A motor operation data analysis method based on intelligent perception

Abstract

The present invention relates to the field of motor control technology, specifically a method for analyzing motor operation data based on intelligent perception, which is as follows: obtaining the original operating parameters of the motor and preprocessing the original operating parameters; establishing an evaluation model through the preprocessed original operating parameters to obtain a characteristic vector; analyzing the characteristic vector and determining the control parameters to be adjusted corresponding to the motor based on the analysis results; setting a calculation node based on the position of the motor, and adjusting the control parameters of the motor based on the control parameters to be adjusted corresponding to the motor. The present invention can accurately predict the operating status and potential faults of the motor, realize early warning and health management, and at the same time, flexibly select the type of mathematical model according to the specific structure and operating environment of the motor, and dynamically adjust the model parameters, thereby ensuring the high adaptability and flexibility of the control strategy.

Description

Technical Field

[0001] The present invention relates to the field of motor control technology, and in particular to a motor operation data analysis method based on intelligent perception. Background Art

[0002] Intelligence and networking have become mainstream trends in the development of the manufacturing industry. As core components of industrial automation and intelligent equipment, the operating efficiency and stability of motors directly impact the performance and energy consumption of the entire system. Traditional motor monitoring methods rely primarily on regular physical inspections and basic sensor monitoring, but this approach has many limitations:

[0003] 1) Monitoring limitations: Existing monitoring methods mostly focus on basic electrical parameters such as motor current and voltage. They do not adequately monitor non-electrical parameters such as vibration and temperature that reflect the motor's health, making it impossible to fully assess the motor's actual operating status.

[0004] 2) Data isolation: The collected data is often isolated point measurements, lacking continuity and systematicity. This makes it difficult to effectively judge the motor's operating trends and can easily lead to missed opportunities for early fault warning.

[0005] 3) Analysis lag: Data processing and analysis usually rely on manual intervention or simple algorithms, which are slow and cannot provide real-time feedback on motor operation anomalies, easily affecting the speed of fault response and maintenance efficiency.

[0006] 4) Insufficient energy efficiency optimization: There is a lack of intelligent energy consumption analysis tools. Optimization of motor operating efficiency mainly relies on empirical adjustments, making it difficult to achieve refined management and resulting in energy waste.

[0007] 5) High maintenance costs: Due to the lack of predictive maintenance strategies, motor failures often lead to sudden downtime, which makes repair costs high and affects production plans, thereby increasing the company's operating costs.

[0008] Therefore, there is an urgent need for a more advanced and intelligent motor operation data analysis method to address the limitations of traditional motor monitoring methods. Summary of the Invention

[0009] The purpose of the present invention is to provide a motor operation data analysis method based on intelligent perception to solve the problems raised in the above background technology.

[0010] The technical solution of the present invention is: a motor operation data analysis method based on intelligent perception, the analysis method is specifically as follows:

[0011] Acquiring original operating parameters of the motor and preprocessing the original operating parameters;

[0012] Establishing an evaluation model and obtaining a feature vector by preprocessing the original operating parameters;

[0013] Analyzing the characteristic vector and determining the control parameters to be adjusted corresponding to the motor according to the analysis result;

[0014] A calculation node is set according to the position of the motor, and the control parameters of the motor are adjusted by using the control parameters to be adjusted corresponding to the motor.

[0015] Furthermore, the original operating parameters are pre-processed as follows:

[0016] Classify the original data through a data intelligent screening algorithm to determine the original data to be transmitted;

[0017] The original data to be transmitted is compressed using a compression algorithm.

[0018] Furthermore, the feature vector is obtained as follows:

[0019] Determining optimal hyperparameters from the pre-processed original operating parameters using a Bayesian optimization method;

[0020] Establish an evaluation model through time series analysis algorithm;

[0021] The feature vector is obtained according to the optimal hyperparameter and the evaluation model.

[0022] Furthermore, the evaluation model is established as follows:

[0023] Sorting the pre-processed original operating parameters in chronological order;

[0024] The model type is determined according to the historical operating parameters of the motor, and the evaluation model is constructed according to the sorted original operating parameters.

[0025] Furthermore, the evaluation model is tested for applicability through a residual analysis algorithm, and the evaluation model is tested for generalization ability through a cross-validation algorithm.

[0026] Furthermore, the feature vector is analyzed as follows:

[0027] Normalizing the eigenvectors and determining a final eigenvector based on the operating conditions of the motor;

[0028] A learning model is trained using the final feature vector, and a threshold is set in the learning model to classify the operating state and fault level of the motor;

[0029] According to the divided operating state and fault level of the motor, the characteristic vector is analyzed, specifically:

[0030] When the motor is in a normal working state, the operating parameters of the motor remain unchanged; otherwise, the control parameters of the motor are adjusted through an adjustment strategy.

[0031] Furthermore, the control parameters of the motor are adjusted as follows:

[0032] Determining a characteristic vector to be adjusted according to the operating state and fault level of the motor;

[0033] Establishing a mathematical model of the motor operation;

[0034] Setting a calculation node according to the motor position;

[0035] The motor parameters are dynamically adjusted through the characteristic vector to be adjusted, the calculation node, and the mathematical model.

[0036] Furthermore, a mathematical model of the motor operation is established as follows:

[0037] Determining the type corresponding to the mathematical model according to the motor type;

[0038] The feature vector to be adjusted is converted into a state space form, specifically:

[0039] ;

[0040] in: and is the state equation, and is the output equation, is the inductor, is the externally applied voltage, is the back electromotive force coefficient, is the angular velocity of the motor, is the armature resistance, is the current in the motor winding, is the moment of inertia of the motor, The torque generated for the external load;

[0041] The characteristic vector to be adjusted, which is converted into a state space form, is input into the mathematical model to establish a mathematical model of the motor operation.

[0042] Furthermore, the computing nodes include:

[0043] Real-time data processing module, used to obtain and process the operating parameters of the motor;

[0044] Intelligent analysis module, used to identify the motor's operating characteristic vector;

[0045] The control strategy adjustment module is used to dynamically adjust the motor parameters according to the motor control parameters output by the mathematical model.

[0046] Furthermore, the intelligent analysis module identifies, predicts and classifies the operating parameters of the motor through a machine learning model.

[0047] The present invention provides a motor operation data analysis method based on intelligent perception through improvement, which has the following improvements and advantages compared with the existing technology:

[0048] First, the present invention utilizes a big data analysis algorithm to identify eigenvectors, combined with time series analysis and Bayesian optimization, to accurately predict the motor's operating status and potential faults, enabling early warning and health management. Furthermore, the system flexibly selects mathematical models and dynamically adjusts model parameters based on the motor's specific structure and operating environment, ensuring the control strategy's high adaptability and flexibility.

[0049] Second: The present invention uses a multi-dimensional sensor array to monitor the motor operating parameters in real time, and combines it with intelligent data screening and compression algorithms to effectively reduce the burden of data transmission and processing, improve the efficiency of data analysis and real-time response capabilities, and at the same time ensure the accuracy and generalization ability of the evaluation model through residual analysis and cross-validation, so that the system can still maintain good stability and reliability when facing different working conditions and environmental changes. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] The present invention will be further explained below in conjunction with the accompanying drawings and embodiments:

[0051] Figure 1 It is a flow chart of the motor operation data analysis method of the present invention. DETAILED DESCRIPTION

[0052] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0053] It should be noted that, in the description of the present invention, the terms "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating the orientation or position relationship, are based on the orientation or position relationship shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description. They do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention.

[0054] Furthermore, it should be understood that for the sake of ease of description, the sizes of the various components shown in the drawings are not drawn according to actual proportions. For example, the thickness or width of certain layers may be exaggerated relative to other layers.

[0055] It should be noted that like numbers and letters represent similar items in the following figures, so once an item is defined or described in one figure, it will not need to be further discussed and described in detail in the description of the subsequent figures.

[0056] refer to Figure 1 This embodiment provides a motor operation data analysis method based on intelligent perception, and the motor operation data analysis method specifically includes the following steps:

[0057] Step S1: According to the structure and operating environment of the motor, a multi-dimensional sensor array is set on the motor to obtain the original operating parameters of the motor. It is worth noting that the original operating parameters of the motor include but are not limited to the current parameters, voltage parameters, vibration parameters and temperature parameters when the motor is running. However, in this embodiment, the current parameters, voltage parameters, vibration parameters and temperature parameters when the motor is running are used as an example for specific explanation. Specifically, when obtaining the current and voltage parameters of the motor, Hall effect current sensors and voltage sensors can be selected, which can perform contactless measurements, thereby reducing measurement errors. At the same time, the current sensor should be installed on the power supply line of the motor, and the voltage sensor can be directly connected to the input terminal of the motor. When obtaining the vibration parameters of the motor, an accelerometer or vibration sensor, such as a piezoelectric, magnetoelectric or capacitive sensor, can be selected and installed on the motor housing, bearing seat or other key positions so that it can monitor the vibration frequency and amplitude when the motor is running. When obtaining the temperature parameters of the motor, a thermocouple, resistance temperature detector (RTD) or infrared temperature sensor can be selected and installed on the motor winding, bearing and housing surface to monitor the temperature rise when the motor is running.

[0058] After obtaining the original operating parameters of the motor through the multi-dimensional sensor array, the obtained original operating parameters are pre-processed. The details are as follows:

[0059] Step S1.1: Classify the raw data using an intelligent data screening algorithm. Specifically, the importance and transmission requirements of the motor's operating parameters are determined based on the motor's actual operating status to define data screening criteria. For example, the required operating data, such as current, voltage, vibration, and temperature parameters, can be determined based on parameter fluctuations, historical fault correlations, or the sensitivity of the current operating mode. A preliminary analysis of the classified raw data is then performed to extract key features, such as time series trends, anomalies, and periodic patterns, to facilitate subsequent classification.

[0060] By using the decision tree classification model and the motor's historical operating data, the required data points and time periods are identified, and a final decision tree classification model is established to filter the motor data. The classified raw data is then input into the final decision tree classification model, which automatically identifies and filters the raw data to be transmitted.

[0061] Step S1.2: Based on the characteristics of the raw data to be transmitted, automatically identified and filtered in step S1.1, such as its continuity, repetitiveness, and structure, a corresponding compression algorithm, such as lossless or lossy, is determined. The filtered raw data to be transmitted is then preprocessed, such as by quantization and rearrangement, to improve compression efficiency. It is worth noting that for time series data, differential encoding can be performed first to reduce redundancy between data before compression is performed.

[0062] Specifically, in this embodiment, the original data to be transmitted is compressed, which can be divided into the following two cases:

[0063] For numerical data, Delta encoding can be used to first reduce the differences between consecutive data points, and then compressed using a general compression library such as Zlib, Gzip, or Snappy.

[0064] For specific types of data, such as waveform data of vibration signals, it is possible to consider using compression in a specific format, such as wavelet transform or Fourier transform followed by compression.

[0065] Step S2: Based on the raw operating parameters preprocessed in step S1, a big data analysis algorithm that is consistent with the motor operating data characteristics is selected, and the preprocessed raw operating parameters are identified using the determined big data analysis algorithm. Specifically, the big data analysis algorithms in this embodiment include, but are not limited to, time series analysis algorithms, deep learning models, and ensemble learning models.

[0066] At the same time, after identifying the pre-processed original operating parameters, an evaluation model is established based on the identified pre-processed original operating parameters, and a specific feature vector is obtained from the evaluation model, as follows:

[0067] Step S2.1: Use the random forest learning model to analyze the preprocessed raw operating parameters and determine the objective function for the Bayesian optimization method. Also, define the range of values ​​for each hyperparameter in the Bayesian optimization method, such as the learning rate, regularization strength, and tree depth. Then, select the Gaussian process as the initial prior model and randomly select the initial hyperparameters as the operating data for the initial prior model. Optimize and iterate the model until the objective function value that maximizes is obtained, which is the desired optimal hyperparameter.

[0068] Step S2.2: Establish an evaluation model through time series analysis algorithm. The details are as follows:

[0069] Step S2.2.1: Sort the pre-processed original operating parameters in chronological order using the recorded timestamp information corresponding to the original operating parameters.

[0070] Step S2.2.2: Based on the motor's historical operating parameters, identify failure modes, performance fluctuation patterns, or efficiency trends. Based on the different patterns, determine the corresponding model type. Specifically, for time series data, models such as ARIMA and LSTM can be used. For classification or regression problems, models such as random forest and support vector machines can be considered.

[0071] In this embodiment, after the model type is determined, the model parameters of the corresponding model are initialized, and a feature vector is selected or constructed from the sorted original operating parameters to construct an evaluation model. It is worth noting that after the evaluation model is constructed, its applicability can be further tested using a residual analysis algorithm, and its generalization ability can also be further tested using a cross-validation algorithm.

[0072] Step S2.3: Substitute the optimal hyperparameters determined in step S2.1 into the evaluation model determined in step S2.2.2 to extract the optimal feature combination from the original data to form a feature vector.

[0073] Step S3: Analyze the characteristic vector and determine the control parameters to be adjusted corresponding to the motor based on the analysis results. In this embodiment, the characteristic vector is analyzed as follows:

[0074] Step S3.1: Normalize the feature vectors using Min-Mx scaling or Z-score normalization to ensure that all feature vectors are on the same scale to reduce dimensionality effects between feature vectors. At the same time, based on the motor's operating conditions, such as light load, full load, and startup, the most relevant features are selected or constructed to ensure that the selected feature vectors effectively reflect the motor's state.

[0075] Step S3.2: Based on the final feature vector obtained in Step S3.1, determine the need for motor fault detection or state assessment, thereby clarifying the required learning model and training the required learning model. Simultaneously, within the trained learning model, set corresponding thresholds to classify the motor state into different levels, such as normal, minor fault, and major fault. Evaluation tools such as receiver operating characteristic (ROC) curves and confusion matrices can also be used to find the optimal threshold. The threshold is selected and set based on the actual operation of the motor.

[0076] In this embodiment, the final feature vector obtained in step S3.1 is used as input and substituted into the trained learning model, so that the operating status and fault level of the motor can be directly divided.

[0077] Step S3.3: Analyze the characteristic vector according to the operating status and fault level of the motor divided in step S3.2, specifically:

[0078] When the motor is in normal working state, the operating parameters of the motor remain unchanged. Otherwise, the control parameters of the motor are adjusted through the adjustment strategy.

[0079] Step S4: According to the location of the motor, a computing node is set near it, and the control parameters of the motor are adjusted according to the control parameters to be adjusted corresponding to the motor obtained in step S3.3, as follows:

[0080] Step S4.1: Based on the motor's operating status and fault level, determine the eigenvector most relevant to the current state. For example, if the motor is overheating, the most relevant eigenvectors are temperature and current. If the motor is inefficient, the most relevant eigenvectors are current-to-voltage ratio, speed, and other characteristics. Once the most relevant eigenvectors are determined, determine the eigenvectors to be adjusted using a feature importance assessment method, such as checking feature scores in a random forest model or using the coefficients of a linear regression model.

[0081] Step S4.2: Build a mathematical model of motor operation and dynamically adjust motor parameters based on the eigenvector to be adjusted. This involves using the eigenvector to be adjusted, determined in step S3.3.1, as input to the mathematical model. The mathematical model predicts the motor's next operating state and, based on the prediction, calculates the optimal control parameter sequence, such as voltage, frequency, and torque, to achieve the desired operating state or avoid a fault.

[0082] In this embodiment, a mathematical model of motor operation is established as follows:

[0083] Step S4.2.1: Determine the type of mathematical model corresponding to the motor type. In this embodiment, motor types include, but are not limited to, DC motors, AC induction motors, and permanent magnet synchronous motors. The corresponding mathematical model types can be composed of voltage equations and mechanical motion equations, expressed in both stationary and rotating coordinate systems. Note that this embodiment uses a DC motor and its corresponding mathematical model as an example for detailed description.

[0084] Step S4.2.2: Express the motor's operating state in state space form, that is, convert the eigenvector to be adjusted into state space form, that is, convert it into state equations and output equations, providing a mathematical basis for the mathematical model. Specifically, the state equation is:

[0085] ;

[0086] in: and is the equation of state, is the inductor, is the externally applied voltage, is the back electromotive force coefficient, is the angular velocity of the motor, is the armature resistance, is the current in the motor winding, is the moment of inertia of the motor, The torque generated by the external load.

[0087] The output equation is:

[0088] ;

[0089] in: and is the output equation, is the angular velocity of the motor, is the current in the motor winding.

[0090] Step S4.2.3: Input the eigenvector to be adjusted, which is converted into the state space form, into the mathematical model to establish the mathematical model of the motor operation.

[0091] Step S4.3: Set the calculation node according to the position of the motor. The details are as follows:

[0092] In this embodiment, the computing node includes a real-time data processing module, an intelligent analysis module, and a control strategy adjustment module. The real-time data processing module is used to obtain the operating parameters of the motor and process the obtained operating parameters to remove invalid data points and filter out noise signals, thereby ensuring the quality of subsequent analysis. It is worth noting that in the real-time data processing module, corresponding preset thresholds and rules can be set according to actual usage requirements. The processed operating parameters of the motor can be detected by the Flink detector provided in the real-time data processing module to extract abnormal operating parameters, such as sudden current changes or abnormal vibrations.

[0093] The intelligent analysis module identifies, predicts and classifies the operating parameters of the motor through a machine learning model, and identifies the operating feature vector of the motor.

[0094] The control strategy adjustment module is used to dynamically adjust the motor parameters according to the motor control parameters output by the mathematical model.

[0095] Specifically, for motor speed control, the motor speed can be automatically adjusted according to load changes and predicted efficiency curves to achieve the optimal energy efficiency ratio. Specifically:

[0096] Sensors deployed on the motor collect real-time operating data, including current, voltage, temperature, and speed. After receiving this data through edge computing nodes, the system uses built-in algorithm models to analyze the current load and motor efficiency. Based on the analysis results, the system automatically adjusts the motor speed. For example, as the load increases, if the predicted efficiency curve indicates that a higher speed will result in a better energy efficiency ratio, the speed is automatically increased; otherwise, the speed is reduced.

[0097] For motor torque optimization, it can optimize the torque distribution strategy according to the motor load distribution to reduce energy waste. Specifically:

[0098] Using motor load data collected by sensors, the system analyzes load distribution during different operating phases and identifies operational links where overload or inefficiency may occur. Based on this load analysis, the system adjusts the torque distribution strategy. For example, in scenarios where multiple motors operate in concert, the system optimizes torque distribution between each motor. Notably, edge computing nodes can continuously learn and adjust torque optimization strategies to adapt to varying operating environments and load requirements.

[0099] For motor fault warning, it can detect potential faults in advance, adjust the operating mode or send maintenance requests to avoid unplanned downtime. Specifically:

[0100] Edge computing nodes continuously monitor the motor's health, including vibration levels, temperature fluctuations, and current fluctuations. Pre-trained fault prediction models, such as neural networks trained on historical data, analyze current status data to predict potential failure risks. If a high-risk fault signal is detected, the system automatically adjusts the operating mode to a more conservative state or sends an early warning notification to maintenance personnel, arranging preventive maintenance and preventing unplanned downtime.

[0101] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A motor operation data analysis method based on intelligent perception, characterized in that: The analysis method is as follows: Acquiring original operating parameters of the motor and preprocessing the original operating parameters; Preprocessing the original operating parameters also includes: Through the data intelligent screening algorithm, the original data is classified and the original data to be transmitted is determined; Compressing the original data to be transmitted by using a compression algorithm; Through the decision tree classification model and the historical operation data of the motor, the required data points and time periods are clarified, and the final decision tree classification model that can be used to filter the motor data is established. The classified raw data is then input into the final decision tree classification model, from which the raw data to be transmitted is automatically identified and filtered; By using the pre-processed original operating parameters and a time series analysis algorithm, an evaluation model is established to obtain a feature vector; Analyzing the characteristic vector according to the divided operating state and fault level of the motor, and determining the control parameters to be adjusted corresponding to the motor according to the analysis results; According to the position of the motor, a calculation node is set, and the control parameters of the motor are adjusted by the control parameters to be adjusted corresponding to the motor, including establishing a mathematical model of the motor operation; The computing nodes include: Real-time data processing module, used to obtain and process the operating parameters of the motor; In the real-time data processing module, corresponding preset thresholds and rules can be set according to actual usage requirements to detect the operating parameters of the processed motor, thereby extracting abnormal operating parameters; Intelligent analysis module, which applies machine learning models to identify the motor's operating characteristic vectors; A control strategy adjustment module, configured to dynamically adjust the motor parameters according to the motor control parameters output by the mathematical model; Sensors deployed on the motor collect real-time operating data, including current, voltage, temperature, and speed. After receiving the acquired data through the edge computing node, the current load conditions and motor efficiency are analyzed through the built-in algorithm model, and the motor speed is automatically adjusted based on the analysis results.

2. The method for analyzing motor operation data based on intelligent perception according to claim 1, characterized in that: The feature vector is obtained as follows: Determining optimal hyperparameters from the pre-processed original operating parameters using a Bayesian optimization method; Establish an evaluation model through time series analysis algorithm; The feature vector is obtained according to the optimal hyperparameter and the evaluation model.

3. The motor operation data analysis method based on intelligent perception according to claim 1 or 2, characterized in that: The evaluation model is established as follows: Sorting the pre-processed original operating parameters in chronological order; The model type is determined according to the historical operating parameters of the motor, and the evaluation model is constructed according to the sorted original operating parameters.

4. The method for analyzing motor operation data based on intelligent perception according to claim 3, characterized in that: The evaluation model is tested for applicability using a residual analysis algorithm, and the evaluation model is tested for generalization ability using a cross-validation algorithm.

5. The motor operation data analysis method based on intelligent perception according to claim 1 or 2, characterized in that: The characteristic vector is analyzed as follows: Normalizing the eigenvectors and determining a final eigenvector based on the operating conditions of the motor; A learning model is trained using the final feature vector, and a threshold is set in the learning model to classify the operating state and fault level of the motor; According to the divided operating state and fault level of the motor, the characteristic vector is analyzed as follows: When the motor is in a normal working state, the operating parameters of the motor remain unchanged; otherwise, the control parameters of the motor are adjusted through an adjustment strategy.

6. The method for analyzing motor operation data based on intelligent perception according to claim 5, characterized in that: The control parameters of the motor are adjusted as follows: Determining a characteristic vector to be adjusted according to the operating state and fault level of the motor; Establishing a mathematical model of the motor operation; Setting a calculation node according to the motor position; The motor parameters are dynamically adjusted through the characteristic vector to be adjusted, the calculation node, and the mathematical model.

7. The method for analyzing motor operation data based on intelligent perception according to claim 6, characterized in that: The mathematical model of the motor operation is established as follows: Determining the type corresponding to the mathematical model according to the motor type; The feature vector to be adjusted is converted into a state space form, specifically: ; in: and is the state equation, and is the output equation, is the inductor, is the externally applied voltage, is the back electromotive force coefficient, is the angular velocity of the motor, is the armature resistance, is the current in the motor winding, is the moment of inertia of the motor, The torque generated for the external load; The characteristic vector to be adjusted, which is converted into a state space form, is input into the mathematical model to establish a mathematical model of the motor operation.

8. The method for analyzing motor operation data based on intelligent perception according to claim 1, characterized in that: The intelligent analysis module identifies, predicts and classifies the operating parameters of the motor through a machine learning model.

Images