Method, device and equipment for monitoring running state of electromagnetic suspension system and medium

Abstract

The invention discloses an electromagnetic suspension system operation state monitoring method and device, equipment and a medium, and relates to the technical field of traffic and artificial intelligence crossing, and the method comprises the steps: constructing an attribute configuration table containing the corresponding relation between a train operation state and a state attribute based on the engineering knowledge and empirical data of a maglev train; determining a preset feature extractor based on the attention mechanism and a wavelet function, performing feature extraction on the test sample in combination with a sliding window function, and performing multi-scale analysis on feature fragments based on a wavelet transform technology; training the initial attribute learner based on a generalized zero sample learning method and an attribute configuration table, and judging a test sample based on a gating model constructed in the training process; and performing attribute prediction on the feature vector by using a target attribute learner, determining a monitoring result based on a judgment result, an attribute value and a nearest neighbor search rule, and taking early warning and maintenance measures. The accuracy of monitoring the operation state of the electromagnetic suspension system is improved.

Description

Technical Field

[0001] The present invention relates to the field of intersectional technologies of transportation and artificial intelligence, and in particular to a method, device, equipment and medium for monitoring the operating status of an electromagnetic suspension system. Background Art

[0002] Maglev trains are fast, quiet, safe and environmentally friendly, and are becoming a research hotspot in the field of modern transportation. Electromagnetic suspension systems need to build a complete state monitoring system to ensure the safety and stable operation of trains. However, current state monitoring research has rarely discussed the problem of extremely unbalanced samples. The probability of some states of maglev trains occurring is extremely low, and the number of samples available for training models is extremely low (or even nonexistent). The system may be disturbed by the external environment and appear in a completely new state; moreover, traditional monitoring methods cannot effectively monitor it. At the same time, the time series of the electromagnetic suspension system has a large amount of information and low information density, which makes it difficult to capture the features contained in the sequence.

[0003] It can be seen from the above that how to prevent the extreme imbalance of samples with low information density in the electromagnetic suspension system from leading to inaccurate monitoring of the operating status of the electromagnetic suspension system is a problem that needs to be solved urgently. Summary of the invention

[0004] In view of this, the purpose of the present invention is to provide a method, device, equipment and medium for monitoring the operating status of an electromagnetic suspension system, which can prevent the extremely unbalanced samples with low information density of the electromagnetic suspension system from causing inaccurate monitoring of the operating status of the electromagnetic suspension system. The specific scheme is as follows:

[0005] In a first aspect, the present application provides a method for monitoring the operating status of an electromagnetic suspension system, comprising:

[0006] Based on the engineering knowledge and experience data of the maglev train, an attribute configuration table including the correspondence between the train running state and the state attribute is constructed; the train running state includes normal operation, passing the joint, low-frequency shaking, high-frequency vibration, and rail hitting;

[0007] The test samples are obtained by using the running state data of the maglev train during operation, and then the test samples are subjected to feature extraction based on a preset feature extractor and a sliding window function to obtain feature segments, and the feature segments are subjected to multi-scale analysis based on a wavelet transform technology to obtain feature vectors; the running state data includes known running states and unknown running states; the preset feature extractor is a feature extractor determined based on an attention mechanism and a wavelet function;

[0008] Based on the generalized zero-shot learning method and the attribute configuration table, an initial attribute learner is trained to obtain a target attribute learner, and a gating model is constructed during the training process to determine whether the test sample is in a known operating state or an unknown operating state based on the gating model to obtain a judgment result;

[0009] The target attribute learner is used to perform attribute prediction on the feature vector to obtain the corresponding attribute value, and then the monitoring result is determined based on the judgment result and the attribute value in combination with the nearest neighbor search rule, so as to take corresponding early warning and maintenance measures based on the monitoring result.

[0010] Optionally, obtaining a test sample by using various operating state data of the maglev train during operation includes:

[0011] Using a suspension data acquisition system to collect various operating status data of the maglev train during operation, and determining each of the operating status data as a test sample;

[0012] Among them, the suspension data acquisition system includes a data collector, a suspension gap sensor, a current sensor, an acceleration sensor, a temperature sensor, a WIFI network center and an embedded host computer software center; the operating status data includes the suspension gap, suspension acceleration, lateral acceleration, suspension electromagnet working current and suspension electromagnet coil surface temperature of the maglev train during operation.

[0013] Optionally, the extracting features of the test sample based on a preset feature extractor and a sliding window function to obtain feature fragments includes:

[0014] The test sample is segmented based on a preset sliding window length and a preset sliding step size to obtain each sliding window, and the mean, the first energy value and the difference between the first peak values ​​of each sliding window are calculated by using an attention mechanism to obtain a corresponding calculation result;

[0015] Determine a first attention weighting function using the calculation result, so as to determine a target window segment with a maximum weight value based on the first attention weighting function;

[0016] A second attention weighting function is determined based on the difference between the second energy value and the second peak value of the first-order difference sequence corresponding to the sliding window, and the key features in the target window segment are determined using the second attention weighting function to obtain a feature segment.

[0017] Optionally, the performing multi-scale analysis on the characteristic segment based on wavelet transform technology to obtain a characteristic vector includes:

[0018] The preset wavelet function is scaled and translated using preset scale parameters to obtain a target wavelet function, and inner product calculation is performed based on the target wavelet function and the characteristic fragment to obtain wavelet coefficients at different scales, and the wavelet coefficients are extracted to obtain a feature vector.

[0019] Optionally, the initial attribute learner is trained based on the generalized zero-sample learning method and the attribute configuration table to obtain a target attribute learner, and a gating model is constructed during the training process to determine whether the test sample is a known operating state or an unknown operating state based on the gating model to obtain a judgment result, including:

[0020] Training the initial attribute learner based on the generalized zero-shot learning method, the machine learning algorithm and the attribute configuration table to obtain the target attribute learner;

[0021] In the process of training the initial attribute learner, a gating model is constructed, and a feature center value of each known operating state in the gating model in a feature space is obtained;

[0022] Performing distribution direction similarity calculation and distribution distance similarity calculation on the test sample and the feature center value respectively to obtain a first similarity calculation result and a second similarity calculation result;

[0023] The first similarity calculation result and the second similarity calculation result are integrated to obtain a comprehensive similarity score, and the one with the highest comprehensive similarity score is determined as a target similarity score, and it is determined whether the target similarity score is greater than a preset similarity threshold;

[0024] If the target similarity score is less than a preset similarity threshold, it indicates that the test sample is in an unknown running state;

[0025] If the target similarity score is not less than a preset similarity threshold, it indicates that the test sample is in a known running state.

[0026] Optionally, respectively performing distribution direction similarity calculation and distribution distance similarity calculation on the test sample and the feature center value to obtain a first similarity calculation result and a second similarity calculation result includes:

[0027] Determine a first vector and a second vector based on the feature center value and the test sample;

[0028] By calculating the cosine value of the first vector and the second vector, a first similarity calculation result representing the similarity between the feature center value and the distribution direction of the test sample is obtained;

[0029] A second similarity calculation result representing the similarity between the feature center value and the test sample distribution distance is obtained by calculating the ratio of the norms of the first vector and the second vector.

[0030] Optionally, the using the target attribute learner to perform attribute prediction on the feature vector to obtain a corresponding attribute value, and then determining a monitoring result based on the judgment result and the attribute value in combination with a nearest neighbor search rule, so as to take corresponding early warning and maintenance measures based on the monitoring result, including:

[0031] Utilize the target attribute learner to perform attribute prediction on the feature vector to obtain a corresponding attribute value; if the judgment result indicates that the test sample is in a known operating state, calculate a first distance between the attribute value and the attribute vector corresponding to each known operating state, and select the known operating state with the smallest first distance as the target known operating state of the test sample based on a nearest neighbor search rule, and take corresponding early warning and maintenance measures based on the target known operating state;

[0032] If the judgment result characterizes that the test sample is in an unknown operating state, the second distance between the attribute value and the attribute vector corresponding to each unknown operating state is calculated, and the unknown operating state with the smallest second distance is selected as the target unknown operating state of the test sample based on the nearest neighbor search rule, and corresponding warning and maintenance measures are taken based on the target unknown operating state.

[0033] In a second aspect, the present application provides an electromagnetic suspension system operation status monitoring device, comprising:

[0034] An attribute configuration table construction module is used to construct an attribute configuration table containing the correspondence between the train running state and the state attributes based on the engineering knowledge and experience data of the maglev train; the train running state includes normal operation, seam crossing, low-frequency shaking, high-frequency vibration, and rail smashing;

[0035] A feature extraction module is used to obtain a test sample using the running state data of the maglev train during operation, and then perform feature extraction on the test sample based on a preset feature extractor and a sliding window function to obtain a feature segment, and perform multi-scale analysis on the feature segment based on a wavelet transform technology to obtain a feature vector; the running state data includes a known running state and an unknown running state; the preset feature extractor is a feature extractor determined based on an attention mechanism and a wavelet function;

[0036] An attribute learner training module is used to train an initial attribute learner based on a generalized zero-shot learning method and the attribute configuration table to obtain a target attribute learner, and to construct a gating model during the training process to determine whether the test sample is in a known operating state or an unknown operating state based on the gating model to obtain a judgment result;

[0037] The attribute value prediction unit is used to use the target attribute learner to perform attribute prediction on the feature vector to obtain the corresponding attribute value, and then determine the monitoring result based on the judgment result and the attribute value and in combination with the nearest neighbor search rule, so as to take corresponding early warning and maintenance measures based on the monitoring result.

[0038] In a third aspect, the present application provides an electronic device, including:

[0039] Memory, used to store computer programs;

[0040] The processor is used to execute the computer program to implement the aforementioned electromagnetic suspension system operation status monitoring method.

[0041] In a fourth aspect, the present application provides a computer-readable storage medium for storing a computer program, wherein the computer program, when executed by a processor, implements the aforementioned electromagnetic suspension system operating status monitoring method.

[0042] The present application constructs an attribute configuration table containing the correspondence between the train running state and the state attributes based on the engineering knowledge and experience data of the maglev train; the train running state includes normal operation, seam crossing, low-frequency shaking, high-frequency vibration, and rail smashing; the test samples are obtained by using the running state data of the maglev train during operation, and then the test samples are feature extracted based on a preset feature extractor and a sliding window function to obtain feature fragments, and the feature fragments are multi-scale analyzed based on the wavelet transform technology to obtain feature vectors; the running state data includes known running states and unknown running states; the preset feature extraction The detector is a feature extractor determined based on the attention mechanism and the wavelet function; the initial attribute learner is trained based on the generalized zero-sample learning method and the attribute configuration table to obtain the target attribute learner, and a gating model is constructed during the training process to judge whether the test sample is a known operating state or an unknown operating state based on the gating model to obtain a judgment result; the target attribute learner is used to predict the attribute of the feature vector to obtain the corresponding attribute value, and then the monitoring result is determined based on the judgment result and the attribute value, and combined with the nearest neighbor search rule, so as to take corresponding early warning and maintenance measures based on the monitoring result.

[0043] As can be seen from the above, this application constructs an attribute configuration table by integrating the engineering knowledge and experience data of the maglev train, and then uses the preset feature extractor determined by the attention mechanism and the wavelet function and the wavelet transform technology to extract and multi-scale analyze the various operating state data during the operation of the maglev train to obtain a feature vector, and then trains the initial attribute learner based on the generalized zero-sample learning method and the attribute configuration table to obtain a target attribute learner, and constructs a gating model during the training process to use the gating model to determine whether the test sample is a known operating state or an unknown operating state; then, based on the attribute learner, the attribute prediction of the feature vector is performed to obtain the attribute value. In this way, based on the judgment result and the attribute value, and combined with the nearest neighbor search rule, the final monitoring result is determined. Through the monitoring result, the potential faults or abnormal conditions of the maglev train can be discovered in time, and corresponding early warning and maintenance measures can be taken, which greatly reduces the failure risk and operating cost of the maglev train and improves the safety and reliability of operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0045] Figure 1 A flow chart of a method for monitoring the operating status of an electromagnetic suspension system disclosed in this application;

[0046] Figure 2 A schematic diagram of the structure of an EMS type maglev train provided in this application;

[0047] Figure 3 An attribute configuration representation provided for this application;

[0048] Figure 4 A schematic diagram of the meaning of attributes corresponding to different working conditions provided for this application;

[0049] Figure 5 A structural diagram of a suspension data acquisition system provided for this application;

[0050] Figure 6 A flowchart of capturing feature fragments using an attention mechanism provided for this application;

[0051] Figure 7 A schematic diagram of a generalized zero-shot learning process for attributes provided in this application;

[0052] Figure 8A schematic diagram of a gating model calculation process provided in this application;

[0053] Fig. 9 A specific electromagnetic suspension system operation status monitoring method disclosed in this application;

[0054] Fig.10 A schematic diagram of an evaluation index provided for this application;

[0055] Fig.11 A schematic diagram of the accuracy of online evaluation of operating status monitoring based on traditional machine learning provided in this application;

[0056] Fig.12 A schematic diagram of online evaluation accuracy of operating status monitoring based on generalized zero samples provided in this application;

[0057] Fig.13 A confusion matrix diagram of an online evaluation result of operation status monitoring based on traditional machine learning provided in this application;

[0058] Fig.14 A schematic diagram of a confusion matrix of online evaluation results of operating status monitoring based on generalized zero samples provided in this application;

[0059] Fig.15 A schematic diagram of the scores of the online evaluation results of the operation status monitoring based on generalized zero samples provided by this application;

[0060] Fig.16 This is a schematic diagram of the structure of an electromagnetic suspension system operating status monitoring device disclosed in this application;

[0061] Fig.17 This is a structural diagram of an electronic device disclosed in this application. DETAILED DESCRIPTION

[0062] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0063] At present, there is little discussion on the problem of extremely unbalanced samples in the state monitoring research of electromagnetic levitation systems. The probability of occurrence of some states of maglev trains is extremely low, and the number of samples that can be used to train models is extremely low. The system may be disturbed by the external environment and appear in a completely new state; and traditional monitoring methods cannot effectively monitor it. The time series information of the electromagnetic levitation system is large and the information density is low, which makes it difficult to capture the characteristics contained in the sequence. To this end, the present application provides a method for monitoring the operating status of an electromagnetic levitation system, which is based on the judgment results and the attribute values, and combines the nearest neighbor search rule to determine the final monitoring results. Through the monitoring results, potential faults or abnormal conditions of maglev trains can be discovered in time, and corresponding early warning and maintenance measures can be taken, which greatly reduces the failure risk and operating costs of maglev trains and improves the safety and reliability of operation.

[0064] See also Figure 1 As shown, an embodiment of the present invention discloses a method for monitoring the operating status of an electromagnetic suspension system, comprising:

[0065] Step S11, constructing an attribute configuration table including the correspondence between train running status and status attributes based on engineering knowledge and experience data of maglev trains; the train running status includes normal operation, seam crossing, low-frequency shaking, high-frequency vibration, and rail hitting.

[0066] In this embodiment, Figure 2 This is a structural diagram of an EMS type maglev train provided in this embodiment. A maglev train consists of three carriages and multiple suspension frame modules, and is suspended by the interaction between electromagnets and guide rails. When the train is running, it will encounter abnormal conditions such as seams, low-frequency shaking, high-frequency vibration, and rail smashing. These conditions will cause different changes in parameters such as gaps, acceleration, and current, thereby affecting the stability and safety of the train. Figure 3 This is a property configuration representation provided in this embodiment. Based on the engineering knowledge and experience data of maglev trains, an property configuration table containing the correspondence between train running status and status attributes is constructed, wherein grade#1, grade#2, grade#3, grade#4, and grade#5 represent five operating conditions: normal train operation, seam crossing, low-frequency shaking, high-frequency vibration, and rail smashing, respectively. Figure 4 A schematic diagram of the meaning of attributes corresponding to different working conditions provided for this embodiment summarizes seven suspension system characteristics, including gap abnormality, jitter, and strong correlation between gap, current, and acceleration, which are used to evaluate the system status.

[0067] Step S12, using the operating status data of the maglev train during operation to obtain a test sample, then extracting features from the test sample based on a preset feature extractor and a sliding window function to obtain a feature segment, and performing multi-scale analysis on the feature segment based on wavelet transform technology to obtain a feature vector; the operating status data includes a known operating status and an unknown operating status; the preset feature extractor is a feature extractor determined based on an attention mechanism and a wavelet function.

[0068] In this embodiment, Figure 5 A structural diagram of a suspension data acquisition system provided in this embodiment is shown in FIG. A plurality of sensors are arranged near the bogie and guide rails of a maglev train to monitor the running status of the train, wherein: It represents the air gap value of the maglev train, in mm. is the acceleration of the maglev train, in units of , is the current in the electromagnet, in A, and u is the voltage on the electromagnet, in V. The suspension data acquisition system composed of a data collector, a suspension gap sensor, a current sensor, an acceleration sensor, a temperature sensor, a WIFI network center, and an embedded host computer software center is used to collect the suspension gap, vertical acceleration, lateral acceleration, electromagnet working current, and electromagnet coil surface temperature during the debugging process of the maglev train to obtain the running state data, and the running state data is used as a test sample. Specifically, the test sample is obtained by using the various running state data of the maglev train during operation, including: using the suspension data acquisition system to collect the various running state data of the maglev train during operation, and determining each of the running state data as a test sample; wherein the suspension data acquisition system includes a data collector, a suspension gap sensor, a current sensor, an acceleration sensor, a temperature sensor, a WIFI network center, and an embedded host computer software center; the running state data includes the suspension gap, suspension acceleration, lateral acceleration, suspension electromagnet working current, and suspension electromagnet coil surface temperature of the maglev train during operation.

[0069] Furthermore, after obtaining the test sample, the test sample is segmented based on a preset sliding window length and a preset sliding step size to obtain each sliding window, and the attention mechanism is used to calculate the mean, the first energy value and the difference between the first peak values ​​of each sliding window, so as to determine the first attention weighting function based on the difference between the mean, the first energy value and the first peak value of the sliding window, and determine the target window segment with the largest weight value based on the first attention weighting function, and then calculate the second energy value and the second peak value difference of the first-order difference sequence corresponding to the sliding window, so as to determine the second attention weighting function based on the difference between the second energy value and the second peak value of the first-order difference sequence corresponding to the sliding window, and use the second attention weighting function to determine the key features in the target window segment to obtain a feature segment.

[0070] Specifically, the feature extraction of the test sample based on the preset feature extractor and the sliding window function to obtain the feature fragment includes: segmenting the test sample based on the preset sliding window length and the preset sliding step size to obtain each sliding window, and using the attention mechanism to calculate the mean, the first energy value and the difference between the first peak values ​​of each sliding window to obtain the corresponding calculation results; using the calculation results to determine the first attention weighting function to determine the target window fragment with the largest weight value based on the first attention weighting function; based on the difference between the second energy value and the second peak value of the first-order difference sequence corresponding to the sliding window, determine the second attention weighting function, and use the second attention weighting function to determine the key features in the target window fragment to obtain the feature fragment. It is worth mentioning that the preset sliding window length and the preset sliding step size can be adjusted according to actual conditions and are not specifically limited here.

[0071] Figure 6 The present embodiment provides a flowchart of capturing feature segments with an attention mechanism, which combines mathematical information such as mean and standard deviation with the attention mechanism, and constructs different types of attention weighting functions and sliding windows to capture regions with significantly different attention weight values, thereby completing the task of capturing feature segments with different requirements. In a specific implementation, for the sliding window , the length is N, and the corresponding mean calculation formula is: ; The corresponding calculation formula for the first energy value is:

[0072] ;

[0073] in, The calculation formula of the difference between the first peak values ​​is:

[0074] ;

[0075] in, is the maximum value of the sliding window; is the minimum value of the sliding window. The calculation formula of the first attention weighting function obtained based on the above three formulas is:

[0076] ;

[0077] in, and The parameters used to dynamically adjust the power in combination with data distribution characteristics and task requirements can be adjusted according to actual needs and are not specifically limited here. The first attention weighting function is used to calculate the first attention weight value of each sliding window, and the sliding window with the largest first attention weight value is determined as the target window fragment. Then, the corresponding first-order difference sequence is determined based on the sliding window as: ; The calculation formula for calculating the second energy value of the first-order difference sequence is:

[0078] ;

[0079] The calculation formula for calculating the difference of the second peak of the first-order difference sequence is: ;in, is the maximum value in the first-order difference sequence; is the minimum value in the first-order difference sequence. The second attention weighting function obtained by combining the second energy value and the difference between the second peak values ​​of the first-order difference sequence is: ;in, and The parameters used to dynamically adjust the power in combination with data distribution characteristics and task requirements can be adjusted according to actual needs and are not specifically limited here.

[0080] In this embodiment, after obtaining the characteristic segment, the preset wavelet function is scaled and translated using the preset scale parameter to obtain the target wavelet function, and the inner product of the continuous wavelet function and the continuous signal in the characteristic segment is determined based on the continuous wavelet function in the target wavelet function to obtain the first wavelet coefficient; the inner product of the discrete wavelet function and the discrete signal in the characteristic segment is determined using the discrete wavelet function in the target wavelet function to obtain the second wavelet coefficient, and the feature vector is obtained by extracting the first wavelet coefficient and the second wavelet coefficient. Among them, common wavelet functions include: Haar (i.e., Haar wavelet function), Daubechies (i.e., Daubechies wavelet function), and Biorthogonal (i.e., biorthogonal wavelet function). Specifically, the multi-scale analysis of the feature segment based on wavelet transform technology to obtain a feature vector includes: scaling and translating a preset wavelet function using preset scale parameters to obtain a target wavelet function, and performing inner product calculation based on the target wavelet function and the feature segment to obtain wavelet coefficients at different scales, and extracting the wavelet coefficients to obtain a feature vector.

[0081] It can be understood that the preset scale parameter is determined according to the actual situation, and the preset scale parameter is used to perform scaling and translation transformation on the preset wavelet function to obtain the target wavelet function. The formula of the target wavelet function is:

[0082] ;

[0083] in, is the scaling parameter, is the translation parameter, is the translation factor, is the target wavelet function, is the preset scale parameter.

[0084] The formula of the continuous wavelet function is: ;

[0085] in, is the complex conjugate of the target wavelet function; for the discrete signal in the characteristic segment, the common discrete parameter selection is: ;in, and are integer indices for scaling and translation respectively.

[0086] The formula of the discrete wavelet function is: ; Calculate the inner product of the continuous signal in the feature segment and different wavelet functions to obtain the corresponding wavelet coefficients; extract each of the wavelet coefficients to obtain a feature vector. The calculation formula of the feature vector is:

[0087] ;

[0088] in, is the energy of the wavelet coefficients, is the standard deviation of wavelet coefficients, is the mean of the wavelet coefficients. The corresponding energy feature extraction formula is:

[0089] ;

[0090] in, Represents the wavelet coefficients within this frequency band; It is worth mentioning that, according to actual needs, some or all of the wavelet coefficients can be selected as extraction objects, and the wavelet coefficients can also be processed by dimension reduction, normalization, etc., which are not specifically limited here.

[0091] Step S13: train the initial attribute learner based on the generalized zero-sample learning method and the attribute configuration table to obtain the target attribute learner, and construct a gating model during the training process to judge whether the test sample is a known operating state or an unknown operating state based on the gating model to obtain a judgment result.

[0092] In this embodiment, after obtaining the feature vector, the initial attribute learner is trained based on the generalized zero-sample learning method, the machine learning algorithm and the attribute configuration table to obtain the target attribute learner, and in the process of training the initial attribute learner, a gating model for operating state diversion is constructed, and the feature center value of each known operating state in the gating model in the feature space is obtained, so as to calculate the distribution direction similarity and distribution distance similarity of the test sample and the feature center value based on the feature center value, respectively, and the first similarity calculation result and the second similarity calculation result are synthesized to obtain a comprehensive similarity score, and it is judged whether the target similarity score with the highest comprehensive similarity score is greater than a preset similarity threshold, so as to determine whether the test sample is an unknown operating state or a known operating state based on the judgment result.

[0093] Specifically, the method trains the initial attribute learner based on the generalized zero-shot learning method and the attribute configuration table to obtain a target attribute learner, and constructs a gating model during the training process to determine whether the test sample is a known operating state or an unknown operating state based on the gating model to obtain a judgment result, including: training the initial attribute learner based on the generalized zero-shot learning method, the machine learning algorithm and the attribute configuration table to obtain a target attribute learner; constructing a gating model in the process of training the initial attribute learner, and obtaining the feature center value of each known operating state in the gating model in the feature space; respectively The test sample and the feature center value are subjected to distribution direction similarity calculation and distribution distance similarity calculation to obtain a first similarity calculation result and a second similarity calculation result; the first similarity calculation result and the second similarity calculation result are integrated to obtain a comprehensive similarity score, and the one with the highest comprehensive similarity score is determined as a target similarity score, and it is determined whether the target similarity score is greater than a preset similarity threshold; if the target similarity score is less than the preset similarity threshold, the test sample is characterized as an unknown operating state; if the target similarity score is not less than the preset similarity threshold, the test sample is characterized as a known operating state.

[0094] It can be understood that the initial attribute learner is trained based on the generalized zero-shot learning method, the machine learning algorithm and the attribute configuration table, the machine learning algorithm includes probabilistic naive Bayes (NB, Naive Bayes), polynomial kernel-support vector machine (SVC or SVM, Support Vector Machine with Polynomial Kernel), and decision tree (Decision Tree); the training set in the generalized zero-shot learning method only includes training samples with known operating states, but the test set includes both known operating states and unknown operating states. In a specific implementation, the training set can be recorded as ; The test set can be recorded as ; Further, the data set of the training set is recorded as ; , are the training set samples and the corresponding state attributes respectively. The data set of the test set is recorded as ; , They are the test set samples and the corresponding state attributes respectively.

[0095] Figure 7 A schematic diagram of an attribute generalized zero-shot learning process provided in this embodiment, wherein: is the initial attribute learner, Based on the generalized zero-sample learning method, the known operating state and the corresponding state attributes are transferred to the unknown operating state, so as to learn the known operating state through the known operating state and the corresponding state attributes. To full operation status The mapping model between ; The mapping formula of the state attributes corresponding to all running states is: ;in, is the state attribute description matrix of all running states. The attribute values ​​predicted by the attribute learner after training The difference from the actual state attribute is , when the gap When the preset gap condition is met, the trained attribute learner is determined as the target attribute learner.

[0096] In this embodiment, during the training of the initial attribute learner, a gating model for operating state diversion is constructed, and the feature center value of each known operating state in the gating model in the feature space is obtained to determine a first vector and a second vector based on the feature center value and the corresponding test sample, and a first similarity calculation result representing the similarity between the feature center value and the distribution direction of the test sample is obtained by calculating the cosine value between the first vector and the second vector; the ratio of the norms of the first vector and the second vector is calculated to obtain a second similarity calculation result representing the similarity between the feature center value and the distribution distance of the test sample.

[0097] Specifically, the distribution direction similarity calculation and the distribution distance similarity calculation are performed on the test sample and the feature center value respectively to obtain a first similarity calculation result and a second similarity calculation result, including: determining a first vector and a second vector based on the feature center value and the test sample; calculating the cosine value of the first vector and the second vector to obtain a first similarity calculation result representing the similarity in the distribution direction of the feature center value and the test sample; and calculating the ratio of the norms of the first vector and the second vector to obtain a second similarity calculation result representing the similarity in the distribution distance of the feature center value and the test sample.

[0098] Figure 8 A schematic diagram of a gating model calculation process provided in this embodiment, and is the coordinate axis in the feature space, and the mean of the sample features belonging to the i-th known operating state in the training set is ; The characteristics of the test sample are recorded as . Taking the origin of the coordinate system as the starting point, and The corresponding coordinates are used as the end point to make two vectors to obtain the first vector and the second vector ; Then use the cosine formula Determine a first similarity calculation result that represents the similarity between the feature center value and the distribution direction of the test sample. By introducing the norm ratio of the first vector and the second vector , the formula corresponding to the norm ratio of the first vector and the second vector is as follows:

[0099] ;

[0100] in, is the first vector; is the second vector; is the norm of the first vector; is the norm of the second vector. The above formula is used to obtain a second similarity calculation result that characterizes the similarity between the feature center value and the test sample distribution distance; then the first similarity calculation result and the second similarity calculation result are combined to obtain a comprehensive similarity score. , the formula corresponding to the comprehensive similarity score is as follows:

[0101] ;

[0102] After obtaining the comprehensive similarity score, the highest comprehensive similarity score is taken as the target similarity score. The corresponding calculation formula is as follows:

[0103] ;

[0104] After obtaining the target similarity score, determine whether the target similarity score is greater than a preset similarity threshold , the corresponding judgment formula is:

[0105] ;

[0106] If the target similarity score is less than the preset similarity threshold, it indicates that the test sample is in an unknown running state; if the target similarity score is not less than the preset similarity threshold, it indicates that the test sample is in a known running state.

[0107] Step S14: Use the target attribute learner to predict the attribute of the feature vector to obtain the corresponding attribute value, and then determine the monitoring result based on the judgment result and the attribute value, and combine the nearest neighbor search rule to take corresponding early warning and maintenance measures based on the monitoring result.

[0108] In this embodiment, after obtaining the target attribute learner, the target attribute learner is used to perform attribute prediction on the feature vector to obtain an attribute value. If the target similarity score is not less than the preset similarity threshold, the test sample is characterized as a known operating state, and the first distance between the attribute value and the attribute vector corresponding to each known operating state is calculated, and the known operating state with the smallest first distance is selected as the target known operating state of the test sample based on the nearest neighbor search rule, and corresponding early warning and maintenance measures are taken based on the target known operating state; if the target similarity score is less than the preset similarity threshold, the test sample is characterized as an unknown operating state, and the second distance between the attribute value and the attribute vector corresponding to each unknown operating state is calculated, and the unknown operating state with the smallest second distance is selected as the target unknown operating state of the test sample based on the nearest neighbor search rule, so as to take corresponding early warning and maintenance measures based on the target unknown operating state. The distances referred to by the first distance and the second distance can be Mahalanobis distances. It should be noted that if the target known operating state is normal operation, no early warning and maintenance measures are required.

[0109] Specifically, the target attribute learner is used to predict the attribute of the feature vector to obtain the corresponding attribute value, and then the monitoring result is determined based on the judgment result and the attribute value, and combined with the nearest neighbor search rule, so as to take corresponding early warning and maintenance measures based on the monitoring result, including: using the target attribute learner to predict the attribute of the feature vector to obtain the corresponding attribute value; if the judgment result indicates that the test sample is a known operating state, the first distance between the attribute value and the attribute vector corresponding to each known operating state is calculated, and the known operating state with the smallest first distance is selected as the target known operating state of the test sample based on the nearest neighbor search rule, and corresponding early warning and maintenance measures are taken based on the target known operating state; if the judgment result indicates that the test sample is an unknown operating state, the second distance between the attribute value and the attribute vector corresponding to each unknown operating state is calculated, and the unknown operating state with the smallest second distance is selected as the target unknown operating state of the test sample based on the nearest neighbor search rule, and corresponding early warning and maintenance measures are taken based on the target unknown operating state.

[0110] As can be seen from the above, this application constructs an attribute configuration table by integrating the engineering knowledge and experience data of the maglev train, and then uses the preset feature extractor determined by the attention mechanism and the wavelet function and the wavelet transform technology to extract and multi-scale analyze the various operating state data during the operation of the maglev train to obtain a feature vector, and then trains the initial attribute learner based on the generalized zero-sample learning method and the attribute configuration table to obtain a target attribute learner, and constructs a gating model during the training process to use the gating model to determine whether the test sample is a known operating state or an unknown operating state; then, based on the attribute learner, the attribute prediction of the feature vector is performed to obtain the attribute value. In this way, based on the judgment result and the attribute value, and combined with the nearest neighbor search rule, the final monitoring result is determined. Through the monitoring result, the potential faults or abnormal conditions of the maglev train can be discovered in time, and corresponding early warning and maintenance measures can be taken, which greatly reduces the failure risk and operating cost of the maglev train and improves the safety and reliability of operation.

[0111] It can be seen from the above embodiments that the present application implements operating status monitoring of the electromagnetic suspension system based on feature extraction and training of the initial attribute learner. Therefore, the process of implementing operating status monitoring of the electromagnetic suspension system based on feature extraction and training of the initial attribute learner is described.

[0112] See also Fig. 9 As shown, the embodiment of the present invention discloses a specific method for monitoring the operating status of an electromagnetic suspension system, including:

[0113] This embodiment first constructs an attribute configuration table containing the correspondence between the train running state and the state attributes based on the engineering knowledge and experience data of the maglev train, and can selectively perform data preprocessing and normalization on the train running state, fit the training set data containing the known running state with the corresponding attribute label, and determine the preset feature extractor based on the obtained fitted data, the attention mechanism and the wavelet function, and then use the preset feature extractor to reduce the dimension of the fitted data to obtain the training set data features, use the training set data features to train the initial attribute learner to obtain the target attribute learner, and construct a gating model during the training process to calculate the feature center value of each known running state in the feature space.

[0114] Then, the operating state data of the maglev train during operation are used to obtain test samples containing known operating states and unknown operating states, and the test samples are subjected to feature extraction based on the preset feature extractor to obtain feature segments, and the feature segments are subjected to multi-scale analysis through wavelet transform technology to obtain feature vectors; then, the feature vectors are predicted using the target attribute learner to obtain corresponding attribute values. The first similarity calculation result is obtained by calculating the distribution direction similarity between the test sample and the feature center value; the second similarity calculation result is obtained by calculating the distribution distance similarity between the test sample and the feature center value, and the first similarity calculation result and the second similarity calculation result are synthesized to obtain a comprehensive similarity score, and the highest comprehensive similarity score is determined as the target similarity score, and it is determined whether the target similarity score is greater than the preset similarity threshold.

[0115] Furthermore, if the target similarity score is not less than a preset similarity threshold, the test sample is characterized as a known operating state, and the first distance between the attribute value corresponding to the test sample and the attribute vector corresponding to each known operating state is calculated, and based on the nearest neighbor search rule, the known operating state with the smallest first distance is selected as the target known operating state of the test sample, the target known operating state is the state to which the test sample is most likely to belong among the known operating states, and corresponding early warning and maintenance measures are taken based on the target known operating state; if the target similarity score is less than the preset similarity threshold, the test sample is characterized as an unknown operating state, and the second distance between the attribute value corresponding to the test sample and the attribute vector corresponding to each unknown operating state is calculated, and based on the nearest neighbor search rule, the unknown operating state with the smallest second distance is selected as the target unknown operating state of the test sample, and corresponding early warning and maintenance measures are taken based on the target unknown operating state.

[0116] There are six main evaluation indicators in this embodiment: Fig.10 This is a schematic diagram of an evaluation index provided in this embodiment, where each symbol represents a different meaning and is used to judge the final monitoring result. Figures 11 to 15 It can be seen that the monitoring effect of traditional machine learning operation status on unknown operation status is very poor; but after the introduction of generalized zero-shot learning, the average F1-score of known and unknown operation status is significantly improved, and the precision and recall rate are also high. It can effectively distinguish and accurately monitor known and unknown operation status samples, and the effect is significantly better than traditional methods.

[0117] As can be seen from the above, this embodiment uses the attribute value predicted by the target attribute learner and combines it with the judgment result to determine the target operating state corresponding to the test sample, and takes corresponding response measures based on the target operating state, breaking the limitation of traditional methods that it is difficult to monitor unknown operating states, and realizing accurate monitoring of samples of all operating states, which provides strong support for improving operational efficiency, ensuring passenger safety and reducing maintenance costs.

[0118] Accordingly, see Fig.16 As shown, the present application also provides an electromagnetic suspension system operation status monitoring device, comprising:

[0119] The attribute configuration table construction module 11 is used to construct an attribute configuration table containing the correspondence between the train running state and the state attribute based on the engineering knowledge and experience data of the maglev train; the train running state includes normal operation, seam crossing, low-frequency shaking, high-frequency vibration, and rail smashing;

[0120] The feature extraction module 12 is used to obtain a test sample using the running state data of the maglev train during operation, and then perform feature extraction on the test sample based on a preset feature extractor and a sliding window function to obtain a feature segment, and perform multi-scale analysis on the feature segment based on a wavelet transform technology to obtain a feature vector; the running state data includes a known running state and an unknown running state; the preset feature extractor is a feature extractor determined based on an attention mechanism and a wavelet function;

[0121] The attribute learner training module 13 is used to train the initial attribute learner based on the generalized zero-shot learning method and the attribute configuration table to obtain a target attribute learner, and to construct a gating model during the training process to determine whether the test sample is a known operating state or an unknown operating state based on the gating model to obtain a judgment result;

[0122] The attribute value prediction module 14 is used to use the target attribute learner to perform attribute prediction on the feature vector to obtain the corresponding attribute value, and then determine the monitoring result based on the judgment result and the attribute value and in combination with the nearest neighbor search rule, so as to take corresponding early warning and maintenance measures based on the monitoring result.

[0123] As can be seen from the above, this application constructs an attribute configuration table by integrating the engineering knowledge and experience data of the maglev train, and then uses the preset feature extractor determined by the attention mechanism and the wavelet function and the wavelet transform technology to extract and multi-scale analyze the various operating state data during the operation of the maglev train to obtain a feature vector, and then trains the initial attribute learner based on the generalized zero-sample learning method and the attribute configuration table to obtain a target attribute learner, and constructs a gating model during the training process to use the gating model to determine whether the test sample is a known operating state or an unknown operating state; then, based on the attribute learner, the attribute prediction of the feature vector is performed to obtain the attribute value. In this way, based on the judgment result and the attribute value, and combined with the nearest neighbor search rule, the final monitoring result is determined. Through the monitoring result, the potential faults or abnormal conditions of the maglev train can be discovered in time, and corresponding early warning and maintenance measures can be taken, which greatly reduces the failure risk and operating cost of the maglev train and improves the safety and reliability of operation.

[0124] In some specific implementations, the feature extraction module 12 may specifically include:

[0125] The data acquisition unit is used to collect various operation status data of the maglev train during operation by using a suspension data acquisition system, and determine each of the operation status data as a test sample.

[0126] In some specific implementations, the feature extraction module 12 may specifically include:

[0127] A test sample segmentation unit, used for segmenting the test sample based on a preset sliding window length and a preset sliding step length to obtain each sliding window, and using an attention mechanism to calculate the mean, the first energy value and the difference between the first peak values ​​of each sliding window to obtain a corresponding calculation result;

[0128] a first weighting function determining unit, configured to determine a first attention weighting function using the calculation result, so as to determine a target window segment having a maximum weight value based on the first attention weighting function;

[0129] The second weighting function determination unit is used to determine a second attention weighting function based on the difference between the second energy value and the second peak of the first-order difference sequence corresponding to the sliding window, and use the second attention weighting function to determine the key features in the target window segment to obtain a feature segment.

[0130] In some specific implementations, the feature extraction module 12 may specifically include:

[0131] The feature segment decomposition unit is used to use preset scale parameters to scale and translate the preset wavelet function to obtain a target wavelet function, and to perform inner product calculation based on the target wavelet function and the feature segment to obtain wavelet coefficients at different scales, and to extract the wavelet coefficients to obtain a feature vector.

[0132] In some specific implementations, the attribute learner training module 13 may specifically include:

[0133] A learner training unit, used for training an initial attribute learner based on a generalized zero-shot learning method, a machine learning algorithm and the attribute configuration table to obtain a target attribute learner;

[0134] A gating model building unit, used to build a gating model in the process of training the initial attribute learner, and obtain the feature center value of each known operating state in the gating model in the feature space;

[0135] A similarity calculation unit, used to perform distribution direction similarity calculation and distribution distance similarity calculation on the test sample and the feature center value respectively, so as to obtain a first similarity calculation result and a second similarity calculation result;

[0136] a target similarity determination unit, configured to synthesize the first similarity calculation result and the second similarity calculation result to obtain a comprehensive similarity score, determine the highest comprehensive similarity score as the target similarity score, and determine whether the target similarity score is greater than a preset similarity threshold;

[0137] A first target similarity judgment unit, configured to indicate that the test sample is in an unknown running state if the target similarity score is less than a preset similarity threshold;

[0138] The first target similarity judgment unit is used to indicate that the test sample is in a known running state if the target similarity score is not less than a preset similarity threshold.

[0139] In some specific implementations, the attribute learner training module 13 may specifically include:

[0140] A vector determination unit, configured to determine a first vector and a second vector based on a feature center value and the test sample;

[0141] A first similarity calculation unit, configured to obtain a first similarity calculation result representing the similarity between the feature center value and the distribution direction of the test sample by calculating the cosine value of the first vector and the second vector;

[0142] The second similarity calculation unit is used to obtain a second similarity calculation result representing the similarity between the feature center value and the test sample distribution distance by calculating the ratio of the norms of the first vector and the second vector.

[0143] In some specific implementations, the attribute value prediction module 14 may specifically include:

[0144] An attribute prediction unit is used to perform attribute prediction on the feature vector using the target attribute learner to obtain a corresponding attribute value. If the judgment result indicates that the test sample is in a known operating state, a first distance between the attribute value and the attribute vector corresponding to each known operating state is calculated, and based on a nearest neighbor search rule, the known operating state with the smallest first distance is selected as the target known operating state of the test sample, and corresponding warning and maintenance measures are taken based on the target known operating state;

[0145] An early warning and maintenance unit is used to calculate the second distance between the attribute value and the attribute vector corresponding to each unknown operating state if the judgment result indicates that the test sample is in an unknown operating state, and select the unknown operating state with the smallest second distance as the target unknown operating state of the test sample based on the nearest neighbor search rule, and take corresponding early warning and maintenance measures based on the target unknown operating state.

[0146] Furthermore, the present application also discloses an electronic device. Fig.17 This is a structural diagram of an electronic device 20 shown according to an exemplary embodiment, and the content in the figure cannot be regarded as any limitation on the scope of use of this application. The electronic device 20 may specifically include: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. The memory 22 is used to store a computer program, which is loaded and executed by the processor 21 to implement the relevant steps in the electromagnetic suspension system operation status monitoring method disclosed in any of the aforementioned embodiments. In addition, the electronic device 20 in this embodiment may specifically be an electronic computer.

[0147] In this embodiment, the power supply 23 is used to provide working voltage for each hardware device on the electronic device 20; the communication interface 24 can create a data transmission channel between the electronic device 20 and the external device, and the communication protocol it follows is any communication protocol that can be applied to the technical solution of the present application, and is not specifically limited here; the input and output interface 25 is used to obtain external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs and is not specifically limited here.

[0148] In addition, the memory 22, as a carrier for storing resources, can be a read-only memory, a random access memory, a disk or an optical disk, etc. The resources stored thereon can include an operating system 221, a computer program 222, etc., and the storage method can be temporary storage or permanent storage.

[0149] The operating system 221 is used to manage and control the hardware devices and computer program 222 on the electronic device 20, and can be Windows Server, Netware, Unix, Linux, etc. In addition to including a computer program that can be used to complete the electromagnetic suspension system operation status monitoring method performed by the electronic device 20 disclosed in any of the aforementioned embodiments, the computer program 222 can further include a computer program that can be used to complete other specific tasks.

[0150] Furthermore, the present application also discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, the aforementioned electromagnetic suspension system operation status monitoring method is implemented. For the specific steps of the method, reference may be made to the corresponding contents disclosed in the aforementioned embodiments, and no further description will be given here.

[0151] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the 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.

[0152] Professionals may further appreciate 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. Professionals and technicians may 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 this application.

[0153] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein may be implemented directly using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0154] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0155] The technical solution provided by the present application is introduced in detail above. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea. At the same time, for general technicians in this field, according to the idea of ​​the present application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

1. A method for monitoring the operating status of an electromagnetic suspension system, characterized in that: include: Based on the engineering knowledge and experience data of maglev trains, an attribute configuration table including the correspondence between train running status and status attributes is constructed; The train running status includes normal operation, crossing joints, low-frequency shaking, high-frequency vibration, and rail striking; The test samples are obtained by using the running state data of the maglev train during operation, and then the test samples are subjected to feature extraction based on a preset feature extractor and a sliding window function to obtain feature segments, and the feature segments are subjected to multi-scale analysis based on a wavelet transform technology to obtain feature vectors; the running state data includes known running states and unknown running states; the preset feature extractor is a feature extractor determined based on an attention mechanism and a wavelet function; Based on the generalized zero-shot learning method and the attribute configuration table, an initial attribute learner is trained to obtain a target attribute learner, and a gating model is constructed during the training process to determine whether the test sample is in a known operating state or an unknown operating state based on the gating model to obtain a judgment result; The target attribute learner is used to perform attribute prediction on the feature vector to obtain the corresponding attribute value, and then the monitoring result is determined based on the judgment result and the attribute value in combination with the nearest neighbor search rule, so as to take corresponding early warning and maintenance measures based on the monitoring result.

2. The method for monitoring the operating status of an electromagnetic suspension system according to claim 1, characterized in that: The method of obtaining a test sample by using the running state data of the maglev train during operation includes: Using a suspension data acquisition system to collect various operating status data of the maglev train during operation, and determining each of the operating status data as a test sample; Among them, the suspension data acquisition system includes a data collector, a suspension gap sensor, a current sensor, an acceleration sensor, a temperature sensor, a WIFI network center and an embedded host computer software center; the operating status data includes the suspension gap, suspension acceleration, lateral acceleration, suspension electromagnet working current and suspension electromagnet coil surface temperature of the maglev train during operation.

3. The method for monitoring the operating status of an electromagnetic suspension system according to claim 1, characterized in that: The extracting features of the test sample based on a preset feature extractor and a sliding window function to obtain feature fragments includes: The test sample is segmented based on a preset sliding window length and a preset sliding step size to obtain each sliding window, and the mean, the first energy value and the difference between the first peak values ​​of each sliding window are calculated by using an attention mechanism to obtain a corresponding calculation result; Determine a first attention weighting function using the calculation result, so as to determine a target window segment with a maximum weight value based on the first attention weighting function; A second attention weighting function is determined based on the difference between the second energy value and the second peak value of the first-order difference sequence corresponding to the sliding window, and the key features in the target window segment are determined using the second attention weighting function to obtain a feature segment.

4. The method for monitoring the operating status of an electromagnetic suspension system according to any one of claims 1 to 3, characterized in that: The multi-scale analysis of the characteristic segment based on wavelet transform technology to obtain a characteristic vector includes: The preset wavelet function is scaled and translated using preset scale parameters to obtain a target wavelet function, and inner product calculation is performed based on the target wavelet function and the characteristic fragment to obtain wavelet coefficients at different scales, and the wavelet coefficients are extracted to obtain a feature vector.

5. The method for monitoring the operating status of an electromagnetic suspension system according to claim 1, characterized in that: The initial attribute learner is trained based on the generalized zero-sample learning method and the attribute configuration table to obtain a target attribute learner, and a gating model is constructed during the training process to determine whether the test sample is a known operating state or an unknown operating state based on the gating model to obtain a judgment result, including: Training the initial attribute learner based on the generalized zero-shot learning method, the machine learning algorithm and the attribute configuration table to obtain the target attribute learner; In the process of training the initial attribute learner, a gating model is constructed, and a feature center value of each known operating state in the gating model in a feature space is obtained; Performing distribution direction similarity calculation and distribution distance similarity calculation on the test sample and the feature center value respectively to obtain a first similarity calculation result and a second similarity calculation result; The first similarity calculation result and the second similarity calculation result are integrated to obtain a comprehensive similarity score, and the one with the highest comprehensive similarity score is determined as a target similarity score, and it is determined whether the target similarity score is greater than a preset similarity threshold; If the target similarity score is less than a preset similarity threshold, it indicates that the test sample is in an unknown running state; If the target similarity score is not less than a preset similarity threshold, it indicates that the test sample is in a known running state.

6. The method for monitoring the operating status of an electromagnetic suspension system according to claim 5, characterized in that: The performing distribution direction similarity calculation and distribution distance similarity calculation on the test sample and the feature center value respectively to obtain a first similarity calculation result and a second similarity calculation result includes: Determine a first vector and a second vector based on the feature center value and the test sample; By calculating the cosine value of the first vector and the second vector, a first similarity calculation result representing the similarity between the feature center value and the distribution direction of the test sample is obtained; A second similarity calculation result representing the similarity between the feature center value and the test sample distribution distance is obtained by calculating the ratio of the norms of the first vector and the second vector.

7. The method for monitoring the operating status of an electromagnetic suspension system according to claim 5, characterized in that: The method of using the target attribute learner to perform attribute prediction on the feature vector to obtain a corresponding attribute value, and then determining a monitoring result based on the judgment result and the attribute value and in combination with a nearest neighbor search rule, so as to take corresponding early warning and maintenance measures based on the monitoring result, includes: Utilize the target attribute learner to perform attribute prediction on the feature vector to obtain a corresponding attribute value; if the judgment result indicates that the test sample is in a known operating state, calculate a first distance between the attribute value and the attribute vector corresponding to each known operating state, and select the known operating state with the smallest first distance as the target known operating state of the test sample based on a nearest neighbor search rule, and take corresponding early warning and maintenance measures based on the target known operating state; If the judgment result characterizes that the test sample is in an unknown operating state, the second distance between the attribute value and the attribute vector corresponding to each unknown operating state is calculated, and the unknown operating state with the smallest second distance is selected as the target unknown operating state of the test sample based on the nearest neighbor search rule, and corresponding warning and maintenance measures are taken based on the target unknown operating state.

8. An electromagnetic suspension system operation status monitoring device, characterized in that: include: An attribute configuration table construction module is used to construct an attribute configuration table containing the correspondence between the train running state and the state attributes based on the engineering knowledge and experience data of the maglev train; the train running state includes normal operation, seam crossing, low-frequency shaking, high-frequency vibration, and rail smashing; A feature extraction module is used to obtain a test sample using the running state data of the maglev train during operation, and then perform feature extraction on the test sample based on a preset feature extractor and a sliding window function to obtain a feature segment, and perform multi-scale analysis on the feature segment based on a wavelet transform technology to obtain a feature vector; the running state data includes a known running state and an unknown running state; the preset feature extractor is a feature extractor determined based on an attention mechanism and a wavelet function; An attribute learner training module is used to train an initial attribute learner based on a generalized zero-shot learning method and the attribute configuration table to obtain a target attribute learner, and to construct a gating model during the training process to determine whether the test sample is in a known operating state or an unknown operating state based on the gating model to obtain a judgment result; The attribute value prediction unit is used to use the target attribute learner to perform attribute prediction on the feature vector to obtain the corresponding attribute value, and then determine the monitoring result based on the judgment result and the attribute value and in combination with the nearest neighbor search rule, so as to take corresponding early warning and maintenance measures based on the monitoring result.

9. An electronic device, characterized in that: include: Memory, used to store computer programs; A processor is used to execute the computer program to implement the method for monitoring the operating status of an electromagnetic suspension system according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: Used to store a computer program, wherein when the computer program is executed by a processor, the method for monitoring the operating status of an electromagnetic suspension system according to any one of claims 1 to 7 is implemented.

Images