Permanent magnet synchronous motor comprehensive control method and system based on inverter

Abstract

The present invention provides a method and system for controlling a permanent magnet synchronous motor based on a frequency converter, the method comprising the following steps: obtaining system topology information of a public power system and historical operation data of all power loads; dividing all power loads into a plurality of regional load groups, and deploying regional servers for each regional load group; constructing a neural network-based fault detection model in the regional server, and training the fault detection model; obtaining real-time operation data of all power loads, determining a faulty power load having a fault in the regional load group according to the output result of the fault detection model, and marking the regional load group with the faulty power load determined as a faulty load group; for each faulty load group, combining real-time operation data and load correlation, and using the regional server to diffusely calculate with the faulty power load as the center to obtain PID control parameters of all power loads. The present invention has the effect of intelligently regulating all motor loads in the system.

Description

Technical Field

[0001] The present invention belongs to the technical field of motor control, and in particular relates to a permanent magnet synchronous motor comprehensive control method and system based on a frequency converter. Background Art

[0002] With the continuous improvement of industrial automation, more and more production lines and equipment are beginning to use permanent magnet synchronous motors as power sources. These motors are widely used in various industrial scenarios due to their high efficiency and reliability. Each motor is usually equipped with a frequency converter. The motor equipped with a frequency converter can adjust the speed according to the load demand, thereby reducing energy consumption and achieving energy saving goals. Even if some frequency converters or motors fail, the parameters of other normally running motors can be adjusted manually to keep the load demand under control and reduce energy waste.

[0003] However, as the scale and complexity of the system increase, especially in some large factories and public facilities, the power system usually contains a large number of motor loads. The motor operation data in the complex power system is large and highly correlated. When a load failure occurs in a complex power system, it is difficult to adjust all other normal loads to the optimal state to reduce energy waste by manual adjustment alone. Therefore, an intelligent control method for motors in complex power systems is urgently needed. Summary of the invention

[0004] The present invention provides a permanent magnet synchronous motor comprehensive control method and system based on a frequency converter, so as to solve the problem that when a load failure occurs in a complex power system, it is difficult to manually adjust all other normal loads to an optimal state.

[0005] In a first aspect, the present invention provides a method for controlling a permanent magnet synchronous motor based on a frequency converter, which is applied to a public power system, wherein the public power system includes a plurality of permanent magnet synchronous motors as power loads, and all the permanent magnet synchronous motors are equipped with a frequency converter and a PID controller for controlling the frequency converter. The method comprises the following steps:

[0006] Acquiring system topology information of the public power system and historical operating data of all the power loads;

[0007] Divide all the power loads into a plurality of regional load groups according to the load association in combination with the system topology information and the historical operation data, and deploy a regional server for each of the regional load groups, wherein the regional server is in communication connection with all the power loads in the regional load group;

[0008] Constructing a neural network-based fault detection model in the regional server, and training the fault detection model using historical operating data of all the power loads in the same regional load group;

[0009] Acquire the real-time operation data of all the electric loads, input the real-time operation data into the corresponding fault detection model according to the regional load group to which they belong, determine the faulty electric load in the regional load group according to the output result of the fault detection model, and mark the regional load group determined to have the faulty electric load as a faulty load group;

[0010] For each of the faulty load groups, the real-time operation data and the load correlation are combined, and the regional server is used to diffusely calculate with the faulty power load as the center to obtain the PID control parameters of all the power loads.

[0011] Optionally, the historical operation data includes historical motor operation data of the permanent magnet synchronous motor, historical inverter parameter data of the inverter and historical PID parameter data of the PID controller, and the historical motor operation data, the historical inverter parameter data and the historical PID parameter data are all time series data; the real-time operation data includes real-time motor operation data of the permanent magnet synchronous motor, real-time inverter control parameters and real-time capacity of the inverter, and real-time PID control parameters of the PID controller.

[0012] Optionally, the system topology information includes the connection relationship between all the power loads and the load location information of all the power loads, and the combining of the system topology information and the historical operation data to divide all the power loads into a plurality of regional load groups according to the load association, and respectively deploying a regional server for each of the regional load groups includes the following steps:

[0013] Selecting a plurality of the electric loads as regional center loads according to the load location information of all the electric loads, wherein a first load distance between any two of the regional center loads exceeds a preset first distance threshold;

[0014] For each of the regional center loads, traverse all other power loads and calculate the second load distances between all other power loads and the regional center load, and divide the power loads whose distances between the regional center load and the second load are less than a preset second distance threshold into a regional load group, where the second distance threshold is less than the first distance threshold;

[0015] marking all the power loads that do not belong to any of the regional load groups as loads to be divided;

[0016] For each of the regional load groups, a first load correlation between all the power loads other than the regional center load and the regional center load is calculated based on the historical operation data, and all the power loads whose first load correlation is less than a preset correlation threshold are removed from the regional load group and marked as the load to be divided;

[0017] For each of the loads to be divided, the second load correlation between the loads to be divided and each of the regional center loads is calculated based on the historical operation data, the regional load group corresponding to the regional center load with the largest second load correlation is marked as the target regional load group, and the loads to be divided are divided into the target regional load group.

[0018] Optionally, the calculating, based on the historical operation data, a first load correlation between all the power loads other than the regional center load and the regional center load comprises the following steps:

[0019] Preprocessing first historical operation data of the regional center load and second historical operation data of other regional loads in the regional load group other than the regional center load;

[0020] determining a plurality of first time lag orders between the first historical operation data and the second historical operation data by using an AICc information criterion;

[0021] respectively calculating first cross-correlation coefficients of the first historical operation data and the second historical operation data at each first time lag order;

[0022] performing a cross-correlation analysis on the first historical operation data and the second historical operation data according to all the first cross-correlation coefficients, and determining whether there is a synchronization relationship between the first historical operation data and the second historical operation data;

[0023] If the synchronization relationship does not exist between the first historical operation data and the second historical operation data, determining that the first load correlation between the central load corresponding to the region and other loads corresponding to the region is 0;

[0024] If the synchronization relationship exists between the first historical operating data and the second historical operating data, the causal relationship between the first historical operating data and the second historical operating data is determined based on the first time lag order, and the first load correlation between the central load corresponding to the area and other loads corresponding to the area is calculated.

[0025] Optionally, judging the causal relationship between the first historical operation data and the second historical operation data based on the first time lag order, and calculating a first load correlation between the central load corresponding to the region and other loads corresponding to the region comprises the following steps:

[0026] constructing a first regression analysis model between the first historical operation data and the second historical operation data based on the first time lag order;

[0027] Performing a causal regression analysis on the first historical operation data and the second historical operation data by using the first regression analysis model to obtain a first regression analysis result;

[0028] Determining whether there is a causal relationship between the first historical operation data and the second historical operation data according to the first regression analysis result;

[0029] If the causal relationship does not exist between the first historical operation data and the second historical operation data, determining that the first load correlation between the central load corresponding to the region and other loads corresponding to the region is 0;

[0030] If the causal relationship exists between the first historical operation data and the second historical operation data, a first regression analysis correlation value and a second regression analysis correlation value are calculated based on the first regression analysis result;

[0031] A normalized weighted calculation is performed on the first regression analysis correlation value and the second regression analysis correlation value to obtain a first load correlation degree between the central load corresponding to the area and other loads corresponding to the area.

[0032] Optionally, the calculating, based on the historical operation data, respectively the second load correlation between the load to be divided and each of the regional center loads comprises the following steps:

[0033] preprocessing the third historical operation data of the load to be divided;

[0034] determining a plurality of second time lag orders between the first historical operation data and the third historical operation data by using an AICc information criterion;

[0035] respectively calculating the second cross-correlation coefficients of the first historical operation data and the third historical operation data at each second time lag order;

[0036] performing a cross-correlation analysis on the first historical operation data and the third historical operation data according to all the second cross-correlation coefficients, and determining whether there is a synchronization relationship between the first historical operation data and the third historical operation data;

[0037] If the synchronization relationship does not exist between the first historical operation data and the third historical operation data, determining that the second load correlation between the corresponding regional center load and the corresponding load to be divided is 0;

[0038] If the synchronization relationship exists between the first historical operation data and the third historical operation data, the causal relationship between the first historical operation data and the third historical operation data is determined based on the second time lag order, and the second load correlation between the corresponding regional center load and the corresponding load to be divided is calculated.

[0039] Optionally, judging the causal relationship between the first historical operation data and the third historical operation data based on the second time lag order, and calculating the second load correlation between the corresponding regional center load and the corresponding load to be divided comprises the following steps:

[0040] constructing a second regression analysis model between the first historical operation data and the third historical operation data based on the second time lag order;

[0041] Performing a causal regression analysis on the first historical operation data and the third historical operation data by using the second regression analysis model to obtain a second regression analysis result;

[0042] Determining whether there is a causal relationship between the first historical operation data and the third historical operation data according to the second regression analysis result;

[0043] If the causal relationship does not exist between the first historical operation data and the third historical operation data, determining that the second load correlation between the corresponding regional center load and the corresponding load to be divided is 0;

[0044] If the causal relationship exists between the first historical operation data and the third historical operation data, a third regression analysis correlation value and a fourth regression analysis correlation value are calculated based on the second regression analysis result;

[0045] A normalized weighted calculation is performed on the third regression analysis correlation value and the fourth regression analysis correlation value to obtain a second load correlation degree between the corresponding regional center load and the corresponding load to be divided.

[0046] Optionally, constructing a neural network-based fault detection model in the regional server and using historical operation data of all the power loads in the same regional load group to train the fault detection model comprises the following steps:

[0047] For each of the regional load groups, a neural network-based fault detection model is constructed in the regional server deployed in the regional load group, wherein the fault detection model includes a motor fault detection model and an inverter fault detection model;

[0048] Using the historical motor operation data of all the electric loads in the same regional load group to complete the model training of the motor fault detection model;

[0049] Extracting model parameters of the last fully connected layer in the motor fault detection model after model training is completed;

[0050] Using the model parameters to replace the initial model parameters of the last fully connected layer in the inverter fault detection model;

[0051] After the inverter fault detection model completes parameter replacement, the model training of the inverter fault detection model is completed using the historical inverter parameter data and the historical PID parameter data of all the power loads in the same regional load group.

[0052] Optionally, combining the real-time operation data and the load correlation, and using the regional server to diffusely calculate with the faulty power load as the center to obtain the PID control parameters of all the power loads comprises the following steps:

[0053] Determine the fault area topology information of the fault load group based on the system topology information;

[0054] Taking the faulty power load as a target power load, and determining neighboring power loads of the target power load according to the fault area topology information;

[0055] Executing a PID control parameter calculation step on the neighboring power load to obtain the PID control parameters of the neighboring power load;

[0056] Repeating the neighboring power load for which the PID control parameters have been obtained as the new target power load, and repeatedly executing the above-mentioned neighboring power load determination step and parameter calculation step, until the PID control parameters of all the power loads are calculated;

[0057] The PID control parameter calculation steps are as follows:

[0058] The PID control parameters of the neighboring power load are calculated by combining the real-time operation data of the target power load and the neighboring power load, and the load correlation between the target power load and the neighboring power load.

[0059] In the second aspect, the present invention also provides a permanent magnet synchronous motor integrated control system based on a frequency converter, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, it implements the permanent magnet synchronous motor integrated control method based on the frequency converter as described in the first aspect.

[0060] The beneficial effects of the present invention are:

[0061] Compared with the traditional manual adjustment scheme in the technical background, the technical scheme of the present invention has the advantages of higher automation and intelligence, more accurate fault detection and positioning, more efficient control parameter optimization, stronger system adaptability and scalability, and higher data utilization. By introducing a neural network-based fault detection model and an automatic control system, real-time monitoring and intelligent adjustment of power loads are achieved, reducing the need for manual intervention and improving the system response speed and adjustment accuracy. The faulty load can be detected and located more accurately, ensuring that other motors in the system can be quickly adjusted to the optimal operating state when a fault occurs, thereby minimizing energy consumption and maintaining system stability. In addition, the scheme has strong adaptability and scalability through the deployment of regional servers and the application of neural network models. It can flexibly respond to changes in system scale regardless of changes, ensuring efficient operation and fault response capabilities of the system. At the same time, through the deep mining and utilization of historical operation data, not only the accuracy of the fault detection model is improved, but also rich data support is provided for control parameter optimization, thereby improving the overall control effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 It is a flow chart of a method for integrated control of a permanent magnet synchronous motor based on a frequency converter in one embodiment of the present application.

[0063] Figure 2 This is a schematic diagram of the division process of regional load groups in one implementation manner of the present application.

[0064] Figure 3 This is a topological diagram of the connection of power loads in a fault load group in one embodiment of the present application. DETAILED DESCRIPTION

[0065] The following will be combined with the drawings in the embodiments of the present application to clearly describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments in the present application belong to the scope of protection of this application.

[0066] The terms "first", "second", etc. in the specification and claims of the present application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described here, and the objects distinguished by "first", "second", etc. are generally of one type, and the number of objects is not limited. For example, the first object can be one or more. In addition, "and / or" in the specification and claims represents at least one of the connected objects, and the character " / " generally indicates that the objects associated with each other are in an "or" relationship.

[0067] Figure 1 FIG. 1 is a flow chart of a method for controlling a permanent magnet synchronous motor based on a frequency converter in one embodiment. Figure 1 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, Figure 1 At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.

[0068] The present invention discloses a method for controlling a permanent magnet synchronous motor based on a frequency converter, which is applied to a public power system. The public power system includes a plurality of permanent magnet synchronous motors as power loads, and all permanent magnet synchronous motors are equipped with a frequency converter and a PID controller for controlling the frequency converter. Figure 1 As shown, the present invention discloses a permanent magnet synchronous motor integrated control method based on a frequency converter, which specifically includes the following steps:

[0069] S101. Obtain system topology information of a public power system and historical operating data of all power loads.

[0070] Among them, the system topology information refers to the connection relationship and layout between the various power loads in the entire power system, which includes the specific locations and relationships between motors, inverters and PID controllers. The historical operation data includes the historical motor operation data of permanent magnet synchronous motors, the historical inverter parameter data of inverters and the historical PID parameter data of PID controllers. These data are all time series data, that is, data records that change over time. Specifically, the historical motor operation data may include parameters such as motor speed, torque, current and voltage, the historical parameter data of the inverter may include output frequency, voltage, current, etc., and the historical parameter data of the PID controller includes the historical values ​​of proportional, integral and differential parameters. By collecting these historical data, we can fully understand the operating status of each power load at different time points and the changes in control parameters, which provides basic data support for model training and fault detection in subsequent steps.

[0071] S102. Divide all power loads into multiple regional load groups according to load correlation in combination with system topology information and historical operation data, and deploy regional servers for each regional load group.

[0072] Among them, load correlation refers to the degree of mutual influence between different power loads in operation and control, which can usually be determined through correlation analysis in historical operation data. Specifically, statistical methods such as the Pearson correlation coefficient or a clustering algorithm based on machine learning can be used to divide power loads with high correlation into the same regional load group. The purpose of this is to centrally manage loads with strong correlation to facilitate subsequent fault detection and control optimization. Each regional load group deploys a regional server, which is connected to all power loads in the regional load group, responsible for collecting and processing real-time data in the area, and executing corresponding control strategies. Through such division and deployment, the management efficiency and fault response speed of the system can be effectively improved, and it is also convenient to realize distributed control and management.

[0073] S103. Construct a neural network-based fault detection model in the regional server, and use the historical operating data of all power loads in the same regional load group to train the fault detection model.

[0074] Among them, neural network is a machine learning model that simulates the neuron structure of the human brain. It has powerful nonlinear mapping ability and self-learning ability, and is suitable for fault detection of complex systems. Specifically, an appropriate neural network structure can be selected, such as a multi-layer perceptron (MLP), a convolutional neural network (CNN), or a recurrent neural network (RNN), and the model design and training are carried out according to the characteristics of historical operation data. During the training process, the input data is the historical operation data of the power load, including the speed, torque, current and voltage of the motor, the output frequency, voltage, current of the inverter, and the proportional, integral and differential parameters of the PID controller, etc., and the output is the prediction result of the fault state. Through the training of a large amount of historical data, the neural network model can learn to identify different features under normal and fault conditions, thereby realizing accurate detection of faults. The trained model is deployed in the regional server to monitor the operating status of the power load in the area in real time.

[0075] S104. Obtain the real-time operating data of all power loads, input the real-time operating data into the corresponding fault detection model according to the regional load group to which they belong, determine the faulty power loads in the regional load group according to the output result of the fault detection model, and mark the regional load group with the faulty power load as a faulty load group.

[0076] Among them, the real-time operation data includes the real-time motor operation data of the permanent magnet synchronous motor, the real-time inverter control parameters and real-time capacity of the inverter, and the real-time PID control parameters of the PID controller. These data are transmitted to the regional server in real time through sensors and communication networks, and input into the fault detection model for analysis. The fault detection model outputs the fault prediction result based on the characteristics of the input data. When a power load is detected as a fault, the system marks it as a faulty load and marks the regional load group where it is located as a faulty load group. The purpose of this is to quickly identify the fault location when a fault occurs, and to perform corresponding processing and adjustments to avoid fault propagation and energy waste.

[0077] S105. For each faulty load group, the real-time operation data and load correlation are combined, and the regional server is used to diffusely calculate the faulty power load to obtain the PID control parameters of all power loads.

[0078] Among them, when a regional load group is marked as a faulty load group, the regional server will recalculate the PID control parameters of all power loads in the area based on real-time operating data and load correlation. Specifically, the diffusion calculation method can be used, that is, taking the faulty power load as the center and gradually spreading to the surrounding area, adjusting the PID parameters of power loads with higher correlation to achieve the effect of overall optimization. This method can ensure that when a fault occurs, other motors in the system can be quickly adjusted to the optimal operating state, thereby minimizing energy consumption and maintaining system stability. Through the analysis and calculation of real-time data, the control parameters can be dynamically adjusted to adapt to different operating conditions and load requirements, thereby improving the operating efficiency and reliability of the system.

[0079] In one embodiment, the system topology information includes the connection relationship between all power loads and the load location information of all power loads. Step S102 specifically includes the following steps:

[0080] Selecting multiple power loads as regional center loads according to load location information of all power loads, wherein a first load distance between any two regional center loads exceeds a preset first distance threshold;

[0081] For each regional center load, traverse all other power loads and calculate the second load distance between all other power loads and the regional center load, and divide the power loads whose distance between the regional center load and the second load is less than a preset second distance threshold into a regional load group, where the second distance threshold is less than the first distance threshold;

[0082] Mark all power loads that do not belong to any regional load group as loads to be divided;

[0083] For each regional load group, a first load correlation between all power loads other than the regional center load and the regional center load is calculated based on historical operation data, and all power loads with a first load correlation less than a preset correlation threshold are removed from the regional load group and marked as loads to be divided;

[0084] For each load to be divided, the second load correlation between the load to be divided and each regional center load is calculated based on the historical operation data, the regional load group corresponding to the regional center load with the maximum second load correlation is marked as the target regional load group, and the load to be divided is divided into the target regional load group.

[0085] In this embodiment, refer to Figure 2, the coordinates of each power load can be determined by calculating the geographical location or topological location of each power load. For example, assuming that the location of the power load is represented by a two-dimensional coordinate, for each power load, the Euclidean distance between it and other power loads can be calculated. First, a power load is randomly selected as the first regional center load, such as Figure 2 The load A0 in the selected area is selected, and then the other power loads are traversed in turn, and the distance between them and the selected area center load is calculated. If the distance is greater than the preset first distance threshold, it is selected as the new area center load. Repeat this process until all power loads are traversed. The purpose of this is to ensure that the distance between the center loads of each area is far enough so that in the subsequent area division process, each area can cover a certain range and do not overlap.

[0086] Then you can continue to use the Euclidean distance as the calculation standard. For each regional center load, calculate the distance between it and other power loads, and divide the power loads whose distance is less than the preset second distance threshold into the regional load group where the regional center load is located. The purpose of this is to form a load group with the regional center load as the core, ensuring that the loads in each group are close to each other. And because the second distance threshold is less than the first distance threshold, the same power load will not be divided into different load groups. After the initial division is completed, all power loads are traversed to check whether they have been divided into a regional load group. If a power load is not divided into any regional load group, it is marked as a load to be divided, such as Figure 2 The load D is shown.

[0087] Next, for each regional load group, the first load correlation between all power loads other than the regional center load and the regional center load is calculated based on the historical operating data, and all power loads whose first load correlation is less than the preset correlation threshold are removed from the regional load group and marked as loads to be divided. In specific implementation, the load correlation can be calculated by statistical analysis and machine learning methods. Specifically, the Pearson correlation coefficient can be used to measure the correlation between two power loads. For the power loads in each regional load group, the correlation coefficient of the historical operating data between it and the regional center load is calculated. If the correlation coefficient is less than the preset correlation threshold, it is removed from the regional load group and marked as a load to be divided. The purpose of this is to ensure that the power loads in each regional load group have a high degree of correlation in operation and control, which is convenient for unified management and optimized control.

[0088] For each load to be divided, the second load correlation between the load to be divided and each regional center load is calculated based on the historical operation data, the regional load group corresponding to the regional center load with the maximum second load correlation is marked as the target regional load group, and the load to be divided is divided into the target regional load group. In specific implementation, the Pearson correlation coefficient can continue to be used as the calculation standard. For each load to be divided, the correlation coefficient of the historical operation data between it and all regional center loads is calculated, the maximum value is found, and the load to be divided is divided into the corresponding regional load group. Figure 2 As shown in the figure, if the correlation coefficients between the load D to be divided and the three regional center loads are P1=0.3, P2=0.6 and P3=0.5 respectively, it will be divided into the regional load group C with the correlation coefficient P3. The purpose of this is to ensure that each load to be divided can find the most suitable regional load group, so as to achieve optimization and efficient management of the overall system.

[0089] In one embodiment, the step of calculating the first load correlation between all power loads other than the regional center load and the regional center load based on the historical operation data specifically includes the following steps:

[0090] Preprocessing first historical operation data of the regional center load and second historical operation data of other regional loads other than the regional center load in the regional load group;

[0091] determining a plurality of first time lag orders between the first historical operating data and the second historical operating data by using an AICc information criterion;

[0092] Calculate the first cross-correlation coefficient of the first historical operation data and the second historical operation data at each first time lag order respectively;

[0093] Performing a cross-correlation analysis on the first historical operation data and the second historical operation data according to all first cross-correlation coefficients, and determining whether there is a synchronization relationship between the first historical operation data and the second historical operation data;

[0094] If there is no synchronization relationship between the first historical operation data and the second historical operation data, determining that the first load correlation between the central load of the corresponding area and other loads of the corresponding area is 0;

[0095] If there is a synchronization relationship between the first historical operating data and the second historical operating data, the causal relationship between the first historical operating data and the second historical operating data is determined based on the first time lag order, and the first load correlation between the central load of the corresponding area and other loads in the corresponding area is calculated.

[0096] In this embodiment, first, the first historical operation data of the regional center load and the second historical operation data of other regional loads other than the regional center load in the regional load group are preprocessed. The specific implementation method of this step includes operations such as data cleaning, data smoothing, data alignment, and data stationarity test. Data cleaning is to remove noise and outliers and ensure the quality of data. For example, if the load data at a certain time point deviates significantly from the normal range, it can be corrected by interpolation or mean substitution. Data smoothing can be achieved by moving average method to reduce the impact of random fluctuations on the analysis results. For example, a three-point moving average method is used to calculate the average value of each three consecutive data points and replace the data value of the middle point. Data standardization is to eliminate the influence of dimensions and make data of different loads comparable. The Z-score standardization method can be used to subtract the mean from each data point and divide it by the standard deviation so that the mean of the data is 0 and the standard deviation is 1. Data alignment can ensure that the time series data of the two motors are aligned in time, and interpolation or resampling is performed when necessary. The purpose of data stationarity test is to ensure that the time series data is stationary. Unit root tests (such as ADF tests) can be used to detect whether the data is stationary. If the data is not stationary, it can be made stationary by difference or logarithmic transformation. Through these preprocessing steps, the quality and consistency of historical operation data are ensured, providing a reliable basis for subsequent analysis.

[0097] Next, the AICc information criterion is used to determine multiple first time lag orders between the first historical operating data and the second historical operating data. The AICc information criterion is an improved Akaike information criterion, which is suitable for situations with a small sample size. In specific implementation, the stepwise regression method can be used to calculate the AICc values ​​at different lag orders respectively. Assuming that the maximum lag order is 10, the AICc values ​​of lags 1 to 10 are calculated in sequence, and the lag order that minimizes the AICc value is selected as the optimal lag order. For example, if the AICc value of lag 3 is the smallest, the optimal lag order is determined to be 3. In this way, the optimal time lag order between the first historical operating data and the second historical operating data can be effectively determined, reflecting the dynamic relationship between the two.

[0098] Then, the first cross-correlation coefficient of the first historical operation data and the second historical operation data at each first time lag order is calculated respectively. The cross-correlation coefficient is used to measure the correlation between two time series, especially the correlation at different lag times. In specific implementation, the Pearson correlation coefficient can be used for calculation. For example, assuming that the first historical operation data is X t , the second historical running data is Y t , then the cross-correlation coefficient r at lag k is k It can be expressed as:

[0099]

[0100] in and X t and Y t The mean value of . By calculating the cross-correlation coefficient under different lag orders, the change of the correlation between the two can be obtained, providing data support for the subsequent cross-correlation analysis. Then, the first historical operation data and the second historical operation data are cross-correlated according to all the first cross-correlation coefficients, and it is determined whether there is a synchronization relationship between the first historical operation data and the second historical operation data. In specific implementation, a cross-correlation function diagram can be drawn to observe the change of the cross-correlation coefficient under different lag orders. If the cross-correlation coefficient reaches a significant level (for example, exceeds a certain critical value) under a certain lag order, it is considered that there is a synchronization relationship between the two. For example, if the cross-correlation coefficient is 0.8 under lag 3 and the significance level is 0.05, it is considered that there is a synchronization relationship between the first historical operation data and the second historical operation data. In this way, the dynamic relationship between the two can be intuitively judged, providing a basis for the subsequent correlation calculation. When the cross-correlation analysis results show that the cross-correlation coefficient does not reach a significant level under all lag orders, it is considered that there is no significant dynamic relationship between the two. Therefore, the load correlation between the two can be directly set to 0. If there is a synchronization relationship between the first historical operation data and the second historical operation data, the Granger causality test method can be used to determine the causal relationship.

[0101] In one embodiment, the step of determining the causal relationship between the first historical operation data and the second historical operation data based on the first time lag order, and calculating the first load correlation between the central load of the corresponding area and other loads of the corresponding area specifically includes the following steps:

[0102] constructing a first regression analysis model between the first historical operation data and the second historical operation data based on the first time lag order;

[0103] Performing causal regression analysis on the first historical operation data and the second historical operation data by using a first regression analysis model to obtain a first regression analysis result;

[0104] Determining whether there is a causal relationship between the first historical operation data and the second historical operation data according to the first regression analysis result;

[0105] If there is no causal relationship between the first historical operation data and the second historical operation data, determining that the first load correlation between the central load of the corresponding area and other loads of the corresponding area is 0;

[0106] If there is a causal relationship between the first historical operation data and the second historical operation data, a first regression analysis correlation value and a second regression analysis correlation value are calculated based on the first regression analysis result;

[0107] A normalized weighted calculation is performed on the first regression analysis correlation value and the second regression analysis correlation value to obtain a first load correlation degree between the central load of the corresponding area and other loads in the corresponding area.

[0108] In this embodiment, a regression analysis model is constructed using the determined optimal lag order to explain the relationship between the two sets of data. Assume that the first historical operation data is X t , the second historical running data is Y t , the optimal lag order is k 0 , we can construct the following regression model:

[0109]

[0110] Among them, α is a constant term, β i is the regression coefficient, ∈ t is the error term. In specific implementation, the least squares method (0LS) can be used to estimate the regression coefficient β i The principle of the least squares method is to find the optimal regression coefficient by minimizing the sum of squares of the error terms. For example, if the lag order is 3, the model can be expressed as:

[0111] Y t =α+β 1 X t-1 +β 2 X t-2 +β 3 X t-3 +∈ t

[0112] In this way, a regression model that can reflect the impact of the first historical operating data on the second historical operating data can be constructed, providing a basis for subsequent causal analysis. Next, a causal regression analysis is performed on the first historical operating data and the second historical operating data through the first regression analysis model to obtain the first regression analysis result. The specific implementation method of this step is to use the constructed regression model to perform regression analysis to obtain the regression coefficient and statistical significance level. In specific implementation, statistical software (such as R, Python) can be used to perform regression analysis. The results of the regression analysis include the regression coefficient β i , standard error, t value and p value, etc. For example, if the regression analysis results show that β 1=0.5, standard error is 0.1, t value is 5, and p value is 0.001, which means that the regression coefficient of lag 1 is significant. In this way, the detailed results of the first regression analysis can be obtained, providing a basis for judging the causal relationship.

[0113] According to the first regression analysis result, it is determined whether there is a causal relationship between the first historical operation data and the second historical operation data. The specific implementation method of this step is to determine whether the regression coefficient is significant based on the p-value in the regression analysis result. In general, if the p-value is less than the significance level (for example, 0.05), the regression coefficient is considered significant and a causal relationship exists. In specific implementation, the significance of the regression coefficients can be tested one by one. In this way, the causal relationship between the first historical operation data and the second historical operation data can be accurately determined, providing a basis for the subsequent load correlation calculation.

[0114] If there is no causal relationship between the first historical operation data and the second historical operation data, it is considered that there is no causal relationship between the two. Therefore, the load correlation between the two can be directly set to 0. If there is a causal relationship between the first historical operation data and the second historical operation data, the first regression analysis correlation value and the second regression analysis correlation value are calculated based on the first regression analysis result. The specific implementation method of this step is to calculate the correlation value using a significant regression coefficient. In specific implementation, the t value and p value in the regression analysis results can be used as preliminary indicators of the correlation value. Provide a basis for subsequent normalized weighted calculations.

[0115] The preliminary correlation value is then normalized to be between 0 and 1. In specific implementation, the maximum and minimum normalization method can be used to subtract the minimum value from the correlation value and divide it by the maximum value. For example, if the range of all correlation values ​​is between 0 and 1, the value is directly used as the normalization result. Then, the normalized result can be weighted to obtain the final load correlation. In this way, a comprehensive load correlation can be obtained, which accurately reflects the correlation between the regional center load and other regional loads, and provides a scientific basis for load management and optimization.

[0116] In one embodiment, calculating the second load correlation between the load to be divided and the central load of each region based on the historical operation data specifically includes the following steps:

[0117] Preprocessing the third historical operation data of the load to be divided;

[0118] determining a plurality of second time lag orders between the first historical operation data and the third historical operation data by using an AICc information criterion;

[0119] respectively calculating the second cross-correlation coefficients of the first historical operation data and the third historical operation data at each second time lag order;

[0120] Performing a cross-correlation analysis on the first historical operation data and the third historical operation data according to all the second cross-correlation coefficients, and determining whether there is a synchronization relationship between the first historical operation data and the third historical operation data;

[0121] If there is no synchronization relationship between the first historical operation data and the third historical operation data, determining that the second load correlation between the corresponding regional center load and the corresponding load to be divided is 0;

[0122] If there is a synchronization relationship between the first historical operation data and the third historical operation data, the causal relationship between the first historical operation data and the third historical operation data is determined based on the second time lag order, and the second load correlation between the corresponding regional center load and the corresponding load to be divided is calculated.

[0123] In this embodiment, the specific implementation of the second load correlation calculation process may be the same as the specific implementation of the first load correlation calculation process.

[0124] In one embodiment, the step of determining the causal relationship between the first historical operation data and the third historical operation data based on the second time lag order, and calculating the second load correlation between the corresponding regional center load and the corresponding load to be divided specifically includes the following steps:

[0125] constructing a second regression analysis model between the first historical operation data and the third historical operation data based on the second time lag order;

[0126] Performing causal regression analysis on the first historical operation data and the third historical operation data through a second regression analysis model to obtain a second regression analysis result;

[0127] Determine whether there is a causal relationship between the first historical operation data and the third historical operation data according to the second regression analysis result;

[0128] If there is no causal relationship between the first historical operation data and the third historical operation data, determining that the second load correlation between the corresponding regional center load and the corresponding load to be divided is 0;

[0129] If there is a causal relationship between the first historical operation data and the third historical operation data, a third regression analysis correlation value and a fourth regression analysis correlation value are calculated based on the second regression analysis result;

[0130] A normalized weighted calculation is performed on the third regression analysis correlation value and the fourth regression analysis correlation value to obtain a second load correlation degree between the corresponding regional center load and the corresponding load to be divided.

[0131] In this embodiment, the specific implementation of the second load correlation calculation process may be the same as the specific implementation of the first load correlation calculation process.

[0132] In one implementation manner, step S103 specifically includes the following steps:

[0133] For each regional load group, a neural network-based fault detection model is constructed in a regional server deployed by the regional load group, and the fault detection model includes a motor fault detection model and an inverter fault detection model;

[0134] The model training of the motor fault detection model is completed using the historical motor operation data of all electric loads in the same regional load group;

[0135] Extract the model parameters of the last fully connected layer in the motor fault detection model after model training is completed;

[0136] Use the model parameters to replace the initial model parameters of the last fully connected layer in the inverter fault detection model;

[0137] After the inverter fault detection model completes parameter replacement, the model training of the inverter fault detection model is completed using the historical inverter parameter data and historical PID parameter data of all power loads in the same regional load group.

[0138] In this embodiment, it is first necessary to build a regional server and deploy a neural network framework, such as TensorFlow or PyTorch, on the server. Both the motor fault detection model and the inverter fault detection model are multi-layer neural network models, including an input layer, several hidden layers, and an output layer. The input layer receives the operating data of the motor or inverter, the hidden layer performs feature extraction and pattern recognition through a nonlinear activation function (such as ReLU), and the output layer gives the fault detection result. The specific structure of the model can be adjusted according to actual needs, such as using a convolutional neural network (CNN) to process time series data, or using a long short-term memory network (LSTM) to capture time dependencies. In this way, a neural network model that can automatically detect motor and inverter faults is established, providing a basis for subsequent model training and fault detection.

[0139] The model training of the motor fault detection model is completed using the historical motor operation data of all electrical loads in the same regional load group. In specific implementation, it is first necessary to collect the historical motor operation data of all electrical loads in the regional load group, which include current, voltage, speed, temperature, etc. After the data collection is completed, data preprocessing is required. Then, the preprocessed data is divided into a training set and a test set. The training set is used for model training, and the test set is used to verify the model performance. During the model training process, the back propagation algorithm and gradient descent method are used to update the model parameters to minimize the loss function (such as mean square error). After the training is completed, the model performance is evaluated through the test set, and indicators such as accuracy, recall rate and F1 value are calculated. In this way, a neural network model that can accurately detect motor faults can be trained.

[0140] After the motor fault detection model is trained, the model parameters of the last fully connected layer need to be extracted. A fully connected layer means that each neuron is connected to all neurons in the previous layer, and the parameters include weight matrix and bias vector. These parameters can be extracted using the API provided by the neural network framework. After extracting the parameters of the last fully connected layer of the motor fault detection model, these parameters need to be replaced in the corresponding layer of the inverter fault detection model. The replaced inverter fault detection model will share the last layer parameters of the motor fault detection model, so that the features learned in the motor fault detection model can be used to improve the initial performance of the inverter fault detection model. In this way, the knowledge of the motor fault detection model can be effectively transferred to the inverter fault detection model, providing a good starting point for subsequent model training.

[0141] After the inverter fault detection model completes parameter replacement, the model training of the inverter fault detection model is completed using the historical inverter parameter data and historical PID parameter data of all power loads in the same regional load group. In specific implementation, it is first necessary to collect the historical inverter parameter data and historical PID parameter data of all power loads in the regional load group, which include output voltage, output current, frequency, PID control parameters, etc. After the data collection is completed, data preprocessing is required. Then, the preprocessed data is divided into a training set and a test set, and the training set is used for model training, and the test set is used to verify the model performance. During the model training process, the back propagation algorithm and gradient descent method are used to update the model parameters to minimize the loss function. After the training is completed, the model performance is evaluated through the test set, and indicators such as accuracy, recall rate and F1 value are calculated. In this way, a neural network model that can accurately detect inverter faults can be trained.

[0142] In one embodiment, the step of combining the real-time operation data and the load correlation in step S105 and using the regional server to diffusely calculate and obtain the PID control parameters of all power loads with the fault power load as the center specifically includes the following steps:

[0143] Determine fault area topology information of the faulty load group based on the system topology information;

[0144] Taking the faulty power load as the target power load, and determining the neighboring power loads of the target power load according to the topological information of the faulty area;

[0145] Execute a PID control parameter calculation step on the neighbor power load to obtain the PID control parameter of the neighbor power load;

[0146] Repeating the neighboring power loads for which PID control parameters have been obtained as new target power loads, and repeatedly executing the above-mentioned neighboring power load determination step and parameter calculation step, until the PID control parameters of all power loads are calculated;

[0147] The specific steps for calculating PID control parameters are as follows:

[0148] The PID control parameters of the neighboring power load are calculated by combining the real-time operating data of the target power load and the neighboring power load, as well as the load correlation between the target power load and the neighboring power load.

[0149] In this embodiment, refer to Figure 3 First, obtain the topological information of the entire power system, which usually includes the connection relationship between power loads, the configuration of transmission lines, the location of substations, etc. This data can be obtained through the SCADA (Supervisory Control and Data Acquisition System) or GIS (Geographic Information System) of the power system. In the event of a fault, the location and type of the fault point can be monitored in real time through the SCADA system. After determining the faulty load group, it is necessary to analyze the topological structure of the load group in the system, clarify its connection relationship with other load groups, and the scope of the fault's impact on the entire system. In this way, the fault area can be accurately located, providing a basis for subsequent fault handling.

[0150] The faulty power load is taken as the target power load, and the neighboring power loads of the target power load are determined according to the topological information of the faulty area. In specific implementation, after determining the topological information of the faulty area, the faulty power load needs to be taken as the target power load, such as Figure 3 The target power load refers to the load that needs to be monitored and processed at present. According to the topology information, the neighboring power loads directly connected to the target power load can be determined, such as Figure 3The neighboring power load refers to the load that has a direct electrical connection with the target power load in the power network.

[0151] The PID control parameter calculation step is performed on the neighbor power load to obtain the PID control parameters of the neighbor power load. In the specific implementation, it is necessary to calculate the PID control parameters of the neighbor power load according to the real-time operation data and historical data of the neighbor power load, combined with the load correlation between the target power load and the neighbor power load. The real-time operation data includes the real-time motor operation data of the permanent magnet synchronous motor, the real-time inverter control parameters of the inverter and the real-time capacity of the inverter, and the real-time PID control parameters of the PID controller. The real-time operation data of the permanent magnet synchronous motor includes the speed, current, voltage, etc. of the motor; the real-time control parameters of the inverter include the output frequency, voltage, etc.; the real-time capacity of the inverter refers to the maximum power that the inverter can currently provide; the real-time PID control parameters of the PID controller include three parameters: proportional (P), integral (I) and differential (D). The PID parameters are adjusted by an optimization algorithm (such as a genetic algorithm and a particle swarm optimization algorithm).

[0152] Specifically, the dynamic behavior of the permanent magnet synchronous motor and the inverter is first described by a mathematical model. The state space model or the transfer function model is usually used to represent the dynamic characteristics of the system. The state space model describes the dynamic behavior of the system through a set of state variables, while the transfer function model describes the dynamic characteristics of the system through the input-output relationship. These models can be established through experimental data and theoretical analysis, and the model parameters are estimated and calibrated through system identification methods (such as least squares method, Kalman filtering, etc.). The purpose of establishing a system model is to accurately predict the dynamic behavior of the system during the control and optimization process, so as to achieve precise control.

[0153] The performance of the PID controller is then measured by the objective function. The objective function usually includes aspects such as steady-state error, response speed, overshoot, and smoothness of the control signal. The steady-state error is the error of the system in steady state, the response speed is the time required for the system to reach steady state, the overshoot is the maximum amplitude of the system response exceeding the target value, and the smoothness of the control signal is the rate of change of the control signal. The objective function can be expressed as a weighted sum, where the weight of each indicator is adjusted according to the specific application scenario. The purpose of the objective function is to comprehensively consider various performance indicators in the control and optimization process to achieve the best control effect of the system.

[0154] Then, the PID control parameters are adjusted through the optimization algorithm. Genetic algorithm and particle swarm optimization algorithm are two commonly used optimization algorithms. Genetic algorithm gradually optimizes PID parameters by simulating natural selection and genetic variation process. The specific steps include initializing the population, fitness evaluation, selection, crossover and mutation. Particle swarm optimization algorithm gradually optimizes PID parameters by simulating the intelligent behavior of the group. The specific steps include initializing the particle swarm, speed and position update, fitness evaluation and optimal position update. The selection and implementation of the optimization algorithm need to be adjusted according to the specific application scenario and system characteristics to achieve the best optimization effect.

[0155] During real-time operation, the PID control parameters are continuously adjusted through the optimization algorithm to ensure that the system can maintain stable and efficient operation under different working conditions. The operating status of the system is monitored in real time to verify the effectiveness of the optimized PID parameters. The purpose of real-time adjustment of PID parameters is to dynamically adapt to system changes during system operation and achieve the best control effect. Through real-time monitoring and feedback, problems in the system can be discovered and corrected in a timely manner to improve the robustness and adaptability of the system. The optimized PID parameters are applied to the actual control system to ensure that the system can still maintain stable operation under fault conditions. Through real-time monitoring and feedback, the PID parameters are further optimized and adjusted to improve the robustness and adaptability of the system. In practical applications, it is necessary to consider the actual operating environment and working conditions of the system, reasonably set the parameters and objective function of the optimization algorithm, and ensure that the system can maintain the best control effect under various operating conditions. Through the above steps, efficient control of permanent magnet synchronous motors and inverters can be achieved, and the performance and reliability of the system can be improved.

[0156] Repeat the neighbor power loads that have obtained PID control parameters as new target power loads, and repeat the above-mentioned neighbor power load determination steps and parameter calculation steps until the PID control parameters of all power loads are calculated. In specific implementation, after calculating the PID control parameters of the initial neighbor power loads, these neighbor power loads need to be used as new target power loads, and the neighbor power load determination steps and parameter calculation steps need to be continued. This process is an iterative process until the PID control parameters of all power loads are calculated. In each iteration, it is necessary to re-determine the neighbor power loads (such as load a2) of the new target power load (such as load a1), and calculate its PID control parameters based on the real-time operating data and historical data of these loads, combined with the load correlation. In this way, all loads of the entire power system can be gradually covered to ensure that in the event of a fault, the PID control parameters of all loads can be reasonably adjusted, thereby improving the stability and reliability of the system.

[0157] The specific steps for calculating the PID control parameters are as follows: Combine the real-time operating data of the target power load and the neighboring power load, as well as the load correlation between the target power load and the neighboring power load, and calculate the PID control parameters of the neighboring power load. The least squares method can be used to fit the relationship between the error signal and the control output to determine the PID parameters. When calculating specifically, the following formula can be used:

[0158]

[0159] Among them, J is the objective function, e j is the error signal at the jth moment, K p , K j and K d is the PID parameter to be solved. By optimizing the objective function J, the optimal PID control parameters can be obtained. In this way, it can be ensured that the calculated PID parameters can effectively control the operating state of the neighboring power load and improve the stability and reliability of the system.

[0160] The present invention also discloses a permanent magnet synchronous motor integrated control system based on a frequency converter, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the permanent magnet synchronous motor integrated control method based on the frequency converter described in any one of the above-mentioned embodiments is implemented.

[0161] Among them, the processor can adopt a central processing unit (CPU). Of course, according to actual usage, other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. can also be adopted. The general-purpose processor can adopt a microprocessor or any conventional processor, etc., and this application does not impose any restrictions on this.

[0162] Among them, the memory can be an internal storage unit of a computer device, such as a hard disk or memory of a computer device, or an external storage device of a computer device, such as a plug-in hard disk, a smart memory card (SMC), a secure digital card (SD) or a flash memory card (FC) equipped on the computer device, etc., and the memory can also be a combination of an internal storage unit and an external storage device of a computer device. The memory is used to store computer programs and other programs and data required by the computer device. The memory can also be used to temporarily store data that has been output or is to be output, and this application does not impose any restrictions on this.

[0163] A person skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of protection of the present application is limited to these examples. In line with the concept of the present application, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of different aspects of one or more embodiments of the present application as above, which are not provided in detail for the sake of simplicity.

[0164] One or more embodiments of the present application are intended to cover all such substitutions, modifications and variations that fall within the broad scope of the present application. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of one or more embodiments of the present application should be included in the protection scope of the present application.

Claims

1. A permanent magnet synchronous motor comprehensive control method based on a frequency converter, characterized in that: Applied to a public power system, the public power system includes a plurality of permanent magnet synchronous motors as power loads, all of the permanent magnet synchronous motors are equipped with a frequency converter and a PID controller for controlling the frequency converter, the method comprises the following steps: Obtaining system topology information of the public power system and historical operation data of all the power loads, wherein the historical operation data includes historical motor operation data of the permanent magnet synchronous motor, historical inverter parameter data of the inverter, and historical PID parameter data of the PID controller, wherein the historical motor operation data, the historical inverter parameter data, and the historical PID parameter data are all time series data, and the system topology information includes connection relationships between all the power loads and load position information of all the power loads; Selecting a plurality of the electric loads as regional center loads according to the load location information of all the electric loads, wherein a first load distance between any two of the regional center loads exceeds a preset first distance threshold; For each of the regional center loads, traverse all other power loads and calculate the second load distances between all other power loads and the regional center load, and divide the power loads whose distances between the regional center load and the second load are less than a preset second distance threshold into a regional load group, where the second distance threshold is less than the first distance threshold; marking all the power loads that do not belong to any of the regional load groups as loads to be divided; For each of the regional load groups, a first load correlation between all the power loads other than the regional center load and the regional center load is calculated based on the historical operation data, and all the power loads whose first load correlation is less than a preset correlation threshold are removed from the regional load group and marked as the load to be divided; For each of the loads to be divided, respectively calculate the second load association between the load to be divided and each of the regional center loads based on the historical operation data, mark the regional load group where the regional center load with the largest second load association is located as a target regional load group, and divide the load to be divided into the target regional load group; A regional server is deployed for each of the regional load groups respectively, and the regional server is communicatively connected with all the power loads in the regional load group; Constructing a neural network-based fault detection model in the regional server, and training the fault detection model using historical operating data of all the power loads in the same regional load group; Acquire the real-time operation data of all the electric loads, input the real-time operation data into the corresponding fault detection model according to the regional load group to which they belong, determine the faulty electric load in the regional load group according to the output result of the fault detection model, and mark the regional load group determined to have the faulty electric load as a faulty load group; For each of the faulty load groups, the real-time operation data and the load correlation are combined, and the regional server is used to diffusely calculate with the faulty power load as the center to obtain the PID control parameters of all the power loads.

2. The permanent magnet synchronous motor integrated control method based on frequency converter according to claim 1 is characterized in that: The real-time operation data includes the real-time motor operation data of the permanent magnet synchronous motor, the real-time inverter control parameters and real-time inverter capacity of the inverter, and the real-time PID control parameters of the PID controller.

3. The permanent magnet synchronous motor integrated control method based on frequency converter according to claim 1 is characterized in that: The calculating, based on the historical operation data, the first load correlation between all the power loads other than the regional center load and the regional center load comprises the following steps: Preprocessing first historical operation data of the regional center load and second historical operation data of other regional loads in the regional load group other than the regional center load; determining a plurality of first time lag orders between the first historical operation data and the second historical operation data by using an AICc information criterion; respectively calculating first cross-correlation coefficients of the first historical operation data and the second historical operation data at each first time lag order; performing a cross-correlation analysis on the first historical operation data and the second historical operation data according to all the first cross-correlation coefficients, and determining whether there is a synchronization relationship between the first historical operation data and the second historical operation data; If the synchronization relationship does not exist between the first historical operation data and the second historical operation data, determining that the first load correlation between the central load corresponding to the region and other loads corresponding to the region is 0; If the synchronization relationship exists between the first historical operating data and the second historical operating data, the causal relationship between the first historical operating data and the second historical operating data is determined based on the first time lag order, and the first load correlation between the central load corresponding to the area and other loads corresponding to the area is calculated.

4. The permanent magnet synchronous motor integrated control method based on frequency converter according to claim 3 is characterized in that: The determining the causal relationship between the first historical operation data and the second historical operation data based on the first time lag order, and calculating the first load correlation between the central load corresponding to the region and other loads corresponding to the region comprises the following steps: constructing a first regression analysis model between the first historical operation data and the second historical operation data based on the first time lag order; Performing a causal regression analysis on the first historical operation data and the second historical operation data by using the first regression analysis model to obtain a first regression analysis result; Determining whether there is a causal relationship between the first historical operation data and the second historical operation data according to the first regression analysis result; If the causal relationship does not exist between the first historical operation data and the second historical operation data, determining that the first load correlation between the central load corresponding to the region and other loads corresponding to the region is 0; If the causal relationship exists between the first historical operation data and the second historical operation data, a first regression analysis correlation value and a second regression analysis correlation value are calculated based on the first regression analysis result; A normalized weighted calculation is performed on the first regression analysis correlation value and the second regression analysis correlation value to obtain a first load correlation degree between the central load corresponding to the area and other loads corresponding to the area.

5. The permanent magnet synchronous motor integrated control method based on frequency converter according to claim 3 is characterized in that: The step of respectively calculating the second load correlation between the load to be divided and each of the regional center loads based on the historical operation data comprises the following steps: preprocessing the third historical operation data of the load to be divided; determining a plurality of second time lag orders between the first historical operation data and the third historical operation data by using an AICc information criterion; respectively calculating the second cross-correlation coefficients of the first historical operation data and the third historical operation data at each second time lag order; performing a cross-correlation analysis on the first historical operation data and the third historical operation data according to all the second cross-correlation coefficients, and determining whether there is a synchronization relationship between the first historical operation data and the third historical operation data; If the synchronization relationship does not exist between the first historical operation data and the third historical operation data, determining that the second load correlation between the corresponding regional center load and the corresponding load to be divided is 0; If the synchronization relationship exists between the first historical operation data and the third historical operation data, the causal relationship between the first historical operation data and the third historical operation data is determined based on the second time lag order, and the second load correlation between the corresponding regional center load and the corresponding load to be divided is calculated.

6. The permanent magnet synchronous motor integrated control method based on frequency converter according to claim 5 is characterized in that: The determining the causal relationship between the first historical operation data and the third historical operation data based on the second time lag order, and calculating the second load correlation between the corresponding regional center load and the corresponding load to be divided comprises the following steps: constructing a second regression analysis model between the first historical operation data and the third historical operation data based on the second time lag order; Performing a causal regression analysis on the first historical operation data and the third historical operation data by using the second regression analysis model to obtain a second regression analysis result; Determining whether there is a causal relationship between the first historical operation data and the third historical operation data according to the second regression analysis result; If the causal relationship does not exist between the first historical operation data and the third historical operation data, determining that the second load correlation between the corresponding regional center load and the corresponding load to be divided is 0; If the causal relationship exists between the first historical operation data and the third historical operation data, a third regression analysis correlation value and a fourth regression analysis correlation value are calculated based on the second regression analysis result; A normalized weighted calculation is performed on the third regression analysis correlation value and the fourth regression analysis correlation value to obtain a second load correlation degree between the corresponding regional center load and the corresponding load to be divided.

7. The permanent magnet synchronous motor integrated control method based on frequency converter according to claim 2 is characterized in that: The step of constructing a neural network-based fault detection model in the regional server and training the fault detection model using the historical operation data of all the power loads in the same regional load group comprises the following steps: For each of the regional load groups, a neural network-based fault detection model is constructed in the regional server deployed in the regional load group, wherein the fault detection model includes a motor fault detection model and an inverter fault detection model; Using the historical motor operation data of all the electric loads in the same regional load group to complete the model training of the motor fault detection model; Extracting model parameters of the last fully connected layer in the motor fault detection model after model training is completed; Using the model parameters to replace the initial model parameters of the last fully connected layer in the inverter fault detection model; After the inverter fault detection model completes parameter replacement, the model training of the inverter fault detection model is completed using the historical inverter parameter data and the historical PID parameter data of all the power loads in the same regional load group.

8. The inverter-based permanent magnet synchronous motor comprehensive control method according to claim 2 is characterized in that: The step of combining the real-time operation data and the load correlation and using the regional server to diffusely calculate and obtain the PID control parameters of all the power loads with the faulty power load as the center comprises the following steps: Determine the fault area topology information of the fault load group based on the system topology information; Taking the faulty power load as a target power load, and determining neighboring power loads of the target power load according to the fault area topology information; Executing a PID control parameter calculation step on the neighboring power load to obtain the PID control parameters of the neighboring power load; Repeating the neighboring power load for which the PID control parameters have been obtained as the new target power load, and repeatedly executing the above-mentioned neighboring power load determination step and parameter calculation step, until the PID control parameters of all the power loads are calculated; The PID control parameter calculation steps are as follows: The PID control parameters of the neighboring power load are calculated by combining the real-time operation data of the target power load and the neighboring power load, and the load correlation between the target power load and the neighboring power load.

9. A permanent magnet synchronous motor integrated control system based on a frequency converter, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the permanent magnet synchronous motor comprehensive control method based on the inverter is implemented as described in any one of claims 1 to 8.

Images