Superconducting levitator internal temperature rise detection method and apparatus based on deep learning

Abstract

A superconducting levitator internal temperature rise detection method based on deep learning. The method comprises: acquiring initial vibration acceleration information of a superconducting levitator, initial temperature rise information, and detection information of the vibration acceleration of the superconducting levitator (S1); performing feature extraction on the initial vibration acceleration information to obtain a high-frequency feature parameter set and a low-frequency feature parameter set (S2); calculating the high-frequency feature parameter set and the low-frequency feature parameter set to obtain wavelet frequency band energy information (S3); inputting the wavelet frequency band energy information and the initial temperature rise information of a superconductor in the superconducting levitator into a preset deep learning network for training (S4); and inputting the detection information of the vibration acceleration of the superconducting levitator into a superconducting levitator internal temperature rise detection model to obtain superconducting levitator internal temperature rise prediction information (S5). The levitation performance of a superconductor is not affected, and the temperature rise detection accuracy is also guaranteed. Further disclosed is a superconducting levitator internal temperature rise detection apparatus based on deep learning.

Description

A method and device for detecting internal temperature rise of a superconducting levitator based on deep learning

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application claims priority to Chinese patent application number 202410033606.9, filed with the Patent Office of China on January 10, 2024, entitled “A method and device for detecting internal temperature rise of a superconducting levitator based on deep learning,” the entire contents of which are incorporated herein by reference. Technical Field

[0003] The present invention relates to the field of temperature rise detection inside a superconducting levitator, and in particular to a method and device for detecting temperature rise inside a superconducting levitator based on deep learning. Background Art

[0004] During the operation of high-temperature superconducting pinned maglev trains, AC losses within the superconductor are generated due to factors such as the uneven magnetic field of the permanent magnet track, random demagnetization of the magnets, and train vibrations, which in turn cause temperature rise. The temperature rise of the superconductor within the superconducting levitator is a key parameter reflecting the levitation state of the superconducting levitator. Therefore, during the operation of high-temperature superconducting pinned maglev trains, timely and real-time monitoring of the internal temperature rise of the superconductor is necessary to prevent the train from experiencing quenching. Current methods for detecting the temperature rise of superconductors within superconducting levitators primarily involve drilling holes into the superconductor and inserting temperature sensors to directly measure the internal temperature rise. However, this method damages the superconductor structure, which, on the one hand, affects the superconductor's levitation performance; on the other hand, the existing method for measuring the internal temperature rise of superconductors mostly uses platinum resistance thermometers, which are significantly affected by magnetic field fluctuations, resulting in large errors in the final temperature rise measurement results. Therefore, a method for detecting the internal temperature rise of superconducting levitators is urgently needed that does not affect the levitation performance of the superconductor and, on the other hand, ensures the accuracy of temperature rise detection.

[0005] Summary of the Invention

[0006] The purpose of the present invention is to provide a method and device for detecting the internal temperature rise of a superconducting levitator based on deep learning to improve the above-mentioned problem. To achieve the above-mentioned purpose, the technical solution adopted by the present invention is as follows:

[0007] In a first aspect, the present application provides a method for detecting temperature rise inside a superconducting levitator based on deep learning, the method comprising:

[0008] Acquiring initial vibration acceleration information of the superconducting levitator, initial temperature rise information of the superconductor in the superconducting levitator, and detection information of the vibration acceleration of the superconducting levitator;

[0009] Extracting features of the initial vibration acceleration information of the superconducting levitator to obtain a high-frequency feature parameter set and a low-frequency feature parameter set corresponding to the initial vibration acceleration information;

[0010] Performing wavelet band energy calculation on the high-frequency feature parameter set and the low-frequency feature parameter set to obtain wavelet band energy information;

[0011] Inputting the wavelet frequency band energy information and the initial temperature rise information of the superconductor in the superconducting levitator into a preset deep learning network for training to obtain a temperature rise detection model inside the superconducting levitator;

[0012] The detection information of the superconducting levitator vibration acceleration is input into the superconducting levitator internal temperature rise detection model to obtain the superconducting levitator internal temperature rise prediction information, which is used to reflect the real-time temperature rise of the superconductor inside the superconducting levitator.

[0013] In a second aspect, the present application further provides a device for detecting internal temperature rise of a superconducting levitator based on deep learning, the device comprising:

[0014] An acquisition module, configured to acquire initial vibration acceleration information of the superconducting levitator, initial temperature rise information of the superconductor in the superconducting levitator, and detection information of vibration acceleration of the superconducting levitator;

[0015] a first processing module, configured to extract features from the initial vibration acceleration information of the superconducting levitator to obtain a high-frequency feature parameter set and a low-frequency feature parameter set corresponding to the initial vibration acceleration information;

[0016] A second processing module is used to perform wavelet band energy calculation on the high-frequency feature parameter set and the low-frequency feature parameter set to obtain wavelet band energy information;

[0017] a third processing module, configured to input the wavelet frequency band energy information and the initial temperature rise information of the superconductor in the superconducting levitator into a preset deep learning network for training, thereby obtaining a temperature rise detection model for the superconducting levitator;

[0018] The prediction module is configured to input the detection information of the superconducting levitator's vibration acceleration into the superconducting levitator's internal temperature rise detection model to obtain superconducting levitator's internal temperature rise prediction information, wherein the superconducting levitator's internal temperature rise prediction information is used to reflect the real-time temperature rise of the superconductor inside the superconducting levitator.

[0019] The beneficial effects of the present invention are:

[0020] The present invention introduces a superconducting levitator internal temperature rise detection model based on deep learning. The model extracts the initial vibration acceleration information of the superconducting levitator and explores the nonlinear mapping relationship between the temperature rise of the superconductor inside the superconductor and the time-frequency domain characteristics of the vibration acceleration. A contactless method for detecting the internal temperature rise of the superconductor inside the superconducting levitator is proposed. This method does not affect the suspension performance of the superconductor. On the other hand, based on deep learning, the detection accuracy of the superconductor temperature rise inside the superconducting levitator can be guaranteed, avoiding the introduction of temperature detection sensors and eliminating the need to set up a specific installation space corresponding to the temperature detection sensor.

[0021] Other features and advantages of the present invention will be described in the following description, and in part will become apparent from the description, or understood by practicing the embodiments of the present invention. The purposes and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description, claims, and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without paying any creative work.

[0023] FIG1 is a schematic flow chart of a method for detecting temperature rise inside a superconducting levitator based on deep learning according to an embodiment of the present invention;

[0024] FIG2 is a schematic structural diagram of a device for detecting temperature rise inside a superconducting levitator based on deep learning according to an embodiment of the present invention;

[0025] FIG3 is a schematic structural diagram of a device for detecting internal temperature rise of a superconducting levitator based on deep learning according to an embodiment of the present invention;

[0026] FIG4 is a schematic structural diagram of a temperature rise detection model for a superconducting levitator according to the present invention;

[0027] Markings in the figure: 1. Acquisition module; 2. First processing module; 3. Second processing module; 4. Third processing module; 5. Prediction module; 800. Internal temperature rise detection equipment of superconducting levitator based on deep learning; 801. Processor; 802. Memory; 803. Multimedia component; 804. I / O interface; 805. Communication component. DETAILED DESCRIPTION

[0028] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0029] It should be noted that similar reference numerals and letters represent similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined or explained in subsequent drawings. At the same time, in the description of the present invention, the terms "first", "second", etc. are used only to distinguish the description and should not be understood as indicating or implying relative importance.

[0030] Example 1:

[0031] This embodiment provides a method for detecting the internal temperature rise of a superconducting levitator based on deep learning.

[0032] Referring to FIG1 , the method includes steps S1 to S5, specifically:

[0033] S1: Acquiring initial vibration acceleration information of the superconducting levitator, initial temperature rise information of the superconductor in the superconducting levitator, and detection information of vibration acceleration of the superconducting levitator;

[0034] In step S1, both initial vibration acceleration information and superconducting levitator vibration acceleration detection information can be collected using a vibration acceleration sensor. Since superconductors are high-temperature superconducting materials with a critical temperature below 93K, they can be cooled using liquid nitrogen. Once in a superconducting state, the pinning centers within the superconductor can capture magnetic flux lines above the permanent magnet track, thereby providing self-stabilization. The initial vibration acceleration information in this method is used to reflect vibration acceleration values ​​within the range from liquid nitrogen temperature to the quench temperature. The superconducting levitator vibration acceleration detection information is used to reflect the current real-time vibration acceleration of the superconducting levitator.

[0035] The initial temperature rise information of the superconductor in the superconducting levitator is collected by a temperature sensor to facilitate data training in the later stage.

[0036] S2: extracting features from the initial vibration acceleration information of the superconducting levitator to obtain a high-frequency feature parameter set and a low-frequency feature parameter set corresponding to the initial vibration acceleration information;

[0037] In step S2, to clarify the specific process of feature extraction, step S2 includes S21 to S23, specifically:

[0038] S21: Obtaining the scale factor and translation factor of the wavelet transform;

[0039] S22: Decomposing the scale factor, the translation factor, and the initial vibration acceleration information of the superconducting levitator using a preset wavelet decomposition model to obtain a plurality of decomposition characteristic parameters;

[0040] In step S22, the wavelet decomposition model is:

[0041] In the above formula (1), w f (a, b) represents a single decomposition characteristic parameter when the scale factor of the wavelet transform is a and the translation factor of the wavelet transform is b. a represents the scale factor of the wavelet transform, b represents the translation factor of the wavelet transform, and f(t) represents the initial vibration acceleration information of the superconducting levitator. represents the complex conjugate of ψ, ψ represents the preset basic wavelet, and t represents the time variable.

[0042] S23: Sort the multiple decomposed feature parameters in sequence to obtain a high-frequency feature parameter set and a low-frequency feature parameter set.

[0043] In step S23, in order to clarify the specific process of sequentially sorting the multiple decomposition feature parameters, step S23 includes S231 to S235, specifically:

[0044] S231: Obtain the number of forks and the number of fork levels in the complete binary tree;

[0045] S232: constructing the first decomposition feature parameter among the plurality of decomposition feature parameters as a stack top;

[0046] In step S22, the role of the heap is to express the decomposition feature parameters in the form of an array. After the top of the heap is set, it is convenient to insert the remaining decomposition feature parameters in sequence, so as to associate each decomposition feature parameter through the node position in the binary tree.

[0047] S233: inserting the remaining decomposition feature parameters in sequence according to the number of forks and the number of fork levels to obtain an unsorted decomposition feature parameter pile;

[0048] In step S233, all decomposition feature parameters are simply inserted into the corresponding order to obtain an unsorted decomposition feature parameter pile.

[0049] S234: Sorting the unsorted decomposition feature parameter pile according to a preset minimum heap model to obtain a sorted decomposition feature parameter pile;

[0050] In step S234, the preset minimum heap model is: ① the minimum element value in the minimum heap appears at the root node, that is, the smallest low-frequency feature parameter after sorting appears at the top of the heap; ② the element value of each parent node in the heap is smaller than the element value of the child node.

[0051] When sorting, the existing heap sorting algorithm can be used to sort to obtain a sorted decomposition feature parameter heap.

[0052] S235: Divide the sorted decomposition feature parameter stack according to a preset frequency threshold to obtain a high-frequency feature parameter set and a low-frequency feature parameter set.

[0053] In step S235, the division is performed according to the preset frequency threshold. When the element value corresponding to the feature parameter is greater than the preset frequency threshold, it is saved in the high-frequency feature parameter set; when the element value corresponding to the feature parameter is less than the preset frequency threshold, it is saved in the low-frequency feature parameter set.

[0054] S3: performing wavelet band energy calculation on the high-frequency feature parameter set and the low-frequency feature parameter set to obtain wavelet band energy information;

[0055] In step S3, the specific calculation process of the wavelet band energy includes steps S31 to S35, specifically:

[0056] S31: solving each high-frequency feature parameter in the high-frequency feature parameter set by a preset wavelet basis function to obtain a first wavelet basis point corresponding to each high-frequency feature parameter;

[0057] In step S31, the wavelet basis function is an existing method, which aims to decompose each input high-frequency feature parameter into corresponding wavelet basis point coordinates.

[0058] S32: Calculate each of the first wavelet base points to obtain a high-frequency band energy set;

[0059] In step S32, the calculation formula is: E H ={b1 2 +b2 2 +…+b r 2}(2);

[0060] In the above formula (2), E H represents the high frequency band energy set; {b1,b2,…,b r} represents the first wavelet basis point corresponding to the high-frequency characteristic parameter.

[0061] The first wavelet basis point corresponding to the characteristic parameter.

[0062] S33: solving each low-frequency feature parameter in the low-frequency feature parameter set by using a preset wavelet basis function to obtain a second wavelet basis point corresponding to each low-frequency feature parameter;

[0063] In step S33, the wavelet basis function is an existing method, and its purpose is also to decompose each input low-frequency feature parameter into corresponding wavelet basis point coordinates.

[0064] S34: Calculate each of the second wavelet base points to obtain a low-frequency band energy set;

[0065] In step S34, the calculation formula is: E L ={a1 2 +a2 2 +…+a s 2}(3);

[0066] In the above formula (3), E L represents the low-frequency band energy set; {a1,a2,…,a s} represents the second wavelet basis point corresponding to the low-frequency characteristic parameter.

[0067] S35: Solving the high-frequency band energy set and the low-frequency band energy set through a preset normalization model to obtain wavelet band energy information.

[0068] In step S35, the normalized model is:

[0069] In the above formula (4), A n Indicates the wavelet band energy information corresponding to n characteristic parameters; E L represents the low-frequency band energy set; E H Represents the high-frequency band energy set; E n 2 Represents the frequency band energy corresponding to n characteristic parameters.

[0070] S4: Inputting the wavelet frequency band energy information and the initial temperature rise information of the superconductor in the superconducting levitator into a preset deep learning network for training to obtain a temperature rise detection model inside the superconducting levitator;

[0071] As shown in FIG4 , in step S4 , the preset deep learning network is an error back propagation neural network. Step S4 includes S41 to S44 , specifically:

[0072] S41: Obtain the number of neurons in the input layer, the number of neurons in the hidden layer, and the number of neurons in the output layer;

[0073] S42: constructing a preset hidden layer output model according to the number of neurons in the input layer and the number of neurons in the hidden layer to obtain output information of the hidden layer neurons;

[0074] In step S42, the hidden layer output model is:

[0075] In the above formula (5), Represents the output information of neuron j in the hidden layer; σ h represents the activation function of the hidden layer neurons; m represents the number of neurons in the input layer; q represents the number of neurons in the hidden layer; represents the connection weight between the i-th neuron in the input layer and the j-th neuron in the hidden layer; x i represents the i-th neuron in the input layer; is the offset of the jth neuron in the hidden layer.

[0076] S43: constructing a preset output layer output model according to the number of neurons in the hidden layer and the number of neurons in the output layer to obtain output information of the output layer neurons.

[0077] In step S43, the output layer output model is:

[0078] In the above formula (6), y k Represents the output information of neuron k in the output layer; σ 0 represents the activation function of the output layer neurons; j represents the value range of the number of neurons in the hidden layer; q represents the number of neurons in the hidden layer; n represents the number of neurons in the output layer; represents the connection weight between the jth neuron in the hidden layer and the kth neuron in the output layer; Represents the output information of neuron j in the hidden layer; represents the offset of the kth neuron in the output layer; y A Represents the output information of the output layer neurons, Represents the output information of the kth neuron in the output layer.

[0079] In this method, the training parameters of the error back propagation neural network are: the number of neurons in the hidden layer is 5, the number of neurons in the input layer is 5, the learning rate coefficient is 0.001, the training frequency is 1000, and the minimum error of the training target is 10 -6 .

[0080] In this method, in order to overcome the problems of poor generalization ability and poor robustness of the error back propagation neural network, the preset deep learning network is an error back propagation neural network, and the error back propagation neural network is multiple, including steps S44 to S46, specifically:

[0081] S44: Training each of the error back propagation neural networks to obtain prediction error information;

[0082] In step S44, the calculation formula of the prediction error information is: f =∑ l D(l)l=2,…,M(7);

[0083] In the above formula (7), e f Represents the prediction error information; D(l) represents the output error of the output information of the lth error back propagation neural network; the value range of l is 2 to M.

[0084] S45: Calculating the prediction error information using a preset weight prediction model to obtain prediction sequence weight information corresponding to the prediction error information;

[0085] In step S45, the weight prediction model is:

[0086] In the above formula (8), α g Indicates the prediction sequence weight information corresponding to the prediction error information; e f Represents prediction error information.

[0087] S46: constructing a strong classification model in the superconducting levitator internal temperature rise detection model based on the multiple error back propagation neural networks and the prediction sequence weight information, and the strong classification model is used for supplementary prediction.

[0088] In step S46, the strong classification model is:

[0089] In the above formula (6), h(y) represents the output information of the strong classification model after supplementary prediction; z represents the number of training rounds, which ranges from 1 to T; α g Represents the prediction sequence weight information corresponding to the prediction error information; f[V(z),α g ] represents the preset training function of the error back propagation neural network in each round of training.

[0090] S5: Inputting the detection information of the superconducting levitator vibration acceleration into the superconducting levitator internal temperature rise detection model to obtain superconducting levitator internal temperature rise prediction information, wherein the superconducting levitator internal temperature rise prediction information is used to reflect the real-time temperature rise of the superconductor inside the superconducting levitator.

[0091] Because the superconducting levitator's internal temperature rise detection model reflects the nonlinear mapping relationship between the superconductor's internal temperature rise and the time-frequency characteristics of its vibration acceleration, once the superconductor's vibration acceleration detection information is obtained, it can be input into the detection model to obtain the superconducting levitator's internal temperature rise prediction information. This method is a non-contact method for detecting the superconducting levitator's internal temperature rise, which does not affect the superconductor's suspension performance. Furthermore, based on deep learning, it can ensure the detection accuracy of the superconductor's internal temperature rise, avoid the introduction of temperature detection sensors, and do not require the establishment of a specific installation space for the temperature detection sensors.

[0092] Example 2:

[0093] As shown in FIG2 , this embodiment provides a device for detecting temperature rise inside a superconducting levitator based on deep learning, the device comprising:

[0094] Acquisition module 1, used to acquire initial vibration acceleration information of the superconducting levitator, initial temperature rise information of the superconductor in the superconducting levitator, and detection information of vibration acceleration of the superconducting levitator;

[0095] A first processing module 2 is configured to extract features from the initial vibration acceleration information of the superconducting levitator to obtain a high-frequency feature parameter set and a low-frequency feature parameter set corresponding to the initial vibration acceleration information;

[0096] A second processing module 3 is configured to perform wavelet band energy calculation on the high-frequency feature parameter set and the low-frequency feature parameter set to obtain wavelet band energy information;

[0097] The third processing module 4 is configured to input the wavelet frequency band energy information and the initial temperature rise information of the superconductor in the superconducting levitator into a preset deep learning network for training, thereby obtaining a temperature rise detection model for the superconducting levitator.

[0098] The prediction module 5 is configured to input the detection information of the superconducting levitator vibration acceleration into the superconducting levitator internal temperature rise detection model to obtain the superconducting levitator internal temperature rise prediction information, which is used to reflect the real-time temperature rise of the superconductor inside the superconducting levitator.

[0099] It should be noted that, regarding the apparatus in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated on here.

[0100] Example 3:

[0101] Corresponding to the above method embodiment, this embodiment also provides a superconducting levitator internal temperature rise detection device based on deep learning. The superconducting levitator internal temperature rise detection device based on deep learning described below and the superconducting levitator internal temperature rise detection method based on deep learning described above can refer to each other.

[0102] FIG3 is a block diagram illustrating a deep learning-based device 800 for detecting the internal temperature rise of a superconducting levitator, according to an exemplary embodiment. As shown in FIG3 , the deep learning-based device 800 may include a processor 801 and a memory 802 . The deep learning-based device 800 may also include one or more of a multimedia component 803 , an I / O interface 804 , and a communication component 805 .

[0103] The processor 801 is used to control the overall operation of the deep learning-based superconducting levitator internal temperature rise detection device 800 to complete all or part of the steps of the deep learning-based superconducting levitator internal temperature rise detection method described above. The memory 802 is used to store various types of data to support the operation of the deep learning-based superconducting levitator internal temperature rise detection device 800. This data may include, for example, instructions for any application or method operating on the deep learning-based superconducting levitator internal temperature rise detection device 800, as well as application-related data such as contact information, sent and received messages, images, audio, video, and so on. The memory 802 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk. The multimedia component 803 may include a screen and an audio component. The screen may be, for example, a touch screen, and the audio component is used to output and / or input audio signals. For example, the audio component may include a microphone for receiving external audio signals. The received audio signal may be further stored in the memory 802 or transmitted via the communication component 805. The audio component also includes at least one speaker for outputting audio signals. The I / O interface 804 provides an interface between the processor 801 and other interface modules, which may be a keyboard, a mouse, buttons, etc. These buttons may be virtual buttons or physical buttons. The communication component 805 is used for wired or wireless communication between the deep learning-based superconducting levitator internal temperature rise detection device 800 and other devices. Wireless communication, such as Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G or 4G, or a combination of one or more thereof, therefore the corresponding communication component 805 may include: a Wi-Fi module, a Bluetooth module, an NFC module.

[0104] In an exemplary embodiment, the deep learning-based superconducting levitator internal temperature rise detection device 800 can be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above-described deep learning-based superconducting levitator internal temperature rise detection method.

[0105] In another exemplary embodiment, a computer-readable storage medium including program instructions is also provided. When executed by a processor, the program instructions implement the steps of the above-described method for detecting the internal temperature rise of a superconducting levitator based on deep learning. For example, the computer-readable storage medium may be the aforementioned memory 802 including the program instructions. The program instructions may be executed by the processor 801 of the apparatus 800 for detecting the internal temperature rise of a superconducting levitator based on deep learning to implement the above-described method for detecting the internal temperature rise of a superconducting levitator based on deep learning.

[0106] Example 4:

[0107] Corresponding to the above method embodiment, this embodiment further provides a readable storage medium. The readable storage medium described below and the method for detecting the internal temperature rise of a superconducting levitator based on deep learning described above can refer to each other.

[0108] A readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the method for detecting the internal temperature rise of a superconducting levitator based on deep learning according to the above method embodiment.

[0109] The readable storage medium may specifically be any readable storage medium that can store program code, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0110] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be within the scope of protection of the present invention.

[0111] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A method for detecting the internal temperature rise of a superconducting suspension device based on deep learning, characterized in that, Including: Obtaining the initial vibration acceleration information of the superconducting suspension device, the initial temperature rise information of the superconductor inside the superconducting suspension device, and the detection information of the vibration acceleration of the superconducting suspension device; Performing feature extraction on the initial vibration acceleration information of the superconducting suspension device to obtain a high-frequency feature parameter set and a low-frequency feature parameter set corresponding to the initial vibration acceleration information; Performing wavelet band energy calculation on the high-frequency feature parameter set and the low-frequency feature parameter set to obtain wavelet band energy information; Inputting the wavelet band energy information and the initial temperature rise information of the superconductor inside the superconducting suspension device into a preset deep learning network for training to obtain a temperature rise detection model inside the superconducting suspension device; Inputting the detection information of the vibration acceleration of the superconducting suspension device into the temperature rise detection model inside the superconducting suspension device to obtain the predicted temperature rise information inside the superconducting suspension device, and the predicted temperature rise information inside the superconducting suspension device is used to reflect the real-time temperature rise of the superconductor inside the superconducting suspension device.

2. The method for detecting the internal temperature rise of a superconducting magnetic levitator based on deep learning according to claim 1, characterized in that Performing feature extraction on the initial vibration acceleration information of the superconducting suspension device to obtain a high-frequency feature parameter set and a low-frequency feature parameter set corresponding to the initial vibration acceleration information, including: Obtaining the scale factor of wavelet transform and the translation factor of wavelet transform; Decomposing the scale factor, the translation factor, and the initial vibration acceleration information of the superconducting suspension device through a preset wavelet decomposition model to obtain a plurality of decomposition feature parameters; Sequentially sorting the plurality of decomposition feature parameters to obtain a high-frequency feature parameter set and a low-frequency feature parameter set.

3. The method for detecting the internal temperature rise of a superconducting levitator based on deep learning according to claim 2, characterized in that, Sequentially sorting the plurality of decomposition feature parameters to obtain a high-frequency feature parameter set and a low-frequency feature parameter set, including: Obtaining the number of bifurcations and the bifurcation levels in the complete binary tree; Constructing the first decomposition feature parameter among the plurality of decomposition feature parameters as the heap top; Sequentially inserting the remaining decomposition feature parameters according to the number of bifurcations and the bifurcation levels to obtain an unsorted decomposition feature parameter heap; Sorting the unsorted decomposition feature parameter heap according to a preset minimum heap model to obtain a sorted decomposition feature parameter heap; Dividing the sorted decomposition feature parameter heap according to a preset frequency threshold to obtain a high-frequency feature parameter set and a low-frequency feature parameter set.

4. The method for detecting the internal temperature rise of a superconducting magnetic levitator based on deep learning according to claim 1, wherein Performing wavelet band energy calculation on the high-frequency feature parameter set and the low-frequency feature parameter set to obtain wavelet band energy information, including: Solving each high-frequency feature parameter in the high-frequency feature parameter set through a preset wavelet basis function to obtain the first wavelet base point corresponding to each high-frequency feature parameter; Calculating each of the first wavelet base points to obtain a high-frequency band energy set; Solving each low-frequency feature parameter in the low-frequency feature parameter set through a preset wavelet basis function to obtain the second wavelet base point corresponding to each low-frequency feature parameter; Calculating each of the second wavelet base points to obtain a low-frequency band energy set; Solving the high-frequency band energy set and the low-frequency band energy set through a preset normalization model to obtain wavelet band energy information.

5. The method for detecting the internal temperature rise of a superconducting levitator based on deep learning according to claim 1, wherein Input the wavelet band energy information and the initial temperature rise information of the superconductor in the superconducting suspension into a preset deep learning network for training to obtain a temperature rise detection model inside the superconducting suspension. Among them, the preset deep learning network is an error backpropagation neural network, including: Obtain the number of neurons in the input layer, the number of neurons in the hidden layer, and the number of neurons in the output layer; Construct according to the number of neurons in the input layer and the number of neurons in the hidden layer through a preset hidden layer output model to obtain the output information of the hidden layer neurons; Construct according to the number of neurons in the hidden layer and the number of neurons in the output layer through a preset output layer output model to obtain the output information of the output layer neurons.

6. The method for detecting the internal temperature rise of a superconducting suspension device based on deep learning according to claim 2, wherein The wavelet decomposition model is as follows: In the above formula, w f (a, b) represents a single decomposition characteristic parameter when the scaling factor of the wavelet transform is a and the translation factor of the wavelet transform is b. a represents the scaling factor of the wavelet transform, b represents the translation factor of the wavelet transform, and f(t) represents the initial vibration acceleration information of the superconducting suspension device. Represents the complex conjugate of ψ, ψ represents a preset basic wavelet, and t represents a time variable.

7. The method for detecting the internal temperature rise of a superconducting levitator based on deep learning according to claim 5, characterized in that, The output model of the hidden layer is as follows: In the above formula, represents the output information of neuron j in the hidden layer; σ h represents the activation function of the hidden layer neurons; m represents the number of neurons in the input layer; q represents the number of neurons in the hidden layer; represents the connection weight between the $i$-th neuron in the input layer and the $j$-th neuron in the hidden layer; $x$ i represents the $i$-th neuron in the input layer; Is the offset of the jth neuron in the hidden layer.

8. The method for detecting the internal temperature rise of a superconducting levitator based on deep learning according to claim 5, characterized in that The output layer output model is as follows: In the above formula, y k represents the output information of neuron k in the output layer; σ 0 represents the activation function of the output layer neurons; j represents the value range of the number of neurons in the hidden layer; q represents the number of neurons in the hidden layer; n represents the number of neurons in the output layer; Denotes the connection weight between the j-th neuron in the hidden layer and the k-th neuron in the output layer; Represents the output information of neuron j in the hidden layer; denotes the offset of the k-th neuron in the output layer; y A denotes the output information of the neurons in the output layer; Represents the output information of the kth neuron in the output layer.

9. The method for detecting the internal temperature rise of a superconducting levitator based on deep learning according to claim 5, wherein The training parameters of the error backpropagation neural network are as follows: the number of neurons in the hidden layer is 5, the number of neurons in the input layer is 5, the learning rate coefficient is 0.001, the training frequency is 1000, and the minimum error of the training target is 10 -6 .

10. An internal temperature rise detection device for a superconducting levitator based on deep learning, characterized in that, Includes: An acquisition module for acquiring the initial vibration acceleration information of the superconducting suspension, the initial temperature rise information of the superconductor in the superconducting suspension, and the detection information of the vibration acceleration of the superconducting suspension; A first processing module for extracting features from the initial vibration acceleration information of the superconducting suspension to obtain a high-frequency feature parameter set and a low-frequency feature parameter set corresponding to the initial vibration acceleration information; A second processing module for calculating the wavelet band energy of the high-frequency feature parameter set and the low-frequency feature parameter set to obtain wavelet band energy information; A third processing module for inputting the wavelet band energy information and the initial temperature rise information of the superconductor in the superconducting suspension into a preset deep learning network for training to obtain a temperature rise detection model inside the superconducting suspension; A prediction module for inputting the detection information of the vibration acceleration of the superconducting suspension into the temperature rise detection model inside the superconducting suspension to obtain temperature rise prediction information inside the superconducting suspension, and the temperature rise prediction information inside the superconducting suspension is used to reflect the real-time temperature rise of the superconductor inside the superconducting suspension.

Images