Engineering wheel cleaning automatic spraying system, control method, equipment and medium

Abstract

The invention relates to the technical field of engineering equipment control, in particular to an engineering wheel cleaning automatic spraying system, a control method, equipment and a medium. Outputting a first area sludge coverage rate of the upper surface of the wheel part, a second area sludge coverage rate of the front surface of the wheel part and a third area sludge coverage rate of the inner side surface of the wheel part; and establishing a wheel smudginess evaluation model, calculating and outputting a current evaluation value, and outputting a control signal for the spraying system according to a threshold range in which the evaluation value is located. According to the method, comprehensive scoring consideration is carried out in combination with the smudginess condition and the smudginess position of the wheel part of the engineering vehicle, and control over different powers of the spraying system is output according to the obtained comprehensive analysis result, so that the amount of water sprayed by the current spraying system conforms to the smudginess condition of the wheel part of the engineering vehicle as much as possible, and the spraying efficiency of the spraying system is improved. And finally, the spraying effect of the spraying system is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of engineering equipment control, in particular to an automatic spraying system for cleaning wheels of engineering vehicles, a control method, equipment and medium. BACKGROUND

[0002] When engineering vehicles are used in the engineering field, the wheels are covered with a large amount of sludge or other dirt. Due to current environmental protection requirements, all engineering vehicles need to use a spraying system to flush the wheels on site to meet the requirements for normal road use.

[0003] However, in the prior art, the spraying system is not controlled in detail, and the power of the spraying system is not automatically controlled according to the degree of dirt on the wheels of the engineering vehicle. Under the same power of the spraying system, the engineering vehicle with less dirt does not need a large amount of spraying, which causes waste of water resources, and the engineering vehicle with serious dirt may not meet the normal road use standard, resulting in poor flushing effect of the spraying system. SUMMARY

[0004] The present application aims to provide an automatic spraying system for cleaning wheels of engineering vehicles, a control method, equipment and medium to solve the above problems in the prior art.

[0005] The present application is achieved by the following technical solutions:

[0006] In a first aspect, the present application provides an automatic spraying system control method for cleaning wheels of engineering vehicles, comprising:

[0007] A first shooting area is set, and when the engineering vehicle enters the first shooting area, a first basic image of the wheel part of the current engineering vehicle is obtained. After the first basic image is preprocessed, a first target image is obtained. The first target image is identified, and the first area sludge coverage rate of the upper surface of the wheel part, the second area sludge coverage rate of the front surface of the wheel part and the third area sludge coverage rate of the inner surface of the wheel part are output.

[0008] A wheel part dirt evaluation model is established, and the first sludge coverage rate, the second area sludge coverage rate and the third area sludge coverage rate are used to calculate the current evaluation value based on the wheel part dirt evaluation model.

[0009] A plurality of threshold ranges are set, and the plurality of threshold ranges correspond to different working powers of the spraying system. The control signal of the spraying system is output according to the threshold range in which the evaluation value is located.

[0010] The second shooting area is arranged between the first shooting area and the spraying system, when the engineering vehicle enters the second shooting area, if the control signal of the spraying system has been received at this time, the control signal is executed normally, if the control signal of the spraying system has not been received, the spraying system is directly started to spray at any working power;

[0011] The third shooting area is arranged, when the engineering vehicle enters the first shooting area, the second basic image of the wheel part of the current engineering vehicle is obtained, the second basic image is preprocessed to obtain the second target image, the second target image is identified, and the first area sludge coverage rate, the second area sludge coverage rate and the third area sludge coverage rate are outputted;

[0012] Whether the wheel part dirt evaluation model is corrected is judged based on the first area sludge coverage rate, the second area sludge coverage rate and the third area sludge coverage rate of the third shooting area.

[0013] Preferably, the identification of the first target image or the identification of the second target image comprises:

[0014] The historical image of the wheel with sludge is obtained, the historical image is preprocessed to extract features, and the feature image after extraction is divided into a data set, the data set comprising a training set and a test set;

[0015] An image recognition neural network model is established, the structure of the neural network is optimized, the training set is input into the image recognition neural network model for training, the number of iterations is set and the learning rate is updated, and the trained image recognition neural network model is outputted;

[0016] The first target image or the second target image is identified by the image recognition neural network model, the sludge area in the first target image or the second target image is marked, and the identification result is outputted.

[0017] Preferably, the feature extraction after pre-processing of the historical image comprises:

[0018] The module based on channel and spatial double attention comprises:

[0019] F C =[W2·ReLU(W1·G c +b1)]+b2

[0020] Further comprising structured sparse regularization, comprising:

[0021]

[0022] In the formula, F c is the channel attention weight, W1 and W2 are learnable weight matrices, and Gc is the average pooling result of the input feature map, b1 and b2 are bias terms, L SSR is the structured sparse regularization term, λ is the regularization coefficient, N is the number of feature maps, C is the number of channels, F ij is the value of the i-th feature map in the j-th channel, and ε is a constant.

[0023] Preferably, the optimizing the structure of the neural network comprises:

[0024] constructing a hybrid convolution model, comprising:

[0025]

[0026] wherein y(p) is the value of the output feature map at position p, w k is the k-th convolution kernel, x is the input feature map, Δp k is the offset of the k-th convolution kernel.

[0027] Preferably, the first area sludge coverage rate of the outer surface of the wheel part, the second area sludge coverage rate of the front surface of the wheel part, and the third area sludge coverage rate of the inner surface of the wheel part comprise:

[0028] obtaining the area of the sludge area of the other mark in the first target image or the second target image, obtaining the external surface area of the current wheel part, and outputting the first area sludge coverage rate, the second area sludge coverage rate, and the third area sludge coverage rate based on the external surface area, the area of the sludge area, and the position of the sludge area in the wheel part;

[0029]

[0030] wherein, is the first area sludge coverage rate, S1 is the area of the sludge area located on the upper surface of the wheel part, is the first area sludge coverage rate, S2 is the area of the sludge area located on the front surface of the wheel part, is the first area sludge coverage rate, S3 is the area of the sludge area located on the inner surface of the wheel part, S z is the external surface area of the wheel part.

[0031] Preferably, the establishing the wheel part dirt evaluation model comprises:

[0032]

[0033] wherein, K l is the evaluation value, S a is the area of the upper surface of the wheel part, S b is the area of the front surface of the wheel part, S c is the surface area of the inner surface of the wheel part.

[0034] Preferably, the judging whether to correct the wheel dirty evaluation model comprises:

[0035] A judgment threshold is set, and it is respectively judged whether the first area sludge coverage rate, the second area sludge coverage rate and the third area sludge coverage rate of the third shooting area are less than the judgment threshold, when all the sludge coverage rates are less than the judgment threshold, no correction is performed, otherwise, correction is performed;

[0036] The correction comprises obtaining the sludge coverage rate not less than the judgment threshold, and setting the product of the sludge coverage rate in the evaluation model and a calculation coefficient greater than 1.

[0037] In the second aspect, the application further provides an automatic spraying system for cleaning wheels of engineering vehicles, comprising:

[0038] An image processing module is arranged to set a first shooting area, when the engineering vehicle enters the first shooting area, a first basic image of the wheel part of the current engineering vehicle is obtained, and after the first basic image is preprocessed, a first target image is obtained, the first target image is identified, and the first area sludge coverage rate on the upper surface of the wheel part, the second area sludge coverage rate of the front surface of the wheel part and the third area sludge coverage rate of the inner surface of the wheel part are output; a wheel dirty evaluation model is established, and the current evaluation value is calculated based on the wheel dirty evaluation model through the first sludge coverage rate, the second area sludge coverage rate and the third area sludge coverage rate;

[0039] An output module is arranged, and a plurality of threshold ranges are set, the plurality of threshold ranges correspond to different working powers of the spraying system, and a control signal for the spraying system is output according to the threshold range where the evaluation value is located; a second shooting area is set, the second shooting area is located between the first shooting area and the spraying system, when the engineering vehicle enters the second shooting area, if the control signal for the spraying system has been received at this time, the control signal is normally executed, and if the control signal for the spraying system has not been received, any working power of the spraying system is directly started to spray; a third shooting area is set, when the engineering vehicle enters the first shooting area, a second basic image of the wheel part of the current engineering vehicle is obtained, and after the second basic image is preprocessed, a second target image is obtained, the second target image is identified, and the first area sludge coverage rate, the second area sludge coverage rate and the third area sludge coverage rate are output; whether to correct the wheel dirty evaluation model is judged based on the first area sludge coverage rate, the second area sludge coverage rate and the third area sludge coverage rate of the third shooting area;

[0040] A main control module is connected with the image processing module and the output module, and is used for executing the control method of the automatic spraying system for cleaning wheels of engineering vehicles.

[0041] In a third aspect, the present application 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 automatic wheel cleaning and spraying method for engineering vehicles when executing the computer program.

[0042] In a fourth aspect, the present application provides a computer readable storage medium, wherein the computer readable storage medium stores a computer program, and the computer program is executed by a processor to implement the above-mentioned automatic wheel cleaning and spraying system control method for engineering vehicles.

[0043] The technical scheme of the present application has at least the following advantages and beneficial effects:

[0044] The control method provided by the present application mainly comprises identifying the first target image, outputting the first area sludge coverage rate of the upper surface of the wheel part, the second area sludge coverage rate of the front surface of the wheel part, and the third area sludge coverage rate of the inner surface of the wheel part; establishing a wheel part dirt evaluation model, calculating and outputting the current evaluation value based on the wheel part dirt evaluation model through the first sludge coverage rate, the second area sludge coverage rate, and the third area sludge coverage rate, and outputting the control signal of the spraying system according to the threshold range where the evaluation value is located. Through the above method, the dirt condition of the wheel part of the engineering vehicle and the position of the dirt are comprehensively scored and considered, and according to the obtained comprehensive analysis result, the control of different powers of the spraying system is outputted, so that the water amount of the current spraying system is as much as possible to meet the dirt condition of the wheel part of the engineering vehicle, and finally the spraying effect of the spraying system is improved. BRIEF DESCRIPTION OF DRAWINGS

[0045] In order to more clearly illustrate the technical scheme of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as a limitation on the scope, and for those skilled in the art, other related drawings can also be obtained without creative labor on the basis of these drawings.

[0046] Fig. 1 The flowchart of the present application;

[0047] Fig. 2 The system structure diagram of the present application. DETAILED DESCRIPTION

[0048] In order to make the purposes, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some but not all of the embodiments of the present application. The components of the embodiments of the present application described and shown in the drawings can be arranged and designed in various different configurations.

[0049] The terms "first", "second", and the like in the specification and claims of the present application and the above drawings are used to distinguish similar objects, and do not necessarily indicate a specific order or a chronological sequence. The naming or numbering of the steps appearing in the present application does not mean that the steps in the method flow must be executed in the order / time sequence indicated by the naming or numbering. The execution order of the steps that have been named or numbered can be changed according to the technical purpose to be achieved, as long as the same or similar technical effects can be achieved.

[0050] Please refer to Figs. 1-2 An engineering vehicle wheel cleaning automatic spraying system control method, comprising:

[0051] S101: Set a first shooting area, when the engineering vehicle enters the first shooting area, obtain a first basic image of the wheel part of the current engineering vehicle, and after pre-processing the first basic image, obtain a first target image, identify the first target image, and output the first area sludge coverage rate of the upper surface of the wheel part, the second area sludge coverage rate of the front surface of the wheel part, and the third area sludge coverage rate of the inner side surface of the wheel part;

[0052] The first shooting area is the first area that the engineering vehicle needs to reach before washing, and the wheel part dirt identification before washing is performed. In addition, different degrees of dirt coverage are given for different positions of the wheel part, because the degree of washing required by different areas of different positions of the wheel part is different, and the above method provides a basis for subsequent evaluation.

[0053] Secondly, the upper surface of the wheel part is defined as the circumferential surface in contact with the ground in the embodiment, the front surface of the wheel part is the entire surface facing the outside of the tire, and the inner side surface of the wheel part is the surface of the hub exposed to the outside.

[0054] S102: Establish a wheel part dirt evaluation model, and calculate and output the current evaluation value based on the wheel part dirt evaluation model through the first sludge coverage rate, the second area sludge coverage rate, and the third area sludge coverage rate;

[0055] S103: Set a plurality of threshold ranges, and the plurality of threshold ranges correspond to different working powers of the spraying system, and output the control signal of the spraying system according to the threshold range where the evaluation value is located;

[0056] In the embodiment, the entire numerical range is divided into three threshold ranges by two numerical values, and the three threshold ranges can correspond to three gears of the spraying system, i.e., low, medium and high, and the three gears are different working powers of the spraying system, i.e., different water spraying amounts.

[0057] S104: A second shooting area is set, and the second shooting area is located between the first shooting area and the spraying system. When the engineering vehicle enters the second shooting area, if a control signal for the spraying system has been received at this time, the control signal is normally executed, and if no control signal for the spraying system has been received, any working power of the spraying system is directly started to spray.

[0058] In the embodiment, since image recognition needs time, if the engineering vehicle needs to wait in the first shooting area during the process of advancing, a large amount of congestion of the following engineering vehicles will be caused, so the engineering vehicle generally does not stop during the process of advancing. Therefore, the purpose of setting the second shooting area is to inform the current control system that the engineering vehicle is about to enter the spraying area, and if the generation of the control signal has not been completed at this time, any power is started to spray without waiting for the control signal, so as to avoid the stop of the engineering vehicle and cause congestion of the following engineering vehicles.

[0059] S104: A third shooting area is set, when the engineering vehicle enters the first shooting area, a second basic image of the wheel part of the current engineering vehicle is acquired, and after the second basic image is preprocessed, a second target image is obtained. The second target image is identified, and a first area sludge coverage rate, a second area sludge coverage rate and a third area sludge coverage rate are output.

[0060] S105: Whether the wheel part dirtiness evaluation model is corrected is judged based on the first area sludge coverage rate, the second area sludge coverage rate and the third area sludge coverage rate of the third shooting area.

[0061] The purpose of setting the third shooting area is to verify the previous control signal, judge how the cleaning effect is, and whether the model needs to be corrected. In addition, it can also be judged that the engineering vehicle has driven out of the spraying area, and the spraying area can be closed.

[0062] The control method provided by the application mainly comprises the following steps: recognizing the first target image, outputting the first area sludge coverage rate of the upper surface of the wheel part, the second area sludge coverage rate of the front surface of the wheel part and the third area sludge coverage rate of the inner surface of the wheel part; establishing a wheel part dirtiness evaluation model, calculating and outputting the current evaluation value based on the wheel part dirtiness evaluation model through the first sludge coverage rate, the second area sludge coverage rate and the third area sludge coverage rate, and outputting the control signal of the spraying system according to the threshold range where the evaluation value is located. Through the above method, the dirtiness of the wheel part of the engineering vehicle and the position of the dirtiness are comprehensively scored and considered, and according to the obtained comprehensive analysis result, the control of different powers of the spraying system is output, so that the water amount of the current spraying system is as much as possible to meet the dirtiness of the wheel part of the engineering vehicle, and finally the spraying effect of the spraying system is improved

[0063] In an example embodiment of the application, the recognizing the first target image or the recognizing the second target image comprises:

[0064] A historical image of the wheel with sludge is obtained, feature extraction is performed on the preprocessed historical image, and the feature image after extraction is divided into a data set, the data set comprising a training set and a test set;

[0065] An image recognition neural network model is established, the structure of the neural network is optimized, the training set is input into the image recognition neural network model for training, the number of iterations is set and the learning rate is updated, and the trained image recognition neural network model is output;

[0066] The first target image or the second target image is recognized by the image recognition neural network model, the sludge area in the first target image or the second target image is marked, and the recognition result is output.

[0067] In the present embodiment, the image recognition neural network model can adopt conventional prior art, for example, VGG, ResNet and the like deep learning network.

[0068] In an example embodiment of the application, the feature extraction after pre-processing the historical image comprises:

[0069] In the field of image recognition, feature extraction and representation optimization are extremely critical, which directly affects the performance of the model. In order to more efficiently extract features with strong distinguishing ability and enhance their representation ability, a channel and spatial double attention based module (CSAM) is introduced. This module can adaptively adjust the feature map weight by explicitly modeling the dependence relationship between channel and spatial dimensions.

[0070] The module based on channel and spatial double attention comprises:

[0071] Fc =[W2·ReLU(w1·G c +b1)]+b2

[0072] To facilitate the network to learn more compact feature representation, structured sparse regularization is also included, including:

[0073]

[0074] In the formula, F c is the channel attention weight, W1 and W2 are learnable weight matrices, G C is the average pooling result of the input feature map, b1 and b2 are bias terms, L SSR is a structured sparse regularization term, λ is a regularization coefficient, N is the number of feature maps, C is the number of channels, F ij is the value of the i-th feature map in the j-th channel, and ε is a constant.

[0075] In an example embodiment of the present application, in the field of image recognition, although deep learning network structures such as VGG and ResNet have achieved excellent performance, they still have deficiencies in dealing with complex scenes and fine-grained recognition. In view of the influence of network depth, width and module design on performance, the following optimization strategies are proposed: fuse dilated convolution and deformable convolution to construct a hybrid convolution module, thereby expanding the receptive field and improving model adaptability, and optimize the structure of the neural network, including:

[0076]

[0077] In the formula, y(p) is the value of the output feature map at position p, w k is the k-th convolution kernel, x is the input feature map, Δp k is the offset of the k-th convolution kernel.

[0078] In an example embodiment of the present application, the first area sludge coverage rate of the output wheel upper surface, the second area sludge coverage rate of the wheel front surface and the third area sludge coverage rate of the wheel inner surface include:

[0079] The area of the sludge area of the other mark in the first target image or the second target image is obtained, the external surface area of the current wheel is obtained, and the first area sludge coverage rate, the second area sludge coverage rate and the third area sludge coverage rate are output based on the external surface area, the area of the sludge area and the position of the sludge area in the wheel;

[0080]

[0081] In the formula, is the first area sludge coverage rate, S1 is the area of the sludge area located on the wheel upper surface, S2 is the sludge area on the front surface of the wheel part, S3 is the sludge area on the inner side surface of the wheel part, S z S4 is the outer surface area of the wheel part.

[0082] In one example embodiment of the present application, the establishing the wheel part dirty evaluation model comprises:

[0083]

[0084] wherein, K l S is the evaluation value, S a S is the area of the upper surface of the wheel part, S b S is the area of the front surface of the wheel part, S c S is the surface area of the inner side surface of the wheel part.

[0085] In the present embodiment, each coverage rate is weighted calculated according to the area of each position of the wheel part, to obtain a relatively reasonable value.

[0086] In one example embodiment of the present application, the judging whether to correct the wheel part dirty evaluation model comprises:

[0087] A judgment threshold is set, and it is respectively judged whether the first area sludge coverage rate, the second area sludge coverage rate and the third area sludge coverage rate of the third shooting area are less than the judgment threshold, when all the sludge coverage rates are less than the judgment threshold, no correction is performed, otherwise, correction is performed;

[0088] The correction comprises obtaining the sludge coverage rate not less than the judgment threshold, and setting a product of a calculation coefficient greater than 1 after the sludge coverage rate in the evaluation model.

[0089] For example, if the current second area sludge coverage rate is not less than the judgment threshold, the model is corrected as:

[0090]

[0091] wherein, N is the calculation coefficient.

[0092] An automatic spraying system for cleaning the wheels of engineering vehicles, comprising:

[0093] The image processing module is configured to set a first shooting area, obtain a first basic image of a wheel part of the engineering vehicle when the engineering vehicle enters the first shooting area, obtain a first target image after preprocessing the first basic image, identify the first target image, and output a first area sludge coverage rate of an upper surface of the wheel part, a second area sludge coverage rate of a front surface of the wheel part, and a third area sludge coverage rate of an inner surface of the wheel part; and establish a wheel part dirtiness evaluation model, calculate and output a current evaluation value based on the wheel part dirtiness evaluation model through the first sludge coverage rate, the second area sludge coverage rate, and the third area sludge coverage rate.

[0094] The output module is configured to set a plurality of threshold ranges corresponding to different working powers of the spraying system, output a control signal for the spraying system according to a threshold range in which the evaluation value is located, set a second shooting area between the first shooting area and the spraying system, execute the control signal normally when the engineering vehicle enters the second shooting area and the control signal for the spraying system has been received, and directly start spraying at any working power of the spraying system when the engineering vehicle enters the second shooting area and the control signal for the spraying system has not been received; and set a third shooting area, obtain a second basic image of the wheel part of the engineering vehicle when the engineering vehicle enters the first shooting area, obtain a second target image after preprocessing the second basic image, identify the second target image, and output the first area sludge coverage rate, the second area sludge coverage rate, and the third area sludge coverage rate; and determine whether to correct the wheel part dirtiness evaluation model based on the first area sludge coverage rate, the second area sludge coverage rate, and the third area sludge coverage rate of the third shooting area.

[0095] The main control module is connected with the image processing module and the output module, and is configured to execute the engineering vehicle wheel cleaning automatic spraying system control method in any one of claims 1-7.

[0096] In addition, each functional unit in each embodiment of the present application can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0097] The integrated unit, if implemented in the form of a software function unit and sold or used as an independent product, can be stored in a computer readable storage medium. The computer software product stored in a storage medium includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the embodiments of the method of the present application. The aforementioned storage medium includes various media that can store program codes, such as a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0098] The above merely describes the preferred embodiments of the present application and is not intended to limit the present application. For those skilled in the art, the present application can have various modifications and changes. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A control method for an automatic spraying system for cleaning engineering wheels, characterized in that: include: A first shooting area is set. When an engineering vehicle enters the first shooting area, a first basic image of the wheel of the current engineering vehicle is obtained. After preprocessing the first basic image, a first target image is obtained. The first target image is recognized to output the sludge coverage rate of a first area on the upper surface of the wheel, the sludge coverage rate of a second area on the front of the wheel, and the sludge coverage rate of a third area on the inner side of the wheel; Establishing a wheel contamination evaluation model, and calculating and outputting a current evaluation value based on the wheel contamination evaluation model through the first sludge coverage rate, the second area sludge coverage rate, and the third area sludge coverage rate; And set a number of threshold ranges, which correspond to different working powers of the sprinkler system, and output a control signal for the sprinkler system according to the threshold range where the evaluation value is located; A second shooting area is set, and the second shooting area is located between the first shooting area and the sprinkler system. When the engineering vehicle enters the second shooting area, if the control signal for the sprinkler system has been received at this time, the control signal is normally executed; if the control signal for the sprinkler system is not received, the sprinkler system is directly started at any working power to spray; Set a third shooting area. When the engineering vehicle enters the first shooting area, obtain a second basic image of the wheel of the current engineering vehicle, pre-process the second basic image to obtain a second target image, recognize the second target image, and output the sludge coverage rate of the first area, the sludge coverage rate of the second area, and the sludge coverage rate of the third area; Whether to modify the wheel contamination assessment model is determined based on the first region sludge coverage rate, the second region sludge coverage rate, and the third region sludge coverage rate of the third imaging region.

2. The control method of an automatic spraying system for cleaning engineering wheels according to claim 1, characterized in that: The identifying of the first target image or the identifying of the second target image includes: Obtain historical images of wheels with mud, perform feature extraction on the historical images after preprocessing, and divide the extracted feature images into a data set, wherein the data set includes a training set and a test set; Establish an image recognition neural network model and optimize the structure of the neural network. Input the training set into the image recognition neural network model for training. Set the number of iterations and update the learning rate. Output the trained image recognition neural network model. The first target image or the second target image is recognized by an image recognition neural network model, the sludge area in the first target image or the second target image is marked, and the recognition result is output.

3. The control method of an automatic spraying system for cleaning engineering wheels according to claim 2, characterized in that: The feature extraction after pre-processing the historical image includes: Includes modules based on channel and spatial dual attention, including: F c =[W2·ReLU(W1·G c +b1)]+b2 Also included is structured sparsity regularization, including: Where, F c is the channel attention weight, W1 and W2 are learnable weight matrices, G c is the average pooling result of the input feature map, b1 and b2 are bias terms, L SSR is the structured sparse regularization term, λ is the regularization coefficient, N is the number of feature maps, C is the number of channels, and F ij is the value of the i-th feature map at the j-th channel, and ε is a constant.

4. The control method of an automatic spraying system for cleaning engineering wheels according to claim 3 is characterized in that: The optimization of the structure of the neural network includes: Build a hybrid convolutional model, including: Where y(p) is the value of the output feature map at position p, w k is the kth convolution kernel, x is the input feature map, Δp k is the offset of the kth convolution kernel.

5. The control method of an automatic spraying system for cleaning engineering wheels according to claim 4, characterized in that: The sludge coverage rate of the first area on the upper surface of the output wheel, the sludge coverage rate of the second area on the front surface of the wheel, and the sludge coverage rate of the third area on the inner side of the wheel include: Obtaining the area of ​​a marked sludge region in the first target image or the second target image, obtaining the external surface area of ​​the current wheel, and outputting the sludge coverage rate of the first region, the sludge coverage rate of the second region, and the sludge coverage rate of the third region based on the external surface area, the area of ​​the sludge region, and the position of the wheel where the sludge region is located; Where, is the sludge coverage rate of the first area, S1 is the area of ​​the sludge area on the upper surface of the wheel, is the sludge coverage rate of the first area, S2 is the area of ​​the sludge area in front of the wheel, is the sludge coverage rate of the first area, S3 is the area of ​​the sludge area on the inner side of the wheel, S z is the external surface area of ​​the wheel.

6. The control method of an automatic spraying system for cleaning engineering wheels according to claim 5, characterized in that: The establishment of the wheel contamination evaluation model comprises: Where K l is the evaluation value, S a is the area of ​​the upper surface of the wheel, S b is the front area of ​​the wheel, S c It is the surface area of ​​the inner side of the wheel.

7. The control method of an automatic spraying system for cleaning engineering wheels according to claim 5, characterized in that: The determining whether to modify the wheel contamination evaluation model includes: Setting a judgment threshold to respectively judge whether the sludge coverage rate of the first area, the second area, and the third area of ​​the third shooting area is less than the judgment threshold; if all the sludge coverage rates are less than the judgment threshold, no correction is performed; otherwise, correction is performed; The correction includes obtaining a sludge coverage ratio that is not less than a judgment threshold, and setting a product of calculation coefficients greater than 1 after the sludge coverage ratio in the evaluation model.

8. An automatic spraying system for cleaning wheels for engineering purposes, characterized in that: include: An image processing module is provided, wherein a first shooting area is set. When an engineering vehicle enters the first shooting area, a first basic image of the wheel of the current engineering vehicle is obtained. The first basic image is preprocessed to obtain a first target image. The first target image is recognized and the sludge coverage rate of the first area on the upper surface of the wheel, the sludge coverage rate of the second area on the front of the wheel, and the sludge coverage rate of the third area on the inner side of the wheel are output; a wheel contamination evaluation model is established, and a current evaluation value is calculated based on the wheel contamination evaluation model using the first sludge coverage rate, the second sludge coverage rate, and the third sludge coverage rate; An output module is provided, and several threshold ranges are set, and the several threshold ranges correspond to different working powers of the spray system, and a control signal for the spray system is output according to the threshold range where the evaluation value is located; a second shooting area is set, and the second shooting area is located between the first shooting area and the spray system. When the engineering vehicle enters the second shooting area, if the control signal for the spray system has been received at this time, the control signal is executed normally. If the control signal for the spray system is not received, the spray system is directly started to spray at any working power; a third shooting area is set, and when the engineering vehicle enters the first shooting area, a second basic image of the wheel of the current engineering vehicle is obtained, and the second basic image is pre-processed to obtain a second target image, the second target image is recognized, and the sludge coverage rate of the first area, the sludge coverage rate of the second area, and the sludge coverage rate of the third area are output; based on the sludge coverage rate of the first area, the sludge coverage rate of the second area, and the sludge coverage rate of the third area of ​​the third shooting area, it is determined whether to modify the wheel contamination evaluation model; The main control module is connected to the image processing module and the output module, and is used to execute the control method of the automatic spraying system for cleaning engineering wheels as described in any one of claims 1 to 7.

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 automatic spraying method for cleaning engineering wheels as described in any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by the processor, the method for controlling an automatic spraying system for washing engineering wheels according to any one of claims 1 to 7 is implemented.

Images