Fault detection method and device, electronic equipment and storage medium

Abstract

Embodiments of the application disclose a fault detection method and device, electronic equipment and a storage medium. The method comprises: acquiring real-time running data of an automation device; analyzing the real-time running data every preset time period to detect whether a running parameter of the automation device has changed; if it is detected that the running parameter has not changed, processing the real-time running data corresponding to the current time period by a fault detection model to obtain a fault detection result corresponding to the automation device; and if it is detected that the running parameter has changed, training the fault detection model by an incremental learning method according to the real-time running data acquired within a first preset time period after the first preset time period. The embodiment can improve the accuracy of fault detection.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of fault detection, and in particular to a fault detection method and device, an electronic device, and a storage medium. BACKGROUND

[0002] With the continuous improvement of industrial automation level, more and more automation equipment is widely used in the production process, thereby greatly improving the production efficiency and product quality. However, due to the complexity and high integration of automation equipment, once a fault occurs, a large amount of time and manpower is required for maintenance and replacement, which not only causes production stagnation and loss, but also increases the maintenance and replacement costs of enterprises.

[0003] At present, the fault detection technology for automation equipment includes a fault detection technology based on a machine learning method. However, the fault detection technology based on the machine learning method requires a large amount of data samples and model training time, but in the case of limited data samples, the accuracy of the trained model is low, thereby resulting in low accuracy of fault detection using the model. SUMMARY

[0004] The embodiments of the present application disclose a fault detection method, device, electronic device, and storage medium, which can improve the accuracy of fault detection.

[0005] The embodiments of the present application disclose a fault detection method, which comprises:

[0006] obtaining real-time running data of an automation equipment;

[0007] analyzing the real-time running data every preset time period to detect whether a running parameter of the automation equipment changes;

[0008] if it is detected that the running parameter does not change, processing the real-time running data corresponding to the current time period by a fault detection model to obtain a fault detection result corresponding to the automation equipment;

[0009] if it is detected that the running parameter changes, training the fault detection model by an incremental learning method according to the real-time running data obtained within a first preset time length after the first preset time length.

[0010] In one embodiment, the analyzing the real-time running data every preset time period to detect whether a running parameter of the automation equipment changes comprises:

[0011] analyzing the real-time running data corresponding to the current time period to determine a running main frequency corresponding to the current time period;

[0012] The running main frequency corresponding to the current time period is compared with the running main frequency corresponding to the previous time period to detect whether the running parameter of the automation equipment changes.

[0013] In one embodiment, the real-time running data corresponding to the current time period is analyzed to determine the running main frequency corresponding to the current time period, including:

[0014] The real-time running data corresponding to the current time period is subjected to Fourier transform to obtain a running spectrum corresponding to the current time period;

[0015] The amplitude of each frequency in the running spectrum is determined;

[0016] The amplitudes of the frequencies are compared, and the frequency with the largest amplitude is determined as the running main frequency corresponding to the current time period.

[0017] In one embodiment, the running main frequency corresponding to the current time period is compared with the running main frequency corresponding to the previous time period to detect whether the running parameter of the automation equipment changes, including:

[0018] The frequency difference between the running main frequency corresponding to the current time period and the running main frequency of the previous time period is calculated;

[0019] If the frequency difference is greater than a frequency threshold, it is detected that the running parameter of the automation equipment changes;

[0020] If the frequency difference is not greater than the frequency threshold, it is detected that the running parameter of the automation equipment does not change.

[0021] In one embodiment, the real-time running data corresponding to the current time period is processed by a fault detection model to obtain a fault detection result corresponding to the automation equipment, including:

[0022] The initial time series data is input into the fault detection model, and the initial time series data is processed by the fault detection model to obtain predicted time series data; wherein the initial time series data is data obtained by arranging the running data corresponding to the current time period in time order; the fault detection model is trained according to target running data, and the running parameter corresponding to the target running data is the same as the running parameter corresponding to the real-time running data of the current time period;

[0023] The initial time series data is compared with the predicted time series data to obtain a data difference between the initial time series data and the predicted time series data;

[0024] If the data difference is not greater than the difference threshold value, it is determined that the fault detection result corresponding to the automation equipment is that the automation equipment is operating normally.

[0025] If the data difference is greater than the difference threshold value, it is determined that the fault detection result corresponding to the automation equipment is that the automation equipment has a fault or is about to have a fault.

[0026] In one embodiment, the fault detection model comprises a self-encoder model, the self-encoder model comprising an encoder and a decoder, the inputting of the initial time series data into the fault detection model, the processing of the initial time series data by the fault detection model to obtain predicted time series data, comprising:

[0027] inputting the initial time series into the encoder, performing dimensionality increasing encoding processing on the initial time series by the encoder to obtain an intermediate variable;

[0028] inputting the intermediate variable into the decoder, performing dimensionality decreasing decoding processing on the intermediate variable by the decoder to obtain predicted time series data.

[0029] In one embodiment, before the obtaining of the real-time running data of the automation equipment, the method further comprises:

[0030] in the case that the automation equipment is in an experimental environment, the running data of the automation equipment obtained within a second preset time length is taken as initial training data set;

[0031] based on the initial training data set, pre-training the fault detection model to be trained to obtain a pre-trained fault detection model.

[0032] Embodiments of the present application disclose a fault detection device, the device comprising:

[0033] a data acquisition module for acquiring real-time running data of an automation equipment;

[0034] a parameter detection module for analyzing the real-time running data every preset time period to detect whether the running parameters of the automation equipment have changed;

[0035] a model processing module for, if it is detected that the running parameters have not changed, processing the real-time running data corresponding to the current time period by a fault detection model to obtain a fault detection result corresponding to the automation equipment;

[0036] a model training module, configured to, if it is detected that the operation parameter changes, train the fault detection model by an incremental learning manner according to the real-time operation data acquired within a first preset time length after the operation parameter changes.

[0037] The embodiment of the present application discloses an electronic device, comprising:

[0038] a memory storing executable program codes;

[0039] a processor coupled with the memory;

[0040] The processor invokes the executable program codes stored in the memory to execute the method of any of the above embodiments.

[0041] The embodiment of the present application discloses a computer readable storage medium storing a computer program, wherein the computer program, when executed by a processor, causes the processor to execute the method of any of the above embodiments.

[0042] The fault detection method, device, electronic device and storage medium disclosed by the embodiment of the present application, the electronic device can analyze the real-time operation data of the automatic device, and can detect the fault of the automatic device when the operation parameter of the automatic device does not change, which can ensure the timeliness of fault detection, and the fault detection model is incrementally trained when the operation parameter of the automatic device changes, which can make the fault detection model better adapt to the parameter change of the automatic device, improve the precision of the fault detection model, and thus improve the accuracy of fault detection. In addition, the electronic device detects the fault based on the real-time operation data, without the need for automatic device downtime or other interference operations, so as to avoid the impact and loss of the fault detection operation on the production of the automatic device, thereby improving the availability and production efficiency of the automatic device. Moreover, the electronic device can automatically detect whether the operation parameter of the automatic device changes, and automatically judge whether the fault detection model needs to be trained, thereby reducing the need for manual intervention and improving the automation degree and efficiency of fault detection. BRIEF DESCRIPTION OF DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiments will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0044] Figure 1 is an application scenario diagram of a fault detection method disclosed by the embodiment of the present application;

[0045] Figure 2 This is a flowchart illustrating a fault detection method disclosed in an embodiment of this application;

[0046] Figure 3 This is a schematic diagram of a model training process disclosed in an embodiment of this application;

[0047] Figure 4 This is a flowchart illustrating another fault detection method disclosed in an embodiment of this application;

[0048] Figure 5 This is a modular schematic diagram of a fault detection device disclosed in an embodiment of this application;

[0049] Figure 6 This is a structural block diagram of an electronic device disclosed in an embodiment of this application. Detailed Implementation

[0050] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0051] It should be noted that the terms "comprising" and "having" and any variations thereof in the embodiments of this application are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to these processes, methods, products, or devices.

[0052] It is understood that the terms "first," "second," etc., used in this application may be used herein to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish one element from another. For example, without departing from the scope of this application, a first preset duration may be referred to as a second preset duration, and similarly, a second preset duration may be referred to as a first preset duration. Both the first preset duration and the second preset duration are preset durations, but they are not the same preset duration.

[0053] This application discloses a fault detection method, apparatus, electronic device, and storage medium, which can improve the accuracy of fault detection.

[0054] The following will be described in detail with reference to the accompanying drawings.

[0055] like Figure 1 As shown, Figure 1FIG. 1 is a schematic diagram of an application scenario of a fault detection method disclosed by an embodiment of the present application. The application scenario can include an electronic device 110 and an automated device 120. The electronic device 110 can include, but is not limited to, a mobile phone, a tablet computer, a wearable device, a notebook computer, a PC (Personal Computer), and the like. The automated device 120 can include, but is not limited to, a winding machine, a conveyor belt, a mechanical arm, a robot, and the like. The present application is not limited in this regard.

[0056] The electronic device 110 can be in communication connection with the automated device 120. The automated device 120 can send real-time running data to the electronic device 110, so that the electronic device 110 performs fault detection according to the real-time running data. Alternatively, the automated device 120 can not send real-time running data to the electronic device 110, and the processor of the automated device 120 can execute the fault detection method. In other words, the automated device 120 can be regarded as an electronic device 110 that executes the fault detection method.

[0057] In one embodiment, the electronic device 110 can obtain real-time running data of the automated device 120, and analyze the real-time running data every preset time period to detect whether the running parameters of the automated device 120 change. If it is detected that the running parameters do not change, the electronic device 110 can process the real-time running data corresponding to the current time period by using a fault detection model to obtain a fault detection result corresponding to the automated device 120. If it is detected that the running parameters change, the electronic device 110 can train the fault detection model by using an incremental learning method according to the real-time running data obtained within a first preset time period after the change is detected.

[0058] In one embodiment, as shown in FIG. 2, the electronic device 110 can obtain real-time running data of the automated device 120, and analyze the real-time running data every preset time period to detect whether the running parameters of the automated device 120 change. If it is detected that the running parameters do not change, the electronic device 110 can process the real-time running data corresponding to the current time period by using a fault detection model to obtain a fault detection result corresponding to the automated device 120. If it is detected that the running parameters change, the electronic device 110 can train the fault detection model by using an incremental learning method according to the real-time running data obtained within a first preset time period after the change is detected. Figure 2 Figure 2 FIG. 3 is a flowchart of a fault detection method disclosed by an embodiment of the present application. The fault detection method can be applied to the electronic device in the above embodiment. The fault detection method can include the following steps.

[0059] In step 210, real-time running data of an automated device is obtained.

[0060] ​In order to detect faults of the automation equipment, the electronic device can acquire real-time running data of the automation equipment. The electronic device can establish a communication connection with the automation equipment, which can include a wired connection and a wireless connection. In the case of a wired connection, the automation equipment and the electronic device can be connected by a cable and support a related communication protocol, which can include a serial communication protocol, an Ethernet protocol, etc. In the case of a wireless connection, the automation equipment and the electronic device can be connected through a wireless network, which can include Wi-Fi, Bluetooth, Zigbee, etc.

[0061] As an optional implementation, the automation equipment can be provided with a sensor for collecting real-time running data. The sensor can be in communication connection with the electronic device, and the sensor can transmit the collected real-time running data to the electronic device.

[0062] The real-time running data of the automation equipment can be used to represent the current state of the automation equipment. The real-time running data can include external running data of the automation equipment, which refers to data that can be measured by a sensor, such as the speed and torque of a winding machine. The real-time running data can not include internal running data of the automation equipment, such as input signals and output signals of the automation equipment.

[0063] It should be noted that the real-time running data of the automation equipment acquired by the electronic device in step 210 is in a production environment. Different manufacturers can use different production materials and shaft motion parameters in the production environment, and the data of different manufacturers is not disclosed to each other. Therefore, the technical personnel need to pre-train the fault detection model in an experimental environment, and then send the pre-trained fault detection model to each manufacturer for further training in the production environment of each manufacturer in an incremental learning manner. Thus, the parameters in the fault detection model obtained by each manufacturer can be different and adapted to the production environment of each manufacturer.

[0064] As shown in Figure 3 , the electronic device can be connected to the automation equipment through a wired connection or a wireless connection. The electronic device can acquire real-time running data of the automation equipment through the communication connection. The real-time running data can be used to represent the current state of the automation equipment. The real-time running data can include external running data of the automation equipment, which refers to data that can be measured by a sensor, such as the speed and torque of a winding machine. The real-time running data can not include internal running data of the automation equipment, such as input signals and output signals of the automation equipment. Figure 3is a flowchart of model training disclosed by an embodiment of the present application, wherein the real-time running data 310 of the automation equipment in the experimental environment is obtained through the automatic equipment, a pre-trained fault detection model 320 is obtained, the pre-trained fault detection model 320 enters each manufacturer respectively, the real-time running data 330 of the automation equipment in the production environment of the first manufacturer is obtained, the fault detection model 340 is obtained in an incremental learning manner, the real-time running data 350 of the automation equipment in the production environment of the second manufacturer is obtained, the fault detection model 360 is obtained in an incremental learning manner, the real-time running data 370 of the automation equipment in the production environment of the third manufacturer is obtained, and the fault detection model 380 is obtained in an incremental learning manner.

[0065] In one embodiment, when the automation equipment is in the experimental environment, the electronic device can obtain the real-time running data of the automation equipment within a second preset time length as an initial training data set, and based on the initial training data set, the electronic device can pre-train the fault detection model to be trained to obtain a pre-trained fault detection model.

[0066] In the pre-training process, the electronic device can input the data in the initial training data set into the fault detection model, obtain a result through encoding and decoding, calculate a model loss according to the result, and update the model parameters according to the model loss. The training of the fault detection model is not limited in the present application. Since the data of each manufacturer is not disclosed, the real-time running data cannot be labeled by the technical personnel. Optionally, the fault detection model can be an unsupervised model.

[0067] Due to the privacy problem of each manufacturer, sufficient running data of the automation equipment cannot be obtained, and a general fault detection model that can handle all cases cannot be directly obtained. By implementing the embodiment, a pre-trained fault detection model can be obtained in the experimental environment, and incremental learning training can be performed through the data of each manufacturer. The reliability of the fault detection model and the privacy safety of the manufacturer can be ensured.

[0068] In step 220, the real-time running data is analyzed every preset time period to detect whether the running parameters of the automation equipment change.

[0069] The electronic device can analyze the real-time running data every preset time period to detect whether the running parameter of the automation device changes, the running parameter being used to represent the state and performance of the automation device, and the running parameter can include but is not limited to a temperature parameter, a pressure parameter, a current parameter, etc., without limitation. The length of the preset time period is pre-set and fixed, for example, the preset time period is 2 seconds.

[0070] The current time period can be a preset time period with a current time as a terminal time, and the historical time period can include one or more. Optionally, the historical time period can include a previous preset time period of the current time period. In one example, the length of the preset time period is 2 seconds, the current time is the 10th second, the current time period is from the 8th second to the 10th second, and the historical time period is from the 6th second to the 8th second. Optionally, the historical time period can include a plurality of preset time periods closest to the current time period, and the electronic device can compare the analysis result corresponding to the current time period with the analysis result corresponding to the plurality of preset time periods closest to the current time period, respectively. In one example, the length of the preset time period is 2 seconds, the current time is the 10th second, the current time period is from the 8th second to the 10th second, and the historical time period includes 2 preset time periods closest to the current time period. The historical time period includes from the 6th second to the 8th second and from the 4th second to the 6th second.

[0071] In the case where the historical time period includes a plurality of time periods, the electronic device can compare the analysis result corresponding to the current time period with the analysis result corresponding to each historical time period, respectively, to obtain a plurality of comparison results. The comparison result is used to represent the difference between the analysis result corresponding to the target historical time period and the analysis result corresponding to the current time period, and the target historical time period is any historical time period. If any comparison result in the plurality of comparison results does not satisfy the preset requirement, it is detected that the running parameter of the automation device changes, and if the plurality of comparison results all satisfy the preset requirement, it is detected that the running parameter of the automation device does not change, wherein the preset requirement is used to represent the comparison result corresponding to the automation running parameter not changing.

[0072] Optionally, the electronic device can perform weighting processing on the analysis results corresponding to each historical time period to obtain a target analysis result, and then compare the analysis result corresponding to the current time period with the target analysis result. The weights corresponding to each historical time period can be the same, or can be determined according to the proximity of each historical time period to the current time period. The closer the historical time period to the current time period, the greater the weight corresponding to the historical time period. For example, the current time period is from the 8th second to the 10th second, the historical time periods include from the 6th second to the 8th second, from the 4th second to the 6th second, and from the 2nd second to the 4th second. The weight corresponding to the historical time period from the 6th second to the 8th second is 0.5, the weight corresponding to the historical time period from the 4th second to the 6th second is 0.3, and the weight corresponding to the historical time period from the 2nd second to the 4th second is 0.2.

[0073] In step 230, if it is detected that the operating parameter has not changed, the real-time operating data corresponding to the current time period is processed by the fault detection model to obtain a fault detection result corresponding to the automated device.

[0074] If the electronic device detects that the operating parameter has not changed, the electronic device can process the real-time operating data corresponding to the current time period by the fault detection model to obtain a fault detection result corresponding to the automated device. The fault detection model has the function of predicting the operating data of the automated device in a normal operating state (i.e., without failure) by using the real-time operating data of the automated device. By comparing the real-time operating data of the automated device with the predicted operating data, it can be determined whether the automated device has failed.

[0075] In some embodiments, the fault detection model is trained according to target operating data, the operating parameter corresponding to the target operating data being the same as the operating parameter corresponding to the real-time operating data of the current time period. The fault detection result can include normal operation of the automated device and failure or impending failure of the automated device. Optionally, the fault detection model can be trained by an incremental learning method according to the target operating data, i.e., the fault detection model has been trained by the real-time operating data of the automated device in a production environment. The fault detection model can also be pre-trained according to the target operating data, i.e., the fault detection model is trained by the real-time operating data of the automated device in an experimental environment.

[0076] In step 240, if it is detected that the operating parameter has changed, after a first preset time period, the fault detection model is trained by an incremental learning method according to the real-time operating data obtained within the first preset time period.

[0077] If the electronic device detects that the running parameter changes, it means that the running parameter of the automation device changes from the first running parameter to the second running parameter, and the training data set corresponding to the first running parameter is used when the fault detection model is last trained. The training data set corresponding to the first running parameter includes the running data of the automation device under the first running parameter. The fault detection model trained by the training data set corresponding to the first running parameter is not applicable to processing the running data of the automation device under the second running parameter. Therefore, the electronic device can maintain the acquisition of real-time running data of the automation device under the second running parameter. After a first preset time period, the fault detection model is trained by incremental learning based on the real-time running data acquired within the first preset time period, to update the parameters in the fault detection model, so that the trained fault detection model is applicable to processing the running data of the automation device under the second running parameter, i.e. to improve the accuracy of the fault detection model, thereby improving the accuracy of fault detection. The first preset time period and the length of the preset time period can be different, and since the amount of data required for model training is greater than the amount of data required for fault detection, the first preset time period can be greater than the length of the preset time period.

[0078] Within the first preset time period after detecting the change in the running parameter, the electronic device can continuously acquire real-time running data of the automation device, so that the real-time running data acquired within the first preset time period can be added to the training data set corresponding to the second running parameter after the first preset time period. Optionally, the electronic device can also acquire part of the data in the training data set corresponding to the first running parameter and add it to the training data set corresponding to the second running parameter, so that the size of the training data set corresponding to the second running parameter is the same as that of the training data set corresponding to the first running parameter. The part of the data acquired from the training data set corresponding to the first running parameter can be randomly selected data, or data that differs little from the mean value of the training data set corresponding to the first running parameter, and this is not limited. Based on the updated training data set, the incremental learning method can enable the fault detection model to continuously learn from new samples and retain most of the previously learned, so that the fault detection model trained by the incremental learning method is more suitable for real-time running data corresponding to the changed running parameter. By training the model by incremental learning, not only can the time and resources required for model training be reduced to meet the needs of the production environment, but also the prediction ability for the old running parameter can be retained while learning the new running parameter, so that the fault detection model has self-learning ability, thereby ensuring the accuracy of fault detection.

[0079] In the embodiments of the present application, the electronic device can analyze the real-time running data of the automation equipment, perform fault detection on the automation equipment when the running parameters of the automation equipment do not change, ensure the timeliness of fault detection, and perform incremental training on the fault detection model when the running parameters of the automation equipment change, so that the fault detection model can better adapt to the parameter changes of the automation equipment, improve the accuracy of the fault detection model, and thus improve the accuracy of fault detection. In addition, the electronic device performs fault detection based on real-time running data without the need for automation equipment downtime or other interference operations, so it can avoid the impact and loss of fault detection operations on automation equipment production, thereby improving the usability and production efficiency of the automation equipment. Moreover, the electronic device can automatically detect whether the running parameters of the automation equipment have changed and automatically determine whether the fault detection model needs to be trained, reducing the need for human intervention and improving the automation level and efficiency of fault detection.

[0080] As shown in Figure 4 , Figure 4 is a flow diagram of another fault detection method disclosed in the embodiments of the present application. The fault detection method can be applied to the electronic device in the above embodiments. The fault detection method can include the following steps:

[0081] Step 410, obtaining real-time running data of the automation equipment.

[0082] The description of step 410 can refer to the description of step 210 in the above embodiments, which will not be repeated here.

[0083] Step 420, analyzing the real-time running data corresponding to the current time period to determine the running main frequency corresponding to the current time period.

[0084] In one embodiment, the analysis result corresponding to the current time period can be the running main frequency corresponding to the current time period. The electronic device can analyze the real-time running data corresponding to the current time period to determine the running main frequency corresponding to the current time period. Optionally, the method of the electronic device for analyzing the real-time running data corresponding to the current time period can include but is not limited to Fourier transform, autocorrelation analysis, and support vector machine. Fourier transform can convert the real-time running data corresponding to the current time period from time domain to frequency domain, thereby determining the running main frequency corresponding to the current time period. Autocorrelation analysis can calculate the autocorrelation function of the real-time running data corresponding to the current time period based on the time domain, and obtain the autocorrelation function image. By calculating the time interval between the wave crests in the autocorrelation function image, the running main frequency corresponding to the current time period can be determined. The support vector machine can classify and predict the real-time running data corresponding to the current time period, thereby determining the running main frequency corresponding to the current time period.

[0085] In one embodiment, the electronic device can perform Fourier transform on the real-time running data corresponding to the current time period to obtain a running spectrum corresponding to the current time period, determine the amplitude of each frequency in the running spectrum, compare the amplitudes of the frequencies, and determine the frequency with the largest amplitude as the main running frequency corresponding to the current time period.

[0086] The running spectrum can be a set of frequencies exhibited by the automation device in the current time period, and can be obtained by the electronic device by performing Fourier transform on the real-time running data corresponding to the current time period. Optionally, the real-time running data corresponding to the current time period is a time-domain signal of the running state in the current time period, and the electronic device can perform Fourier transform on the real-time running data corresponding to the current time period to convert the running state in the current time period into a frequency-domain signal of the running state in the current time period. In the running spectrum, the amplitude of each frequency is used to represent the relative strength of the frequency in the current time period, and can reflect the degree of influence of the frequency on the running state in the current time period.

[0087] As an example, the real-time running data of the automation device corresponding to the current time period can be x(t), where t represents time, the electronic device can sample and discretize x(t) to obtain a time-domain discrete signal x(n), n can be a natural number not greater than the sampling point number N, which is determined by the sampling frequency fs and the time length of the current time period, and the electronic device can perform Fourier transform on x(n) to obtain a frequency-domain discrete signal X(k), where k represents frequency. The frequency-domain discrete signal X(k) can be used to represent the amplitude and phase of the time-domain discrete signal x(n) at a frequency of k*fs / N. The frequency-domain discrete signal X(k) is subjected to amplitude spectrum analysis or power spectrum analysis to obtain a running spectrum corresponding to the current time period. The running spectrum can include multiple peaks, each peak corresponding to a running frequency of the automation device, and the size of each peak representing the amplitude of each running frequency. The electronic device determines the frequency corresponding to the highest peak in the multiple peaks as the main running frequency corresponding to the current time period.

[0088] Step 430, compare the main running frequency corresponding to the current time period with the main running frequency corresponding to the previous time period to detect whether the running parameter of the automation device has changed.

[0089] The previous time period refers to a time period before the current time period. The electronic device can compare the running main frequency corresponding to the current time period with the running main frequency corresponding to the previous time period to obtain a comparison result. Then, whether the comparison result meets a preset requirement is determined. If the comparison result does not meet the preset requirement, it is detected that the running parameter of the automation device changes. If the comparison result meets the preset requirement, it is detected that the running parameter of the automation device does not change.

[0090] Optionally, the electronic device can determine whether the running main frequency corresponding to the current time period is the same as the running main frequency corresponding to the previous time period. The comparison result can include that the running main frequencies are the same or the running main frequencies are different. The preset requirement can be that the running main frequencies are the same. If the comparison result is that the running main frequencies are the same, it indicates that the preset requirement is met, and the electronic device detects that the running parameter of the automation device does not change. If the comparison result is that the running main frequencies are different, it indicates that the preset requirement is not met, and the electronic device detects that the running parameter of the automation device changes.

[0091] Optionally, the electronic device can calculate a frequency difference between the running main frequency corresponding to the current time period and the running main frequency of the previous time period, and compare the frequency difference with a preset frequency threshold. The comparison result can include that the frequency difference is greater than the frequency threshold and that the frequency difference is not greater than the frequency threshold. The preset requirement can be that the frequency difference is not greater than the frequency threshold. If the comparison result is that the frequency difference is not greater than the frequency threshold, it indicates that the preset requirement is met, and the electronic device detects that the running parameter of the automation device does not change. If the comparison result is that the frequency difference is greater than the frequency threshold, it indicates that the preset requirement is not met, and the electronic device detects that the running parameter of the automation device changes.

[0092] Step 440, if it is detected that the running parameter does not change, the real-time running data corresponding to the current time period is processed by the fault detection model to obtain a fault detection result corresponding to the automation device.

[0093] In an embodiment, if a change in the operating parameter of the automation device is detected, the electronic device can input the initial time series data to the fault detection model, process the initial time series data through the fault detection model to obtain predicted time series data, compare the initial time series data with the predicted time series data, and obtain a data difference between the initial time series data and the predicted time series data. If the data difference is not greater than a difference threshold, it is determined that the fault detection result corresponding to the automation device is that the automation device is operating normally. If the data difference is greater than the difference threshold, it is determined that the fault detection result corresponding to the automation device is that the automation device has a fault or is about to have a fault. Optionally, the difference threshold can include a first difference threshold and a second difference threshold, the first difference threshold being less than the second difference threshold. If the data difference is greater than the first difference threshold and not greater than the second difference threshold, the fault detection result corresponding to the automation device is that the automation device is about to have a fault. If the data difference is greater than the second difference threshold, the fault detection result corresponding to the automation device is that the automation device has a fault.

[0094] The initial time series data is data obtained by arranging the operating data corresponding to the current time period in chronological order, and the difference threshold is a. Optionally, the initial time series data can include initial data corresponding to a plurality of time points, and the predicted time series data also includes predicted data corresponding to a plurality of time points. The plurality of time points included in the initial time series data can be the same as the plurality of time points included in the predicted time series data. The electronic device can calculate the data difference between the initial data corresponding to each time point and the corresponding predicted data. The electronic device can perform mean calculation on the data difference between the initial data corresponding to each time point and the corresponding predicted data to obtain the data difference between the initial time series data and the predicted time series data.

[0095] As an optional implementation, the electronic device can calculate the data similarity between the initial time series data and the predicted time series data, and determine whether the data similarity is greater than a similarity threshold. If the data similarity is greater than the similarity threshold, it is determined that the fault detection result corresponding to the automation device is that the automation device is operating normally. If the data similarity is not greater than the similarity threshold, it is determined that the fault detection result corresponding to the automation device is that the automation device has a fault or is about to have a fault.

[0096] In an embodiment, the fault detection model can include a self-encoder model, which can include an encoder and a decoder. Since the feature dimension of the real-time running data is usually small, the encoder in the embodiment can be used for dimensionality increasing, and the decoder can be used for dimensionality reduction. Optionally, the electronic device can input the initial time series to the encoder, perform dimensionality increasing coding processing on the initial time series through the encoder to obtain an intermediate variable, and then input the intermediate variable to the decoder, perform dimensionality reduction decoding processing on the intermediate variable through the decoder, and obtain the predicted time series data.

[0097] The dimensionality increasing coding can refer to a process of encoding the initial data to generate a high-dimensional feature vector. In the embodiment, the initial data is the initial time series, and the high-dimensional feature vector is the intermediate variable. The high-dimensional feature vector can include the main features of the initial data and can better represent the initial data. The dimensionality increasing coding can include multiple transformation operations, such as linear transformation, nonlinear transformation, activation function, etc., so as to map the initial time series to a high-dimensional space. The dimensionality reduction decoding can refer to a process of remapping the high-dimensional feature vector that has undergone the dimensionality increasing coding back to the dimension of the initial data. The dimensionality reduction decoding can include multiple inverse transformation operations, such as linear inverse transformation, nonlinear inverse transformation, etc., to convert the high-dimensional feature vector into an output result having the same dimension as the initial data. In the embodiment, the high-dimensional feature vector is the intermediate variable, and the output result is the predicted time series data.

[0098] By implementing the embodiment, dimensionality increasing can be performed through the encoder of the self-encoder model to obtain more information in a high-dimensional space, and then dimensionality reduction can be performed through the decoder, so as to improve the prediction effect of the self-encoder model.

[0099] At step 450, if it is detected that the running parameter has changed, the fault detection model is trained through an incremental learning manner according to the real-time running data obtained within the first preset time length after the first preset time length.

[0100] The method of step 450 is the same as the method of step 240 in the above embodiment, which will not be described here.

[0101] In the embodiment, the electronic device can also analyze the real-time running data corresponding to the current time period, determine the running main frequency corresponding to the current time period, and compare the running main frequency corresponding to the current time period with the running main frequency corresponding to the previous time period, to detect whether the running parameter of the automation device has changed. Since the running main frequency is an important indicator of the automation device, comparing the running main frequencies can improve the accuracy of detecting whether the running parameter of the automation device has changed.

[0102] As Figure 5As shown, Figure 5 is a modular schematic diagram of a fault detection device disclosed by an embodiment of the present application. The fault detection device 500 can include a data acquisition module 510, a parameter detection module 520, a model processing module 530, and a model training module 540, wherein:

[0103] The data acquisition module 510 is configured to acquire real-time running data of the automation equipment.

[0104] The parameter detection module 520 is configured to analyze the real-time running data every preset time period to detect whether the running parameter of the automation equipment changes.

[0105] The model processing module 530 is configured to, if it is detected that the running parameter does not change, process the real-time running data corresponding to the current time period by using the fault detection model to obtain the fault detection result corresponding to the automation equipment.

[0106] The model training module 540 is configured to, if it is detected that the running parameter changes, train the fault detection model by using an incremental learning manner according to the real-time running data acquired within a first preset time length after the first preset time length.

[0107] In an embodiment, the parameter detection module 520 is further configured to analyze the real-time running data corresponding to the current time period to determine the running main frequency corresponding to the current time period; compare the running main frequency corresponding to the current time period with the running main frequency corresponding to the previous time period to detect whether the running parameter of the automation equipment changes.

[0108] In an embodiment, the parameter detection module 520 is further configured to perform Fourier transform on the real-time running data corresponding to the current time period to obtain the running spectrum corresponding to the current time period; determine the amplitude of each frequency in the running spectrum; compare the amplitudes of the frequencies, and determine the frequency with the largest amplitude as the running main frequency corresponding to the current time period.

[0109] In an embodiment, the parameter detection module 520 is further configured to calculate the frequency difference between the running main frequency corresponding to the current time period and the running main frequency of the previous time period; if the frequency difference is greater than a frequency threshold, it is detected that the running parameter of the automation equipment changes; if the frequency difference is not greater than the frequency threshold, it is detected that the running parameter of the automation equipment does not change.

[0110] In an embodiment, the model processing module 530 is further configured to input the initial time series data into the fault detection model, process the initial time series data through the fault detection model, and obtain predicted time series data; the initial time series data is data obtained by arranging running data corresponding to a current time period in chronological order; the fault detection model is trained according to target running data, and the running parameters corresponding to the target running data are the same as the running parameters corresponding to real-time running data of the current time period; the initial time series data is compared with the predicted time series data to obtain a data difference between the initial time series data and the predicted time series data; if the data difference is not greater than a difference threshold, it is determined that the fault detection result corresponding to the automation equipment is normal operation of the automation equipment; and if the data difference is greater than the difference threshold, it is determined that the fault detection result corresponding to the automation equipment is that the automation equipment has a fault or is about to have a fault.

[0111] In an embodiment, the fault detection model includes a self-encoder model, the self-encoder model includes an encoder and a decoder, the initial time series data is input into the fault detection model, and the model processing module 530 is further configured to input the initial time series into the encoder, perform dimensionality increasing coding processing on the initial time series through the encoder, and obtain an intermediate variable; the intermediate variable is input into the decoder, dimensionality decreasing decoding processing is performed on the intermediate variable through the decoder, and predicted time series data is obtained.

[0112] In an embodiment, the fault detection apparatus 500 further includes a pre-training module configured to, in a case where the automation equipment is in an experimental environment, take running data of the automation equipment acquired within a second preset time length as initial training data set; and pre-train a fault detection model to be trained based on the initial training data set, to obtain a pre-trained fault detection model.

[0113] In the embodiments of the present application, the electronic device can analyze real-time running data of the automation equipment, perform fault detection on the automation equipment when the running parameters of the automation equipment do not change, ensure the timeliness of fault detection, perform incremental training on the fault detection model when the running parameters of the automation equipment change, enable the fault detection model to better adapt to changes in the parameters of the automation equipment, improve the accuracy of the fault detection model, and thus improve the accuracy of fault detection. In addition, the electronic device performs fault detection based on real-time running data, without the need for shutdown or other interference operations of the automation equipment, so that the impact and loss of the fault detection operation on the production of the automation equipment can be avoided, and the usability and production efficiency of the automation equipment are improved. Furthermore, the electronic device can automatically detect whether the running parameters of the automation equipment change and automatically determine whether the fault detection model needs to be trained, reducing the need for manual intervention and improving the automation level and efficiency of fault detection.

[0114] As Figure 6 shown in one embodiment, an electronic device can include:

[0115] a memory 610 storing executable program code; and a processor 620 coupled with the memory 610;

[0116] The processor 620 invokes the executable program code stored in the memory 610 to implement the fault detection method provided in each of the above embodiments.

[0117] The memory 610 can include a Random Access Memory (RAM) and a Read-Only Memory (ROM). The memory 610 can be used to store instructions, programs, codes, code sets or instruction sets. The memory 610 can include a program storage area and a data storage area, where the program storage area can store instructions for implementing an operating system, instructions for implementing at least one function (such as a touch function, a sound playing function, an image playing function, etc.), instructions for implementing each of the above methods, etc. The data storage area can also store data created by the electronic device in use, etc.

[0118] The processor 620 can include one or more processing cores. The processor 620 connects various parts within the entire electronic device through various interfaces and lines, executes various functions of the electronic device and processes data by running or executing instructions, programs, code sets or instruction sets stored in the memory 610, and invoking data stored in the memory 610. Optionally, the processor 620 can be implemented in at least one of a hardware form of a Digital Signal Processing (DSP), a Field-Programmable Gate Array (FPGA), and a Programmable Logic Array (PLA). The processor 620 can integrate a combination of one or more of a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), and a modem, etc. Among them, the CPU is mainly responsible for processing an operating system, a user interface, and an application program, etc.; the GPU is responsible for rendering and drawing display content; and the modem is used for processing wireless communication. It can be understood that the above-mentioned modem can also not be integrated into the processor 620, but can be implemented by a separate communication chip.

[0119] It can be understood that the electronic device can include more or less structural elements than the above structural block diagram, for example, including a power module, a physical key, a WiFi (Wireless Fidelity) module, a speaker, a Bluetooth module, a sensor, and the like, and can not be limited herein.

[0120] The embodiments of the present application disclose a computer readable storage medium storing a computer program, wherein the computer program causes a computer to execute the method described in the above embodiments.

[0121] In addition, the embodiments of the present application further disclose a computer program product, when the computer program product runs on a computer, causes the computer to execute all or part of steps in any one of the fault detection methods described in the above embodiments.

[0122] A person of ordinary skill in the art can understand that all or part of the steps in the above embodiments can be completed by instructing the relevant hardware through a program, and the program can be stored in a computer readable storage medium, including a Read-Only Memory (ROM), a Random Access Memory (RAM), a Programmable Read-Only Memory (PROM), an Erasable Programmable Read-Only Memory (EPROM), a One-time Programmable Read-Only Memory (OTPROM), an Electrically-Erasable Programmable Read-Only Memory (EEPROM), a Compact Disc Read-Only Memory (CD-ROM) or other optical disk memories, magnetic disk memories, magnetic tape memories, or any other medium that can be used to carry or store data and can be read by a computer.

[0123] The above describes in detail the fault detection method, device, electronic device and storage medium disclosed by the embodiments of the present application. The principles and implementation manners of the present application are described by applying specific examples in this paper. The above embodiment is only used to help understand the method and core idea of the present application; at the same time, for a person of ordinary skill in the art, according to the idea of the present application, the specific implementation manner and application range will be changed, and the above description should not be understood as a limitation of the present application.

Claims

1. A fault detection method characterized by, The method comprises: acquiring real-time running data of an automation device; analyzing the real-time running data every preset time period to detect whether a running parameter of the automation device changes, comprising: analyzing real-time running data corresponding to a current time period to determine a running main frequency corresponding to the current time period; comparing the running main frequency corresponding to the current time period with a running main frequency corresponding to a previous time period to detect whether the running parameter of the automation device changes; if it is detected that the running parameter does not change, processing the real-time running data corresponding to the current time period by a fault detection model to obtain a fault detection result corresponding to the automation device, comprising: inputting initial time series data into the fault detection model, processing the initial time series data by the fault detection model to obtain predicted time series data; wherein the initial time series data is data obtained by arranging running data corresponding to the current time period in time sequence; the fault detection model is trained according to target running data, and a running parameter corresponding to the target running data is the same as a running parameter corresponding to the real-time running data of the current time period; comparing the initial time series data with the predicted time series data to obtain a data difference value between the initial time series data and the predicted time series data; if the data difference value is not greater than a difference threshold value, determining that the fault detection result corresponding to the automation device is that the automation device operates normally; if the data difference value is greater than the difference threshold value, determining that the fault detection result corresponding to the automation device is that the automation device has a fault or is about to have a fault; if it is detected that the running parameter changes, training the fault detection model by an incremental learning manner according to real-time running data acquired within a first preset time length after the first preset time length; the analyzing the real-time running data corresponding to the current time period to determine the running main frequency corresponding to the current time period, comprising: performing Fourier transform on the real-time running data corresponding to the current time period to obtain a running spectrum corresponding to the current time period; determining amplitudes of each frequency in the running spectrum; comparing the amplitudes of each frequency, and determining a frequency with the largest amplitude as the running main frequency corresponding to the current time period.

2. The method of claim 1, wherein, the comparing the running main frequency corresponding to the current time period with the running main frequency corresponding to the previous time period to detect whether the running parameter of the automation device changes, comprising: calculating a frequency difference value between the running main frequency corresponding to the current time period and the running main frequency of the previous time period; if the frequency difference value is greater than a frequency threshold value, detecting that the running parameter of the automation device changes; if the frequency difference value is not greater than the frequency threshold value, detecting that the running parameter of the automation device does not change.

3. The method of claim 1, wherein, The fault detection model comprises a self-encoder model comprising an encoder and a decoder, the initial time series data is input into the fault detection model, the initial time series data is processed through the fault detection model to obtain predicted time series data, comprising: the initial time series is input into the encoder, and the initial time series is processed through the encoder to obtain intermediate variables through dimensionality increasing encoding; the intermediate variables are input into the decoder, and the intermediate variables are processed through the decoder to obtain predicted time series data through dimensionality reducing decoding.

4. The method of claim 1, wherein, Before the real-time running data of the automation equipment is obtained, the method further comprises: in the case that the automation equipment is in an experimental environment, the running data of the automation equipment obtained within a second preset time length is taken as initial training data set; based on the initial training data set, the fault detection model to be trained is pre-trained to obtain a pre-trained fault detection model.

5. A fault detection apparatus characterized by comprising: The device comprises: a data acquisition module for acquiring real-time running data of an automation equipment; a parameter detection module for analyzing the real-time running data every preset time period to detect whether the running parameters of the automation equipment change; the parameter detection module is specifically configured to analyze the real-time running data corresponding to the current time period, determine the running main frequency corresponding to the current time period, and compare the running main frequency corresponding to the current time period with the running main frequency corresponding to the previous time period to detect whether the running parameters of the automation equipment change; a model processing module for processing the real-time running data corresponding to the current time period through a fault detection model to obtain a fault detection result corresponding to the automation equipment if it is detected that the running parameters do not change; the model processing module is specifically configured to input initial time series data into the fault detection model, process the initial time series data through the fault detection model to obtain predicted time series data; wherein the initial time series data is data obtained by arranging the running data corresponding to the current time period in time sequence; the fault detection model is trained according to target running data, the running parameters corresponding to the target running data are the same as the running parameters corresponding to the real-time running data of the current time period, the initial time series data and the predicted time series data are compared to obtain a data difference value between the initial time series data and the predicted time series data, if the data difference value is not greater than a difference threshold value, it is determined that the fault detection result corresponding to the automation equipment is that the automation equipment operates normally, and if the data difference value is greater than the difference threshold value, it is determined that the fault detection result corresponding to the automation equipment is that the automation equipment fails or is about to fail; The model training module is configured to, if it is detected that the operation parameter changes, train the fault detection model according to the real-time operation data obtained within the first preset time length by an incremental learning manner after the first preset time length. The parameter detection module is specifically configured to perform Fourier transform on the real-time operation data corresponding to a current time period to obtain an operation frequency spectrum corresponding to the current time period, determine the amplitude of each frequency in the operation frequency spectrum, compare the amplitudes of the frequencies, and determine the frequency with the largest amplitude as the operation main frequency corresponding to the current time period.

6. An electronic device, comprising: Comprise: a memory storing executable program codes; a processor coupled with the memory; The processor invokes the executable program codes stored in the memory to execute the method of any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a computer program, wherein the computer program, when executed by the processor, causes the processor to execute the method of any one of claims 1 to 4.

Images