Capacitor capacity abnormality dynamic monitoring method based on edge computing

Abstract

The application relates to the technical field of power system state monitoring and intelligent operation and maintenance, and discloses a capacitor capacity abnormality dynamic monitoring method based on edge computing. By connecting an edge monitoring unit in parallel and connecting a current-limiting resistor in series in a target capacitor loop, a unified voltage sampling period and a time stamp are set, safe and controllable discharging and accurate timing recording are realized. The edge node continuously collects the capacitor voltage, constructs full-sequence original data, and aligns the data point by point through a theoretical discharge model and a sampling value to form a capacity error evaluation function. The system optimizes in the capacity search range, judges the convergence and error distribution stability, and automatically outputs the capacity abnormality judgment without manual intervention. All monitoring results are stored synchronously with a unique identifier and a time stamp, a sliding window algorithm fuses historical data in real time, a capacity continuous degradation trend is captured, and the real-time performance, accuracy and early warning capability of monitoring are improved.

Description

Technical Field

[0001] This invention relates to the field of power system condition monitoring and intelligent operation and maintenance technology, specifically to a method for dynamic monitoring of capacitor capacity anomalies based on edge computing. Background Technology

[0002] With the continuous expansion of power system scale and the improvement of its intelligence level, a large number of parallel capacitors are widely used in scenarios such as power quality management and reactive power compensation. The accuracy of capacitor capacity is directly related to the reactive power balance and equipment safety of the distribution system. However, during long-term operation, capacitors are prone to capacity decay or even failure due to dielectric aging, overvoltage impacts, partial discharge, etc., resulting in a deviation between their actual capacity and nominal value, which in turn affects the stable operation of the power system. Therefore, timely and accurate detection and monitoring of capacitor capacity anomalies is of great significance.

[0003] Current technologies primarily rely on manual periodic inspections, offline testing instruments, or back-end big data statistical analysis to detect capacitor capacity. For example, some substations use portable capacity testers for periodic spot checks, requiring power outages and wire disconnection for manual measurement, resulting in long cycles and a high risk of missed detections. Some systems attempt to estimate capacitor operating status based on voltage and current data collected from the main station, using static models or empirical criteria; however, the results are highly susceptible to load fluctuations and grid disturbances, limiting accuracy and real-time performance. In recent years, some systems have utilized communication networks to transmit raw data back to a back-end server, processing capacity information with centralized algorithms. However, this approach faces practical problems such as bandwidth bottlenecks, network latency, and centralized storage of massive datasets, making it difficult to meet the needs of distributed large-scale capacitor monitoring. Furthermore, traditional methods often neglect dynamic modeling and real-time parameter identification of the equipment's physical characteristics, relying solely on single-point data or static rules, lacking specificity and failing to achieve early detection and trend tracking of capacity anomalies. Meanwhile, due to the complex on-site conditions and numerous environmental interferences, existing detection methods have uncertainties in data acquisition, boundary criterion setting, and capacity change curve analysis, leading to frequent misjudgments and omissions, which seriously affect the operation and maintenance efficiency and safety level of the power system.

[0004] Therefore, this case aims to propose a dynamic monitoring method for capacitor capacity anomalies based on edge computing. It obtains high-resolution voltage sequences by pre-setting sampling periods, and then generates corresponding theoretical voltage sequences by combining them with RC discharge theoretical models. The capacity is estimated by residual sum of squares function, and the method utilizes two mechanisms, residual convergence judgment and sliding window trend analysis, to achieve accurate, real-time, and low-latency capacity anomaly detection and degradation early warning. Summary of the Invention

[0005] This invention provides a method for dynamic monitoring of capacitor capacity anomalies based on edge computing, which helps to solve the problems mentioned in the background art.

[0006] This invention provides the following technical solution: a method for dynamic monitoring of capacitor capacitance anomalies based on edge computing, comprising:

[0007] In the electrical circuit where the target capacitor is located, an edge monitoring unit is connected in parallel and a current-limiting resistor is connected in series. A preset voltage sampling period is set and the start timestamp is recorded to obtain the initial voltage value and nominal capacitance value at the relative time zero point.

[0008] Starting from the zero point of relative time, the voltage across the target capacitor is collected according to the preset voltage sampling period to form an original voltage sequence data set;

[0009] A theoretical discharge voltage model is established with the initial voltage value across the target capacitor, the resistance value of the current limiting resistor, and the candidate capacitance value as parameters. The theoretical voltage at each sampling time is calculated, and a set of theoretical voltage sequence data corresponding one-to-one with the original sequence is generated.

[0010] For each candidate capacitance value, the residual between the measured voltage and the corresponding theoretical voltage is calculated and the sum of squares is obtained to obtain the capacitance error evaluation function with the capacitance value as the independent variable.

[0011] Within the preset capacity search range, the capacity error evaluation function is calculated, and the one with the smallest total residual sum of squares is selected as the optimal capacity estimate. The convergence and residual distribution stability are determined by comparing the difference between the minimum and the second smallest value with the preset residual difference threshold.

[0012] When convergence and stability are achieved, calculate the absolute deviation of the capacity from the nominal capacitance value and the ratio of the relative error of the capacity, and compare it with the threshold of the relative error of the capacity, and output whether the capacity is abnormal or normal.

[0013] The capacity monitoring results record is generated using the optimal capacity estimate, the relative capacity error ratio, and the capacity status marker. It is then written into the edge node database along with the unique identifier of the target capacitor and the start timestamp.

[0014] Set the sliding window length, select the most recent optimal capacity estimates within the window, calculate the moving average capacity value and the ratio of the sliding capacity error, compare it with the capacity relative error ratio threshold, identify the continuous degradation trend, and output a capacity degradation early warning flag.

[0015] Optionally, the step of connecting an edge monitoring unit in parallel and a current-limiting resistor in series in the electrical circuit containing the target capacitor, setting a preset voltage sampling period and recording the start timestamp, and obtaining the initial voltage value and nominal capacitance value at the relative time zero point specifically includes:

[0016] An edge monitoring unit is connected in parallel across the target capacitor. The edge monitoring unit has voltage acquisition function and local computing processing capability, and each edge monitoring unit is assigned a unique number.

[0017] A current-limiting resistor is connected in series in the electrical circuit where the target capacitor is located. The resistance value of the current-limiting resistor is preset and recorded, and the resistance value of the current-limiting resistor is used as the fixed resistance parameter in the discharge circuit.

[0018] Configure the voltage sampling period in the edge monitoring unit and set the voltage sampling period to a preset voltage sampling period.

[0019] Before performing the target capacitor discharge test, record the actual calendar time at that time, and record the actual calendar time as the discharge test start timestamp, and define the discharge start time as the relative time zero point.

[0020] At the instant the target capacitor begins to discharge, the initial voltage value across the target capacitor is acquired by the edge monitoring unit and recorded as the initial voltage value at the relative time zero point.

[0021] Obtain the nominal capacitance value of the target capacitor from its nameplate or manufacturer's technical data, and use the nominal capacitance value as a reference capacitance parameter for this capacitance identification and anomaly determination process.

[0022] Optionally, the step of acquiring the voltage across the target capacitor at a preset voltage sampling period starting from the relative time zero point to form an original voltage sequence data set specifically includes:

[0023] During the discharge process of the target capacitor, continuous and equally spaced sampling is performed starting from the zero point of relative time according to the preset voltage sampling period to obtain multiple sampling moments arranged in chronological order. The total number of time and voltage sampling point pairs is not less than two.

[0024] At each sampling moment, the measured voltage across the target capacitor is obtained by the edge monitoring unit, and each sampling moment and its corresponding measured voltage value are combined into a time-voltage sampling point pair.

[0025] All time and voltage sampling point pairs are stored sequentially into the data set according to time order, forming the original voltage sequence data set describing the discharge process of the target capacitor.

[0026] Optionally, the step of establishing a theoretical discharge voltage model with parameters including the initial voltage value across the target capacitor, the resistance value of the current-limiting resistor, and the candidate capacitance value, calculating the theoretical voltage at each sampling time, and generating a theoretical voltage sequence data set corresponding one-to-one with the original sequence specifically includes:

[0027] Based on the discharge circuit consisting of the target capacitor and the current-limiting resistor connected in series, a theoretical discharge voltage model is constructed using the exponential decay characteristic of the series discharge of the resistor and capacitor. The theoretical discharge voltage model takes the initial voltage value across the target capacitor, the resistance value of the current-limiting resistor, and the candidate capacitance value as input parameters, and outputs the theoretical voltage value across the target capacitor under a given time condition.

[0028] For each sampling time recorded in the original voltage sequence data set, under the same candidate capacitance value, the theoretical discharge voltage model is called to calculate the theoretical voltage value across the target capacitor corresponding to the sampling time.

[0029] According to the time order of the original voltage sequence data set, the theoretical voltage values ​​corresponding to each sampling time are arranged sequentially to generate a theoretical voltage sequence data set that maintains a one-to-one correspondence with the original voltage sequence data set at the sampling time.

[0030] Optionally, the step of calculating the residual between the measured voltage and the corresponding theoretical voltage for each candidate capacitance value and summing the squares to obtain a capacitance error evaluation function with capacitance as the independent variable specifically includes:

[0031] Under each candidate capacitance value condition, for each time and voltage sampling point pair in the original voltage sequence data set, calculate the difference between the measured voltage value in the corresponding time and voltage sampling point pair and the theoretical voltage value in the corresponding time and voltage sampling point pair in the theoretical voltage sequence data set, and use the difference between the measured voltage value and the theoretical voltage value as the single-point residual of the corresponding time and voltage sampling point pair;

[0032] Under the same candidate capacitance value, the single-point residuals of all time and voltage sampling point pairs are squared, and the summation operation is performed over the entire range of all time and voltage sampling point pairs to obtain the total residual sum of squares for the current candidate capacitance value;

[0033] Within the entire range of candidate capacitance values, a capacity error evaluation function is formed, with the candidate capacitance value as the independent variable and the corresponding total residual sum of squares as the function value. The capacity error evaluation function only changes with the candidate capacitance value, while the other parameters remain unchanged during the capacity identification process.

[0034] Optionally, the step of calculating a capacity error evaluation function within a preset capacity search range, selecting the smallest total residual sum of squares as the optimal capacity estimate, and comparing the difference between the minimum and second smallest values ​​with a preset residual difference threshold to determine convergence and residual distribution stability, specifically includes:

[0035] Based on the nominal capacitance value of the target capacitor, the lower limit of the capacity search is set to half of the nominal capacitance value, and the upper limit of the capacity search is set to the nominal capacitance value. Multiple discrete capacity points are divided between the lower limit and the upper limit of the capacity search according to a preset capacity step size, and a capacity search set consisting of multiple candidate capacitance values ​​is constructed.

[0036] For each candidate capacitance value in the capacity search set, the capacity error evaluation function is called to calculate the total residual sum of squares corresponding to each candidate capacitance value, and the correspondence between the candidate capacitance value and the corresponding total residual sum of squares is recorded.

[0037] Find the candidate capacitor value with the smallest total residual square sum in the capacity search set, take the candidate capacitor value with the smallest total residual square sum as the optimal capacity estimate for this discharge process, and record the corresponding minimum total residual square sum as the current minimum total residual square sum;

[0038] After excluding the candidate capacitor value with the smallest total residual sum of squares, find the candidate capacitor value with the second smallest total residual sum of squares in the set of remaining candidate capacitor values, and record the total residual sum of squares corresponding to the second smallest candidate capacitor value as the current second smallest total residual sum of squares;

[0039] Calculate the absolute value of the difference between the current second smallest total residual sum of squares and the current smallest total residual sum of squares. Compare the absolute value of the difference with a preset residual difference threshold. The physical quantity unit of the preset residual difference threshold is voltage square, and its numerical value is on the order of one millionth. When the difference is less than the preset residual difference threshold, it is determined that the optimal capacity estimate meets the convergence condition and the residual distribution is stable. When the difference is greater than or equal to the preset residual difference threshold, it is determined that the optimal capacity estimate does not meet the stability requirement.

[0040] Optionally, the step of calculating the absolute deviation of the capacitance value from the nominal capacitance value and the ratio of the relative capacitance error when the capacitance is converged and stable, and comparing it with the relative capacitance error ratio threshold, and outputting whether the capacitance is abnormal or normal, specifically includes:

[0041] After the optimal capacity estimate is obtained and the convergence condition is met, the absolute value of the difference between the optimal capacity estimate and the nominal capacitance value is calculated to obtain the absolute capacity deviation.

[0042] The ratio of absolute capacitance deviation to nominal capacitance value is used as the relative capacitance error ratio. The relative capacitance error ratio is obtained by dividing the absolute capacitance deviation by the nominal capacitance value.

[0043] The preset capacity relative error ratio threshold is set to 0.1, which corresponds to 10% in percentage form.

[0044] The relative error ratio of the capacity is compared with the relative error ratio threshold of the capacity. When the relative error ratio of the capacity is greater than the relative error ratio threshold of the capacity, the capacity status of the target capacitor is judged as abnormal. When the relative error ratio of the capacity is less than or equal to the relative error ratio threshold of the capacity, the capacity status of the target capacitor is judged as normal.

[0045] Optionally, the step of generating a capacity monitoring result record using the optimal capacity estimate, the relative capacity error ratio, and the capacity status marker, and then writing it into the edge node database along with the unique identifier of the target capacitor and the start timestamp, specifically includes:

[0046] The capacity monitoring results record is constructed based on the optimal capacity estimate, the relative error ratio of capacity, and the capacity status marker during the current discharge process. The capacity status marker adopts binary encoding, with a value of 1 indicating an abnormal capacity state and a value of 0 indicating a normal capacity state.

[0047] Obtain the unique identifier of the target capacitor and the discharge test start time stamp. Use the unique identifier of the target capacitor and the discharge test start time stamp as index information, and combine them with the optimal capacity estimate, the relative capacity error ratio and the capacity status mark to form a capacity monitoring result record.

[0048] Capacity monitoring results are recorded and written to the edge node database and stored according to the index information, maintaining a structured set of capacity monitoring results records in the edge node database.

[0049] Optionally, the step of setting the sliding window length, selecting the most recent optimal capacity estimates within the window, calculating the moving average capacity value and the ratio of the sliding capacity error, comparing it with a capacity relative error ratio threshold, identifying a continuous degradation trend, and outputting a capacity degradation early warning flag specifically includes:

[0050] In the edge node database, the optimal capacity estimate corresponding to the most recent discharge tests of the same target capacitor is selected in chronological order. The sliding window length is preset and is greater than one. The most recent optimal capacity estimates are combined into a sliding window capacity estimate set.

[0051] In each degradation trend analysis, the optimal capacity estimate within the sliding window is calculated using the arithmetic mean to obtain the moving average capacity value;

[0052] Calculate the absolute value of the difference between the sliding average capacitance value and the nominal capacitance value to obtain the absolute deviation of the sliding capacitance, and use the ratio of the absolute deviation of the sliding capacitance to the nominal capacitance value as the sliding capacitance error ratio.

[0053] The sliding capacity error ratio is compared with the relative capacity error ratio threshold. When the sliding capacity error ratio is greater than the relative capacity error ratio threshold, it is determined that the target capacitor capacity has a continuous degradation trend and a capacity degradation warning mark is generated. When the sliding capacity error ratio is less than or equal to the relative capacity error ratio threshold, it is determined that the change in the target capacitor capacity is within the allowable range and no capacity degradation warning mark is generated.

[0054] The present invention has the following beneficial effects:

[0055] 1. An edge monitoring unit with local computing capabilities is connected in parallel in the electrical circuit containing the target capacitor, and a current-limiting resistor of known resistance is connected in series. The initial voltage and nominal capacitance value are obtained by synchronously acquiring voltage and clock stamps through a preset voltage sampling period. This miniaturizes and modularizes the functions of the instrument-grade testing equipment, directly embedding it into the operating circuit, avoiding contact errors and personnel safety risks associated with manual oscilloscope measurements on-site. Simultaneously, by synchronously recording the absolute clock and relative zero point, the basis for unified time calibration across multiple nodes and cross-test comparisons is achieved. Compared to traditional sampling modes, this method requires no downtime or wiring modifications, and the current-limiting resistor isolates surges, while the independent computing power of the edge nodes avoids data transmission bottlenecks.

[0056] 2. After discharge startup, the edge monitoring unit automatically and continuously collects the voltage across the capacitor according to a pre-set fixed sampling frequency, generating a time-ordered and evenly spaced raw voltage sequence data. Utilizing the embedded real-time clock and local cache of the edge device, the raw data is preprocessed locally, avoiding the communication overhead of high-frequency wireless and wired transmission while ensuring data real-time performance and integrity. Traditional technologies typically rely on power-off testing or periodic oscilloscope access, resulting in limited sampling frequency and massive data volumes—which then require queuing and processing after being sent to the cloud. This solution achieves a zero-latency closed loop from sampling to local storage.

[0057] 3. Based on the characteristics of a resistor-capacitor series discharge circuit, and combining the initial voltage, the resistance value of the current-limiting resistor, and the candidate capacitance values, an online exponential decay theoretical voltage model is constructed, and the corresponding theoretical voltage sequence is calculated for each sampling time. The solution of continuous differential equations in traditional simulation software is transformed into edge node kernel function calls, which calculate and generate theoretical sequences that correspond one-to-one with the original sequences in real time. Compared with existing technologies that use static table lookups or offline fitting in the cloud, localized model calling reduces reliance on cloud computing and can dynamically adjust model accuracy based on on-site errors.

[0058] 4. Under each candidate capacitance value, the residuals of the difference between the measured voltage sequence and the theoretical voltage sequence are calculated point by point and summed and squared to form an evaluation function with the capacitance value as the independent variable. Using the sum of squared residuals as the objective function, unlike traditional methods that only calculate single-shot errors, this method integrates information from the entire sequence, enhancing robustness to noise and transient fluctuations. Furthermore, edge nodes can compute multiple candidate values ​​in parallel, accelerating the overall search process. Unlike existing technologies that rely heavily on linear regression or single least squares methods for coarse estimation, this evaluation function fully utilizes all sampling points, improving estimation accuracy, avoiding error jitter caused by occasional fluctuations, and ensuring reliable convergence of the capacity estimation results under conditions of high noise and wide-range load variations.

[0059] 5. Within a predefined capacity search interval, a candidate set is generated with a uniform step size. For each candidate value, the aforementioned evaluation function is quickly calculated, and the value with the smallest sum of squared errors is selected as the optimal estimate. Simultaneously, the difference between the minimum and second-minimum evaluation function values ​​is compared with a threshold to determine convergence and residual distribution stability. A second-minimum difference metric is introduced as a stability criterion, compensating for the vulnerability of simply minimizing the objective function to local optima. Furthermore, this iterative search is parallelized at edge nodes and can be completed within tens of milliseconds. Unlike traditional methods that only compare error magnitude, this scheme directly judges the peak steepness of the function curve, effectively avoiding erroneous convergence caused by noise or model mismatch. While ensuring estimation speed, it improves the reliability of the results, providing a stable basis for the next step of capacity determination.

[0060] 6. When the optimal estimate meets the convergence condition, calculate the absolute deviation and relative error ratio between the estimated value and the nominal value, and compare them with a preset threshold to output a judgment of normal or abnormal capacity. By combining the relative error ratio threshold with physical meaning, considering both absolute capacity changes and relative amplitude differences, the tolerance for different application scenarios can be flexibly set. Furthermore, real-time local calculation at the edge eliminates the need for cloud-based parameter distribution and polling comparisons. Unlike common experience-based judgments or manual threshold settings, this solution can dynamically adjust the threshold based on historical data and can be customized online for different capacitor models. This achieves seamless integration from computational quantification to application decision-making, making anomaly judgments more objective and traceable, and improving on-site operation and maintenance efficiency.

[0061] 7. Combine the estimated value, relative error ratio, binary state marker, capacitor unique identifier, and timestamp into a structured tuple and write it to the local database. By combining the computing power and storage capacity of edge nodes, historical monitoring data can be continuously recorded in offline or offline environments, and rapid retrieval can be achieved through database indexing; it also supports subsequent batch uploads to the cloud or manual on-site access. Compared with traditional methods that only allow querying historical reports in the cloud, this solution reduces communication pressure and improves data security. It ensures data integrity and reliability and provides sufficient historical evidence for subsequent trend analysis and operation and maintenance decisions.

[0062] 8. Based on historical capacity estimation records, a sliding window strategy is used to calculate the moving average capacity value and error ratio, which is then compared with a relative error threshold to determine whether a continuous degradation trend exists and trigger an early warning. This extends single-point anomaly detection to multi-point trend fusion, avoiding misjudgments caused by occasional noise; simultaneously, the moving average is used to filter short-term fluctuations, making the early warning more focused on the true degradation trend. Compared to traditional manual periodic analysis or single-time monitoring, this solution can output trend early warnings in real time, improving preventative maintenance and extending equipment lifespan. It achieves a leap from current status alarms to future trend warnings, providing decision-makers with a more forward-looking operational perspective. Attached Figure Description

[0063] Figure 1 This is a schematic diagram of the process of the present invention. Detailed Implementation

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

[0065] Example, refer to Figure 1 A method for dynamic monitoring of capacitor capacitance anomalies based on edge computing, comprising:

[0066] In the electrical circuit where the target capacitor is located, an edge monitoring unit is connected in parallel and a current-limiting resistor is connected in series. A preset voltage sampling period is set and the start timestamp is recorded to obtain the initial voltage value and nominal capacitance value at the relative time zero point.

[0067] Starting from the zero point of relative time, the voltage across the target capacitor is collected according to the preset voltage sampling period to form an original voltage sequence data set;

[0068] A theoretical discharge voltage model is established with the initial voltage value across the target capacitor, the resistance value of the current limiting resistor, and the candidate capacitance value as parameters. The theoretical voltage at each sampling time is calculated, and a set of theoretical voltage sequence data corresponding one-to-one with the original sequence is generated.

[0069] For each candidate capacitance value, the residual between the measured voltage and the corresponding theoretical voltage is calculated and the sum of squares is obtained to obtain the capacitance error evaluation function with the capacitance value as the independent variable.

[0070] Within the preset capacity search range, the capacity error evaluation function is calculated, and the one with the smallest total residual sum of squares is selected as the optimal capacity estimate. The convergence and residual distribution stability are determined by comparing the difference between the minimum and the second smallest value with the preset residual difference threshold.

[0071] When convergence and stability are achieved, calculate the absolute deviation of the capacity from the nominal capacitance value and the ratio of the relative error of the capacity, and compare it with the threshold of the relative error of the capacity, and output whether the capacity is abnormal or normal.

[0072] The capacity monitoring results record is generated using the optimal capacity estimate, the relative capacity error ratio, and the capacity status marker. It is then written into the edge node database along with the unique identifier of the target capacitor and the start timestamp.

[0073] Set the sliding window length, select the most recent optimal capacity estimates within the window, calculate the moving average capacity value and the ratio of the sliding capacity error, compare it with the capacity relative error ratio threshold, identify the continuous degradation trend, and output a capacity degradation early warning flag.

[0074] By connecting the edge monitoring unit and the series current-limiting resistor in parallel within the same circuit, both measurement safety and controllable discharge conditions are ensured. The introduction of a preset voltage sampling period and recording the start timestamp solves the problems of clock asynchrony and data timing disorder at different monitoring nodes, making subsequent sequence alignment and model fitting more accurate. Compared to traditional methods that require power outages to disconnect wiring or sending the specimen to a laboratory for measurement, this solution eliminates the need for on-site power outages and manual measurement, improving monitoring efficiency and safety. Continuous voltage acquisition at preset periods forms a raw voltage sequence data set, replacing discontinuous or non-uniform sampling, ensuring the temporal integrity of the sequence data and laying the foundation for subsequent discrete modeling. By effectively combining the physical model with measured data through a one-to-one correspondence between the theoretical discharge voltage model and the raw sequence, model-driven error minimization is achieved. Compared to existing techniques that use simple attenuation rates or empirical formulas for rough estimation, this innovative approach incorporates the entire sequence residual into the evaluation function, improving estimation accuracy. Furthermore, by using a dual verification of convergence—the minimum residual sum of squares and the difference of the second smallest value—the problem of misjudgment caused by noise or local minima is solved, ensuring the robustness and reliability of the judgment results. Finally, the sliding window trend fusion breaks through the limitations of single-judgment, replacing traditional periodic statistical analysis with time series trend early warning, enabling earlier detection of capacity degradation trends and reducing the risk of sudden failures without warning. This entire process automates and localizes data acquisition, model fitting, and judgment and early warning, reducing communication latency and bandwidth consumption, and providing an efficient, real-time, and scalable solution for power system operation and maintenance.

[0075] The step of connecting an edge monitoring unit in parallel and a current-limiting resistor in series in the electrical circuit containing the target capacitor, setting a preset voltage sampling period and recording the start timestamp, and obtaining the initial voltage value and nominal capacitance value at the relative time zero point specifically includes:

[0076] An edge monitoring unit is connected in parallel across the target capacitor. The edge monitoring unit has voltage acquisition function and local computing processing capability, and each edge monitoring unit is assigned a unique number.

[0077] A current-limiting resistor is connected in series in the electrical circuit where the target capacitor is located. The resistance value of the current-limiting resistor is preset and recorded, and the resistance value of the current-limiting resistor is used as the fixed resistance parameter in the discharge circuit.

[0078] Configure the voltage sampling period in the edge monitoring unit and set the voltage sampling period to a preset voltage sampling period.

[0079] Before performing the target capacitor discharge test, record the actual calendar time at that time, and record the actual calendar time as the discharge test start timestamp, and define the discharge start time as the relative time zero point.

[0080] At the instant the target capacitor begins to discharge, the initial voltage value across the target capacitor is acquired by the edge monitoring unit and recorded as the initial voltage value at the relative time zero point.

[0081] Obtain the nominal capacitance value of the target capacitor from its nameplate or manufacturer's technical data, and use the nominal capacitance value as a reference capacitance parameter for this capacitance identification and anomaly determination process.

[0082] Further specific implementation steps include:

[0083] An edge monitoring unit with voltage acquisition and local computing capabilities is deployed in parallel across the target capacitor. The unit number is denoted as... ;

[0084] In the circuit containing the capacitor, a current-limiting resistor of a known value is connected in series. The resistance value is set to be... ;in, This is the resistance value of the current-limiting resistor connected in series in the discharge circuit;

[0085] Set the voltage sampling period of the edge nodes to ;

[0086] Before conducting the discharge test, record the actual calendar time as the experiment start timestamp. At the same time, the relative time origin of the discharge start time is set as ;

[0087] The initial voltage value was collected at the moment the discharge began. ;in, Indicates that at time The instantaneous voltage across the capacitor. Time is the independent variable; In time The initial voltage value measured at that time;

[0088] Obtain the nominal capacitance value of the capacitor as specified by the manufacturer, and denot it as... .

[0089] The process of acquiring the voltage across the target capacitor at a preset voltage sampling period starting from the relative time zero point to form an original voltage sequence data set specifically includes:

[0090] During the discharge process of the target capacitor, continuous and equally spaced sampling is performed starting from the zero point of relative time according to the preset voltage sampling period to obtain multiple sampling moments arranged in chronological order. The total number of time and voltage sampling point pairs is not less than two.

[0091] At each sampling moment, the measured voltage across the target capacitor is obtained by the edge monitoring unit, and each sampling moment and its corresponding measured voltage value are combined into a time-voltage sampling point pair.

[0092] All time and voltage sampling point pairs are stored sequentially into the data set according to time order, forming the original voltage sequence data set describing the discharge process of the target capacitor.

[0093] Further specific implementation steps include:

[0094] During the discharge process, at equal intervals from Start continuous sampling, set the number of samples... The time for each sampling moment is: , ;in, For the first The relative time points corresponding to each discrete sampling point; The sampling sequence number; This represents the total number of sampling points predetermined during this discharge process;

[0095] Obtain the corresponding voltage sample value: ;in, In the first The capacitor terminal voltage value measured by the edge monitoring unit at each sampling time;

[0096] Construct the original set of sample point pairs: ;in, This is the original sampled data set for this discharge process.

[0097] The process of establishing a theoretical discharge voltage model using the initial voltage value across the target capacitor, the resistance value of the current-limiting resistor, and the candidate capacitance value as parameters, calculating the theoretical voltage at each sampling time, and generating a theoretical voltage sequence data set corresponding one-to-one with the original sequence specifically includes:

[0098] Based on the discharge circuit consisting of the target capacitor and the current-limiting resistor connected in series, a theoretical discharge voltage model is constructed using the exponential decay characteristic of the series discharge of the resistor and capacitor. The theoretical discharge voltage model takes the initial voltage value across the target capacitor, the resistance value of the current-limiting resistor, and the candidate capacitance value as input parameters, and outputs the theoretical voltage value across the target capacitor under a given time condition.

[0099] For each sampling time recorded in the original voltage sequence data set, under the same candidate capacitance value, the theoretical discharge voltage model is called to calculate the theoretical voltage value across the target capacitor corresponding to the sampling time.

[0100] According to the time order of the original voltage sequence data set, the theoretical voltage values ​​corresponding to each sampling time are arranged sequentially to generate a theoretical voltage sequence data set that maintains a one-to-one correspondence with the original voltage sequence data set at the sampling time.

[0101] Further specific implementation steps include:

[0102] Build Discharge model, calculates at time... The theoretical voltage value is: ;in, Given a candidate capacitance value At that time, from ideal Discharge model in the first The theoretical voltage value calculated at each sampling time; The estimated capacity value; The unit is Equivalent to a second;

[0103] Constructing the theoretical voltage sequence: ;in, To achieve a given capacitance value At that time, by A complete set of theoretical voltage sequences generated by the model.

[0104] The step of calculating the residual between the measured voltage and the corresponding theoretical voltage for each candidate capacitance value and summing the squares to obtain a capacitance error evaluation function with capacitance as the independent variable specifically includes:

[0105] Under each candidate capacitance value condition, for each time and voltage sampling point pair in the original voltage sequence data set, calculate the difference between the measured voltage value in the corresponding time and voltage sampling point pair and the theoretical voltage value in the corresponding time and voltage sampling point pair in the theoretical voltage sequence data set, and use the difference between the measured voltage value and the theoretical voltage value as the single-point residual of the corresponding time and voltage sampling point pair;

[0106] Under the same candidate capacitance value, the single-point residuals of all time and voltage sampling point pairs are squared, and the summation operation is performed over the entire range of all time and voltage sampling point pairs to obtain the total residual sum of squares for the current candidate capacitance value;

[0107] Within the entire range of candidate capacitance values, a capacity error evaluation function is formed, with the candidate capacitance value as the independent variable and the corresponding total residual sum of squares as the function value. The capacity error evaluation function only changes with the candidate capacitance value, while the other parameters remain unchanged during the capacity identification process.

[0108] Further specific implementation steps include:

[0109] Construct a single-point residual function, specifically as follows: ;in, To achieve a given capacitance value At that time, the first Measured voltage at each sampling point Compared with theoretically predicted voltage The difference between them;

[0110] Construct the total residual sum of squares function as the objective function to minimize it, specifically as follows: ;in, Given a capacitance value When the sum of the squared residuals of all sampling points is the objective function for capacity identification.

[0111] The process of calculating a capacity error evaluation function within a preset capacity search range, selecting the smallest total residual sum of squares as the optimal capacity estimate, and comparing the difference between the minimum and second smallest values ​​with a preset residual difference threshold to determine convergence and residual distribution stability, specifically includes:

[0112] Based on the nominal capacitance value of the target capacitor, the lower limit of the capacity search is set to half of the nominal capacitance value, and the upper limit of the capacity search is set to the nominal capacitance value. Multiple discrete capacity points are divided between the lower limit and the upper limit of the capacity search according to a preset capacity step size, and a capacity search set consisting of multiple candidate capacitance values ​​is constructed.

[0113] For each candidate capacitance value in the capacity search set, the capacity error evaluation function is called to calculate the total residual sum of squares corresponding to each candidate capacitance value, and the correspondence between the candidate capacitance value and the corresponding total residual sum of squares is recorded.

[0114] Find the candidate capacitor value with the smallest total residual square sum in the capacity search set, take the candidate capacitor value with the smallest total residual square sum as the optimal capacity estimate for this discharge process, and record the corresponding minimum total residual square sum as the current minimum total residual square sum;

[0115] After excluding the candidate capacitor value with the smallest total residual sum of squares, find the candidate capacitor value with the second smallest total residual sum of squares in the set of remaining candidate capacitor values, and record the total residual sum of squares corresponding to the second smallest candidate capacitor value as the current second smallest total residual sum of squares;

[0116] Calculate the absolute value of the difference between the current second smallest total residual sum of squares and the current smallest total residual sum of squares. Compare the absolute value of the difference with a preset residual difference threshold. The physical quantity unit of the preset residual difference threshold is voltage square, and its numerical value is on the order of one millionth. When the difference is less than the preset residual difference threshold, it is determined that the optimal capacity estimate meets the convergence condition and the residual distribution is stable. When the difference is greater than or equal to the preset residual difference threshold, it is determined that the optimal capacity estimate does not meet the stability requirement.

[0117] Further specific implementation steps include:

[0118] Constructing the capacity search interval, specifically: ;in, This is the set of candidate capacitance values ​​used for the trial search; For the search set of the first One candidate capacitance value; This is the capacity search sequence number; This is the minimum capacitance value within the search range; This is the capacity search step size; The total number of discrete points to search;

[0119] For each Calculate the objective function value In the set Find the minimum value in the range, and let: At the same time, the index that achieves this minimum value is set as: That is, ;in, The capacitance estimate is obtained by minimizing the objective function; Capacity index to achieve the minimum value of the objective function;

[0120] Set the current minimum residual to: ;in, Let be the minimum value of the objective function obtained in this search;

[0121] In the index set Remove index postscript: ;in, To exclude the minimum value index Afterwards, Obtain the index number of the second smallest value;

[0122] Calculate the absolute value of the difference between the second smallest residual and the smallest residual. Specifically: ;

[0123] when When the capacity estimation results converge and the residual distribution is stable, then... The residual difference threshold used for convergence assessment;

[0124] when If the result does not meet the stability requirements, the capacity estimation result is determined to be incorrect.

[0125] The process of calculating the absolute deviation of the capacitance value from the nominal capacitance value and the ratio of the relative capacitance error when the capacitance is converged and stable, and comparing it with the relative capacitance error ratio threshold, and outputting whether the capacitance is abnormal or normal, specifically includes:

[0126] After the optimal capacity estimate is obtained and the convergence condition is met, the absolute value of the difference between the optimal capacity estimate and the nominal capacitance value is calculated to obtain the absolute capacity deviation.

[0127] The ratio of absolute capacitance deviation to nominal capacitance value is used as the relative capacitance error ratio. The relative capacitance error ratio is obtained by dividing the absolute capacitance deviation by the nominal capacitance value.

[0128] The preset capacity relative error ratio threshold is set to 0.1, which corresponds to 10% in percentage form.

[0129] The relative error ratio of the capacity is compared with the relative error ratio threshold of the capacity. When the relative error ratio of the capacity is greater than the relative error ratio threshold of the capacity, the capacity status of the target capacitor is judged as abnormal. When the relative error ratio of the capacity is less than or equal to the relative error ratio threshold of the capacity, the capacity status of the target capacitor is judged as normal.

[0130] Further specific implementation steps include:

[0131] The absolute deviation of the calculated capacity is as follows: ;in, To estimate the absolute difference between the capacity and the nominal capacity;

[0132] The relative error ratio of the capacity is calculated as follows: ;in, This is the ratio of the relative error of the capacitor;

[0133] Set the threshold ratio for capacity deviation. ;

[0134] when When this occurs, it is determined to be a capacity anomaly;

[0135] when The capacity was determined to be normal at that time.

[0136] The process of generating capacity monitoring result records using the optimal capacity estimate, the relative capacity error ratio, and the capacity status marker, and then writing them into the edge node database along with the target capacitor's unique identifier and start timestamp, specifically includes:

[0137] The capacity monitoring results record is constructed based on the optimal capacity estimate, the relative error ratio of capacity, and the capacity status marker during the current discharge process. The capacity status marker adopts binary encoding, with a value of 1 indicating an abnormal capacity state and a value of 0 indicating a normal capacity state.

[0138] Obtain the unique identifier of the target capacitor and the discharge test start time stamp. Use the unique identifier of the target capacitor and the discharge test start time stamp as index information, and combine them with the optimal capacity estimate, the relative capacity error ratio and the capacity status mark to form a capacity monitoring result record.

[0139] Capacity monitoring results are recorded and written to the edge node database and stored according to the index information, maintaining a structured set of capacity monitoring results records in the edge node database.

[0140] Further specific implementation steps include:

[0141] Construct the capacity estimation results into a structured tuple, specifically as follows: ;in, A structured record tuple for the results of a single capacity analysis; A binary label variable for the capacitor state. This indicates that the capacity is abnormal. This indicates that the capacity is normal.

[0142] Will With timestamp and capacitor number As an index, write to the edge node database: ;in, These are database record entries used within the edge nodes to store capacitance monitoring results. This is a unique identifier for the capacitor.

[0143] The process of setting the sliding window length, selecting the most recent optimal capacity estimates within the window, calculating the moving average capacity value and the ratio of the sliding capacity error, comparing it with a capacity relative error ratio threshold, identifying a continuous degradation trend, and outputting a capacity degradation early warning flag specifically includes:

[0144] In the edge node database, the optimal capacity estimate corresponding to the most recent discharge tests of the same target capacitor is selected in chronological order. The sliding window length is preset and is greater than one. The most recent optimal capacity estimates are combined into a sliding window capacity estimate set.

[0145] In each degradation trend analysis, the optimal capacity estimate within the sliding window is calculated using the arithmetic mean to obtain the moving average capacity value;

[0146] Calculate the absolute value of the difference between the sliding average capacitance value and the nominal capacitance value to obtain the absolute deviation of the sliding capacitance, and use the ratio of the absolute deviation of the sliding capacitance to the nominal capacitance value as the sliding capacitance error ratio.

[0147] The sliding capacity error ratio is compared with the relative capacity error ratio threshold. When the sliding capacity error ratio is greater than the relative capacity error ratio threshold, it is determined that the target capacitor capacity has a continuous degradation trend and a capacity degradation warning mark is generated. When the sliding capacity error ratio is less than or equal to the relative capacity error ratio threshold, it is determined that the change in the target capacitor capacity is within the allowable range and no capacity degradation warning mark is generated.

[0148] Further specific implementation steps include:

[0149] Set the sliding window length to Take the nearest Set of secondary capacity estimates Specifically: ;in, This is a set of capacity estimates used for sliding window analysis; It is a time-order index; Numbered in chronological order Secondary capacity estimation results;

[0150] The moving average capacity is calculated as follows: ;in, Based on the most recent The moving average capacity is calculated from the results of the second capacity estimation.

[0151] The slip error ratio is calculated as follows: ;in, This is the ratio of relative deviations calculated based on the moving average capacity;

[0152] when If this occurs, it is judged as a continuous degradation trend, triggering a capacity degradation warning flag;

[0153] when If the current capacity change is determined to be within the allowable range, no warning flag will be triggered.

[0154] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0155] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for dynamic monitoring of capacitor capacitance anomalies based on edge computing, characterized in that, include: In the electrical circuit where the target capacitor is located, an edge monitoring unit is connected in parallel and a current-limiting resistor is connected in series. A preset voltage sampling period is set and the start timestamp is recorded to obtain the initial voltage value and nominal capacitance value at the relative time zero point. Starting from the zero point of relative time, the voltage across the target capacitor is collected according to the preset voltage sampling period to form an original voltage sequence data set; A theoretical discharge voltage model is established with the initial voltage value across the target capacitor, the resistance value of the current limiting resistor, and the candidate capacitance value as parameters. The theoretical voltage at each sampling time is calculated, and a set of theoretical voltage sequence data corresponding one-to-one with the original sequence is generated. For each candidate capacitance value, the residual between the measured voltage and the corresponding theoretical voltage is calculated and the sum of squares is obtained to obtain the capacitance error evaluation function with the capacitance value as the independent variable. Within the preset capacity search range, the capacity error evaluation function is calculated, and the one with the smallest total residual sum of squares is selected as the optimal capacity estimate. The convergence and residual distribution stability are determined by comparing the difference between the minimum and the second smallest value with the preset residual difference threshold. When convergence and stability are achieved, calculate the absolute deviation of the capacity from the nominal capacitance value and the ratio of the relative error of the capacity, and compare it with the threshold of the relative error of the capacity, and output whether the capacity is abnormal or normal. The capacity monitoring results record is generated using the optimal capacity estimate, the relative capacity error ratio, and the capacity status marker. It is then written into the edge node database along with the unique identifier of the target capacitor and the start timestamp. Set the sliding window length, select the most recent optimal capacity estimates within the window, calculate the moving average capacity value and the ratio of the sliding capacity error, compare it with the capacity relative error ratio threshold, identify the continuous degradation trend, and output a capacity degradation early warning flag.

2. The method for dynamic monitoring of capacitor capacitance anomalies based on edge computing according to claim 1, characterized in that, The step of connecting an edge monitoring unit in parallel and a current-limiting resistor in series in the electrical circuit containing the target capacitor, setting a preset voltage sampling period and recording the start timestamp, and obtaining the initial voltage value and nominal capacitance value at the relative time zero point specifically includes: An edge monitoring unit is connected in parallel across the target capacitor. The edge monitoring unit has voltage acquisition function and local computing processing capability, and each edge monitoring unit is assigned a unique number. A current-limiting resistor is connected in series in the electrical circuit where the target capacitor is located. The resistance value of the current-limiting resistor is preset and recorded, and the resistance value of the current-limiting resistor is used as the fixed resistance parameter in the discharge circuit. Configure the voltage sampling period in the edge monitoring unit and set the voltage sampling period to a preset voltage sampling period. Before performing the target capacitor discharge test, record the actual calendar time at that time, and record the actual calendar time as the discharge test start timestamp, and define the discharge start time as the relative time zero point. At the instant the target capacitor begins to discharge, the initial voltage value across the target capacitor is acquired by the edge monitoring unit and recorded as the initial voltage value at the relative time zero point. Obtain the nominal capacitance value of the target capacitor from its nameplate or manufacturer's technical data, and use the nominal capacitance value as a reference capacitance parameter for this capacitance identification and anomaly determination process.

3. The method for dynamic monitoring of capacitor capacitance anomalies based on edge computing according to claim 2, characterized in that, The process of acquiring the voltage across the target capacitor at a preset voltage sampling period starting from the relative time zero point to form an original voltage sequence data set specifically includes: During the discharge process of the target capacitor, continuous and equally spaced sampling is performed starting from the zero point of relative time according to the preset voltage sampling period to obtain multiple sampling moments arranged in chronological order. The total number of time and voltage sampling point pairs is not less than two. At each sampling moment, the measured voltage across the target capacitor is obtained by the edge monitoring unit, and each sampling moment and its corresponding measured voltage value are combined into a time-voltage sampling point pair. All time and voltage sampling point pairs are stored sequentially into the data set according to time order, forming the original voltage sequence data set describing the discharge process of the target capacitor.

4. The method for dynamic monitoring of capacitor capacitance anomalies based on edge computing according to claim 3, characterized in that, The process of establishing a theoretical discharge voltage model using the initial voltage value across the target capacitor, the resistance value of the current-limiting resistor, and the candidate capacitance value as parameters, calculating the theoretical voltage at each sampling time, and generating a theoretical voltage sequence data set corresponding one-to-one with the original sequence specifically includes: Based on the discharge circuit consisting of the target capacitor and the current-limiting resistor connected in series, a theoretical discharge voltage model is constructed using the exponential decay characteristic of the series discharge of the resistor and capacitor. The theoretical discharge voltage model takes the initial voltage value across the target capacitor, the resistance value of the current-limiting resistor, and the candidate capacitance value as input parameters, and outputs the theoretical voltage value across the target capacitor under a given time condition. For each sampling time recorded in the original voltage sequence data set, under the same candidate capacitance value, the theoretical discharge voltage model is called to calculate the theoretical voltage value across the target capacitor corresponding to the sampling time. According to the time order of the original voltage sequence data set, the theoretical voltage values ​​corresponding to each sampling time are arranged sequentially to generate a theoretical voltage sequence data set that maintains a one-to-one correspondence with the original voltage sequence data set at the sampling time.

5. The method for dynamic monitoring of capacitor capacitance anomalies based on edge computing according to claim 4, characterized in that, The step of calculating the residual between the measured voltage and the corresponding theoretical voltage for each candidate capacitance value and summing the squares to obtain a capacitance error evaluation function with capacitance as the independent variable specifically includes: Under each candidate capacitance value condition, for each time and voltage sampling point pair in the original voltage sequence data set, calculate the difference between the measured voltage value in the corresponding time and voltage sampling point pair and the theoretical voltage value in the corresponding time and voltage sampling point pair in the theoretical voltage sequence data set, and use the difference between the measured voltage value and the theoretical voltage value as the single-point residual of the corresponding time and voltage sampling point pair; Under the same candidate capacitance value, the single-point residuals of all time and voltage sampling point pairs are squared, and the summation operation is performed over the entire range of all time and voltage sampling point pairs to obtain the total residual sum of squares for the current candidate capacitance value; Within the entire range of candidate capacitance values, a capacity error evaluation function is formed, with the candidate capacitance value as the independent variable and the corresponding total residual sum of squares as the function value. The capacity error evaluation function only changes with the candidate capacitance value, while the other parameters remain unchanged during the capacity identification process.

6. The method for dynamic monitoring of capacitor capacitance anomalies based on edge computing according to claim 5, characterized in that, The process of calculating a capacity error evaluation function within a preset capacity search range, selecting the smallest total residual sum of squares as the optimal capacity estimate, and comparing the difference between the minimum and second smallest values ​​with a preset residual difference threshold to determine convergence and residual distribution stability, specifically includes: Based on the nominal capacitance value of the target capacitor, the lower limit of the capacity search is set to half of the nominal capacitance value, and the upper limit of the capacity search is set to the nominal capacitance value. Multiple discrete capacity points are divided between the lower limit and the upper limit of the capacity search according to a preset capacity step size, and a capacity search set consisting of multiple candidate capacitance values ​​is constructed. For each candidate capacitance value in the capacity search set, the capacity error evaluation function is called to calculate the total residual sum of squares corresponding to each candidate capacitance value, and the correspondence between the candidate capacitance value and the corresponding total residual sum of squares is recorded. Find the candidate capacitor value with the smallest total residual square sum in the capacity search set, take the candidate capacitor value with the smallest total residual square sum as the optimal capacity estimate for this discharge process, and record the corresponding minimum total residual square sum as the current minimum total residual square sum; After excluding the candidate capacitor value with the smallest total residual sum of squares, find the candidate capacitor value with the second smallest total residual sum of squares in the set of remaining candidate capacitor values, and record the total residual sum of squares corresponding to the second smallest candidate capacitor value as the current second smallest total residual sum of squares; Calculate the absolute value of the difference between the current second smallest total residual sum of squares and the current smallest total residual sum of squares. Compare the absolute value of the difference with a preset residual difference threshold. The physical quantity unit of the preset residual difference threshold is voltage square, and its numerical value is on the order of one millionth. When the difference is less than the preset residual difference threshold, it is determined that the optimal capacity estimate meets the convergence condition and the residual distribution is stable. When the difference is greater than or equal to the preset residual difference threshold, it is determined that the optimal capacity estimate does not meet the stability requirement.

7. The method for dynamic monitoring of capacitor capacitance anomalies based on edge computing according to claim 6, characterized in that, The process of calculating the absolute deviation of the capacitance value from the nominal capacitance value and the ratio of the relative capacitance error when the capacitance is converged and stable, and comparing it with the relative capacitance error ratio threshold, and outputting whether the capacitance is abnormal or normal, specifically includes: After the optimal capacity estimate is obtained and the convergence condition is met, the absolute value of the difference between the optimal capacity estimate and the nominal capacitance value is calculated to obtain the absolute capacity deviation. The ratio of absolute capacitance deviation to nominal capacitance value is used as the relative capacitance error ratio. The relative capacitance error ratio is obtained by dividing the absolute capacitance deviation by the nominal capacitance value. The preset capacity relative error ratio threshold is set to 0.1, which corresponds to 10% in percentage form. The relative error ratio of the capacity is compared with the relative error ratio threshold of the capacity. When the relative error ratio of the capacity is greater than the relative error ratio threshold of the capacity, the capacity status of the target capacitor is judged as abnormal. When the relative error ratio of the capacity is less than or equal to the relative error ratio threshold of the capacity, the capacity status of the target capacitor is judged as normal.

8. The method for dynamic monitoring of capacitor capacitance anomalies based on edge computing according to claim 7, characterized in that, The process of generating capacity monitoring result records using the optimal capacity estimate, the relative capacity error ratio, and the capacity status marker, and then writing them into the edge node database along with the target capacitor's unique identifier and start timestamp, specifically includes: The capacity monitoring results record is constructed based on the optimal capacity estimate, the relative error ratio of capacity, and the capacity status marker during the current discharge process. The capacity status marker adopts binary encoding, with a value of 1 indicating an abnormal capacity state and a value of 0 indicating a normal capacity state. Obtain the unique identifier of the target capacitor and the discharge test start time stamp. Use the unique identifier of the target capacitor and the discharge test start time stamp as index information, and combine them with the optimal capacity estimate, the relative capacity error ratio and the capacity status mark to form a capacity monitoring result record. Capacity monitoring results are recorded and written to the edge node database and stored according to the index information, maintaining a structured set of capacity monitoring results records in the edge node database.

9. The method for dynamic monitoring of capacitor capacitance anomalies based on edge computing according to claim 8, characterized in that, The process of setting the sliding window length, selecting the most recent optimal capacity estimates within the window, calculating the moving average capacity value and the ratio of the sliding capacity error, comparing it with a capacity relative error ratio threshold, identifying a continuous degradation trend, and outputting a capacity degradation early warning flag specifically includes: In the edge node database, the optimal capacity estimate corresponding to the most recent discharge tests of the same target capacitor is selected in chronological order. The sliding window length is preset and is greater than one. The most recent optimal capacity estimates are combined into a sliding window capacity estimate set. In each degradation trend analysis, the optimal capacity estimate within the sliding window is calculated using the arithmetic mean to obtain the moving average capacity value; Calculate the absolute value of the difference between the sliding average capacitance value and the nominal capacitance value to obtain the absolute deviation of the sliding capacitance, and use the ratio of the absolute deviation of the sliding capacitance to the nominal capacitance value as the sliding capacitance error ratio. The sliding capacity error ratio is compared with the relative capacity error ratio threshold. When the sliding capacity error ratio is greater than the relative capacity error ratio threshold, it is determined that the target capacitor capacity has a continuous degradation trend and a capacity degradation warning mark is generated. When the sliding capacity error ratio is less than or equal to the relative capacity error ratio threshold, it is determined that the change in the target capacitor capacity is within the allowable range and no capacity degradation warning mark is generated.

Images