A method for intelligent identification and early warning of distribution network operation risks

Abstract

This application provides a method for intelligent identification and early warning of distribution network operation risks, including: identifying the probability of equipment contact failure through actual wear and tear; when the probability of equipment contact failure exceeds the requirements for safe operation, determining equipment fault early warning signals through historical fault statistics, identifying potential equipment failure risk points and extracting risk distribution characteristics; identifying high-risk equipment nodes through equipment fault early warning signals, assessing whether the redistribution of power flow after identifying high-risk nodes will trigger cascading faults of adjacent equipment overload, extracting fault propagation paths, and determining a system risk level distribution map; dividing the system risk level distribution map into risk areas, identifying key equipment nodes in high-risk areas, extracting load transfer schemes for high-risk areas, and determining load distribution paths and power transmission directions.

Description

Technical Field

[0001] This invention relates to the field of information technology, and in particular to a method for intelligent identification and early warning of operational risks in power distribution networks. Background Technology

[0002] With the rapid development of smart grids, distribution automation systems need to respond in real time to complex and ever-changing power demands and quickly adjust the operation of the distribution network in case of faults or anomalies to ensure the continuity of power supply. However, the existing operating methods of distribution automation systems are difficult to adapt to the dynamic changes caused by frequent switching operations in the distribution network when dealing with complex scenarios. In particular, when the system needs to perform switching operations in a specific sequence, each operation will cause significant changes in the power flow distribution of the distribution network. The power flow distribution of the distribution network refers to the voltage amplitude, phase, and active and reactive power distribution of each node in the grid. Significant changes in the power flow distribution can cause problems such as line overload, voltage exceeding limits, or a surge in power loss, which may lead to system instability and large-scale power outages in severe cases. Existing solutions usually cannot obtain and comprehensively analyze the impact of these changes on the overall reliability of the system in real time. In addition, the physical degradation of switching equipment has not been fully considered, and the long-term operating status of the equipment is difficult to accurately assess, making it difficult to identify potential risks in a timely manner. When switching equipment executes remote control commands, its contacts experience mechanical wear due to repeated operations. This wear accumulates with each operation, significantly increasing the probability of switch failure. Failure to operate refers to the switch's failure to correctly execute an action after receiving a command. This can not only cause power distribution network outages but also trigger cascading failures. For example, during peak electricity consumption periods, if a critical section switch fails to operate due to wear, it may lead to abnormal power flow distribution in the local power grid, resulting in regional power outages. Each switch operation alters the power flow path of the power grid, and these path changes, in turn, affect the reliability of subsequent switch operations and the overall stability of the system. For instance, in an urban power distribution network, when a line needs to be isolated due to a fault, multiple section switches are operated remotely to reconfigure the power flow path. However, frequent operations may cause severe wear on some switch contacts. Without real-time monitoring of the degree of degradation, the system struggles to determine whether to continue relying on that switch for subsequent tasks. Furthermore, the dynamic changes in power flow distribution may prevent the system from accurately predicting potential power outage risks due to a lack of comprehensive operational data. Therefore, how to collect and comprehensively analyze equipment status and power grid operation data in real time after switching action, so as to dynamically adjust the risk warning mechanism, has become a key issue for improving the safety and reliability of power distribution automation systems. Summary of the Invention

[0003] This invention provides a method for intelligent identification and early warning of operational risks in power distribution networks, mainly including:

[0004] Vibration and current waveform data are collected after switch opening and closing operations. Contact wear depth and spring fatigue are extracted to identify the contact reliability of the switch mechanism and determine the preliminary equipment health status. The operating conditions of the power grid nodes are obtained, and the preliminary equipment health status is matched with the operating conditions to determine the increase in contact resistance. The impact of load fluctuations on the degradation of equipment contact performance is identified to determine the actual wear status of the switchgear. Based on the actual wear status of the switchgear, the probability of equipment contact failure is calculated, and an equipment fault warning signal is generated. Based on the equipment fault warning signal, the possibility of cascading failures due to overload of adjacent equipment after power flow redistribution is assessed, and a system risk level distribution map is determined. The system risk level distribution map is divided into risk areas to determine the load distribution path and power transmission direction. Based on the load distribution path and power transmission direction, the matching degree between equipment wear and the operating environment is analyzed to determine the correspondence between equipment health status and fault probability. Based on the operating conditions after load transfer, the risk assessment results for each equipment node are determined. When the risk assessment results for each equipment node show that the system risk exceeds a threshold, adjustment instructions including load transfer and equipment maintenance timing are generated.

[0005] The acquisition of vibration and current waveform data after switch opening and closing operations, extraction of contact wear depth and spring fatigue degree, identification of switch mechanism contact reliability, and determination of preliminary equipment health status include:

[0006] Vibration waveform data after switch opening and closing operations are acquired, noise is filtered out, vibration characteristic signals are extracted, and the deviation ratio between the vibration amplitude and the standard vibration amplitude is calculated. Based on the relationship between the deviation ratio and the contact wear depth, the contact wear depth data is determined. Current waveform data during the opening and closing process is acquired, harmonic components are analyzed, and the harmonic distortion rate is calculated. Based on the difference between the harmonic distortion rate and historical operation records, the spring fatigue degree value is determined. Based on the contact wear depth data and the spring fatigue degree value, the contact resistance increment is calculated, the total contact resistance growth rate is determined, and the contact reliability level is determined by comparing the total contact resistance growth rate with a preset threshold. Based on the contact reliability level, the cumulative number of operations, and the standard deviation of the operation interval time, the equipment comprehensive deterioration index is calculated by weighted summation to determine the preliminary equipment health status.

[0007] Furthermore, the step of calculating the contact resistance increment based on the contact wear depth data and the spring fatigue value, determining the total contact resistance growth rate, and determining the contact reliability level by comparing the total contact resistance growth rate with a preset threshold includes:

[0008] Based on the contact wear depth data, calculate the resistance increment caused by wear; based on the spring fatigue degree value, calculate the resistance increment caused by fatigue; add the resistance increment caused by wear and the resistance increment caused by fatigue to obtain the total contact resistance growth rate; compare the total contact resistance growth rate with a preset threshold to determine the contact reliability level.

[0009] Furthermore, the process of acquiring the operating conditions of the power grid nodes, matching the preliminary equipment health status with the operating conditions, determining the increase in contact resistance, identifying the impact of load fluctuations on the degradation of equipment contact performance, and determining the actual wear status of the switchgear includes:

[0010] The operating conditions of the power grid nodes are obtained, the voltage deviation rate and current density values ​​of the nodes where the switching equipment is located are read, the number of current peak changes and voltage drop depths are counted within the load fluctuation cycle, the load fluctuation intensity is determined by the product of the number of current peak changes and the voltage drop depth, and the distribution characteristics of the load change frequency and the statistical mean of the load change amplitude are analyzed by the proportional relationship between the load fluctuation intensity and the contact performance degradation, so as to determine the actual wear condition of the switching equipment.

[0011] Furthermore, after determining the actual wear condition of the switching equipment, the process includes:

[0012] Based on the actual wear status, the probability of equipment contact failure is calculated; if the probability of equipment contact failure exceeds a threshold, the fault time series and type are extracted from historical fault statistics to generate the equipment fault warning signal; based on the equipment fault warning signal, nodes with similar operating environments and wear characteristics are identified, and they are grouped by clustering algorithm to obtain a set of potential equipment failure risk points; based on the set of potential equipment failure risk points, geographical coordinates and load levels are extracted to construct a risk distribution feature vector.

[0013] Furthermore, the step of assessing the likelihood of cascading failures due to overload of adjacent equipment after power flow redistribution based on the equipment fault warning signal, and determining the system risk level distribution map, includes:

[0014] By using the failure probability value and fault type identifier in the equipment fault warning signal, a risk score is calculated, high-risk equipment nodes with risk scores exceeding the risk score threshold are marked, and the location information of the high-risk equipment nodes in the power grid topology is obtained. The power grid topology changes after the failure of the high-risk equipment nodes are simulated, the power distribution of each line is calculated, the power is compared with the rated capacity, the overload risk of adjacent equipment is determined, and potential cascading fault nodes are identified. Based on the potential cascading fault nodes, a node adjacency matrix is ​​constructed, and the fault propagation path is traversed through a search algorithm to extract the path sequence. Based on the path sequence, the path risk value is calculated, and combined with the overload severity, the system risk level distribution map is determined.

[0015] Furthermore, the step of dividing the system risk level distribution map into risk areas and determining the load distribution path and power transmission direction includes:

[0016] The system risk level distribution map is divided into risk areas using a clustering algorithm, and areas where the average risk level exceeds the average risk level threshold are marked as high-risk areas. The nodes with the largest load and connecting lines in the high-risk areas are identified as key equipment nodes. Based on the transferable capacity of the key equipment nodes, a load transfer scheme is extracted. Based on the load transfer scheme, the electrical distance of the connecting lines is calculated, and the load distribution path and the power transmission direction are determined.

[0017] Furthermore, the step of extracting a load transfer scheme based on the transferable capacity of the key equipment nodes includes:

[0018] Obtain the available power supply capacity and backup power location information of the key equipment nodes; determine the load transfer scheme based on the matching relationship between the available power supply capacity and the load to be transferred.

[0019] Furthermore, the process involves analyzing the degree of equipment wear and its compatibility with the operating environment based on the load distribution path and the power transmission direction, determining the correlation between equipment health status and failure probability, and determining the risk assessment results for each equipment node based on the operating conditions after load transfer, including:

[0020] Based on the load distribution path and power transmission direction, the degree of equipment wear and the matching degree with the operating environment are calculated. Based on the matching degree, fault cases under similar operating conditions are queried from historical fault records, and the health status feature value sequence before the fault occurs is extracted to determine the correspondence between equipment health status and fault probability. Based on the correspondence, the new operating conditions of each node after load transfer are simulated. If the operating conditions exceed the safe operating range of the equipment, the fault probability value is increased by the product of the excess range and a preset coefficient to obtain the corrected fault probability. The comprehensive risk index of each equipment node is calculated using the corrected fault probability, and the risk assessment result of each equipment node is determined based on the preset range in which the risk index is located.

[0021] Furthermore, when the risk assessment result shows that the system risk exceeds a threshold, the generation of adjustment instructions including load transfer and equipment maintenance timing includes:

[0022] When the risk assessment result exceeds the threshold, the load value and transferable capacity of the load allocation path are read, the power flow transfer node is determined according to the power transmission direction, and an operation constraint matrix is ​​constructed in combination with the actual wear status; a load transfer timing table is calculated according to the operation constraint matrix; the equipment maintenance priority is determined according to the load transfer timing table and the actual wear status, and a maintenance timing table is generated; and adjustment instructions are compiled according to the load transfer timing table and the maintenance timing table.

[0023] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects:

[0024] This invention discloses an intelligent identification and early warning method for distribution network operation risks. This method addresses the unique business scenario of cascading failure risks caused by contact wear and failure of power grid switching equipment under frequent load fluctuations. It achieves accurate identification and proactive intervention through multi-level data analysis and state matching. First, real-time waveform data is collected using vibration and current sensors. Combined with historical data and filtered, contact wear depth and spring fatigue are extracted to preliminarily assess the equipment's health status. Second, voltage and current operating conditions from power flow distribution monitoring are integrated to perform state matching calculations on the impact of contact resistance growth and load fluctuations on contact performance, determining the actual wear state. Then, based on the actual wear, the probability of contact failure is identified, and a warning signal is generated using historical fault statistics to identify potential failure risk points. Subsequently, for high-risk nodes, the cascading overloads caused by power flow redistribution are assessed, fault propagation paths are extracted, and a system risk level distribution map is generated. Based on this, risk areas are divided, key nodes are identified, and load transfer schemes are extracted, determining the allocation path and power transmission direction. Finally, by matching wear degree with the operating environment, the correlation between equipment health and fault probability is analyzed to assess the risk outcome after load transfer. When a threshold is exceeded, the path, direction, and wear state are integrated to generate a timing-based adjustment command, achieving coordinated optimization of load transfer and equipment maintenance. This invention, through the above closed-loop mechanism, significantly improves power grid stability, reduces fault propagation probability, and ensures safe and reliable system operation. Attached Figure Description

[0025] Figure 1 This is a flowchart of a method for intelligent identification and early warning of distribution network operation risks according to the present invention.

[0026] Figure 2 This is a schematic diagram of a method for intelligent identification and early warning of distribution network operation risks according to the present invention. Detailed Implementation

[0027] To further understand the content of this invention, a detailed description of the invention is provided in conjunction with the accompanying drawings and embodiments. The specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention. It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0028] like Figures 1-2 This embodiment of a method for intelligent identification and early warning of distribution network operation risks may specifically include:

[0029] Step S101: Real-time vibration waveform data and current waveform data after the switch opening and closing operation are collected by vibration sensor and current sensor. Combined with historical opening and closing operation records, the contact wear depth and spring fatigue degree are extracted through filtering to identify the contact reliability of the switch mechanism and assess the preliminary equipment health status.

[0030] The raw vibration sensor data after the switch opening and closing operations are acquired. A Butterworth filter is used to filter the vibration waveform, removing low-frequency noise and high-frequency interference. Vibration characteristic signals within a preset frequency band are extracted, and the deviation ratio between the vibration amplitude and the standard vibration amplitude is calculated. Based on the linear relationship between the deviation ratio and the contact wear depth, the contact wear depth data is obtained. Current waveform data during the opening and closing process is collected from the current sensor. Fourier transform analysis is used to analyze the harmonic components of the current waveform, extracting the amplitudes of the third and fifth harmonics, and calculating the harmonic distortion rate. Based on the difference between the harmonic distortion rate and the initial harmonic distortion rate recorded in the historical operation record, the percentage decrease in spring clamping force is determined, obtaining the spring fatigue level value. Based on the contact wear depth data and the spring fatigue level value, the resistance increment caused by wear and the resistance increment caused by fatigue are calculated using the linear relationship between contact wear depth and contact resistance growth, and the correspondence between spring fatigue level and contact pressure decrease. The two are added together to obtain the total contact resistance growth rate. If the total contact resistance growth rate exceeds a preset threshold, the contact reliability is determined to be reduced, and the contact reliability level is obtained. The contact reliability level is used as the primary weighting factor, the cumulative number of switching operations is used as the secondary weighting factor, and the standard deviation of the operation interval time is used as the correction factor. The comprehensive equipment degradation index is calculated by weighted summation, and the preliminary equipment health status is determined according to the preset range of the degradation index.

[0031] Specifically, in one embodiment, a vibration sensor is installed at the moving contact support of the switching mechanism. When the switch performs opening and closing operations, the sensor continuously collects vibration signals at a sampling frequency of 10kHz. A Butterworth filter, as a filter with the flattest frequency response within its passband, effectively separates vibration components of different frequency bands due to the attenuation characteristics of its transfer function at the cutoff frequency. By combining a fourth-order Butterworth high-pass filter and a low-pass filter, a band-pass filtering characteristic is formed, filtering out low-frequency noise components generated by power grid interference and high-frequency noise introduced by electromagnetic interference. The vibration characteristic signal is mainly concentrated in the mid-frequency band generated by contact collision, which contains mechanical vibration information during the contact process.

[0032] For example, in the monitoring of circuit breakers in a 500kV substation, the standard vibration amplitude is obtained through multiple tests on brand-new switchgear and stored as a benchmark reference value in the system database. When the actual vibration amplitude deviates from the standard value, the deviation ratio directly reflects the degree of wear on the contact surface. There is a linear relationship between the contact wear depth and the deviation ratio. This relationship was obtained by fitting a large amount of experimental data. The physical mechanism is that contact wear leads to a reduction in the contact area, which alters the vibration transmission characteristics during collision.

[0033] It should be noted that harmonic analysis of the current waveform can reveal the deterioration state of the internal spring mechanism of the switch. The Fourier transform decomposes the current signal in the time domain into sinusoidal components of different frequencies. The fundamental component represents the normal power frequency current, while the harmonic components reflect the degree of distortion in the current waveform. The third and fifth harmonics, as the main odd harmonic components, show amplitude variations closely related to the decrease in spring clamping force. When spring fatigue leads to insufficient clamping force, the contacts will experience slight bouncing and unstable contact during closure. This phenomenon manifests as an increase in harmonic content in the current waveform. The harmonic distortion rate is defined as the ratio of the sum of the effective values ​​of all harmonic components to the effective value of the fundamental component. By comparing the current harmonic distortion rate with the historical records during the initial commissioning of the equipment, the degree of spring fatigue can be quantified.

[0034] In one possible implementation, the mechanism of contact resistance growth involves two main aspects. The increase in resistance due to contact wear primarily stems from the reduction in contact area and the increase in surface roughness; for every 0.1 mm increase in wear depth, the contact resistance increases by approximately 5% to 8% of its original value. The increase in resistance due to spring fatigue arises from the decrease in contact pressure. According to Hertzian contact theory, contact resistance is inversely proportional to contact pressure; for every 10% decrease in spring clamping force, the contact resistance increases by approximately 15% to 20%. The resistance increases from these two factors are linearly superimposed to obtain the total contact resistance growth rate; this simplified approach offers sufficient accuracy in engineering practice.

[0035] Preferably, the contact reliability level is divided into five levels, each corresponding to a different resistance growth rate range. When the total contact resistance growth rate is less than 20%, it is determined to be a Level 1 reliable state; a growth rate between 20% and 40% is a Level 2 state; 40% to 60% is a Level 3 state; 60% to 80% is a Level 4 state; and exceeding 80% is a Level 5 severely degraded state. This classification method facilitates maintenance personnel in quickly assessing the equipment status and formulating corresponding maintenance strategies.

[0036] Specifically, the comprehensive equipment degradation index is calculated using a hierarchical weighted method. Contact reliability level is the primary weighting factor, with a weighting coefficient set at 0.6, reflecting the decisive impact of contact performance on the overall health of the equipment. The cumulative number of switching operations has a weighting coefficient of 0.3, reflecting the cumulative effect of mechanical lifespan. The standard deviation of the operation interval is used as a correction factor, with a weighting coefficient of 0.1, to reflect the uniformity of equipment usage frequency. When operations are too frequent or the equipment is idle for extended periods, the standard deviation increases, negatively correcting the degradation index.

[0037] For example, in the monitoring of 10kV switchgear in urban power distribution networks, after three years of operation, the contact wear depth of a certain sectional switch was calculated to be 0.35 mm, the spring fatigue level reached 72% of the initial clamping force, and the total contact resistance growth rate was 48%, corresponding to a level three contact reliability. Considering that the switch had accumulated 2156 operations, with an average operation interval of 12.8 hours and a standard deviation of 3.2 hours, the final calculated comprehensive equipment degradation index was 0.62.

[0038] In one embodiment, the preliminary equipment health status is divided into four intervals based on the degradation index: a degradation index of less than 0.3 indicates a healthy state, with the equipment operating normally and requiring no special maintenance; 0.3 to 0.5 indicates a sub-healthy state, and it is recommended to increase the monitoring frequency; 0.5 to 0.7 indicates a degraded state, requiring planned maintenance; and a degradation index greater than 0.7 indicates a severely degraded state, requiring immediate repair or replacement.

[0039] Step S102: Obtain the operating conditions of the current power grid node, match the preliminary equipment health status with the operating conditions of the current power grid node to obtain the contact resistance growth rate, identify the impact of power grid load fluctuations on the equipment contact performance degradation, and determine the actual wear status of the switchgear based on the frequency and magnitude of load changes.

[0040] Real-time voltage amplitude and current phasor data of each node in the power grid are collected by the power flow distribution monitoring module. The voltage deviation rate and current density values ​​of each node at different times are extracted. The electrical connection strength is determined based on the line impedance values ​​between the node and the switching equipment. A node operating condition matrix is ​​constructed to obtain the power grid operating condition distribution. Based on this power grid operating condition distribution, the voltage deviation rate and current density values ​​of the nodes where the switching equipment is located are read. Combined with the preliminary equipment health status, the quadratic function relationship between current density and contact resistance temperature rise can be expressed as:

[0041] ΔT represents the temperature rise, J represents the current density, and R represents the current density. contact Let a, b, and c represent the contact resistance, and a, b, and c be fitting coefficients. The temperature rise under current operating conditions is calculated. If the temperature rise exceeds a preset threshold, the contact resistance increase is determined based on the linear relationship between the temperature rise and the contact resistance. Using this contact resistance increase, the number of peak current changes and voltage dip depth within the load fluctuation cycle are statistically analyzed. The load fluctuation intensity is determined by the product of the number of changes and the voltage dip depth. The decrease in equipment contact performance is obtained by the proportional relationship between the load fluctuation intensity and the decrease in contact performance. Based on this decrease in equipment contact performance, the distribution characteristics of the load change frequency and the statistical mean of the load change amplitude are analyzed. The fatigue accumulation rate is determined by the frequency distribution characteristics. Combined with the mean load change amplitude and historical operating time, the actual wear state of the switchgear is determined.

[0042] Specifically, in one implementation, the power flow distribution monitoring module is deployed at the distribution automation master station. It acquires voltage amplitude and current phasor information by collecting telemetry data from each feeder node in real time. Electrical connection strength reflects the degree of electrical coupling between nodes; its value is determined by the reciprocal of the line impedance. A smaller impedance value indicates a tighter electrical connection, and the greater the electrical stress on the switching equipment. The node operating condition matrix is ​​a two-dimensional array structure. Rows represent different monitoring nodes, and columns contain data in three dimensions: voltage deviation rate, current density, and electrical connection strength. Matrix operations can quickly locate abnormal nodes and assess their impact on adjacent equipment.

[0043] It should be noted that the voltage deviation rate is defined as the difference between the actual voltage and the rated voltage divided by the rated voltage, reflecting the degree of voltage fluctuation at nodes. The current density value is the ratio of the actual current to the conductor cross-sectional area, characterizing the current-carrying capacity of the conductor. Under normal operation of the distribution network, these two parameters fluctuate within the allowable range, but when sudden load changes or faults occur, the parameters will change significantly, and this change directly affects the operating status of the switching equipment.

[0044] Specifically, the quadratic function relationship of contact resistance temperature rise originates from the physical mechanism of the Joule heating effect. The heat generated when current passes through the contact resistance is proportional to the square of the current. The temperature rise calculation formula includes a quadratic term of the current density, a baseline value of the contact resistance, and a heat dissipation coefficient. In actual monitoring at a 220kV substation, when the current density reaches 1.2 times the rated value, the contact point temperature rise can reach 30 degrees Celsius. At this point, the contact resistance increases by approximately 12% due to thermal expansion. The linear relationship between temperature rise and contact resistance was obtained through fitting a large amount of experimental data; for every 10 degrees Celsius increase, the contact resistance increases by approximately 4%, and this relationship remains stable within the engineering-allowed temperature range. The combination of preliminary equipment health status and current operating conditions enables a shift from static to dynamic assessment, making the calculation of the contact resistance increase more closely reflect actual operating conditions.

[0045] For example, the process of determining the intensity of load fluctuations involves statistical analysis across multiple time scales. Within a 15-minute load fluctuation cycle, the magnitude and duration of each current peak exceeding the average value are recorded, while the depth and recovery speed of voltage drops are monitored. The product of the number of changes and the depth of voltage drops forms a load fluctuation intensity index, which comprehensively reflects the severity of load changes.

[0046] In one possible implementation, the degradation of device contact performance is determined by a proportional relationship. For every unit increase in load fluctuation intensity, the degradation of contact performance increases by 0.8% to 1.2%, a proportionality that takes into account the combined effects of mechanical and electrical stresses. Frequent load fluctuations accelerate oxidation and wear on the contact surface, leading to a further increase in contact resistance.

[0047] Preferably, the distribution characteristics of load change frequency are obtained using a statistical histogram method. The 24 hours are divided into 96 time periods, each 15 minutes long, and the number and magnitude of load changes within each time period are counted. The fatigue accumulation rate is determined based on the standard deviation of the frequency distribution; a larger standard deviation indicates more uneven load changes and more concentrated fatigue stress on the equipment.

[0048] For example, in the power distribution network of a city's central business district, load changes frequently during daytime business hours, while remaining relatively stable at night. Analysis revealed that the frequency of load changes exhibits a bimodal distribution between 10:00 AM and 2:00 PM, during which the fatigue accumulation rate is 3.5 times higher than at night. The statistical mean of load change amplitude, obtained by calculating the arithmetic mean of all recorded data, reflects the average stress level experienced by the equipment. Furthermore, the determination of the actual wear condition comprehensively considers three factors: fatigue accumulation rate, average load change amplitude, and historical operating time. The product of the fatigue accumulation rate and operating time yields the cumulative fatigue degree, while the average load change amplitude determines the intensity of a single stress event. Together, these factors form the assessment value of the actual wear condition of the equipment.

[0049] Step S103: Identify the probability of equipment contact failure by actual wear condition. When the probability of equipment contact failure exceeds the safe operation requirements, determine the equipment fault warning signal by using historical fault statistics, identify potential equipment failure risk points and extract risk distribution characteristics.

[0050] Based on the actual wear condition, the contact failure probability of the equipment in different operating stages is calculated using the Weibull distribution function. The failure probability is then fitted with shape and scale parameters. If the failure probability exceeds a preset safety threshold, the time series and type of failures under similar wear conditions are extracted from historical fault statistics to determine the equipment fault warning signal. Using the fault feature parameters in the equipment fault warning signal, switchgear nodes with similar operating environments and wear characteristics in the distribution network are identified. The nodes are grouped according to wear degree, operating years, and load level using the K-means clustering algorithm to obtain a set of potential equipment failure risk points. Based on the clustering label of each risk point in the set of equipment failure risk points, the geographical coordinates, feeder number, and power supply load level of each risk point are extracted from the distribution automation master station database to construct a risk distribution feature vector containing spatial location distribution density and temporal fault probability distribution, thus obtaining the risk distribution characteristics.

[0051] Specifically, in one implementation, the application of the Weibull distribution function in power distribution equipment fault prediction is based on its ability to accurately describe the wear and tear failure patterns of equipment. The shape parameter of the Weibull distribution reflects the trend of the failure rate over time; a shape parameter less than 1 indicates early failure, equal to 1 indicates random failure, and greater than 1 indicates wear failure. For power distribution switchgear, due to long-term arc erosion and mechanical wear, its shape parameter is typically between 1.5 and 3. The scale parameter characterizes the characteristic lifespan of the equipment, i.e., the operating time when approximately 63.2% of the equipment fails.

[0052] Specifically, the calculation of failure probability first requires collecting historical failure data for similar switchgear, including the failure time and the number of operating hours prior to the failure. The parameters of the Weibull distribution are then fitted using the maximum likelihood estimation method to obtain a distribution function applicable to the specific equipment type. The equivalent operating time corresponding to the actual wear state is then substituted into the distribution function to calculate the contact failure probability. When the failure probability exceeds a preset safety threshold, failure cases with similar wear characteristics are retrieved from the historical failure database, and the changing patterns of symptom parameters before the failure are analyzed.

[0053] It should be noted that the application of the K-means clustering algorithm in identifying potentially risky equipment groups involves the partitioning of a multi-dimensional feature space. The algorithm first represents each switchgear as a three-dimensional feature vector, where wear level is comprehensively assessed through a combination of contact resistance growth rate and mechanical action frequency, operating years are directly obtained from the equipment ledger, and load level is calculated based on the average load rate over the past year. The number of clusters K is selected using the elbow rule; by calculating the sum of squares within each group under different K values, the K value that significantly slows down the rate of decrease in the sum of squares within each group is chosen.

[0054] Preferably, the construction of the risk distribution feature vector comprehensively considers both spatial and temporal dimensions. Spatial location distribution density is derived by calculating the number of risk points per unit area, reflecting the geographical concentration of faults. Densely distributed risk points may indicate the presence of common environmental factors in the area, such as high temperature, high humidity, or severe pollution. Temporal fault probability distribution, on the other hand, predicts the probability of faults in future periods by statistically analyzing historical fault frequencies over different time periods and combining this with the current wear and tear of the equipment.

[0055] For example, in a practical application of a city's power distribution network, three sets of switchgear located in an industrial area were identified as exhibiting similar high wear characteristics. These devices shared common features: frequent load fluctuations and high operating ambient temperatures. By constructing risk distribution characteristics, it was found that the probability of failure for these devices significantly increased during the summer peak electricity consumption period. Based on this, a targeted maintenance plan was developed to complete preventative maintenance before the peak failure period.

[0056] Step S104: Identify high-risk equipment nodes through equipment fault warning signals, assess whether the redistribution of power flow will trigger a chain of faults causing overload of adjacent equipment based on the identified high-risk nodes, extract the fault propagation path, and determine the system risk level distribution map.

[0057] By analyzing the failure probability value and fault type identifier in the equipment fault warning signal, a risk score is calculated by multiplying the failure probability value by the fault type weighting coefficient. If the score exceeds a preset high-risk threshold, the equipment node is marked as high-risk, and a set of high-risk nodes and their location information in the power grid topology are obtained. Based on the set of high-risk nodes, the power grid topology changes after the failure of each high-risk node are simulated. The power distribution of each line is recalculated using a DC power flow calculation method. The redistributed line power is compared with the line's rated capacity. If the power exceeds the safety threshold of the rated capacity, it is determined that adjacent equipment has an overload risk, and potential cascading fault nodes are identified. Using the potential cascading fault nodes, a node adjacency matrix is ​​constructed based on the reciprocal of the line impedance between nodes as the connection weight. A breadth-first search algorithm is used to traverse from the high-risk nodes, recording all reachable paths for the fault to propagate from the source node to adjacent nodes, and extracting the fault propagation path sequence. Based on the fault propagation path sequence, the number of nodes and path length on each path are counted. The path risk value is calculated by multiplying the path length by the node failure probability. Combined with the severity of overload, corresponding risk levels are assigned to each area of ​​the power grid, and a system risk level distribution map is constructed.

[0058] Specifically, in one implementation, the equipment fault warning signal includes two key parameters: a failure probability value and a fault type identifier. The failure probability value is a time-related probability calculated based on the Weibull distribution function, reflecting the likelihood of equipment failure within a specific time period. The fault type identifier is categorized into three main types based on historical fault patterns: poor contact, mechanical jamming, and insulation degradation. Each type corresponds to a different weighting coefficient: poor contact has a weight of 0.8, mechanical jamming has a weight of 1.0, and insulation degradation has a weight of 0.6. The risk score is obtained by multiplying the failure probability value by the corresponding fault type weighting coefficient. When the score exceeds a high-risk threshold of 0.7, the node is marked as a high-risk equipment node.

[0059] It should be noted that the DC power flow calculation method is a simplified algorithm in power system analysis. Its basic principle is to ignore line resistance and reactive power, and only consider the distribution of active power in the network. In the scenario of distribution network fault analysis, when a high-risk node fails in a simulation, the load carried by that node needs to be redistributed through other paths. DC power flow calculation obtains the new power distribution state by solving the node power balance equations. In the specific calculation process, the node admittance matrix is ​​first constructed, and then a linear equation system is established according to the power balance condition. The phase angle of each node is obtained by solving the Gaussian elimination method, and then the power flow of each branch is calculated. When the power of a line exceeds 1.3 times its rated capacity safety threshold, the line is judged to be in an overload state, and the equipment nodes connected to it become potential cascading failure nodes. This judgment method considers the short-term overload capacity of the distribution equipment, which avoids overly conservative judgment and ensures the safety margin of the system.

[0060] Specifically, the node adjacency matrix is ​​constructed based on the physical topology of the power grid. The rows and columns of the matrix represent nodes in the power grid, and the matrix element values ​​represent the connection strength between nodes. If there is a direct electrical connection between two nodes, the corresponding matrix element value is the reciprocal of the line impedance; the smaller the impedance, the stronger the electrical coupling and the greater the possibility of fault propagation. If there is no direct connection between two nodes, the matrix element value is zero.

[0061] For example, in a practical application of a 35kV distribution network, when a substation outgoing switch is identified as a high-risk node, a scenario of switch failure is simulated. The loads of the three 10kV feeders originally powered by this switch need to be transferred to an adjacent substation via a tie switch. A breadth-first search algorithm starts from the failed node and traverses all reachable nodes hierarchically. The first layer contains nodes directly connected to the failed node, the second layer contains nodes connected to nodes in the first layer but not the failed node, and so on.

[0062] Preferably, the fault propagation path sequence is recorded using a path vector method, with each path represented by a sequence of node numbers. The calculation of the path risk value comprehensively considers two factors: path length and node failure probability. A longer path indicates that the fault needs to go through more intermediate links to propagate, and the risk is relatively lower; however, if there are nodes with a high failure probability on the path, the overall risk value of the path will increase significantly.

[0063] In one possible implementation, the system risk level distribution map adopts a five-level risk classification system. Level 1 is the normal operating area, with a risk value less than 0.2; Level 2 is the low-risk area, with a risk value between 0.2 and 0.4; Level 3 is the medium-risk area, with a risk value between 0.4 and 0.6; Level 4 is the high-risk area, with a risk value between 0.6 and 0.8; and Level 5 is the extremely high-risk area, with a risk value greater than 0.8. The risk level of each area is visually displayed in the distribution network geographic information system using color coding, allowing maintenance personnel to quickly identify areas requiring key attention. Furthermore, the assessment of overload severity considers not only the magnitude of the overload but also its duration. Short-term minor overloads may not immediately lead to equipment failure, but prolonged overloads accelerate equipment aging. Quantifying the overload severity by accumulating the product of overload time and overload magnitude provides a more accurate basis for determining the risk level.

[0064] For example, during periods of high temperature and heavy load in summer, multiple switching devices in the power distribution network of an industrial park simultaneously experienced health deterioration. Analysis using the methods described above revealed that if the main power supply switch failed, load transfer would lead to overload of two interconnecting lines, potentially triggering a cascading power outage at three substations. The risk level distribution map constructed based on this indicated that the area was an extremely high-risk zone, requiring immediate load adjustment or equipment maintenance measures.

[0065] Understandably, by constructing a system risk level distribution map, a shift from analyzing single equipment failures to preventing system-level cascading failures has been achieved, providing a decision-making basis for proactive operation and maintenance and risk management of the power distribution network.

[0066] Step S105: Divide the system risk level distribution map into risk areas, identify key equipment nodes in high-risk areas, extract load transfer schemes for high-risk areas, and determine load distribution paths and power transmission directions.

[0067] The K-means clustering algorithm is used to divide the system risk level distribution map into risk areas. Based on the similarity of risk values ​​and geographical proximity of adjacent nodes, nodes with similar risk levels and adjacent spatial locations are grouped into the same area. If the average risk level within an area exceeds a preset threshold, it is marked as a high-risk area, thus obtaining the risk area boundary. Based on the node distribution within the risk area boundary, the node with the largest power supply load and connected to the outside area by a tie line is identified as a critical equipment node. The transferable line capacity and backup power supply location information of the critical node are obtained from the distribution automation master station. Based on the matching relationship between the transferable capacity and the load to be transferred, a load transfer scheme for the high-risk area is extracted. Using the backup power supply node and the load node to be transferred in the load transfer scheme, the shortest electrical distance path from the backup power supply to the load node is determined by the Dijkstra algorithm based on the remaining capacity and line impedance value of each tie line. The power transmission direction is determined based on the voltage phase angle difference of each node on the path.

[0068] Specifically, in one implementation, the K-means clustering algorithm automatically divides risk areas through iterative optimization. Initially, k nodes are randomly selected as cluster centers, with the value of k determined based on the scale of the distribution network, typically 1.5 times the number of feeders. Each node calculates its comprehensive distance to each cluster center based on its risk value and geographical coordinates. The comprehensive distance is composed of a weighted average of risk value difference and physical distance, with a risk value difference weight of 0.7 and a physical distance weight of 0.3, reflecting the dominant role of risk similarity.

[0069] It should be noted that the identification of critical equipment nodes is based on two core indicators: the size of the power supply load and the strength of the interconnection capability. The power supply load is obtained by summing the rated capacity of all users downstream of the node, reflecting the scope of the power outage impact in the event of a node failure. The interconnection capability is assessed by statistically analyzing the number and total capacity of interconnection lines connecting to areas outside the region. The more interconnection lines and the larger the capacity, the stronger the load transfer capability of the node.

[0070] Specifically, the process of extracting load transfer schemes requires comprehensive consideration of multiple constraints. The distribution automation master station stores complete network topology information, including real-time data such as the rated capacity, current load rate, and switch status of each line. Once a key node is identified, the system queries all connecting lines of that node and calculates the remaining capacity of each line, i.e., the difference between the rated capacity and the current load. Only lines with remaining capacity greater than the load to be transferred can be considered as alternative transfer paths.

[0071] Preferably, the application of Dijkstra's algorithm in distribution networks requires adaptive modifications to traditional algorithms. The algorithm uses line impedance as the weight of edges; a smaller impedance value indicates a shorter electrical distance and lower power transmission loss. Starting from a backup power source node, the algorithm gradually expands the search range, adding the node with the smallest cumulative impedance to the determined set at each step, until it reaches the load node to be transferred.

[0072] For example, the determination of the power transmission direction is based on the fundamental principle of power systems: power flows from nodes with leading voltage phase angles to nodes with lagging phase angles. After determining the load distribution path, the voltage phase angles of each node on the path are obtained through power flow calculations. The phase angle difference between adjacent nodes determines the power flow direction, ensuring that the transferred load can be smoothly transmitted from the backup power source to the target node.

[0073] Step S106: Based on the load distribution path, power transmission direction, and key equipment nodes in high-risk areas, determine the correspondence between equipment health status and failure probability by analyzing the degree of equipment wear and the matching degree of the operating environment, and determine the risk assessment results of each equipment node based on the operating conditions after load transfer.

[0074] Based on the load distribution path and power transmission direction, current density, voltage deviation, and temperature data of each node along the path are extracted. Combined with historical wear records of key equipment nodes in high-risk areas, the absolute value of the difference between actual operating parameters and rated parameters is divided by the rated parameters to obtain the equipment wear degree and operating environment matching degree. Using this matching degree, fault cases under similar operating conditions are queried from historical fault records. A sequence of health state feature values ​​before the fault occurs is extracted. Using Bayesian inference, the posterior probability of a fault occurring under the current health state is calculated based on prior probability and likelihood function, determining the correspondence curve between equipment health state and fault probability. Based on this correspondence curve, the new operating conditions of each node after load transfer are simulated, including changes in load factor, voltage level, and power factor after the transfer. If the operating conditions exceed the equipment's safe operating range, the fault probability value is increased by the product of the excess magnitude and a preset coefficient to obtain the corrected fault probability. The location weight is determined by using the corrected failure probability and the electrical distance of the equipment from the power source in the load distribution path. The reliability weight is set according to the power supply reliability requirements. The comprehensive risk index of each equipment node is calculated by weighted summation of failure probability, location weight and reliability weight. The risk assessment result of each equipment node is determined according to the preset range of the risk index.

[0075] Specifically, in one implementation, the calculation of the equipment wear level and its matching degree with the operating environment involves a comprehensive evaluation of multi-dimensional parameters. After the load distribution path is determined, the current density, voltage deviation, and temperature data of each node on the path are extracted from real-time monitoring data. The current density is obtained by measuring the current flowing through the conductor and dividing it by the conductor's cross-sectional area, reflecting the conductor's current-carrying capacity; the voltage deviation is calculated by the difference between the actual voltage and the rated voltage, reflecting the voltage quality; and the temperature data is obtained through infrared thermography or contact temperature sensors, directly reflecting the equipment's thermal state. The matching degree value is calculated using a normalization method, dividing the absolute value of the difference between the actual value and the rated value of each parameter by the rated value, and then weighting the three normalized parameters by 0.4 for current density, 0.3 for voltage deviation, and 0.3 for temperature, to obtain the comprehensive matching degree value.

[0076] It should be noted that the application of Bayesian inference in calculating equipment failure probability is based on conditional probability theory. The prior probability is obtained from historical failure statistics and represents the basic probability of equipment failure without current health status information. The likelihood function describes the probability of observing current health status characteristics given that the equipment is about to fail. Using Bayes' theorem, the prior probability is multiplied by the likelihood function and then divided by the normalization constant to obtain the posterior probability, which is the updated probability of equipment failure given the observed current health status characteristics. The health status characteristic value sequence includes time-series data in multiple dimensions such as vibration amplitude, temperature rise rate, and contact resistance growth rate. This data is recorded starting 30 days before the failure, sampled daily, forming a feature evolution trajectory. The corresponding relationship curve is obtained through statistical analysis and fitting of a large number of failure cases. The horizontal axis represents the comprehensive health status score, and the vertical axis represents the failure probability. The curve exhibits an exponential growth trend; when the health status score falls below 40 points, the failure probability increases sharply.

[0077] Specifically, simulating operating conditions after load transfer requires recalculating the power flow distribution of the power grid. When loads in high-risk areas are transferred via backup paths, lines that were previously lightly loaded may become heavily loaded, leading to increased line losses and voltage drops. The load factor is calculated as the ratio of the actual load after transfer to the line's rated capacity; the voltage level is obtained from the voltage amplitude at each node through power flow calculations; and the power factor reflects the ratio of active power to apparent power, demonstrating the effect of reactive power compensation.

[0078] For example, the safe operating range of equipment is defined based on technical parameters and industry standards provided by the manufacturer. For 10kV switchgear, the safe range for load factor is 0 to 0.8, the safe range for voltage level is 0.93 to 1.07 times the rated voltage, and the safe range for power factor is 0.85 to 1.0. When any parameter exceeds the safe range, the excess is calculated, which is the difference between the actual value and the safe boundary divided by the width of the safe range.

[0079] Preferably, the fault probability is corrected using a piecewise linear adjustment method. When the deviation is less than 10%, the fault probability increases by 5% of the original value; when the deviation is between 10% and 20%, the fault probability increases by 15% of the original value; and when the deviation is greater than 20%, the fault probability increases by 30% of the original value. This piecewise adjustment method avoids abrupt changes in the fault probability and ensures the continuity of the evaluation results.

[0080] In one possible implementation, location weights are determined by considering the importance of the equipment in the power supply path. Electrical distance is obtained by summing the impedance values ​​of all lines along the path from the power source to the equipment node. A smaller electrical distance indicates that the equipment is closer to the power source, resulting in a greater impact on downstream systems during a fault, thus assigning a higher location weight. Location weights are calculated using an inverse proportional function, with the weight value equal to the maximum electrical distance divided by the electrical distance of the node. Furthermore, power supply reliability requirements are determined based on the importance level of the load. Level 1 loads, such as hospitals and data centers, have a reliability weight of 1.0; Level 2 loads, such as commercial centers and residential areas, have a weight of 0.7; and Level 3 loads, such as general industrial users, have a weight of 0.4. The comprehensive risk index is obtained by weighted summation of three factors: fault probability (50%), location weight (30%), and reliability weight (20%).

[0081] For example, in a risk assessment of a power distribution network in the core area of ​​a city, a 10kV switchgear supplying power to a hospital was calculated to have a failure probability of 0.15, a location weight of 0.8, a reliability weight of 1.0, and a comprehensive risk index of 0.15×0.5+0.8×0.3+1.0×0.2. Based on the preset risk range, this equipment was assessed as a medium-to-high risk level and required priority for maintenance.

[0082] Step S107: When the dynamic risk assessment results show that the system risk exceeds the safety threshold, the load distribution path, power transmission direction and actual wear status of the switching equipment are integrated to generate adjustment instructions that include load transfer and equipment maintenance timing, thereby optimizing the overall system stability.

[0083] When the dynamic risk assessment results show that the system risk exceeds the preset safety threshold, the current load value and transferable capacity of each node on the load distribution path are read. The source and target nodes for power flow transfer are determined based on the power transmission direction. Combining the ratio of the actual wear status score of the switching equipment to its rated life, an equipment operation constraint matrix is ​​constructed, including node load limits and equipment operation frequency restrictions. Based on the equipment operation constraint matrix, a dynamic programming algorithm is used to calculate the optimal order of load transfer. Under the conditions that the line capacity does not exceed the rated value and the node voltage deviation does not exceed the preset range, the load transfer amount and transfer path for each time period are determined, resulting in a load transfer sequence table. Using the load transfer sequence table, the priority sequence of equipment maintenance is determined based on the equipment wear degree and failure probability. If the equipment is on a load transfer path, its maintenance start time is set to the load transfer completion time of that path plus a preset safety interval, resulting in an equipment maintenance sequence table. Based on the load transfer timing table and the equipment maintenance timing table, an adjustment instruction sequence is compiled in chronological order. Each instruction includes the equipment number, opening and closing action, execution time, and target status parameters. These instructions are then sent to each execution terminal through the distribution automation master station to optimize the overall system stability.

[0084] Specifically, in one implementation, the construction of the equipment operation constraint matrix comprehensively considers multiple constraints on power grid operation. When system risk exceeds the safety threshold, the distribution automation master station immediately activates the emergency response mechanism, extracting the current load value and transferable capacity of all nodes on the load distribution path from the real-time database. The current load value is collected in real time through telemetry terminals and updated every 15 seconds; the transferable capacity is calculated by subtracting the current load value from the line's rated capacity. The ratio of wear status score to rated life reflects the remaining service life of the equipment; the smaller the ratio, the closer the equipment is to the end of its life, requiring stricter operational constraints. The equipment operation constraint matrix is ​​a two-dimensional array, with rows representing each node and columns containing constraint parameters such as node load limit, equipment action limit, and minimum action interval. The node load limit is dynamically adjusted according to the current health status of the equipment; healthy equipment can carry 100% of the rated load, while the load limit of severely worn equipment is reduced to 70% to 80% of the rated value. The equipment action limit is based on mechanical life considerations; the total number of actions of each switchgear during its lifespan should not exceed the upper limit specified by the manufacturer.

[0085] It should be noted that the application of dynamic programming in load transfer optimization follows the principle of optimal substructure. The algorithm divides the load transfer process into multiple stages, each corresponding to a time slice of 5 minutes. In each stage, it determines which loads to transfer and along which path. State variables include the load distribution and switching states of each node, while decision variables are the amount of load transferred and the chosen transfer path. The state transition equation describes the change in load distribution from the current stage to the next, and the objective function is to minimize power loss and the number of switching actions during the transfer process. Line capacity constraints require that the power on the line never exceed its rated capacity; voltage deviation constraints require that the voltage at each node remain within ±7% of its rated value. The algorithm solves this by recursively solving backwards from the final state to the initial state, obtaining the globally optimal load transfer sequence.

[0086] Specifically, a multi-attribute decision-making method is used to determine equipment maintenance priorities. Wear level is quantified using indicators such as contact wear depth and spring fatigue; equipment with more severe wear has a higher maintenance priority. Failure probability reflects the likelihood of equipment failure within a future period; a higher probability necessitates more prompt maintenance. When equipment is on a load transfer path, its maintenance must wait for the load transfer to complete to avoid power outages due to equipment maintenance during the transfer process.

[0087] Preferably, the adjustment of the maintenance time window follows the principle of safety. The load transfer completion time is obtained by accumulating the transfer time of each stage, generally requiring 3 to 5 minutes for every 100 kW of load transferred. The safety interval is set to 30 minutes to ensure that the system operates stably after the load transfer and that all parameters return to normal before equipment maintenance begins.

[0088] For example, the instruction sequence is compiled using a standardized format. Equipment numbers are encoded using a six-digit code: the first two digits represent the substation number, the middle two represent the feeder number, and the last two represent the switch serial number. Opening and closing actions are represented in binary, with 0 representing opening and 1 representing closing. Execution timestamps are used, accurate to the second. Target status parameters include the expected current value, voltage value, and power factor, serving as the basis for post-execution verification.

[0089] In one possible implementation, the distribution automation master station sends instructions to each execution terminal via a dedicated communication network. The communication uses the IEC 61850 protocol to ensure the real-time performance and reliability of instruction transmission. After receiving the instruction, the execution terminal first performs a safety check to confirm that the current state allows the operation, and then executes the action at the specified time. Furthermore, the optimization of system stability is reflected in several aspects. Reasonable load transfer avoids cascading failures caused by local overloads; orderly equipment maintenance reduces the probability of sudden failures; and precise timing control reduces disturbances during operation.

[0090] For example, in a practical application of a power distribution network in an industrial park, when the risk value of the switchgear on the main power supply line reaches 0.85, an automatic adjustment program is initiated. Using a dynamic programming algorithm, the 2000 kW load within the park is transferred to the backup line in three batches, with a 10-minute interval between each batch. Simultaneously, maintenance of high-risk switches is scheduled to begin 45 minutes after the load transfer is completed. The entire process achieves uninterrupted power supply, ensuring the normal production of enterprises within the park.

[0091] Understandably, through intelligent scheduling decisions and refined execution control, the organic integration of distribution network risk management and operation optimization has been achieved, thereby improving power supply reliability and equipment utilization efficiency.

[0092] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application 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 this application.

Claims

1. A method for intelligent identification and early warning of operational risks in power distribution networks, characterized in that, include: Vibration waveform data and current waveform data are collected after the switch opening and closing operations. The wear depth of the contacts and the fatigue degree of the springs are extracted to identify the contact reliability of the switch mechanism and determine the preliminary health status of the equipment. The operating conditions of the power grid nodes are obtained, and the preliminary health status of the equipment is matched with the operating conditions to determine the increase in contact resistance, identify the impact of load fluctuations on the degradation of equipment contact performance, and determine the actual wear status of the switch equipment. Based on the actual wear condition of the switchgear, the probability of contact failure is calculated, and a fault warning signal is generated. Based on the equipment fault warning signal, assess the possibility of cascading failures due to overload of adjacent equipment after the power flow redistribution, and determine the system risk level distribution map; The system risk level distribution map is divided into risk areas to determine the load distribution path and power transmission direction. Based on the load distribution path and power transmission direction, the degree of equipment wear and the degree of matching with the operating environment are analyzed to determine the correspondence between equipment health status and failure probability. Based on the operating conditions after load transfer, the risk assessment results of each equipment node are determined. When the risk assessment results of each equipment node show that the system risk exceeds the threshold, an adjustment instruction including load transfer and equipment maintenance timing is generated.

2. The intelligent identification and early warning method for distribution network operation risks according to claim 1, characterized in that, The acquisition of vibration and current waveform data after switch opening and closing operations, extraction of contact wear depth and spring fatigue degree, identification of switch mechanism contact reliability, and determination of preliminary equipment health status include: Vibration waveform data after switch opening and closing operations are acquired, noise is filtered out, vibration characteristic signals are extracted, and the deviation ratio between the vibration amplitude and the standard vibration amplitude is calculated. Based on the relationship between the deviation ratio and the contact wear depth, the contact wear depth data is determined. Current waveform data during the opening and closing process is acquired, harmonic components are analyzed, and the harmonic distortion rate is calculated. Based on the difference between the harmonic distortion rate and historical operation records, the spring fatigue degree value is determined. Based on the contact wear depth data and the spring fatigue degree value, the contact resistance increment is calculated, the total contact resistance growth rate is determined, and the contact reliability level is determined by comparing the total contact resistance growth rate with a preset threshold. Based on the contact reliability level, the cumulative number of operations, and the standard deviation of the operation interval time, the equipment comprehensive deterioration index is calculated by weighted summation to determine the preliminary equipment health status.

3. The intelligent identification and early warning method for distribution network operation risks according to claim 2, characterized in that, The step of calculating the contact resistance increment based on the contact wear depth data and the spring fatigue value, determining the total contact resistance growth rate, and comparing the total contact resistance growth rate with a preset threshold to determine the contact reliability level includes: Based on the contact wear depth data, calculate the resistance increment caused by wear; based on the spring fatigue degree value, calculate the resistance increment caused by fatigue; add the resistance increment caused by wear and the resistance increment caused by fatigue to obtain the total contact resistance growth rate; compare the total contact resistance growth rate with a preset threshold to determine the contact reliability level.

4. The intelligent identification and early warning method for distribution network operation risks according to claim 1, characterized in that, The process of obtaining the operating conditions of the power grid nodes, matching the preliminary equipment health status with the operating conditions, determining the increase in contact resistance, identifying the impact of load fluctuations on the degradation of equipment contact performance, and determining the actual wear status of the switchgear includes: The operating conditions of the power grid nodes are obtained, the voltage deviation rate and current density values ​​of the nodes where the switching equipment is located are read, the number of current peak changes and voltage drop depths are counted within the load fluctuation cycle, the load fluctuation intensity is determined by the product of the number of current peak changes and the voltage drop depth, and the distribution characteristics of the load change frequency and the statistical mean of the load change amplitude are analyzed by the proportional relationship between the load fluctuation intensity and the contact performance degradation, so as to determine the actual wear condition of the switching equipment.

5. The intelligent identification and early warning method for distribution network operation risks according to claim 1, characterized in that, After determining the actual wear condition of the switchgear, the process includes: Based on the actual wear status, the probability of equipment contact failure is calculated; if the probability of equipment contact failure exceeds a threshold, the fault time series and type are extracted from historical fault statistics to generate the equipment fault warning signal; based on the equipment fault warning signal, nodes with similar operating environments and wear characteristics are identified, and they are grouped by clustering algorithm to obtain a set of potential equipment failure risk points; based on the set of potential equipment failure risk points, geographical coordinates and load levels are extracted to construct a risk distribution feature vector.

6. The intelligent identification and early warning method for distribution network operation risks according to claim 1, characterized in that, The step of assessing the likelihood of cascading failures due to overload of adjacent equipment after power flow redistribution based on the equipment fault warning signal, and determining the system risk level distribution map, includes: By using the failure probability value and fault type identifier in the equipment fault warning signal, a risk score is calculated, high-risk equipment nodes with risk scores exceeding the risk score threshold are marked, and the location information of the high-risk equipment nodes in the power grid topology is obtained. The power grid topology changes after the failure of the high-risk equipment nodes are simulated, the power distribution of each line is calculated, the power is compared with the rated capacity, the overload risk of adjacent equipment is determined, and potential cascading fault nodes are identified. Based on the potential cascading fault nodes, a node adjacency matrix is ​​constructed, and the fault propagation path is traversed through a search algorithm to extract the path sequence. Based on the path sequence, the path risk value is calculated, and combined with the overload severity, the system risk level distribution map is determined.

7. The intelligent identification and early warning method for distribution network operation risks according to claim 1, characterized in that, The step of dividing the system risk level distribution map into risk areas and determining load distribution paths and power transmission directions includes: The system risk level distribution map is divided into risk areas using a clustering algorithm, and areas where the average risk level exceeds the average risk level threshold are marked as high-risk areas. The nodes with the largest load and connecting lines in the high-risk areas are identified as key equipment nodes. Based on the transferable capacity of the key equipment nodes, a load transfer scheme is extracted. Based on the load transfer scheme, the electrical distance of the connecting lines is calculated, and the load distribution path and the power transmission direction are determined.

8. The intelligent identification and early warning method for distribution network operation risks according to claim 7, characterized in that, The step of extracting a load transfer scheme based on the transferable capacity of the key equipment nodes includes: Obtain the available power supply capacity and backup power location information of the key equipment nodes; determine the load transfer scheme based on the matching relationship between the available power supply capacity and the load to be transferred.

9. The intelligent identification and early warning method for distribution network operation risks according to claim 1, characterized in that, The process involves analyzing the degree of equipment wear and its compatibility with the operating environment based on the load distribution path and the power transmission direction, determining the correlation between equipment health status and failure probability, and determining the risk assessment results for each equipment node based on the operating conditions after load transfer. Based on the load distribution path and power transmission direction, the degree of equipment wear and the matching degree with the operating environment are calculated. Based on the matching degree, fault cases under similar operating conditions are queried from historical fault records, and the health status feature value sequence before the fault occurs is extracted to determine the correspondence between equipment health status and fault probability. Based on the correspondence, the new operating conditions of each node after load transfer are simulated. If the operating conditions exceed the safe operating range of the equipment, the fault probability value is increased by the product of the excess range and a preset coefficient to obtain the corrected fault probability. The comprehensive risk index of each equipment node is calculated using the corrected fault probability, and the risk assessment result of each equipment node is determined based on the preset range in which the risk index is located.

10. The method for intelligent identification and early warning of distribution network operation risks according to claim 1, characterized in that, When the risk assessment results of each device node show that the system risk exceeds the threshold, an adjustment instruction including load transfer and equipment maintenance timing is generated, including: When the risk assessment result exceeds the threshold, the load value and transferable capacity of the load allocation path are read, the power flow transfer node is determined according to the power transmission direction, and an operation constraint matrix is ​​constructed in combination with the actual wear status; a load transfer timing table is calculated according to the operation constraint matrix; the equipment maintenance priority is determined according to the load transfer timing table and the actual wear status, and a maintenance timing table is generated; and adjustment instructions are compiled according to the load transfer timing table and the maintenance timing table.

Images