Passive vector parameter real-time monitoring method in AC-DC conversion process

Abstract

The invention provides a passive vector parameter real-time monitoring method in an AC-DC conversion process, and belongs to the field of power system detection, and the method comprises the steps: monitoring the operation data of an AC-DC conversion system in a low-voltage metering cabinet in real time, constructing a hexagonal diagram according to the operation data, and extracting the characteristic parameters of the hexagonal diagram; the hexagonal diagram characteristic parameters comprise a voltage and current phase relation and a harmonic component; if it is determined that the parameter change of the passive device affects the AC-DC conversion system, generating a passive device degradation early warning signal, adjusting the reactive compensation capacity, the reactive compensation device parameter and the filter parameter, and determining the adjustment amount; monitoring the change of the characteristic parameters of the hexagonal diagram in real time, performing trend fitting on the characteristic parameters of the historical hexagonal diagram by adopting a time sequence analysis method, judging a degradation acceleration point of the passive device, and adjusting the operation parameters of the reactive power compensation device according to the degradation acceleration point of the passive device; degradation of a passive device can be effectively early warned, a reactive compensation strategy is optimized, and operation reliability and metering precision of the low-voltage metering cabinet are improved.

Description

Technical Field

[0001] The invention relates to a real-time monitoring method for passive vector parameters in an AC / DC conversion process, and belongs to the field of power system detection. Background Art

[0002] In the low-voltage metering cabinet system, the voltage and current signals of the AC / DC conversion system are sampled, and the AC voltage and current in the three-phase stationary coordinate system are converted into DC quantities in the two-phase rotating coordinate system. The obtained voltage and current vectors are then plotted on the complex plane. As time goes by, these vector points will form an approximately hexagonal trajectory, namely a hexagonal graph. The shape, size and position of the hexagonal graph reflect the operating status of the AC / DC conversion system and the parameter changes of the passive components. The characteristic parameters of the hexagonal graph of the AC / DC conversion system are important indicators reflecting the status of the passive components.

[0003] However, due to the complex electrical coupling between multiple low-voltage metering cabinets, changes in the passive device parameters of a single metering cabinet can affect the operating status of the entire cluster. This often leads to the following technical difficulties in the extraction and analysis of hexagonal map feature parameters:

[0004] First, the passive device parameters of different metering cabinets show nonlinear changes under different load conditions, which makes it difficult to accurately capture the dynamic characteristics of the hexagonal diagram feature parameters.

[0005] Secondly, the electrical coupling effect between multiple metering cabinets causes the changes in the characteristic parameters of the hexagonal diagram of a single metering cabinet to be transmitted to adjacent metering cabinets, forming a chain reaction, further exacerbating the complexity of parameter identification.

[0006] In addition, the degradation process of passive components is gradual and hidden, and their parameter changes are often difficult to detect through conventional monitoring methods in the early stages, resulting in the gradual degradation of the AC / DC conversion system performance. During the collaborative operation of the cluster, due to the differences in the status of the passive components of each metering cabinet, the coordinated control of the reactive compensation device faces challenges.

[0007] For example, if the passive device parameters of a metering cabinet exhibit abnormalities, the adjustment capabilities of its reactive power compensation device may be limited, affecting the power factor optimization and metering accuracy of the entire cluster. Furthermore, the degradation of passive devices increases the potential failure risk of the metering cabinet. However, the hidden nature of the degradation process makes the establishment of early warning mechanisms difficult.

[0008] The existence of these technical problems makes the research on low-voltage metering cabinet passive device status assessment and cluster collaborative optimization operation strategy based on hexagonal graph characteristic parameters face many challenges. Summary of the Invention

[0009] According to the problem described in the background, the problem to be solved by the present invention is: to provide a real-time monitoring method for passive vector parameters in the AC-DC conversion process to solve the above-mentioned problems.

[0010] To achieve the above object, the present invention provides the following technical solution: a method for real-time monitoring of passive vector parameters during AC / DC conversion is provided, comprising the following steps:

[0011] (1) Collect the operating data of the AC / DC conversion system in the low-voltage metering cabinet, generate a hexagonal diagram to extract its characteristic parameters, and analyze the parameters to evaluate the status of passive components;

[0012] (2) The historical operating parameters of the passive device are combined with the characteristic parameters of the hexagonal diagram to establish a mapping relationship with the AC / DC conversion system, and the failure mode of the passive device is obtained. The influence of the change of the passive device parameters on the AC / DC conversion system is judged based on the preset performance threshold;

[0013] (3) If it is determined that the change in the parameters of the passive device has an impact on the AC / DC conversion system, a passive device degradation warning signal is generated, and the reactive compensation capacity, reactive compensation device parameters, and filter parameters are adjusted to determine the adjustment amount;

[0014] (4) Using a machine learning algorithm, the adjustment amount and historical parameter data of passive devices are used as input to predict the trend of passive device parameter changes. The compensation method, compensation capacity configuration and switching method of the reactive compensation device are determined based on the prediction results, and the power factor of the entire cluster is controlled through dynamic compensation and coordination.

[0015] (5) Based on the compensation method, compensation capacity configuration and switching method of the reactive compensation device, as well as the overall power factor of the cluster and the early warning signal of passive device degradation, the particle swarm optimization algorithm is used to coordinate the reactive compensation devices of several low-voltage metering cabinets and adjust the output capacity and operating status of the reactive compensation device;

[0016] (6) Monitor the changes of the characteristic parameters of the hexagonal diagram in real time, use the time series analysis method to perform trend fitting on the characteristic parameters of the historical hexagonal diagram, determine the acceleration point of the degradation of the passive device, and adjust the operating parameters of the reactive compensation device according to the acceleration point of the degradation of the passive device.

[0017] Preferably, the step (1) includes the following steps:

[0018] (1.1) Sampling the AC / DC conversion system of the low-voltage metering cabinet according to the preset sampling frequency to obtain AC voltage and current waveform data;

[0019] (1.2) processing the AC voltage and current waveform data by Fourier transform, and extracting amplitude and phase characteristics of the fundamental signal and the harmonic signal from the waveform data;

[0020] (1.3) constructing a hexagonal coordinate matrix based on the characteristics of the fundamental signal and the harmonic signal, wherein the hexagonal coordinate matrix includes voltage amplitude, current amplitude, phase difference value, fundamental frequency, total harmonic distortion rate, and grid frequency deviation value;

[0021] (1.4) Performing principal component analysis on the hexagonal coordinate matrix, clustering the characteristic components obtained by the principal component analysis using a support vector machine, and obtaining evaluation indicators of monitoring parameters.

[0022] Preferably, the step (2) includes the following steps:

[0023] (2.1) Obtaining historical operation records of passive devices in a memory, wherein the historical operation records include device operation time, temperature parameters, voltage stress parameters, and current stress parameters;

[0024] (2.2) constructing a device stress parameter matrix based on the historical operation records, and obtaining a failure mode classification result by using Gaussian mixture clustering on the stress parameter matrix;

[0025] (2.3) using a recursive neural network to establish a mapping relationship between the stress parameter matrix and operating parameters of the AC / DC conversion system, wherein the operating parameters include a power factor value and a metering accuracy value;

[0026] (2.4) Determining the degree of impact of passive components on the performance of the AC / DC conversion system, specifically including: constructing a component performance degradation index based on the failure mode classification results;

[0027] (2.5) extracting temperature threshold parameters, voltage threshold parameters, and current threshold parameters based on the performance degradation indicator;

[0028] (2.6) Calculate the power factor deviation value and the metering accuracy deviation value using the threshold parameters.

[0029] Preferably, the step (3) includes the following steps:

[0030] (3.1) receiving a warning data matrix indicating that at least one parameter of the device's temperature change, voltage change, and current change exceeds a preset threshold value, and classifying the warning data matrix using a multi-layer perceptron to obtain a warning level value;

[0031] (3.2) extracting a reactive compensation capacity reference value, a compensation device reference parameter, and a compensation operating point reference value from a compensation parameter database according to the warning level value, and calculating a compensation capacity adjustment value using a reactive power compensation algorithm;

[0032] (3.3) extracting a frequency range reference value, a bandwidth reference value, and a gain reference value from a filter parameter database for the warning level value and the compensation capacity adjustment value, and obtaining a filter parameter adjustment value by performing a least squares operation;

[0033] (3.4) constructing a correction parameter matrix based on the compensation capacity adjustment value and the filter parameter adjustment value, and optimizing the correction parameter matrix using a random forest regression method to obtain a final adjustment value of the reactive compensation capacity, a final adjustment value of the compensation device parameter, and a final adjustment value of the filter parameter.

[0034] Preferably, the step (4) includes the following steps:

[0035] (4.1) constructing a training data matrix based on the historical parameter data of the passive device, processing the training data matrix through a long short-term memory network, and obtaining a passive device parameter change trend prediction curve;

[0036] (4.2) extracting a time distribution sequence of parameter changes based on the prediction curve, constructing a parameter change matrix, wherein the parameter change matrix includes temperature changes, voltage changes, and current changes, and calculating a compensation method selection sequence number based on the parameter change matrix;

[0037] (4.3) extracting an initial configuration value of compensation capacity corresponding to the compensation mode selection number from a compensation device parameter database, optimizing the initial configuration value of compensation capacity using a Kalman filter, calculating an optimal estimate based on the measurement noise variance and the process noise variance, and obtaining an optimized configuration value of compensation capacity;

[0038] (4.4) generating a compensation device switching schedule based on the optimized compensation capacity configuration value, the compensation device switching schedule including a switching sequence number, a switching time interval, and a switching hold time, adjusting the compensation device switching schedule in real time using an adaptive feedback compensation algorithm, and calculating a compensation deviation based on a real-time power factor measurement value;

[0039] (4.5) If the overall power factor is lower than a preset threshold, the reactive compensation adjustment range is determined, and the reactive compensation device operating parameter adjustment value is calculated based on the adjustment range. The reactive compensation device operating strategy is updated using the parameter adjustment value, and the overall power factor of the cluster is iteratively updated. It is determined whether the updated overall power factor reaches the preset optimization target value. If the optimization target value is not reached, the operating strategy is updated again until the overall power factor reaches the optimization target value.

[0040] Preferably, the step (5) includes the following steps:

[0041] (5.1) Obtaining the operating data of the compensation device of the low-voltage meter cabinet to construct an operating status matrix, wherein the operating status matrix includes the compensation mode attribute value, the capacity configuration value, and the switching mode sequence number;

[0042] (5.2) constructing a fitness function based on the operating state matrix, wherein the fitness function includes a compensation effect constraint term and a device operation constraint term;

[0043] (5.3) Parameter search of the fitness function is performed using a particle swarm optimization algorithm, where the particle position represents the compensation capacity allocation ratio, and the particle velocity represents the capacity adjustment step size;

[0044] (5.4) Calculating the device load rate based on the compensation capacity allocation ratio, where the load rate is the ratio of the real-time capacity to the rated capacity, and grouping and controlling the compensation devices according to the load rate values;

[0045] (5.5) A deep neural network is used to predict the load rate. If the predicted load rate exceeds a preset threshold, a control instruction sequence of the compensation device is corrected according to the predicted load rate to obtain a corrected control instruction sequence.

[0046] Preferably, the step (6) includes the following steps:

[0047] (6.1) collecting characteristic parameters of the hexagonal diagram using a real-time monitoring device, wherein the characteristic parameters include a voltage phase parameter, a current phase parameter, and a harmonic content parameter, and extracting sampling points from the characteristic parameters at fixed time intervals to obtain a characteristic parameter time series data set;

[0048] (6.2) fitting the characteristic parameter time series data set using an exponential smoothing algorithm to obtain a parameter change trend curve, calculating the slope of the trend curve to obtain a parameter change rate curve, calculating the acceleration value based on the change rate curve and marking abnormal points;

[0049] (6.3) Using a long short-term memory network to receive the parameter change rate and acceleration data, verifying the abnormal points through the parameter change law extraction unit in the network hidden layer to obtain the degradation acceleration point;

[0050] (6.4) calculating a compensation parameter adjustment amount based on the parameter change amplitude of the degradation acceleration point position, correcting the compensation device operating parameters based on the compensation parameter adjustment amount, and generating a compensation device operating parameter update instruction;

[0051] (6.5) Obtain the time series data of the characteristic parameters of the historical hexagonal graph, calculate the rate of change of the characteristic parameters at each time point using the sliding window method, perform trend fitting on the rate of change of the characteristic parameters, construct a linear regression model of the time series, and obtain the slope value of the trend line. If the slope value of the trend line exceeds the preset threshold, mark the time point as a potential degradation point, calculate the goodness of fit index, perform second-order difference calculation on the potential degradation point, and determine whether the second-order derivative of the rate of change shows a turning point. If so, determine the corresponding point as the acceleration point of passive device degradation.

[0052] Preferably, the step (4.5) includes the following steps:

[0053] (4.5.1) Collecting a real-time power factor value using a power factor measurement device and determining whether the real-time power factor value is lower than a preset power factor threshold;

[0054] (4.5.2) If the real-time power factor value is lower than a preset power factor threshold, obtaining compensation capacity parameters and operating status parameters from a compensation device parameter library;

[0055] (4.5.3) Constructing a compensation parameter optimization objective function based on the compensation capacity parameter and the operating status parameter, and solving the compensation parameter optimization objective function using a gradient descent method to obtain a reactive compensation adjustment range;

[0056] (4.5.4) Based on the reactive compensation adjustment range, an adaptive learning rate algorithm is used to optimize the compensation capacity value and the switch control value to obtain a compensation parameter adjustment matrix;

[0057] (4.5.5) extracting a compensation capacity correction value and a switch timing correction value from the compensation parameter adjustment matrix, and updating a compensation device control instruction sequence based on the compensation capacity correction value and the switch timing correction value;

[0058] (4.5.6) Executing reactive power compensation operation according to the compensation device control instruction sequence to collect power factor measurement values, and determining whether the power factor measurement values ​​reach a preset optimization target value.

[0059] Preferably, the step (6.5) includes the following steps:

[0060] (6.5.1) Collect historical hexagonal graph characteristic parameters from the low-voltage metering cabinet measurement device, sample the characteristic parameters using a fixed-length sliding window, and obtain a characteristic parameter change rate dataset;

[0061] (6.5.2) Fit the characteristic parameter change rate data set using a linear regression algorithm and calculate the trend line slope of the fitted curve. If the slope is greater than a preset change rate threshold, the sampling point is determined to be a potential outlier.

[0062] (6.5.3) Calculate the goodness-of-fit index for the potential outlier. If the coefficient of determination, standard deviation of fit, and correlation coefficient all meet the preset goodness-of-fit criteria, record the sampling point in the outlier dataset.

[0063] (6.5.4) Perform a first-order difference operation on the outlier data set using the central difference method to obtain a first-order derivative sequence. Perform a second-order difference operation on the first-order derivative sequence to obtain a second-order derivative sequence. If the second-order derivative changes from negative to positive, the sampling point is determined to be a degradation acceleration point.

[0064] The beneficial effects of the present invention are:

[0065] 1. By real-time monitoring of the operating data of the AC / DC conversion system and generating a hexagonal diagram, the parameter changes of passive components can be effectively captured, laying the foundation for state assessment and optimized control.

[0066] 2. The fusion of hexagonal diagram characteristic parameters with historical data can clearly reveal the causal relationship between passive device parameter changes and system performance. Moreover, after stress parameters are recorded in matrix form, their degradation trends can be intuitively reflected. The prediction results of the recursive neural network further quantify the specific impact of these changes on power factor and metering accuracy.

[0067] 3. By adjusting the reactive power compensation capacity, reactive power compensation device parameters and filter parameters, the impact of device degradation on power factor and metering accuracy can be effectively addressed, while avoiding conflicts or system oscillations during parameter adjustment.

[0068] 4. By determining the compensation method, compensation capacity configuration, switching method, and power factor of the reactive compensation device, the entire process from parameter prediction to compensation control is automated. This not only reduces compensation lag, but also avoids equipment shock through multi-level progressive control, significantly improving the stability and efficiency of cluster operation.

[0069] 5. The particle swarm optimization algorithm is used to coordinate the reactive power compensation devices of several low-voltage metering cabinets, adjust the output capacity and operating status of the reactive power compensation devices, respond to degradation warnings, and reduce the risk of high-load operation of equipment through algorithm collaborative optimization. The application of sliding time windows and characteristic curves further enhances the system's adaptability to load changes.

[0070] 6. By adopting the time series analysis method, trend fitting is performed on the characteristic parameters of the historical hexagonal diagram to determine the acceleration point of passive device degradation. According to the acceleration point of passive device degradation, the operating parameters of the reactive compensation device are adjusted to achieve accurate monitoring and dynamic optimization of passive device degradation. By combining real-time data with historical trends, not only the diagnostic efficiency is improved, but also the equipment life is extended through timely adjustment of operating parameters.

[0071] 7. The present invention can effectively warn of degradation of passive components, optimize reactive power compensation strategies, and improve operational reliability and metering accuracy of low-voltage metering cabinets. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] Figure 1 Schematic diagram of the steps of the present invention;

[0073] Figure 2 Schematic diagram of the process of step (1);

[0074] Figure 3 is a flow chart of step (2);

[0075] Figure 4 Schematic diagram of the process of step (3);

[0076] Figure 5 is a flow chart of step (4);

[0077] Figure 6 is a flow chart of step (5);

[0078] Figure 7 is a flow chart of step (6); DETAILED DESCRIPTION

[0079] The embodiments of the present invention are further described below with reference to the accompanying drawings:

[0080] Example 1

[0081] like Figure 1 As shown, the present invention provides a method for real-time monitoring of passive vector parameters during AC-DC conversion, comprising the following steps:

[0082] (1) Real-time collection of operating data of the AC / DC conversion system in the low-voltage metering cabinet, and the generation of a hexagonal diagram based on the collected operating data to extract its characteristic parameters, and the evaluation of the passive device status through parameter analysis;

[0083] Among them, characteristic parameters include indicators such as voltage amplitude, current phase relationship and harmonic components;

[0084] Step (1) includes the following steps:

[0085] (1.1) Sampling the AC / DC conversion system of the low-voltage metering cabinet according to the preset sampling frequency to obtain AC voltage and current waveform data;

[0086] Sampling is to collect the voltage amplitude and current amplitude of each port of the AC / DC conversion system of the low-voltage meter cabinet, perform regular sampling and recording of the AC waveform under the control of the waveform sampling frequency, and obtain the first set of AC voltage and current waveform data at a fixed sampling interval;

[0087] (1.2) processing the AC voltage and current waveform data by Fourier transform, and extracting amplitude and phase characteristics of the fundamental signal and the harmonic signal from the waveform data;

[0088] The first set of AC voltage and current waveform data is processed by Fourier transform, the amplitude of the fundamental signal and each harmonic signal is calculated, and the phase difference value between the voltage signal and the current signal, namely the phase feature, is extracted from the fundamental signal.

[0089] (1.3) constructing a hexagonal coordinate matrix based on the characteristics of the fundamental signal and the harmonic signal, wherein the hexagonal coordinate matrix includes voltage amplitude, current amplitude, phase difference value, fundamental frequency, total harmonic distortion rate, and grid frequency deviation value;

[0090] The first hexagonal coordinate matrix is ​​constructed based on the phase difference value. The voltage amplitude, current amplitude, phase difference value, fundamental frequency, total harmonic distortion (THD), and grid frequency deviation values ​​are annotated along the six dimensions of the coordinate matrix. Amplitude and phase information is extracted from each harmonic signal, and the voltage and current total harmonic distortion rates are calculated. The second hexagonal coordinate matrix is ​​then constructed. For both hexagonal coordinate matrices, numerical features are extracted along each dimension, including amplitude, phase, and frequency features, to produce a set of eigenvectors.

[0091] (1.4) Performing principal component analysis on the hexagonal coordinate matrix, clustering the characteristic components obtained by the principal component analysis using a support vector machine, and obtaining evaluation indicators of monitoring parameters.

[0092] Before principal component analysis, the eigenvector group is subjected to dimensionality reduction processing, and then the main characteristic components are extracted through principal component analysis method to construct a feature classification space. Based on the feature classification space, the support vector machine is used to perform cluster analysis on the characteristic components to obtain the feature clustering results and form the evaluation indicators of the monitoring parameters.

[0093] In step (1.1), the sampling frequency is usually set to 128 times the fundamental frequency. For example, in a 50 Hz power grid, the sampling frequency can reach 6400 Hz to ensure the accuracy of the waveform data. 144 sampling points are collected in each sampling cycle, of which 72 sampling points are for voltage and 72 points are for current. The data are arranged in chronological order to form a waveform sequence.

[0094] In step (1.2), the waveform data is decomposed into the fundamental wave and 2nd to 13th harmonic components through Fourier transform, among which the higher harmonics can be ignored due to their small content; the phase difference extracted from the fundamental wave component is calculated by the zero-crossing time difference of the voltage and current waveforms, reflecting the load characteristics. For example, the phase difference of the resistive load is close to 0 degrees, the inductive load is positive, and the capacitive load is negative.

[0095] In step (1.3), during the hexagonal diagram construction phase, two hexagonal diagram coordinate matrices are constructed based on the extracted feature data. The first matrix reflects the basic operating status of the system and includes six dimensions: voltage RMS, current RMS, phase difference, fundamental frequency, total harmonic distortion, and grid frequency deviation. For example, when the voltage is 220 volts, the current is 5 amperes, and the phase difference is 30 degrees, the normalized values ​​are marked on the corresponding coordinate axes. The second matrix focuses on harmonic features, and the coordinate axes represent the 2nd to 7th harmonic contents, for example, the 2nd harmonic accounts for 2.5% and the 3rd harmonic accounts for 4.2%. The hexagonal diagram is drawn with the origin as the center using polar coordinates.

[0096] In step (1.4), the eigenvector groups of the two matrices are subjected to dimensionality reduction processing, and the first three principal components are extracted using principal component analysis, with a cumulative contribution rate of more than 85%. The feature classification space is constructed, and the feature components are clustered using the radial basis kernel function through the support vector machine. The parameters are optimized based on cross-validation, and the system status is finally divided into three categories: normal, subhealthy, and abnormal.

[0097] In step (1) of the present invention, the generation and analysis process of the hexagonal diagram fully reflects the dynamic characteristics of the system operation, and the sampling frequency and harmonic analysis range can be adjusted according to the actual scenario.

[0098] By following the above steps to monitor the operating data of the AC / DC conversion system in real time and generate a hexagonal diagram, the parameter changes of passive components can be effectively captured, laying the foundation for state assessment and optimized control.

[0099] (2) The historical operating parameters of the passive devices are combined with the characteristic parameters of the hexagonal diagram, and data fusion and machine learning algorithms are used to process them. A mapping relationship with the AC / DC conversion system is established, and the failure mode of the passive devices is obtained. The influence of the change of the passive device parameters on the AC / DC conversion system is judged based on the preset performance threshold.

[0100] The step (2) includes the following steps:

[0101] (2.1) Obtaining historical operation records of passive devices in a memory, wherein the historical operation records include device operation time, temperature parameters, voltage stress parameters, and current stress parameters;

[0102] After reading each parameter, the current hexagonal diagram characteristic parameter set and the power factor and metering accuracy parameters in the metering cabinet are collected, the data are marked according to the time series, and outliers and missing values ​​are removed to obtain the first data set;

[0103] (2.2) constructing a device stress parameter matrix based on the historical operation records, and obtaining a failure mode classification result by using Gaussian mixture clustering on the stress parameter matrix;

[0104] Constructing a device stress parameter matrix based on the historical operation records specifically involves extracting temperature change curves, voltage stress curves, and current stress curves for the passive device parameters in the first data set, constructing a device stress parameter matrix, and obtaining a corresponding relationship between the device stress parameters and the operation time; then, based on the device failure records in the historical operation data records of the passive devices, using Gaussian mixture clustering to classify the device failure data, obtaining failure mode classification results, and establishing a corresponding relationship between the failure mode and the stress parameter matrix;

[0105] (2.3) using a recursive neural network to establish a mapping relationship between the stress parameter matrix and operating parameters of the AC / DC conversion system, wherein the operating parameters include a power factor value and a metering accuracy value;

[0106] Specific operating parameters include input parameters and output parameters. Input parameters include temperature change, voltage stress value, and current stress value. Output parameters include power factor and metering accuracy value.

[0107] (2.4) Determining the degree of impact of passive components on the performance of the AC / DC conversion system, specifically including: constructing a component performance degradation index based on the failure mode classification results;

[0108] Calculate the failure probability distribution curves of various types based on the failure mode classification results, and construct the device performance degradation index based on the stress parameter matrix, including temperature degradation index, voltage degradation index, and current degradation index;

[0109] (2.5) extracting temperature threshold parameters, voltage threshold parameters, and current threshold parameters based on the performance degradation indicator;

[0110] The specific extraction method is to construct a degradation curve based on the device performance degradation index, and extract the state parameters corresponding to the preset performance threshold from the degradation curve, including temperature threshold parameters, voltage threshold parameters, and current threshold parameters;

[0111] (2.6) Calculating the power factor deviation value and the metering accuracy deviation value using the threshold parameters;

[0112] The power factor deviation and metering accuracy deviation of the AC / DC conversion system are calculated based on the state parameters, and the degree of influence of the passive device parameter changes on the performance of the AC / DC conversion system is judged by the deviation values.

[0113] In step (2.1), the collection of historical operating data can be based on fixed time intervals, such as recording temperature and stress values ​​once every hour, to ensure the continuity and representativeness of the data;

[0114] In step (2.2), the correspondence between device stress parameters and operating time is that each row of the matrix corresponds to the parameter value at a certain time point, and each column represents a different stress dimension, such as temperature deviation, voltage ratio, and current peak ratio, thereby reflecting the stress state of the device at different operating stages;

[0115] In the failure mode analysis stage, the Gaussian mixture clustering algorithm is used to classify the device failure records in the historical operation data to obtain the distribution characteristics of various failure modes. For example, for electrolytic capacitors, failure types such as electrolyte leakage, internal short circuit and open circuit may be identified. Each type of failure mode is associated with a specific stress parameter combination. Subsequently, the probability distribution curve of each type of failure mode is calculated based on the classification results, and the performance degradation index is constructed in combination with the device stress parameter matrix. These indicators include temperature degradation factor, voltage degradation factor and current degradation factor, which are used to quantify the degree of device degradation over time. In practical applications, Gaussian mixture clustering divides the failure data into multiple clusters through iterative optimization of the expectation maximization algorithm. The center point of each cluster represents a typical failure mode, and the probability distribution curve reveals the occurrence trend of different failure modes with increasing operating time.

[0116] In step (2.3), during the mapping relationship construction phase, a recursive neural network is used to train the device stress parameter matrix and the operating parameters of the AC / DC conversion system to establish a nonlinear mapping relationship between the two. The input layer receives stress parameters such as temperature change, voltage stress value, and current stress value, and the output layer generates system performance indicators such as power factor value and metering accuracy value. Through the hidden layer processing of the network, long short-term memory units are used to capture time dependence to ensure that the model can reflect the dynamic impact of the device state evolving over time. After training is completed, the mapping relationship is used to predict the potential impact of passive device parameter changes on system performance. In this step, the structure of the recursive neural network can contain 128 hidden layer neurons, and the weights are optimized through multiple iterations to keep the prediction error within an acceptable range.

[0117] In step (2.6), the calculation of the deviation value can be obtained by running the data through multiple simulations to ensure the reliability of the judgment results.

[0118] In step (2) of the present invention, the hexagonal diagram characteristic parameters are integrated with historical data through the above steps, which can clearly reveal the causal relationship between the changes in passive device parameters and system performance. After the stress parameters are recorded in matrix form, their degradation trends can be intuitively reflected, and the prediction results of the recursive neural network further quantify the specific impact of these changes on power factor and metering accuracy. The algorithm implementation details of this step can be adjusted by technicians according to actual scenarios to meet the needs of different types of passive devices.

[0119] (3) If it is determined that the change in the parameters of the passive device has an impact on the AC / DC conversion system, a passive device degradation warning signal is generated, and the reactive compensation capacity, reactive compensation device parameters, and filter parameters are adjusted to determine the adjustment amount;

[0120] The step (3) includes the following steps:

[0121] (3.1) receiving a warning data matrix indicating that at least one parameter of the device's temperature change, voltage change, and current change exceeds a preset threshold value, and classifying the warning data matrix using a multi-layer perceptron to obtain a warning level value;

[0122] The device parameter change is determined based on the passive device degradation warning threshold. When any of the temperature change, voltage change, and current change exceeds the preset threshold range, a warning data matrix containing the change value is generated. The warning data matrix is ​​classified using a multi-layer perceptron to obtain a warning level value.

[0123] (3.2) Constructing a compensation parameter calculation matrix for the warning level value, extracting a reactive compensation capacity reference value, a compensation device reference parameter, and a compensation operating point reference value from a compensation parameter database based on the warning level value, and calculating a compensation capacity adjustment value using a reactive power compensation algorithm;

[0124] (3.3) extracting a frequency range reference value, a bandwidth reference value, and a gain reference value from a filter parameter database for the warning level value and the compensation capacity adjustment value, and obtaining a filter parameter adjustment value by performing a least squares operation;

[0125] (3.4) constructing a correction parameter matrix based on the compensation capacity adjustment value and the filter parameter adjustment value, and optimizing the correction parameter matrix using a random forest regression method to obtain a final adjustment value of the reactive compensation capacity, a final adjustment value of the compensation device parameter, and a final adjustment value of the filter parameter.

[0126] The data standardization method is used to normalize the correction parameter matrix to obtain the normalized parameter matrix. The compensation parameter optimization vector is constructed based on the normalized parameter matrix. Then, the optimization vector is solved by the random forest regression method to obtain the final adjustment values ​​of the reactive compensation capacity, the compensation device parameters, and the filter parameters.

[0127] In step (3.1), the multilayer perceptron adopts a three-layer structure. The input layer receives the change data, the hidden layer processes the nonlinear relationship through the activation function such as ReLU, and the output layer gives a warning level value from 1 to 4, indicating the degree of degradation from mild to severe.

[0128] In step (3.2), during the parameter adjustment phase, the reactive power compensation algorithm determines the specific amount of compensation capacity that needs to be increased through iterative optimization based on the difference between the target power factor value and the current value. The core is to calculate the compensation demand through real-time power factor deviation and calibrate it based on historical adjustment results to ensure that the adjustment amount is neither over-compensated nor under-compensated.

[0129] In step (3.3), during the filter optimization phase, the least squares method is used to calculate the filter parameter adjustment values ​​to adapt to the harmonic changes caused by the increase in compensation capacity. The application of the least squares method ensures that the adjusted filter parameters are closely matched to the system operating state by minimizing the sum of squared errors. In actual scenarios, the adjustment of filter parameters is closely related to reactive power compensation, aiming to suppress additional harmonic interference caused by device degradation and improve system stability.

[0130] In step (3.4), during the optimization calculation phase, random forest regression can effectively avoid the overfitting problem of a single model through multiple sampling and voting mechanisms, ensuring the robustness of the adjustment results.

[0131] In step (3) of the present invention, closed-loop control from early warning signal generation to parameter adjustment is achieved through the above steps. Taking an electrolytic capacitor as an example, when its temperature change exceeds 15 degrees Celsius and triggers an early warning, the system not only generates a corresponding signal, but also quickly classifies the degradation level through a multi-layer perceptron, combines the database reference value and algorithm operation, and dynamically adjusts the reactive compensation and filtering parameters. This adjustment can effectively deal with the impact of device degradation on power factor and metering accuracy, while avoiding conflicts or system oscillations in parameter adjustment. The implementation of this step does not require manual intervention, and all calculation processes are completed automatically, ensuring the efficiency and consistency of the adjustment. The specific parameter adjustment range can be flexibly set by technicians based on the actual equipment characteristics.

[0132] (4) Using a machine learning algorithm, the adjustment amount and historical parameter data of passive devices are used as input to predict the trend of passive device parameter changes. The compensation method, compensation capacity configuration and switching method of the reactive compensation device are determined based on the prediction results, and the power factor of the entire cluster is controlled through dynamic compensation and coordination.

[0133] The step (4) comprises the following steps:

[0134] (4.1) constructing a training data matrix based on the historical parameter data of the passive device, processing the training data matrix through a long short-term memory network, and obtaining a passive device parameter change trend prediction curve;

[0135] The training data matrix contains the time series values ​​of temperature parameters, voltage parameters, and current parameters;

[0136] (4.2) extracting a time distribution sequence of parameter changes of the passive device based on the prediction curve, constructing a parameter change matrix, wherein the parameter change matrix includes temperature changes, voltage changes, and current changes, and calculating a compensation method selection sequence number based on the parameter change matrix;

[0137] (4.3) extracting an initial configuration value of compensation capacity corresponding to the compensation mode selection number from a compensation device parameter database, optimizing the initial configuration value of compensation capacity using a Kalman filter, calculating an optimal estimate based on the measurement noise variance and the process noise variance, and obtaining an optimized configuration value of compensation capacity;

[0138] The initial configuration value of compensation capacity includes reactive compensation, harmonic compensation, and unbalance compensation;

[0139] (4.4) generating a compensation device switching schedule based on the optimized compensation capacity configuration value, the compensation device switching schedule including a switching sequence number, a switching time interval, and a switching hold time, adjusting the compensation device switching schedule in real time using an adaptive feedback compensation algorithm, and calculating a compensation deviation based on a real-time power factor measurement value;

[0140] Adjust the switching parameters of the compensation device to obtain the final compensation control instructions;

[0141] (4.5) If the overall power factor is lower than a preset threshold, the reactive compensation adjustment range is determined, and the reactive compensation device operating parameter adjustment value is calculated based on the adjustment range. The reactive compensation device operating strategy is updated using the parameter adjustment value, and the overall power factor of the cluster is iteratively updated. It is determined whether the updated overall power factor reaches the preset optimization target value. If the optimization target value is not reached, the operating strategy is updated again until the overall power factor reaches the optimization target value.

[0142] The step (4.5) includes the following steps:

[0143] (4.5.1) Collecting a real-time power factor value using a power factor measurement device and determining whether the real-time power factor value is lower than a preset power factor threshold;

[0144] (4.5.2) If the real-time power factor value is lower than a preset power factor threshold, obtaining compensation capacity parameters and operating status parameters from a compensation device parameter library;

[0145] (4.5.3) Constructing a compensation parameter optimization objective function based on the compensation capacity parameter and the operating status parameter, and solving the compensation parameter optimization objective function using a gradient descent method to obtain a reactive compensation adjustment range;

[0146] (4.5.4) Based on the reactive compensation adjustment range, an adaptive learning rate algorithm is used to optimize the compensation capacity value and the switch control value to obtain a compensation parameter adjustment matrix; the input parameters include the compensation capacity value, the switch control value, and the response time.

[0147] (4.5.5) extracting a compensation capacity correction value and a switch timing correction value from the compensation parameter adjustment matrix, and updating a compensation device control instruction sequence based on the compensation capacity correction value and the switch timing correction value;

[0148] (4.5.6) Executing reactive power compensation operation according to the compensation device control instruction sequence to collect power factor measurement values, and determining whether the power factor measurement values ​​reach a preset optimization target value.

[0149] Compare the power factor measurement value with the preset optimization target value. If the measurement value does not reach the target value and does not exceed the maximum number of iterations, return to the second step to recalculate the compensation adjustment interval until the termination condition is met.

[0150] In step (4.1), during the parameter trend prediction phase, historical parameter data for passive components is retrieved from memory to construct a multidimensional training data matrix. This matrix is ​​then modeled using a long short-term memory (LSTM) network. The network employs a three-layer structure: the input layer receives the daily parameter time series, the hidden layer is equipped with 64 neurons to capture long-term dependencies, and the output layer generates a parameter change trend prediction curve for the next seven days. The time distribution series of temperature, voltage, and current changes are then extracted from the prediction curve to construct a parameter change matrix. In this step, the LSTM network, through its forget gate and update gate mechanisms, effectively retains key patterns in the historical data, thereby improving prediction accuracy.

[0151] In step (4.2), during the compensation strategy determination phase, the compensation method selection sequence is calculated based on the parameter change matrix. When the parameters change rapidly, the fast compensation method is selected, while when the parameters change slowly, the conventional compensation method is selected.

[0152] In step (4.3), the introduction of the Kalman filter aims to smooth the measurement error and model uncertainty, ensuring that the compensation capacity is closer to the actual needs of the system;

[0153] In step (4.4), the switching timing table of the compensation device is adjusted in real time by an adaptive feedback compensation algorithm, and the compensation deviation is calculated according to the real-time measurement value of the power factor to form a final compensation control instruction to achieve smooth regulation.

[0154] In step (4) of the present invention, the entire process from parameter prediction to compensation control is automated through the above steps, which not only reduces compensation lag, but also avoids equipment shock through multi-level progressive control, significantly improving the stability and efficiency of cluster operation. The specific parameter configuration can be further adjusted according to the actual load characteristics.

[0155] (5) Based on the compensation method, compensation capacity configuration and switching method of the reactive compensation device, as well as the overall power factor of the cluster and the early warning signal of passive device degradation, the particle swarm optimization algorithm is used to coordinate the reactive compensation devices of several low-voltage metering cabinets, adjust the output capacity and operating status of the reactive compensation device, and optimize its output capacity and operating status to improve system performance;

[0156] The step (5) comprises the following steps:

[0157] (5.1) Obtaining the operating data of the compensation device of the low-voltage meter cabinet to construct an operating status matrix, wherein the operating status matrix includes the compensation mode attribute value, the capacity configuration value, and the switching mode sequence number;

[0158] (5.2) constructing a fitness function based on the operating state matrix, wherein the fitness function includes a compensation effect constraint term and a device operation constraint term;

[0159] The fitness function of the particle swarm optimization algorithm is constructed to collect cluster power factor measurements and passive device degradation warning signals.

[0160] (5.3) Parameter search of the fitness function is performed using a particle swarm optimization algorithm. The particle position represents the compensation capacity allocation ratio, the particle velocity represents the capacity adjustment step, and the search range is limited to the rated capacity range of the compensation device.

[0161] (5.4) Calculating the device load rate based on the compensation capacity allocation ratio, where the load rate is the ratio of the real-time capacity to the rated capacity, and grouping and controlling the compensation devices according to the load rate values;

[0162] (5.5) A deep neural network is used to predict the load rate. If the predicted load rate exceeds a preset threshold, a control instruction sequence of the compensation device is corrected according to the predicted load rate to obtain a corrected control instruction sequence.

[0163] Based on the results of particle swarm optimization, a control instruction sequence for the compensation device is generated, including the switching order, capacity allocation value, and response time. The operating status of each compensation device is controlled hierarchically. A sliding time window is used to record the operating data of the compensation device, calculate the device input-output characteristic curve, and construct a training dataset for a deep neural network. The deep neural network predicts the future operating status of the compensation device. The input features include the historical load rate, capacity allocation value, and operating time, and the output feature is the predicted load rate. The control instruction sequence for the compensation device is corrected based on the predicted load rate. When the predicted load rate exceeds the preset threshold, the capacity allocation value and switching order are adjusted to generate the corrected control instructions.

[0164] In step (5.1), the compensation mode attribute value includes local compensation or centralized compensation;

[0165] In step (5.2), the compensation effect constraint and the device operation constraint are designed to balance compensation efficiency and equipment safety. In this step, the fitness function is designed to integrate multiple objectives in a weighted manner to ensure that the optimization result meets the power factor requirement while avoiding overload.

[0166] In step (5.4), group control realizes hierarchical management, and the particle swarm optimization algorithm can quickly converge to a reasonable allocation solution by simulating group collaboration, thereby improving computational efficiency;

[0167] In step (5.5), a deep neural network training set is constructed based on historical operating data. The input features include 24-hour load rate, capacity allocation value, and operating time. The network adopts a three-layer structure with 64 hidden nodes. The load rate for the next four hours is predicted through nonlinear mapping. The control instruction sequence is then corrected based on the prediction results to reduce operating pressure and keep other device parameters stable. In this step, the deep neural network can identify load anomalies in advance by learning historical patterns, ensuring the timeliness and accuracy of adjustments.

[0168] In step (5) of the present invention, dynamic coordination of cluster reactive power compensation is achieved through the above steps. This method not only responds to degradation warnings, but also reduces the risk of high-load operation of equipment through algorithm collaborative optimization. The application of sliding time windows and characteristic curves further enhances the system's adaptability to load changes. The specific grouping strategy can be adjusted according to the actual scenario.

[0169] (6) Monitor the changes of the characteristic parameters of the hexagonal diagram in real time, use the time series analysis method to perform trend fitting on the characteristic parameters of the historical hexagonal diagram, determine the acceleration point of the degradation of the passive device, and adjust the operating parameters of the reactive compensation device according to the acceleration point of the degradation of the passive device to maintain the stability of the system.

[0170] The step (6) includes the following steps:

[0171] (6.1) collecting characteristic parameters of the hexagonal diagram using a real-time monitoring device, wherein the characteristic parameters include a voltage phase parameter, a current phase parameter, and a harmonic content parameter, and extracting sampling points from the characteristic parameters at fixed time intervals to obtain a characteristic parameter time series data set;

[0172] (6.2) fitting the characteristic parameter time series data set using an exponential smoothing algorithm to obtain a parameter change trend curve, calculating the slope of the trend curve to obtain a parameter change rate curve, calculating the acceleration value based on the change rate curve and marking abnormal points;

[0173] When the acceleration exceeds the preset threshold, it is marked as an outlier point, and the outlier points are clustered to obtain the initial degraded acceleration point set;

[0174] (6.3) Using a long short-term memory network to receive the parameter change rate and acceleration data, verifying the abnormal points through the parameter change law extraction unit in the network hidden layer to obtain the degradation acceleration point;

[0175] A long short-term memory network is used to train the historical sequence of characteristic parameters. The input layer receives the parameter change rate and acceleration data, the hidden layer contains a parameter change law extraction unit, and the output layer provides the parameter change prediction value. The initial set of degradation acceleration points is verified based on the network prediction results, and abnormal points that do not conform to the change law are eliminated to obtain the final position of the degradation acceleration point.

[0176] (6.4) calculating a compensation parameter adjustment amount based on the parameter change amplitude of the degradation acceleration point position, correcting the compensation device operating parameters based on the compensation parameter adjustment amount, and generating a compensation device operating parameter update instruction;

[0177] Read the compensation parameter reference value from the reactive power compensation device parameter library, calculate the compensation parameter adjustment amount according to the proportional relationship based on the parameter change amplitude at the degradation acceleration point, and modify the compensation device operating parameters based on the compensation parameter adjustment amount, including the compensation capacity increase or decrease value, switching delay adjustment value, and response rate change value, to generate a compensation device operating parameter update instruction;

[0178] (6.5) Obtain the time series data of the characteristic parameters of the historical hexagonal graph, calculate the rate of change of the characteristic parameters at each time point using the sliding window method, perform trend fitting on the rate of change of the characteristic parameters, construct a linear regression model of the time series, and obtain the slope value of the trend line. If the slope value of the trend line exceeds the preset threshold, mark the time point as a potential degradation point, calculate the goodness of fit index, perform second-order difference calculation on the potential degradation point, and determine whether the second-order derivative of the rate of change shows a turning point. If so, determine the corresponding point as the acceleration point of passive device degradation.

[0179] The step (6.5) includes the following steps:

[0180] (6.5.1) Collect historical hexagonal graph characteristic parameters from the low-voltage metering cabinet measurement device, sample the characteristic parameters using a fixed-length sliding window, and obtain a characteristic parameter change rate dataset;

[0181] The characteristic parameters of the hexagonal graph include voltage harmonic parameters, current harmonic parameters, and phase parameters. The window length is set to a multiple of the sampling period. The parameter change rate within the window is calculated to construct a characteristic parameter change rate data set.

[0182] (6.5.2) Fit the characteristic parameter change rate data set using a linear regression algorithm and calculate the trend line slope of the fitted curve. If the slope is greater than a preset change rate threshold, the sampling point is determined to be a potential outlier.

[0183] (6.5.3) Calculate the goodness-of-fit index for the potential outlier. If the coefficient of determination, standard deviation of fit, and correlation coefficient all meet the preset goodness-of-fit criteria, record the sampling point in the outlier dataset.

[0184] (6.5.4) Perform a first-order difference operation on the outlier dataset using the central difference method to obtain a first-order derivative sequence. Perform a second-order difference operation on the first-order derivative sequence to obtain a second-order derivative sequence. If the second-order derivative changes from negative to positive, the sampling point is determined to be a degradation acceleration point.

[0185] Before performing the first-order difference operation on the smooth sequence by the central difference method, spline interpolation is performed on the abnormal point data set to generate a continuous rate of change curve and construct a smooth rate of change sequence.

[0186] During the acceleration point verification and adjustment phase, a long-short-term memory network (LSTM) is used to conduct in-depth analysis of parameter change rates and accelerations. Subsequently, baseline values ​​are extracted from the reactive compensation device parameter library. Adjustments are calculated proportionally based on the magnitude of parameter changes at the acceleration point. Finally, update instructions are generated to adjust operating parameters, ensuring that the system responds promptly to degradation effects. The combination of network verification and parameter adjustment improves the accuracy and practicality of acceleration point determination.

[0187] In step (6) of the present invention, accurate monitoring and dynamic optimization of passive device degradation are achieved through the above steps. This method not only improves diagnostic efficiency by combining real-time data with historical trends, but also extends the life of the equipment through timely adjustment of operating parameters. The specific threshold and window length can be optimized according to actual needs.

Claims

1. A method for real-time monitoring of passive vector parameters during AC / DC conversion, characterized in that: The following steps are involved: (1) Collect the operating data of the AC / DC conversion system in the low-voltage metering cabinet, generate a hexagonal diagram to extract its characteristic parameters, and analyze the parameters to evaluate the status of passive components; (2) The historical operating parameters of the passive device are combined with the characteristic parameters of the hexagonal diagram to establish a mapping relationship with the AC / DC conversion system, and the failure mode of the passive device is obtained. The influence of the change of the passive device parameters on the AC / DC conversion system is judged based on the preset performance threshold; (3) If it is determined that the change in the parameters of the passive device has an impact on the AC / DC conversion system, a passive device degradation warning signal is generated, and the reactive compensation capacity, reactive compensation device parameters, and filter parameters are adjusted to determine the adjustment amount; (4) Using a machine learning algorithm, the adjustment amount and historical parameter data of passive devices are used as input to predict the trend of passive device parameter changes. The compensation method, compensation capacity configuration and switching method of the reactive compensation device are determined based on the prediction results, and the power factor of the entire cluster is controlled through dynamic compensation and coordination. (5) Based on the compensation method, compensation capacity configuration and switching method of the reactive compensation device, as well as the overall power factor of the cluster and the early warning signal of passive device degradation, the particle swarm optimization algorithm is used to coordinate the reactive compensation devices of several low-voltage metering cabinets and adjust the output capacity and operating status of the reactive compensation device; (6) Monitor the changes of the characteristic parameters of the hexagonal diagram in real time, use the time series analysis method to perform trend fitting on the characteristic parameters of the historical hexagonal diagram, determine the acceleration point of the degradation of the passive device, and adjust the operating parameters of the reactive compensation device according to the acceleration point of the degradation of the passive device.

2. The method for real-time monitoring of passive vector parameters during AC / DC conversion according to claim 1, characterized in that: The step (1) comprises the following steps: (1.1) Sampling the AC / DC conversion system of the low-voltage metering cabinet according to the preset sampling frequency to obtain AC voltage and current waveform data; (1.2) processing the AC voltage and current waveform data by Fourier transform, and extracting amplitude and phase characteristics of the fundamental signal and the harmonic signal from the waveform data; (1.3) constructing a hexagonal coordinate matrix based on the characteristics of the fundamental signal and the harmonic signal, wherein the hexagonal coordinate matrix includes voltage amplitude, current amplitude, phase difference value, fundamental frequency, total harmonic distortion rate, and grid frequency deviation value; (1.4) Performing principal component analysis on the hexagonal coordinate matrix, clustering the characteristic components obtained by the principal component analysis using a support vector machine, and obtaining evaluation indicators of monitoring parameters.

3. The method for real-time monitoring of passive vector parameters during AC / DC conversion according to claim 1, characterized in that: The step (2) includes the following steps: (2.1) Obtaining historical operation records of passive devices in a memory, wherein the historical operation records include device operation time, temperature parameters, voltage stress parameters, and current stress parameters; (2.2) constructing a device stress parameter matrix based on the historical operation records, and obtaining a failure mode classification result by using Gaussian mixture clustering on the stress parameter matrix; (2.3) using a recursive neural network to establish a mapping relationship between the stress parameter matrix and operating parameters of the AC / DC conversion system, wherein the operating parameters include a power factor value and a metering accuracy value; (2.4) Determining the degree of impact of passive components on the performance of the AC / DC conversion system, specifically including: constructing a component performance degradation index based on the failure mode classification results; (2.5) extracting temperature threshold parameters, voltage threshold parameters, and current threshold parameters based on the performance degradation indicator; (2.6) Calculate the power factor deviation value and the metering accuracy deviation value using the threshold parameters.

4. The method for real-time monitoring of passive vector parameters during AC / DC conversion according to claim 1, characterized in that: The step (3) includes the following steps: (3.1) receiving a warning data matrix indicating that at least one parameter of the device's temperature change, voltage change, and current change exceeds a preset threshold value, and classifying the warning data matrix using a multi-layer perceptron to obtain a warning level value; (3.2) extracting a reactive compensation capacity reference value, a compensation device reference parameter, and a compensation operating point reference value from a compensation parameter database according to the warning level value, and calculating a compensation capacity adjustment value using a reactive power compensation algorithm; (3.3) extracting a frequency range reference value, a bandwidth reference value, and a gain reference value from a filter parameter database for the warning level value and the compensation capacity adjustment value, and obtaining a filter parameter adjustment value by performing a least squares operation; (3.4) constructing a correction parameter matrix based on the compensation capacity adjustment value and the filter parameter adjustment value, and optimizing the correction parameter matrix using a random forest regression method to obtain a final adjustment value of the reactive compensation capacity, a final adjustment value of the compensation device parameter, and a final adjustment value of the filter parameter.

5. The method for real-time monitoring of passive vector parameters during AC / DC conversion according to claim 1, characterized in that: The step (4) comprises the following steps: (4.1) constructing a training data matrix based on the historical parameter data of the passive device, processing the training data matrix through a long short-term memory network, and obtaining a passive device parameter change trend prediction curve; (4.2) extracting a time distribution sequence of parameter changes based on the prediction curve, constructing a parameter change matrix, wherein the parameter change matrix includes temperature changes, voltage changes, and current changes, and calculating a compensation method selection sequence number based on the parameter change matrix; (4.3) extracting an initial configuration value of compensation capacity corresponding to the compensation mode selection number from a compensation device parameter database, optimizing the initial configuration value of compensation capacity using a Kalman filter, calculating an optimal estimate based on the measurement noise variance and the process noise variance, and obtaining an optimized configuration value of compensation capacity; (4.4) generating a compensation device switching schedule based on the optimized compensation capacity configuration value, the compensation device switching schedule including a switching sequence number, a switching time interval, and a switching hold time, adjusting the compensation device switching schedule in real time using an adaptive feedback compensation algorithm, and calculating a compensation deviation based on a real-time power factor measurement value; (4.5) If the overall power factor is lower than a preset threshold, the reactive compensation adjustment range is determined, and the reactive compensation device operating parameter adjustment value is calculated based on the adjustment range. The reactive compensation device operating strategy is updated using the parameter adjustment value, and the overall power factor of the cluster is iteratively updated. It is determined whether the updated overall power factor reaches the preset optimization target value. If the optimization target value is not reached, the operating strategy is updated again until the overall power factor reaches the optimization target value.

6. The method for real-time monitoring of passive vector parameters during AC / DC conversion according to claim 1, characterized in that: The step (5) comprises the following steps: (5.1) Obtaining the operating data of the compensation device of the low-voltage meter cabinet to construct an operating status matrix, wherein the operating status matrix includes the compensation mode attribute value, the capacity configuration value, and the switching mode sequence number; (5.2) constructing a fitness function based on the operating state matrix, wherein the fitness function includes a compensation effect constraint term and a device operation constraint term; (5.3) Parameter search of the fitness function is performed using a particle swarm optimization algorithm, where the particle position represents the compensation capacity allocation ratio, and the particle velocity represents the capacity adjustment step size; (5.4) Calculating the device load rate based on the compensation capacity allocation ratio, where the load rate is the ratio of the real-time capacity to the rated capacity, and grouping and controlling the compensation devices according to the load rate values; (5.5) A deep neural network is used to predict the load rate. If the predicted load rate exceeds a preset threshold, a control instruction sequence of the compensation device is corrected according to the predicted load rate to obtain a corrected control instruction sequence.

7. The method for real-time monitoring of passive vector parameters during AC / DC conversion according to claim 1, characterized in that: The step (6) includes the following steps: (6.1) collecting characteristic parameters of the hexagonal diagram using a real-time monitoring device, wherein the characteristic parameters include a voltage phase parameter, a current phase parameter, and a harmonic content parameter, and extracting sampling points from the characteristic parameters at fixed time intervals to obtain a characteristic parameter time series data set; (6.2) fitting the characteristic parameter time series data set using an exponential smoothing algorithm to obtain a parameter change trend curve, calculating the slope of the trend curve to obtain a parameter change rate curve, calculating the acceleration value based on the change rate curve and marking abnormal points; (6.3) Using a long short-term memory network to receive the parameter change rate and acceleration data, verifying the abnormal points through the parameter change law extraction unit in the network hidden layer to obtain the degradation acceleration point; (6.4) calculating a compensation parameter adjustment amount based on the parameter change amplitude of the degradation acceleration point position, correcting the compensation device operating parameters based on the compensation parameter adjustment amount, and generating a compensation device operating parameter update instruction; (6.5) Obtain the time series data of the characteristic parameters of the historical hexagonal graph, calculate the rate of change of the characteristic parameters at each time point using the sliding window method, perform trend fitting on the rate of change of the characteristic parameters, construct a linear regression model of the time series, and obtain the slope value of the trend line. If the slope value of the trend line exceeds the preset threshold, mark the time point as a potential degradation point, calculate the goodness of fit index, perform second-order difference calculation on the potential degradation point, and determine whether the second-order derivative of the rate of change shows a turning point. If so, determine the corresponding point as the acceleration point of passive device degradation.

8. The method for real-time monitoring of passive vector parameters during AC / DC conversion according to claim 5, characterized in that: The step (4.5) includes the following steps: (4.5.1) Collecting a real-time power factor value using a power factor measurement device and determining whether the real-time power factor value is lower than a preset power factor threshold; (4.5.2) If the real-time power factor value is lower than a preset power factor threshold, obtaining compensation capacity parameters and operating status parameters from a compensation device parameter library; (4.5.3) Constructing a compensation parameter optimization objective function based on the compensation capacity parameter and the operating status parameter, and solving the compensation parameter optimization objective function using a gradient descent method to obtain a reactive compensation adjustment range; (4.5.4) Based on the reactive compensation adjustment range, an adaptive learning rate algorithm is used to optimize the compensation capacity value and the switch control value to obtain a compensation parameter adjustment matrix; (4.5.5) extracting a compensation capacity correction value and a switch timing correction value from the compensation parameter adjustment matrix, and updating a compensation device control instruction sequence based on the compensation capacity correction value and the switch timing correction value; (4.5.6) Executing reactive power compensation operation according to the compensation device control instruction sequence to collect power factor measurement values, and determining whether the power factor measurement values ​​reach a preset optimization target value.

9. The method for real-time monitoring of passive vector parameters during AC / DC conversion according to claim 7, characterized in that: The step (6.5) includes the following steps: (6.5.1) Collect historical hexagonal graph characteristic parameters from the low-voltage metering cabinet measurement device, sample the characteristic parameters using a fixed-length sliding window, and obtain a characteristic parameter change rate dataset; (6.5.2) Fit the characteristic parameter change rate data set using a linear regression algorithm and calculate the trend line slope of the fitted curve. If the slope is greater than a preset change rate threshold, the sampling point is determined to be a potential outlier. (6.5.3) Calculate the goodness-of-fit index for the potential outlier. If the coefficient of determination, standard deviation of fit, and correlation coefficient all meet the preset goodness-of-fit criteria, record the sampling point in the outlier dataset. (6.5.4) Perform a first-order difference operation on the outlier data set using the central difference method to obtain a first-order derivative sequence. Perform a second-order difference operation on the first-order derivative sequence to obtain a second-order derivative sequence. If the second-order derivative changes from negative to positive, the sampling point is determined to be a degradation acceleration point.

Images