Motor operation cycle monitoring and warning method and system

Abstract

The invention discloses a motor operation cycle monitoring and warning method and system, and relates to the technical field of motor monitoring, and the method comprises the steps: collecting motor operation data in real time through employing a distributed sensor network, and carrying out the analysis and processing of the collected original data through employing an edge computing device; transmitting the analyzed and processed data to a cloud data center, carrying out deep learning model training and feature extraction, and dividing a motor operation cycle based on the extracted features; and performing multi-dimensional anomaly detection on each divided operation cycle to generate an anomaly index, and performing motor operation cycle monitoring alarm according to the obtained anomaly index. The real-time performance and the reliability of the system can be improved. Compared with a traditional rule-based method, the method is more flexible and accurate, state changes of the motor can be evaluated more comprehensively, an early warning signal is sent out in time, and unnecessary downtime and maintenance cost are reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of motor monitoring, and in particular to a motor operation cycle monitoring and alarming method and system. Background Art

[0002] As a core component of industrial automation, motors play a vital role in modern manufacturing. With the introduction and promotion of the concept of Industry 4.0, intelligent monitoring of motor systems has become an important means to improve production efficiency and ensure equipment safety. Traditional motor monitoring mainly relies on regular manual inspections and simple alarm systems based on fixed thresholds. However, this monitoring method has many limitations, such as low efficiency of manual inspections and susceptibility to human factors, and the alarm mechanism based on fixed thresholds is difficult to adapt to complex and changing working environments, often leading to false alarms or missed alarms. In recent years, with the continuous improvement of Internet of Things (IoT) technology and big data analysis capabilities, intelligent monitoring systems have gradually become a new trend in motor status monitoring. These systems usually include sensor networks for collecting real-time data and central processing units to perform data analysis tasks. Despite this, most intelligent monitoring solutions on the market still face problems such as data processing delays and untimely model updates, especially when facing large-scale motor groups, the bottleneck of centralized architecture becomes more and more obvious.

[0003] Furthermore, for the accurate monitoring and fault prediction of the motor operation cycle, existing technical solutions often fail to fully utilize the advantages of edge computing. Edge computing can process sensor data locally, reduce data transmission delays and alleviate cloud computing pressure. However, the existing technology only performs simple preprocessing on the edge side and lacks the application of effective feature extraction and deep learning models, which limits its applicability and accuracy in complex scenarios. In addition, most of the existing cycle monitoring methods focus on a single indicator and ignore the correlation between multi-dimensional data during the operation of the motor, making anomaly detection not comprehensive enough and difficult to effectively identify potential risks. Therefore, how to build an efficient and accurate motor operation cycle monitoring system has become one of the key issues that need to be solved urgently. Summary of the invention

[0004] In view of the problems existing in the existing motor operation cycle monitoring and alarming methods, the present invention is proposed. Therefore, the problem to be solved by the present invention is how to provide a motor operation cycle monitoring and alarming method and system.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0006] In the first aspect, the present invention provides a motor operation cycle monitoring and alarm method, which includes using a distributed sensor network to collect motor operation data in real time, and using edge computing equipment to analyze and process the collected raw data; transmitting the analyzed and processed data to a cloud data center for deep learning model training and feature extraction, and dividing the motor operation cycle based on the extracted features; performing multi-dimensional anomaly detection on each divided operation cycle to generate an anomaly index, and performing motor operation cycle monitoring and alarming according to the obtained anomaly index.

[0007] As a preferred solution of the motor operation cycle monitoring and alarm method described in the present invention, the analysis and processing of the collected raw data includes the following steps: constructing a lightweight anomaly detection algorithm, running the lightweight anomaly detection algorithm in real time on the edge device to perform sliding window segmentation on the raw data, and calculating the statistical features within each window; using a local sensitive hashing algorithm to perform a fast similarity search on the feature vectors, and identifying potential anomalies based on the similarity search results; performing time series prediction on the identified anomalies, marking the data with priorities according to the prediction results, and determining the transmission order and compression ratio.

[0008] As a preferred solution of the motor operation cycle monitoring and alarm method described in the present invention, the construction of a lightweight anomaly detection algorithm includes the following steps: denoising and standardizing the original sensor data, segmenting the data using a sliding window method, and extracting multidimensional statistical features from each window; for each sliding window, calculating the local mean and standard deviation within the sliding window, the distance between the data point and the local mean, the neighbor density of the data point, and the time correlation between the data points; using an improved local sensitive hashing algorithm to perform a fast similarity search on the feature vector, estimate the local density of each data point, construct L hash tables, each table uses K hash functions, and calculates a hash value for each feature vector; searching for similar vectors in the hash table to calculate the local density, performing a time correlation analysis, and comprehensively considering the local density, distance deviation and time correlation to calculate the anomaly score of each data point, the expression is:

[0009] S(r)=w1*D(x)+w2*N(r)+w3*T(r)

[0010] Where D(x) represents the distance between the data point x and the local mean, N(x) represents the density of x’s k nearest neighbors, T(x) represents the time correlation of x, and w1, w2, and w3 are weight coefficients.

[0011] As a preferred solution of the motor operation cycle monitoring and alarm method described in the present invention, the division of the motor operation cycle includes the following steps: constructing a reference template library containing standard operation cycles under different working conditions, and normalizing the input time series; calculating the dynamic time regularization distance between the input sequence and each template in the template library, introducing an adaptive window mechanism, and dynamically adjusting the comparison window size; designing a multi-scale feature fusion strategy to comprehensively consider local and global features, and using a hierarchical clustering algorithm based on the fusion features to segment the time series, identifying the start and end points of the operation cycle according to the segmentation results, and assigning a unique identifier to each cycle. The calculation formula of the multi-scale feature fusion strategy is:

[0012] F=α*F local +(1-α)*F global

[0013]

[0014] Among them, F is the fusion feature, F local is the local feature, F global is the global feature, α is the fusion weight, f i is the i-th local feature, w i is the corresponding weight, and m is the number of local features.

[0015] As a preferred solution of the motor operation cycle monitoring and alarm method described in the present invention, the multi-dimensional anomaly detection for each divided operation cycle includes the following steps: constructing an LSTM model, using historical normal operation data for training, using Adam optimizer and MSE loss function, for each operation cycle, using sliding window method to predict to obtain a prediction value sequence, calculating prediction error sequence and time domain anomaly index; performing wavelet transform on the signal, performing wavelet decomposition, obtaining approximate coefficients and detail coefficients to extract spectral features, calculating the energy of each layer of wavelet coefficients and the statistical characteristics of the wavelet coefficients; using grid search method to optimize SVM parameters, classifying the spectral features of each operation cycle, obtaining anomaly probability, and calculating frequency domain anomaly index; designing a fusion strategy for anomaly detection results, comprehensively considering the detection results of each view, calculating a comprehensive anomaly index based on the fusion result, and triggering an alarm of corresponding level according to the comprehensive anomaly index value and the updated threshold.

[0016] As a preferred solution of the motor operation cycle monitoring and alarm method of the present invention, the calculation formula of the comprehensive abnormality index is:

[0017] AI=β1*AI t +β2*AI f +β3*AI p

[0018] Among them, AI is the comprehensive abnormality index, AI t 、AI f 、AI p are the anomaly indices of the time domain, frequency domain, and phase space views, respectively, and β1, β2, and β3 are the corresponding weights.

[0019] As a preferred solution of the motor operation cycle monitoring and alarm method described in the present invention, the motor operation cycle monitoring and alarming according to the obtained abnormal index includes: if the comprehensive abnormal index is below the first alarm threshold, the normal operating state is maintained; if the comprehensive abnormal index is greater than the first alarm threshold and below the second alarm threshold, the motor operation cycle first-level alarm is triggered, the data acquisition frequency is increased, the LSTM model weight and the SVM classifier are updated, the phase space analysis weight is increased for secondary detection, and if the comprehensive abnormal index of the secondary detection is below the first alarm threshold, the alarm is lifted; if the comprehensive abnormal index of the secondary detection is greater than the first alarm threshold, the motor operation cycle second-level alarm is triggered, the abnormal detection model is retrained, and the short-term prediction model is used for prediction; if the comprehensive abnormal index is greater than the second alarm threshold, the motor operation cycle third-level alarm is triggered, the historical database is called for comparative analysis, and the operating unit is shut down urgently.

[0020] In the second aspect, the present invention provides a motor operation cycle monitoring and alarm system, which includes: an analysis module, which is used to use a distributed sensor network to collect motor operation data in real time, and use an edge computing device to analyze and process the collected raw data; an extraction module, which is used to transmit the analyzed and processed data to a cloud data center, perform deep learning model training and feature extraction, and divide the motor operation cycle based on the extracted features; an alarm module, which is used to perform multi-dimensional anomaly detection on each divided operation cycle to generate an anomaly index, and perform motor operation cycle monitoring and alarm according to the obtained anomaly index.

[0021] In a third aspect, the present invention provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the processor executes the computer program, the steps of the motor operation cycle monitoring and alarm method are implemented.

[0022] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, the steps of the motor operation cycle monitoring and alarm method are implemented.

[0023] The beneficial effect of the present invention is that the real-time performance and reliability of the system are improved by reducing the data transmission bandwidth requirements and reducing the data processing time. Through deep learning technology, the system can automatically learn and identify complex nonlinear relationships, which is more flexible and accurate than traditional rule-based methods. It can achieve refined management of different stages in the motor life cycle, thereby extending the equipment life and optimizing maintenance strategies. The state changes of the motor can be evaluated more comprehensively. According to the obtained abnormal index, the system can issue early warning signals in time to reduce unnecessary downtime and maintenance costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0025] Figure 1 The figure is a flow chart of the motor operation cycle monitoring and alarm method. DETAILED DESCRIPTION

[0026] In order to make the above-mentioned purposes, features and advantages of the present invention more understandable, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0027] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0028] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0029] Example 1

[0030] Reference Figure 1 , which is the first embodiment of the present invention, provides a motor operation cycle monitoring and alarm method, comprising:

[0031] S1: Use a distributed sensor network to collect motor operation data in real time, and use edge computing devices to analyze and process the collected raw data;

[0032] Specifically, a distributed sensor network is used to collect motor operation data in real time, including multi-dimensional parameters such as current, voltage, speed, vibration and temperature; edge computing devices are used to pre-process and preliminarily analyze the collected raw data to reduce the data transmission load;

[0033] A lightweight anomaly detection algorithm is run in real time on edge devices to segment the raw data into sliding windows and calculate the statistical features within each window.

[0034] The locality sensitive hashing (LSH) algorithm is used to perform fast similarity search on feature vectors and identify potential outliers based on similarity search results.

[0035] Perform time series forecasting on the identified anomalies, evaluate their development trends, and based on the forecast results, prioritize the data, determine the transmission order and compression ratio.

[0036] Define the anomaly scoring function S(x), the expression is:

[0037] S(x)=w1*D(r)+w2*N(r)+w3*T(x)

[0038] Where D(x) represents the distance between the data point x and the local mean, N(x) represents the density of the k nearest neighbors of x, T(x) represents the time correlation of x, and w1, w2 and w3 are weight coefficients;

[0039] Compute the local mean and standard deviation within a sliding window:

[0040]

[0041]

[0042] Among them, μ is the local mean in the sliding window, σ is the standard deviation in the sliding window, n is the window size, and xi is the data point in the window; the distance between the data point x and the local mean is calculated as follows:

[0043]

[0044] The LSH algorithm is used to calculate the k nearest neighbor density of x, and the expression is:

[0045]

[0046] Where r is the radius of the hypersphere centered on x, so that the hypersphere contains exactly k neighboring points; calculate the time correlation, the expression is:

[0047]

[0048] Where x' is the data at the previous time point, Δt is the time interval; if S(x)>θ, x is marked as a potential outlier, where θ is the preset threshold.

[0049] S2: The analyzed and processed data is transmitted to the cloud data center for deep learning model training and feature extraction. Based on the extracted features, the improved dynamic time warping algorithm is used to divide the motor operation cycle;

[0050] Specifically, the preprocessed data is transmitted to a cloud data center for deep learning model training and feature extraction, and based on the extracted features, an improved dynamic time warping algorithm is used to divide the motor operation cycle;

[0051] Build a reference template library containing standard operating cycles under different working conditions and normalize the input time series;

[0052] Calculate the dynamic time warping distance between the input sequence and each template in the template library, introduce an adaptive window mechanism, and dynamically adjust the comparison window size;

[0053] A multi-scale feature fusion strategy is designed, which comprehensively considers local and global features. Based on the fusion features, a hierarchical clustering algorithm is used to segment the time series. The segmentation results are used to identify the start and end points of the operation cycle, and a unique identifier is assigned to each cycle. The calculation formula is:

[0054] F=α*F local +(1-α)*F global

[0055]

[0056] Among them, F is the fusion feature, F local is the local feature, F global is the global feature, α is the fusion weight, f i is the i-th local feature, w i is the corresponding weight, m ​​is the number of local features, CNN represents convolutional neural network, and X is the input time series.

[0057] S3: Perform multi-dimensional anomaly detection on each divided operation cycle to generate an anomaly index, adjust the alarm threshold and trigger an alarm of the corresponding level.

[0058] Specifically, multi-dimensional anomaly detection is performed on each divided operation cycle to generate an anomaly index. According to the severity of the anomaly index, the alarm threshold is adaptively adjusted and the corresponding level of alarm is triggered. The graph neural network model is used to predict potential failure risks and give maintenance recommendations based on historical data and expert knowledge.

[0059] Build a multi-view anomaly detection framework, including time domain view, frequency domain view and phase space view;

[0060] In the time domain view, Long Short-Term Memory Network (LSTM) is applied to capture abnormalities in timing patterns;

[0061] In the frequency domain view, wavelet transform is used to extract spectral features, and support vector machine (SVM) is used to detect frequency domain anomalies;

[0062] In the phase space view, phase space reconstruction is constructed and dynamical anomalies are identified by calculating complexity metrics of the recurrence graph;

[0063] Build an LSTM model, use historical normal operation data for training, use the Adam optimizer and MSE loss function, use the sliding window method to predict each operation cycle to obtain the predicted value sequence, and calculate the prediction error sequence and time domain anomaly index;

[0064] Perform wavelet transform on the signal, perform wavelet decomposition, obtain approximate coefficients and detail coefficients to extract spectrum features, calculate the energy of each layer of wavelet coefficients and the statistical characteristics of the wavelet coefficients;

[0065] The grid search method is used to optimize the SVM parameters, classify the spectrum characteristics of each operation cycle, obtain the abnormal probability, and calculate the frequency domain abnormality index;

[0066] Design a fusion strategy for anomaly detection results, comprehensively consider the detection results of each view, calculate the comprehensive anomaly index based on the fusion results, and trigger the corresponding level of alarm according to the comprehensive anomaly index value and the updated threshold. The calculation formula for the comprehensive anomaly index is:

[0067] AI=β1*AI t +β2*AI f +β3*AI p

[0068] Among them, AI is the comprehensive abnormality index, AI t 、AI f 、AI p are the anomaly indices of the time domain, frequency domain, and phase space views, respectively, and β1, β2, and β3 are the corresponding weights.

[0069] The first alarm threshold and the second alarm threshold for triggering an alarm are obtained based on the comprehensive abnormal index in the historical database. The calculation formula is:

[0070]

[0071] In the formula, Al i is the alarm threshold, ω jis the weight of the comprehensive abnormality index obtained in the jth alarm monitoring, Al j is the comprehensive abnormality index obtained in the jth alarm monitoring; when i=1, it represents the first alarm threshold, and n is the comprehensive abnormality index of the last five times; when i=2, it represents the second alarm threshold, and n is the comprehensive abnormality index of the last three times;

[0072] If the comprehensive abnormality index is below the first alarm threshold, the normal operation state is maintained;

[0073] If the comprehensive abnormality index is greater than the first alarm threshold and below the second alarm threshold, the first-level alarm of the motor operation cycle is triggered, the data collection frequency is increased, the LSTM model weight and SVM classifier are updated, and the phase space analysis weight is increased for secondary detection. If the comprehensive abnormality index of the secondary detection is below the first alarm threshold, the alarm is lifted;

[0074] If the comprehensive abnormality index of the secondary detection is greater than the first alarm threshold, the second alarm of the motor operation cycle is triggered, the abnormality detection model is retrained, and the short-term prediction model is used for prediction;

[0075] If the comprehensive abnormality index is greater than the second alarm threshold, the third-level alarm of the motor operation cycle is triggered, the historical database is called for comparative analysis, and the running unit is shut down urgently.

[0076] Furthermore, the present embodiment also provides a motor operation cycle monitoring and alarm system, including: an analysis module, which is used to adopt a distributed sensor network to collect motor operation data in real time, and use an edge computing device to analyze and process the collected raw data; an extraction module, which is used to transmit the analyzed and processed data to a cloud data center for deep learning model training and feature extraction, and divide the motor operation cycle based on the extracted features; an alarm module, which is used to perform multi-dimensional anomaly detection on each divided operation cycle to generate an anomaly index, and perform motor operation cycle monitoring and alarm according to the obtained anomaly index.

[0077] This embodiment also provides a computer device, which is suitable for the motor operation cycle monitoring and alarm method, including: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement all or part of the steps of the method described in the embodiment of the present invention as proposed in the above embodiment.

[0078] This embodiment also provides a storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the method in any optional implementation of the above embodiment is executed. Wherein, the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read-Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic memory, flash memory, disk or optical disk.

[0079] The storage medium proposed in this embodiment and the data storage method proposed in the above embodiment belong to the same inventive concept. The technical details not fully described in this embodiment can be found in the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

[0080] As can be seen from the above, this method improves the real-time performance and reliability of the system by reducing the data transmission bandwidth requirements and reducing the data processing time. Through deep learning technology, the system can automatically learn and identify complex nonlinear relationships, which is more flexible and accurate than traditional rule-based methods. It can achieve refined management of different stages in the motor life cycle, thereby extending the equipment life and optimizing maintenance strategies. The state changes of the motor can be evaluated more comprehensively. According to the obtained abnormal index, the system can issue early warning signals in time to reduce unnecessary downtime and maintenance costs.

[0081] Example 2

[0082] Referring to Table 1, which is a second embodiment of the present invention, this embodiment provides a motor operation cycle monitoring and alarm method. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0083] Table 1 Test data table

[0084]

[0085] As can be seen from the above, this method improves the real-time performance and reliability of the system by reducing the data transmission bandwidth requirements and reducing the data processing time. Through deep learning technology, the system can automatically learn and identify complex nonlinear relationships, which is more flexible and accurate than traditional rule-based methods. It can achieve refined management of different stages in the motor life cycle, thereby extending the equipment life and optimizing maintenance strategies. The state changes of the motor can be evaluated more comprehensively. According to the obtained abnormal index, the system can issue early warning signals in time to reduce unnecessary downtime and maintenance costs.

[0086] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A motor operation cycle monitoring and alarm method, characterized in that: include, Use a distributed sensor network to collect motor operation data in real time, and use edge computing devices to analyze and process the collected raw data; The analyzed and processed data is transmitted to the cloud data center for deep learning model training and feature extraction, and the motor operation cycle is divided based on the extracted features; Multi-dimensional anomaly detection is performed on each divided operation cycle to generate an anomaly index, and motor operation cycle monitoring and alarm are performed based on the obtained anomaly index.

2. The motor operation cycle monitoring and alarm method according to claim 1, characterized in that: The analysis and processing of the collected raw data includes the following steps: Build a lightweight anomaly detection algorithm and run it in real time on edge devices to segment the original data into sliding windows and calculate the statistical features within each window. Use the local sensitive hashing algorithm to quickly search for similarity on feature vectors and identify potential outliers based on similarity search results; Perform time series prediction on the identified anomalies, mark the data with priority based on the prediction results, and determine the transmission order and compression ratio.

3. The motor operation cycle monitoring and alarm method according to claim 2, characterized in that: The construction of a lightweight anomaly detection algorithm includes the following steps: The raw sensor data is denoised and standardized, the data is segmented using the sliding window method, and multidimensional statistical features are extracted from each window; For each sliding window, calculate the local mean and standard deviation within the sliding window, the distance between the data point and the local mean, the neighbor density of the data point, and the time correlation between the data points; Use the improved local sensitive hashing algorithm to perform fast similarity search on feature vectors, estimate the local density of each data point, build L hash tables, each table uses K hash functions, and calculate the hash value for each feature vector; Find similar vectors in the hash table to calculate local density, perform time correlation analysis, and comprehensively consider local density, distance deviation, and time correlation to calculate the anomaly score of each data point. The expression is: S(x)=w1*D(x)+w2*N(x)+w3*T(x) Where D(x) represents the distance between the data point x and the local mean, N(x) represents the density of x’s k nearest neighbors, T(x) represents the time correlation of x, and w1, w2, and w3 are weight coefficients.

4. The motor operation cycle monitoring and alarm method according to claim 3, characterized in that: The division of the motor operation cycle comprises the following steps: Build a reference template library containing standard operating cycles under different working conditions and normalize the input time series; Calculate the dynamic time warping distance between the input sequence and each template in the template library, introduce an adaptive window mechanism, and dynamically adjust the comparison window size; A multi-scale feature fusion strategy is designed to comprehensively consider local and global features. A hierarchical clustering algorithm is used to segment the time series based on the fused features. The start and end points of the operation cycle are identified according to the segmentation results, and a unique identifier is assigned to each cycle.

5. The motor operation cycle monitoring and alarm method according to claim 4, characterized in that: The multi-dimensional anomaly detection for each divided operation cycle comprises the following steps: Build an LSTM model, use historical normal operation data for training, use the Adam optimizer and MSE loss function, use the sliding window method to predict each operation cycle to obtain the predicted value sequence, and calculate the prediction error sequence and time domain anomaly index; Perform wavelet transform on the signal, perform wavelet decomposition, obtain approximate coefficients and detail coefficients to extract spectrum features, calculate the energy of each layer of wavelet coefficients and the statistical characteristics of the wavelet coefficients; The grid search method is used to optimize the SVM parameters, classify the spectrum characteristics of each operation cycle, obtain the abnormal probability, and calculate the frequency domain abnormality index; Design a fusion strategy for anomaly detection results, comprehensively consider the detection results of each view, calculate the comprehensive anomaly index based on the fusion results, and trigger the corresponding level of alarm according to the comprehensive anomaly index value and the updated threshold.

6. The motor operation cycle monitoring and alarm method according to claim 5, characterized in that: The calculation formula of the comprehensive abnormality index is: AI=β1*AI t +β2*AI f +β3*AI p Among them, AI is the comprehensive abnormality index, AI t 、AI f 、AI p are the anomaly indices of the time domain, frequency domain, and phase space views, respectively, and β1, β2, and β3 are the corresponding weights.

7. The motor operation cycle monitoring and alarm method according to claim 6, characterized in that: The motor operation cycle monitoring and alarming according to the obtained abnormal index includes maintaining a normal operation state if the comprehensive abnormal index is below a first alarm threshold; If the comprehensive abnormality index is greater than the first alarm threshold and below the second alarm threshold, the first-level alarm of the motor operation cycle is triggered, the data collection frequency is increased, the LSTM model weight and SVM classifier are updated, and the phase space analysis weight is increased for secondary detection. If the comprehensive abnormality index of the secondary detection is below the first alarm threshold, the alarm is lifted; If the comprehensive abnormality index of the secondary detection is greater than the first alarm threshold, the second alarm of the motor operation cycle is triggered, the abnormality detection model is retrained, and the short-term prediction model is used for prediction; If the comprehensive abnormality index is greater than the second alarm threshold, the third-level alarm of the motor operation cycle is triggered, the historical database is called for comparative analysis, and the running unit is shut down urgently.

8. A motor operation cycle monitoring and alarm system, based on the motor operation cycle monitoring and alarm method according to any one of claims 1 to 7, characterized in that: include, An analysis module is used to collect motor operation data in real time using a distributed sensor network and analyze and process the collected raw data using edge computing devices; The extraction module is used to transmit the analyzed and processed data to the cloud data center for deep learning model training and feature extraction, and divide the motor operation cycle based on the extracted features; The alarm module is used to perform multi-dimensional anomaly detection on each divided operation cycle to generate an anomaly index, and perform motor operation cycle monitoring and alarm according to the obtained anomaly index.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the motor operation cycle monitoring and alarm method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the motor operation cycle monitoring and alarm method according to any one of claims 1 to 7 are implemented.

Images