Characteristic spectrum construction method for armored vehicle state monitoring and diagnosis
By constructing a feature map of the armored vehicle status, the one-dimensional data is converted into two-dimensional image grayscale values, which solves the problem that sensor information cannot be applied to deep learning models, and improves the accuracy and fault diagnosis capabilities of armored vehicle status monitoring.
Patent Information
- Application Number
- CN202510289289.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-12
- Publication Date
- 2025-07-04
AI Technical Summary
The armored vehicle status information collected by the sensor cannot be directly applied to the deep learning model, resulting in insufficient accuracy of monitoring and diagnosis of armored vehicle status.
Construct a feature map, extract and convert the armored vehicle status information collected by the sensor into image grayscale values, generate a two-dimensional feature map, and apply it to deep learning models for status monitoring and diagnosis.
The accuracy of armored vehicle status monitoring and diagnosis is improved, and the high learning ability and recognition accuracy of deep learning models are used to achieve more accurate status monitoring and fault diagnosis.
Smart Images
Figure CN120256859A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of armored vehicle condition monitoring and diagnosis, and in particular to a method for constructing a feature map for armored vehicle condition monitoring and diagnosis and an armored vehicle condition monitoring and diagnosis method based on the feature map. Background Art
[0002] In the aspect of equipment condition monitoring and fault diagnosis, usually one or more sensors are used to obtain equipment condition information, and then the equipment condition information is input into a pattern recognition model for condition or fault discrimination.
[0003] In terms of pattern recognition models, the current mainstream models are artificial intelligence models, which have been widely applied because of their learning ability and stronger non-linear data processing ability.
[0004] Traditional artificial intelligence models are machine learning models, and their input is a one-dimensional data sequence. With the development of artificial intelligence models, the current mainstream artificial intelligence models are deep learning models, that is, convolutional neural network models. Compared with traditional machine learning models, deep learning models have higher learning ability and recognition accuracy.
[0005] However, the input of deep learning models is usually images (two-dimensional data), so the equipment condition information collected by sensors cannot be applied to deep learning models, and thus the accuracy of equipment condition monitoring and diagnosis cannot be improved. Summary of the Invention
[0006] The present invention provides a method for constructing a feature map for armored vehicle condition monitoring and diagnosis and an armored vehicle condition monitoring and diagnosis method, which solve the problem that the vehicle condition information collected by one or more sensors cannot be applied to the deep learning model for armored vehicle condition monitoring and diagnosis and cannot improve the vehicle condition diagnosis accuracy.
[0007] The present invention provides a method for constructing a feature map for armored vehicle condition monitoring and diagnosis, and the method includes: obtaining vehicle condition information for reflecting the condition of an armored vehicle; performing feature extraction on the vehicle condition information for reflecting the condition of the armored vehicle to obtain various feature information and feature values of the armored vehicle; respectively converting the feature values of the various feature information into image gray values of the various feature information; and constructing a feature map for armored vehicle condition monitoring and diagnosis by using the image gray values of the various feature information.
[0008] Preferably, the vehicle condition information includes at least one of engine cylinder compression pressure information, engine speed information, engine vibration information, and transmission vibration information.
[0009] Preferably, when the vehicle state information includes the engine cylinder compression pressure information, the multiple characteristic information includes the peak value of the engine cylinder compression pressure, the pulse index of the engine cylinder compression pressure, and the waveform index of the engine cylinder compression pressure. The feature extraction of the vehicle state information used to reflect the state of the armored vehicle to obtain the multiple characteristic information and characteristic values of the armored vehicle includes: determining the absolute value of the engine cylinder compression pressure, the average value of the engine cylinder compression pressure, the absolute average value of the engine cylinder compression pressure, and the mean square value of the engine cylinder compression pressure according to the compression pressure signal sequence of the engine cylinder within one working cycle; extracting the maximum absolute value from the absolute values of the engine cylinder compression pressure as the peak value of the engine cylinder compression pressure; determining the pulse index of the engine cylinder compression pressure according to the peak value of the engine cylinder compression pressure and the average value of the engine cylinder compression pressure; and determining the waveform index of the engine cylinder compression pressure according to the absolute average value of the engine cylinder compression pressure and the mean square value of the engine cylinder compression pressure.
[0010] Preferably, when the vehicle state information includes the engine speed information, the multiple characteristic information includes the maximum engine speed, the engine speed fluctuation value, and the engine acceleration time. The feature extraction of the vehicle state information used to reflect the state of the armored vehicle to obtain the multiple characteristic information and characteristic values of the armored vehicle includes: obtaining the maximum speed value of the engine in the no-load state, the maximum speed value and the minimum speed value in the no-load idle state, and the engine acceleration time when the engine accelerates to the maximum speed when the accelerator is depressed at the maximum speed in the no-load idle state from the engine speed information; and obtaining the engine speed fluctuation value according to the maximum speed value and the minimum speed value of the engine in the no-load idle state.
[0011] Preferably, when the vehicle state information includes the engine vibration information, the multiple characteristic information includes the peak value of the engine vibration, the peak-to-peak value of the engine vibration, the effective value of the engine vibration, the waveform index of the engine vibration, and the pulse index of the engine vibration. The feature extraction of the vehicle state information used to reflect the state of the armored vehicle to obtain the multiple characteristic information and characteristic values of the armored vehicle includes: obtaining the absolute value of the engine vibration, the absolute average value of the engine vibration, the maximum and minimum values of the engine vibration, and the mean square value of the engine vibration according to the engine vibration signal sequence; extracting the maximum absolute value from the absolute values of the engine vibration as the peak value of the engine vibration; determining the peak-to-peak value of the engine vibration according to the maximum and minimum values of the engine vibration; determining the effective value of the engine vibration according to the mean square value of the engine vibration; determining the waveform index of the engine vibration according to the mean square value of the engine vibration and the absolute average value of the engine vibration; and determining the engine pulse index according to the peak value of the engine vibration and the absolute average value of the engine vibration.
[0012] Preferably, when the vehicle state information includes the transmission vibration information, the multiple characteristic information includes the transmission vibration peak value, the transmission vibration peak-to-peak value, the transmission vibration effective value, the transmission vibration waveform index, and the transmission vibration pulse index. The feature extraction of multiple vehicle state information for reflecting the state of the armored vehicle to obtain the multiple characteristic information and characteristic values of the armored vehicle includes: obtaining the absolute value of the transmission vibration, the absolute average value of the transmission vibration, the maximum and minimum values of the transmission vibration, and the mean square value of the transmission vibration according to the transmission vibration signal sequence; extracting the maximum absolute value from the absolute value of the transmission vibration as the transmission vibration peak value; determining the transmission vibration peak-to-peak value according to the maximum and minimum values of the transmission vibration; determining the transmission vibration effective value according to the mean square value of the transmission vibration; determining the transmission vibration waveform index according to the mean square value of the transmission vibration and the absolute average value of the transmission vibration; and determining the transmission pulse index according to the transmission vibration peak value and the absolute average value of the transmission vibration.
[0013] Preferably, the conversion of the characteristic values of the multiple characteristic information into the image gray values of the multiple characteristic information includes: performing normalization processing on the characteristic values of each characteristic information to obtain the normalized characteristic values of each characteristic information; and performing gray conversion on the normalized characteristic values of each characteristic information to obtain the image gray values of each characteristic information.
[0014] Preferably, the construction of the characteristic map for the state monitoring and diagnosis of the armored vehicle by using the image gray values of the multiple characteristic information includes: arranging the image gray values of the multiple characteristic information of the armored vehicle according to a preset arrangement method to obtain a two-dimensional array composed of the image gray values of the multiple characteristic information of the armored vehicle; mapping each image gray value of the two-dimensional array to the pixel value of the image, and using an image encoder to save the two-dimensional array as a gray image as the characteristic map.
[0015] The present invention also provides a method for monitoring and diagnosing the state of an armored vehicle, and the method includes: generating a characteristic map of an armored vehicle by using the above-mentioned method for constructing a characteristic map for monitoring and diagnosing the state of an armored vehicle; and analyzing the characteristic map of the armored vehicle by using a trained deep learning model for monitoring and diagnosing the state of an armored vehicle to determine the vehicle state of the armored vehicle, where the state includes a normal state and at least one type of fault state.
[0016] Preferably, the trained deep learning model for armored vehicle status monitoring and diagnosis is obtained through the following steps: using the above-mentioned method for constructing feature maps for armored vehicle status monitoring and diagnosis, processing the vehicle status information in the normal state and each type of fault state in the dataset respectively to obtain the normal state feature map and the feature map of each type of fault state; using the normal state feature map and the feature map of each type of fault state to train the deep learning model for armored vehicle status monitoring and diagnosis, and obtaining the trained deep learning model for armored vehicle status monitoring and diagnosis.
[0017] The present invention is based on converting one-dimensional multi-type information for armored vehicle status monitoring and diagnosis into two-dimensional multi-type information fusion feature maps for armored vehicle status monitoring and diagnosis. These feature maps are applied to the deep learning model for armored vehicle status monitoring and diagnosis, thereby improving the accuracy of armored vehicle status monitoring and diagnosis. Brief Description of the Drawings
[0018] Figure 1 is a schematic flow chart of the method for constructing feature maps for armored vehicle status monitoring and diagnosis provided by the present invention;
[0019] Figure 2 is a schematic structural diagram of the feature map provided by the present invention;
[0020] Figure 3 is a schematic diagram of the conversion from grayscale value to grayscale image provided by the present invention;
[0021] Figure 4 is a schematic flow chart of the armored vehicle status monitoring and diagnosis method provided by the present invention. Detailed Description of the Embodiments
[0022] The following is a detailed description of the embodiments of the present invention with reference to the accompanying drawings. It should be understood that the embodiments described below are only used to illustrate and explain the present invention, and are not used to limit the present invention.
[0023] The present invention can fuse multi-type feature information of armored vehicles into feature maps, which can be applied to the deep learning model for armored vehicle status monitoring and diagnosis, thereby improving the accuracy of armored vehicle status monitoring and diagnosis.
[0024] Among them, the feature map described in the present invention refers to a series of two-dimensional plane images composed of the arrangement of vehicle status information. Taking a two-dimensional plane image of N×N pixels as an example, the total number of feature information is N×N = N 2The value of N can be adjusted according to the testability of the specific vehicle. If the vehicle has M types of faults, this series of spectra includes a normal state diagram and M types of state fault diagrams, a total of M + 1 types of images. This series of spectra characterizes the M + 1 states of the vehicle and can be used as the input of a deep learning network for the state monitoring and fault diagnosis of armored vehicles.
[0025] Embodiment 1
[0026] See Figure 1 , the method for constructing a characteristic spectrum for the state monitoring and diagnosis of armored vehicles provided by the present invention may include the following steps:
[0027] Step S101: Obtain vehicle state information for reflecting the state of the armored vehicle, where the vehicle state information includes at least one of engine cylinder compression pressure information, engine speed information, engine vibration information, and transmission vibration information.
[0028] Step S102: Extract features from the vehicle state information for reflecting the state of the armored vehicle to obtain various feature information and feature values of the armored vehicle.
[0029] 1. If the vehicle state information includes the engine cylinder compression pressure information, correspondingly, the various feature information includes the peak value of the engine cylinder compression pressure, the pulse index of the engine cylinder compression pressure, and the waveform index of the engine cylinder compression pressure.
[0030] Among them, the peak value of the engine cylinder compression pressure is the maximum value of the air pressure in the cylinder detected during the process of driving the engine crankshaft to rotate by the starting motor when the engine is not supplying fuel, and is obtained through the following steps: (1) According to the compression pressure signal sequence p(n) of the engine cylinder in one working cycle, determine the absolute value of the engine cylinder compression pressure |p(n)|; (2) Extract the maximum absolute value from the absolute value of the engine cylinder compression pressure |p(n)| As the peak value p of the engine cylinder compression pressure max , specifically
[0031] Among them, the pulse index of the engine cylinder compression pressure is obtained through the following steps: (1) According to the compression pressure signal sequence p(n) of the engine cylinder in one working cycle, determine the absolute value of the engine cylinder compression pressure |p(n)| and the average value of the engine cylinder compression pressure (2) According to the absolute value of the engine cylinder compression pressure |p(n)|, determine the peak value of the engine cylinder compression pressure (3) According to the peak value of the engine cylinder compression pressure and the average value of the engine cylinder compression pressure Determine the pulse index p of the compression pressure of the engine cylinder If , specifically
[0032] Among them, the waveform index of the compression pressure of the engine cylinder is obtained through the following steps: (1) According to the compression pressure signal sequence p(n) of the engine cylinder in one working cycle, determine the absolute average value of the compression pressure of the engine cylinder The mean square value of the compression pressure of the engine cylinder (2) According to the absolute average value of the compression pressure of the engine cylinder and the mean square value of the compression pressure of the engine cylinder Determine the waveform index p of the compression pressure of the engine cylinder Sf , specifically
[0033] Among them, n = 1, 2, …, K, and K is the number of samples in one working cycle of the cylinder
[0034] 2. If the vehicle state information includes the engine speed information, correspondingly, the multiple characteristic information includes the maximum engine speed, the engine speed fluctuation value, and the engine acceleration time
[0035] Among them, the maximum engine speed refers to the maximum speed value that can be reached in the no-load state, and the maximum speed value of the engine in the no-load state can be obtained from the engine speed information
[0036] Among them, the engine speed fluctuation value refers to the difference between the maximum value and the minimum value of the engine speed in the no-load idle state. The maximum speed value and the minimum speed value in the no-load idle state can be obtained from the engine speed information, and then the difference between the two is used as the engine speed fluctuation value
[0037] Among them, the engine acceleration time refers to the time when the driver steps on the accelerator pedal at the maximum speed in the no-load idle state to quickly accelerate the engine speed to the maximum speed. This process time is the engine acceleration time
[0038] 3. If the vehicle state information includes the engine vibration information, correspondingly, the multiple characteristic information includes the engine vibration peak value, the engine vibration peak-to-peak value, the engine vibration effective value, the engine vibration waveform index, and the engine vibration pulse index
[0039] Let the engine vibration signal sequence be V e (n), n = 1, 2, …, K
[0040] Among them, the peak value of the engine vibration is obtained through the following steps: (1) According to the engine vibration signal sequence V e (n), the absolute value of the engine vibration |V e (n)|, the absolute average value of the engine vibration, the maximum and minimum values of the engine vibration, and the root mean square value of the engine vibration are obtained. (2) The maximum absolute value is extracted from the absolute value of the engine vibration |V e (n)| as the peak value V of the engine vibration emax , specifically
[0041] Among them, the peak-to-peak value of the engine vibration is obtained through the following steps: (1) According to the engine vibration signal sequence, the maximum value of the engine vibration and the minimum value are obtained. (2) According to the maximum value of the engine vibration and the minimum value , the peak-to-peak value (or peak-to-peak value) V of the engine vibration is determined eP-P , specifically
[0042] Among them, the effective value of the engine vibration is obtained through the following steps: (1) According to the engine vibration signal sequence, the root mean square value of the engine vibration is obtained. (2) According to the root mean square value of the engine vibration , the effective value V of the engine vibration is determined erms , specifically
[0043] Among them, the waveform index of the engine vibration is obtained through the following steps: (1) According to the engine vibration signal sequence, the absolute average value of the engine vibration and the root mean square value of the engine vibration are obtained. (2) According to the root mean square value of the engine vibration and the absolute average value of the engine vibration , the waveform index V of the engine vibration is determined eSf , specifically
[0044] Among them, the impulse index of the engine vibration is obtained through the following steps: (1) According to the engine vibration signal sequence, the absolute value of the engine vibration |V e (n)| and the absolute average value of the engine vibration are obtained. (2) According to the absolute value of the engine vibration |V e (n)|, the peak value of the engine vibration is obtained, and according to the peak value of the engine vibration and the absolute average value of the engine vibration Determine the engine pulse index V eIf , specifically
[0045] 4. If the vehicle state information includes the transmission vibration information, correspondingly, the multiple characteristic information includes the transmission vibration peak value, the transmission vibration peak-to-peak value, the transmission vibration effective value, the transmission vibration waveform index, and the transmission vibration pulse index.
[0046] Let the transmission vibration signal sequence be V t (n), where n = 1, 2, …, K.
[0047] Among them, the transmission vibration peak value is obtained through the following steps: (1) According to the transmission vibration signal sequence V t (n), obtain the absolute value of the transmission vibration |V t (n)|, (2) Extract the maximum absolute value from the absolute value of the transmission vibration |V t (n)| as the transmission vibration peak value V tmax , specifically
[0048] Among them, the transmission vibration peak-to-peak value is obtained through the following steps: (1) According to the transmission vibration signal sequence V t (n), obtain the maximum value and the minimum value (2) Determine the transmission vibration peak-to-peak value (or called) V and the minimum value tP-P , specifically
[0049] Among them, the transmission vibration effective value is obtained through the following steps: (1) According to the transmission vibration signal sequence V t (n), obtain the mean square value of the transmission vibration (2) Determine the transmission vibration effective value V according to the mean square value of the transmission vibration trms , specifically
[0050] Among them, the transmission vibration waveform index is obtained through the following steps: (1) According to the transmission vibration signal sequence V t (n), obtain the absolute mean value of the transmission vibration and the mean square value of the transmission vibration (2) According to the mean square value of the transmission vibration and the absolute mean value of the transmission vibration Determine the vibration waveform index V of the gearbox tSf , specifically
[0051] Among them, the vibration pulse index of the gearbox is obtained through the following steps: (1) According to the gearbox vibration signal sequence V t (n), obtain the absolute value of gearbox vibration |V t (n)| and the absolute mean value of gearbox vibration (2) According to the absolute value of gearbox vibration |V t (n)|, obtain the peak value of gearbox vibration And according to the peak value of the gearbox vibration and the absolute mean value of the gearbox vibration Determine the pulse index V of the gearbox tIf , specifically
[0052] Step S103: Convert the feature values of multiple feature information into the image gray values of multiple feature information respectively.
[0053] This step includes a normalization operation step and a gray conversion step, specifically as follows:
[0054] (1) Normalization method
[0055] Since the dimensions of different types of feature values are different, and there are several orders of magnitude differences between some types of feature values, it is not conducive to pattern recognition. Therefore, it is necessary to perform a normalization operation on the feature value data of each type of feature information to obtain the normalized feature value data of each type of feature information. The purpose is to remove the dimension and make different feature value data comparable. The normalized data are all distributed between 0 and 1, and the maximum value is 1 and the minimum value is 0.
[0056] Let the i-th type of feature value be x i (j), i = 1, 2,..., N 2 , j = 1, 2,..., J, where J is the number of data samples collected by various sensors.
[0057] The normalization is carried out by the following method:
[0058]
[0059] Among them, is the feature value of the i-th type of feature after normalization. x imax is the maximum value of the i-th type of feature in all M + 1 states, and x imin is the minimum value of the i-th type of feature in all M + 1 states. These two values can be comparison values or empirical values.
[0060] (2) Method for converting feature values into image gray values
[0061] In a grayscale image, the grayscale value range is 0 to 255, and each pixel is one of 256 grayscales between black and white. Therefore, it is necessary to perform grayscale conversion on the normalized feature values of each type of feature information to obtain the image grayscale values of each type of feature information, so as to generate a grayscale image.
[0062] Let the grayscale value be c i (j). The conversion method for converting the normalized feature values of each type of feature information into the image grayscale values of each type of feature information is as follows:
[0063]
[0064] Step S104: Use the image grayscale values of multiple types of feature information to construct a feature map for armored vehicle status monitoring and diagnosis.
[0065] First, arrange the image grayscale values of multiple types of feature information of the armored vehicle according to a preset arrangement method to obtain a two-dimensional array composed of the image grayscale values of multiple types of feature information of the armored vehicle. Taking the 16 feature information obtained in step S102 as an example, arrange the image grayscale values of the 16 feature information obtained in step S103 according to Figure 2 the arrangement shown to obtain a two-dimensional array structure based on the image grayscale of 16 feature information. Among them, Figure 2 C1, C2,... C16 in represent the image grayscale values of 16 feature information respectively. Then map each image grayscale value of the two-dimensional array to an image pixel value, and use an image encoder to save the two-dimensional array as a grayscale image, that is, a feature map. Figure 3 shows the conversion process from grayscale value to feature map.
[0066] In this embodiment, a 4*4 feature map formed by 16 feature information is taken as an example for detailed description. According to actual needs, the number of feature information can be more or less than 16, and correspondingly, the size and shape of the feature map can also be different.
[0067] Embodiment 2
[0068] Refer to Figure 4 , a method for armored vehicle status monitoring and diagnosis provided by the present invention may include the following steps:
[0069] Step S201: Use the feature map construction method for armored vehicle status monitoring and diagnosis in Embodiment 1 to generate a feature map of an armored vehicle.
[0070] Step S202: Analyze the feature map of the armored vehicle by using the trained deep learning model for armored vehicle status monitoring and diagnosis to determine the vehicle status of the armored vehicle, where the status includes a normal status and at least one type of fault status.
[0071] Among them, the trained deep learning model for armored vehicle status monitoring and diagnosis is obtained through the following steps: Use the feature map construction method for armored vehicle status monitoring and diagnosis in Embodiment 1 to process the vehicle status information in the normal status and each type of fault status in the vehicle status information dataset respectively to obtain the normal status feature map and each type of fault status feature map; Use the normal status feature map and each type of fault status feature map to train the deep learning model for armored vehicle status monitoring and diagnosis to obtain the trained deep learning model for armored vehicle status monitoring and diagnosis.
[0072] Similarly, taking the 16 types of feature information in Embodiment 1 as an example, obtain the engine cylinder compression pressure information, engine speed information, engine vibration information, and transmission vibration information from the vehicle status information dataset, and perform feature extraction on the engine cylinder compression pressure information, engine speed information, engine vibration information, and transmission vibration information through the processing method in step S102 of Embodiment 1 to obtain 16 types of feature information and corresponding feature values. Convert the feature values of the 16 types of feature information into corresponding image gray values through the processing method in step S103 of Embodiment 1, and then construct a feature map for armored vehicle status monitoring and diagnosis through the processing method in step S104 of Embodiment 1. Since the aforementioned engine cylinder compression pressure information, engine speed information, engine vibration information, and transmission vibration information contain information in the normal state of the vehicle and information in M types of fault states, where M is greater than or equal to 1, the constructed feature map contains a normal map corresponding to the normal state and fault maps corresponding to each of the M types of fault states. After training the deep learning model with the constructed feature maps, the trained deep learning model can accurately identify M + 1 types of vehicle statuses. The model training process is a common technology and will not be elaborated here. When actually applying the trained deep learning model, the engine cylinder compression pressure information, engine speed information, engine vibration information, and transmission vibration information of any armored vehicle to be diagnosed can be obtained, and a feature map integrating 16 types of feature information can be obtained through steps S102 to S104 of Embodiment 1. Input this feature map into the trained deep learning model for status recognition, and the normal or specific fault type of the armored vehicle can be determined.
[0073] This embodiment takes a 4*4 feature map formed by 16 feature information as an example for detailed description. According to actual needs, the number of feature information can be more or less than 16, and correspondingly, the size and shape of the feature map can also be different.
[0074] The present invention constructs a feature map based on one-dimensional vehicle state data, enabling the high learning ability and high recognition accuracy of a deep learning model to be applied to the state monitoring and diagnosis of armored vehicles. Compared with traditional machine learning models that use one-dimensional data sequences for monitoring and diagnosis, the accuracy of state monitoring and diagnosis of armored vehicles is improved.
[0075] Although the present invention has been described in detail above, the present invention is not limited thereto, and those skilled in the art of the present technology can make various modifications according to the principles of the present invention. Therefore, all modifications made according to the principles of the present invention should be understood to fall within the protection scope of the present invention.
Claims
1. A method for constructing a characteristic map for the status monitoring and diagnosis of armored vehicles, characterized in that, The method includes: Obtaining vehicle state information for reflecting the state of an armored vehicle; Performing feature extraction on the vehicle state information for reflecting the state of an armored vehicle to obtain various feature information and feature values of the armored vehicle; Converting the feature values of various feature information into image gray values of various feature information respectively; Using the image gray values of various feature information to construct a feature map for monitoring and diagnosing the state of an armored vehicle.
2. The method according to claim 1, wherein The vehicle state information includes at least one of engine cylinder compression pressure information, engine speed information, engine vibration information, and transmission vibration information.
3. The method according to claim 2, characterized in that When the vehicle state information includes the engine cylinder compression pressure information, the various feature information includes the peak value of the engine cylinder compression pressure, the pulse index of the engine cylinder compression pressure, and the waveform index of the engine cylinder compression pressure. The performing feature extraction on the vehicle state information for reflecting the state of an armored vehicle to obtain various feature information and feature values of the armored vehicle includes: Determining the absolute value of the engine cylinder compression pressure, the average value of the engine cylinder compression pressure, the absolute average value of the engine cylinder compression pressure, and the mean square value of the engine cylinder compression pressure according to the compression pressure signal sequence of the engine cylinder in one working cycle; Extracting the maximum absolute value from the absolute values of the engine cylinder compression pressure as the peak value of the engine cylinder compression pressure; Determining the pulse index of the engine cylinder compression pressure according to the peak value of the engine cylinder compression pressure and the average value of the engine cylinder compression pressure; Determining the waveform index of the engine cylinder compression pressure according to the absolute average value of the engine cylinder compression pressure and the mean square value of the engine cylinder compression pressure.
4. The method according to claim 2, wherein When the vehicle state information includes the engine speed information, the various feature information includes the maximum engine speed, the engine speed fluctuation value, and the engine acceleration time. The performing feature extraction on the vehicle state information for reflecting the state of an armored vehicle to obtain various feature information and feature values of the armored vehicle includes: Obtaining the maximum speed value of the engine in the no-load state, the maximum speed value and the minimum speed value in the no-load idle state, and the engine acceleration time for the engine to accelerate to the maximum speed when the throttle is depressed at the maximum speed in the no-load idle state from the engine speed information; Obtaining the engine speed fluctuation value according to the maximum speed value and the minimum speed value of the engine in the no-load idle state.
5. The method according to claim 2, characterized in that, When the vehicle state information includes the engine vibration information, the various feature information includes the peak value of the engine vibration, the peak-to-peak value of the engine vibration, the effective value of the engine vibration, the waveform index of the engine vibration, and the pulse index of the engine vibration. The performing feature extraction on the vehicle state information for reflecting the state of an armored vehicle to obtain various feature information and feature values of the armored vehicle includes: Obtaining the absolute value of the engine vibration, the absolute average value of the engine vibration, the maximum and minimum values of the engine vibration, and the mean square value of the engine vibration according to the engine vibration signal sequence; Extracting the maximum absolute value from the absolute values of the engine vibration as the peak value of the engine vibration; Determine the peak-to-peak value of the engine vibration based on the maximum and minimum values of the engine vibration; Determine the effective value of the engine vibration based on the mean square value of the engine vibration; Determine the waveform index of the engine vibration based on the mean square value of the engine vibration and the absolute mean value of the engine vibration; Determine the impulse index of the engine vibration based on the peak value of the engine vibration and the absolute mean value of the engine vibration; 6. The method according to claim 2, characterized in that, When the vehicle state information includes the transmission vibration information, the multiple characteristic information includes the peak value of the transmission vibration, the peak-to-peak value of the transmission vibration, the effective value of the transmission vibration, the waveform index of the transmission vibration, and the impulse index of the transmission vibration. The feature extraction of multiple vehicle state information used to reflect the state of the armored vehicle to obtain the multiple characteristic information and characteristic values of the armored vehicle includes: Obtain the absolute value of the transmission vibration, the absolute mean value of the transmission vibration, the maximum and minimum values of the transmission vibration, and the mean square value of the transmission vibration according to the transmission vibration signal sequence; Extract the maximum absolute value from the absolute values of the transmission vibration as the peak value of the transmission vibration; Determine the peak-to-peak value of the transmission vibration based on the maximum and minimum values of the transmission vibration; Determine the effective value of the transmission vibration based on the mean square value of the transmission vibration; Determine the waveform index of the transmission vibration based on the mean square value of the transmission vibration and the absolute mean value of the transmission vibration; Determine the transmission pulse index based on the peak value of the transmission vibration and the absolute mean value of the transmission vibration; 7. The method according to any one of claims 1-6, characterized in that The conversion of the characteristic values of multiple characteristic information into the image gray values of multiple characteristic information includes: Perform normalization processing on the characteristic values of each characteristic information to obtain the normalized characteristic values of each characteristic information; Perform gray conversion on the normalized characteristic values of each characteristic information to obtain the image gray values of each characteristic information.
8. The method according to any one of claims 1 to 6, characterized in that, The construction of a characteristic map for armored vehicle state monitoring and diagnosis using the image gray values of multiple characteristic information includes: Arrange the image gray values of multiple characteristic information of the armored vehicle according to a preset arrangement method to obtain a two-dimensional array composed of the image gray values of multiple characteristic information of the armored vehicle; Map each image gray value of the two-dimensional array to the pixel value of the image, and use an image encoder to save the two-dimensional array as a gray image as the characteristic map.
9. A method for monitoring and diagnosing the status of an armored vehicle, characterized in that, The method includes: Generate a characteristic map of an armored vehicle using the characteristic map construction method for armored vehicle state monitoring and diagnosis according to any one of claims 1-8; Use a trained deep learning model for armored vehicle state monitoring and diagnosis to analyze the characteristic map of the armored vehicle to determine the vehicle state of the armored vehicle, and the state includes a normal state and at least one type of fault state.
10. The armored vehicle status monitoring and diagnosis method according to claim 9, characterized in that, Obtain the trained deep learning model for armored vehicle state monitoring and diagnosis through the following steps: Use the characteristic map construction method for armored vehicle state monitoring and diagnosis according to any one of claims 1-8 to process the vehicle state information in the normal state and each type of fault state in the dataset respectively to obtain a normal state characteristic map and each type of fault state characteristic map; Using the normal state feature map and the feature maps of each type of fault state, train the deep learning model for armored vehicle condition monitoring and diagnosis to obtain a trained deep learning model for armored vehicle condition monitoring and diagnosis.