Detection method and system for wet brake of engineering vehicle drive axle

Abstract

The invention provides a detection method and system for a wet brake of an engineering vehicle drive axle. The method comprises the steps that basic feature information of a processed original image data set is extracted based on an MDSM network structure; generating a dynamic image content according to the processed original image data set, and performing feature recombination on the basic feature information to obtain multi-scale feature information; mapping the multi-scale feature to the processed original image data set to obtain a mapping position, and predicting a distance difference between the mapping position and a prediction frame to output a prediction result; identifying a frequency component in the processed vibration signal; and calculating a peak average amplitude based on the signal amplitude, and judging whether the peak average amplitude is within a preset amplitude range or not to obtain a balance result of the wet brake. According to the method, the multi-scale feature information on the wet brake disc can be obtained, the defect feature position can be obtained, and whether defects exist on the surface of the wet brake or not can be obtained according to the vibration signals.

Description

Technical Field

[0001] The present invention relates to the technical field of brake detection, and in particular to a detection method and system for a wet brake of a drive axle of an engineering vehicle. Background Art

[0002] The use of wet brakes on drive axles addresses three key pain points: First, construction vehicles often operate on slopes, and traditional dry brakes are prone to overheating and failure during continuous downhill descents; second, the dusty environment in mining areas causes significant wear on dry brakes; and third, the heavy weight of construction vehicles requires extremely high braking torque. Multi-disc wet brakes address thermal decay through oil dissipation, a sealed design to prevent contamination, and a compact structure to provide high braking force.

[0003] A wet brake is a braking system that operates by completely immersing the brake friction elements in fluid. Compared to common "dry" brakes (such as automotive disc or drum brakes exposed to air), it has significantly different operating environments and characteristics.

[0004] Among existing technologies, most rely on manual visual inspection. However, due to the diversity of wet brake defects and quality issues, manual subjective judgment inspection is not only time-consuming and labor-intensive, but also affects inspection efficiency and results. Traditional machine vision inspection methods rely on manually designed features and cannot effectively detect complex defects. Summary of the Invention

[0005] Based on this, the purpose of the present invention is to provide a detection method and system for wet brakes of engineering vehicle drive axles to solve the above-mentioned deficiencies in the prior art.

[0006] In a first aspect, the present invention provides a method for detecting a wet brake of a drive axle of an engineering vehicle, the method comprising:

[0007] collecting a raw image data set of a wet brake, acquiring a vibration signal of a brake rotor of the wet brake when the wet brake is in operation, and processing the raw image data set and the vibration signal;

[0008] Based on the MDSM network structure, multiple dilation rate convolution residual blocks are connected in series to extract the basic feature information of the processed original image dataset;

[0009] generating dynamic image content according to the processed original image data set, and performing feature reorganization on the basic feature information based on an upsampling kernel and the dynamic image content to obtain multi-scale feature information;

[0010] Mapping the multi-scale features to the processed original image dataset based on the FCOS detection network to obtain a mapping position, and predicting the distance difference between the mapping position and the prediction box to output a prediction result;

[0011] identifying frequency components in the processed vibration signal, and obtaining a signal amplitude of the wet brake based on all the frequency components;

[0012] A peak-to-average amplitude is calculated based on the signal amplitude, and it is determined whether the peak-to-average amplitude is within a preset amplitude range, so as to obtain a balancing result of the wet brake.

[0013] Compared with the prior art, the present invention has the following advantages: basic feature information with defects of a wet brake can be extracted through the MDSM network structure and by connecting multiple void rate convolution residual blocks in series. Multi-scale feature information is obtained by reorganizing the information through an upsampling kernel, and by mapping it to the original image, the defect features and defect feature locations of the wet brake can be obtained. The signal amplitude can be obtained by identifying the vibration signal of the wet brake, and the balance result of the wet brake can be obtained through the signal amplitude, so as to determine whether the thickness of the wet brake is uneven and whether the surface of the wet brake has defects.

[0014] Furthermore, the steps of collecting a raw image dataset of a wet brake, acquiring a vibration signal of a brake rotor of the wet brake when the wet brake is in operation, and processing the raw image dataset and the vibration signal include:

[0015] Collecting images of the wet brake taken from an industrial camera to obtain a raw image dataset;

[0016] collecting a vibration signal of the brake rotor of the wet brake during operation, which is collected by a piezoelectric acceleration sensor, and performing denoising on the vibration signal;

[0017] The original image datasets are sequentially integrated and cropped, and information annotation and data enhancement are sequentially performed on the original image datasets.

[0018] Furthermore, the step of extracting basic feature information of the processed original image dataset by connecting multiple dilated convolution residual blocks in series based on the MDSM network structure includes:

[0019] Defect feature information of the original image data set after extraction and processing based on the MDSM network structure;

[0020] A plurality of dilated convolution residual blocks are connected in series to obtain an enlarged receptive field, and basic feature information is extracted based on the enlarged receptive field and the defect feature information.

[0021] Furthermore, the expression for increasing the receptive field is:

[0022]

[0023] Where n out Indicates the output feature map size, n in Represents the input feature map size, k represents the convolution kernel size, p represents the size of the additional padding value, and s represents the step size.

[0024] Furthermore, the step of generating dynamic image content according to the processed original image data set, and performing feature reorganization on the basic feature information based on the upsampling kernel and the dynamic image content to obtain multi-scale feature information includes:

[0025] combining the processed raw image data sets to obtain dynamic image content;

[0026] The basic feature information is reorganized by combining the upsampling kernel and the dynamic image content between adjacent dilation rate convolution residual blocks to obtain the multi-scale feature information.

[0027] Furthermore, the step of predicting the distance difference between the mapping position and the prediction box to output the prediction result includes:

[0028] Using the mapped positions as training samples, and training the FCOS detection network based on the training samples to obtain a trained FCOS detection network;

[0029] The four-dimensional vector of the mapping position is predicted based on the trained FCOS detection network to obtain the distance difference between the mapping position and the prediction box, and the prediction result is output. The expression of the center score of the prediction result is:

[0030]

[0031] Wherein, Center-ness score represents the center score, min(l,r) represents the minimum value of l and r, max(l,r) represents the maximum value of l and r, min(t,b) represents the minimum value of t and b in the up and down directions, max(t,b) represents the maximum value of t and b in the up and down directions, l represents the distance from the predicted point to the left boundary of the target bounding box, r represents the distance from the predicted point to the right boundary of the target bounding box, t represents the distance from the predicted point to the upper boundary of the target bounding box, and b represents the distance from the predicted point to the lower boundary of the target bounding box.

[0032] Furthermore, the step of identifying the frequency components in the processed vibration signal and obtaining the signal amplitude of the wet brake based on all the frequency components includes:

[0033] Decomposing the processed vibration signal based on ICEEMDAN decomposition to obtain a decomposed vibration signal, and identifying frequency components in the decomposed vibration signal;

[0034] The decomposed vibration signal is reconstructed based on the VMD decomposition and the frequency components to obtain a signal amplitude.

[0035] In a second aspect, the present invention further provides a detection system for a wet brake of a drive axle of an engineering vehicle, the system comprising:

[0036] a collection and processing module, configured to collect an original image data set of a wet brake, acquire a vibration signal of a brake rotor of the wet brake when the wet brake is in operation, and process the original image data set and the vibration signal;

[0037] A series extraction module, which is based on the MDSM network structure and connects multiple dilation rate convolution residual blocks in series to extract basic feature information of the processed original image dataset;

[0038] a generating and recombining module, configured to generate dynamic image content according to the processed original image data set, and perform feature recombining on the basic feature information based on an upsampling kernel and the dynamic image content to obtain multi-scale feature information;

[0039] A mapping prediction module is used to map the multi-scale features onto the processed original image dataset based on the FCOS detection network to obtain a mapping position, and predict the distance difference between the mapping position and the prediction box to output a prediction result;

[0040] an identification and generation module, configured to identify frequency components in the processed vibration signal and obtain a signal amplitude of the wet brake based on all the frequency components;

[0041] The calculation and judgment module is used to calculate the peak-to-average amplitude based on the signal amplitude and judge whether the peak-to-average amplitude is within a preset amplitude range to obtain a balancing result of the wet brake.

[0042] In a third aspect, the present invention also provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned detection method for wet brakes of drive axles of engineering vehicles when executing the computer program.

[0043] In a third aspect, the present invention further provides a storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-mentioned detection method for the wet brake of the drive axle of an engineering vehicle. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 The present invention is a method for detecting a wet brake of a drive axle of an engineering vehicle in a first embodiment;

[0045] Figure 2 A detection system for a wet brake of a drive axle of an engineering vehicle according to a second embodiment of the present invention;

[0046] Figure 3 This is the electronic device in the second embodiment of the present invention.

[0047] Description of main component symbols:

[0048] 10. Collection and processing module; 20. Series extraction module; 30. Generation and reorganization module; 40. Mapping prediction module; 50. Recognition and generation module; 60. Calculation and judgment module;

[0049] 70. Bus; 71. Processor; 72. Memory; 73. Communication interface.

[0050] The following specific embodiments will further illustrate the present invention in conjunction with the above-mentioned drawings. DETAILED DESCRIPTION

[0051] To facilitate understanding of the present invention, the present invention will be described more fully below with reference to the accompanying drawings. The drawings illustrate several embodiments of the present invention. However, the present invention may be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to provide a more thorough and comprehensive understanding of the present invention.

[0052] It should be noted that when an element is referred to as being "fixed to" another element, it may be directly on the other element or there may be an intermediate element. When an element is referred to as being "connected to" another element, it may be directly connected to the other element or there may be an intermediate element. The terms "vertical," "horizontal," "left," "right," and similar expressions used herein are for illustrative purposes only.

[0053] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this invention pertains. The terms used in this specification of the present invention are for the purpose of describing specific embodiments only and are not intended to limit the present invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0054] Example 1

[0055] See also Figure 1 , which shows a detection method for a wet brake of a drive axle of an engineering vehicle in an embodiment of the present invention, the method includes steps S1 to S6:

[0056] S1, collecting an original image dataset of a wet brake, acquiring a vibration signal of a brake rotor of the wet brake when the wet brake is in operation, and processing the original image dataset and the vibration signal;

[0057] Specifically, the step S1 includes steps S11 to S13:

[0058] S11, collecting images of the wet brake taken from an industrial camera to obtain an original image dataset;

[0059] S12, collecting a vibration signal of the brake rotor of the wet brake during operation, which is collected by a piezoelectric acceleration sensor, and denoising the vibration signal;

[0060] S13, sequentially integrating and cropping the original image datasets, and sequentially performing information annotation and data enhancement on the original image datasets;

[0061] It can be understood that an industrial camera is used to capture images of the wet brake surface at multiple angles to obtain a raw image dataset of the wet brake, and a piezoelectric accelerometer is used to collect vibration signals of the brake rotor of the wet brake during operation. In a specific implementation, a vehicle can be started in a laboratory or a repair shop, and then the brake system can be operated, and the vibration signals of the brake rotor of the wet brake can be collected using the piezoelectric accelerometer.

[0062] It is worth noting that after obtaining the original image dataset, multi-angle images can be integrated to obtain a larger image that can include information on all directions of the wet brake disc. The integrated image has a width of 20,000 resolution and a height of 4,000 resolution, and is cropped to the required size. In this embodiment, the main means of data enhancement is to flip and scale the image.

[0063] S2, based on the MDSM network structure, connects multiple dilation rate convolution residual blocks in series to extract the basic feature information of the processed original image dataset;

[0064] Specifically, step S3 includes steps S31 to S32:

[0065] S31, extracting defect feature information of the processed original image data set based on the MDSM network structure;

[0066] S32, connecting multiple dilated convolution residual blocks in series to obtain an enlarged receptive field, and extracting basic feature information based on the enlarged receptive field and the defect feature information;

[0067] The expression for increasing the receptive field is:

[0068]

[0069] Where n out Indicates the output feature map size, n in Represents the input feature map size, k represents the convolution kernel size, p represents the size of the additional padding value, and s represents the step size;

[0070] It can be understood that in this embodiment, the MDSM network structure includes three residual networks, four semantic supplementation branches, and four feature map outputs of different sizes. The MDSM network structure can extract the defect feature information of the original image dataset, and by using the void convolution residual block in a serial manner to gradually increase the network receptive field, the basic feature information of the defects in the wet brake can be extracted, thereby improving the effectiveness of feature extraction.

[0071] S3, generating dynamic image content according to the processed original image data set, and performing feature reorganization on the basic feature information based on the upsampling kernel and the dynamic image content to obtain multi-scale feature information;

[0072] Specifically, step S3 includes steps S31 to S32:

[0073] S31, combining the processed original image data sets to obtain dynamic image content;

[0074] S32, performing feature recombination on the basic feature information by combining the upsampling kernel and the dynamic image content between adjacent dilation rate convolution residual blocks to obtain the multi-scale feature information;

[0075] It can be understood that by linearly and dynamically combining images from multiple angles in the original image dataset, the dynamic image content of the wet brake can be obtained. Between adjacent dilated convolution residual blocks, upsampling kernels are generated based on the input dynamic image content to achieve feature recombination, allowing the network to have a sufficient receptive field while retaining a large amount of spatial information. This enables multi-scale prediction output and obtain multi-scale feature information.

[0076] S4, mapping the multi-scale features to the processed original image dataset based on the FCOS detection network to obtain a mapping position, and predicting the distance difference between the mapping position and the prediction box to output a prediction result;

[0077] Specifically, step S4 includes steps S41 to S42:

[0078] S41, using the mapped positions as training samples, and training the FCOS detection network based on the training samples to obtain a trained FCOS detection network;

[0079] S42, predicting a four-dimensional vector of the mapping position based on the trained FCOS detection network to obtain a distance difference between the mapping position and the prediction box, and outputting a prediction result;

[0080] It can be understood that the FCOS detection network is a feature point regression-based detection network without anchor boxes, including classification branches, center branches, and regression branches. The FCOS detection network can use the mapping position as a training sample to avoid the access of anchor points. The trained FCOS detection network predicts the four-dimensional vector of the mapping position, which represents the distance difference between the current position and the left, top, right, and bottom of the predicted box, thereby outputting the position result of the defect feature.

[0081] It is worth noting that the FCOS algorithm uses a pixel-by-pixel regression strategy, which improves the recall rate while generating low-quality predicted bounding boxes that require a large offset from the center. In this embodiment, a center score is learned for each position by using a center branch to calculate the relative distance between the center of the true value box and the center score. The accuracy of the prediction result can be obtained through the center score.

[0082] The expression of the center score of the prediction result is:

[0083]

[0084] Wherein, Center-ness score represents the center score, min(l,r) represents the minimum value of l and r, max(l,r) represents the maximum value of l and r, min(t,b) represents the minimum value of t and b in the up and down directions, max(t,b) represents the maximum value of t and b in the up and down directions, l represents the distance from the predicted point to the left boundary of the target bounding box, r represents the distance from the predicted point to the right boundary of the target bounding box, t represents the distance from the predicted point to the upper boundary of the target bounding box, and b represents the distance from the predicted point to the lower boundary of the target bounding box.

[0085] S5, identifying frequency components in the processed vibration signal, and obtaining a signal amplitude of the wet brake based on all the frequency components;

[0086] Specifically, step S5 includes steps S51 to S52:

[0087] S51, decomposing the processed vibration signal based on ICEEMDAN decomposition to obtain a decomposed vibration signal, and identifying frequency components in the decomposed vibration signal;

[0088] S52, reconstructing the decomposed vibration signal based on VMD decomposition and the frequency component to obtain a signal amplitude;

[0089] It can be understood that the collected braking vibration signal is decomposed using the ICEEMDAN algorithm to obtain multiple decomposed vibration signals, and by identifying the frequency components in the decomposed vibration signal, the frequency components and time characteristics of the signal can be obtained, and the frequency components are reconstructed through VMD decomposition to obtain the signal amplitude.

[0090] S6, calculating a peak-to-average amplitude based on the signal amplitude, and determining whether the peak-to-average amplitude is within a preset amplitude range, so as to obtain a balancing result of the wet brake;

[0091] It is understood that by calculating the peak-to-average amplitude, the current operating mode of the wet brake can be determined from the peak-to-average amplitude. If a mismatching mode occurs, for example, the amplitude corresponding to normal modes such as braking and releasing does not correspond to the amplitude of the current mode, a balance result of the wet brake can be obtained, indicating that the surface thickness difference of the wet brake is large and not smooth, which may cause defects in the wet brake during long-term operation, thereby further determining the surface defects of the wet brake.

[0092] In summary, the wet brake detection method for an engineering vehicle drive axle in the above-mentioned embodiment of the present invention can extract basic feature information of a wet brake with defects through an MDSM network structure and a series connection of multiple void rate convolution residual blocks. The information is then reorganized through an upsampling kernel to obtain multi-scale feature information. By mapping this information onto the original image, the defect features and defect feature locations of the wet brake can be obtained. The signal amplitude can be obtained by identifying the vibration signal of the wet brake, and the balance result of the wet brake can be obtained from the signal amplitude, thereby determining whether the wet brake has uneven thickness and further whether the surface of the wet brake has defects.

[0093] Example 2

[0094] The second embodiment of the present invention also provides a detection system for wet brakes of a drive axle of an engineering vehicle, see Figure 2, shown is a detection system for wet brakes of a drive axle of an engineering vehicle in a second embodiment of the present invention, the system comprising:

[0095] a collection and processing module 10 for collecting a raw image dataset of a wet brake, acquiring a vibration signal of a brake rotor of the wet brake when the wet brake is in operation, and processing the raw image dataset and the vibration signal;

[0096] A series extraction module 20 is configured to connect multiple dilation rate convolution residual blocks in series based on the MDSM network structure to extract basic feature information of the processed original image dataset;

[0097] a generating and recombining module 30 for generating dynamic image content according to the processed original image data set, and performing feature recombining on the basic feature information based on an upsampling kernel and the dynamic image content to obtain multi-scale feature information;

[0098] A mapping prediction module 40 is configured to map the multi-scale features onto the processed original image dataset based on the FCOS detection network to obtain a mapping position, and predict a distance difference between the mapping position and a prediction box to output a prediction result;

[0099] an identification and generation module 50 for identifying frequency components in the processed vibration signal and obtaining a signal amplitude of the wet brake based on all the frequency components;

[0100] The calculation and judgment module 60 is configured to calculate a peak-to-average amplitude based on the signal amplitude and determine whether the peak-to-average amplitude is within a preset amplitude range, so as to obtain a balancing result of the wet brake.

[0101] In some optional embodiments, the collection and processing module 10 includes:

[0102] a collecting unit, configured to collect images of the wet brake taken from an industrial camera to obtain an original image data set;

[0103] an acquisition unit, configured to acquire a vibration signal of the brake rotor of the wet brake during operation, acquired by a piezoelectric acceleration sensor, and to denoise the vibration signal;

[0104] The integration unit is used to sequentially integrate and crop the original image datasets, and sequentially perform information annotation and data enhancement on the original image datasets.

[0105] In some optional embodiments, the serial extraction module 20 includes:

[0106] A first extraction unit, configured to extract defect feature information of the processed original image data set based on the MDSM network structure;

[0107] A second extraction unit is configured to connect in series a plurality of dilated convolution residual blocks to obtain an enlarged receptive field, and extract basic feature information based on the enlarged receptive field and the defect feature information;

[0108] The expression for increasing the receptive field is:

[0109]

[0110] Where n out Indicates the output feature map size, n in Represents the input feature map size, k represents the convolution kernel size, p represents the size of the additional padding value, and s represents the step size.

[0111] In some optional embodiments, the generating and recombining module 30 includes:

[0112] a combining unit, configured to combine the processed original image data sets to obtain dynamic image content;

[0113] A recombining unit is used to perform feature recombining on the basic feature information by combining the upsampling kernel and the dynamic image content between adjacent dilation rate convolution residual blocks to obtain the multi-scale feature information.

[0114] In some optional embodiments, the mapping prediction module 40 includes:

[0115] a training unit, configured to use the mapped positions as training samples and train the FCOS detection network based on the training samples to obtain a trained FCOS detection network;

[0116] A prediction unit is configured to predict the four-dimensional vector of the mapping position based on the trained FCOS detection network to obtain a distance difference between the mapping position and the prediction box, and output a prediction result, wherein the expression of the center score of the prediction result is:

[0117]

[0118] Wherein, Center-ness score represents the center score, min(l,r) represents the minimum value of l and r, max(l,r) represents the maximum value of l and r, min(t,b) represents the minimum value of t and b in the up and down directions, max(t,b) represents the maximum value of t and b in the up and down directions, l represents the distance from the predicted point to the left boundary of the target bounding box, r represents the distance from the predicted point to the right boundary of the target bounding box, t represents the distance from the predicted point to the upper boundary of the target bounding box, and b represents the distance from the predicted point to the lower boundary of the target bounding box.

[0119] In some optional embodiments, the identification generation module 50 includes:

[0120] a decomposition unit, configured to decompose the processed vibration signal based on ICEEMDAN decomposition to obtain a decomposed vibration signal, and identify frequency components in the decomposed vibration signal;

[0121] The reconstruction unit is configured to reconstruct the decomposed vibration signal based on the VMD decomposition and the frequency component to obtain a signal amplitude.

[0122] The functions or operation steps implemented when the above modules and units are executed are substantially the same as those in the above method embodiments and will not be repeated here.

[0123] The detection system for wet brakes of engineering vehicle drive axles provided in the embodiments of the present invention has the same implementation principle and technical effects as those of the aforementioned method embodiments. For the sake of brief description, for matters not mentioned in the system embodiment, reference may be made to the corresponding contents in the aforementioned method embodiments.

[0124] Example 3

[0125] The present invention also provides an electronic device, see Figure 3 , shown is an electronic device in a third embodiment of the present invention.

[0126] The electronic device may include a processor 71 and a memory 72 storing computer program instructions.

[0127] Specifically, the processor 71 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the present application.

[0128] Among them, the memory 72 may include a large-capacity memory for data or instructions. By way of example and not limitation, the memory 72 may include a hard disk drive (HDD), a floppy disk drive, a solid-state drive (SSD), a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive, or a combination of two or more of these. Where appropriate, the memory 72 may include a removable or non-removable (or fixed) medium. Where appropriate, the memory 72 may be inside or outside the data processing device. In a specific embodiment, the memory 72 is a non-volatile memory. In a specific embodiment, the memory 72 includes a read-only memory (ROM) and a random access memory (RAM). Where appropriate, the ROM may be a mask-programmed ROM, a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable PROM (EEPROM), an electrically alterable ROM (EAROM) or a flash memory (FLASH), or a combination of two or more of these. Under appropriate circumstances, the RAM can be a static random access memory (SRAM) or a dynamic random access memory (DRAM), where the DRAM can be a fast page mode dynamic random access memory (FPMDRAM), an extended data output dynamic random access memory (EDODRAM), a synchronous dynamic random access memory (SDRAM), etc.

[0129] The memory 72 may be used to store or cache various data files required for processing and / or communication, as well as possible computer program instructions executed by the processor 71 .

[0130] The processor 71 reads and executes the computer program instructions stored in the memory 72 to implement the closed space environment control method of the first embodiment.

[0131] In some embodiments, the electronic device may further include a communication interface 73 and a bus 70. Figure 3 As shown, the processor 71, the memory 72, and the communication interface 73 are connected via a bus 70 and communicate with each other.

[0132] The communication interface 73 is used to implement communication between the various modules, devices, units and / or equipment in this application. The communication interface 73 can also implement data communication with other components such as: external devices, image / data acquisition equipment, databases, external storage, and image / data processing workstations.

[0133] The bus 70 includes hardware, software, or both, and couples the components of the device to each other. The bus 70 includes, but is not limited to, at least one of the following: a data bus, an address bus, a control bus, an expansion bus, and a local bus. By way of example and not limitation, bus 70 may include an Accelerated Graphics Port (AGP) or other graphics bus, an Extended Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a Hyper Transport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an InfiniBand interconnect, a Low Pin Count (LPC) bus, a memory bus, a Micro Channel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local Bus (VLB) bus, or other suitable buses, or a combination of two or more of the above. Bus 70 may include one or more buses, where appropriate. Although this application describes and illustrates a particular bus, this application contemplates any suitable bus or interconnect.

[0134] The electronic device can obtain the closed space environment control system and execute the closed space environment control method of the first embodiment.

[0135] In addition, in conjunction with the enclosed space environment control method in the first embodiment, the present application may provide a storage medium for implementation. The storage medium stores computer program instructions; when the computer program instructions are executed by a processor, the enclosed space environment control method in the first embodiment is implemented.

[0136] Throughout this specification, reference to terms such as "one embodiment," "some embodiments," "examples," "specific examples," or "some examples" means that a specific feature, structure, material, or characteristic described in conjunction with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0137] The above-described embodiments merely illustrate several implementations of the present invention, and while their descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art would be able to make numerous variations and improvements without departing from the spirit of the present invention, all of which fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be determined by the appended claims.

Claims

1. A method for detecting a wet brake of a drive axle of an engineering vehicle, characterized in that: The method comprises: collecting a raw image data set of a wet brake, acquiring a vibration signal of a brake rotor of the wet brake when the wet brake is in operation, and processing the raw image data set and the vibration signal; Based on the MDSM network structure, multiple dilation rate convolution residual blocks are connected in series to extract the basic feature information of the processed original image dataset; generating dynamic image content according to the processed original image data set, and performing feature reorganization on the basic feature information based on an upsampling kernel and the dynamic image content to obtain multi-scale feature information; Mapping the multi-scale features to the processed original image dataset based on the FCOS detection network to obtain a mapping position, and predicting the distance difference between the mapping position and the prediction box to output a prediction result; identifying frequency components in the processed vibration signal, and obtaining a signal amplitude of the wet brake based on all the frequency components; A peak-to-average amplitude is calculated based on the signal amplitude, and it is determined whether the peak-to-average amplitude is within a preset amplitude range, so as to obtain a balancing result of the wet brake.

2. The detection method for wet brakes of engineering vehicle drive axles according to claim 1, characterized in that: The steps of collecting an original image dataset of a wet brake, acquiring a vibration signal of a brake rotor of the wet brake when the wet brake is in operation, and processing the original image dataset and the vibration signal include: collecting images of the wet brake taken from an industrial camera to obtain a raw image dataset; collecting a vibration signal of the brake rotor of the wet brake during operation, which is collected by a piezoelectric acceleration sensor, and performing denoising on the vibration signal; The original image datasets are sequentially integrated and cropped, and information annotation and data enhancement are performed on the original image datasets sequentially.

3. The detection method for wet brakes of engineering vehicle drive axles according to claim 1, characterized in that: The step of extracting basic feature information of the processed original image dataset based on the MDSM network structure and connecting multiple dilated convolution residual blocks in series includes: Defect feature information of the original image data set after extraction and processing based on the MDSM network structure; A plurality of dilated convolution residual blocks are connected in series to obtain an enlarged receptive field, and basic feature information is extracted based on the enlarged receptive field and the defect feature information.

4. The detection method for wet brakes of engineering vehicle drive axles according to claim 3, characterized in that: The expression for increasing the receptive field is: Where n out Indicates the output feature map size, n in Represents the input feature map size, k represents the convolution kernel size, p represents the size of the additional padding value, and s represents the step size.

5. The detection method for wet brakes of engineering vehicle drive axles according to claim 1, characterized in that: The step of generating dynamic image content according to the processed original image data set, and performing feature reorganization on the basic feature information based on the upsampling kernel and the dynamic image content to obtain multi-scale feature information includes: combining the processed raw image data sets to obtain dynamic image content; The basic feature information is reorganized by combining the upsampling kernel and the dynamic image content between adjacent dilation rate convolution residual blocks to obtain the multi-scale feature information.

6. The detection method for wet brakes of engineering vehicle drive axles according to claim 1, characterized in that: The step of predicting the distance difference between the mapping position and the prediction frame to output the prediction result includes: Using the mapped positions as training samples, and training the FCOS detection network based on the training samples to obtain a trained FCOS detection network; The four-dimensional vector of the mapping position is predicted based on the trained FCOS detection network to obtain the distance difference between the mapping position and the prediction box, and the prediction result is output. The expression of the center score of the prediction result is: Wherein, Center-ness score represents the center score, min(l,r) represents the minimum value of l and r, max(l,r) represents the maximum value of l and r, min(t,b) represents the minimum value of t and b in the up and down directions, max(t,b) represents the maximum value of t and b in the up and down directions, l represents the distance from the predicted point to the left boundary of the target bounding box, r represents the distance from the predicted point to the right boundary of the target bounding box, t represents the distance from the predicted point to the upper boundary of the target bounding box, and b represents the distance from the predicted point to the lower boundary of the target bounding box.

7. The detection method for wet brakes of engineering vehicle drive axles according to claim 1, characterized in that: The step of identifying the frequency components in the processed vibration signal and obtaining the signal amplitude of the wet brake based on all the frequency components includes: Decomposing the processed vibration signal based on ICEEMDAN decomposition to obtain a decomposed vibration signal, and identifying frequency components in the decomposed vibration signal; The decomposed vibration signal is reconstructed based on the VMD decomposition and the frequency components to obtain a signal amplitude.

8. A detection system for wet brakes of engineering vehicle drive axles, characterized in that: The system comprises: a collection and processing module, configured to collect an original image data set of a wet brake, acquire a vibration signal of a brake rotor of the wet brake when the wet brake is in operation, and process the original image data set and the vibration signal; A series extraction module, which is based on the MDSM network structure and connects multiple dilation rate convolution residual blocks in series to extract basic feature information of the processed original image dataset; a generating and recombining module, configured to generate dynamic image content according to the processed original image data set, and perform feature recombining on the basic feature information based on an upsampling kernel and the dynamic image content to obtain multi-scale feature information; A mapping prediction module is used to map the multi-scale features onto the processed original image dataset based on the FCOS detection network to obtain a mapping position, and predict the distance difference between the mapping position and the prediction box to output a prediction result; an identification and generation module, configured to identify frequency components in the processed vibration signal and obtain a signal amplitude of the wet brake based on all the frequency components; The calculation and judgment module is used to calculate the peak-to-average amplitude based on the signal amplitude and judge whether the peak-to-average amplitude is within a preset amplitude range to obtain a balancing result of the wet brake.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the method for detecting a wet brake for a drive axle of an engineering vehicle according to any one of claims 1 to 7 is implemented.

10. A storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the detection method for a wet brake of a drive axle of an engineering vehicle according to any one of claims 1 to 7 is implemented.

Images