A new energy intelligent box-type substation operation monitoring method and system

Abstract

The application relates to the technical field of data processing, and discloses a new energy intelligent box-type substation operation monitoring method and system. The method comprises the following steps: functionally dividing a substation space according to a new energy access capacity to obtain power density distribution parameters; performing denoising processing on electrical parameters through wavelet transform to obtain purified operation data; establishing a fuzzy feature analysis model to obtain an equipment operation state fuzzy feature vector; calculating new energy output-load coupling features to obtain a power fluctuation abnormal factor and a local abnormal factor; and processing the local abnormal factor through BP neural network training to obtain an equipment fatigue accumulation monitoring result. Through the establishment of the equipment fatigue monitoring system based on the new energy power fluctuation abnormal factor and the BP neural network, the accuracy of new energy substation equipment state evaluation and the timeliness of fault early warning are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, and in particular to a new energy intelligent box-type substation operation monitoring method and system. BACKGROUND

[0002] The new energy intelligent box-type substation operation monitoring method in the prior art mainly adopts a traditional threshold judgment and rule matching method for device state monitoring. By deploying temperature sensors, current transformers, voltage transformers and other monitoring devices inside the substation, the operating parameters of key devices such as transformers and switching devices are collected in real time. When the monitoring parameters exceed the preset threshold, an alarm signal is triggered, and the alarm information is transmitted to the monitoring center through the SCADA system. These traditional monitoring methods mainly rely on single parameter value comparison and simple logical judgment to qualitatively evaluate the device operating state, which can to some extent find obvious faults and abnormal states of the device.

[0003] However, the traditional monitoring method lacks specific analysis of the intermittent and fluctuating characteristics of new energy power generation, and cannot effectively identify the cumulative impact of new energy power fluctuations on substation equipment. Secondly, the existing method is mainly based on single parameter threshold judgment, lacks multi-parameter fusion analysis capability, and cannot accurately reflect the comprehensive operating state of the device. Thirdly, the traditional monitoring system lacks intelligent fault prediction function, and can only find problems after the device parameters are obviously abnormal, and cannot realize early warning and preventive maintenance.

[0004] Due to the inability to accurately identify the new energy power fluctuation mode, the sensor layout lacks pertinence, which affects the effectiveness of data collection. In the data preprocessing process, due to the lack of a special new energy power fluctuation filtering algorithm, the collected data contains a large amount of noise interference. In the device state evaluation stage, due to the lack of fuzzy feature analysis model considering new energy working conditions, the fatigue degree of the device cannot be accurately quantified. Finally, in the abnormal detection and prediction link, due to the lack of intelligent algorithm based on new energy power fluctuation characteristics, it is difficult to realize accurate device fatigue accumulation prediction and residual life evaluation. SUMMARY

[0005] The present application provides a new energy intelligent box-type substation operation monitoring method and system, which is used to solve the problem that the existing new energy intelligent box-type substation operation monitoring method lacks specialized monitoring technology for new energy power fluctuation characteristics, and cannot accurately identify and predict the device fatigue accumulation state. By establishing a device fatigue monitoring system based on new energy power fluctuation abnormal factors and BP neural network, the accuracy of new energy substation device state evaluation and the timeliness of fault warning are improved.

[0006] In a first aspect, the application provides a new energy intelligent box-type substation operation monitoring method, which comprises: functionally dividing a box-type substation space according to a new energy access capacity to obtain power density distribution parameters of a transformer area, a switch cabinet area and a control room area; performing denoising processing on collected electrical parameters through wavelet transform to obtain purified operation data eliminating power fluctuation interference; establishing a fuzzy feature analysis model based on the purified operation data to obtain a fuzzy feature vector of a device operation state; calculating a new energy output-load coupling feature according to the fuzzy feature vector to obtain a new energy power fluctuation abnormal factor and a local abnormal factor; and training and processing the local abnormal factor through a BP neural network to obtain a deviation value of a current operation state from a normal benchmark state and a device fatigue accumulation monitoring result.

[0007] In a second aspect, the application provides a new energy intelligent box-type substation operation monitoring system, which comprises:

[0008] A division module is configured to functionally divide a box-type substation space according to a new energy access capacity to obtain power density distribution parameters of a transformer area, a switch cabinet area and a control room area;

[0009] A denoising module is configured to perform denoising processing on collected electrical parameters through wavelet transform to obtain purified operation data eliminating power fluctuation interference;

[0010] An establishment module is configured to establish a fuzzy feature analysis model based on the purified operation data to obtain a fuzzy feature vector of a device operation state;

[0011] A calculation module is configured to calculate a new energy output-load coupling feature according to the fuzzy feature vector to obtain a new energy power fluctuation abnormal factor and a local abnormal factor;

[0012] A monitoring module is configured to train and process the local abnormal factor through a BP neural network to obtain a deviation value of a current operation state from a normal benchmark state and a device fatigue accumulation monitoring result.

[0013] In a third aspect, a new energy intelligent box-type substation operation monitoring device is provided, which comprises a memory and at least one processor, the memory storing instructions; the at least one processor invokes the instructions in the memory to enable the new energy intelligent box-type substation operation monitoring device to perform the new energy intelligent box-type substation operation monitoring method described above.

[0014] In a fourth aspect, a computer readable storage medium is provided, which stores instructions, when executed on a computer, enabling the computer to perform the new energy intelligent box-type substation operation monitoring method described above.

[0015] The technical solution provided in this application functionally divides the space of the prefabricated substation according to the renewable energy access capacity, obtaining power density distribution parameters for the transformer area, switchgear area, and control room area. This solves the problem of the lack of specificity in sensor layout in traditional monitoring methods, and can accurately determine the location of monitoring points based on the characteristics of renewable energy power transmission and the equipment's carrying capacity, effectively improving the coverage and accuracy of the monitoring system. Wavelet transform is used to denoise the collected electrical parameters, obtaining purified operating data that eliminates power fluctuation interference. This specifically addresses the power oscillation interference problem caused by intermittent renewable energy generation. Compared to traditional simple filtering methods, wavelet transform can perform analysis in both the time and frequency domains, effectively preserving the true operating status information of the equipment while eliminating renewable energy power fluctuation noise, providing a high-quality data foundation for subsequent analysis. A fuzzy feature analysis model is established based on the purified operating data to obtain fuzzy feature vectors of the equipment operating status. This overcomes the limitation of traditional monitoring methods that can only make qualitative judgments. By using fuzzy logic theory to handle the uncertainty and boundary fuzziness of the equipment status, accurate quantitative assessment of the equipment operating status is achieved.

[0016] An algorithm for calculating the coupling characteristics of renewable energy output-load based on fuzzy feature vectors is designed to address the intermittent and fluctuating nature of renewable energy generation. It accurately identifies the matching relationship between photovoltaic and wind power generation and load demand. By calculating standardized deviations, it obtains renewable energy power fluctuation anomaly factors and local anomaly factors, providing crucial feature inputs for subsequent intelligent analysis. This coupling feature analysis method has stronger predictive capabilities and higher accuracy compared to traditional single-parameter monitoring. An intelligent algorithm that trains local anomaly factors using a BP neural network fully considers the complex impact of renewable energy power fluctuations on equipment fatigue accumulation. By using deep learning to extract implicit patterns from historical equipment operating data, it accurately predicts the deviation between the current operating state and the normal baseline state and quantifies the degree of equipment fatigue accumulation. This achieves a shift from passive monitoring to proactive prediction, significantly improving the timeliness of equipment fault warnings. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a schematic diagram of one embodiment of the operation monitoring method for new energy intelligent prefabricated substations in this application.

[0019] Figure 2 This is a schematic diagram of one embodiment of the new energy intelligent prefabricated substation operation monitoring system in this application.

[0020] Figure 3 This is a schematic block diagram of the operation monitoring equipment for new energy intelligent box-type substations in an embodiment of the present invention. Detailed Implementation

[0021] This application provides a method and system for monitoring the operation of a new energy intelligent prefabricated substation. The terms "first," "second," "third," "fourth," etc. (if present)," in the specification, claims, and accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0022] For ease of understanding, the specific process of the embodiments of this application is described below. Please refer to [link / reference]. Figure 1 One embodiment of the operation monitoring method for new energy intelligent prefabricated substations in this application includes:

[0023] Step S101: Based on the new energy access capacity, the space of the prefabricated substation is functionally divided to obtain the power density distribution parameters of the transformer area, switchgear area, and control room area.

[0024] Step S102: Denoise the collected electrical parameters by wavelet transform to obtain purified operation data with power fluctuation interference eliminated;

[0025] Step S103: Establish a fuzzy feature analysis model based on the purification operation data to obtain the fuzzy feature vector of the equipment operating status;

[0026] Step S104: Calculate the new energy output-load coupling characteristics based on the fuzzy feature vector to obtain the new energy power fluctuation anomaly factor and local anomaly factor;

[0027] Step S105: Train the local abnormal factors through a BP neural network to obtain the deviation between the current operating state and the normal baseline state and the equipment fatigue accumulation monitoring results.

[0028] It is understood that the executing entity of this application can be a new energy intelligent prefabricated substation operation monitoring system, or it can be a terminal or a server; the specific implementation is not limited here. This application's embodiment uses a server as an example for illustration.

[0029] Specifically, the power density distribution parameters are obtained by analyzing historical output data of photovoltaic and wind power generation to establish a power fluctuation amplitude-frequency distribution model. This model identifies the spatial distribution pattern of new energy power transmission within the substation. The transformer area bears the main power conversion load and therefore has the highest power density. The switchgear area bears the impact of frequent switching operations and has a moderate power density. The control room area mainly bears the power density of control signal transmission and has the lowest power density. The sensor deployment density is determined by calculating the equipment load-bearing capacity coefficient of each area. Four sensors are deployed per square meter in the transformer area, three sensors are deployed per square meter in the switchgear area, and two sensors are deployed per square meter in the control room area. Finally, a sensor coordinate position matrix containing accurate coordinate information is generated.

[0030] Wavelet transform denoising is specifically designed to address electrical parameter oscillation interference caused by power fluctuations in new energy sources. The multi-scale wavelet decomposition algorithm decomposes the collected electrical parameter signals into wavelet coefficients at different frequency scales. Power fluctuations caused by intermittent power generation from new energy sources are mainly concentrated in the low-frequency band, while the normal operation signals of the equipment are distributed in the mid-to-high frequency band. By identifying the distribution characteristics of wavelet coefficients, the power oscillation interference components are adaptively filtered out. The reconstructed electrical signal removes the power instability caused by changes in sunlight and wind speed fluctuations. The generated purified operation data retains the true operating status information of the equipment while eliminating the noise from power generation fluctuations in new energy sources.

[0031] The fuzzy feature analysis model establishment process extracts features from the purified operation data, including transformer temperature rise parameters, transformer load current parameters, transformer insulation resistance parameters, switch equipment operation frequency parameters, switch equipment contact resistance parameters, and switch equipment mechanical characteristic parameters. The fuzzy feature analysis model adopts a five-level fatigue level classification standard, including slight fatigue, moderate fatigue, severe fatigue, dangerous fatigue, and ultimate fatigue. Each operating parameter is mapped to the corresponding fatigue level according to its numerical range, and the membership degree value is calculated. The fuzzy inference rule base establishes the correlation between parameter status and fatigue level. By comprehensively evaluating the membership degree values ​​of multiple parameters, the overall operating status assessment of the equipment is obtained. The final output fuzzy feature vector contains three dimensions of information: status level membership degree value and change trend.

[0032] The calculation process of renewable energy output-load coupling characteristics first aligns the fuzzy feature vector of equipment operating status with the photovoltaic output curve and wind power curve in time series to generate a renewable energy output time series data sequence. This sequence is then correlated with the load demand curve to establish a coupling relationship matrix. The coupling relationship matrix reflects the dynamic correlation between renewable energy output changes and load changes. When renewable energy output and load demand do not match, a power fluctuation anomaly factor is generated. The degree of power fluctuation is quantified by calculating the standardized deviation between the current power value and the historical moving average power value. Local anomaly factors are obtained by analyzing the equipment stress distribution data caused by power mismatch. The local anomaly detection algorithm identifies abnormal points in the equipment operating parameters that deviate from the normal distribution. These abnormal points reflect the local stress concentration impact caused by renewable energy power fluctuations on specific equipment.

[0033] The BP neural network training process uses local anomaly factors as network input to establish a three-layer neural network architecture. The input layer receives local anomaly factor data, the hidden layer performs feature mapping transformation, and the output layer generates state deviation prediction values. The training dataset is labeled with historical normal operating state data to obtain labels containing normal and abnormal states. During the forward propagation calculation, the local anomaly factor data is processed by weight matrix transformation and activation function to generate the network output prediction value. The error function calculates the difference between the predicted value and the actual label value. The backpropagation algorithm adjusts the network weight parameters according to the error gradient until convergence. The trained neural network predicts the state of the current equipment operating data and outputs the numerical deviation value between the current operating state and the normal baseline state. Combined with the equipment fatigue accumulation calculation, the deviation value is converted into fatigue accumulation percentage and remaining life assessment to form a complete equipment fatigue accumulation monitoring result.

[0034] In one specific embodiment, the process of performing step S101 may specifically include the following steps:

[0035] Based on historical output data of photovoltaic and wind power generation, power fluctuation amplitude-frequency distribution modeling is performed to obtain the new energy power transmission path and equipment carrying capacity coefficient;

[0036] The internal space of the prefabricated substation is divided into zones based on the equipment load-bearing capacity coefficient, resulting in functional zoning boundaries for the transformer zone, switchgear zone, control room zone, and cable interlayer zone.

[0037] The power density distribution algorithm is used to calculate the power impact intensity by inputting the functional zone boundary, and the power impact intensity coefficient of each monitoring area is obtained.

[0038] Based on the power impact intensity coefficient and the importance of the equipment, the sensor density configuration is processed to obtain a sensor deployment density of four per square meter in the transformer area, three per square meter in the switch cabinet area, and two per square meter in the control room area.

[0039] The location of monitoring points is optimized based on the sensor deployment density to obtain the sensor coordinate position matrix, monitoring parameter configuration table, and data acquisition frequency setting value.

[0040] Specifically, this is achieved by analyzing the power variation patterns in historical output data of photovoltaic and wind power generation. Historical output data of photovoltaic power generation includes power output curves under different weather conditions, while historical output data of wind power generation covers power generation records under various wind speed conditions. The modeling process classifies and statistically analyzes these historical data according to the power variation amplitude. The statistical results show that power fluctuation amplitudes below 10% of rated power are considered low fluctuations, 10% to 30% of rated power are considered medium fluctuations, and above 30% of rated power are considered high fluctuations. Frequency distribution analysis calculates the number of occurrences of each fluctuation amplitude per unit time and establishes a mathematical relationship model between power fluctuation amplitude and frequency. This model identifies the transmission path of new energy power within the substation, including the main transmission path from the new energy access point to the transformer and the branch transmission paths from the transformer to each load point. The equipment carrying capacity coefficient is calculated by analyzing the load response characteristics of each piece of equipment under different power fluctuation conditions. The transformer carrying capacity coefficient reflects its overload tolerance under power fluctuations, while the switchgear carrying capacity coefficient reflects its mechanical tolerance performance under frequent operation. The zoning process scientifically divides the internal space of the prefabricated substation based on the equipment load-bearing capacity coefficient. The transformer zone is divided according to the transformer load-bearing capacity coefficient and installation space requirements to determine the boundary range. Zones with high transformer load-bearing capacity coefficients require larger safety distances and heat dissipation space. The switchgear zone is divided considering the switchgear load-bearing capacity coefficient and operation and maintenance space requirements. The switchgear load-bearing capacity coefficient affects its layout density and aisle width design. The control room zone is divided based on the electromagnetic compatibility requirements of the control equipment and personnel operating space. The cable interlayer zone is divided according to the cable current carrying capacity and heat dissipation requirements to determine the space configuration. The functional zone boundaries are determined by calculating the power density of equipment and space utilization of each zone to optimize the process. The boundary division results form a spatial distribution map containing the coordinate range and area information of each zone.

[0041] The power impact intensity calculation process inputs the functional zone boundary data into the power density distribution algorithm for precise calculation. The power density distribution algorithm calculates the spatial density distribution of power based on the new energy power transmission path and the distribution of equipment in each area. The transformer area, which undertakes the main power conversion task, has the highest power density, so the power impact intensity coefficient is set to the maximum value. The switchgear area, which is subjected to frequent switching operations, has a power impact intensity coefficient set to a medium value. The control room area, which mainly transmits control signals, has a relatively small power impact intensity coefficient set to a low value. The cable interlayer area, which is subjected to cable transmission power impact intensity coefficient, is calculated and determined based on cable specifications and current carrying capacity. The calculation of the power impact intensity coefficient takes into account the comprehensive influence of power fluctuation amplitude and frequency distribution and equipment response characteristics.

[0042] The sensor density configuration process comprehensively considers two key factors: power surge intensity coefficient and equipment importance. Equipment importance assessment is based on the equipment's role in the substation and the scope of its fault impact. Transformers, as core equipment, have the highest importance, and based on their high power surge intensity coefficient, a sensor density of four per square meter is determined. Switchgear has a relatively high importance and a moderate power surge intensity, so a sensor density of three per square meter is determined. Control room equipment has a moderate importance and a low power surge intensity, so a sensor density of two per square meter is determined. The sensor density configuration process also considers sensor coverage and monitoring accuracy requirements. The density configuration results ensure comprehensive monitoring of key parameters and timely detection of abnormal conditions in each area.

[0043] The monitoring point location optimization calculation is performed based on the sensor deployment density to design a precise spatial layout. The optimization calculation considers two main objectives: the coverage effect of the sensor monitoring range and the signal transmission quality. The sensor coordinate position matrix generates three-dimensional position information including horizontal, vertical, and height coordinates by calculating the optimal spatial coordinates of each monitoring point. The monitoring parameter configuration table determines the monitoring parameter type of each sensor according to the characteristics of the equipment in each area and the monitoring requirements. Sensors in the transformer area mainly monitor temperature, current, and vibration parameters. Sensors in the switch cabinet area focus on monitoring the number of switching operations, contact resistance, and mechanical characteristic parameters. Sensors in the control room area monitor environmental temperature, humidity, and electromagnetic interference parameters. The data acquisition frequency setpoint is determined based on the variation characteristics of each parameter and the monitoring accuracy requirements. High-frequency acquisition is used for rapidly changing electrical parameters, and low-frequency acquisition is used for slowly changing environmental parameters.

[0044] In one specific embodiment, the process of performing step S102 may specifically include the following steps:

[0045] Based on the power density distribution parameters, the voltage, current, power, temperature and vibration parameters in the substation are collected and processed in real time to obtain the original electrical parameter data under the new energy operation conditions;

[0046] The original electrical parameter data is input into a timestamp synchronization algorithm for time alignment processing to obtain a synchronized electrical parameter sequence with a unified time base;

[0047] Based on the synchronous electrical parameter sequence, a multi-scale wavelet decomposition algorithm is used to perform frequency domain decomposition processing to obtain the wavelet coefficient distribution of the new energy power fluctuation frequency band.

[0048] Based on the wavelet coefficient distribution, an adaptive filtering process is performed on the power oscillation interference caused by intermittent power generation from new energy sources to obtain a reconstructed electrical signal with power fluctuation noise removed.

[0049] The reconstructed electrical signals are encapsulated and processed according to the new energy monitoring time sequence format to obtain purification operation data containing equipment identification, parameter type and value.

[0050] Specifically, real-time data acquisition and processing continuously monitors electrical parameters based on the sensor layout determined by power density distribution parameters. Power density distribution parameters guide the precise deployment of sensors in the transformer area, switchgear area, and control room area. Voltage parameter acquisition obtains real-time voltage values ​​of each node through voltage transformers. Current parameter acquisition uses current transformers to measure the instantaneous current flowing through each device. Power parameter acquisition calculates active and reactive power through power measurement devices. Temperature parameter acquisition uses temperature sensors to monitor the surface and internal temperature rise of equipment. Vibration parameter acquisition uses vibration sensors to detect the mechanical vibration characteristics of equipment during operation. The raw electrical parameter data under new energy operation conditions reflects the direct impact of photovoltaic power generation and wind power generation on the substation equipment operation status. The data acquisition frequency is set according to the parameter change characteristics: electrical parameters are acquired 1,000 times per second, temperature parameters are acquired once every 10 seconds, and vibration parameters are acquired 500 times per second. The raw electrical parameter data contains a large amount of signal noise and interference components caused by new energy power fluctuations.

[0051] The timestamp synchronization algorithm addresses the inconsistency in timing of data collected from different sensors. The original electrical parameter data exhibits slight timestamp deviations due to differences in sensor response times and data transmission delays. The algorithm establishes a unified time base using a network time protocol, calibrating all sensor data with millisecond-level precision. The synchronization process first identifies the original timestamps of each sensor's data, then calculates the deviation from the reference time, and finally rearranges all data according to the unified time base. Synchronizing the electrical parameter sequence ensures that different parameter data collected at the same time have the same timestamp. This timing alignment process eliminates data correlation errors caused by time asynchrony, and the generated synchronized electrical parameter sequence provides an accurate data foundation for subsequent frequency domain analysis.

[0052] The multi-scale wavelet decomposition algorithm decomposes the synchronous electrical parameter sequence into multiple levels in the frequency domain. The wavelet decomposition algorithm selects appropriate wavelet basis functions to perform joint time-frequency domain analysis of the signal. The decomposition process decomposes the original signal into wavelet coefficients of different frequency scales. The low-frequency wavelet coefficients reflect the main trend and slowly changing components of the signal, while the high-frequency wavelet coefficients contain detailed information and rapidly changing components of the signal. The fluctuation of new energy power is mainly concentrated in a specific frequency range. The frequency of photovoltaic power generation fluctuation is related to cloud cover and changes in sunlight, while the frequency of wind power generation fluctuation is related to changes in wind speed and wind direction. The wavelet coefficient distribution identifies the frequency domain characteristics of new energy power fluctuation by analyzing the energy distribution of each frequency band. The decomposition results show that the amplitude of the wavelet coefficients in the frequency band of new energy power fluctuation is significantly higher than that in the normal operation frequency band. The frequency domain decomposition process provides accurate frequency positioning information for subsequent interference filtering.

[0053] The adaptive filtering process intelligently identifies and eliminates power oscillation interference caused by intermittent power generation from new energy sources. Based on the frequency domain characteristics of wavelet coefficient distribution, the filtering parameters are adaptively adjusted. Intermittent power generation is characterized by discontinuity and fluctuation in power output. Photovoltaic power generation experiences a sharp drop in power when obscured by clouds and a rapid increase when the weather is clear. Wind power generation exhibits periodic power fluctuations when wind speed changes. Power oscillation interference is identified by analyzing the amplitude changes and frequency distribution of wavelet coefficients. The adaptive filtering algorithm dynamically adjusts the filtering strength and frequency range according to the interference characteristics. The filtering process retains the true operating status information of the equipment while eliminating power fluctuation noise from new energy sources. The reconstructed electrical signal is generated by inverse transforming the wavelet coefficients after filtering. The reconstruction process ensures the integrity and continuity of the signal. The reconstructed electrical signal, with power fluctuation noise removed, truly reflects the operating characteristics of the equipment under stable power conditions.

[0054] Data encapsulation and processing reconstructs electrical signals and organizes them in a standardized manner according to the new energy monitoring time sequence format. This new energy monitoring time sequence format is specifically designed for the data characteristics of new energy substations and includes key information such as equipment identification, parameter type, numerical value, and timestamp. Equipment identification uses a unique code to distinguish different monitoring equipment and sensors. Parameter type identifies specific monitoring parameters such as voltage, current, power, temperature, and vibration. The numerical field stores the actual measured values ​​after filtering and reconstruction processing. The timestamp records the precise time of data acquisition. Encapsulation processing also includes data format conversion and storage optimization. Data format conversion unifies different types of parameter data into a standard data format. Storage optimization improves data storage and retrieval efficiency through data compression and indexing. The purified operating data, after a complete preprocessing process, possesses high quality and high reliability characteristics.

[0055] In one specific embodiment, the process of executing step S103 may specifically include the following steps:

[0056] Based on the purification operation data, feature extraction processing is performed on the operating parameters of transformers and switchgear to obtain a multi-dimensional set of operating features reflecting changes in equipment status.

[0057] The multidimensional operating characteristic set is classified into state levels according to slight fatigue, moderate fatigue, severe fatigue, dangerous fatigue and ultimate fatigue to obtain the equipment fatigue level classification standard;

[0058] Based on the equipment fatigue level classification standard, the operating parameters are fuzzy mapped to obtain the fuzzy membership degree values ​​corresponding to each equipment state level.

[0059] Based on the fuzzy membership degree values, fuzzy inference rules are used for feature fusion processing to obtain a fuzzy feature analysis model that comprehensively reflects the equipment's operating status.

[0060] The fuzzy feature analysis model is used to quantitatively evaluate the current equipment status, resulting in a fuzzy feature vector of equipment operating status that includes status level, membership degree value, and trend of change.

[0061] Specifically, feature extraction processing deeply mines and analyzes the key equipment operating parameters in the purified operation data. Transformer operating parameter feature extraction focuses on temperature rise characteristics, current load characteristics, and insulation resistance characteristics. Temperature rise characteristics are calculated by analyzing the changing trends of transformer oil temperature, winding temperature, and core temperature to determine the temperature rise rate and magnitude. Current load characteristics reflect the transformer's load-bearing status by statistically analyzing the peak, average, and rate of change of load current. Insulation resistance characteristics assess the degree of insulation performance degradation by monitoring the decreasing trend of insulation resistance values. Switchgear operating parameter feature extraction focuses on mechanical, electrical, and operational characteristics. Mechanical characteristics include switch operating torque, operating time, and mechanical wear degree. Electrical characteristics cover contact resistance, breaking capacity, and withstand voltage level. Operational characteristics statistically analyze the number of switch actions, operating frequency, and fault trip records. The multi-dimensional operating feature set combines various individual features to form a feature matrix that comprehensively reflects the equipment status. The feature matrix contains integrated information in terms of time, space, and status dimensions.

[0062] The condition-based classification system establishes a five-level classification system for the fatigue characteristics of equipment in new energy substations. The "Minor Fatigue" level corresponds to a slight deviation of equipment operating parameters from the normal range, but without affecting equipment performance. The "Medium Fatigue" level indicates that equipment parameters have changed significantly and require attention, but have not yet reached the maintenance threshold. The "Severe Fatigue" level indicates that equipment parameters have deteriorated significantly and a maintenance plan needs to be developed. The "Dangerous Fatigue" level warns that equipment parameters are approaching the fault threshold and require emergency handling. The "Ultimate Fatigue" level indicates that the equipment has reached its service life limit and must be shut down immediately for maintenance. The equipment fatigue level classification standard is based on the technical specifications and operating experience of transformers and switchgear. For transformers, a temperature rise exceeding 10% of the rated value is considered minor fatigue; exceeding 20% ​​is moderate fatigue; and exceeding 30% is severe fatigue. For switchgear, 70% of the rated number of operations is considered minor fatigue; 85% is moderate fatigue; and 95% is severe fatigue. The classification standard considers the accelerating effect of new energy power fluctuations on equipment fatigue accumulation.

[0063] Fuzzy mapping converts discrete operating parameter values ​​into continuous fuzzy membership degree representations. The mapping process uses a combination of trapezoidal and triangular membership functions. Trapezoidal membership functions are suitable for state ranges where parameter changes are relatively gradual, while triangular membership functions are suitable for critical ranges where parameter changes are more sensitive. Fuzzy mapping of transformer temperature rise parameters converts temperature values ​​into membership degrees for each fatigue level. When the temperature rise is 15% of the rated value, the membership degree for slight fatigue is 0.7, the membership degree for moderate fatigue is 0.3, and the membership degree for other levels is zero. Fuzzy mapping of the number of switching equipment operations calculates membership degrees based on the ratio of the cumulative number of operations to the rated number of operations. The fuzzy membership degree values ​​reflect the degree to which the equipment state belongs to each fatigue level. The membership degree values ​​range from zero to one, and the larger the value, the closer the equipment state is to that fatigue level. Fuzzy mapping effectively handles the uncertainty of parameter measurement and the fuzziness of state boundaries.

[0064] Feature fusion processing uses fuzzy membership values ​​and fuzzy inference rules to comprehensively evaluate multiple parameters. The fuzzy inference rule base contains decision rules formed by expert experience and historical fault cases. The inference rules adopt an if-then conditional structure, with a typical rule being "If the transformer temperature rise belongs to moderate fatigue and the load current belongs to severe fatigue, then the overall state of the transformer belongs to severe fatigue." The feature fusion algorithm combines the weighted average method and the maximum membership method. The weighted average method assigns weight coefficients according to the importance of each parameter, with the temperature rise parameter having a weight of 0.4, the load current having a weight of 0.3, and the insulation resistance having a weight of 0.3. The maximum membership method selects the fatigue level with the highest membership degree as the dominant state. The fuzzy feature analysis model integrates all inference rules and fusion algorithms to form a complete equipment condition assessment system. The model has self-learning and adaptive capabilities and can adjust the inference rules and weight parameters according to new operating data.

[0065] The quantitative assessment process utilizes a fuzzy feature analysis model to accurately numerically evaluate the current equipment status. The assessment first inputs the current equipment operating parameters into the fuzzy feature analysis model. The model calculates the membership degree values ​​for each fatigue level based on established inference rules. The status level is determined by comparing the membership degrees of each level to identify the equipment's current primary fatigue state. The membership degree values ​​provide information on the confidence level and uncertainty of the status judgment. The trend is calculated by analyzing the direction and rate of change of the membership degree values ​​over a continuous time period; an upward trend indicates increasing fatigue, while a downward trend indicates improved equipment status. The fuzzy feature vector of the equipment operating status integrates the three core elements: status level, membership degree values, and trend. This feature vector provides standardized input data for subsequent coupled feature calculations and anomaly factor extraction. The quantitative assessment results are objective and comparable, facilitating comparative analysis of the status of different equipment and at different times.

[0066] In one specific embodiment, the process of performing feature fusion processing based on the fuzzy membership degree values ​​using fuzzy inference rules can specifically include the following steps:

[0067] The transformer temperature rise parameter, transformer load current parameter, and transformer insulation resistance parameter are weighted based on fuzzy membership values ​​to obtain the weighting matrix of the transformer operating status.

[0068] The weight allocation matrix is ​​fused with the parameters of the number of operation times of the switchgear, the contact resistance parameters of the switchgear, and the mechanical characteristic parameters of the switchgear to obtain a comprehensive evaluation matrix of the operating status of the switchgear.

[0069] Based on the comprehensive evaluation matrix, fatigue reasoning rules for transformers and fatigue reasoning rules for switching equipment are established, resulting in a fuzzy reasoning rule base for equipment condition judgment.

[0070] Based on the fuzzy inference rule base, the transformer operation data and switchgear operation data are comprehensively evaluated and processed to obtain the fuzzy evaluation results of the overall operation status of each device;

[0071] The fuzzy evaluation results were compared and verified with the historical temperature rise curve of the transformer and the historical operation records of the switchgear to obtain a corrected and optimized fuzzy feature analysis model.

[0072] Specifically, the importance of key transformer operating parameters is quantified based on the distribution characteristics of fuzzy membership values. The weight allocation of transformer temperature rise parameters considers their direct impact on equipment lifespan and sensitivity to fault warning. The membership variation of temperature rise parameters under different fatigue levels determines their weight coefficient. When the membership variation of temperature rise is large between mild and moderate fatigue, the weight coefficient increases accordingly. The weight allocation of transformer load current parameters reflects their impact on the equipment's load-bearing capacity. The peak value variation and duration of the load current affect the calculation of the weight coefficient. The weight allocation of transformer insulation resistance parameters reflects their key role in the safe operation of the equipment. The rate and magnitude of insulation resistance decrease determine their proportion in the weight allocation. The weight allocation matrix is ​​normalized to ensure that the sum of the weights of each parameter equals one. The matrix is ​​in the form of a three-row, three-column structure, with rows representing parameter types and columns representing fatigue levels. The weight allocation process dynamically adjusts the importance ranking of each parameter under different operating conditions.

[0073] The fusion processing comprehensively analyzes the transformer weight allocation matrix and the operating parameters of the switchgear. The switchgear operation frequency parameter reflects the degree of mechanical fatigue accumulation, and the ratio of operation frequency to rated life frequency directly affects the fatigue level judgment of the equipment. The switchgear contact resistance parameter reflects the degradation of its electrical performance, and the increase magnitude and rate of increase of contact resistance determine the degree of electrical fatigue of the equipment. The switchgear mechanical characteristic parameters include operating torque, operating time, and mechanical wear degree. The changing trend of mechanical characteristic parameters reflects the deterioration of the mechanical performance of the equipment. The fusion processing uses matrix multiplication to mathematically combine the transformer weight information and the switchgear parameter information. The fusion algorithm considers the operational correlation between the transformer and the switchgear. The transformer load change will affect the operating frequency of the switchgear, and the switchgear failure will affect the operational stability of the transformer. The comprehensive evaluation matrix integrates the operating status information of the two types of equipment to form a unified evaluation system.

[0074] The reasoning rule establishment process is based on the data correlation in the comprehensive evaluation matrix and expert experience knowledge. The transformer fatigue reasoning rule establishes condition judgment logic for the three core parameters of temperature rise, load and insulation. A typical rule is "If the temperature rise belongs to moderate fatigue, the load belongs to severe fatigue and the insulation belongs to slight fatigue, then the overall state of the transformer belongs to moderate fatigue". The switchgear fatigue reasoning rule builds reasoning logic around the three key parameters of operation number, contact resistance and mechanical characteristics. A representative rule is "If the operation number belongs to dangerous fatigue and the contact resistance belongs to severe fatigue, then the overall state of the switchgear belongs to dangerous fatigue". The reasoning rule library adopts a hierarchical structure design, including parameter-level rules, equipment-level rules and system-level rules. Parameter-level rules handle the state judgment of a single parameter, equipment-level rules integrate multiple parameters to evaluate the equipment state, and system-level rules coordinate multiple devices to conduct overall state analysis. The rule library has scalability and maintainability, supporting the dynamic addition and modification of rules.

[0075] The comprehensive evaluation and processing utilizes a fuzzy inference rule base to intelligently analyze real-time collected equipment operation data. Transformer operation data includes the current temperature rise, load current, and insulation resistance values. After the data is input into the inference rule base, corresponding inference rules are triggered to perform state calculations. Switchgear operation data covers the latest operation count statistics, contact resistance measurements, and mechanical characteristic test results. The inference process employs fuzzy synthesis operations to integrate the conclusions of multiple rules. The synthesis operations use a combination of the maximum-minimum synthesis method and the weighted average synthesis method. The maximum-minimum synthesis method handles the synthesis of mutually exclusive rules, while the weighted average synthesis method handles the synthesis of continuous rules. The fuzzy evaluation results represent the degree of belonging of each device to different fatigue levels in the form of a membership vector. The evaluation results also include confidence information on the equipment status and uncertainty quantification indicators.

[0076] The comparative verification process verifies the accuracy and reliability of the fuzzy evaluation results through historical data analysis. The transformer's historical temperature rise curve records the temperature rise variation patterns of the equipment under different operating conditions. Historical curve analysis identifies the time points and magnitudes of abnormal temperature rise changes. The historical operation records of the switchgear contain all operation actions and fault events since the equipment was put into operation. The operation record analysis statistically analyzes the equipment's fault frequency and fault type distribution. The comparative verification process compares and analyzes the current fuzzy evaluation results with historical data from the same period. The verification process calculates the degree of agreement and deviation between the evaluation results and historical trends. Evaluation results with high agreement maintain the original model parameters, while evaluation results with large deviations trigger the model correction mechanism. The correction mechanism adjusts the weight coefficients and membership function parameters of the inference rules. The optimization process improves the accuracy of the fuzzy feature analysis model based on the feedback of the verification results. The corrected and optimized model has stronger adaptability and higher prediction accuracy.

[0077] In one specific embodiment, the process of executing step S104 may specifically include the following steps:

[0078] Data alignment processing is performed on the photovoltaic power output curve and the wind power generation curve based on the fuzzy feature vector of equipment operation status to obtain the time series data sequence of new energy power output.

[0079] Correlation analysis was performed on the time series data of new energy output and the load demand curve to obtain the coupling relationship matrix between new energy output and load changes;

[0080] The standardized deviation between the current power value and the moving average power value is calculated based on the coupling relationship matrix to obtain the value of the new energy power fluctuation anomaly factor.

[0081] Based on the numerical values ​​of abnormal factors in new energy power fluctuations, the stress cycle mode of equipment is identified and processed to obtain equipment stress distribution data caused by power mismatch.

[0082] The equipment stress distribution data is input into a local anomaly detection algorithm for anomaly point identification, resulting in local anomaly factors that reflect local operational anomalies of the equipment.

[0083] Specifically, new energy power generation data is precisely synchronized based on the time reference of the fuzzy feature vector of equipment operation status. The fuzzy feature vector of equipment operation status contains three dimensions of information: the membership value of the status level and the trend of change. The timestamp of the vector serves as the reference time for data alignment. The photovoltaic power output curve records the power output data of photovoltaic power generation equipment at different times. The curve data includes the power generation timestamp and meteorological condition information. The wind power generation curve covers the real-time power output of wind power generation equipment and wind condition parameters. The data alignment process uses a time interpolation algorithm to unify the power generation data of different sampling frequencies to the time reference of the equipment status vector. The interpolation algorithm calculates the power estimate of the intermediate time based on the power values ​​of adjacent time points. The alignment of photovoltaic power output data takes into account the nonlinear characteristics of light intensity changes, and the alignment of wind power generation data handles the randomness of wind speed fluctuations. The new energy power output time series data sequence integrates the comprehensive power output information of photovoltaic and wind power generation. The sequence data is arranged according to a uniform time interval to form a continuous power output time series.

[0084] Correlation analysis reveals the inherent relationship between the time-series data of renewable energy output and the load demand curve. The load demand curve reflects the temporal variation of electricity load within the substation's power supply area, encompassing the combined demand of industrial, commercial, and residential loads. The correlation analysis employs the Pearson correlation coefficient method to quantify the linear correlation strength between renewable energy output and load demand. The analysis process groups and statistically analyzes the time-series data according to different time scales such as hourly, daily, weekly, and monthly. Hourly-scale analysis identifies the matching relationship between intraday power supply and demand, daily-scale analysis reveals the supply and demand differences between weekdays and rest days, and monthly-scale analysis reflects the impact of seasonal changes on the supply and demand balance. The coupling relationship matrix records the correlation coefficients at different time scales in matrix form. The matrix rows represent the time periods of renewable energy output, and the columns represent the time periods of load demand. The values ​​of the matrix elements represent the coupling strength between the corresponding time periods; a positive correlation indicates synchronous changes in supply and demand, while a negative correlation indicates inverse changes in supply and demand.

[0085] The standardized deviation calculation quantifies the degree of power fluctuation anomaly based on the statistical characteristics in the coupling relationship matrix. The current power value is obtained through real-time monitoring, including the total output of new energy sources and the total load demand. The moving average power value is calculated using the time window method to calculate the power average over a certain period of time. The time window length is set according to the power change characteristics of new energy sources, using a 15-minute window for photovoltaic power generation and a 30-minute window for wind power generation. The standardized deviation calculation formula divides the difference between the current power value and the moving average by the historical standard deviation of the power. The calculation process eliminates the influence of power magnitude differences and enables the comparability of equipment with different capacities. The value of the new energy power fluctuation anomaly factor reflects the degree to which the current power status deviates from the normal operating mode. The larger the absolute value of the anomaly factor, the more severe the power fluctuation. A positive value indicates that the power is above the average level, and a negative value indicates that the power is below the average level. The calculation of the anomaly factor also considers the seasonality and weather-related effects of new energy power generation.

[0086] Stress cycle pattern recognition and processing analysis analyzes the impact of abnormal factors in new energy power fluctuations on the mechanical and electrical stresses of substation equipment. The equipment stress cycle pattern describes the load change law of the equipment under power fluctuation conditions. During the power rise phase, the equipment bears positive stress loads, and during the power fall phase, the equipment bears negative stress loads. The amplitude, frequency, and duration of the stress cycle determine the fatigue accumulation rate of the equipment. The recognition and processing uses the rainflow counting method to statistically analyze the amplitude and frequency distribution of stress cycles. The rainflow counting method decomposes the complex stress time history into several independent stress cycles. Each stress cycle has a clear start point, end point, and amplitude characteristics. Power mismatch leads to overload stress when supply exceeds demand and underload stress when supply falls short of demand. The equipment stress distribution data records the stress level of each device under different power fluctuation conditions. The distribution data includes the load stress of transformers, the operating stress of switchgear, and the action stress of protection devices.

[0087] The local anomaly detection algorithm identifies anomalous concentration points and anomalous change points in equipment stress distribution data. Under normal circumstances, equipment stress distribution data exhibits a uniform distribution or a distribution that conforms to historical statistical patterns. Local anomalies manifest as stress levels in certain equipment or at certain times significantly deviating from the normal distribution range. The anomaly detection algorithm uses a density-based clustering method to identify outliers in the stress data. The algorithm calculates the data density in the neighborhood surrounding each data point. Data points with a density significantly lower than that of the neighboring area are identified as anomalies. Anomaly identification also considers the influence of equipment type and installation location. Abnormal stress in transformers may originate from load changes or internal faults, while abnormal stress in switchgear may be due to frequent operation or mechanical jamming. The local anomaly factor quantifies the degree to which anomalies deviate from the normal distribution; the larger the factor value, the more severe the anomaly. The local anomaly factor provides key anomaly feature inputs for subsequent neural network training.

[0088] In one specific embodiment, the process of executing step S105 may specifically include the following steps:

[0089] The historical normal operation status data is labeled based on local anomaly factors to obtain a training dataset containing normal and abnormal status identifiers.

[0090] The training dataset is processed into a BP neural network structure consisting of an input layer, a hidden layer, and an output layer to obtain a three-layer neural network architecture and initial values ​​for node connection weights.

[0091] Based on the neural network architecture, forward propagation calculations are performed on the local anomaly factor data to obtain the error function between the network output predicted value and the actual label value.

[0092] The backpropagation algorithm is used to adjust the weights based on the error function, resulting in iteratively optimized neural network weight parameters.

[0093] The neural network weight parameters are used to perform state prediction processing on the current device operating data to obtain the numerical deviation between the current operating state and the normal baseline state.

[0094] The fatigue quantification is performed based on the numerical deviation value and the equipment fatigue accumulation calculation formula to obtain the equipment fatigue accumulation monitoring results, which include the fatigue accumulation percentage and remaining life assessment.

[0095] Specifically, historical operating status data is intelligently classified and labeled based on the numerical characteristics of local anomaly factors. The historical normal operating status data covers the continuous operating records of substation equipment over the past three years. The data includes the monitoring values ​​of operating parameters of each device and the corresponding values ​​of local anomaly factors. The labeling process uses a combination of threshold discrimination and expert experience to determine the status labels. Data with local anomaly factor values ​​less than 1.0 are labeled as normal, values ​​between 1.0 and 2.5 are labeled as slightly abnormal, and values ​​greater than 2.5 are labeled as severely abnormal. The labeling process is also verified by combining actual equipment fault records and maintenance history to ensure the accuracy and consistency of labeling. The normal status is represented by the value zero, and the abnormal status is represented by the value one. The training dataset is grouped and organized according to time order and equipment type. The dataset contains an input feature matrix and a target label vector. The input feature matrix contains local anomaly factors, power fluctuation anomaly factors, and equipment operating parameters. The target label vector corresponds to the equipment status classification results at each time point.

[0096] The BP neural network architecture is designed based on the feature dimensions of the training dataset and the requirements of the classification task. The number of nodes in the input layer is equal to the number of columns in the feature matrix, containing multiple input variables such as local anomaly factors, power fluctuation anomaly factors, equipment temperature, and equipment current. The hidden layer adopts a single-layer structure with twice the number of nodes in the input layer to enhance the network's nonlinear fitting ability. The number of nodes in the output layer is equal to the number of classification categories, using binary classification to output normal and abnormal states. The network architecture uses a fully connected approach, with each node establishing connections with all nodes in adjacent layers. The initial values ​​of the node connection weights are randomly initialized, with weight values ​​evenly distributed between -0.5 and +0.5. The initial value of the bias parameter is set to 0.1. The activation function is the Sigmoid function in the hidden layer and the Softmax function in the output layer. The three-layer neural network architecture forms a complete mapping relationship from input features to state classification. The network structure has sufficient complexity to capture the nonlinear correlation between new energy power fluctuations and equipment states.

[0097] Forward propagation computation processes local anomaly factor data through a neural network architecture, propagating the computation from the input layer to the output layer. The input layer receives local anomaly factors and related feature data and passes them to the hidden layer. The hidden layer calculates the weighted input value of each node and generates the node output value through the sigmoid activation function. The output layer receives the output value of the hidden layer and calculates the final network prediction result. The network output prediction value represents the probability that the device state is abnormal. The actual label value comes from the annotation results in the training dataset. The error function uses the mean squared error method to quantify the difference between the predicted value and the actual value. The smaller the error function value, the more accurate the network prediction. The forward propagation process also records the output value and activation value of each layer node to provide the computation basis for subsequent backpropagation. The computation process uses batch processing to process multiple training samples simultaneously to improve computational efficiency.

[0098] The backpropagation algorithm optimizes and adjusts network weights based on gradient information from the error function. The algorithm propagates the error gradient layer by layer from the output layer to the input layer. The output layer error gradient is calculated using the partial derivative of the error function with respect to the output value. The hidden layer error gradient is calculated by backpropagating the error from the output layer using the chain rule. Weight adjustment employs gradient descent to update weight values ​​based on the error gradient and the learning rate. The learning rate is set to 0.01 to ensure network convergence stability. The weight update formula is: new weight equals old weight minus the learning rate multiplied by the error gradient. The iterative optimization process repeats forward and backward propagation until the error function converges to a preset threshold. The convergence condition is set to an error change of less than 0.001 for ten consecutive iterations. The iteratively optimized neural network weight parameters possess the ability to accurately identify the status of equipment in new energy substations. The optimized weight parameters are saved as a model file for subsequent state prediction.

[0099] The state prediction processing utilizes the weight parameters of a trained neural network to perform real-time state assessment of the current equipment operating data. The current equipment operating data includes the latest collected local anomaly factors, power fluctuation anomaly factors, and various equipment operating parameters. The prediction processing inputs the current data into the trained neural network for forward computation, and the network outputs the probability value that the current equipment state is abnormal. The normal baseline state is determined by the statistical average of historical normal operating data. The numerical deviation value is calculated as the difference between the current state probability and the normal baseline probability. A positive deviation value indicates that the equipment state is biased towards an abnormal direction, and a negative deviation value indicates that the equipment state is better than normal. The state prediction also provides confidence information of the prediction results, reflecting the reliability of the prediction. The prediction processing is real-time and can continuously update the equipment state assessment results as new data is input.

[0100] The fatigue quantification process converts numerical deviation values ​​into intuitive equipment fatigue accumulation indicators. The equipment fatigue accumulation calculation takes into account the accelerated impact of new energy power fluctuations on equipment lifespan. The calculation process uses the linear cumulative damage theory to convert the stress cycles borne by the equipment into a fatigue damage ratio. The fatigue accumulation percentage represents the proportion of the current fatigue level of the equipment to the designed fatigue life. The higher the percentage value, the closer the equipment is to fatigue failure. The remaining life assessment calculates the expected service life of the equipment based on the current fatigue accumulation rate and remaining fatigue capacity. The assessment results are expressed in months as the expected remaining service time of the equipment under the current operating mode. The equipment fatigue accumulation monitoring results also include fatigue accumulation trend analysis and maintenance recommendations. The trend analysis predicts the development direction of fatigue accumulation in the future, and the maintenance recommendations provide corresponding maintenance strategies and schedules based on the degree of fatigue accumulation.

[0101] The above describes the operation monitoring method for new energy intelligent prefabricated substations in the embodiments of this application. The following describes the operation monitoring system for new energy intelligent prefabricated substations in the embodiments of this application. Please refer to [link / reference]. Figure 2 One embodiment of the new energy intelligent prefabricated substation operation monitoring system in this application includes:

[0102] The partitioning module is used to functionally partition the space of the prefabricated substation according to the new energy access capacity, and obtain the power density distribution parameters of the transformer area, switchgear area, and control room area.

[0103] The noise reduction module is used to perform noise reduction processing on the collected electrical parameters through wavelet transform to obtain purified operation data that eliminates power fluctuation interference;

[0104] A module is established to build a fuzzy feature analysis model based on the purification operation data to obtain the fuzzy feature vector of the equipment operating status;

[0105] The calculation module is used to calculate the new energy output-load coupling characteristics based on the fuzzy feature vector, and obtain the new energy power fluctuation anomaly factor and local anomaly factor;

[0106] The monitoring module is used to train the local abnormal factors through a BP neural network to obtain the deviation value between the current operating state and the normal reference state and the equipment fatigue accumulation monitoring results.

[0107] above Figure 2 The operation monitoring system of the intelligent prefabricated substation of the new energy in this embodiment of the invention is described in detail from the perspective of modular functional entities. The operation monitoring equipment of the intelligent prefabricated substation of the new energy in this embodiment of the invention is described in detail from the perspective of hardware processing.

[0108] Reference Figure 3 This invention also provides a new energy intelligent prefabricated substation operation monitoring device, which can be a server, and its internal structure can be as follows: Figure 3 As shown, the new energy intelligent prefabricated substation operation monitoring equipment includes a processor, memory, display screen, input device, network interface, and database connected via a system bus. The processor in this computer design provides computing and control capabilities. The memory of the new energy intelligent prefabricated substation operation monitoring equipment includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database of the new energy intelligent prefabricated substation operation monitoring equipment is used to store the data corresponding to this embodiment. The network interface of the new energy intelligent prefabricated substation operation monitoring equipment is used for communication with external terminals via network connection. When the computer program is executed by the processor, it implements the above-described method.

[0109] Those skilled in the art will understand that Figure 3 The structure shown is merely a block diagram of a portion of the structure related to the present invention and does not constitute a limitation on the operation monitoring equipment of the new energy intelligent box-type substation to which the present invention is applied.

[0110] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when the instructions are executed on a computer, cause the computer to perform the steps of the new energy intelligent box-type substation operation monitoring method.

[0111] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0112] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a new energy intelligent prefabricated substation operation monitoring device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0113] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for monitoring the operation of a new energy intelligent prefabricated substation, characterized in that, The method includes: Based on the new energy access capacity, the space of the prefabricated substation is functionally divided to obtain the power density distribution parameters of the transformer area, switchgear area, and control room area; The collected electrical parameters are denoised by wavelet transform to obtain purified operation data that eliminates power fluctuation interference. A fuzzy feature analysis model is established based on the purification operation data to obtain the fuzzy feature vector of the equipment operating status; The calculation of renewable energy output-load coupling characteristics based on the fuzzy feature vector yields renewable energy power fluctuation anomaly factors and local anomaly factors. This includes: aligning the photovoltaic output curve and wind power curve based on the fuzzy feature vector of the equipment operating state to obtain a renewable energy output time-series data sequence; performing correlation analysis between the renewable energy output time-series data sequence and the load demand curve to obtain a coupling relationship matrix between renewable energy output and load changes; calculating the standardized deviation between the current power value and the moving average power value based on the coupling relationship matrix to obtain a renewable energy power fluctuation anomaly factor value; identifying the equipment stress cycle mode based on the renewable energy power fluctuation anomaly factor value to obtain equipment stress distribution data caused by power mismatch; and inputting the equipment stress distribution data into a local anomaly detection algorithm for anomaly point identification to obtain local anomaly factors reflecting local equipment operating anomalies. The local anomaly factors are trained using a BP neural network to obtain the deviation between the current operating state and the normal baseline state, as well as the equipment fatigue accumulation monitoring results.

2. The method for monitoring the operation of a new energy intelligent prefabricated substation according to claim 1, characterized in that, The functional division of the prefabricated substation space based on the new energy access capacity yields power density distribution parameters for the transformer area, switchgear area, and control room area, including: Based on historical output data of photovoltaic and wind power generation, power fluctuation amplitude-frequency distribution modeling is performed to obtain the new energy power transmission path and equipment carrying capacity coefficient; The internal space of the prefabricated substation is divided into zones based on the equipment load-bearing capacity coefficient, resulting in functional zoning boundaries for the transformer zone, switchgear zone, control room zone, and cable interlayer zone. The power density distribution algorithm is used to calculate the power impact intensity of the functional zone boundary to obtain the power impact intensity coefficient of each monitoring area. Based on the power impact intensity coefficient and the importance of the equipment, the sensor density configuration is processed to obtain a sensor deployment density of four sensors per square meter in the transformer area, three sensors per square meter in the switch cabinet area, and two sensors per square meter in the control room area. The location of monitoring points is optimized based on the sensor deployment density to obtain the sensor coordinate position matrix, monitoring parameter configuration table, and data acquisition frequency setting value.

3. The method for monitoring the operation of a new energy intelligent prefabricated substation according to claim 1, characterized in that, The process of denoising the collected electrical parameters using wavelet transform to obtain purified operation data free from power fluctuation interference includes: Based on the power density distribution parameters, the voltage, current, power, temperature, and vibration parameters in the substation are collected and processed in real time to obtain the original electrical parameter data under the new energy operation conditions. The original electrical parameter data is input into a timestamp synchronization algorithm for time alignment processing to obtain a synchronized electrical parameter sequence with a unified time base; Based on the synchronous electrical parameter sequence, a multi-scale wavelet decomposition algorithm is used to perform frequency domain decomposition processing to obtain the wavelet coefficient distribution of the new energy power fluctuation frequency band. Based on the wavelet coefficient distribution, an adaptive filtering process is performed on the power oscillation interference caused by intermittent power generation from new energy sources to obtain a reconstructed electrical signal with power fluctuation noise removed. The reconstructed electrical signals are encapsulated and processed according to the new energy monitoring time sequence format to obtain purification operation data containing equipment identification, parameter type and value.

4. The method for monitoring the operation of a new energy intelligent prefabricated substation according to claim 1, characterized in that, The step of establishing a fuzzy feature analysis model based on the purification operation data to obtain a fuzzy feature vector of the equipment operating status includes: Based on the purification operation data, feature extraction processing is performed on the operating parameters of transformers and switching equipment to obtain a multi-dimensional set of operating features reflecting changes in equipment status. The multidimensional operating characteristic set is classified into state levels according to slight fatigue, moderate fatigue, severe fatigue, dangerous fatigue and extreme fatigue to obtain the equipment fatigue level classification standard; Based on the equipment fatigue level classification standard, the operating parameters are fuzzy mapped to obtain the fuzzy membership degree values ​​corresponding to each equipment state level. Based on the fuzzy membership degree values, fuzzy inference rules are used to perform feature fusion processing to obtain a fuzzy feature analysis model that comprehensively reflects the operating status of the equipment. The fuzzy feature analysis model is used to quantitatively evaluate the current equipment status, resulting in a fuzzy feature vector of equipment operating status that includes status level, membership degree value, and trend of change.

5. The method for monitoring the operation of a new energy intelligent prefabricated substation according to claim 4, characterized in that, The feature fusion processing based on the fuzzy membership degree values ​​using fuzzy inference rules yields a fuzzy feature analysis model that comprehensively reflects the equipment's operating status, including: Based on the fuzzy membership values, the transformer temperature rise parameter, transformer load current parameter, and transformer insulation resistance parameter are weighted and assigned to obtain the weight assignment matrix of the transformer operating status. The weight allocation matrix is ​​fused with the switching equipment operation frequency parameters, switching equipment contact resistance parameters, and switching equipment mechanical characteristic parameters to obtain a comprehensive evaluation matrix of the switching equipment operating status. Based on the comprehensive evaluation matrix, transformer fatigue reasoning rules and switchgear fatigue reasoning rules are established to obtain a fuzzy reasoning rule base for equipment status judgment. Based on the fuzzy inference rule base, the transformer operation data and switchgear operation data are comprehensively evaluated and processed to obtain the fuzzy evaluation results of the overall operation status of each device. The fuzzy evaluation results are compared and verified with the historical temperature rise curve of the transformer and the historical operation records of the switching equipment to obtain a corrected and optimized fuzzy feature analysis model.

6. The method for monitoring the operation of a new energy intelligent prefabricated substation according to claim 1, characterized in that, The step of training the local anomaly factors using a BP neural network to obtain the deviation between the current operating state and the normal baseline state, and the equipment fatigue accumulation monitoring results, includes: Based on the local anomaly factors, the historical normal operation status data is labeled to obtain a training dataset containing normal and anomaly status identifiers. The training dataset is processed into a BP neural network structure consisting of an input layer, a hidden layer, and an output layer to obtain a three-layer neural network architecture and initial values ​​for node connection weights. Based on the neural network architecture, forward propagation calculations are performed on the local anomaly factor data to obtain the error function between the network output predicted value and the actual label value. Based on the error function, the backpropagation algorithm is used to adjust the weights, resulting in iteratively optimized neural network weight parameters. The neural network weight parameters are used to perform state prediction processing on the current device operating data to obtain the numerical deviation value between the current operating state and the normal baseline state. The fatigue quantification is performed based on the numerical deviation value and the equipment fatigue accumulation calculation formula to obtain the equipment fatigue accumulation monitoring results, which include the fatigue accumulation percentage and remaining life assessment.

7. A new energy intelligent prefabricated substation operation monitoring system, characterized in that, For implementing the new energy intelligent prefabricated substation operation monitoring method as described in any one of claims 1-6, the new energy intelligent prefabricated substation operation monitoring system comprises: The partitioning module is used to functionally partition the space of the prefabricated substation according to the new energy access capacity, and obtain the power density distribution parameters of the transformer area, switchgear area, and control room area. The noise reduction module is used to perform noise reduction processing on the collected electrical parameters through wavelet transform to obtain purified operation data that eliminates power fluctuation interference; A module is established to build a fuzzy feature analysis model based on the purification operation data to obtain the fuzzy feature vector of the equipment operating status; The calculation module is used to calculate the new energy output-load coupling characteristics based on the fuzzy feature vector, and obtain new energy power fluctuation anomaly factors and local anomaly factors. This includes: performing data alignment processing on the photovoltaic output curve and wind power generation curve based on the fuzzy feature vector of the equipment operating state to obtain a new energy output time-series data sequence; performing correlation analysis processing on the new energy output time-series data sequence and the load demand curve to obtain a coupling relationship matrix between new energy output and load changes; calculating the standardized deviation between the current power value and the moving average power value based on the coupling relationship matrix to obtain a new energy power fluctuation anomaly factor value; identifying the equipment stress cycle mode based on the new energy power fluctuation anomaly factor value to obtain equipment stress distribution data caused by power mismatch; and inputting the equipment stress distribution data into a local anomaly detection algorithm for anomaly point identification processing to obtain local anomaly factors reflecting local equipment operating anomalies. The monitoring module is used to train the local abnormal factors through a BP neural network to obtain the deviation value between the current operating state and the normal reference state and the equipment fatigue accumulation monitoring results.

8. A new energy intelligent box-type substation operation monitoring device, characterized in that, The method includes a memory and a processor, wherein the memory stores a computer program that can run on the processor, and the processor executes the computer program to implement the new energy intelligent box-type substation operation monitoring method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is run by the processor, it causes the processor to execute the new energy intelligent box-type substation operation monitoring method as described in any one of claims 1 to 6.

Images