Motor bearing operation state detection method and system

Abstract

The invention relates to the technical field of intelligent monitoring and fault diagnosis, and discloses a motor bearing operation state detection method and system. The method comprises the following steps: acquiring bearing operation data in real time through a sensor arranged on a motor, carrying out data preprocessing, constructing a bearing state detection model by adopting a method of combining a convolutional neural network and a long-short term memory network, inputting the preprocessed real-time data into the model, outputting an operation state detection result, and triggering an early warning mechanism according to the result. The system comprises a data acquisition module, a data preprocessing module, a model training module, a state detection module and an early warning output module. According to the invention, the data processing flow is simplified, the accuracy and the efficiency are improved, the operation state of the bearing can be accurately judged, a perfect early warning mechanism is designed, the motor shutdown accident is effectively avoided, the stability and the safety of a production line are improved, and the method has a wide application prospect.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent monitoring and fault diagnosis, and in particular to a method and system for detecting the operating state of a motor bearing. Background Art

[0002] As the core equipment in industrial production, the operating status of the motor directly affects the efficiency and safety of the entire production line. As an important part of the motor, the operating status of the motor bearing is directly related to the overall performance and life of the motor. Traditional motor bearing operating status monitoring methods often rely on manual inspections and empirical judgments. This method is not only inefficient, but also difficult to accurately judge the early failure of bearings. With the development of sensor technology and artificial intelligence technology, more and more intelligent monitoring methods are being applied to the operating status monitoring of motor bearings. However, most of the existing intelligent monitoring methods have problems such as complex data processing, difficult model construction, and imperfect early warning mechanisms, which are difficult to meet the needs of industrial production for real-time monitoring of motor bearing operating status and fault warning. Summary of the invention

[0003] In view of the above-mentioned technical deficiencies, the purpose of the present invention is to provide a method and system for detecting the operating status of a motor bearing, so as to solve the problems existing in the prior art such as complex data processing, difficult model building, and imperfect early warning mechanism.

[0004] In order to solve the above technical problems, the present invention adopts the following technical solutions: In a first aspect, the present invention provides a method for detecting the operating state of a motor bearing, the method comprising: Step S100: collecting the bearing operation data of the motor bearing in real time during operation through a sensor arranged on the motor; Step S200: performing data preprocessing on the collected bearing operation data; Step S300: using a method combining a convolutional neural network and a long short-term memory network, and using historical bearing operation data to construct a bearing state detection model; Step S400: inputting the preprocessed real-time bearing operation data into the trained bearing state detection model, and outputting the operation state detection result of the motor bearing; Step S500: triggering an early warning mechanism according to the operating status detection result.

[0005] Preferably, in a possible implementation manner of the first aspect, the bearing operation data includes bearing vibration signal data, bearing temperature data, bearing sound signal data and bearing speed data.

[0006] Preferably, in a possible implementation manner of the first aspect, the data preprocessing includes removing noise, data cleaning, standardization and feature extraction of the bearing operation data.

[0007] Preferably, in a possible implementation manner of the first aspect, the feature extraction is frequency domain feature extraction, and the frequency spectrum components of the bearing vibration signal data and the bearing sound signal data are analyzed by Fourier transform to extract frequency features related to the bearing state.

[0008] Preferably, in a possible implementation manner of the first aspect, the historical bearing operation data includes historical bearing vibration signal data, historical bearing temperature data, historical bearing sound signal data, historical bearing speed data and historical fault data.

[0009] Preferably, in a possible implementation manner of the first aspect, the structure of the bearing state detection model comprises an input layer, a convolution layer, a pooling layer, a long short-term memory layer, a fully connected layer and an output layer; The input layer receives the preprocessed bearing operation data; Convolutional layers and pooling layers are used for feature extraction of bearing operation data; The long short-term memory layer is used to capture the dynamic changes of time series; The fully connected layer is used to map features to the output layer; The output layer outputs the fault type and fault probability of the motor bearing.

[0010] Preferably, in a possible implementation manner of the first aspect, the fault types include surface damage, wear, poor lubrication, improper assembly, overheating, abnormal noise and speed fluctuation.

[0011] Preferably, in a possible implementation manner of the first aspect, the operating state detection result includes a normal state, a minor fault state, a medium fault state and a major fault state; Normal state means the bearing runs smoothly without abnormality; A minor fault state means that the bearing begins to wear but does not affect normal operation; Medium fault status, indicating that bearing wear has affected the operation of the motor and requires immediate maintenance; A serious fault state indicates that the bearing has failed and needs to be shut down for inspection and maintenance immediately.

[0012] Preferably, in a possible implementation manner of the first aspect, the early warning mechanism includes: When the operating status detection result is normal, no warning is triggered and monitoring continues; When the operating status detection result is a minor fault state, a primary warning is triggered to prompt maintenance personnel to pay attention; When the operating status detection result is a medium fault state, an intermediate warning is triggered and maintenance personnel immediately perform maintenance; When the operating status detection result is a serious fault state, an emergency warning is triggered, and the machine is shut down immediately and the bearing is inspected and repaired.

[0013] In a second aspect, the present invention provides a motor bearing operation status detection system, the system comprising: A data acquisition module is used to collect the bearing operation data of the motor bearing in real time during operation through sensors arranged on the motor; Data preprocessing module, used to preprocess the collected bearing operation data, including noise removal, data cleaning, standardization and feature extraction; A model training module is used to build a bearing condition detection model using historical bearing operation data by combining a convolutional neural network with a long short-term memory network. A state detection module is used to input the preprocessed real-time bearing operation data into the trained bearing state detection model and output the operation state detection result of the motor bearing; The early warning output module is used to trigger the early warning mechanism according to the operation status detection results.

[0014] The beneficial effects of the present invention are as follows: a motor bearing operating state detection method proposed by the present invention realizes real-time monitoring and accurate diagnosis of the motor bearing operating state by integrating sensor technology, data preprocessing technology, convolutional neural network and long short-term memory network. This method not only simplifies the data processing flow and improves the efficiency and accuracy of data processing, but also can accurately judge the operating state of the bearing, including normal state, minor fault state, medium fault state and serious fault state, by constructing an efficient bearing state detection model. In addition, the present invention also designs a perfect early warning mechanism, which can trigger corresponding early warning measures according to the detection results of the operating state of the bearing, effectively avoid motor shutdown accidents caused by bearing failures, and improve the stability and safety of the production line. This method has broad application prospects and can be applied to motor bearing operating state monitoring and fault early warning in various industrial fields. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0016] Figure 1A flow chart of a method for detecting the operating status of a motor bearing is provided for the present application.

[0017] Figure 2 A system structure diagram for detecting the operating status of a motor bearing is provided for this application.

[0018] Explanation of the accompanying drawings: 1-data acquisition module, 2-data preprocessing module, 3-model training module, 4-state detection module, 5-early warning output module. DETAILED DESCRIPTION

[0019] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. 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 creative work are within the scope of protection of the present invention.

[0020] Embodiment 1: Figure 1 As shown, the present invention provides a method for detecting the operating state of a motor bearing, the method comprising: Step S100: collecting the bearing operation data of the motor bearing in real time during operation through sensors arranged on the motor.

[0021] Specifically, the bearing operation data includes bearing vibration signal data, bearing temperature data, bearing sound signal data and bearing speed data, and the sensors used include vibration sensors, temperature sensors, sound sensors and speed sensors.

[0022] Vibration sensors are used to detect the vibration of bearings during operation. By measuring parameters such as the amplitude and frequency of the vibration, it is possible to determine whether the bearings are worn, loose, or unbalanced.

[0023] Temperature sensors are used to monitor the operating temperature of bearings and prevent bearing damage caused by excessive temperature by measuring the temperature of the bearing surface.

[0024] The sound sensor is used to capture the sound signals generated when the bearing is running. By analyzing the frequency, timbre and other characteristics of the sound, it can determine whether the bearing has abnormal conditions such as abnormal noise or friction.

[0025] The speed sensor is used to measure the speed of the motor bearing. By monitoring the changes in speed, it can understand whether the operating state of the bearing is stable, and whether there are problems such as overload and jamming.

[0026] Step S200: preprocessing the collected bearing operation data.

[0027] Specifically, data preprocessing includes noise removal, data cleaning, standardization and feature extraction of bearing operation data.

[0028] In this embodiment, noise usually manifests itself as high-frequency or random signal fluctuations, and has a particularly significant impact on vibration signals and sound signals. In order to effectively remove these noises, a bandpass filter is used to filter the vibration signal. During the processing, the low cutoff frequency (lowcut) and high cutoff frequency (highcut) of the filter are first set, and the sampling frequency (fs) and filter order (order) are determined. Subsequently, the bandpass filter is designed using the butter function in the SciPy library, and the vibration signal is bidirectionally filtered using the filtfilt function, and finally the denoised vibration signal is obtained.

[0029] To implement the above filtering process, two functions are defined. The first function is called butter_bandpass and accepts four parameters: lowcut, highcut, fs, and order. The function first calculates the Nyquist frequency and then converts the low and high cutoff frequencies to normalized frequencies relative to the Nyquist frequency. Next, the butter function is used to design a bandpass filter and return the numerator (b) and denominator (a) coefficients of the filter. The second function is called bandpass_filter and it also accepts four parameters, the same as the butter_bandpass function. The function first calls the butter_bandpass function to obtain the filter coefficients and then uses the filtfilt function to perform bidirectional filtering on the data. The filtfilt function effectively eliminates the phase distortion introduced by the filtering process by applying the filter in the forward and reverse directions and finally returns the filtered data.

[0030] The data cleaning phase mainly deals with missing values ​​and outliers. For missing values, the nan_to_num function in the numpy library is used to fill in the missing values, replacing the NaN value with 0 or the interpolation of the adjacent value. For outliers, a simple threshold method is used to identify and process them. Specifically, the mean value of the data set plus three times the standard deviation is first calculated as the threshold, and then the outliers that exceed the threshold are replaced with the mean value of the data set, thereby achieving smoothing of outliers.

[0031] The standardization process aims to scale the data to a specific range or to have zero mean and unit variance. The MinMaxScaler and StandardScaler classes in the sklearn library are used for standardization. Among them, MinMaxScaler scales the data to between 0 and 1, while StandardScaler makes the data have zero mean and unit variance. In this embodiment, an instance of MinMaxScaler is first created, and its fit_transform method is used to fit and transform the data. Subsequently, an instance of StandardScaler is created, and the same method is used to process the data. The converted data are stored in data_scaled_min_max and data_scaled_standard variables respectively, and converted to a one-dimensional array by the .flatten() method for subsequent processing.

[0032] The feature extraction link is to extract key information that can reflect the bearing status from the original data. Frequency domain features are extracted for the bearing vibration signal data and the bearing sound signal data, and the fast Fourier transform (FFT) is used to extract the dominant frequency. In this embodiment, the welch function in the scipy.signal module is first used to perform frequency domain analysis on the data set. The welch function receives data as input and needs to specify the sampling frequency (fs) of the data. The function returns two values: freqs and Pxx. freqs contains the frequency values ​​obtained by analysis, while Pxx contains the power spectrum density values ​​at each frequency. By comparing the values ​​in the Pxx array, the frequency value with the largest power spectrum density is found, and it is named dominant_freq and extracted as the dominant frequency feature.

[0033] Step S300: Using a method combining a convolutional neural network and a long short-term memory network, a bearing state detection model is constructed using historical bearing operation data.

[0034] Specifically, the historical bearing operation data includes historical bearing vibration signal data, historical bearing temperature data, historical bearing sound signal data, historical bearing speed data and historical fault data, which are obtained through sensors set on the motor and stored in a database of the control system.

[0035] The bearing status detection model consists of an input layer, a convolutional layer, a pooling layer, a long short-term memory layer, a fully connected layer, and an output layer. The input layer is responsible for receiving the preprocessed bearing operation data; the convolutional layer and the pooling layer work together to extract key features from the input data; the long short-term memory layer focuses on capturing dynamic change information in the time series; the fully connected layer further maps these features to the output layer; and finally, the output layer is responsible for outputting the fault type and fault probability of the motor bearing.

[0036] In this example, we first define a time step of 100, and each input sample will contain feature values ​​of 100 consecutive time points. The features variable extracts the number of features from the preprocessed data.

[0037] Then, the input data X is constructed through a loop. For each sample in preprocessed_data (except the last time_steps-1 samples that cannot form a complete input sequence due to insufficient length), take the continuous time_steps feature values ​​starting from the current position and add them as input samples to X. Finally, X is presented as a three-dimensional array with a shape of [number of samples, time steps, number of features].

[0038] Next is the model construction phase. The Sequential model is used, which is a linear stacked model structure that makes it easy to add different layers in sequence.

[0039] The first layer added is a 1D convolutional layer Conv1D, which contains 32 filters, each with a kernel size of 3 and a ReLU activation function. This layer takes the shape of the input data (timestep, number of features) and outputs a tensor of shape [timestep, number of filters]. Next is a max pooling layer MaxPooling1D, with a pooling size of 2, which aims to reduce the dimensionality of the data while retaining the most important feature information. Next is a Dropout layer with a dropout rate of 0.25 to reduce the risk of overfitting. This is followed by a stack of two LSTM layers, each with 64 units. The first LSTM layer sets return_sequences=True to return the output of the entire sequence; the second LSTM layer only returns the output of the last time step. This is followed by a fully connected layer Dense, which contains 32 units and also uses ReLU as the activation function. Another Dropout layer is added with a dropout rate of 0.5 to further reduce the possibility of overfitting.

[0040] Finally, the output layer is a fully connected layer Dense with a softmax activation function. The number of its units is equal to the number of fault types. The probability of each fault type is output to determine the fault type of the bearing and its corresponding fault probability.

[0041] In this embodiment, the fault types include surface damage, wear, poor lubrication, overheating, abnormal noise, and speed fluctuation.

[0042] Surface damage, such as spalling, cracks, pitting, etc., can cause sudden shock pulse forces to occur during the operation of the bearing, thus showing obvious periodic shock components in the vibration signal.

[0043] Wear: After long-term operation, wear will occur between the rolling elements and raceways of the bearing, resulting in an increase in the amplitude of the vibration signal, which may be accompanied by changes in the frequency components.

[0044] Poor lubrication, insufficient lubrication or deterioration of lubricating oil can cause the bearing to run unevenly and produce abnormal vibration signals.

[0045] Overheating. Excessive bearing temperature may be caused by poor lubrication, overload operation or internal bearing failure. Excessive temperature will accelerate the wear and aging of the bearing.

[0046] Abnormal noise. Abnormal noise generated by bearings during operation may be caused by surface damage, wear or improper assembly. These noises are identified and diagnosed through sound signal analysis.

[0047] Speed ​​fluctuations and bearing failures may lead to speed instability, which is manifested as fluctuations or sudden changes in speed. Such fluctuations are analyzed and identified through speed data.

[0048] Step S400: input the preprocessed real-time bearing operation data into the trained bearing state detection model, and output the operation state detection result of the motor bearing.

[0049] Specifically, the real-time bearing operation data preprocessed in step S200, including the denoised vibration signal, cleaned and standardized temperature data, sound signal and speed data, are organized into the same input format as that used in training. These data are divided according to the time step time_steps defined in step S300 to ensure that each input sample contains the feature values ​​of consecutive and equal time points.

[0050] After the data is prepared, it is input one by one into the trained bearing condition detection model. The model first receives the data through the input layer, and then the data is processed by the convolution layer and the pooling layer to extract the key spatial features. Then, the long short-term memory layer (LSTM) captures the dynamic change information of these features in the time series, which is a key step in understanding the evolution of bearing status over time.

[0051] After being processed by the LSTM layer, the data is passed to the fully connected layer and further mapped to the output layer. The output layer uses the softmax activation function to convert the input feature vector into a probability distribution of the fault type. This probability distribution indicates the probability of various fault types that may currently exist in the motor bearing.

[0052] The current operating state detection result of the motor bearing is obtained according to the probability of various fault types output by the model. The state detection results include normal state, minor fault state, medium fault state and serious fault state.

[0053] Normal state means that the bearing runs smoothly without abnormalities; minor fault state means that the bearing begins to wear but does not affect normal operation; moderate fault state means that the bearing wear has had a certain impact on the motor operation and requires immediate maintenance; severe fault state means that the bearing has failed and needs to be shut down for inspection and maintenance.

[0054] In this embodiment, the normal state is when the probability of all fault types is less than 0.1, which is determined to be a normal state, indicating that the bearing runs smoothly without abnormality.

[0055] A minor fault state is when the probability of one or more fault types existing is greater than or equal to 0.1 and less than 0.3, and there is no fault type with a probability greater than or equal to 0.3. This indicates that the bearing begins to wear but does not affect normal operation.

[0056] A medium fault state is when the probability of one fault type is greater than or equal to 0.3 but less than 0.6, or the sum of the probabilities of multiple fault types is greater than or equal to 0.4 but the probabilities of all fault types are less than 0.6. This indicates that bearing wear has had a certain impact on motor operation and requires immediate maintenance.

[0057] Serious fault state: When the probability of one or more fault types is greater than or equal to 0.6, it is judged as a serious fault state, indicating that the bearing has failed and needs to be shut down for maintenance immediately.

[0058] Step S500: triggering an early warning mechanism according to the status detection result.

[0059] Specifically, the early warning mechanism includes: when the operating status detection result is normal, no early warning is triggered and monitoring continues; when the operating status detection result is a minor fault state, a primary early warning is triggered to prompt maintenance personnel to pay attention; when the operating status detection result is a moderate fault state, an intermediate early warning is triggered and maintenance personnel immediately perform maintenance; when the operating status detection result is a severe fault state, an emergency early warning is triggered, the machine is immediately shut down and the bearings are inspected.

[0060] In this embodiment, when the state detection result is normal, it indicates that the bearing is running smoothly and all data are within the normal range. At this time, the early warning system does not trigger any early warning signal, and continues to monitor the bearing operation data in real time to ensure the continuous stability of the bearing state.

[0061] If the test result is a minor fault, it means that the bearing has begun to wear and other abnormal conditions, but has not yet had a significant impact on normal operation. At this time, the early warning mechanism will trigger a primary warning signal and notify maintenance personnel through the system interface or SMS, email, etc. The primary warning is to remind maintenance personnel to pay attention to changes in the bearing status and prepare for preventive maintenance, such as checking lubrication conditions, adjusting operating parameters, etc., to prevent further development of the fault.

[0062] When the detection result is a medium fault state, it means that the bearing wear has had a certain impact on the operation of the motor, and the vibration, temperature, sound and other signals have obvious abnormalities. At this time, the early warning mechanism will trigger an intermediate warning signal and immediately send an emergency notice to the maintenance personnel. The intermediate warning requires the maintenance personnel to take immediate action to conduct detailed inspection and maintenance of the bearing, such as replacing the lubricating oil, adjusting the assembly accuracy, etc., to restore the normal operation of the bearing as soon as possible to prevent the fault from further deteriorating.

[0063] If the test result is a serious fault state, it means that the bearing has failed, strong periodic impact components appear in the vibration signal, sharp abnormal noise appears in the sound signal, the temperature data seriously exceeds the standard, and the speed data fluctuates violently or even changes suddenly. At this time, the early warning mechanism will immediately trigger an emergency warning signal and issue a shutdown command to the operator and maintenance personnel through the system interface, sound and light alarms, etc. The emergency warning requires immediate shutdown and bearing inspection to prevent the fault from causing greater damage to the motor or even causing a safety accident.

[0064] While triggering the early warning mechanism, the early warning system will also record information such as the time, location, type and severity of the fault, providing data support for subsequent fault analysis and processing.

[0065] Embodiment 2: Figure 2 As shown, the present invention provides a motor bearing operation status detection system, the system comprising: The data acquisition module 1 is used to collect the bearing operation data of the motor bearing in real time during operation through the sensor arranged on the motor; Data preprocessing module 2, used for preprocessing the collected bearing operation data, including noise removal, data cleaning, standardization and feature extraction; Model training module 3, for building a bearing condition detection model using historical bearing operation data by combining a convolutional neural network with a long short-term memory network; The state detection module 4 is used to input the pre-processed real-time bearing operation data into the trained bearing state detection model and output the operation state detection result of the motor bearing; The early warning output module 5 is used to trigger the early warning mechanism according to the operation status detection result.

[0066] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for detecting the operating state of a motor bearing, characterized in that: The method comprises: Step S100: collecting the bearing operation data of the motor bearing in real time during operation through a sensor arranged on the motor; Step S200: performing data preprocessing on the collected bearing operation data; Step S300: using a method combining a convolutional neural network and a long short-term memory network, and using historical bearing operation data to construct a bearing state detection model; Step S400: inputting the preprocessed real-time bearing operation data into the trained bearing state detection model, and outputting the operation state detection result of the motor bearing; Step S500: triggering an early warning mechanism according to the operating status detection result.

2. The motor bearing operating state detection method according to claim 1, characterized in that: The bearing operation data includes bearing vibration signal data, bearing temperature data, bearing sound signal data and bearing speed data.

3. The motor bearing operating state detection method according to claim 2, characterized in that: The data preprocessing includes removing noise, cleaning data, standardizing data and extracting features from the bearing operation data.

4. The motor bearing operating state detection method according to claim 3, characterized in that: The feature extraction is frequency domain feature extraction, which uses Fourier transform to analyze the frequency spectrum components of the bearing vibration signal data and the bearing sound signal data to extract frequency features related to the bearing state.

5. The motor bearing operating state detection method according to claim 1, characterized in that: The historical bearing operation data includes historical bearing vibration signal data, historical bearing temperature data, historical bearing sound signal data, historical bearing speed data and historical fault data.

6. The motor bearing operating state detection method according to claim 1, characterized in that: The structure of the bearing state detection model includes an input layer, a convolution layer, a pooling layer, a long short-term memory layer, a fully connected layer and an output layer; The input layer receives the preprocessed bearing operation data; Convolutional layers and pooling layers are used for feature extraction of bearing operation data; The long short-term memory layer is used to capture the dynamic changes of time series; The fully connected layer is used to map features to the output layer; The output layer outputs the fault type and fault probability of the motor bearing.

7. The motor bearing operating state detection method according to claim 6, characterized in that: The fault types include surface damage, wear, poor lubrication, improper assembly, overheating, abnormal noise and speed fluctuation.

8. The motor bearing operating state detection method according to claim 1, characterized in that: The operating state detection results include normal state, minor fault state, medium fault state and major fault state; Normal state means the bearing runs smoothly without abnormality; A minor fault state means that the bearing begins to wear but does not affect normal operation; Medium fault status, indicating that bearing wear has affected the operation of the motor and requires immediate maintenance; A serious fault state indicates that the bearing has failed and needs to be shut down for inspection and maintenance immediately.

9. The motor bearing operating state detection method according to claim 1, characterized in that: The early warning mechanism includes: When the operating status detection result is normal, no warning is triggered and monitoring continues; When the operating status detection result is a minor fault state, a primary warning is triggered to prompt maintenance personnel to pay attention; When the operating status detection result is a medium fault state, an intermediate warning is triggered and maintenance personnel immediately perform maintenance; When the operating status detection result is a serious fault state, an emergency warning is triggered, and the machine is shut down immediately and the bearing is inspected and repaired.

10. A motor bearing operation status detection system, characterized in that: The system comprises: A data acquisition module is used to collect the bearing operation data of the motor bearing in real time during operation through sensors arranged on the motor; Data preprocessing module, used to preprocess the collected bearing operation data, including noise removal, data cleaning, standardization and feature extraction; A model training module is used to build a bearing condition detection model using historical bearing operation data by combining a convolutional neural network with a long short-term memory network. A state detection module is used to input the preprocessed real-time bearing operation data into the trained bearing state detection model and output the operation state detection result of the motor bearing; The early warning output module is used to trigger the early warning mechanism according to the operation status detection results.

Images