovoltaic power output data from the multi-source dataset to generate photovoltaic power output probability distribution data. Using the electrical quantity data in the multi-source dataset, multiple risk level membership degrees are calculated. The risk level membership degree is weighted and defuzzified to generate initial risk severity data for each level; The environmental feature data and topological state data in the multi-source dataset are used to perform association rule mining to generate high-risk, low-probability factors and corresponding risk labels. The photovoltaic output probability distribution data, all the initial risk severity data, the high-risk low-probability factors and the corresponding risk labels are standardized and aligned to generate a hierarchical fusion dataset.

4. The active power distribution network risk assessment and optimization control method according to claim 2, characterized in that, The step of using the hierarchical fusion dataset to perform probabilistic fuzzy coupled inference to generate probabilistic risk data and target risk severity data at each level includes: Using the aforementioned hierarchical fusion dataset, the probabilistic flow of non-fault scenarios is calculated through a multinomial chaotic expansion surrogate model, generating probabilistic risk data at each level. The probability risk data at each level are compared with the corresponding preset probability thresholds. When any probability risk data exceeds the corresponding preset probability threshold, the corresponding level of fuzzy inference unit is invoked to calculate and generate the corresponding target risk severity data. When all probability risk data does not exceed the corresponding preset probability threshold, the probability risk data of each level is directly used as the corresponding target risk severity data.

5. The active power distribution network risk assessment and optimization control method according to claim 2, characterized in that, The step of converting the probability risk data and target risk severity data corresponding to each level into risk indicator values ​​for the corresponding level and integrating them to obtain the first risk assessment data includes: The probability risk data and target risk severity data corresponding to the transformer substation level are converted into transformer substation level risk index values, which include the severity of reverse heavy overload of distribution transformers and the qualification rate of low-voltage lines. The probability risk data and target risk severity data corresponding to the feeder level are converted into feeder level risk index values, which include branch power flow over-limit risk value and feeder reverse load. The probabilistic risk data and target risk severity data corresponding to the substation level are converted into substation-level risk index values, which include node voltage deviation risk values ​​and total reverse load of the substation. The risk index values ​​at the distribution area level, feeder level, and substation level are integrated to obtain the first risk assessment data.

6. The active power distribution network risk assessment and optimization control method according to claim 1, characterized in that, The steps of constructing a topology reconstruction optimization model based on the multi-source dataset and solving it to obtain multiple candidate solutions include: Construct a multi-objective optimization model with the objectives of minimizing network loss, minimizing the amount of sensitive load not restored, and minimizing the risk of reverse heavy overload; Under the constraints of radial topology, bidirectional photovoltaic power flow limitation, equipment tolerance and number of switching operations, a commercial integer programming solver was used to obtain multiple candidate topology schemes. Based on the network topology corresponding to the candidate topology scheme, photovoltaic power output optimization and load optimization allocation are performed to generate adjustment instructions corresponding to the candidate topology scheme. Based on the switch status corresponding to the candidate topology scheme, a topology switch action command is generated, and combined with the high-risk, low-probability factors in the first risk assessment data, equipment maintenance priority suggestion data corresponding to the candidate topology scheme is constructed. Multiple candidate solutions are constructed by using the adjustment instructions corresponding to the candidate topology scheme, the equipment operation and maintenance priority suggestion data, and the candidate topology scheme respectively.

7. The active power distribution network risk assessment and optimization control method according to claim 1, characterized in that, The step of performing a secondary risk assessment by coupling a polynomial chaotic expansion surrogate model with precise power flow calculation to obtain the second risk assessment data corresponding to each candidate solution includes: Based on the updated network state corresponding to each candidate solution, determine the operating scenario corresponding to the candidate solution; When the operating scenario is a non-fault steady-state scenario, the electrical output is approximated by a polynomial chaotic expansion proxy model, and the risk data corresponding to the candidate solution is calculated. When the operating scenario is a fault scenario or a scenario after topology reconfiguration, accurate power flow calculation is used to capture risk fluctuations and obtain the risk data corresponding to the candidate solution; Convert the risk data in this scenario into risk indicators of the corresponding level; Based on the risk indicators and the multi-source dataset, the objective weights and subjective weights of the risk indicators are calculated using the entropy weight method and the analytic hierarchy process, respectively, and the combined weights are obtained by solving the Nash equilibrium. The second comprehensive risk value is calculated using the aforementioned risk indicators and their corresponding combined weights. The second risk assessment data corresponding to each candidate solution is constructed by using the risk index and the second comprehensive risk value corresponding to each candidate solution.

8. An active power distribution network risk assessment and optimization control system, characterized in that, include: The parameter and data acquisition module is used to acquire the static structural parameters, photovoltaic access parameters and multi-source heterogeneous operation data of the active power distribution network, and construct a multi-source dataset. The first risk assessment module is used to perform multi-source data fusion and risk pre-assessment on the multi-source dataset to obtain the first risk assessment data. The risk assessment module is a control strategy used to take the first risk assessment data as the target risk assessment data and maintain the current operating mode when the first risk assessment data does not exceed the preset risk exceedance threshold. The topology reconstruction module is used to trigger dynamic topology reconstruction based on the first risk assessment data when the first risk assessment data exceeds the preset risk exceedance threshold, and then construct a topology reconstruction optimization model based on the multi-source dataset and solve it to obtain multiple candidate solutions. The second risk assessment and optimization module is used to perform a second risk assessment based on the updated network state corresponding to each candidate solution and the multi-source dataset, using a multinomial chaotic expansion proxy model coupled with precise power flow calculation, to obtain the second risk assessment data corresponding to each candidate solution. The result output module is used to verify the validity of the second risk assessment data corresponding to each candidate solution, take the second risk assessment data corresponding to the candidate solution that meets the preset validity conditions as the target risk assessment data, and output the optimized control scheme corresponding to the candidate solution.

9. An electronic device, characterized in that, The system includes a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor causes the processor to perform the steps of the active power distribution network risk assessment and optimization control method as described in any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed, it implements the active power distribution network risk assessment and optimization control method as described in any one of claims 1-7.

Images