Motor fault detection method and system based on machine learning

Abstract

The invention discloses a motor fault detection method and system based on machine learning. The method comprises the steps of data acquisition, data preprocessing, feature extraction, model training, fault detection and diagnosis, and result output and feedback. The system comprises a data acquisition module, a data processing module, a model training module, a fault detection module and an output module. Compared with the prior art, the system has the advantages that comprehensive data acquisition is realized; fine data processing; an efficient and accurate fault detection model is provided; and remote monitoring and maintenance are convenient.

Description

Technical Field

[0001] The present invention relates to the technical field of motor fault detection, and in particular to a motor fault detection method and system based on machine learning. Background Art

[0002] As a key equipment in industrial production, the operating status of motors directly affects production efficiency and safety. Traditional motor fault detection methods mostly rely on manual inspections and regular maintenance, which have problems such as long detection cycles, low efficiency, and easy missed detection. With the rapid development of machine learning technology, its application in the field of fault diagnosis is becoming more and more extensive. However, existing motor fault detection systems based on machine learning often have problems such as incomplete data collection, imprecise data processing, and poor model adaptability, which make it difficult to meet the needs of complex and changing industrial environments. Summary of the invention

[0003] The technical problems to be solved by the present invention are: 1) Incomplete data collection: Traditional methods often only focus on a single or a few operating parameters of the motor, resulting in omission of fault information and affecting detection accuracy;

[0004] 2) Imprecise data processing: The raw data may contain interference factors such as noise and outliers. If it is directly used for model training without effective processing, it will reduce the model performance;

[0005] 3) Poor model adaptability: The fault characteristics of different motor types and working environments vary greatly, and existing models are often difficult to adapt to a variety of complex situations;

[0006] 4) Inconvenience in remote monitoring and maintenance: Fault information cannot be transmitted to the remote management center in a timely manner, affecting the timeliness and efficiency of fault handling.

[0007] In order to solve the above technical problems, the technical solution provided by the present invention is: a motor fault detection method based on machine learning, comprising the following steps:

[0008] S1. Collect the motor's operating data, including current, voltage, vibration, and temperature parameters;

[0009] S2, preprocessing the collected data, including data cleaning, denoising, and normalization operations;

[0010] S3, constructing a motor fault detection model, using a machine learning algorithm to train the preprocessed data to obtain a trained motor fault detection model;

[0011] S4, inputting the motor operation data to be detected into the trained motor fault detection model to obtain the motor fault detection result;

[0012] S5. According to the motor fault detection result, the corresponding fault information and maintenance suggestions are output.

[0013] It also includes an auxiliary motor fault detection system based on machine learning, including a data acquisition module, a data processing module, a model training module, a fault detection module, and an output module based on the system architecture. The data acquisition module is used to collect the operating data of the motor; the data processing module is used to preprocess the collected data; the model training module is used to build and train the motor fault detection model; the fault detection module is used to input the motor operating data to be detected into the trained motor fault detection model to obtain the motor fault detection result; the output module is used to output the fault detection result and maintenance suggestions. The output module includes a communication module, and the communication module is used for data transmission and communication between modules within the system and between the system and external systems.

[0014] The advantages of the present invention compared with the prior art are: 1) Comprehensive data collection: by integrating multiple sensors such as current, voltage, vibration, temperature, etc., comprehensive collection of motor operation data is achieved to ensure the integrity of fault information;

[0015] 2) Fine data processing: Use pre-processing techniques such as data cleaning, denoising, and normalization to effectively remove interference factors in the data, improve data quality, and lay a solid foundation for subsequent model training;

[0016] 3) Efficient and accurate fault detection model: Combine convolutional neural networks and long short-term memory networks to extract spatial and temporal features, use support vector machines or deep neural networks for classification, and build an efficient and accurate motor fault detection model that can accurately identify a variety of motor faults;

[0017] 4) Convenient remote monitoring and maintenance: Through the supporting communication module, fault information and maintenance suggestions are sent to the remote management center in real time, realizing remote monitoring and maintenance, and improving the timeliness and efficiency of fault handling.

[0018] Furthermore, in step S3, the motor fault detection model includes a feature extraction unit, a classifier unit and an output unit; the feature extraction unit is used to extract fault features from the preprocessed data; the classifier unit is used to classify the motor fault according to the extracted fault features; and the output unit is used to output fault detection results and maintenance suggestions.

[0019] Furthermore, the feature extraction unit includes a convolutional neural network and a long short-term memory network, which are used to extract spatial and temporal features respectively; the classifier unit adopts a support vector machine or a deep neural network machine learning algorithm to classify the extracted fault features.

[0020] Furthermore, step S5 sends the fault information and maintenance suggestions to the remote management center through the matching communication module for remote monitoring and maintenance.

[0021] Furthermore, the data acquisition module includes a current sensor, a voltage sensor, a vibration sensor and a temperature sensor, which are used to collect various operating parameters of the motor.

[0022] Furthermore, the data processing module includes a data cleaning unit, a denoising unit and a normalization unit, which are used to pre-process the collected data; the data cleaning unit is composed of an input interface, a data processing logic module and an output interface in terms of architecture, the input interface receives the original data from the data acquisition module, the data processing logic module performs data cleaning operations, and the output interface passes the cleaned data to the subsequent processing unit; the denoising unit is composed of a signal receiving module, a noise detection module and a denoising processing module in terms of architecture, the signal receiving module receives the original signal from the data acquisition module, the noise detection module analyzes the noise component in the signal, and the denoising processing module removes the noise according to the detection result; the normalization unit is composed of a data receiving module, a normalization processing module and a data output module in terms of architecture, the data receiving module receives data from the denoising unit or other data sources, the normalization processing module performs normalization processing on the data, and the data output module passes the normalized data to the subsequent processing unit.

[0023] Furthermore, both the model training module and the fault detection module are integrated with a machine learning algorithm library to support the selection and application of multiple machine learning algorithms. The machine learning algorithm library is architecturally composed of an algorithm interface module, an algorithm implementation module, and a parameter configuration module. The algorithm interface module provides a unified algorithm calling method, the algorithm implementation module contains specific algorithm logic, and the parameter configuration module allows users to adjust algorithm parameters as needed. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 Schematic diagram of the steps of the motor fault detection method based on machine learning.

[0025] Figure 2 It is a schematic diagram of the architecture of the motor fault detection system based on machine learning.

[0026] As shown in the figure: 1. Data acquisition module, 101. Current sensor, 102. Voltage sensor, 103. Vibration sensor, 104. Temperature sensor, 2. Data processing module, 201. Data cleaning unit, 202. De-noising unit, 203. Normalization unit, 3. Model training module, 4. Fault detection module, 5. Output module, 501. Communication module. DETAILED DESCRIPTION

[0027] The present invention is further described in detail below in conjunction with the accompanying drawings.

[0028] Embodiment 1

[0029] like Figure 1 and Figure 2 As shown, the functions of the data cleaning unit 201 include:

[0030] 1) Missing value processing: Identify and fill missing values ​​in the data, which can be filled using the mean, median, mode, or a specific algorithm.

[0031] 2) Outlier detection: Identify and handle outliers in the data, which may include removing outliers or making corrections.

[0032] 3) Duplicate value processing: Identify and delete duplicate records in the data.

[0033] 4) Unified data format: Ensure that the data format is consistent, such as date, time, numerical value and other types of data meet unified standards.

[0034] In terms of interaction, the data cleaning unit 201 interacts with other units of the data acquisition module (1) and the data processing module (2) (such as the denoising unit 202 and the normalization unit 203), and works in coordination with subsequent processing units by receiving raw data and outputting cleaned data.

[0035] The functions of the denoising unit 202 include:

[0036] 1) Noise detection: Use signal processing techniques (such as Fourier transform, wavelet transform, etc.) to identify the noise components in the signal.

[0037] 2) Denoising: Apply denoising algorithms (such as filtering algorithms, wavelet denoising, etc.) to remove noise from the signal and improve the signal-to-noise ratio.

[0038] In terms of interaction, the denoising unit 202 interacts with the data acquisition module 1 and the data processing module 2, and works in coordination with subsequent processing units by receiving original signals and outputting denoised signals.

[0039] The normalization unit 203 has the following functions:

[0040] 1) Data standardization: Scale the data according to a certain ratio so that it falls into a small specific interval (such as [0,1] or [-1,1]).

[0041] 2) Eliminate dimensionality effects: Through normalization processing, the dimensional differences between different data dimensions are eliminated so that each data dimension has the same weight in model training.

[0042] In terms of interaction, the normalization unit 203 interacts with the data processing module 2 and the machine learning algorithm library, receives original data or processed data, and outputs normalized data, thereby providing standardized input for the machine learning algorithm.

[0043] Embodiment 2

[0044] Based on the first embodiment, Figure 1 and Figure 2 As shown, the functions of the machine learning algorithm library include:

[0045] 1) Algorithm selection: Provides a variety of machine learning algorithms for users to choose from, such as support vector machine, decision tree, random forest, neural network, etc.

[0046] 2) Model training: Use training data to train the selected algorithm and build a machine learning model.

[0047] 3) Model evaluation: Evaluate the trained model and its performance indicators (such as accuracy, recall, F1 score, etc.).

[0048] In terms of interaction, the machine learning algorithm library interacts with the data processing module 2, the fault detection module 4 and the end user, receives the processed data as input, and outputs the trained machine learning model or the prediction result of the model. The user can call the algorithm in the algorithm library through the interface, and perform parameter configuration and model training.

[0049] The architecture of the communication module 501 includes a communication interface, a communication protocol processing and a data transmission module, and its functions include:

[0050] 1) Data transmission: realize data exchange between modules within the system and between the system and external systems.

[0051] 2) Communication protocol support: Support multiple communication protocols (such as TCP / IP, HTTP, MQTT, etc.) to ensure reliable data transmission.

[0052] 3) Data encryption and decryption: Encrypt data to ensure the security of data transmission.

[0053] In terms of interaction, the communication module 501 interacts with various modules within the system (such as the data processing module 2, the machine learning algorithm library, the fault detection module 4) and external systems (such as the remote management center, other data processing systems), and realizes data flow and sharing within and outside the system by receiving and sending data.

[0054] Embodiment 3

[0055] Based on the second embodiment, Figure 1 and Figure 2As shown, the working principle of this system includes:

[0056] 1) Data collection

[0057] The operation data of the motor is collected in real time by various sensors installed on the motor (such as vibration sensor 103, temperature sensor 104, current sensor 101, etc.). The collected data includes vibration signals, temperature signals, current signals, etc. These signals reflect the operation status of the motor.

[0058] 2) Data preprocessing

[0059] Improve data quality, reduce noise and interference, and lay the foundation for subsequent feature extraction and model training. Data cleaning is achieved by removing duplicate data, processing missing values, identifying and processing outliers; signal processing technology (such as filtering algorithms, wavelet transforms, etc.) is used to remove noise components in the signal; normalization processing is achieved by scaling the data according to a certain ratio so that it falls into a small specific interval (such as [0,1]) to eliminate the dimensional differences between different data dimensions.

[0060] 3) Feature extraction

[0061] Extract feature information useful for fault detection from preprocessed data. Time domain analysis is achieved by extracting time domain features such as mean, variance, peak-to-peak value, etc.; frequency domain analysis is achieved by extracting spectrum features through Fourier transform and other methods; for non-stationary signals, time-frequency analysis is performed using methods such as wavelet transform to extract time-frequency domain features.

[0062] 4) Model training

[0063] Select the appropriate machine learning algorithm according to specific needs, and divide the preprocessed data set into training set and test set; use the training set data to train the selected machine learning algorithm to determine the optimal parameters of the model; use the test set data to evaluate the performance of the model, such as accuracy, recall rate, F1 score, etc.

[0064] 5) Fault detection and diagnosis

[0065] The motor data collected in real time is input into the trained model, and the model outputs the fault detection results based on the input data; the fault type, fault degree, etc. are identified based on the output results of the model; and a detailed fault diagnosis report is generated, including the fault type, fault location, possible causes and recommended maintenance measures.

[0066] 6) Result output and feedback

[0067] The fault diagnosis results are displayed to the operation and maintenance personnel or management personnel through the user interface; a feedback mechanism is established to allow the operation and maintenance personnel or management personnel to confirm or correct the diagnosis results, and the feedback information is used for continuous optimization and improvement of the model.

[0068] The above shows and describes the basic principles and main features of the present invention and the advantages of the invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A motor fault detection method based on machine learning, characterized in that The following steps are involved: S1. Collect the motor's operating data, including current, voltage, vibration, and temperature parameters; S2, preprocessing the collected data, including data cleaning, denoising, and normalization operations; S3, constructing a motor fault detection model, using a machine learning algorithm to train the preprocessed data to obtain a trained motor fault detection model; S4, inputting the motor operation data to be detected into the trained motor fault detection model to obtain the motor fault detection result; S5. According to the motor fault detection result, the corresponding fault information and maintenance suggestions are output.

2. A motor fault detection system based on machine learning, comprising a data acquisition module (1), a data processing module (2), a model training module (3), a fault detection module (4), and an output module (5) based on a system architecture, characterized in that: The data acquisition module (1) is used to acquire the operation data of the motor; the data processing module (2) is used to pre-process the acquired data; the model training module (3) is used to construct and train the motor fault detection model; the fault detection module (4) is used to input the motor operation data to be detected into the trained motor fault detection model to obtain the motor fault detection result; the output module (5) is used to output the fault detection result and maintenance suggestions, the output module (5) comprises a communication module (501), and the communication module (501) is used for data transmission and communication between modules within the system and between the system and an external system.

3. The motor fault detection method based on machine learning according to claim 1, characterized in that: The motor fault detection model in step S3 includes a feature extraction unit, a classifier unit and an output unit; the feature extraction unit is used to extract fault features from the preprocessed data; the classifier unit is used to classify the motor fault according to the extracted fault features; and the output unit is used to output fault detection results and maintenance suggestions.

4. The motor fault detection method based on machine learning according to claim 3 is characterized in that: The feature extraction unit includes a convolutional neural network and a long short-term memory network, which are used to extract spatial and temporal features respectively; the classifier unit adopts a support vector machine or a deep neural network machine learning algorithm to classify the extracted fault features.

5. The motor fault detection method based on machine learning according to claim 1, characterized in that: The step S5 sends the fault information and maintenance suggestions to the remote management center through the matching communication module (501) for remote monitoring and maintenance.

6. The motor fault detection system based on machine learning according to claim 2, characterized in that: The data acquisition module (1) comprises a current sensor (101), a voltage sensor (102), a vibration sensor (103) and a temperature sensor (104), and is used to collect various operating parameters of the motor.

7. The motor fault detection system based on machine learning according to claim 2, characterized in that: The data processing module (2) comprises a data cleaning unit (201), a denoising unit (202) and a normalization unit (203), and is used to pre-process the collected data; the data cleaning unit (201) is composed of an input interface, a data processing logic module and an output interface in terms of architecture, the input interface receives the original data from the data collection module, the data processing logic module performs a data cleaning operation, and the output interface transmits the cleaned data to a subsequent processing unit; the denoising unit (202) is composed of a signal receiving module, a noise detection module and a denoising processing module in terms of architecture, the signal receiving module receives the original signal from the data collection module (1), the noise detection module analyzes the noise component in the signal, and the denoising processing module removes the noise according to the detection result; the normalization unit (203) is composed of a data receiving module, a normalization processing module and a data output module in terms of architecture, the data receiving module receives data from the denoising unit or other data sources, the normalization processing module performs normalization processing on the data, and the data output module transmits the normalized data to a subsequent processing unit.

8. The motor fault detection system based on machine learning according to claim 2, characterized in that: The model training module (3) and the fault detection module (4) are both integrated with a machine learning algorithm library, supporting the selection and application of a variety of machine learning algorithms. The machine learning algorithm library is composed of an algorithm interface module, an algorithm implementation module and a parameter configuration module in terms of architecture. The algorithm interface module provides a unified algorithm calling method, the algorithm implementation module contains specific algorithm logic, and the parameter configuration module allows the user to adjust the algorithm parameters as required.

Images