Power distribution network operation risk assessment and fault early warning method and system based on risk perception overload prediction

Abstract

The application provides a power distribution network operation risk assessment and fault early warning method and system based on risk perception super overload prediction, and belongs to the technical field of power system operation analysis and electric vehicle access to the power distribution network. The method comprises the following steps: multi-source data acquisition and feature construction, risk perception super overload prediction, power distribution network operation capacity, fault risk correlation and hierarchical early warning, collaborative response and closed-loop feedback; the technical scheme of the application constructs a risk perception load prediction and operation risk assessment closed-loop mechanism for the operation safety of the power distribution network, improves the accuracy and interpretability of the operation evaluation of the power distribution network, enhances the engineering applicability, and thus reduces the operation risk and equipment failure probability caused by the access of electric vehicles.

Description

Technical Field

[0001] This invention belongs to the field of power system operation analysis and electric vehicle access to distribution network technology, and relates to a method and system for distribution network operation risk assessment and fault early warning based on risk perception and overcharging load prediction. Background Technology

[0002] my country's energy structure and end-use energy patterns are rapidly transforming towards low-carbon and electrification. As a crucial component of energy consumption and carbon emissions, the large-scale promotion of electric vehicles in the transportation sector has become a vital path for energy transition and low-carbon transportation development. Research indicates that, under my country's power structure, pure electric vehicles can significantly reduce carbon emissions throughout their entire lifecycle compared to traditional gasoline-powered vehicles. Furthermore, as the proportion of renewable energy in the power system continues to increase, their emission reduction advantages will be further enhanced.

[0003] However, with the rapid growth in the number of electric vehicles and the continuous increase in charging demand, electric vehicle charging load has gradually evolved into a new type of electricity load in the distribution network, characterized by large scale, rapid growth, and complex temporal and spatial characteristics. In particular, the concentrated access of high-power fast charging / supercharging stations has made the charging load exhibit obvious randomness, concentration, and impact. During typical peak electricity consumption periods, the charging load overlaps with existing residential and commercial loads, which may lead to increased local line load rates, aggravated voltage deviations, heavy or even overloaded distribution transformers, and aggravated three-phase imbalance, posing new challenges to the safe operation and planning design of the distribution network.

[0004] Currently, some progress has been made in the research and engineering application of the impact of electric vehicles on the operation of power distribution networks, but the following shortcomings still exist in engineering practice: (1) Insufficient data authenticity: Some methods rely on idealized or hypothetical charging behavior models, which make it difficult to realistically depict the temporal fluctuations and spatial differences of charging load under different regions and site conditions, resulting in limited applicability of the analysis results in engineering applications; (2) Insufficient characterization of high-risk peak periods: Traditional load forecasting or regression methods often take the overall error as the optimization target, which is prone to the problem of underestimating peak load, resulting in the potential for missed judgment of distribution network operation risks; (3) Lack of a correlation mechanism between “operating stress and failure risk”: Existing assessments usually remain at the level of voltage and current over-limit or imbalance calculation, and fail to establish a mapping relationship between operating stress and specific equipment failure mechanisms (such as thermal aging, joint overheating, protection maloperation, insulation flashover, etc.), resulting in assessment results lacking clear engineering orientation and interpretability, and making it difficult to directly support preventive operation and maintenance decisions. (4) Lack of closed-loop handling and feedback mechanism: Existing methods are mostly post-event analysis or offline assessment, lacking a closed-loop operation mechanism from risk identification, early warning triggering to response control and model update.

[0005] Therefore, in the context of large-scale electric vehicle access, there is an urgent need for an analysis method and system that can integrate multi-source real data and be oriented towards the safety goals of power distribution network operation. This would enable risk perception and prediction of charging load, quantification of power distribution network operation stress, fault risk classification and early warning, and coordinated response, thereby providing reliable technical support for power distribution network planning and evaluation, operation monitoring, and scheduling and maintenance. Summary of the Invention

[0006] In view of this, the purpose of this invention is to provide a method and system for risk assessment and fault early warning of distribution network operation based on risk perception overcharging load prediction. By constructing a closed-loop mechanism for risk perception load prediction and operation risk assessment oriented towards distribution network operation safety, the accuracy and interpretability of distribution network operation assessment are improved, and the engineering applicability is enhanced, thereby reducing the operation risks and equipment failure probability caused by electric vehicle access.

[0007] To achieve the above objectives, the present invention provides the following technical solution: A method for risk assessment and fault early warning of distribution network operation based on risk perception and overcharging load prediction, the method specifically includes the following steps: S1. Multi-source data acquisition and feature construction: Collect supercharging station operation data, distribution network operation and topology data, meteorological environment data and spatial attribute data, preprocess them and construct a multi-dimensional feature dataset for load forecasting; S2, Risk-aware overcharging load prediction: The multidimensional feature dataset is input into the pre-constructed risk-aware overcharging load prediction model, and the overcharging station load prediction sequence within the future prediction time window is output; wherein, the risk-aware overcharging load prediction model adopts an asymmetric risk-aware loss function during training, which assigns higher weight to the prediction error of the load period exceeding the preset safety threshold. S3. Distribution network operation stress quantification: The supercharging station load prediction sequence is mapped to the corresponding node of the distribution network, superimposed with the base load, and then time-series power flow calculation is performed. The operation stress index of the distribution network is quantified based on the power flow calculation results. S4. Fault Risk Correlation and Graded Early Warning: Based on the operating stress index, according to the preset correlation rules between operating stress and fault risk, the potential fault risk type and its level are determined, and corresponding graded early warning information is generated.

[0008] Furthermore, in step S1, the collection of supercharging station operation data, distribution network operation and topology data, meteorological environment data, and spatial attribute data specifically includes: Supercharging station side operation data: Collect electrical quantity information such as voltage, current, active power, reactive power and power factor at the supercharging station level and pile level; the data sampling frequency is set to the second level or higher, and the data storage granularity is set to the minute level time scale according to application requirements to take into account the needs of load impact characteristic capture and system operation analysis; obtain charging session logs based on charging pile communication protocols (such as OCPP), including vehicle access time, charging start and end time, vehicle departure time, as well as charging demand information such as initial state of charge, target state of charge, battery capacity and maximum allowable charging power, and collect charging mode and equipment operation status identifiers; Distribution network operation and topology data: Collect the topology and equipment parameter information of the distribution network, including line type, length, impedance parameters, transformer rated capacity and wiring method, etc., to build the distribution network model; at the same time, collect the active and reactive load and voltage information of the gateway and key nodes, and obtain the setting parameters of the relay protection device for subsequent operation risk assessment and boundary determination. Meteorological and environmental data: Collect meteorological data of the area where the supercharging station is located, including ambient temperature, humidity, rainfall, etc., and introduce corresponding environmental status indicators when extreme weather events occur for subsequent risk analysis; Spatial and functional attribute data: Based on the spatial location of the supercharging station, data on surrounding points of interest are obtained, and the distribution characteristics of different types of points of interest are statistically analyzed to characterize the functional attributes and potential charging demand patterns of the area where the station is located.

[0009] Furthermore, in step S1, the data preprocessing and multidimensional feature construction specifically include: 1) Timing alignment and resampling processing Using the unified time base of the power distribution network operation system as a reference, data with different time resolutions are aligned to ensure that all types of data have a one-to-one correspondence at the same time scale. 2) Missing value repair and outlier removal To address missing or abnormal data during the data collection process, data is repaired or removed based on historical statistical characteristics and physical constraints to prevent abnormal data from interfering with load forecasting and risk assessment results. 3) Multidimensional feature construction After data cleaning, a high-dimensional feature set that reflects the historical state of the load, meteorological environment characteristics, spatial functional attributes and time series information is constructed to describe the spatiotemporal evolution characteristics of the overcharge load. 4) Feature normalization processing To eliminate the influence of features with different dimensions on model training, the constructed features are normalized to improve the stability of model training and prediction accuracy.

[0010] Furthermore, in step S2, based on the constructed multidimensional spatiotemporal feature dataset, a risk-aware overcharging load prediction model integrating signal decomposition and deep learning is constructed. This model introduces an asymmetric risk-oriented mechanism during the training phase to address the problems of traditional prediction methods underestimating peak overcharging loads and easily missing operational risks. Specifically, this includes: S21. Construct an adaptive multi-scale load decomposition module and perform the load decomposition sub-step: Perform adaptive multi-scale decomposition on the original overcharge load sequence to obtain multiple sub-sequence components with different time scale characteristics; including: by introducing a parameter adaptive optimization mechanism, optimize the key parameters of the load decomposition model so that the decomposition result reaches the optimal in terms of energy concentration or feature correlation; based on the optimized parameter combination, decompose the original overcharge load sequence into several sub-sequences with different time scale characteristics, including low-frequency components reflecting the long-term evolution trend of the load and high-frequency components reflecting the characteristics of random shocks; subsequently, align each decomposed component with the multi-dimensional feature vector constructed in step S1 to form a sample set for load prediction at different scales. S22. Construct a spatiotemporal feature extraction and fusion prediction network, and perform feature extraction and prediction sub-steps: Construct a prediction network to extract spatiotemporal features from the subsequence components and the multidimensional feature dataset, and output the prediction results for each component; the prediction network includes at least: a local feature extraction unit, used to extract load fluctuation features and coupling relationships between features from a short time window; a time-series dependency modeling unit, used to characterize the long-term memory characteristics and time correlation of the load sequence; and a key information weighting unit, used to dynamically weight the importance of different time steps and feature dimensions; S23. Design an asymmetric risk-perception loss function: Introduce a risk-perception loss function during the model training phase to enhance the model's ability to fit high-risk load intervals; the risk-perception loss function can be expressed as:

[0011] in: Indicates time The actual overcharge load value; This represents the predicted load value at the corresponding time. This represents a dynamic weighting coefficient related to operational risk; the weighting coefficient The model is adaptively adjusted based on the relationship between the load level and the preset safety threshold. When the load level exceeds the safety warning threshold, the prediction error is given a higher weight, thereby guiding the model to focus on high-risk load periods during the training process. S24. Load Reconfiguration and Forecast Output: The forecast results of each component are reconfigured to obtain the supercharging station load forecast sequence. The supercharging station load forecast sequence in the future forecast time domain is output for subsequent distribution network time-series power flow calculation and operation risk assessment.

[0012] Furthermore, step S3 specifically includes: S31. Overcharge Mapping and Node Injection Modeling Based on the actual connection location of the supercharging station in the distribution network, the supercharging station load prediction sequence output in step S2 is mapped to the corresponding distribution network nodes to construct a node load injection model. At the same time scale, the predicted supercharging station load is superimposed with the original basic load of the distribution network to form the total node load including the supercharging load, expressed as follows:

[0013] in, Indicates the node at time [time]. The base load, This represents the predicted load of the supercharging station; under the condition of three-phase operation, the load can also be injected into the corresponding phase according to the access phase of the supercharging load to construct a three-phase unbalanced load injection model; S32, Distribution Network Time-Sequence Power Flow Calculation Based on the distribution network topology and equipment parameters obtained in step S1, a distribution network operation model is constructed, and time-series power flow calculations are performed within the prediction time window. In each time step, the distribution network power flow equations are solved to obtain the voltage magnitude and phase angle of each node, as well as the current, power, and load status of each line and transformer. The power flow calculations can be implemented using power flow analysis methods suitable for distribution networks, including but not limited to forward backward substitution, Newton-Raphson method, etc. This invention is not limited to specific calculation methods or software implementation platforms. Through time-series power flow analysis, time-series results of key operating state quantities of the distribution network within the prediction time range are obtained. S33, Calculation of Operating Stress Indicators Based on the time-series power flow calculation results, the operating status of the distribution network is quantitatively analyzed, and operating stress indicators that can reflect the operating pressure and potential risks of the distribution network are extracted. These operating stress indicators include at least: voltage operating stress indicators: calculating the degree of voltage deviation of each node relative to the allowable operating range, and statistically analyzing the magnitude and duration of voltage exceedances to characterize voltage stability risk; equipment thermal stress indicators: calculating the line current carrying rate and transformer load rate based on line current and equipment rated capacity, and statistically analyzing the duration exceeding a preset safety threshold to characterize the thermal overload risk of lines and transformers; and three-phase unbalance operating stress indicators: calculating the unbalance index based on the three-phase voltage or current of nodes under three-phase modeling conditions to assess the operating stress level caused by unbalanced load access. Through the calculation of the above operating stress indicators, the operating status of the distribution network is transformed from a single electrical quantity result into a set of stress indicators that can characterize the degree of risk. S34, Operating Stress Timing Output Various operational stress indicators are summarized in chronological order to form a time series dataset of operational stress of the distribution network within the prediction time window, which serves as the input for operational stress-fault risk correlation analysis and graded early warning determination in step S4.

[0014] Furthermore, step S4 specifically includes: S41. Construction of Operating Stress Feature Set The operating stress time series data output in step S3 is processed to construct an operating stress feature set for fault risk analysis. The operating stress feature set includes at least the following elements: node voltage over-limit amplitude and over-limit duration; line current carrying rate, transformer load rate and their over-limit duration; three-phase unbalance index (under three-phase modeling conditions); operating stress change rate and cumulative stress level. S42. Construction of operational stress-failure risk association rules Based on the operating mechanism and engineering experience of power distribution network equipment, a rule base for the association between operating stress and potential fault types is constructed to realize the mapping from "operating status" to "fault risk". The association rules include at least the following types: Thermal stress correlation rule: When the line current carrying rate or transformer load rate continuously exceeds the preset safety threshold and the operating ambient temperature is high, it is determined that the equipment has the risk of thermal-related faults such as insulation aging and joint overheating. Voltage stress correlation rule: When the node voltage continuously exceeds the limit or the voltage fluctuation is large, it is determined that there is a risk of low voltage tripping, decreased voltage stability or malfunction of protection device; Unbalanced stress correlation rule: When the three-phase unbalance exceeds the preset threshold, it is determined that the equipment is at risk of increased additional losses, local overheating and shortened service life; Environmental Co-operation Risk Rule: When the operating stress is high and severe weather conditions occur at the same time, it is determined that there is a risk of environmentally coupled faults such as insulation flashover and tree line discharge. S43. Fault Risk Assessment and Classification Based on whether the operating stress index meets the triggering conditions of the association rule, the type of potential failure risk is determined, and the failure risk is graded and assessed according to the stress level, duration and superposition effect. S44. Tiered Early Warning Output and Decision Support Based on the fault risk level, corresponding graded early warning information is generated and output to the power distribution network operation management or maintenance system; the early warning information includes at least: the potential fault type; the location of the equipment or area where it may occur; the risk level and the recommended period of attention.

[0015] Furthermore, the method also includes step S5: the collaborative response and closed-loop feedback step, which specifically includes: S51. Risk Warning Result Analysis and Response Triggering The graded early warning information generated in step S4 is parsed, and corresponding collaborative response strategies are triggered according to the early warning level and the fault risk type. The mapping relationship between the early warning level and the response method includes: low risk early warning: enter the operation monitoring enhancement state and increase the monitoring frequency of the corresponding area or equipment; medium risk early warning: trigger preventive control or operation adjustment strategies; high risk early warning: trigger emergency response strategies and issue alarm prompts to the scheduling or operation and maintenance system. S52, Cooperative Response Strategy Generation Based on different types of failure risks, corresponding collaborative response strategies are generated, and the strategies include at least one or more of the following: Load-side adjustment strategy: When a high-risk state caused by overcharging load is detected, the operating power of the overcharging station is limited, reduced, or scheduled in a time-sharing manner to reduce the local operating stress level; Grid-side operation adjustment strategy: Based on the location and type of risk, perform operation operations such as network reconfiguration, tap adjustment or reactive power compensation control to improve node voltage levels or alleviate equipment overload; Operation and maintenance side handling strategy: For identified high-risk equipment or areas, generate inspection, maintenance or key attention suggestions to guide operation and maintenance personnel to carry out targeted handling; S53, Response Execution and Running Status Update After the coordinated response strategy is generated, the corresponding control instructions or handling suggestions are sent to the relevant execution units, including but not limited to the charging station control system, the distribution network dispatching system and the operation and maintenance management system; after the response is executed, the updated operation data is collected in real time, and the operation status of the distribution network is reassessed to determine whether the operation stress has been effectively alleviated. S54. Risk Assessment Results Feedback and Model Update The operational status data after the response is executed is fed back to steps S1 and S2 to correct the load prediction model and risk assessment parameters, so as to realize the adaptive update of the model. Through the continuous feedback mechanism, the prediction model and risk criteria can be continuously optimized with the changes in the operating environment and load characteristics, forming a self-learning risk assessment and early warning closed loop, so that the system has the ability to adaptively update with load evolution, equipment aging and changes in the operating environment.

[0016] This invention also provides a distribution network operation risk assessment and fault early warning system based on risk perception and overcharging load prediction, characterized in that the system employs the method described above; the system includes: The multi-source data acquisition module is used to collect supercharging station operation data, distribution network operation and topology data, meteorological environment data, and spatial attribute data; The data preprocessing and feature construction module is used to preprocess the collected multi-source data and construct a multidimensional feature dataset; The risk perception overcharging load prediction module has a built-in risk perception overcharging load prediction model, which is used to output the overcharging station load prediction sequence based on the multidimensional feature dataset. The distribution network operation analysis module is used to perform time-series power flow calculations based on the supercharging station load prediction sequence and distribution network topology parameters; The stress quantification module is used to calculate the operating stress index of the distribution network based on the time-series power flow calculation results. The fault risk assessment and graded early warning module is used to output the fault risk type and early warning level based on the operating stress index and preset association rules. The collaborative response and closed-loop feedback module is used to trigger operational adjustments based on the early warning results and to update the parameters of the risk perception overload prediction module and / or the fault risk assessment and graded early warning module.

[0017] Furthermore, the risk-aware supercharging load prediction module includes: a load decomposition unit for adaptively decomposing the original supercharging load sequence into multiple scales; a fusion prediction network unit for extracting spatiotemporal features and predicting each component after decomposition; and a load reconstruction unit for reconstructing the prediction results of each component into a complete supercharging station load prediction sequence.

[0018] The beneficial effects of this invention are as follows: Compared with the prior art, the present invention has at least the following beneficial effects: 1) To achieve pre-emptive identification and graded early warning of power distribution network operation risks under the condition of electric vehicle access; 2) Improve the accuracy of forecasting peak overcharging load periods and reduce the probability of missing operational risks; 3) Establish the correlation between operating stress and equipment failure risk to improve the engineering interpretability of the assessment results; 4) Supports coordinated response decisions from the load side, grid side, and operation and maintenance side; 5) Construct a closed-loop risk assessment system that can be adaptively updated as the operating environment and load evolve.

[0019] Compared to methods based solely on static load or post-event analysis, this invention is more suitable for actual power distribution network operation monitoring and preventive maintenance scenarios.

[0020] Other advantages, objectives, and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination, or may be learned from practice of the invention. The objectives and other advantages of the invention can be realized and obtained through the following description. Attached Figure Description

[0021] To make the objectives, technical solutions, and advantages of the present invention clearer, the preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings, wherein: Figure 1 This is a schematic diagram of the functional module structure of the system of the present invention; Figure 2 This is a flowchart illustrating the overall process of the method of the present invention. Figure 3 This is a diagram showing the execution sequence of the method steps of the present invention. Detailed Implementation

[0022] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0023] The accompanying drawings are for illustrative purposes only and are schematic diagrams, not actual pictures. They should not be construed as limiting the invention. To better illustrate the embodiments of the invention, some parts in the drawings may be omitted, enlarged, or reduced, and do not represent the actual product dimensions. It is understandable to those skilled in the art that some well-known structures and their descriptions may be omitted in the drawings.

[0024] In the accompanying drawings of the embodiments of the present invention, the same or similar reference numerals correspond to the same or similar components. In the description of the present invention, it should be understood that if terms such as "upper," "lower," "left," "right," "front," and "rear" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, they are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, the terms used to describe positional relationships in the drawings are only for illustrative purposes and should not be construed as limiting the present invention. For those skilled in the art, the specific meaning of the above terms can be understood according to the specific circumstances.

[0025] This invention provides a method and system for risk assessment and fault early warning of electric vehicles (especially high-power charging stations) connected to the power distribution network based on multi-source real data and risk perception prediction. It is used to realize the pre-assessment of the power distribution network operation status, risk classification early warning and collaborative response closed-loop control under the condition of large-scale electric vehicle access. Figure 1 This is a schematic diagram of the functional module structure of the system of the present invention. Figure 2 This is a flowchart illustrating the overall process of the method of the present invention. Figure 3 This is a flowchart illustrating the execution sequence of the method steps of the present invention; its overall flow is as follows: Figure 2 , Figure 3 As shown, the process includes, in sequence, multi-source data acquisition and feature construction, risk perception and overcharge load prediction, distribution network time-series power flow analysis, operational stress quantification, fault risk correlation and graded early warning, and collaborative response and closed-loop feedback.

[0026] The system structure of this invention is as follows: Figure 1 As shown, the system includes at least the following functional modules: 1. Multi-source data acquisition module, used to collect supercharging station operation data, charging session data, power distribution network operation and topology data, meteorological environment data, and spatial attribute data; 2. Data preprocessing and feature construction module, used for time-series alignment, missing value repair, anomaly removal, and multi-dimensional feature construction of multi-source heterogeneous data; 3. Risk perception overcharging load prediction module, used to output the overcharging station load prediction sequence within the prediction time window; 4. Distribution network operation analysis module, used to perform time-series power flow calculations for distribution networks including overcharged load access conditions; 5. The stress quantification module is used to calculate voltage operating stress, equipment thermal stress, and three-phase unbalanced operating stress; 6. Fault risk assessment and graded early warning module, used to output fault risk type and early warning level based on operational stress and correlation rules; 7. The collaborative response and closed-loop feedback module is used to trigger operational adjustments based on the early warning results and update the model parameters.

[0027] The modules mentioned above interact with each other through data interfaces. The system can be deployed in a centralized or distributed manner, and this invention does not limit this.

[0028] like Figure 2 and Figure 3 As shown, the method of the present invention includes the following steps: Step S1: Multi-source spatiotemporal data acquisition and feature construction S1.1 Multi-source heterogeneous data acquisition system To achieve accurate characterization of the load characteristics of supercharging stations and their correlation analysis with the operational risks of distribution networks, this invention constructs a multi-source heterogeneous data acquisition system covering "source-grid-load-environment".

[0029] (1) Operational data of supercharging stations The system collects electrical quantity information such as voltage, current, active power, reactive power, and power factor at both the station and pile levels of the supercharging station. The sampling frequency of the data can be set to the second level or higher, and the data storage granularity can be set to the minute level according to application requirements, so as to take into account both load impact characteristic capture and system operation analysis needs.

[0030] Meanwhile, charging session logs are obtained based on charging pile communication protocols (such as OCPP), including vehicle access time, charging start and end time, vehicle departure time, and charging demand information such as initial state of charge, target state of charge, battery capacity and maximum allowable charging power, and charging mode and equipment operation status identifiers are collected.

[0031] (2) Distribution network side operation and topology data The system collects information on the topology and equipment parameters of the distribution network, including line type, length, impedance parameters, transformer rated capacity and wiring method, to build a distribution network model. At the same time, it collects active and reactive load and voltage information at key points and critical nodes, and obtains the setting parameters of relay protection devices for subsequent operation risk assessment and boundary determination.

[0032] Where conditions permit, historical fault and maintenance records can also be collected to help establish the correlation between operating status and fault risk.

[0033] (3) Meteorological and environmental data Meteorological data, including ambient temperature, humidity, and rainfall, is collected from the area where the supercharging station is located. In the event of extreme weather events, corresponding environmental status indicators are introduced for subsequent risk analysis.

[0034] (4) Spatial and functional attribute data Based on the spatial location of the supercharging station, data on surrounding points of interest are obtained, and the distribution characteristics of different types of points of interest are statistically analyzed to characterize the functional attributes and potential charging demand patterns of the area where the station is located.

[0035] S1.2 Data Preprocessing and High-Dimensional Feature Construction Methods To address the inconsistencies in time scale, sampling frequency, and units of multi-source heterogeneous data, this invention performs unified data preprocessing and feature construction on the collected data to ensure the integrity, consistency, and physical rationality of the subsequent model input data.

[0036] (1) Timing alignment and resampling processing Using the unified time base of the power distribution network operation system as a reference, data with different time resolutions are aligned to ensure that all types of data have a one-to-one correspondence at the same time scale.

[0037] (2) Missing value repair and outlier removal To address missing or abnormal data during the data collection process, data is repaired or removed based on historical statistical characteristics and physical constraints to prevent abnormal data from interfering with load forecasting and risk assessment results.

[0038] (3) Construction of multidimensional features After data cleaning, a high-dimensional feature set is constructed to comprehensively reflect the historical state of the load, meteorological environment characteristics, spatial functional attributes, and time series information, which is used to describe the spatiotemporal evolution characteristics of the overcharging load.

[0039] (4) Feature normalization processing To eliminate the influence of features with different dimensions on model training, the constructed features are normalized to improve the stability of model training and prediction accuracy.

[0040] Through the above steps, a feature dataset is formed that can be directly used as input for the risk perception overload prediction model.

[0041] Step S2: Construct a risk-aware-based prediction model for combined overcharging and load. Based on the multidimensional spatiotemporal feature dataset constructed in step S1, this invention constructs a risk-aware overcharging load combined prediction model that integrates signal decomposition and deep learning. The model introduces an asymmetric risk-oriented mechanism during the training phase to address the problems of traditional prediction methods underestimating peak overcharging loads and easily missing operational risks.

[0042] S2.1 Adaptive Multi-Scale Load Decomposition Module In view of the characteristics of strong non-stationarity and obvious multi-scale fluctuations in overcharge load time series, this invention performs adaptive decomposition processing on the original load series to achieve noise suppression and feature decoupling.

[0043] Specifically, by introducing a parameter adaptive optimization mechanism, the key parameters of the load decomposition model are optimized to achieve the optimal decomposition results in terms of energy concentration or feature correlation. In a preferred embodiment, a swarm intelligence optimization algorithm (such as the sparrow search algorithm) can be used to adaptively adjust the number of modes and the penalty factor of the variational mode decomposition model.

[0044] Based on the optimized parameter combination, the original overcharge load sequence is decomposed into several subsequences with different time scale characteristics, including low-frequency components reflecting the long-term evolution trend of the load and high-frequency components reflecting the characteristics of random shocks.

[0045] Subsequently, each decomposed component is aligned with the multidimensional feature vector constructed in step S1 to form a sample set for load prediction at different scales.

[0046] S2.2 Spatiotemporal Feature Extraction and Fusion Prediction Network For each load component obtained from the decomposition, a prediction network structure is constructed to extract temporal features and achieve multi-dimensional information fusion.

[0047] The prediction network includes at least: a local feature extraction unit for extracting load fluctuation features and coupling relationships between features within a short time window; a time-dependency modeling unit for characterizing the long-term memory characteristics and time correlation of the load sequence; and a key information weighting unit for dynamically weighting the importance of different time steps and feature dimensions. In a preferred embodiment, the above functional units can be implemented by convolutional neural networks, bidirectional recurrent neural networks, and attention mechanisms, respectively, but the present invention is not limited to a specific network structure or implementation.

[0048] S2.3 Design of Asymmetric Risk Perception Loss Function To avoid the problem of traditional prediction models failing to predict high load peaks during training, this invention introduces a risk-aware loss function during the model training phase to enhance the model's ability to fit high-risk load intervals.

[0049] The risk perception loss function can be expressed as:

[0050] in: Indicates time The actual overcharge load value; This represents the predicted load value at the corresponding time. This represents the dynamic weighting coefficient associated with operational risks.

[0051] The weighting coefficient The model adaptively adjusts based on the relationship between the load level and the preset safety threshold. When the load level exceeds the safety warning threshold, the prediction error is given a higher weight, thereby guiding the model to focus on high-risk load periods during the training process.

[0052] S2.4 Load Reconfiguration and Predictive Output The above-mentioned risk perception prediction model outputs the supercharging station load prediction sequence in the future prediction time domain, which is used for subsequent distribution network time-series power flow calculation and operation risk assessment.

[0053] Compared with traditional load forecasting methods, the method described in this invention significantly improves the forecasting accuracy for peak load periods while maintaining overall forecasting accuracy, effectively reducing the problem of missed operational risks caused by underestimation of load.

[0054] S2 Technology Performance Description The risk-aware overcharging load forecasting method described in this step can achieve the following technical effects: enhance the model's ability to learn the peak characteristics of overcharging load and avoid the "peak shaving" phenomenon in the forecast results; match the load forecast results with the safety requirements of distribution network operation and provide reliable input for subsequent risk assessment; and provide a quantifiable load forecasting basis for distribution network operation risk assessment and fault early warning.

[0055] Step S3: Distribution network time-series power flow analysis and operational stress quantification method based on predicted load Based on the future load prediction sequence of the supercharging station obtained in step S2, this step constructs a distribution network time-series operation analysis model that includes the access of supercharging loads. The operation status changes of the distribution network under the impact of supercharging loads are quantified through time-series power flow calculation, and multi-dimensional operation stress indicators reflecting the pressure on equipment and network are extracted, providing a basis for subsequent fault risk correlation and graded early warning.

[0056] S3.1 Overcharge Load Mapping and Node Injection Modeling Based on the actual access location of the supercharging station in the distribution network, the supercharging station load prediction sequence output in step S2 is mapped to the corresponding distribution network node to construct a node load injection model.

[0057] At the same time scale, the predicted supercharging station load is superimposed with the original basic load of the distribution network to form the total node load including the supercharging load, which is expressed as follows:

[0058] in, Indicates the node at time [time]. The base load, This represents the predicted load of the supercharging station. Under three-phase operating conditions, the load can be injected into the corresponding phase based on the connected phase of the supercharging load, thus constructing a three-phase unbalanced load injection model.

[0059] S3.2 Distribution Network Time-Sequence Power Flow Calculation Based on the distribution network topology and equipment parameters obtained in step S1, a distribution network operation model is constructed, and time-series power flow calculations are performed within the prediction time window.

[0060] Within each time step, the power flow equations of the distribution network are solved to obtain the voltage magnitude and phase angle of each node, as well as the current, power and load status of each line and transformer.

[0061] The power flow calculation can be implemented using power flow analysis methods suitable for distribution networks, including but not limited to forward backsubstitution and Newton-Raphson methods. This invention is not limited to specific calculation methods or software implementation platforms.

[0062] By using time-series power flow analysis, time-series results of key operating state quantities of the distribution network within the predicted time range are obtained.

[0063] S3.3 Calculation of Operating Stress Indicators Based on the time-series power flow calculation results, the operating status of the distribution network is quantitatively analyzed, and operating stress indicators that can reflect the operating pressure and potential risks of the distribution network are extracted.

[0064] The operational stress parameters include at least: Voltage operating stress index: Calculate the degree of voltage deviation of each node relative to the allowable operating range, and count the magnitude and duration of voltage exceeding the limit to characterize voltage stability risk.

[0065] Equipment thermal stress index: Based on the line current and the rated capacity of the equipment, the line current carrying rate and transformer load rate are calculated, and the duration of exceeding the preset safety threshold is statistically analyzed to characterize the thermal overload risk of the line and transformer.

[0066] Three-phase unbalanced operating stress index: Under three-phase modeling conditions, the unbalance index is calculated based on the three-phase voltage or current at the nodes and is used to assess the operating stress level caused by unbalanced load connection.

[0067] By calculating the above-mentioned operational stress indicators, the operating status of the distribution network is transformed from a single electrical quantity result into a set of stress indicators that can characterize the degree of risk.

[0068] S3.4 Operating Stress Timing Output Various operational stress indicators are summarized in chronological order to form a time series dataset of operational stress of the distribution network within the prediction time window, which serves as the input for operational stress-fault risk correlation analysis and graded early warning determination in step S4.

[0069] This step enables the mapping from "load forecast results" to "distribution network operation stress state," providing crucial data support for the pre-assessment and proactive defense of distribution network operation risks.

[0070] S3 Technology Performance Description The method for distribution network time-series power flow analysis and operational stress quantification based on predicted load described in this step can achieve the following technical effects: By combining the overcharging load forecast results with the distribution network physical operation model, the disconnect between load forecasting and operation risk assessment can be avoided; the voltage stability and equipment load pressure of the distribution network under overcharging load access conditions can be quantitatively characterized; and a unified and quantifiable input basis can be provided for subsequent operation stress-fault risk correlation analysis.

[0071] Step S4: Operational stress-failure risk correlation and graded early warning method Based on the distribution network operation stress time series data obtained in step S3, this step constructs a correlation analysis method between operation stress and equipment failure risk, transforms the distribution network operation status from a simple numerical calculation result into a failure risk judgment result with engineering orientation, and realizes hierarchical early warning output, providing a basis for operation and maintenance decision-making.

[0072] S4.1 Construction of Operating Stress Feature Set The operating stress time series data output in step S3 are processed to construct an operating stress feature set for fault risk analysis.

[0073] The operational stress feature set includes at least the following elements: The magnitude and duration of the node voltage exceeding the limit; Line current carrying capacity, transformer load rate and their over-limit duration; Three-phase imbalance index (under three-phase modeling conditions); Operating stress change rate and cumulative stress level.

[0074] By performing time-series statistics and feature summarization on the operating stress characteristics, a comprehensive stress description that can reflect the long-term stress state and short-term impact characteristics of the equipment is formed.

[0075] S4.2 Construction of Operational Stress-Failure Risk Association Rules Based on the operating mechanism and engineering experience of power distribution network equipment, a rule base for the association between operating stress and potential fault types is constructed to realize the mapping from "operating status" to "fault risk".

[0076] The association rules include at least the following types: Thermal stress correlation rules When the line current carrying rate or transformer load rate continuously exceeds the preset safety threshold and the operating ambient temperature is high, it is determined that the equipment has the risk of heat-related faults such as insulation aging and joint overheating.

[0077] Voltage stress correlation rules When the node voltage continuously exceeds the limit or the voltage fluctuation is large, it is determined that there is a risk of low voltage tripping, decreased voltage stability, or malfunction of the protection device.

[0078] Unbalanced stress correlation rules When the three-phase imbalance exceeds the preset threshold, the equipment is deemed to have risks such as increased additional losses, local overheating, and shortened service life.

[0079] Environmental Collaborative Risk Rules (Optional) When operating stress is high and severe weather conditions occur simultaneously, it is determined that there is a risk of environmentally coupled faults such as insulation flashover and tree line discharge.

[0080] The association rules can be set based on historical operating data, fault logs, or expert experience, and support subsequent modifications and expansions.

[0081] S4.3 Fault Risk Assessment and Classification Based on the operational stress feature set and the association rule base, the fault risk is determined for the distribution network operation status within the prediction time window.

[0082] Specifically, based on whether the operating stress index meets the triggering conditions of the association rule, the type of potential failure risk is determined, and the failure risk is graded and assessed according to the stress level, duration and superposition effect.

[0083] In one embodiment, the fault risk level can be divided into multiple levels, including but not limited to: Low risk level: Operating stress is close to the threshold but has not been continuously exceeded; Medium risk level: Operating stress exceeds the threshold and persists for a certain period of time; High-risk level: Operating stress significantly exceeds limits or multiple stresses overlap.

[0084] S4.4 Tiered Early Warning Output and Decision Support Based on the fault risk level, corresponding graded early warning information is generated and output to the power distribution network operation management or maintenance system. The early warning information includes at least: the potential fault type; the location of the equipment or area where it may occur; the risk level; and the recommended period of attention.

[0085] By issuing tiered early warnings, maintenance personnel can identify high-risk equipment or areas before a failure occurs, and take targeted inspections, load adjustments, or protective measures in advance, thus shifting from passive response to proactive defense.

[0086] S4 Technology Performance Description The operational stress-fault risk correlation and hierarchical early warning method described in this step can achieve the following technical effects: transforming the operational stress of the distribution network from a numerical indicator into fault risk information with clear engineering orientation; identifying potential equipment fault risks caused by overcharging loads in advance, reducing the probability of sudden faults; and providing interpretable and executable decision support for distribution network operation, maintenance and risk prevention and control.

[0087] Step S5: Distribution Network Collaborative Response and Closed-Loop Operation Method Based on Risk Early Warning Based on the fault risk classification and early warning results output in step S4, this step constructs a risk early warning-driven collaborative response and closed-loop operation method to realize closed-loop management of distribution network operation risks from "identification - early warning - handling - feedback", thereby improving the overall system operation safety and engineering feasibility.

[0088] S5.1 Risk Warning Result Analysis and Response Trigger The graded early warning information generated in step S4 is analyzed, and the corresponding collaborative response strategy is triggered according to the early warning level and the fault risk type.

[0089] The mapping relationship between the warning level and the response method may include: low-risk warning: enter the enhanced operation monitoring state and increase the monitoring frequency of the corresponding area or equipment; medium-risk warning: trigger preventive control or operation adjustment strategies; high-risk warning: trigger emergency response strategies and issue alarm prompts to the scheduling or operation and maintenance system.

[0090] Through the above mechanism, early warning information can be automatically transformed into operational decision-making behavior.

[0091] S5.2 Collaborative Response Strategy Generation Based on different types of failure risks, corresponding collaborative response strategies are generated, and the strategies include at least one or more of the following: Load-side adjustment strategy: When a high-risk state caused by overcharging load is detected, the operating power of the overcharging station can be limited, reduced, or scheduled in a time-sharing manner to reduce the local operating stress level.

[0092] Grid-side operation adjustment strategy: Based on the location and type of risk, perform operation operations such as network reconfiguration, tap adjustment or reactive power compensation control to improve node voltage levels or alleviate equipment overload.

[0093] Operation and maintenance side handling strategy: For identified high-risk equipment or areas, generate inspection, maintenance or key attention suggestions to guide operation and maintenance personnel to carry out targeted handling.

[0094] The collaborative response strategy can be configured and updated based on operational experience, equipment capabilities, or scheduling rules.

[0095] S5.3 Response Execution and Running Status Update After the collaborative response strategy is generated, the corresponding control instructions or handling suggestions will be sent to the relevant execution units, including but not limited to the charging station control system, the power distribution network dispatching system and the operation and maintenance management system.

[0096] After the response is completed, the updated operating data is collected in real time, and the operating status of the distribution network is reassessed to determine whether the operating stress has been effectively alleviated.

[0097] S5.4 Risk Assessment Result Feedback and Model Update The operational status data after the response is executed is fed back to steps S1 and S2 to correct the load forecasting model and risk assessment parameters, thereby achieving adaptive updating of the model.

[0098] Through a continuous feedback mechanism, the prediction model and risk criteria can be continuously optimized as the operating environment and load characteristics change, forming a self-learning risk assessment and early warning closed loop, enabling the system to adaptively update as the load evolves, equipment ages, and the operating environment changes.

[0099] S5 Technology Performance Explanation By adopting the risk-based early warning-based collaborative response and closed-loop operation method described in this step, the following technical effects can be achieved: realizing closed-loop management of the entire process of distribution network operation risk identification, early warning and disposal; improving the initiative and timeliness of distribution network operation control under overcharging load impact conditions; enhancing the system's adaptability to changes in the operating environment and load evolution characteristics, and improving the long-term operation safety level.

[0100] Example 1: Risk Assessment and Early Warning of Distribution Network Operation Based on an Ultra-Fast Charging Station in Chongqing (I) Implementation Scenarios and System Environment In this embodiment, a power distribution network in Chongqing is selected as the application scenario. Chongqing is a typical mountainous city, and its power distribution network is characterized by long feeder lengths, uneven load distribution, and complex three-phase operating conditions, making it suitable for verifying the engineering applicability of the power distribution network operation risk assessment method under ultra-fast charging load access conditions.

[0101] An ultra-fast charging station is connected to the power distribution network, which is connected to a node of the urban power distribution network via a 10kV distribution line. The station is equipped with multiple high-power DC charging devices, and the charging load generated during its operation is characterized by high power level, significant time-series fluctuations, and strong local impact.

[0102] (II) Multi-source data acquisition and feature construction (corresponding step S1) According to step S1, multi-source data related to the ultra-fast charging station and power distribution network are collected and processed.

[0103] Specifically, station-level and pile-level operational data of the ultra-fast charging station are collected, including electrical quantity information such as voltage, current, active power, reactive power and power factor, and charging session-related data are obtained through the charging pile communication protocol; at the same time, the topology of the distribution network, line and transformer parameters, node voltage and basic load data are collected.

[0104] In addition, meteorological data of the area where the ultra-fast charging station is located and information on the distribution of points of interest around the station are collected to characterize environmental factors and regional functional attributes.

[0105] The above multi-source data were time-series aligned, missing values ​​were repaired, and outliers were removed. A multi-dimensional feature dataset reflecting the historical load status, meteorological environment, spatial attributes, and temporal characteristics was constructed as input for the subsequent ultra-fast charging load prediction model.

[0106] (III) Risk Perception and Ultra-Fast Charging Load Prediction (corresponding to step S2) Based on the multidimensional feature dataset constructed in step (II), a risk-aware ultra-fast charging load prediction model is built to predict the load of ultra-fast charging stations in future periods.

[0107] During model construction, the ultrafast charging load time series is decomposed into multiple scales, and the temporal evolution features of the load are extracted using a deep learning network. Simultaneously, a risk awareness mechanism is introduced during model training, assigning higher weights to high-load periods exceeding a preset safety threshold to enhance the model's ability to characterize load spikes.

[0108] Using the above method, the load forecast sequence of the ultra-fast charging station within the forecast time window is output for subsequent distribution network operation analysis.

[0109] (iv) Distribution network time-series power flow analysis and operation stress quantification (corresponding to step S3) The ultra-fast charging load prediction sequence obtained in step (iii) is mapped to the corresponding access node of the distribution network and superimposed with the original basic load of the distribution network to construct a node load model that includes the ultra-fast charging load.

[0110] Within the prediction time window, time-series power flow analysis is performed on the distribution network to obtain the voltage level of each node, the current-carrying state of the lines, and the load status of the transformers. Based on the results of the time-series power flow analysis, indices such as voltage operating stress, equipment thermal operating stress, and three-phase unbalanced operating stress are further calculated to quantify the operating pressure of the distribution network under ultra-fast charging load access conditions.

[0111] (v) Operational stress-failure risk correlation and graded early warning (corresponding step S4) Based on the operating stress index obtained in step (iv), and combined with the operating mechanism of power distribution network equipment and engineering experience, the correlation between operating stress and potential fault risk is constructed.

[0112] When the operating stress index is detected to meet the preset association rule conditions, it is determined that there is a corresponding type of equipment failure risk. Based on the operating stress level, duration and superposition effect, the failure risk is graded and assessed, and corresponding graded early warning information is generated.

[0113] (vi) Coordinated response and closed-loop operation (corresponding to step S5) Based on the graded early warning results output in step (5), corresponding collaborative response strategies are triggered, including adjusting the operating power of ultra-fast charging stations, adjusting the operating parameters of the distribution network, and outputting operation and maintenance suggestions.

[0114] After the response is executed, the updated operational data is collected and fed back to the load forecasting and risk assessment module to correct the model parameters and risk criteria, thereby realizing the closed-loop operation of distribution network operation risk assessment and early warning.

[0115] (vii) Explanation of Implementation Results The method described in this embodiment enables the pre-identification and graded early warning of distribution network operation risks caused by ultra-fast charging load access under the complex urban distribution network conditions in Chongqing. This improves the engineering pertinence and feasibility of distribution network operation risk assessment and provides effective technical support for distribution network safe operation and maintenance decisions.

[0116] Finally, it should be noted that 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 preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for distribution network operation risk assessment and fault early warning based on risk perception and overcharging load prediction, characterized in that, The method specifically includes the following steps: S1. Multi-source data acquisition and feature construction: Collect supercharging station operation data, distribution network operation and topology data, meteorological environment data and spatial attribute data, preprocess them and construct a multi-dimensional feature dataset for load forecasting; S2, Risk-aware overcharging load prediction: The multidimensional feature dataset is input into the pre-constructed risk-aware overcharging load prediction model, and the overcharging station load prediction sequence within the future prediction time window is output; wherein, the risk-aware overcharging load prediction model adopts an asymmetric risk-aware loss function during training, which assigns higher weight to the prediction error of the load period exceeding the preset safety threshold. S3. Distribution network operation stress quantification: The supercharging station load prediction sequence is mapped to the corresponding node of the distribution network, superimposed with the base load, and then time-series power flow calculation is performed. The operation stress index of the distribution network is quantified based on the power flow calculation results. S4. Fault Risk Correlation and Graded Early Warning: Based on the operating stress index, according to the preset correlation rules between operating stress and fault risk, the potential fault risk type and its level are determined, and corresponding graded early warning information is generated.

2. The method for distribution network operation risk assessment and fault early warning based on risk perception and overcharging load prediction according to claim 1, characterized in that, In step S1, the collection of supercharging station operation data, distribution network operation and topology data, meteorological environment data, and spatial attribute data specifically includes: Supercharging station side operation data: Collect electrical quantity information such as voltage, current, active power, reactive power, and power factor at the supercharging station level and pile level; the data sampling frequency is set to the second level or higher, and the data storage granularity is set to the minute level time scale according to application requirements to take into account the needs of load impact characteristic capture and system operation analysis; obtain charging session logs based on the charging pile communication protocol, including vehicle access time, charging start and end time, vehicle departure time, as well as initial state of charge, target state of charge, battery capacity, and maximum allowable charging power charging demand information, and collect charging mode and equipment operation status identifiers; Distribution network operation and topology data: Collect the topology and equipment parameter information of the distribution network, including line type, length, impedance parameters, transformer rated capacity and connection method, to build the distribution network model; at the same time, collect the active and reactive load and voltage information of the gateway and key nodes, and obtain the setting parameters of the relay protection device for subsequent operation risk assessment and boundary determination. Meteorological and environmental data: Collect meteorological data of the area where the supercharging station is located, including ambient temperature, humidity and rainfall, and introduce corresponding environmental status labels when extreme weather events occur for subsequent risk analysis; Spatial and functional attribute data: Based on the spatial location of the supercharging station, data on surrounding points of interest are obtained, and the distribution characteristics of different types of points of interest are statistically analyzed to characterize the functional attributes and potential charging demand patterns of the area where the station is located.

3. The method for distribution network operation risk assessment and fault early warning based on risk perception and overcharging load prediction according to claim 2, characterized in that, In step S1, the data preprocessing and multidimensional feature construction specifically include: 1) Timing alignment and resampling processing Using the unified time base of the power distribution network operation system as a reference, data with different time resolutions are aligned to ensure that all types of data have a one-to-one correspondence at the same time scale. 2) Missing value repair and outlier removal To address missing or abnormal data during the data collection process, data is repaired or removed based on historical statistical characteristics and physical constraints to prevent abnormal data from interfering with load forecasting and risk assessment results. 3) Multidimensional feature construction After data cleaning, a high-dimensional feature set is constructed to comprehensively reflect the historical state of the load, meteorological environment characteristics, spatial functional attributes, and time series information, which is used to describe the spatiotemporal evolution characteristics of the overcharging load. 4) Feature normalization processing To eliminate the influence of features with different dimensions on model training, the constructed features are normalized to improve the stability of model training and prediction accuracy.

4. The method for distribution network operation risk assessment and fault early warning based on risk perception overcharging load prediction according to claim 3, characterized in that, In step S2, based on the constructed multidimensional spatiotemporal feature dataset, a risk-aware overcharging load prediction model integrating signal decomposition and deep learning is constructed. This model introduces an asymmetric risk-oriented mechanism during the training phase to address the problems of traditional prediction methods underestimating peak overcharging loads and easily missing operational risks. Specifically, this includes: S21. Construct an adaptive multi-scale load decomposition module and perform the load decomposition sub-step: Perform adaptive multi-scale decomposition on the original overcharge load sequence to obtain multiple sub-sequence components with different time scale characteristics; including: by introducing a parameter adaptive optimization mechanism, optimize the key parameters of the load decomposition model so that the decomposition result reaches the optimal in terms of energy concentration or feature correlation; based on the optimized parameter combination, decompose the original overcharge load sequence into several sub-sequences with different time scale characteristics, including low-frequency components reflecting the long-term evolution trend of the load and high-frequency components reflecting the characteristics of random shocks; subsequently, align each decomposed component with the multi-dimensional feature vector constructed in step S1 to form a sample set for load prediction at different scales. S22. Construct a spatiotemporal feature extraction and fusion prediction network, and perform feature extraction and prediction sub-steps: Construct a prediction network to extract spatiotemporal features from the subsequence components and the multidimensional feature dataset, and output the prediction results for each component; the prediction network includes at least: a local feature extraction unit, used to extract load fluctuation features and coupling relationships between features from a short time window; a time-series dependency modeling unit, used to characterize the long-term memory characteristics and time correlation of the load sequence; and a key information weighting unit, used to dynamically weight the importance of different time steps and feature dimensions; S23. Design an asymmetric risk-perception loss function: Introduce a risk-perception loss function during the model training phase to enhance the model's ability to fit high-risk load intervals; the risk-perception loss function can be expressed as: in: Indicates time The actual overcharge load value; This represents the predicted load value at the corresponding time. This represents a dynamic weighting coefficient related to operational risk; the weighting coefficient The model is adaptively adjusted based on the relationship between the load level and the preset safety threshold. When the load level exceeds the safety warning threshold, the prediction error is given a higher weight, thereby guiding the model to focus on high-risk load periods during the training process. S24. Load Reconfiguration and Forecast Output: The forecast results of each component are reconfigured to obtain the supercharging station load forecast sequence. The supercharging station load forecast sequence in the future forecast time domain is output for subsequent distribution network time-series power flow calculation and operation risk assessment.

5. The method for distribution network operation risk assessment and fault early warning based on risk perception overcharging load prediction according to claim 4, characterized in that, Step S3 specifically includes: S31. Overcharge Mapping and Node Injection Modeling Based on the actual connection location of the supercharging station in the distribution network, the supercharging station load prediction sequence output in step S2 is mapped to the corresponding distribution network nodes to construct a node load injection model. At the same time scale, the predicted supercharging station load is superimposed with the original basic load of the distribution network to form the total node load including the supercharging load, expressed as follows: in, Indicates the node at time [time]. The base load, This represents the predicted load of the supercharging station; S32, Distribution Network Time-Sequence Power Flow Calculation Based on the distribution network topology and equipment parameters obtained in step S1, a distribution network operation model is constructed, and time-series power flow calculation is performed within the prediction time window. In each time step, the distribution network power flow equation is solved to obtain the voltage amplitude and phase angle of each node, as well as the current, power and load status of each line and transformer. S33, Calculation of Operating Stress Indicators Based on the time-series power flow calculation results, the operating status of the distribution network is quantitatively analyzed, and operating stress indicators that can reflect the operating pressure and potential risks of the distribution network are extracted. These operating stress indicators include at least: voltage operating stress indicators: calculating the degree of voltage deviation of each node relative to the allowable operating range, and statistically analyzing the magnitude and duration of voltage exceedances to characterize voltage stability risk; equipment thermal stress indicators: calculating the line current carrying rate and transformer load rate based on line current and equipment rated capacity, and statistically analyzing the duration exceeding a preset safety threshold to characterize the thermal overload risk of lines and transformers; and three-phase unbalance operating stress indicators: calculating the unbalance index based on the three-phase voltage or current of nodes under three-phase modeling conditions to assess the operating stress level caused by unbalanced load access. Through the calculation of the above operating stress indicators, the operating status of the distribution network is transformed from a single electrical quantity result into a set of stress indicators that can characterize the degree of risk. S34, Operating Stress Timing Output Various operational stress indicators are summarized in chronological order to form a time series dataset of operational stress of the distribution network within the prediction time window, which serves as the input for operational stress-fault risk correlation analysis and graded early warning determination in step S4.

6. The method for distribution network operation risk assessment and fault early warning based on risk perception overcharging load prediction according to claim 5, characterized in that, Step S4 specifically includes: S41. Construction of Operating Stress Feature Set The operating stress time series data output in step S3 is processed to construct an operating stress feature set for fault risk analysis. The operating stress feature set includes at least the following elements: node voltage over-limit amplitude and over-limit duration; line current carrying rate, transformer load rate and their over-limit duration; three-phase unbalance index; operating stress change rate and cumulative stress level. S42. Construction of operational stress-failure risk association rules Based on the operating mechanism and engineering experience of power distribution network equipment, a rule base for the association between operating stress and potential fault types is constructed to realize the mapping from "operating status" to "fault risk". The association rules include at least the following types: Thermal stress correlation rule: When the line current carrying rate or transformer load rate continuously exceeds the preset safety threshold and the operating ambient temperature is high, it is determined that the equipment has the risk of thermal-related faults such as insulation aging and joint overheating. Voltage stress correlation rule: When the node voltage continuously exceeds the limit or the voltage fluctuation is large, it is determined that there is a risk of low voltage tripping, decreased voltage stability or malfunction of protection device; Unbalanced stress correlation rule: When the three-phase unbalance exceeds the preset threshold, it is determined that the equipment is at risk of increased additional losses, local overheating and shortened service life; Environmental Co-operation Risk Rule: When the operating stress is high and severe weather conditions occur at the same time, it is determined that there is a risk of environmentally coupled faults such as insulation flashover and tree line discharge. S43. Fault Risk Assessment and Classification Based on whether the operating stress index meets the triggering conditions of the association rule, the type of potential failure risk is determined, and the failure risk is graded and assessed according to the stress level, duration and superposition effect. S44. Tiered Early Warning Output and Decision Support Based on the fault risk level, corresponding graded early warning information is generated and output to the power distribution network operation management or maintenance system; the early warning information includes at least: the potential fault type; the location of the equipment or area where it may occur; the risk level and the recommended period of attention.

7. The method for distribution network operation risk assessment and fault early warning based on risk perception overcharging load prediction according to claim 6, characterized in that, The method also includes step S5: collaborative response and closed-loop feedback steps, which specifically include: S51. Risk Warning Result Analysis and Response Triggering The graded early warning information generated in step S4 is parsed, and corresponding collaborative response strategies are triggered according to the early warning level and the fault risk type. The mapping relationship between the early warning level and the response method includes: low risk early warning: enter the operation monitoring enhancement state and increase the monitoring frequency of the corresponding area or equipment; medium risk early warning: trigger preventive control or operation adjustment strategies; high risk early warning: trigger emergency response strategies and issue alarm prompts to the scheduling or operation and maintenance system. S52, Cooperative Response Strategy Generation Based on different types of failure risks, corresponding collaborative response strategies are generated, and the strategies include at least one or more of the following: Load-side adjustment strategy: When a high-risk state caused by overcharging load is detected, the operating power of the overcharging station is limited, reduced, or scheduled in a time-sharing manner to reduce the local operating stress level; Grid-side operation adjustment strategy: Based on the location and type of risk, implement network reconfiguration, tap adjustment or reactive power compensation control operation to improve node voltage levels or alleviate equipment overload; Operation and maintenance side handling strategy: For identified high-risk equipment or areas, generate inspection, maintenance or key attention suggestions to guide operation and maintenance personnel to carry out targeted handling; S53, Response Execution and Running Status Update After the coordinated response strategy is generated, the corresponding control instructions or handling suggestions are sent to the relevant execution units, including but not limited to the charging station control system, the distribution network dispatching system and the operation and maintenance management system; after the response is executed, the updated operation data is collected in real time, and the operation status of the distribution network is reassessed to determine whether the operation stress has been effectively alleviated. S54. Risk Assessment Results Feedback and Model Update The operational status data after the response is executed is fed back to steps S1 and S2 to correct the load forecasting model and risk assessment parameters, thereby achieving adaptive updating of the model.

8. A distribution network operation risk assessment and fault early warning system based on risk perception and overcharging load prediction, characterized in that, The system employs the method described in any one of claims 1 to 7; The system includes: The multi-source data acquisition module is used to collect supercharging station operation data, distribution network operation and topology data, meteorological environment data, and spatial attribute data; The data preprocessing and feature construction module is used to preprocess the collected multi-source data and construct a multidimensional feature dataset; The risk perception overcharging load prediction module has a built-in risk perception overcharging load prediction model, which is used to output the overcharging station load prediction sequence based on the multidimensional feature dataset. The distribution network operation analysis module is used to perform time-series power flow calculations based on the supercharging station load prediction sequence and distribution network topology parameters; The stress quantification module is used to calculate the operating stress index of the distribution network based on the time-series power flow calculation results. The fault risk assessment and graded early warning module is used to output the fault risk type and early warning level based on the operating stress index and preset association rules. The collaborative response and closed-loop feedback module is used to trigger operational adjustments based on the early warning results and to update the parameters of the risk perception overload prediction module and / or the fault risk assessment and graded early warning module.

9. A distribution network operation risk assessment and fault early warning system based on risk perception and overcharging load prediction according to claim 8, characterized in that, The risk perception overcharge prediction module includes: The load decomposition unit is used to perform adaptive multi-scale decomposition of the original overcharge load sequence. The fusion prediction network unit is used to extract spatiotemporal features and predict each component after decomposition. The load reconfiguration unit is used to reconstruct the prediction results of each component into a complete supercharging station load prediction sequence.

Images