A cleaning control method and control system for engineering vehicles

Abstract

This invention discloses a cleaning control method and control system for engineering vehicles, belonging to the field of engineering vehicle maintenance and repair technology. The cleaning control method of this invention first acquires a first image and a second image of the engineering vehicle, then generates the nozzle flow rate of the cleaning gun at each scanning moment. The cleaning gun operates along the corresponding scanning trajectory, and the first and second throttling valves maintain corresponding valve orifice areas. Based on the absorption and reflection effects of soil on different wavelengths of light, this invention predicts the contour and thickness of the soil, and then adjusts the flow rate of the cleaning gun and the throttling valves of the water supply branch according to the operating flow rate model, avoiding excessive flow and water waste. This invention updates the working head and cleaning speed through a second distribution map to ensure that the working parameters of the cleaning gun conform to the current engineering environment. Furthermore, this invention uses a pressure relief valve to offset the head loss in the pipeline between the booster pump set and the cleaning gun, ensuring that the hydraulic medium of the cleaning gun reaches the working head.

Description

Technical Field

[0001] This invention relates to the field of engineering vehicle maintenance and repair technology, and in particular to a cleaning control method and control system for engineering vehicles. Background Technology

[0002] Currently, environmental protection requirements for construction projects necessitate the installation of vehicle cleaning equipment in construction areas. Large-scale construction projects involve a large number of vehicles and frequent cleaning, typically employing rapid, automatic cleaning equipment. For example, Chinese Patent Publication No. CN118323050A discloses an intelligent identification and automatic cleaning system for construction vehicles. This system first detects the vehicle's position signal. When the vehicle's position signal matches a preset body cleaning area, it outputs a body cleaning control command to the cleaning mechanism; similarly, when the vehicle's position signal matches a preset tire cleaning area, it outputs a tire cleaning control command to the cleaning mechanism.

[0003] Due to factors such as weather and usage environment, the degree of dirt on different engineering vehicles varies. To save water, existing technologies have developed methods that identify the degree of dirt through image recognition. For example, Chinese Patent Publication No. CN118323050A discloses an intelligent cleaning method for engineering vehicles. This method captures an image of the surface of the engineering vehicle to be cleaned, obtains the dirt recognition results, comprehensively calculates the optimal cleaning scheme, intelligently adjusts cleaning parameters, monitors the cleaning status in real time, and detects abnormalities.

[0004] To improve cleaning efficiency, actual engineering vehicle cleaning is usually completed by multiple sets of cleaning guns operating simultaneously. Some cleaning guns are connected via the same branch circuit, and the flow rates between the cleaning guns inhibit each other. To reduce water consumption, precise control of the flow rate of each cleaning gun and branch circuit is required. Furthermore, engineering vehicles operating under similar conditions for the same project or in the same weather can have their operating head and cleaning speed gradually adjusted based on the current cleaning results. Summary of the Invention

[0005] To address the aforementioned problems, this invention provides a cleaning control method and system for engineering vehicles. This invention adjusts the flow rate of the cleaning gun and the water supply branch valve based on the contour and thickness of the soil, thereby saving water. Furthermore, this invention updates the working head and cleaning speed using soil distribution maps before and after cleaning, ensuring that the cleaning parameters of the cleaning gun are suitable for the current engineering environment.

[0006] The objective of this invention can be achieved through the following technical means:

[0007] A method for controlling the cleaning of engineering vehicles includes the following steps:

[0008] Step 1: The booster pump set is connected to multiple water supply branches, each water supply branch is connected to multiple cleaning guns, and the multiple cleaning guns form a working surface. A working area is allocated to each cleaning gun, and the working head and cleaning speed of the cleaning gun are initialized.

[0009] Step 2: The engineering vehicle enters the first position, and the first and second images of the engineering vehicle are acquired. Based on the working surface, the first and second images are transformed into surface images and thickness images respectively, generating the first distribution map of the soil. According to the working area, the first distribution map is divided into multiple sub-images, and the scanning trajectory of the cleaning gun in the corresponding sub-image is generated.

[0010] Step 3: Generate the nozzle flow rate of the cleaning gun at each scan time based on the sub-image and the operation flow model, calculate the valve orifice area of ​​the first throttle valve of the cleaning gun at each scan time, and calculate the valve orifice area of ​​the second throttle valve of the water supply branch at each scan time.

[0011] Step 4: The engineering vehicle enters the second position, the cleaning gun operates along the corresponding scanning trajectory, the first throttle valve and the second throttle valve maintain the corresponding valve orifice area, and after the cleaning gun finishes its operation, the engineering vehicle exits the second position.

[0012] Step 5: Collect the third and fourth images of the engineering vehicle, generate the second distribution map of the soil, and calculate the stain index. If the stain index is greater than the stain threshold, proceed to step 6; otherwise, the engineering vehicle exits the first position and the task ends.

[0013] Step 6: Adjust the working head and cleaning speed of the cleaning gun according to the first and second distribution diagrams, and return to step 2.

[0014] In this invention, in step 2, the first image is a visible light image, multiple sets of coordinate transformation parameters between the first image and the working surface are generated, and multiple sets of the first images are combined into a surface image based on the coordinate transformation parameters. The second image is a spectral image, and multiple sets of the second images are combined into a thickness image based on the coordinate transformation parameters. Soil contour data is extracted from the surface image, soil thickness data is extracted from the thickness image, and a first distribution map is generated by combining the soil contour data and the soil thickness data.

[0015] In this invention, in step 2, the working area of ​​the cleaning gun n is a rectangular area with coordinates (x, y) satisfying x1≤x≤x2 and y1≤y≤y2. The starting coordinates (x1, y1) and ending coordinates (x2, y2) of the scanning trajectory are determined according to the working area, and the spacing of the scanning trajectory is determined according to the scattering area of ​​the cleaning gun.

[0016] In this invention, in step 3, the scanning time t and the deflection angle β at which the working center line of the cleaning gun n moves to coordinates (x, y) are generated in combination with the cleaning speed ω. tExtract the scattering area of ​​the cleaning gun at scanning time t, and calculate the maximum soil thickness C of this scattering area. t and average soil thickness G t The angle β of the cleaning gun n t Working head H0, maximum soil thickness C t and average soil thickness G t Inputting the data into the operational flow model generates the nozzle flow rate Q of the cleaning gun n at scanning time t. mnt , m is the number of the water supply branch where the cleaning gun n is located.

[0017] In this invention, based on the nozzle flow rate Q mnt Calculate the valve orifice area of ​​the first throttle valve corresponding to cleaning gun n at scanning time t, based on the branch flow rate Q of water supply branch m at scanning time t. mt Calculate the valve port area of ​​the second throttle valve corresponding to water supply branch m.

[0018] In this invention, in step 4, the head loss of the booster pump set and the water supply branch is predicted, and the pressure relief of the booster pump set and the water supply branch is adjusted according to the head loss.

[0019] In this invention, in step 5, the maximum soil area, average soil area, maximum soil thickness, and average soil thickness are extracted from the second distribution map, and the stain index is calculated.

[0020] In this invention, in step 6, a difference map of the first distribution map and the second distribution map is calculated, and the maximum soil change rate and the average soil change rate are extracted from the difference map. The adjustment amount of the cleaning speed is calculated based on the maximum soil area, the average soil area, the maximum soil change rate and the average soil change rate. The adjustment amount of the working head is calculated based on the maximum soil thickness, the average soil thickness, the maximum soil change rate and the average soil change rate.

[0021] A cleaning control system for implementing the cleaning control method for the engineering vehicle includes:

[0022] The conveyor belt is configured as a mobile engineering vehicle, and the conveyor belt has a first sensing unit for sensing a first position and a second sensing unit for sensing a second position.

[0023] The image acquisition device is configured to acquire a first image, a second image, a third image, and a fourth image of the engineering vehicle;

[0024] An image processing apparatus is configured to generate a first distribution map and a second distribution map;

[0025] The data processing unit is configured to generate the nozzle flow rate of the cleaning gun at each scan time.

[0026] Multiple sets of cleaning guns are configured to provide cleaning media to engineering vehicles, and each cleaning gun has a first throttle valve;

[0027] Multiple water supply branches are configured to supply cleaning media to at least two sets of cleaning guns, and each water supply branch has a second throttle valve;

[0028] The booster pump set is configured to provide cleaning medium to the water supply branch;

[0029] The control device is configured to generate the valve orifice area of ​​the first throttle valve, the valve orifice area of ​​the second throttle valve, and the relief pressure of the booster pump set and the water supply branch.

[0030] In this invention, the cleaning control system further includes a truss and a cleaning device. The conveyor belt is mounted on the truss and has a third sensing unit for sensing a third position. The cleaning device is located between the first position and the third position, and the cleaning gun is located between the third position and the second position.

[0031] In this invention, the cleaning control system further includes a first water tank connected to a booster pump set. The booster pump set feeds back to the first water tank via a first pressure relief valve, and a second throttle valve returns to the first water tank via a second pressure relief valve. The head loss of the water supply branch is calculated based on the nozzle flow rate of multiple sets of cleaning guns, and the pressure relief of the second pressure relief valve is then calculated. The head loss of the booster pump set is calculated based on the branch flow rate of multiple sets of water supply branches, and the pressure relief of the first pressure relief valve is then calculated.

[0032] The cleaning control system also includes a second water tank, and the water supply branch also includes a two-position three-way valve. In the cleaning state, the second throttle valve is connected to the first throttle valve through the two-position three-way valve. In the shut-off state, the first throttle valve is connected to the second water tank through the two-position three-way valve.

[0033] The present invention provides a cleaning control method and control system for engineering vehicles, the advantages of which are as follows: The present invention predicts the contour and thickness of the soil based on the absorption and reflection effects of soil on different wavelengths of light, and then adjusts the flow rate of the cleaning gun and the throttling valve of the water supply branch in conjunction with the operating flow model, avoiding excessive flow and water waste. The present invention updates the working head and cleaning speed through a second distribution diagram to ensure that the working parameters of the cleaning gun are consistent with the current engineering environment. Furthermore, the present invention achieves water resource circulation through a first pressure relief valve and a second pressure relief valve, ensuring consistent working head of the cleaning gun. Simultaneously, the pressure relief valve offsets the head loss in the pipeline between the booster pump set and the cleaning gun, ensuring that the hydraulic medium of the cleaning gun reaches the working head. Attached Figure Description

[0034] Figure 1 This is a flowchart of the cleaning control method for engineering vehicles according to the present invention;

[0035] Figure 2This is a structural diagram of the control system for the engineering vehicle of the present invention;

[0036] Figure 3 This is a schematic diagram of one direction of the control system of the engineering vehicle of the present invention;

[0037] Figure 4 This is a partial schematic diagram of the control system of the engineering vehicle of the present invention from another direction;

[0038] Figure 5 This is a schematic diagram of the present invention acquiring a first image at a first position;

[0039] Figure 6 This is a schematic diagram of the present invention acquiring a first image at a second location;

[0040] Figure 7 This is a schematic diagram of a preferred first distribution map according to the present invention;

[0041] Figure 8 This is a schematic diagram illustrating how a first distribution map is segmented into sub-images according to the present invention;

[0042] Figure 9 This is a schematic diagram of a sub-image according to the present invention;

[0043] Figure 10 This is a schematic diagram of the scanning trajectory of the present invention;

[0044] Figure 11 This is a schematic diagram of the valve orifice area of ​​the first throttle valve of the present invention at multiple scanning times;

[0045] Figure 12 This is a hydraulic schematic diagram of the present invention;

[0046] Figure 13 This is a block diagram of the control system for the engineering vehicle of the present invention;

[0047] Figure 14 The spectral diagrams of samples with different soil thicknesses are shown in the present invention.

[0048] Figure 15 This represents the probability of the correctness of the sample spectra at different characteristic wavelengths in this invention.

[0049] Figure 16 This is a schematic diagram illustrating the relationship between the deflection angle and the cleaning flow rate in this invention;

[0050] Figure 17 This is a schematic diagram showing the relationship between the working head and the cleaning flow rate of the present invention;

[0051] Figure 18 This is a schematic diagram illustrating the relationship between soil thickness and washing flow rate in this invention.

[0052] Reference numerals in the attached drawings: Truss 100, Conveyor belt 200, First sensing unit 201, Second sensing unit 202, Third sensing unit 203, Cleaning gun 300, First throttle valve 301, Water supply branch 400, Second throttle valve 401, Second pressure relief valve 402, Booster pump set 500, First pressure relief valve 501, Gate valve 502, Two-position three-way valve 503, One-way valve 504, First water tank 505, Second water tank 506, Image acquisition device 600, Cleaning device 700, Control device 800, Engineering vehicle 900. Detailed Implementation

[0053] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0054] Existing technologies directly feed images into big data models to intelligently adjust cleaning parameters. This method is difficult to train and unsuitable for engineering applications. For example, Chinese Patent Publication No. CN113221727A discloses a method for determining the cleaning status of a vehicle through target detection region images, which relies excessively on training the model classifier. Since the main stain in engineering applications is mud, this invention, based on the reflection effect of mud on different wavelengths of light, acquires first and second images of the engineering vehicle to generate a first mud distribution map, improving the accuracy of image recognition in specific fields. Furthermore, this invention divides the mud distribution map into multiple sub-images, independently adjusting the valve area of ​​each cleaning gun. Example 1

[0055] This invention discloses a cleaning control method for engineering vehicles, which involves acquiring a first image and a second image of the engineering vehicle, and adjusting the flow rates of the cleaning gun and the throttle valve of the water supply branch. For example... Figures 1 to 12 As shown, the cleaning control method for engineering vehicles of the present invention includes the following steps:

[0056] Step 1: The booster pump set is connected to multiple water supply branches, each water supply branch is connected to multiple cleaning guns, and the multiple cleaning guns form a working surface. A working area is assigned to each cleaning gun, and the working head and cleaning speed of the cleaning gun are initialized.

[0057] The booster pump set can include one or more hydraulic pumps. The output power of each hydraulic pump is adjusted by a frequency converter. The output power of the booster pump set should be sufficient to allow each cleaning gun to operate at its maximum head. This embodiment includes five working surfaces, each equipped with multiple cleaning guns. The working area of ​​a cleaning gun is the range within which it rotates around the base of the water supply branch. To facilitate the division of adjacent working areas, the working area is typically a rectangular region. For example, the working area of ​​cleaning gun n is a rectangular region with coordinates (x, y) satisfying x1≤x≤x2 and y1≤y≤y2. The cleaning gun can scan the corresponding working surface of the engineering vehicle row by row along the working area.

[0058] Step 2: The engineering vehicle enters the first position, and first and second images of the engineering vehicle are acquired. Based on the working surface, the first and second images are transformed into surface and thickness images, respectively, generating a first distribution map of the soil. According to the working area, the first distribution map is divided into multiple sub-images, and the scanning trajectory of the cleaning gun in the corresponding sub-images is generated. In this embodiment, a binocular vision algorithm, a deep learning algorithm (such as NeRF), or a 3D reconstruction algorithm with depth information can be used to generate a 3D image of the engineering vehicle. Then, the first image is extracted from the 3D image based on the location of the working surface; details are not provided here. The preferred method for generating the surface and thickness images of this invention is described in Embodiment 2.

[0059] like Figure 8 As shown, the rectangular working area is mapped onto the first distribution map, and the first distribution map is divided into multiple sets of sub-images along the boundary of the working area. Some cleaning guns have sub-images in their working areas; these cleaning guns are in working condition and require further determination of the scanning trajectory. In this embodiment, the starting coordinates (x1, y1) and ending coordinates (x2, y2) of the scanning trajectory are determined based on the working area of ​​cleaning gun n, and the spacing of the scanning trajectory is determined based on the scattering area of ​​the cleaning gun. Some cleaning guns have no sub-images in their working areas; these cleaning guns are not working, and the valve port area is 0. Figure 9 In this diagram, soil thickness data at different coordinate points is represented using soil contour lines. (Refer to...) Figure 10 The cleaning gun scans the sub-image to the right from the starting point coordinates, jumps to the next line after reaching the boundary of the working area, and continues until it reaches the end point coordinates to obtain the scanning trajectory.

[0060] Step 3: Generate the nozzle flow rate of the cleaning gun at each scan time based on the sub-image and the operating flow model; calculate the valve orifice area of ​​the first throttle valve of the cleaning gun at each scan time; and calculate the valve orifice area of ​​the second throttle valve of the water supply branch at each scan time. In this invention, the scanning time t at which the operating centerline of the cleaning gun n moves to coordinates (x, y) and the deflection angle β are generated by combining the cleaning speed ω. t Extract the scattering area of ​​the cleaning gun at scanning time t, and calculate the maximum soil thickness C of this scattering area. t and average soil thickness G t The angle β of the cleaning gun n t Working head H0, maximum soil thickness C t and average soil thickness G t Inputting the flow rate model Q=(β, H, C, G) generates the nozzle flow rate Q of the cleaning gun n at scanning time t. mnt 'm' is the water supply branch number where the cleaning gun 'n' is located. Based on the nozzle flow rate Q... mntCalculate the orifice area of ​​the first throttle valve at scan time t, based on the branch flow rate Q of the water supply branch m at scan time t. mt Calculate the valve orifice area of ​​the second throttle valve corresponding to the water supply branch m. The preferred calculation method for nozzle flow rate and valve orifice area is detailed in Example 3.

[0061] Step 4: The engineering vehicle enters the second position, and the cleaning gun operates along the corresponding scanning trajectory. The first and second throttle valves maintain their corresponding orifice areas. After the cleaning gun finishes its operation, the engineering vehicle exits the second position. The orifice areas of the first and second throttle valves are variables related to the scanning time t. The first and second throttle valves maintain their corresponding orifice areas at each scanning time t. The nozzle flow rate is related to the soil thickness corresponding to the scanning time t, thereby saving water while ensuring cleaning efficiency. This invention can also predict the head loss of the booster pump set and the water supply branch, and adjust the relief pressure of the booster pump set and the water supply branch according to the head loss, as described in Embodiment 3.

[0062] Step 5: Acquire the third and fourth images of the engineering vehicle to generate a second soil distribution map. Calculate the stain index. If the stain index is greater than the stain threshold, proceed to Step 6; otherwise, the engineering vehicle exits the first position, and the task ends. The method for generating the second distribution map from the third and fourth images can refer to Step 2. The stain threshold can be a limitation on the cleanliness of the engineering vehicle at the construction site. The size of the stain threshold can be combined with the requirements of the construction documents, as described in Example 4. Based on the "Construction Site Environmental Management Manual," the stain threshold in this example can be taken as 10%. If the stain index is greater than the stain threshold, re-cleaning is required. If the stain index is less than or equal to the stain threshold, the engineering vehicle exits the first position, and the cleaning work ends.

[0063] Step 6: Adjust the working head and cleaning speed of the cleaning gun according to the first and second distribution maps, and return to Step 2. Since the cleaning effect does not meet the requirements of the construction site, it is necessary to further adjust the working head and cleaning speed, such as increasing the working head and decreasing the cleaning speed, as detailed in Example 4. Example 2

[0064] This embodiment further discloses a preferred method for generating surface images, thickness images, and a first distribution map. The method in this embodiment is based on engineering vehicles in the same construction area having similar shapes and vertex pixels. The first or third image is a visible light image, containing RGB pixels at arbitrary coordinates. The second or fourth image is a spectral image used to measure the linear extinction coefficient at arbitrary coordinates. Soil stains exhibit significant absorption characteristics for specific spectral signals; for example, clay minerals are absorbed at a wavelength of 2200 nm (Al-OH bonds), and iron oxides (rust) are absorbed at a wavelength of 850 nm (Fe³⁺ electron transitions). By combining the spectral image under a reference incident intensity, the linear extinction coefficient is determined, and then compared with the extinction coefficient of the metal surface (the clean surface of the engineering vehicle) to obtain soil thickness data.

[0065] First, multiple sets of coordinate transformation parameters for the first image and the working surface are generated. In this embodiment, the coordinate transformation parameters are homography matrices. Specifically, eight sets of vertex pixel template features w1, w2, w3, w4, w5, w6, w7, and w8 of the engineering vehicle are created, such as tire pixel features and cargo box pixel features. Vertex pixel template features corresponding to a certain working surface are determined, such as w1, w2, w3, and w4. Based on these vertex pixel template features w1, w2, w3, and w4, four vertices of the first image are extracted. The size of the surface image is determined based on the positions of these four vertices. The coordinate transformation parameters (homography matrix) are solved based on the coordinates of these four vertices in the first image and the single surface image to be generated.

[0066] Then, based on the coordinate transformation parameters, multiple sets of first images are combined into a surface image, and multiple sets of second images are combined into a thickness image. Specifically, the first image is cropped based on vertex pixel template features w1, w2, w3, and w4. The coordinate data of any coordinate point in the first image is multiplied by the homography matrix to obtain the coordinate data of the coordinate point in the corresponding surface image. The corresponding pixel values ​​are filled into a single surface image, and each first image can generate a single surface image. The average value of any pixel in the single surface images obtained from multiple sets of first images is calculated to generate the final surface image. The same method can be used to generate the thickness image.

[0067] Finally, soil contour data is extracted from the surface image, and soil thickness data is extracted from the thickness image. The soil contour and thickness data are then combined to generate the first distribution map. Specifically, a soil pixel template feature is created, and corresponding pixel features are extracted from the surface image. The connected regions of these pixel features are then merged to obtain the soil contour data. A soil spectral template feature is also created, and the soil thickness corresponding to the linear extinction coefficient at any coordinate point is calculated based on the soil thickness from the spectral template feature. The soil thickness data at any coordinate point is then mapped to the coordinate points of the surface image to generate the first distribution map. For example... Figure 7 As shown, to clearly display the soil thickness data, different soil thickness data in the first distribution map are represented by grayscale values. Example 3

[0068] This embodiment further discloses the method for calculating soil thickness data at arbitrary coordinate points based on soil spectral template features in step 2. The soil spectral template features in this embodiment consist of a coefficient matrix of multiple characteristic wavelengths.

[0069] Step 210: Discretization of characteristic wavelengths.

[0070] The characteristic wavelength determines the prediction accuracy of soil thickness data. First, the reflectance (linear extinction coefficient) of the spectral signal of metal samples within a continuous wavelength range of 350 nm to 1040 nm is collected. Then, the spectral curves of all metal samples with the same soil thickness are averaged to obtain the average linear extinction coefficient of the metal samples at that soil thickness, i.e., the sample spectral set. Sample spectral sets for different soil thicknesses are shown below. Figure 14 As shown in the figure, the soil thickness data corresponding to the metal samples from top to bottom in the spectral curves are 0.528 mm, 0.288 mm, 0.144 mm, 0.13 mm, 0.005 mm, and 0.001 mm, respectively. Different wavelengths were randomly selected as characteristic wavelengths. The probability (acceptability) of the correct sample spectral set at each characteristic wavelength was calculated using the random frog-jumping algorithm. Wavelengths with a correct probability greater than 0.8 were selected as characteristic wavelengths, and the corresponding spectra are the characteristic spectra. Figure 15 As shown, the characteristic wavelengths include 572nm, 601nm, 679.5nm, 719.5nm, 775.2nm, 800.8nm, 803.4nm (excluding similar wavelengths), 821.4nm, 842.1nm, and 876nm.

[0071] Step 220: Construction of the discretized matrix model.

[0072] Step 221: Matrix Decomposition of Characteristic Spectra. Generate a sample spectral matrix U1 of the characteristic spectra. U1 contains the linear extinction coefficient (reflectivity) for each characteristic wavelength, and the number of rows in the matrix is ​​the same as the number of characteristic wavelengths. Set a wavelength weight matrix U2 for each characteristic wavelength and introduce a wavelength error matrix V2. Calculate the wavelength coefficient matrix V1 according to U1 = V1U2 + V2. Iterate through the wavelength weight matrix U2 to reduce the modulus of the wavelength error matrix V2, and finally obtain the wavelength weight matrix U2 with the minimum error matrix.

[0073] Step 222: Matrix decomposition of soil thickness data. Using the same method, generate a sample thickness matrix E1 for each metal sample, containing the soil thickness data for each sample. Set a thickness weight matrix E2 for each characteristic wavelength and introduce a random thickness error matrix F2. Calculate the thickness coefficient matrix F1 according to E1 = F1E2 + F2. Continuously update the weight matrix E2 to reduce the modulus of the error matrix F2, ultimately obtaining the weight matrix E2 with the minimum error matrix.

[0074] Step 223: Combining the results of steps 221 and 222, establish the regression relationship of the coefficient matrix, F1 = V1W1, where W1 is the regression coefficient matrix. Therefore, E1 = V1W1E2 + F2. Combining with step 121, the matrix model of the thickness spectrum is... .

[0075] Step 230: Generate soil spectral template features. In an optimal scenario, the wavelength error matrix V2 = 0, and the matrix model can be simplified to E1 = U1U2. -1 ·W1E2, U2 -1 Let U1W2 be the inverse matrix of U2. Construct a matrix W2 representing the soil spectral template features, where U1W2 = U1U2 -1 • W1E2. Ultimately, E1 = U1W2 + F2. Different characteristic wavelengths or different types of metals can yield different matrix W2s of soil spectral template characteristics. In this embodiment, a simple multiple linear regression algorithm is chosen to simplify F2 to a constant term (0.94 mm). U1 = [ u1 u2 u3 u4 u5 u6 u7 u8 u9] T Then E1 = U1W2 + 0.94 = 0.94 - 0.39u1 - 6.97u2 + 1.31u3 + 3.04u4 + 3.81u5 - 1.63u6 - 0.42u7 + 0.17u8 + 2.0u9. Where u1, u2, u3, u4, u5, u6, u7, u8, and u9 are the linear extinction coefficients for characteristic wavelengths of 572nm, 601nm, 679.5nm, 719.5nm, 775.2nm, 800.8nm, 821.4nm, 842.1nm, and 876nm, respectively.

[0076] Step 240: Generate soil thickness data. Combining the soil spectral template feature matrix W2 obtained in step 230, the current thickness matrix E1 can be obtained by inputting the spectral matrix U1 of the current coordinate point. The mean value of the current thickness matrix E1 is then calculated to obtain the soil thickness data of the current coordinate point. Example 4

[0077] This embodiment further discloses a preferred method for calculating nozzle flow rate and valve orifice area.

[0078] First, multiple sets of historical data are collected to construct an operational flow rate model Q(β,H,C,G). Each set of historical data includes: for the maximum soil thickness C and the average soil thickness G, when the deflection angle of the cleaning gun is β and the working head is H, the effective flow rate for cleaning this scattering area is Q. The operational flow rate model is a computer model composed of four sets of variables. This embodiment does not limit the specific generation method of the operational flow rate model. The function expression of the operational flow rate model can be generated by piecewise modeling and nonlinear coupling, and the function parameters can be updated through online learning to dynamically adapt to system changes and new stain types.

[0079] Then, by combining the cleaning speed ω, the scanning time t and the deflection angle β of the cleaning gun n moving to the coordinate (x, y) are generated. t Specifically, by scanning sub-images at a fixed cleaning speed ω using the cleaning gun n, the scanning time t at which the working center line of the cleaning gun n crosses the coordinate (x, y) can be determined. For example... Figure 3 As shown, the deflection angle β t Let t be the angle between the center line of the work area and the perpendicular line of the work surface at scan time t.

[0080] The scattering area at scan time t is determined based on the coordinates (x, y). The scattering area can be, for example, a circle centered at coordinates (x, y), or an ellipse centered at coordinates (x, y). The maximum soil thickness C within the scattering area is then extracted. t and average soil thickness G t .

[0081] The angle β of the cleaning gun n t Working head H0, maximum soil thickness C t and average soil thickness G t The input is fed into the working flow model Q(β,H,C,G) to generate the nozzle flow rate Q of the cleaning gun n at scanning time t. mnt , m is the number of the water supply branch where the cleaning gun n is located.

[0082] Based on the nozzle flow rate Q of the cleaning gun n at scanning time t mnt Calculate the orifice area of ​​the first throttle valve corresponding to cleaning gun n. (Orifice area of ​​the first throttle valve) g is the acceleration due to gravity, and the unit is meters per second. 2 .

[0083] Based on the nozzle flow rate Q of water supply branch m at scanning time t mt Calculate the valve orifice area of ​​the second throttle valve corresponding to water supply branch m. (Valve orifice area of ​​the second throttle valve) .

[0084] Due to pipeline losses, a booster pump set and water supply branch require high input pressure; however, excessive input pressure leads to water waste. This invention adjusts the relief pressure of the booster pump set and water supply branch. When the input pressure is too high, a portion of the cleaning medium (exceeding the relief pressure) is guided into the first water tank, saving water. Furthermore, this embodiment also provides a method for adjusting the relief pressure of the booster pump set and water supply branch.

[0085] For the cleaning gun n in the water supply branch m, the branch flow rate of the cleaning gun n at scanning time t is Q. mnt The head loss of water supply branch m is N represents the number of cleaning guns in water supply branch m. The loss coefficient k... 2mn =2λL mn / (1800 2 π 2 D 5 λ is the pipe friction coefficient, L mn Let D be the pipe length from cleaning gun n to water supply branch m, and D be the pipe diameter. The allowable head at the input of water supply branch m at scanning time t is H0 + H. 2mt The pressure relief of the water supply branch is P. 1t =(H0+H 2mt ) ρg, where ρ is the density of the cleaning medium (water).

[0086] The flow rate of water supply branch m at scanning time t is: The head loss of the booster pump set is M represents the number of water supply branches. Loss coefficient k 3m =2λL m / (1800 2 π 2 D 5 L m Let m be the length of the pipeline from the water supply branch to the booster pump unit. The allowable head at the input of the booster pump unit at scan time t is H0 + H. 3t +H 2t , The pressure relief P of the booster pump unit 2t =(H0+H 3t +H 2t )ρg. Example 5

[0087] This embodiment further discloses a preferred method for constructing the operating flow rate model in step 3. The operating flow rate model Q=Q(β,H,C,G) in this embodiment represents the relationship between the deflection angle β, operating head H, maximum soil thickness C, average soil thickness G, and cleaning flow rate Q. To improve the accuracy of the operating flow rate model, this embodiment first obtains the theoretical structure of the operating flow rate model from a theoretical perspective, then uses a single-factor method to construct the relationship between each factor and the cleaning flow rate, determines the coefficient of each variable, and then generates the operating flow rate model. This embodiment uses single-factor analysis; in more specific embodiments, orthogonal analysis can also be used to optimize the coefficients of the variables.

[0088] This embodiment mainly involves the core variables: deflection angle β (unit: rad), working head H (unit: m), maximum soil thickness C (unit: mm), and average soil thickness G (unit: mm). In addition to the variables involved in this embodiment, the cleaning flow rate may also be affected by the valve structure of the cleaning gun, the medium temperature, the wind speed, etc. These are not related to the input variables of this invention, and this embodiment will not consider them in more detail.

[0089] Step 310: Create the model structure for the operational flow rate model. The essence of soil washing is that the impact force of a high-pressure flow rate is greater than the soil adhesion force. The deflection angle β affects the direction of water inflow, thus affecting the direction of the impact force. The deflection angle with the highest impact efficiency should be a value between 0 and π. That is, when the impact force is constant, the closer to a certain deflection angle, the higher the washing efficiency. Therefore, Q∝1 / sin(β+β0). When the impact force per unit area is constant, the water velocity is inversely proportional to the washing flow rate Q, where water velocity = ,therefore Where g is the acceleration due to gravity. The energy required to peel off a unit area of ​​soil is related to the soil thickness, therefore Q∝(r²C+r³G), where r² is the peeling coefficient related to the maximum thickness and r³ is the peeling coefficient related to the average thickness. In summary, ,Right now r4 is the linear regression coefficient.

[0090] Step 310: The relationship between the deflection angle β and the cleaning flow rate Q. (For example...) Figure 16 As the deflection angle β increases, the overall trend of the cleaning flow rate Q is roughly a symmetrical trigonometric function, with an optimal spray angle range of about π / 3 (60°). Therefore, β0 is, for example, π / 6.

[0091] Step 330: The relationship between the working head H and the cleaning flow rate Q. For example... Figure 17 As the working head H increases, the demand for cleaning flow rate Q gradually decreases, showing an overall negative exponential relationship. Figure 17 Some outliers may be detection errors, and they are removed in this embodiment. r1 is, for example, 15 to 20 m / s².

[0092] Step 340: Relationship between average soil thickness G, maximum soil thickness C, and washing flow rate Q. (Refer to...) Figure 18 The average soil thickness G, maximum soil thickness C, and washing flow rate have a nearly linear relationship; when the values ​​are very large, the relationship approaches an exponential one. Since the soil thickness on engineering vehicles is typically between 0 and 10 mm, this can be simplified to a linear relationship. For example, r² is 0.4 × 10⁻⁶. -3 r3 is, for example, 1.5 × 10 -3 .

[0093] Step 350: Workflow Model. Combining the coefficients β0, r1, r2, and r3 determined in steps 320, 330, and 340, a linear regression coefficient r4 is fitted. For example, r4 is 2m. 3 ×s -1 Therefore, a task flow model can be constructed, namely... In this embodiment, .

[0094] For example, given an angle of π / 3, a working head of 5m, a maximum soil thickness of 2mm, and an average soil thickness of 0.5mm, the required cleaning flow rate Q, determined by the operating flow rate model, is 3.1 × 10⁻⁶. -4 m 3 ×s -1 That is, 18.6 L / min. Example 6

[0095] This embodiment further discloses a preferred method for adjusting the working head and cleaning speed. The cleaning speed in this invention is the angular velocity of the cleaning gun's oscillation. Currently, the engineering field mainly uses visual saliency to evaluate the stain index. This embodiment determines the stain index through a weighted algorithm and accordingly determines the adjustment amount of the working head and cleaning speed. The area saliency weight and thickness saliency weight involved in this embodiment can be set to fixed values ​​according to construction management requirements, or they can be continuously updated according to an adaptive learning algorithm to adapt to changes in the engineering construction environment.

[0096] Extract the maximum soil area A from the second distribution map. max Average soil area A avg Maximum soil thickness C max and average soil thickness G avg Calculate the stain index. Stain Index = α1(A max A avg ) / A 2 +(1-α1)(C max G avg ) / C 2α1 is the area significance weight; generally, the area significance of soil is greater than the thickness significance, α1 > 0.6. A is the soil area allowed by the construction management documents; the soil area allowed for dump trucks in urban areas is usually 0.1㎡. C is the soil thickness allowed by the construction management documents, usually 0.2 to 0.5cm.

[0097] Calculate the difference plot of the first and second distribution maps, and extract the maximum and average soil change rates from the difference plot. For any coordinate point of soil thickness data S1 in the first distribution map and the corresponding coordinate point of soil thickness data S2 in the second distribution map, the soil thickness data S3 for the corresponding coordinate point in the difference plot is S3 = (S1-S2) / S1. The maximum soil change rate S... max The maximum value of the soil thickness data in the difference plot, and the average soil change rate S avg This represents the average soil thickness data in the difference image. This embodiment uses an image difference algorithm to predict the cleaning effect of the previous cycle, avoiding situations where a large soil area is encountered by a single engineering vehicle, which could affect the cleaning speed and the accuracy of the working head control.

[0098] The adjustment amount for the washing speed is calculated based on the maximum soil area, average soil area, maximum soil change rate, and average soil change rate. The adjustment amount for the working head is calculated based on the maximum soil thickness, average soil thickness, maximum soil change rate, and average soil change rate. First, the soil area influence value (A) is calculated from the maximum and average soil areas. max A avg ) / A 2 The change impact value exp(-S) is calculated from the maximum soil change rate and the average soil change rate. max S avg The adjustment amount of the cleaning speed is Δω = -K. ω [(A max A avg ) / A 2 ]×exp(S max S avg )ω base K ω For velocity gain, ω base The base angular velocity is typically 2π / min. Then, the soil area influence value (C) is calculated from the maximum soil thickness and the average soil thickness. max G avg ) / C 2 The adjustment amount of the working head ΔH = K H [(C max G avg ) / C 2 ] ×(S max S avg H base KH For head gain, H base The base head is typically 3 to 6 meters. Example 7

[0099] like Figures 2 to 13 As shown, a cleaning control system for implementing the cleaning control method for the engineering vehicle according to the present invention includes: a truss 100, a conveyor belt 200, an image acquisition device 600, an image processing device, a data processing device, multiple sets of cleaning guns 300, multiple sets of water supply branches 400, a booster pump set 500, a control device 800, and a cleaning device 700.

[0100] The truss 100 is assembled from multiple sets of crossbeams and longitudinal beams. A conveyor belt 200 is mounted on the truss 100 and is configured to move the engineering vehicle 900. The conveyor belt 200 has a first sensing unit 201 for sensing a first position and a second sensing unit 202 for sensing a second position. The conveyor belt 200 also has a third sensing unit 203 for sensing a third position. A scrubbing station is located between the first and third positions, where a cleaning device 700, such as a brush, is positioned. This cleaning device can move along the truss 100 to different sides of the engineering vehicle 900. A rinsing station is located between the first and third positions, where a cleaning gun 300 is positioned between the second and third positions.

[0101] Image acquisition devices 600 are configured to acquire first and second images of the engineering vehicle 900. One set of image acquisition devices 600 is located on the crossbeam, and two sets of image acquisition devices 600 are located on the longitudinal beam. (Example) Figure 5 and Figure 6 When the engineering vehicle 900 enters the first position, a first image and a second image are acquired directly in front of the engineering vehicle 900. When the engineering vehicle 900 enters the third position, a first image and a second image are acquired directly behind the engineering vehicle 900, as well as two sets of first images and second images from the oblique sides of the engineering vehicle 900. The image acquisition device 600 includes a visible light component and a spectral component. The first image acquired by the visible light component is used to analyze the contour of the soil, and the second image acquired by the spectral component is used to analyze the density of the soil. The image processing device is configured to generate a first distribution map and a second distribution map, and the relevant algorithm is described in Embodiment 2.

[0102] A data processing unit is configured to generate the nozzle flow rate of the cleaning gun 300 at each scan time. Multiple sets of cleaning guns 300 are configured to supply cleaning medium to the engineering vehicle 900; each cleaning gun 300 has a first throttle valve 301, and the cleaning gun 300 is, for example, a high-pressure water gun. Multiple water supply branches 400 are configured to supply cleaning medium to at least two sets of cleaning guns 300; each water supply branch 400 has a second throttle valve 401. A booster pump assembly 500 is configured to supply cleaning medium to the water supply branches 400. A control unit 800 generates the orifice area of ​​the first throttle valve 301, the orifice area of ​​the second throttle valve 401, and the relief pressure of the booster pump assembly 500.

[0103] like Figure 13 In this invention, the cleaning control system further includes a first water tank 505, which is mainly used to store the cleaning medium, such as industrial tap water. The first water tank 505 is connected to a booster pump assembly 500. The booster pump assembly 500 feeds back to the first water tank 505 via a first pressure relief valve 501, and a first throttle valve 301 returns to the first water tank 505 via a second pressure relief valve 402. After the booster pump assembly 500 operates, the cleaning medium flows through gate valves 502 to different water supply branches 400, and then reaches the cleaning gun 300.

[0104] Because the water supply branch 400 is relatively long and has a large number of throttling valves, the pipeline head loss is significant. This invention calculates the head loss of the water supply branch 400 based on the nozzle flow rates of multiple sets of cleaning guns 300, then calculates the pressure relief of the second pressure relief valve 402, calculates the head loss of the booster pump set 500 based on the branch flow rates of the multiple sets of water supply branches 400, and then calculates the pressure relief of the first pressure relief valve 501. The cleaning medium from the booster pump set 500, with a pressure higher than the pressure relief of the first pressure relief valve 501, returns to the first water tank 505 through the return pipe. Similarly, the cleaning medium from the water supply branch 400, with a pressure higher than the pressure relief of the second pressure relief valve 402, also returns to the first water tank 505 through the return pipe, thus avoiding water waste. A one-way valve 504 is installed between the water supply branch 400 and the second pressure relief valve 402 to prevent sewage backflow.

[0105] When segmenting the first distribution map, some cleaning guns 300 have no sub-image in their working area, meaning these cleaning guns 300 are not working and their valve port area is 0 (closed state). This closed state of the cleaning gun 300 may increase the water supply pressure to other cleaning guns 300, leading to some loss of cleaning media. Therefore, the cleaning control system of this invention also includes a second water tank 506, and the water supply branch 400 includes a two-position three-way valve 503. In the cleaning state, the second throttle valve 401 is connected to the first throttle valve 301 via the two-position three-way valve 503. In the closed state, the second throttle valve 401 is connected to the second water tank 506 via the two-position three-way valve 503, and the cleaning media corresponding to the cleaning gun 300 is not lost but stored in the independent second water tank 506. After the booster pump group 500 finishes working, the second water tank 506 is then connected to the first water tank 505, further conserving water resources.

[0106] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for controlling the cleaning of engineering vehicles, characterized in that, Includes the following steps: Step 1: The booster pump set is connected to multiple water supply branches, each water supply branch is connected to multiple cleaning guns, and the multiple cleaning guns form a working surface. A working area is allocated to each cleaning gun, and the working head and cleaning speed of the cleaning gun are initialized. Step 2: The engineering vehicle enters the first position, and the first and second images of the engineering vehicle are acquired. Based on the working surface, the first and second images are transformed into surface images and thickness images respectively, generating the first distribution map of the soil. According to the working area, the first distribution map is divided into multiple sub-images, and the scanning trajectory of the cleaning gun in the corresponding sub-image is generated. Step 3: Generate the nozzle flow rate of the cleaning gun at each scan time based on the sub-image and the operation flow model, calculate the valve orifice area of ​​the first throttle valve of the cleaning gun at each scan time, and calculate the valve orifice area of ​​the second throttle valve of the water supply branch at each scan time. Step 4: The engineering vehicle enters the second position, the cleaning gun operates along the corresponding scanning trajectory, the first throttle valve and the second throttle valve maintain the corresponding valve orifice area, and after the cleaning gun finishes its operation, the engineering vehicle exits the second position. Step 5: Collect the third and fourth images of the engineering vehicle, generate the second distribution map of the soil, and calculate the stain index. If the stain index is greater than the stain threshold, proceed to step 6; otherwise, the engineering vehicle exits the first position and the task ends. Step 6: Adjust the working head and cleaning speed of the cleaning gun according to the first and second distribution diagrams, and return to step 2. In step 2, the first image is a visible light image. Multiple sets of coordinate transformation parameters between the first image and the working surface are generated. Based on these coordinate transformation parameters, the multiple sets of first images are combined into a surface image. The second image is a spectral image. Based on these coordinate transformation parameters, multiple sets of second images are combined into a thickness image. Soil contour data is extracted from the surface image, and soil thickness data is extracted from the thickness image. The soil contour data and soil thickness data are combined to generate a first distribution map. In step 2, the working area of ​​the cleaning gun n is a rectangular region with coordinates (x, y) satisfying x1≤x≤x2 and y1≤y≤y2. Based on the working area, the starting coordinates (x1, y1) and ending coordinates (x2, y2) of the scanning trajectory are determined. Then, the spacing of the scanning trajectory is determined based on the scattering area of ​​the cleaning gun. In step 3, the scanning time t and the deflection angle β of the working center line of the cleaning gun n moving to coordinates (x, y) are generated by combining the cleaning speed ω. t Extract the scattering area of ​​the cleaning gun at scanning time t, and calculate the maximum soil thickness C of this scattering area. t and average soil thickness G t The angle β of the cleaning gun n t Working head H0, maximum soil thickness C t and average soil thickness G t Inputting the data into the operational flow model generates the nozzle flow rate Q of the cleaning gun n at scanning time t. mnt , m is the number of the water supply branch where the cleaning gun n is located.

2. The cleaning control method for engineering vehicles according to claim 1, characterized in that, According to the nozzle flow rate Q mnt Calculate the valve orifice area of ​​the first throttle valve corresponding to cleaning gun n at scanning time t, based on the branch flow rate Q of water supply branch m at scanning time t. mt Calculate the valve port area of ​​the second throttle valve corresponding to water supply branch m.

3. The cleaning control method for engineering vehicles according to claim 1, characterized in that, In step 4, the head loss of the booster pump set and the water supply branch is predicted, and the pressure relief of the booster pump set and the water supply branch is adjusted according to the head loss.

4. The cleaning control method for engineering vehicles according to claim 1, characterized in that, In step 5, the maximum soil area, average soil area, maximum soil thickness, and average soil thickness are extracted from the second distribution map to calculate the stain index.

5. The cleaning control method for engineering vehicles according to claim 1, characterized in that, In step 6, the difference map of the first distribution map and the second distribution map is calculated, the maximum soil change rate and the average soil change rate are extracted from the difference map, the adjustment amount of the cleaning speed is calculated based on the maximum soil area, the average soil area, the maximum soil change rate and the average soil change rate, and the adjustment amount of the working head is calculated based on the maximum soil thickness, the average soil thickness, the maximum soil change rate and the average soil change rate.

6. A control system for implementing the cleaning control method for engineering vehicles according to claim 1, characterized in that, include: The conveyor belt is configured as a mobile engineering vehicle, and the conveyor belt has a first sensing unit for sensing a first position and a second sensing unit for sensing a second position. The image acquisition device is configured to acquire a first image, a second image, a third image, and a fourth image of the engineering vehicle; An image processing apparatus is configured to generate a first distribution map and a second distribution map; The data processing unit is configured to generate the nozzle flow rate of the cleaning gun at each scan time. Multiple sets of cleaning guns are configured to provide cleaning media to engineering vehicles, and each cleaning gun has a first throttle valve; Multiple water supply branches are configured to supply cleaning media to at least two sets of cleaning guns, and each water supply branch has a second throttle valve; The booster pump set is configured to provide cleaning medium to the water supply branch; The control device is configured to generate the valve orifice area of ​​the first throttle valve, the valve orifice area of ​​the second throttle valve, and the relief pressure of the booster pump set and the water supply branch.

7. The control system according to claim 6, characterized in that, The control system also includes a truss and a cleaning device. The conveyor belt is mounted on the truss and has a third sensing unit that senses a third position. The cleaning device is located between the first position and the third position, and the cleaning gun is located between the third position and the second position.

8. The control system according to claim 6, characterized in that, The control system also includes a first water tank connected to a booster pump set. The booster pump set feeds back to the first water tank via a first pressure relief valve, and a second throttle valve returns to the first water tank via a second pressure relief valve. The head loss of the water supply branch is calculated based on the nozzle flow rate of multiple sets of cleaning guns, and then the pressure relief of the second pressure relief valve is calculated. The head loss of the booster pump set is calculated based on the branch flow rate of multiple sets of water supply branches, and then the pressure relief of the first pressure relief valve is calculated.

Images