Motor fault detection device and system combining edge computing and cloud computing

Abstract

The invention relates to a motor fault detection device and system combining edge computing and cloud computing. The device comprises an edge computing module and a cloud computing module. The edge computing module comprises a data acquisition unit which is used for acquiring motor operation data and forwarding the motor operation data to the cloud computing module; the preliminary detection unit is used for performing preliminary fault detection according to the motor operation data to obtain a preliminary detection result; the cloud computing module comprises a machine learning model which is used for verifying the credibility of the preliminary detection result and identifying the risk level according to the current motor operation data and the historical motor operation data to obtain a final fault detection result. According to the method, the advantages of edge computing and cloud computing are respectively exerted, the preliminary detection result can be quickly obtained, the preliminary detection result can be confirmed and corrected to obtain an accurate fault detection result, and the real-time performance, the sensitivity and the accuracy of fault detection are considered.

Description

Technical Field

[0001] The present invention relates to the field of motor technology, and in particular to a motor fault detection device and system combining edge computing and cloud computing. Background Art

[0002] With the development of industrial automation and intelligent manufacturing, permanent magnet synchronous servo motors (PMSM) are widely used in various mechanical equipment due to their high efficiency, precision and reliability. PMSMs may encounter various types of faults during operation, such as coil failure, bearing damage, overheating, power failure, etc. These faults will hinder the normal operation of the motor and even damage the motor, causing safety hazards.

[0003] Traditional motor fault detection methods mainly rely on manual inspection and maintenance, which has the disadvantages of slow response speed, low detection efficiency and high labor cost. In the prior art, there is a motor fault detection method that collects motor operation data through sensors and automatically determines whether the motor has a fault by analyzing the motor operation data. However, the existing motor fault detection method still has the defects of slow response speed, low detection accuracy and low intelligence level. Summary of the invention

[0004] Based on this, the object of the present invention is to provide a motor fault detection device and system that combines edge computing and cloud computing.

[0005] The present invention provides a motor fault detection device combining edge computing and cloud computing, comprising: an edge computing module and a cloud computing module; the edge computing module comprises: a data acquisition unit, which is used to acquire motor operation data and forward the motor operation data to the cloud computing module; a preliminary detection unit, which is used to perform preliminary fault detection according to the motor operation data to obtain preliminary detection results; the cloud computing module comprises: a machine learning model, which is used to verify the credibility of the preliminary detection result and identify the risk level according to the current motor operation data in combination with historical motor operation data, so as to obtain the final fault detection result.

[0006] The present invention organically combines the advantages of edge computing and cloud computing, allowing the edge computing module to process simple computing tasks with strong real-time performance to obtain approximate preliminary detection results; and allowing the cloud computing module to process complex computing tasks with high computing power consumption, confirm and correct the preliminary detection results, and obtain accurate fault detection results, thereby giving full play to the advantages of edge computing and cloud computing respectively, and taking into account the real-time performance, sensitivity and accuracy of fault detection.

[0007] Furthermore, the edge computing module also includes: a preliminary detection result output unit, which is used to send the preliminary detection result to an external response module after obtaining the preliminary detection result, so that the external response module can make a preliminary response.

[0008] Furthermore, the cloud computing module also includes: a fault detection result output unit, which is used to send the fault detection result to the edge computing module after obtaining the fault detection result; the edge computing module also includes: a fault detection result forwarding unit, which is used to forward the fault detection result to an external response module for the external response module to make a formal response.

[0009] Furthermore, the machine learning model of the cloud computing module is trained by historical motor operation data; when new current motor operation data is obtained, the current motor operation data at the previous moment is merged with the historical motor operation data to generate new historical motor operation data; when the historical motor operation data is updated, the parameters of the machine learning model are also continuously updated accordingly.

[0010] Furthermore, the machine learning model is an LSTM model.

[0011] Furthermore, the preliminary detection unit of the edge computing module specifically includes: a preprocessing subunit, which is used to preprocess the motor operation data to obtain preprocessed data; a feature extraction subunit, which is used to extract features from the preprocessed data to obtain a number of feature parameters; and a classifier, which is used to obtain preliminary detection results based on the feature parameters.

[0012] Furthermore, the preprocessing includes FFT transformation, filtering denoising and normalization.

[0013] Furthermore, the classifier obtains a preliminary detection result according to the characteristic parameter and a type weight corresponding to the characteristic parameter; the type weight depends on the type of motor operation data corresponding to the characteristic parameter.

[0014] Furthermore, the classifier is a random forest classifier, a SVM classifier or an XGBoost classifier.

[0015] Based on the same inventive concept, the present invention also provides a motor fault detection system combining edge computing and cloud computing, including: a sensor module, any one of the above-mentioned motor fault detection devices combining edge computing and cloud computing, and a response module; the sensor module monitors the motor body to collect motor operation data; the fault detection device determines whether the motor body has a potential fault and identifies the risk level based on the motor operation data, and outputs a fault detection result; the response module responds based on the fault detection result.

[0016] For better understanding and implementation, the present invention is described in detail below with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A schematic diagram of a module of a motor fault detection system combining edge computing and cloud computing according to the present invention;

[0018] Figure 2 It is a module schematic diagram of a motor fault detection device combining edge computing and cloud computing of the present invention;

[0019] Figure 3 A schematic flow chart of a motor fault detection method performed by a motor fault detection device combining edge computing and cloud computing according to the present invention. DETAILED DESCRIPTION

[0020] It should be clear that the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the embodiments of the present application.

[0021] The terms used in the embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to limit the embodiments of the present application. The singular forms of "a", "said" and "the" used in the embodiments of the present application and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more associated listed items.

[0022] It should be understood that the embodiments of the present application are not limited to the precise structures described above and shown in the drawings, and various modifications and changes can be made without departing from the scope thereof. The scope of the embodiments of the present application is limited only by the appended claims.

[0023] Edge computing refers to assigning computing tasks to computing devices at the edge of the network for execution. In the scenario of motor fault detection, "network edge" refers to the production site where the motor body is located. Since the edge computing device is set near the motor body, the edge computing device can obtain the motor operation data collected by the sensor with very low latency, with fast response speed and strong real-time performance. However, the computing power of edge computing devices is limited, which is not suitable for processing large-scale data and more complex computing tasks. The fault identification accuracy is low, and it is difficult to identify potential faults and risk levels.

[0024] Cloud computing refers to assigning computing tasks to cloud servers in the network center for execution. In the scenario of motor fault detection, the "network center" refers to a data center far away from the motor itself. Cloud servers have powerful computing power and are suitable for processing large-scale data and more complex computing tasks. They can extract richer features and have higher fault identification accuracy. However, since cloud servers are far away from the production site and data transmission relies on remote communication, communication delays are high, response speeds are slow, and real-time performance is poor.

[0025] The present invention creatively combines edge computing with cloud computing, allowing edge computing devices to process simple computing tasks with strong real-time requirements to obtain approximate preliminary detection results; and allowing cloud servers to process complex computing tasks with high computing power consumption, confirm and correct the preliminary detection results, and obtain accurate fault detection results, thereby giving full play to the advantages of edge computing and cloud computing respectively, and taking into account the real-time, sensitivity and accuracy of fault detection.

[0026] See also Figure 1 , Figure 1 This is a module schematic diagram of the motor fault detection system combining edge computing and cloud computing of the present invention.

[0027] The motor fault detection system combining edge computing and cloud computing of the present invention includes: a sensor module 1, a fault detection device 2 and a response module 3. The sensor module 1 monitors the motor body to collect motor operation data. The fault detection device 2 analyzes the state of the motor body and determines whether the motor body has a potential fault based on the motor operation data, and outputs a fault detection result. The response module 3 responds based on the fault detection result. Specifically, the response module 3 includes an information output unit 31 and a control unit 32. The information output unit 31 displays or broadcasts the motor operation data and fault detection results for user reference. The control unit 32 controls the working parameters of the motor body or controls the start and stop of the motor body based on the fault detection result.

[0028] Specifically, the sensor module 1 includes: a current sensor, a temperature sensor, a vibration sensor and a rotation speed sensor. The motor operation data includes current data, temperature data, vibration data and rotation speed data.

[0029] In one embodiment, the current sensor is arranged on the driving circuit of the motor body to monitor the driving current in the driving circuit of the motor body, thereby obtaining current data. The temperature sensor is arranged on the outer shell of the motor body to monitor the outer shell temperature of the motor body, thereby obtaining temperature data. The vibration sensor is arranged on the bearing of the motor body to monitor the vibration of the bearing of the motor body, thereby obtaining vibration data. The speed sensor is arranged on the rotor or output shaft of the motor body to monitor the speed output by the motor body, thereby obtaining speed data.

[0030] See also Figure 2 and Figure 3 , Figure 2 This is a module schematic diagram of a motor fault detection device combining edge computing and cloud computing of the present invention. Figure 3 A schematic flow chart of a motor fault detection method performed by a motor fault detection device combining edge computing and cloud computing according to the present invention.

[0031] Specifically, the fault detection device 2 is a motor fault detection device combining edge computing and cloud computing according to the present invention, which includes: an edge computing module 21 and a cloud computing module 22 .

[0032] The edge computing module 21 is a computing device (such as a PLC, DSP or FPGA processor) arranged at the motor body site, which includes: a data acquisition unit M1, a preliminary detection unit M2, a preliminary detection result output unit M3 and a fault detection result forwarding unit M6. The cloud computing module 22 is a cloud server arranged in the cloud, which includes: a machine learning model M4 and a fault detection result output unit M5. The edge computing module 21 and the cloud computing module 22 can communicate remotely via the MQTT protocol or the HTTP protocol.

[0033] The data acquisition unit M1 of the edge computing module 21 is used to execute step S1: acquire the motor operation data collected by the sensor module, and forward the motor operation data to the cloud computing module 22.

[0034] The preliminary detection unit M2 of the edge computing module 21 is used to execute step S2: perform preliminary fault detection according to the motor operation data to obtain preliminary detection results.

[0035] Specifically, the preliminary detection unit M2 includes: a preprocessing subunit, a feature extraction subunit and a classifier.

[0036] The preprocessing subunit is used to perform preprocessing such as FFT transformation (Fast Fourier Transformation), filtering, denoising and normalization on the motor operation data to obtain preprocessed data.

[0037] The filtering and denoising may be low-pass filtering, wavelet transform filtering, Butterworth filtering or Kalman filtering, etc. The normalization refers to mapping the numerical values ​​of the motor operation data acquired by each sensor into a unified dimension range to ensure comparability between the data.

[0038] The feature extraction subunit is used to extract features from the preprocessed data to obtain a number of feature parameters.

[0039] In one embodiment, feature extraction is performed on the current data to extract feature parameters such as the mean, variance and / or peak factor of the current signal, which can reflect whether the motor is overloaded or short-circuited. Feature extraction is performed on the temperature data to calculate feature parameters such as the temperature rise rate and / or the abnormal temperature fluctuation amplitude to analyze the heat dissipation system failure or motor overload. Feature extraction is performed on the vibration data to extract feature parameters such as time domain (such as RMS, skewness, kurtosis) and frequency domain features (such as spectrum peak, harmonic components) to analyze the wear or looseness of mechanical parts. Feature extraction is performed on the speed data to extract feature parameters such as the standard deviation and fluctuation frequency of the speed to analyze the operation consistency of the mechanical system.

[0040] The classifier is used to obtain preliminary detection results according to the feature parameters.

[0041] After obtaining the characteristic parameters, multiple characteristic parameters can be fused through a feature fusion algorithm such as a serial splicing algorithm, a principal component analysis method, or a neural network algorithm, and then the fused features are classified and identified through a classifier to obtain a preliminary judgment result of whether the motor is faulty. The classifier can be a classifier with classification and identification capabilities such as a random forest classifier, a SVM classifier, or an XGBoost classifier.

[0042] Furthermore, since different types of motor operation data contribute differently to the detection of different fault types, for example, temperature data contributes more to detecting whether the motor has an overheating fault, and vibration data contributes more to detecting whether the motor has a mechanical fault, the classifier obtains a preliminary detection result based on the characteristic parameters and the type weight corresponding to the characteristic parameters; the type weight depends on the type of motor operation data corresponding to the characteristic parameters. When the classifier preliminarily determines whether the motor has a specified type of fault, it dynamically adjusts the type weight corresponding to each motor operation data. In one embodiment, the classifier adjusts the type weight of the characteristic parameters corresponding to each motor operation data by an entropy weight method.

[0043] The preliminary detection result output unit M3 of the edge computing module 21 is used to execute step S3: sending the preliminary detection result to the response module 3, so that the response module 3 makes a preliminary response according to the preliminary detection result.

[0044] The information output unit 31 makes a preliminary response based on the preliminary detection result, including: displaying or broadcasting the motor operation data and preliminary detection result for user reference. The control unit 32 makes a preliminary response based on the preliminary detection result, including: making preliminary adjustments to the working parameters of the motor body. For example, when the preliminary detection result shows that the temperature of the motor body is too high, the control unit 32 immediately lowers the power of the motor body, thereby reducing the heating power of the motor body and avoiding the temperature of the motor body from continuing to rise. Since the edge computing module 21 has a very fast calculation speed, the information output unit 31 and the control unit 32 have a very low delay in making a preliminary response. When it is preliminarily determined that the motor has a fault, an alarm and control can be immediately performed, thereby timely avoiding the further expansion of the impact of the motor fault and ensuring the safety of the motor operation.

[0045] The machine learning model M4 of the cloud computing module 22 is used to execute step S4: verify the credibility of the preliminary detection result and identify the risk level based on the current motor operation data and the historical motor operation data to obtain the final fault detection result.

[0046] The machine learning model is used to extract and identify features of current motor operation data and historical motor operation data to obtain final fault detection results; the fault detection results include not only the judgment result of whether there is a potential fault in the motor body, but also the identification result of the risk level.

[0047] The machine learning model is trained with historical motor operation data. When the edge computing module 21 uploads new current motor operation data, the current motor operation data at the previous moment is merged with the historical motor operation data to generate new historical motor operation data. When the historical motor operation data is updated, the parameters of the machine learning model are also continuously updated, so as to adapt to complex and changeable motor operating conditions.

[0048] The machine learning model can be a neural network model that can learn historical data features, such as RNN (Recurrent Neural Network), LSTM (Long Short-Term Memory), GRU (Gate Recurrent Unit), Bi-LSTM (Bidirectional Long Short-Term Memory), etc. In this embodiment, the machine learning model is preferably an LSTM model.

[0049] The fault detection result output unit M5 of the cloud computing module 22 is used to execute step S5: transmit the fault detection result to the edge computing module 21. The fault detection result forwarding unit M6 of the edge computing module 21 is used to execute step S6: forward the fault detection result to the response module 3, so that the response module 3 makes a formal response according to the fault detection result.

[0050] The information output unit 31 makes a formal response based on the preliminary detection result, including: displaying or broadcasting the motor operation data and the final fault detection result. The control unit 32 makes a formal response based on the fault detection result, including: adjusting the working parameters of the motor body or controlling the start and stop of the motor body. For example, when the preliminary detection result shows that the temperature of the motor body is too high, the control unit 32 first makes a preliminary response (for example, lowering the power of the motor body); when the final fault detection result confirms that a fault has occurred in the motor body, the control unit 32 makes a formal response (for example, controlling the motor body to stop running).

[0051] The control unit 32 can also adaptively control the motor body according to the risk level displayed by the fault detection result. For example, when the fault detection result shows that the risk level is extremely high, the control unit 32 controls the motor body to stop running to prevent the motor body from continuing to work regardless of the fault and causing further damage. When the fault detection result shows that the risk level is medium, the control unit 32 only adjusts certain operating parameters of the motor body (such as voltage, speed, torque, etc.) and does not shut down. When the fault detection result shows that the risk level is extremely low, the control unit 32 does not perform any action. The present invention identifies the risk level through the cloud computing module and responds in a targeted manner according to the risk level, which not only ensures the safety of the motor operation, but also avoids excessive adjustment that affects the motor working efficiency, thereby improving the intelligence level of fault detection.

[0052] The present invention has the following technical effects: 1. On the one hand, the present invention utilizes the advantages of low communication delay and fast computing speed of the edge computing module, and immediately performs preliminary alarm and control when it is preliminarily determined that the motor has a fault, thereby timely avoiding the further expansion of the impact of the motor fault and ensuring the safety of the motor operation; on the other hand, it also utilizes the advantages of powerful computing power and high accuracy of the cloud computing module to obtain accurate fault detection results and risk levels and perform intelligent response actions, thereby ensuring the accuracy and intelligence of fault detection. Therefore, the fault detection device of the present invention organically combines the advantages of edge computing and cloud computing, taking into account the real-time, sensitivity and accuracy of fault detection. 2. The fault detection model of the cloud computing module of the present invention can continuously update its own parameters according to the updated historical motor operation data, so that it can still maintain a high detection accuracy under the conditions of complex and changeable motor working conditions.

[0053] The above-mentioned embodiments only express several implementation methods of the present invention, and the description is relatively specific and detailed, but it cannot be understood as limiting the scope of the invention patent. It should be pointed out that for ordinary technicians in this field, several modifications and improvements can be made without departing from the concept of the present invention, and the present invention is also intended to include these modifications and modifications.

Claims

1. A motor fault detection device combining edge computing and cloud computing, characterized in that: include: Edge computing module and cloud computing module; The edge computing module includes: a data acquisition unit, which is used to acquire motor operation data and forward the motor operation data to the cloud computing module; a preliminary detection unit, which is used to perform preliminary fault detection according to the motor operation data to obtain preliminary detection results; The cloud computing module includes: a machine learning model, which is used to verify the credibility of the preliminary detection result and identify the risk level based on the current motor operation data and the historical motor operation data to obtain the final fault detection result.

2. The motor fault detection device combining edge computing and cloud computing according to claim 1, characterized in that: The edge computing module also includes: The preliminary detection result output unit is used to send the preliminary detection result to the external response module after obtaining the preliminary detection result, so that the external response module can make a preliminary response.

3. The motor fault detection device combining edge computing and cloud computing according to claim 2, characterized in that: The cloud computing module also includes: A fault detection result output unit, which is used to send the fault detection result to the edge computing module after obtaining the fault detection result; The edge computing module also includes: The fault detection result forwarding unit is used to forward the fault detection result to an external response module so that the external response module can make a formal response.

4. The motor fault detection device combining edge computing and cloud computing according to claim 3, characterized in that: The machine learning model of the cloud computing module is trained by historical motor operation data; when new current motor operation data is obtained, the current motor operation data at the previous moment is merged with the historical motor operation data to generate new historical motor operation data; when the historical motor operation data is updated, the parameters of the machine learning model are also continuously updated accordingly.

5. The motor fault detection device combining edge computing and cloud computing according to claim 4, characterized in that: The machine learning model is an LSTM model.

6. The motor fault detection device combining edge computing and cloud computing according to claim 5, characterized in that: The preliminary detection unit of the edge computing module specifically includes: A preprocessing subunit, which is used to preprocess the motor operation data to obtain preprocessed data; A feature extraction subunit, which is used to extract features from the preprocessed data to obtain a number of feature parameters; A classifier is used to obtain a preliminary detection result according to the feature parameters.

7. The motor fault detection device combining edge computing and cloud computing according to claim 6, characterized in that: The preprocessing includes FFT transformation, filtering denoising and normalization.

8. The motor fault detection device combining edge computing and cloud computing according to claim 7, characterized in that: The classifier obtains a preliminary detection result according to the characteristic parameters and the type weights corresponding to the characteristic parameters; the type weights depend on the types of motor operation data corresponding to the characteristic parameters.

9. The motor fault detection device combining edge computing and cloud computing according to claim 8, characterized in that: The classifier is a random forest classifier, a SVM classifier or an XGBoost classifier.

10. A motor fault detection system combining edge computing and cloud computing, characterized in that: include: A sensor module, a motor fault detection device combining edge computing and cloud computing as described in any one of claims 1 to 9, and a response module; The sensor module monitors the motor body to collect motor operation data; The fault detection device determines whether the motor body has a potential fault and identifies the risk level according to the motor operation data, and outputs a fault detection result; The response module responds according to the fault detection result.

Images