Method and system for detecting running state of electromechanical equipment in real time
Through wavelet denoising, SMOTE expansion, kernel function adaptive synthetic data generation and particle swarm optimization algorithm, the Gaussian kernel function SVM model is constructed, which solves the noise and sample imbalance in the operating state detection of electromechanical equipment, and achieves a more accurate Real-time detection and early detection of faults.
Patent Information
- Application Number
- CN202411408625.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-10
- Publication Date
- 2024-11-08
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing technology has noisy data in the detection of the operating status of electromechanical equipment that affects the generalization capability of the model, sample imbalance leads to inaccurate detection, and traditional algorithm training relies on manual intervention, making it difficult to achieve real-time monitoring and early detection of faults.
The wavelet denoising method is used to eliminate noise data, improve the SMOTE algorithm for non-marginal expansion, and the adaptive synthetic data generation method based on the kernel function is used to perform marginal expansion, and combine the particle swarm optimization algorithm to tune the SVM model to construct the operating state of the Gaussian kernel function. Detect the model.
It improves the generalization ability and detection accuracy of the model, reduces the impact of noise, avoids data marginalization, and realizes accurate and timely detection of the operating status of electromechanical equipment.
Smart Images

Figure CN118916774A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of equipment status detection, and in particular relates to a real-time detection method and system for the operating status of electromechanical equipment. Background Art
[0002] With the rapid development of railway transportation system, the safety and efficiency of railway platforms have received increasing attention.
[0003] As an important part of railway platforms, real-time monitoring and diagnosis of the operating status of electromechanical equipment is crucial to ensuring the safety and reliability of railway transportation.
[0004] At present, railway units still implement a manual regular inspection system for platform electromechanical equipment. They judge the operating status of equipment based on experience, such as listening, touching, and observing. This inspection task makes employees feel exhausted, and they are unable to grasp the equipment inspection status in real time. Once an emergency failure occurs, it is difficult to discover it in time, resulting in an accident. At the same time, early detection and diagnosis of the fault cannot be achieved, resulting in the expansion of the fault and the inability to implement planned maintenance and advanced maintenance.
[0005] With the rapid development of artificial intelligence, the use of machine learning detection models to detect and predict the status of electromechanical equipment has also achieved certain developments. However, before traditional machine learning algorithms can evaluate the operating status of electromechanical equipment, they need to be trained. Due to differences in the sampling environment of electromechanical equipment operating status data, there is usually an imbalance in operating data samples, which affects the training of machine learning algorithms and makes it impossible to accurately evaluate and detect the operating status of electromechanical equipment. Summary of the invention
[0006] The purpose of the embodiments of the present invention is to provide a method and system for real-time detection of the operating status of electromechanical equipment, aiming to solve the technical problems raised in the above-mentioned background technology.
[0007] To achieve the above object, the present invention provides the following technical solutions.
[0008] In a first aspect of the present invention, a method for real-time detection of the operating status of an electromechanical device is provided, the detection method comprising the following steps: The noise data is removed from the monitored electromechanical equipment operation data by using the wavelet denoising method to obtain minority class samples of the original data of the electromechanical equipment working status; The non-marginal expansion of minority class samples is performed by improving the SMOTE algorithm; the adaptive synthetic data generation method based on kernel function is used to marginalize and expand the different class samples of minority class samples to obtain the electromechanical equipment balanced data set; The Gaussian kernel function is selected as the kernel function of the SVM classifier, the penalty factor c and the kernel function parameter g are initialized, and the SVM model is trained using a balanced data set. During the training process, the improved particle swarm optimization algorithm is used to tune the parameters of the SVM model to obtain a trained running status detection model. The real-time data of the operating status of the electromechanical equipment is obtained by using a data collector, and the real-time data of the operating status is used as the input of the trained operating status detection model to output the detection classification results corresponding to the real-time data.
[0009] As a further solution of the present invention, the step of removing noise data from the monitored electromechanical equipment operation data by using a wavelet denoising method includes: Establish the fault model of electromechanical equipment operation data, the fault model is expressed as: ; In formula (1), Represented as electromechanical equipment operating data including noise data; Indicates the initial electromechanical equipment operation data; is the noise data; Perform wavelet transform on the fault model and get: ; The electromechanical equipment operation data is decomposed by wavelet, and the wavelet basis function and layered data are selected for splitting to obtain multiple wavelet levels and corresponding wavelet coefficients; the wavelet coefficients are threshold processed, and the corresponding threshold and threshold function are selected according to the decomposed wavelet coefficients, which are expressed as: ; In formula (2) and formula (3), Indicates the wavelet corresponding to different data; y indicates the maximum decomposition degree, and k indicates the data length; The value range of is [0,1]; represents the threshold; the threshold function is Medium continuous; According to the threshold function According to the continuity of the data and the feature requirements of the data at different scales, the optimal estimate is obtained and the new threshold representation is obtained: ; In formula (4), y represents the degree of decomposition; N is the signal size; Indicates the noise level, It is expressed as: ; The processed coefficients are synthesized into denoised data through wavelet reconstruction.
[0010] As a further solution of the present invention, the step of performing non-marginal expansion on minority class samples by improving the SMOTE algorithm includes: The minority class samples are divided into noise points and non-noise points, and the extended samples are constructed by linear interpolation between the non-noise points and the middle points of the minority class samples. The extended samples are distributed near the center points of the original minority class samples, which can effectively avoid the marginalization of the extended data.
[0011] As a further solution of the present invention, the calculation process of the middle point of the minority class sample includes: taking the mean value of each characteristic attribute of the minority class sample after normalization as the center point of the minority class sample, expressed as: ; In formula (5), represents the i-th sample under the r-th feature attribute after normalization preprocessing, and n represents the number of minority class samples; At the center of the minority class samples Data expansion is performed between non-noise points using the following formula: ; In formula (6), represents the newly expanded minority class samples, Represents the minority class samples in the non-noise point set; Represents a random number that satisfies a uniform distribution on the interval [0,1].
[0012] As a further solution of the present invention, the step of using the kernel function-based adaptive synthetic data generation method to marginalize and expand different class samples of minority class samples includes: The weight is calculated based on the proportion of majority class samples among the k nearest neighbors of minority class samples; Determine the number of synthetic samples that need to be generated for each minority class sample based on the weight and the preset ratio; The probability density of minority class samples is estimated using the kernel function, and the width of the kernel function is controlled by the bandwidth parameter; Using the Markov Chain Monte Carlo method to generate synthetic samples based on kernel density estimation, the initial samples are randomly selected from the minority class samples, and then new sample points are generated by adding random perturbations until the required number of synthetic samples is generated. The generated synthetic samples are merged with the original majority class samples to form the electromechanical device balancing dataset.
[0013] As a further solution of the present invention, the step of using the improved particle swarm optimization algorithm to tune the parameters of the SVM classification model includes: Set the number of particles and the maximum iteration of the particle swarm. At this time, set the penalty factor c and kernel function parameter g of the SVM model to the two-dimensional coordinates of the individual particle position; Randomly initialize the penalty factor c and kernel function parameter g; Use samples from the balanced data set to train the SVM model; Update the position and speed of individual particles, calculate the fitness value of each individual particle after the update, and determine the relationship between the optimal fitness value of the current iteration number and the fitness value of each individual particle after the update to obtain the position of the optimal individual particle in the particle group; When the maximum number of iterations is reached, the iteration is stopped and the optimal penalty factor c and kernel function parameter g are output; The optimal penalty factor c and kernel function parameter g are used to construct an SVM-based running status detection model.
[0014] As a further solution of the present invention, the step of updating the position and velocity of individual particles includes: Suppose in D-dimensional space, the number of particles is n, and the position of the i-th particle is and speed Respectively expressed as: ; ; The optimal position of individual particles and the optimal position of the particle group are expressed as: ; ; When the particle individual optimal position Optimal position of the population When it is better, update the optimal position of the population; For any dimension The particle position and velocity, the velocity and position update formula are as follows: ; ; In equations (11) and (12), V represents the particle velocity, and are acceleration factors, represents a random number in the interval [0,1]; x represents the particle position, represents the optimal position of the particle; Represents the optimal position of the particle swarm.
[0015] In a second aspect of the present invention, a real-time detection system for the operating status of electromechanical equipment is provided, the detection system comprising: In the data processing module, the user uses the wavelet denoising method to remove noise data from the monitored electromechanical equipment operation data, and obtains minority class samples of the original data of the electromechanical equipment working status; The data expansion module is used to perform non-marginal expansion of minority class samples by improving the SMOTE algorithm; the kernel function-based adaptive synthetic data generation method is used to perform marginal expansion of different class samples of minority class samples to obtain a balanced data set of electromechanical equipment; The model training module is used to select the Gaussian kernel function as the kernel function of the SVM classifier, initialize the penalty factor c and the kernel function parameter g, and use the balanced data set to train the SVM model. During the training process, the improved particle swarm optimization algorithm is used to tune the parameters of the SVM model to obtain a trained running status detection model; The status detection module is used to obtain the real-time data of the operating status of the electromechanical equipment using the data collector, use the real-time data of the operating status as the input of the trained operating status detection model, and output the detection classification results corresponding to the real-time data.
[0016] Compared with the prior art, the real-time detection method and system for the operation status of electromechanical equipment of the present invention have the following beneficial effects: First, the present invention removes noise data from the monitored electromechanical equipment operation data through a wavelet denoising method, obtains minority class samples of the original data of the electromechanical equipment working status, and further uses the denoised data as a training set input, which can improve the generalization ability of the model and improve the performance of the classification or regression model; and by reducing the influence of noise data, the complexity and computational cost of subsequent data processing can be reduced; Second, the present invention improves the SMOTE algorithm to perform non-marginal expansion on minority class samples, so that data expansion is performed within the same sample data, and the expanded minority class samples are distributed near the center point of the original sample, which effectively avoids the marginalization of the expanded data and avoids the subjectivity of the k value selection in the traditional SMOTE algorithm; Third, the present invention adopts an adaptive synthetic data generation method based on kernel function to marginalize and expand different types of samples of minority class samples, realizes secondary expansion of sample data, and enhances the boundary distinction between samples of different classes; Fourth, the present invention selects a Gaussian kernel function as the kernel function of the SVM classifier, uses a balanced data set to train the SVM model, and uses an improved particle swarm optimization algorithm to tune the parameters of the SVM model, thereby obtaining an electromechanical equipment operation status detection model with excellent performance, thereby improving the accuracy of electromechanical equipment status operation detection. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0018] Figure 1 is a system architecture diagram of the present invention; Figure 2 This is a flow chart of a method for real-time detection of the operating status of electromechanical equipment according to the present invention; Figure 3 A sub-flow chart of the real-time detection method of the operating status of electromechanical equipment of the present invention; Figure 4 Another sub-flow chart of the real-time detection method of the operating status of electromechanical equipment of the present invention; Figure 5 It is another sub-flow chart of the real-time detection method of the operating status of electromechanical equipment of the present invention; Figure 6 The present invention is a structural block diagram of a real-time detection system for the operating status of electromechanical equipment. DETAILED DESCRIPTION
[0019] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings. In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below; It should be noted that the various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments, and the same or similar parts between the various embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description.
[0020] For those skilled in the art, it can be further appreciated that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.
[0021] It should be noted that the steps of the methods or algorithms described in conjunction with the embodiments disclosed herein may be directly implemented using hardware, a software module executed by a processor, or a combination of the two.
[0022] The software module may be placed in random access memory (RAM), memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the technical field.
[0023] like Figure 1 As shown, it is a system architecture diagram of the present invention, which includes a basic data layer 101, a control layer 102 and an application layer 103; the basic data layer 101 of the present invention is used to collect the operating status data of the electromechanical equipment, and the basic data layer is distributed with a number of data collection nodes, and the task of each data collection node is to collect the working status data of the electromechanical equipment through the data collector and complete the upload processing of the data information; the control layer 102 has the function of processing and controlling the working status data of the electromechanical equipment, and is composed of a data processing module and a control module; the application layer 103 is composed of a server client and a switch, and its function is to centrally manage the working status data of the electromechanical equipment, and can realize the functions of displaying and querying the working status control results of the electromechanical equipment.
[0024] Preferably, data communication transmission is achieved between the basic data layer 101 and the control layer 102, and between the control layer 102 and the application layer 103 through a network, wherein the network may include various types of wired communication links or wireless communication links, for example: wired communication links include optical fibers, twisted pairs or coaxial cables, and wireless communication links include Bluetooth communication links and wireless fidelity (WIreless-FIdelity, Wi-Fi) communication links.
[0025] The specific implementation of the real-time detection method and system for the operating status of electromechanical equipment of the present invention is described in detail below in conjunction with specific embodiments.
[0026] like Figure 2As shown, in one embodiment of the present invention, a real-time detection method for the operating status of an electromechanical device is provided, and the detection method comprises the following steps: Step S201: removing noise data from the monitored electromechanical equipment operation data by using a wavelet denoising method to obtain minority class samples of the original data of the electromechanical equipment working status; Step S202: non-marginal expansion of minority class samples is performed by improving the SMOTE algorithm; and different class samples of minority class samples are marginalized and expanded by using an adaptive synthetic data generation method based on a kernel function to obtain a balanced data set of electromechanical equipment; Step S203: Selecting a Gaussian kernel function as the kernel function of the SVM classifier, initializing the penalty factor c and the kernel function parameter g, and using the balanced data set to train the SVM model. During the training process, an improved particle swarm optimization algorithm is used to tune the parameters of the SVM model to obtain a trained running status detection model; Step S204: using a data collector to obtain real-time data on the operating status of the electromechanical equipment, using the real-time data on the operating status as input to a trained operating status detection model, and outputting a detection classification result corresponding to the real-time data.
[0027] Further, such as Figure 3 As shown, the step of removing noise data from the monitored electromechanical equipment operation data by using the wavelet denoising method includes: Step S301: Establish a fault model for the operation data of the electromechanical equipment. The fault model is expressed as: ; In formula (1), Represented as electromechanical equipment operating data including noise data; Indicates the initial electromechanical equipment operation data; is the noise data; Step S302: Perform wavelet transform on the fault model to obtain: ; Step S303: perform wavelet decomposition on the electromechanical equipment operation data, select wavelet basis functions and layered data for splitting, and obtain multiple wavelet levels and corresponding wavelet coefficients; perform threshold processing on the wavelet coefficients, and select corresponding thresholds and threshold functions according to the decomposed wavelet coefficients, which are expressed as: ; In formula (2) and formula (3), Indicates the wavelet corresponding to different data; y indicates the maximum decomposition degree, and k indicates the data length; The value range of is [0,1]; represents the threshold; the threshold function is Medium continuous; Step S304: According to the threshold function According to the continuity of the data and the feature requirements of the data at different scales, the optimal estimate is obtained and the new threshold representation is obtained: ; In formula (4), y represents the degree of decomposition; N is the signal size; Indicates the noise level, It is expressed as: ; Step S305: synthesize the processed coefficients into denoised data through wavelet reconstruction.
[0028] Preferably, the step of performing non-marginal expansion on minority class samples by improving the SMOTE algorithm provided in an embodiment of the present invention includes: dividing the minority class samples into noise points and non-noise points, constructing expanded samples by using a linear interpolation method between the non-noise points and the middle points of the minority class samples, and distributing the expanded samples near the center points of the original minority class samples, which can effectively avoid the marginalization of the expanded data.
[0029] As a further solution of the present invention, the calculation process of the middle point of the minority class sample includes: taking the mean value of each characteristic attribute of the minority class sample after normalization as the center point of the minority class sample, expressed as: ; In formula (5), represents the i-th sample under the r-th feature attribute after normalization preprocessing, and n represents the number of minority class samples; At the center of the minority class samples Data expansion is performed between non-noise points using the following formula: ; In formula (6), represents the newly expanded minority class samples, Represents the minority class samples in the non-noise point set; Represents a random number that satisfies a uniform distribution on the interval [0,1].
[0030] like Figure 4 As shown, in the embodiment of the present invention, the step of using the kernel function-based adaptive synthetic data generation method to marginalize and expand different class samples of minority class samples includes: Step S401: Calculate the weight according to the proportion of majority class samples in the k nearest neighbors of minority class samples; Step S402: Determine the number of synthetic samples that need to be generated for each minority class sample according to the weight and the preset ratio; Step S403: using a kernel function to estimate the probability density of minority class samples, the width of the kernel function is controlled by a bandwidth parameter; Step S404: using the Markov chain Monte Carlo method to generate synthetic samples based on kernel density estimation; wherein, initial samples are randomly selected from the minority class samples, and then new sample points are generated by adding random perturbations until the required number of synthetic samples is generated; Step S405: merging the generated synthetic samples with the original majority class samples to form a mechatronics equipment balanced data set.
[0031] like Figure 5 As shown, in a preferred embodiment of the present invention, the step of using the improved particle swarm optimization algorithm to perform parameter tuning on the SVM classification model includes: Step S501: Setting the number of particles and the maximum iteration of the particle swarm. At this time, the penalty factor c and the kernel function parameter g of the SVM model are set to the two-dimensional coordinates of the individual particle position; Step S502: Randomly initialize the penalty factor c and kernel function parameter g; Step S503: training the SVM model using samples in the balanced data set; Step S504: updating the position and speed of individual particles, calculating the fitness value of each individual particle after the update, determining the relationship between the optimal fitness value of the current iteration number and the fitness value of each individual particle after the update, and obtaining the position of the optimal individual particle in the particle group; Step S505: When the maximum number of iterations is reached, the iteration is stopped and the optimal penalty factor c and kernel function parameter g are output; Step S506: constructing a running status detection model based on SVM using the optimal penalty factor c and kernel function parameter g.
[0032] Preferably, in an embodiment of the present invention, the step of updating the individual particle positions and velocities comprises: Suppose in D-dimensional space, the number of particles is n, and the position of the i-th particle is and speed Respectively expressed as: ; ; The optimal position of individual particles and the optimal position of the particle group are expressed as: ; ; When the particle individual optimal position Optimal position of the population When it is better, update the optimal position of the population; For any dimension The particle position and velocity, the velocity and position update formula are as follows: ; ; In equations (11) and (12), V represents the particle velocity, and are acceleration factors, represents a random number in the interval [0,1]; x represents the particle position, represents the optimal position of the particle; Represents the optimal position of the particle swarm.
[0033] like Figure 6 As shown, in another embodiment of the present invention, a real-time detection system for the operation status of electromechanical equipment is provided, and the detection system 600 includes: Data processing module 601, the user removes noise data from the monitored electromechanical equipment operation data through the wavelet denoising method to obtain minority class samples of the original data of the electromechanical equipment working status; The data expansion module 602 is used to perform non-marginal expansion on minority class samples by improving the SMOTE algorithm; and to perform marginal expansion on different class samples of minority class samples by using an adaptive synthetic data generation method based on a kernel function to obtain a balanced data set of electromechanical equipment; The model training module 603 is used to select a Gaussian kernel function as the kernel function of the SVM classifier, initialize the penalty factor c and the kernel function parameter g, and train the SVM model using a balanced data set. During the training process, an improved particle swarm optimization algorithm is used to tune the parameters of the SVM model to obtain a trained running status detection model; The state detection module 604 is used to obtain the real-time data of the operating state of the electromechanical equipment using the data collector, use the real-time data of the operating state as the input of the trained operating state detection model, and output the detection classification result corresponding to the real-time data.
[0034] In an embodiment of the present invention, the present invention further provides a computer device.
[0035] The computer device includes a memory, a processor, and computer-readable instructions stored in the memory and executable on the processor. When the processor executes the computer-readable instructions, the real-time detection method for the operating status of electromechanical equipment provided in the above embodiment is implemented.
[0036] Furthermore, in the exemplary embodiments provided by the present disclosure, a computer-readable storage medium is also provided.
[0037] Specifically, in an exemplary embodiment of the present disclosure, the storage medium stores computer-readable instructions, and when the computer-readable instructions are executed by one or more processors, the one or more processors execute the real-time detection method for the operating status of electromechanical equipment provided in the above-mentioned embodiment.
[0038] In various embodiments of the present invention, it should be understood that the size of the serial numbers of the processes does not necessarily mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.
[0039] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, i.e., they may be located in one place or distributed over multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.
[0040] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The integrated unit may be implemented in the form of hardware or in the form of software functional units.
[0041] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-accessible memory.
[0042] Based on this understanding, the technical solution of the present invention can essentially or partly contribute to the prior art or all or part of the technical solution, which can be embodied in the form of a software product. The computer software product is stored in a memory and includes several requests for enabling a computer device (which can be a personal computer, server or network device, etc., specifically a processor in a computer device) to execute part or all of the steps of the method for real-time detection of the operating status of electromechanical equipment described in various embodiments of the present invention.
[0043] The above is a detailed introduction to the real-time detection method for the operating status of electromechanical equipment disclosed in the embodiment of the present invention. Specific examples are used in this article to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea; at the same time, for general technical personnel in this field, according to the idea of the present invention, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.
Claims
1. A real-time detection method for the operating status of electromechanical equipment, characterized in that: The detection method comprises the following steps: The noise data is removed from the monitored electromechanical equipment operation data by using the wavelet denoising method to obtain minority class samples of the original data of the electromechanical equipment working status; The non-marginal expansion of minority class samples is performed by improving the SMOTE algorithm; the adaptive synthetic data generation method based on kernel function is used to marginalize and expand the different class samples of minority class samples to obtain the electromechanical equipment balanced data set; The Gaussian kernel function is selected as the kernel function of the SVM classifier, the penalty factor c and the kernel function parameter g are initialized, and the SVM model is trained using a balanced data set. During the training process, the improved particle swarm optimization algorithm is used to tune the parameters of the SVM model to obtain a trained running status detection model. The real-time data of the operating status of the electromechanical equipment is obtained by using a data collector, and the real-time data of the operating status is used as the input of the trained operating status detection model to output the detection classification results corresponding to the real-time data.
2. The real-time detection method for the operating status of electromechanical equipment according to claim 1 is characterized in that: The steps of removing noise data from the monitored electromechanical equipment operation data by wavelet denoising method include: Establish the fault model of electromechanical equipment operation data, the fault model is expressed as: ; In formula (1), Represented as electromechanical equipment operating data including noise data; Indicates the initial electromechanical equipment operation data; is the noise data; Perform wavelet transform on the fault model and get: ; The electromechanical equipment operation data is decomposed by wavelet, and the wavelet basis function and layered data are selected for splitting to obtain multiple wavelet levels and corresponding wavelet coefficients; the wavelet coefficients are threshold processed, and the corresponding threshold and threshold function are selected according to the decomposed wavelet coefficients, which are expressed as: ; In formula (2) and formula (3), Indicates the wavelet corresponding to different data; y indicates the maximum decomposition degree, and k indicates the data length; The value range of is [0,1]; represents the threshold; the threshold function is Medium continuous; According to the threshold function According to the continuity of the data and the feature requirements of the data at different scales, the optimal estimate is obtained and the new threshold representation is obtained: ; In formula (4), y represents the degree of decomposition; N is the signal size; Indicates the noise level, It is expressed as: ; The processed coefficients are synthesized into denoised data through wavelet reconstruction.
3. The real-time detection method for the operation status of electromechanical equipment according to claim 2 is characterized in that: The steps of non-marginalizing the minority class samples by improving the SMOTE algorithm include: The minority class samples are divided into noise points and non-noise points, and the extended samples are constructed by linear interpolation between the non-noise points and the middle points of the minority class samples. The extended samples are distributed near the center points of the original minority class samples, which can effectively avoid the marginalization of the extended data.
4. The real-time detection method for the operating status of electromechanical equipment according to claim 3 is characterized in that: The calculation process of the middle point of the minority class sample includes: taking the mean of each characteristic attribute of the minority class sample after normalization as the center point of the minority class sample, expressed as: ; In formula (5), represents the i-th sample under the r-th feature attribute after normalization preprocessing, and n represents the number of minority class samples; At the center of the minority class samples Data expansion is performed between non-noise points using the following formula: ; In formula (6), represents the newly expanded minority class samples, Represents the minority class samples in the non-noise point set; Represents a random number that satisfies a uniform distribution on the interval [0,1].
5. The real-time detection method for the operating status of electromechanical equipment according to claim 4 is characterized in that: The steps of using the kernel function-based adaptive synthetic data generation method to marginalize and expand different class samples of minority class samples include: The weight is calculated based on the proportion of majority class samples among the k nearest neighbors of minority class samples; Determine the number of synthetic samples that need to be generated for each minority class sample based on the weight and the preset ratio; The probability density of minority class samples is estimated using the kernel function, and the width of the kernel function is controlled by the bandwidth parameter; Using the Markov Chain Monte Carlo method to generate synthetic samples based on kernel density estimation, the initial samples are randomly selected from the minority class samples, and then new sample points are generated by adding random perturbations until the required number of synthetic samples is generated. The generated synthetic samples are merged with the original majority class samples to form the electromechanical device balancing dataset.
6. The real-time detection method for the operation status of electromechanical equipment according to claim 5 is characterized in that: The steps of using the improved particle swarm optimization algorithm to tune the parameters of the SVM classification model include: Set the number of particles and the maximum iteration of the particle swarm. At this time, set the penalty factor c and kernel function parameter g of the SVM model to the two-dimensional coordinates of the individual particle position; Randomly initialize the penalty factor c and kernel function parameter g; Use samples from the balanced data set to train the SVM model; Update the position and speed of individual particles, calculate the fitness value of each individual particle after the update, and determine the relationship between the optimal fitness value of the current iteration number and the fitness value of each individual particle after the update to obtain the position of the optimal individual particle in the particle group; When the maximum number of iterations is reached, the iteration is stopped and the optimal penalty factor c and kernel function parameter g are output; The optimal penalty factor c and kernel function parameter g are used to construct an SVM-based running status detection model.
7. The real-time detection method for the operation status of electromechanical equipment according to claim 6 is characterized in that: The steps to update the position and velocity of individual particles include: Suppose in D-dimensional space, the number of particles is n, and the position of the i-th particle is and speed Respectively expressed as: ; ; The optimal position of individual particles and the optimal position of the particle group are expressed as: ; ; When the particle individual optimal position Optimal position of the population When it is better, update the optimal position of the population; For any dimension The particle position and velocity, the velocity and position update formula are as follows: ; ; In equations (11) and (12), V represents the particle velocity, and are acceleration factors, represents a random number in the interval [0,1]; x represents the particle position, represents the optimal position of the particle; Represents the optimal position of the particle swarm.
8. A detection system for implementing the real-time detection method for the operation status of electromechanical equipment according to any one of claims 1 to 7, characterized in that: The detection system includes: In the data processing module, the user uses the wavelet denoising method to remove noise data from the monitored electromechanical equipment operation data, and obtains minority class samples of the original data of the electromechanical equipment working status; The data expansion module is used to perform non-marginal expansion of minority class samples by improving the SMOTE algorithm; the kernel function-based adaptive synthetic data generation method is used to perform marginal expansion of different class samples of minority class samples to obtain a balanced data set of electromechanical equipment; The model training module is used to select the Gaussian kernel function as the kernel function of the SVM classifier, initialize the penalty factor c and the kernel function parameter g, and use the balanced data set to train the SVM model. During the training process, the improved particle swarm optimization algorithm is used to tune the parameters of the SVM model to obtain a trained running status detection model; The status detection module is used to obtain the real-time data of the operating status of the electromechanical equipment using the data collector, use the real-time data of the operating status as the input of the trained operating status detection model, and output the detection classification results corresponding to the real-time data.
Citation Information
Patent Citations
Online fault diagnosing method for Fast RVM (relevance vector machine) sewage treatment
CN106681305A
Fluidized bed caking fault monitoring method based on weighted up-sampling method of kernel probability density estimation
CN113032726A
Nuclear power device fault diagnosis method in case of sample imbalance
CN117892136A
Traffic accident cause coupling effect analysis method considering sample imbalance
CN118277913A
