Method for controlling photovoltaic cleaning robot with ultrasonic waves cooperating with intelligent dust collection

Abstract

The invention discloses a control method for a photovoltaic cleaning robot with ultrasonic waves cooperating with intelligent dust collection, and belongs to the technical field of photovoltaic cleaning. The method specifically comprises the steps that photovoltaic panel dirt image information, photovoltaic panel parameters and photovoltaic cleaning robot parameters are collected, ultrasonic vibration parameters acting on a roller brush are calculated, the roller brush is driven to vibrate and clean photovoltaic panel dirt, optical feedback data of the cleaned photovoltaic panel are obtained, and self-adaptive suction parameters are calculated by analyzing the dirt stripping state. The dust collection component is controlled according to the self-adaptive suction parameters, dirt is sucked into the interior through array holes in the surface of the roller brush, and finally the collected dirt is discharged to a designated area; the cleaning efficiency is improved, meanwhile, the damage to the plate surface is avoided, the dependence on water resources and external energy sources is reduced, and the cleaning cost is reduced.

Description

Technical Field

[0001] This invention relates to a control method for a photovoltaic cleaning robot with ultrasonic collaborative intelligent vacuuming, belonging to the field of photovoltaic cleaning technology. Background Technology

[0002] Considering environmental factors and economic benefits, photovoltaic panels are usually installed in high-altitude mountainous areas, Gobi deserts, and coastal areas. However, long-term outdoor operation can lead to the accumulation of dirt such as dust, sand, and bird droppings on the surface of photovoltaic panels, which can significantly reduce their photoelectric conversion efficiency. Due to the geographical influence, these areas are generally sparsely populated, and traditional manual cleaning methods for photovoltaic panels have problems such as low efficiency, high cost, and poor safety. Therefore, there is a greater reliance on intelligent cleaning equipment to clean photovoltaic panels.

[0003] In existing photovoltaic cleaning robot technologies, water spray is typically used to clean photovoltaic panels. While this method can effectively remove dirt, it remains dependent on water resources and carries the risk of the panels freezing at low temperatures. Electric curtain dust removal technology can transport and collect dirt along a specific path under the influence of an electric field. However, existing technologies have the following problems: it is difficult to balance cleaning efficiency and panel protection, the cleaning cost is high and there is secondary pollution, and there is an over-reliance on external resources. Summary of the Invention

[0004] The purpose of this invention is to provide a control method for a photovoltaic cleaning robot with ultrasonic collaborative intelligent vacuuming, so as to solve the problems of poor cleaning efficiency, high cost and safety of the photovoltaic panel in the prior art.

[0005] To solve the above-mentioned technical problems, the present invention is implemented using the following technical solution.

[0006] A control method for a photovoltaic cleaning robot with ultrasonic-assisted intelligent vacuuming includes:

[0007] Collect image information of dirt on photovoltaic panel surface, photovoltaic panel surface parameters and photovoltaic cleaning robot parameters, the photovoltaic cleaning robot parameters include the physical characteristics of roller brush, effective working thickness and pressure response time constant;

[0008] The ultrasonic vibration parameters acting on the roller brush of the photovoltaic cleaning robot are calculated based on the image information of the dirt on the photovoltaic panel, so that the robot vibrates and transmits the vibration to the photovoltaic panel through the roller brush to clean the photovoltaic panel.

[0009] Obtain optical feedback data of the photovoltaic panel after cleaning, analyze the dirt removal status, and calculate adaptive suction parameters based on the dirt removal status;

[0010] Based on the adaptive suction parameters, the dust collection components of the photovoltaic cleaning robot are controlled to generate negative pressure in the internal cavity of the roller brush, which draws dirt into the roller brush through the array of holes on the surface of the roller brush.

[0011] The contaminants sucked into the roller brush are discharged to a designated area.

[0012] Specifically, based on the image information of dirt on the photovoltaic panel surface, the ultrasonic vibration parameters acting on the roller brush of the photovoltaic cleaning robot are calculated, causing it to vibrate and transmit the vibration to the photovoltaic panel surface via the roller brush, thus cleaning the photovoltaic panel surface, including:

[0013] Extract the optical and spatial distribution features of the contaminated areas from the contaminated image information of the photovoltaic panel to determine the contamination type, coverage density, and local average thickness.

[0014] Based on the type of contamination, coverage density, and local average thickness, a mapping relationship is established with the initial ultrasonic vibration frequency parameters and the initial ultrasonic vibration amplitude parameters to construct the initial ultrasonic vibration parameter set.

[0015] Based on the photovoltaic panel parameters and the physical characteristics of the roller brush, an energy transfer attenuation model for ultrasonic vibration energy is constructed. The initial set of ultrasonic vibration parameters is used as the input to the energy transfer attenuation model to calculate the vibration energy.

[0016] Based on the comparison results between the vibration energy and the preset energy threshold, the initial ultrasonic vibration frequency parameters and the initial ultrasonic vibration amplitude parameters are iteratively optimized to obtain the ultrasonic vibration parameter set.

[0017] Based on the spatial distribution characteristics and the set of ultrasonic vibration parameters, the roller brush is driven to generate vibration and the vibration is transmitted to the photovoltaic panel surface for cleaning.

[0018] Specifically, based on the photovoltaic panel parameters and the physical characteristics of the roller brush, an energy transfer attenuation model for ultrasonic vibration energy is constructed. The initial set of ultrasonic vibration parameters is used as input to the energy transfer attenuation model to calculate the vibration energy, including:

[0019] The photovoltaic panel parameters and the physical properties of the roller brush are standardized to generate basic input parameters. Based on the basic input parameters, the propagation speed and attenuation constant of ultrasonic vibration in the roller brush are calculated.

[0020] Based on the propagation speed and attenuation constant, the energy transmittance function of ultrasonic vibration energy transferred from the roller brush to the photovoltaic panel is derived. Combined with the energy reflection coefficient and transmission coefficient of the photovoltaic panel surface, an energy transfer attenuation model is constructed.

[0021] The initial ultrasonic vibration frequency parameters and initial ultrasonic vibration amplitude parameters are converted into initial vibration energy density and input into the energy transfer attenuation model. The remaining vibration energy density is calculated using the energy transmittance function and the attenuation constant.

[0022] The remaining vibration energy density is spatially integrated over the surface of the photovoltaic panel to calculate the total vibration energy value acting on the entire area of ​​the photovoltaic panel as the vibration energy.

[0023] Specifically, the initial ultrasonic vibration frequency parameters and initial ultrasonic vibration amplitude parameters are converted into initial vibration energy density and input into the energy transfer attenuation model. The remaining vibration energy density is calculated using the energy transmittance function and the attenuation constant, including:

[0024] Based on the physical properties of the roller brush and the initial ultrasonic vibration amplitude parameters, the maximum kinetic energy of the roller brush per unit volume during the vibration process is calculated, and the initial mechanical energy density value is calculated in combination with the initial ultrasonic vibration frequency parameters.

[0025] Multiply the attenuation constant by the effective thickness of the roller brush, and then raise the product to the negative power of the natural exponential function to obtain the exponential attenuation factor during the process of ultrasonic vibration penetrating the thickness of the roller brush.

[0026] The theoretical vibration energy density of the photovoltaic panel is obtained by multiplying the initial mechanical energy density value by the energy transmission ratio represented by the energy transmittance function.

[0027] By multiplying the exponential decay factor by the theoretical vibration energy density, the residual vibration energy density formed on the surface of the photovoltaic panel after the ultrasonic vibration has passed through the internal attenuation and transmission loss of the roller brush is determined.

[0028] Specifically, the process involves acquiring optical feedback data from the cleaned photovoltaic panel, analyzing the dirt removal status, and calculating adaptive suction parameters based on the dirt removal status, including:

[0029] Obtain optical feedback data of the photovoltaic panel after cleaning, detect the pixel area of ​​the residual dirt area in the photovoltaic panel, compare it with the photovoltaic panel before cleaning, and calculate the area residual ratio and grayscale change ratio of the dirt.

[0030] The area residual ratio and grayscale change ratio are used as inputs for fuzzification, mapped to linguistic variables and membership functions are set respectively.

[0031] By using the area residual ratio and gray scale change ratio as preconditions and the suction level as postconditions, reasoning rules are established, and the rule set is obtained by taking the largest value operation in combination with the membership function.

[0032] The membership function curve of the constructed rule set is combined with the abscissa value corresponding to the centroid of the area enclosed by the preset coordinate system and mapped to the physical control range of the vacuuming component. The adaptive suction parameters are obtained through linear transformation.

[0033] Specifically, the membership function curve of the constructed rule set is combined with the abscissa value corresponding to the centroid of the area enclosed by the preset coordinate system, and mapped to the physical control range of the vacuum component. Adaptive suction parameters are obtained through linear transformation, including:

[0034] Based on the trigger strength of the reasoning rules in the rule set, the corresponding membership function is truncated, and the maximum value of the truncated membership function is taken on the horizontal axis of the preset coordinate system to generate the membership function curve.

[0035] Set a discrete point sequence on the horizontal axis, discretize the membership function curve, calculate the membership value of the rule set at the discrete points, and transform the membership function curve into a membership value sequence.

[0036] Multiply the x-coordinates of all discrete points by their corresponding membership values ​​and sum them to obtain the static moment of the rule set. Sum the membership values ​​of all discrete points to obtain the total membership.

[0037] Based on the quotient of the static moment and the sum of the membership degrees, determine the abscissa value of the centroid corresponding to the centroid of the area enclosed by the membership function curve of the rule set and the coordinate system.

[0038] The horizontal coordinate value of the center of gravity is input into a preset linear scaling function and converted into an adaptive suction parameter that controls the negative pressure generated by the vacuuming component. The conversion range is the minimum and maximum value of the physical control quantity of the vacuuming component.

[0039] Specifically, based on the trigger strength of the inference rules in the rule set, the corresponding membership function is truncated. The maximum value of the truncated membership function is then taken on the horizontal axis of a preset coordinate system to generate a membership function curve, including:

[0040] Obtain the trigger strength and membership function corresponding to the triggered inference rule, truncate the membership function in the vertical direction, and the height of the upper boundary of the truncation is the trigger strength corresponding to the inference rule to obtain the local membership function;

[0041] Set the local membership function to the horizontal axis, read the corresponding local membership function value for each point on the horizontal axis, and if a point on the horizontal axis is not within the range of the local membership function, the membership degree is zero.

[0042] For each point on the horizontal axis, compare the membership values ​​of the local membership function at that point, select the maximum value as the membership value of that point, connect the maximum membership values ​​of all points, and generate the membership function curve.

[0043] Specifically, based on adaptive suction parameters, the suction components of the photovoltaic cleaning robot are controlled to create negative pressure inside the roller brush cavity, drawing dirt into the roller brush through the array of holes on its surface. This includes:

[0044] By establishing a mapping relationship between adaptive suction parameters and vacuuming components, the adaptive suction parameters are converted into target rotation speed values ​​of the vacuuming components, and corresponding target control signals are generated based on the target rotation speed values.

[0045] The target control signal is output to the dust collection component, and the negative pressure deviation value is obtained by comparing the actual negative pressure value of the internal cavity of the roller brush with the target negative pressure value determined by the adaptive suction parameters.

[0046] The negative pressure deviation value is subjected to proportional-integral control to calculate the control compensation amount used to correct the target control signal. The control compensation amount is superimposed with the target control signal to generate the adjusted real-time control signal and adjust the dust collection component.

[0047] The adjusted suction components bring the internal cavity of the roller brush to the target negative pressure value. Based on the pressure difference between the internal cavity of the roller brush and the external environment, formed by the array of holes on the surface of the roller brush, dirt is sucked in.

[0048] Specifically, proportional-integral control is applied to the negative pressure deviation value to calculate the control compensation amount used to correct the target control signal. This compensation amount is then superimposed on the target control signal to generate an adjusted real-time control signal, which is used to adjust the dust collection components. This includes:

[0049] The sampling period is set according to the pressure response time constant of the internal cavity of the roller brush. The negative pressure deviation value of the current sampling period is multiplied by the preset proportional coefficient to obtain the proportional adjustment component.

[0050] The negative pressure deviation values ​​of all sampling periods are summed up, and the sum is multiplied by the preset integral coefficient to obtain the integral adjustment component. The proportional adjustment component is added to the integral adjustment component to obtain the control compensation amount.

[0051] The control compensation amount is algebraically superimposed with the target control signal and sent to the dust collection component for adjustment.

[0052] A photovoltaic cleaning robot control system for ultrasonic collaborative intelligent vacuuming includes a dirt sensing module, an ultrasonic dirt vibration module, a suction power decision module, a negative pressure adsorption module, and an emission module.

[0053] The dirt sensing module is used to collect dirt image information, photovoltaic panel parameters, and photovoltaic cleaning robot parameters. The photovoltaic cleaning robot parameters include the physical characteristics of the roller brush, effective working thickness, and pressure response time constant.

[0054] The ultrasonic cleaning module is used to calculate the ultrasonic vibration parameters acting on the roller brush of the photovoltaic cleaning robot based on the image information of the dirt on the photovoltaic panel, so that it generates vibration and transmits it to the photovoltaic panel through the roller brush to clean the photovoltaic panel.

[0055] The suction power decision module is used to acquire optical feedback data of the photovoltaic panel after cleaning, analyze the dirt removal status, and calculate adaptive suction power parameters based on the dirt removal status.

[0056] The negative pressure adsorption module is used to control the dust collection component of the photovoltaic cleaning robot to generate negative pressure in the internal cavity of the roller brush according to the adaptive suction parameters, so as to draw dirt into the inside of the roller brush through the array of holes on the surface of the roller brush.

[0057] The discharge module is used to discharge the dirt inside the suction roller brush to a designated area.

[0058] Compared with existing technologies, the beneficial effects achieved by this invention are as follows: The ultrasonic vibration parameters of the photovoltaic cleaning robot are calculated based on the image information of the photovoltaic panel surface dirt, and the photovoltaic panel surface is vibrated to remove dirt. The dirt cleaning status is analyzed based on the optical feedback data of the photovoltaic panel surface before and after cleaning, and adaptive suction parameters are calculated. The photovoltaic cleaning robot is then controlled to vacuum through these adaptive suction parameters, and the vacuumed dirt is discharged to a designated area. This invention solves the problems of poor cleaning efficiency, high cost, and panel safety in existing technologies. It intelligently matches and dynamically adjusts ultrasonic energy and suction negative pressure for different types of dirt, improving cleaning efficiency while avoiding damage to the panel surface, reducing dependence on water resources and external energy, and lowering cleaning costs.

[0059] It should be understood that the above general description and the following detailed description are merely exemplary and explanatory, and are not intended to limit the technical solutions of this disclosure. Attached Figure Description

[0060] Figure 1 A flowchart of a photovoltaic cleaning robot control method for ultrasonic collaborative intelligent vacuuming provided by the present invention;

[0061] Figure 2 A schematic diagram of the membership function provided for this invention;

[0062] Figure 3 This invention provides a structural diagram of a photovoltaic cleaning robot control system for ultrasonic collaborative intelligent vacuuming. Detailed Implementation

[0063] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solution of the present invention, rather than limitations thereof. In the absence of conflict, the embodiments of the present invention and the technical features in the embodiments can be combined with each other.

[0064] The term "and / or" simply describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. Additionally, the character " / " generally indicates that the preceding and following related objects have an "or" relationship.

[0065] Example 1

[0066] Please see Figures 1-3 This invention provides an embodiment of a photovoltaic cleaning robot control method for ultrasonic collaborative intelligent vacuuming, which includes the following specific steps:

[0067] Step S1: Collect image information of the photovoltaic panel surface dirt, photovoltaic panel parameters and photovoltaic cleaning robot parameters. The photovoltaic cleaning robot parameters include the physical characteristics of the roller brush, effective working thickness and pressure response time constant.

[0068] It should be noted that the image information of the photovoltaic panel surface dirt is obtained by taking pictures of the photovoltaic panel surface before cleaning using an industrial camera installed on the photovoltaic cleaning robot. The photovoltaic panel surface parameters are obtained by reading the component model specification table in the photovoltaic power station database and measuring the ambient light intensity and temperature in real time. The photovoltaic cleaning robot parameters are retrieved from the preset storage unit of the robot controller and the posture angle of each joint and the current value of the drive motor are read in real time. The photovoltaic panel surface dirt image information includes, but is not limited to, the original RGB image of dirt distribution. The photovoltaic panel surface parameters include, but are not limited to, the material density, elastic modulus and thickness of the photovoltaic panel surface. The photovoltaic cleaning robot parameters include, but are not limited to, the robot's current travel speed, the physical characteristics of the roller brush, the effective working thickness and the pressure response time constant. The physical characteristics of the roller brush include, but are not limited to, the material density, elastic modulus, internal damping characteristic parameters, bristle length and bristle diameter of the roller brush.

[0069] Step S2: Calculate the ultrasonic vibration parameters acting on the roller brush of the photovoltaic cleaning robot based on the image information of the dirt on the photovoltaic panel, so that it generates vibration and transmits it to the photovoltaic panel through the roller brush to clean the photovoltaic panel.

[0070] The specific steps of step S2 are as follows:

[0071] Step S201: Extract the optical and spatial distribution features of the contaminated areas from the photovoltaic panel contamination image information to determine the contamination type, coverage density, and local average thickness.

[0072] In this embodiment, the acquired photovoltaic panel surface dirt image information is processed by grayscale conversion and noise filtering. Based on the processed image, an edge detection algorithm is used to identify and mark the contours of all dirt areas. According to the grayscale value distribution and texture features of the pixels within the contour, the dirt areas are divided into different dirt categories, such as dust deposition, floating object attachment, and biological residue. The area enclosed by each marked contour is calculated, and the sum of the areas of all dirt areas is compared with the total area of ​​the photovoltaic panel surface image to obtain the dirt coverage density. For each marked dirt area, the local average thickness of the dirt area is estimated by using a preset grayscale thickness mapping relationship based on the difference between the average grayscale value of the pixels within its contour and the reference grayscale value of the dirt of that category at a standard thickness. Specifically, by acquiring standard grayscale images of dirt under the same illumination and imaging conditions, the average grayscale value of the dirt area at each thickness is measured and recorded. By analyzing the grayscale data of samples with different thicknesses, a corresponding data table between the pollutant thickness and the average grayscale value of the image is established, and a continuous function relationship describing the changing trend of the two is fitted to establish a grayscale thickness mapping relationship.

[0073] Step S202: Based on the type of dirt, coverage density and local average thickness, establish a mapping relationship between the initial ultrasonic vibration frequency parameters and the initial ultrasonic vibration amplitude parameters, and construct the initial ultrasonic vibration parameter set.

[0074] In this embodiment, a preset reference vibration frequency correspondence table and reference vibration amplitude correspondence table are consulted according to the type of contamination to determine the initial ultrasonic vibration frequency reference value and the initial ultrasonic vibration amplitude reference value corresponding to that type of contamination. Based on the local average thickness, an amplitude compensation amount proportional to the thickness is added to the initial ultrasonic vibration amplitude reference value. The initial ultrasonic vibration frequency reference value is fine-tuned according to the calculated coverage density; the higher the density, the lower the frequency fine-tuning value. For areas marked as contaminated, a corresponding combination including the adjusted initial ultrasonic vibration frequency parameters and initial ultrasonic vibration amplitude parameters is generated. All combinations constitute the initial ultrasonic vibration parameter set, wherein the amplitude compensation amount is determined by… Those skilled in the art can set up the system according to the actual situation, test the inherent resonant frequency range of each type of dirt, and use ultrasonic vibration sources of different frequencies to test the cleaning effect. They can record the specific frequency range that can cause efficient structural loosening or resonant peeling of the dirt layer, establish the median value of the frequency range as the reference vibration frequency of the type of dirt, and establish a reference vibration frequency correspondence table. By testing the minimum detachment force required to separate various types of dirt per unit area, the minimum detachment force is converted into the minimum vibration acceleration required to generate inertial force to achieve detachment of the dirt. Then, combined with the equivalent mass of the vibration system, the minimum vibration displacement amplitude required is calculated, that is, the initial ultrasonic vibration amplitude reference value, and a reference vibration amplitude correspondence table is established.

[0075] Step S203: Based on the photovoltaic panel parameters and the physical characteristics of the roller brush, construct an energy transfer attenuation model for ultrasonic vibration energy, and use the initial ultrasonic vibration parameter set as the input to the energy transfer attenuation model to calculate the vibration energy.

[0076] The specific steps of step S203 are as follows:

[0077] Step S2031: Standardize the photovoltaic panel parameters and the physical characteristics of the roller brush to generate basic input parameters. Based on the basic input parameters, calculate the propagation speed and attenuation constant of ultrasonic vibration in the roller brush.

[0078] In this embodiment, based on the material density and elastic modulus in the photovoltaic panel parameters and the material density and elastic modulus in the physical properties of the roller brush, the propagation speed of ultrasonic vibration in the roller brush material and the propagation speed in the photovoltaic panel material are calculated using the formula for calculating the propagation speed of longitudinal waves in a uniform solid medium. Based on the internal damping characteristic parameters of the roller brush material and the propagation speed in the roller brush material, the attenuation constant of ultrasonic vibration in the roller brush is obtained by calculating the frequency-related sound absorption coefficient. The speed calculation formula is a prior art technique.

[0079] Step S2032: Based on the propagation speed and attenuation constant, derive the energy transmittance function of ultrasonic vibration energy transferred from the roller brush to the photovoltaic panel. Combine the energy reflection coefficient and transmission coefficient of the photovoltaic panel surface to construct an energy transfer attenuation model.

[0080] In this embodiment, based on the propagation speed of ultrasonic vibration in the roller brush material and the propagation speed in the photovoltaic panel material, the acoustic impedance ratio when the vibration wave travels from the roller brush material to the photovoltaic panel material is calculated. Based on the acoustic impedance ratio, the energy reflection coefficient and energy transmission coefficient of the vibration energy at the interface between the roller brush and the photovoltaic panel are calculated. Based on the density and propagation speed of the roller brush material and the photovoltaic panel material, the acoustic impedance values ​​of the roller brush and the photovoltaic panel are calculated. The ratio of the acoustic impedance value of the photovoltaic panel to the acoustic impedance value of the roller brush is calculated to obtain the acoustic impedance ratio. This acoustic impedance ratio is then substituted into the formula for calculating the acoustic energy reflection coefficient of a vertically incident plane wave. The numerator is the square of the acoustic impedance ratio minus 1, and the denominator is the square of the acoustic impedance ratio plus 1, which gives the energy reflection coefficient of vibration energy at the interface between the roller brush and the photovoltaic panel. Substituting the acoustic impedance ratio into the formula for calculating the acoustic energy transmission coefficient of a vertically incident plane wave, the numerator of the formula is four times the acoustic impedance ratio, and the denominator is the square of the acoustic impedance ratio plus 1, which gives the energy transmission coefficient of vibration energy at the interface between the roller brush and the photovoltaic panel. Multiplying the energy transmission coefficient by an exponential attenuation factor determined by the attenuation constant and the effective thickness of the roller brush, the product constitutes the energy transmittance function, which characterizes the total loss of vibration energy in the process of transferring from the inside of the roller brush to the surface of the photovoltaic panel.

[0081] Step S2033: Convert the initial ultrasonic vibration frequency parameters and initial ultrasonic vibration amplitude parameters into initial vibration energy density, and input them into the energy transfer attenuation model. Calculate the remaining vibration energy density using the energy transmittance function and the attenuation constant.

[0082] The specific steps of step S2033 are as follows:

[0083] Step S20331: Based on the physical characteristics of the roller brush and the initial ultrasonic vibration amplitude parameters, calculate the maximum kinetic energy of the roller brush per unit volume during the vibration process, and calculate the initial mechanical energy density value by combining the initial ultrasonic vibration frequency parameters.

[0084] In this embodiment, the maximum velocity of the roller brush material in simple harmonic motion is calculated based on the roller brush material density and the initial ultrasonic vibration amplitude parameter in the initial ultrasonic vibration parameter set. The maximum velocity is obtained by multiplying the initial ultrasonic vibration amplitude parameter by the initial ultrasonic vibration frequency parameter and then multiplying the product by twice the constant pi. Based on the roller brush material density and the calculated maximum velocity, the roller brush material density is multiplied by the square of the maximum velocity value to obtain the intermediate value of the maximum kinetic energy per unit volume of the roller brush material. The intermediate value of the maximum kinetic energy is multiplied by a preset coefficient to obtain the maximum kinetic energy that the roller brush material per unit volume can achieve within the vibration period. The maximum kinetic energy is defined as the initial mechanical energy density value, which represents the energy intensity at the vibration source. The coefficient is set by those skilled in the art according to the actual situation.

[0085] Step S20332: Multiply the attenuation constant by the effective thickness of the roller brush, and take the negative power of the natural exponential function of the product to obtain the exponential attenuation factor during the process of ultrasonic vibration penetrating the thickness of the roller brush.

[0086] In this embodiment, the attenuation constant is multiplied by the effective thickness of the roller brush to obtain a dimensionless product representing the total attenuation effect. An exponential function with the natural constant as the base of the dimensionless product is then calculated, and the negative power of the exponential function is obtained to obtain the exponential attenuation factor experienced by the ultrasonic vibration during the process of penetrating the entire thickness of the roller brush. Its value ranges from 0 to 1, representing the proportion of energy attenuation caused by the internal damping of the material during the process of the ultrasonic vibration passing through the effective thickness of the roller brush.

[0087] Step S20333: The theoretical vibration energy density of the photovoltaic panel is obtained by multiplying the initial mechanical energy density value with the energy transmission ratio represented by the energy transmittance function.

[0088] In this embodiment, according to the energy transfer attenuation model, the energy transmittance function is a function of the initial ultrasonic vibration frequency parameter. The energy transmittance function describes the proportion of vibration energy passing through the interface between the roller brush and the photovoltaic panel. Its function expression includes the frequency parameter as an independent variable. Substituting the initial ultrasonic vibration frequency parameter into the energy transmittance function, a definite value between 0 and 1 corresponding to the initial ultrasonic vibration frequency parameter is obtained, which is the energy transmittance value at the current frequency. The initial mechanical energy density value, which is a scalar value representing energy density, is multiplied by the energy transmittance value, and the product is the theoretical vibration energy density.

[0089] Step S20334: Multiply the exponential decay factor by the theoretical vibration energy density to determine the residual vibration energy density formed on the photovoltaic panel surface after the ultrasonic vibration has passed through the internal attenuation and transmission loss of the roller brush.

[0090] In this embodiment, the exponential decay factor and the theoretical vibration energy density are calculated by floating-point multiplication. The theoretical remaining energy after interface transmission is determined by multiplication, and the volume loss generated in the propagation path inside the roller brush material is deducted. Thus, after the vibration energy undergoes propagation attenuation in the roller brush body and energy dissipation mechanism of interface transmission loss, the effective energy density that can be formed on the surface of the photovoltaic panel to destroy the dirt adhesion structure is finally obtained. The product is taken as the remaining vibration energy density. The remaining vibration energy density quantifies the energy intensity that actually acts on the cleaning target after two stages of attenuation, including volume attenuation and interface loss.

[0091] Step S2034: Integrate the remaining vibration energy density spatially over the surface of the photovoltaic panel to calculate the total vibration energy value acting on the entire area of ​​the photovoltaic panel as the vibration energy.

[0092] In this embodiment, based on the spatial distribution characteristics of the contaminated areas and the corresponding residual vibration energy density, the geometric area of ​​each contaminated area's contour boundary information is calculated. For regular contours, the corresponding planar geometric area formula is used for calculation. For irregular contours, the pixel statistical method is used to decompose the internal region of the contour into a series of basic geometric shapes, such as triangles. The area of ​​each basic shape is calculated and then summed to obtain the area value of the contaminated area. For each contaminated area, the area value of the contaminated area is multiplied by its corresponding residual vibration energy density value. Through the multiplication operation, the effective vibration energy intensity acting on a unit area of ​​the area is extended to the entire contaminated area area, thereby calculating the total vibration energy required to clean the specific contaminated area and obtaining the local area vibration energy value. The local area vibration energy values ​​of all contaminated areas are calculated and summed to obtain the total vibration energy value acting on all contaminated areas on the photovoltaic panel surface. For areas on the photovoltaic panel surface that are not covered by contamination, since there is no dirt on their surface that needs to be overcome, the required cleaning vibration energy is set to zero.

[0093] Step S204: Based on the comparison results between the vibration energy and the preset energy threshold, iteratively optimize the initial ultrasonic vibration frequency parameters and the initial ultrasonic vibration amplitude parameters to obtain the ultrasonic vibration parameter set.

[0094] In this embodiment, the total vibration energy value is compared with a preset minimum energy threshold for the current type of contamination. If the total vibration energy value is less than the minimum energy threshold, the parameter optimization process is initiated. The initial ultrasonic vibration amplitude parameter is increased by a fixed step size while keeping the adjusted amplitude parameter unchanged. The initial ultrasonic vibration frequency parameter is adjusted to search for a better resonance frequency point. The new parameter combination is used to iteratively perform calculations until the recalculated total vibration energy value is greater than the minimum energy threshold. The final determined initial ultrasonic vibration frequency parameter and initial ultrasonic vibration amplitude parameter that make the total vibration energy value meet the standard are used as the optimized ultrasonic vibration parameter combination. The optimization process is then performed on all contaminated areas, and the optimized parameter combinations of all contaminated areas are collected to form the final ultrasonic vibration parameter set. The minimum energy threshold is set by those skilled in the art according to the actual situation. During iterative optimization, the initial ultrasonic vibration frequency parameter must not exceed the upper limit of the frequency tolerance of the photovoltaic panel material.

[0095] Step S205: Based on the spatial distribution characteristics and the set of ultrasonic vibration parameters, drive the roller brush to generate vibration and transmit it to the photovoltaic panel surface for cleaning.

[0096] In this embodiment, based on the extracted spatial distribution characteristics of dirt, the cleaning path of the photovoltaic cleaning robot's roller brush and the required dwell time in each dirty area are determined. Based on the roller brush cleaning path and the working time, the specific vibration frequency and vibration amplitude values ​​corresponding to each working position are extracted sequentially from the ultrasonic vibration parameter set. The vibration frequency and vibration amplitude values ​​are combined into a time-continuous drive parameter sequence, and the drive parameter sequence is converted into a drive signal. The drive signal is sent to the ultrasonic vibration component linked with the roller brush to generate corresponding high-frequency mechanical vibration. This high-frequency mechanical vibration is transmitted to the surface of the photovoltaic panel through the roller brush body, applying a periodic force to the dirt attached to it, thereby achieving cleaning.

[0097] Step S3: Obtain optical feedback data of the cleaned photovoltaic panel, analyze the dirt removal status, and calculate adaptive suction parameters based on the dirt removal status.

[0098] The specific steps of step S3 are as follows:

[0099] Step S301: Obtain optical feedback data of the photovoltaic panel after cleaning, detect the pixel area of ​​the residual dirt area in the photovoltaic panel, compare it with the photovoltaic panel before cleaning, and calculate the residual dirt area ratio and grayscale change ratio.

[0100] In this embodiment, optical feedback data of the photovoltaic panel after cleaning is acquired, and the same grayscale and noise filtering preprocessing is performed on it as on the image before cleaning. An edge detection algorithm is used to identify and mark the contours of all residual dirt areas in the preprocessed image after cleaning. The pixel area of ​​each marked contour is calculated, and the sum of the pixel areas of all residual dirt areas is recorded as the total dirt area after cleaning. The total dirt area after cleaning is divided by the total dirt area before cleaning, and the quotient is the dirt area residual ratio. For each original dirt area before cleaning, the average grayscale value of the corresponding area in the image after cleaning is calculated. The average grayscale value of the pixel is subtracted from the average grayscale value of the pixel before cleaning, and the difference is divided by the average grayscale value before cleaning. The quotient is the grayscale change ratio of that area. The arithmetic mean of the grayscale change ratios of all areas is taken as the overall grayscale change ratio.

[0101] Step S302: The area residual ratio and grayscale change ratio are used as inputs for fuzzification processing, mapped to linguistic variables respectively, and membership functions are set.

[0102] In this embodiment, linguistic variables are set for the area residual ratio, namely, the first-level residual area, the second-level residual area, and the third-level residual area. A triangle membership function is set for each linguistic variable, defining its mathematical shape and universe of discourse coverage by setting the horizontal coordinates of the three vertices of the triangle. Three linguistic variables are set for the grayscale change ratio, namely, the first-level grayscale change, the second-level grayscale change, and the third-level grayscale change. A trapezoidal membership function is set for each linguistic variable, defining its mathematical shape and universe of discourse coverage by setting the horizontal coordinates of the four vertices of the trapezoid. The area residual ratio values ​​are substituted into the triangle membership functions corresponding to the first-level, second-level, and third-level residual areas. For the triangle membership function, the input value is determined based on the relationship between the triangle and the three vertices of the triangle. The relationship between the x-coordinates of the vertices is used to calculate the linear interpolation membership degree at the left rising edge, the vertex, and the right falling edge. The gray-scale change ratio is substituted into the trapezoidal membership function set for the first-level, second-level, and third-level gray-scale changes. For the trapezoidal membership function, based on the relationship between the input value and the x-coordinates of the four vertices of the trapezoid, the linear interpolation membership degree at the left rising edge, the top platform, and the right falling edge is calculated. If the input value is in the platform interval, the membership degree is 1. Here, the first-level residual area indicates small residual dirt, the second-level residual area indicates medium residual dirt, and the third-level residual area indicates large residual dirt. The first-level gray-scale change indicates a slight change, the second-level gray-scale change indicates a moderate change, and the third-level gray-scale change indicates a significant change.

[0103] exist Figure 2 In this study, the area residue ratio and grayscale change ratio are converted into the corresponding membership degree of linguistic variables. The membership degree ranges from 0 to 1. In the area residue ratio in the left figure, the blue line indicates low residue (<5%), with the membership degree closer to 1 as the ratio decreases. The green line indicates medium residue (around 15%), with the membership degree gradually increasing from 0 to 1 and then decreasing when the ratio is between 10% and 20%. The red line indicates high residue (>40%), with the membership degree increasing as the ratio increases. The closer the degree is to 1, the closer the membership degree is to 1. The horizontal axis represents the area retention ratio, and the vertical axis represents the membership degree. In the grayscale change ratio chart on the right, the yellow line indicates a very small change, corresponding to a change rate of <20%. The lower the change rate, the closer the membership degree is to 1. The magenta line indicates a moderate change, corresponding to a change rate of 50%. When the change rate is around 40%, the membership degree is close to 1. The cyan line indicates a significant change, corresponding to a change rate of >70%. The higher the change rate, the closer the membership degree is to 1. The horizontal axis represents the grayscale change ratio, and the vertical axis represents the membership degree.

[0104] Step S303: Using the area residual ratio and gray scale change ratio as preconditions and the suction level as postconditions, establish inference rules, and combine the membership function to perform the maximum operation to obtain the rule set.

[0105] In this embodiment, a rule set consisting of multiple inference rules is established. The precondition of each inference rule statement is constructed by logically ANDing the linguistic variables of area residual ratio and grayscale change ratio. The postcondition is the linguistic variable of attraction level, which is set to weak, medium, and strong. The rule set includes inference rules, such as "If the residual area is large and the grayscale change is weak, then the attraction level is strong," covering all input combinations. The membership values ​​of the area residual ratio to each linguistic variable and the membership values ​​of the grayscale change ratio to each linguistic variable are combined pairwise. For each inference rule, the smaller of the membership values ​​of the area residual ratio and the grayscale change ratio in the precondition is taken as the trigger strength of the inference rule. The linguistic variable of attraction level in the postcondition of each triggered rule, such as the strong membership function, will be truncated by the trigger strength of the inference rule, forming a truncated local output rule set. The local output fuzzy sets generated by all triggered inference rules together constitute the rule set.

[0106] Step S304: Construct the membership function curve of the rule set, combine it with the abscissa value corresponding to the centroid of the area enclosed by the preset coordinate system, map it to the physical control range of the vacuuming component, and obtain the adaptive suction parameters through linear transformation.

[0107] The specific steps of step S304 are as follows:

[0108] Step S3041: Based on the trigger strength of the reasoning rules in the rule set, the corresponding membership function is truncated, and the maximum value of the truncated membership function is taken on the horizontal axis of the preset coordinate system to generate the membership function curve.

[0109] The specific steps of step S3041 are as follows:

[0110] Step S30411: Obtain the trigger strength and membership function corresponding to the triggered inference rule, truncate the membership function in the vertical direction, and the upper boundary height of the truncation is the trigger strength corresponding to the inference rule to obtain the local membership function.

[0111] In this embodiment, triggered inference rules are extracted from the rule set. For each triggered rule, the membership function of the linguistic variable corresponding to the attraction level of the rule's post-conclusion part is obtained, with the rule trigger strength and the membership function represented by the horizontal axis (attraction level) and the vertical axis (membership degree). The membership function is defined as a triangle and a trapezoid. For each triggered rule, its membership function is modified along the vertical axis. For points on the horizontal axis, if the membership function value corresponding to that point is greater than the rule trigger strength, the membership value of that point is set as the rule trigger strength; otherwise, the membership function value is set to less than or equal to the rule trigger strength. If the original membership function value is retained, the rule triggering strength is used as the upper limit to truncate the membership function. For each triggered inference rule, a new local membership function is generated with the vertical axis value not exceeding its rule triggering strength. The coordinate system is set according to the physical quantity, and the horizontal and vertical axes are determined according to the coordinate system. The vertical axis represents the membership degree, and its value range is [0,1]. It is used to quantify the degree to which the physical quantity belongs to the linguistic variable. For example, the suction level value belongs to the strong suction. The horizontal axis is set according to the actual physical control range of the vacuum system negative pressure generation unit. Its minimum and maximum values ​​correspond to the minimum and maximum negative pressure that the vacuum component driver can generate.

[0112] Step S30412: Set the local membership function to the horizontal axis, read the corresponding local membership function value for each point on the horizontal axis, and if a point on the horizontal axis is not within the range of the local membership function, the membership degree is zero.

[0113] In this embodiment, all local membership functions are defined on the horizontal axis, corresponding one-to-one with the points on the horizontal axis, covering the range from the minimum to the maximum suction level. The corresponding local membership function is checked sequentially for each point on the horizontal axis to determine whether the horizontal coordinate value of the point is within the domain of the local membership function, that is, whether the horizontal coordinate value makes the local membership function defined. If the horizontal coordinate value of the point is within the domain of the local membership function, the vertical coordinate value corresponding to the horizontal coordinate value of the local membership function at that point is read, that is, the membership value. If the horizontal coordinate value of the point is not within the domain of the local membership function, the membership value of the local membership function at that point is zero.

[0114] Step S30413: Compare the membership values ​​of the local membership function at each point on the horizontal axis, select the maximum value as the membership value of that point, connect the maximum membership values ​​of all points, and generate the membership function curve.

[0115] In this embodiment, the membership values ​​of points on the horizontal axis are compared, and the membership value with the largest value is selected as the membership value of that point in the final comprehensive output result. After traversing and processing all points on the horizontal axis, a new sequence is obtained. The horizontal coordinate value of each point corresponds to a final membership value after taking the maximum value. In order of ascending horizontal coordinate values, the final membership values ​​corresponding to adjacent points are connected by line segments to form a membership function curve. The membership function curve represents the overall shape and distribution of suction level obtained based on all input conditions.

[0116] Step S3042: Set a discrete point sequence on the horizontal axis, discretize the membership function curve, calculate the membership value of the rule set at the discrete points, and transform the membership function curve into a membership value sequence.

[0117] In this embodiment, on the horizontal axis representing the suction level, starting from the minimum suction level and progressing towards the maximum suction level in fixed steps, a series of equally spaced horizontal coordinate points are generated sequentially, forming a discrete point sequence covering the entire universe of discourse. For each point in this discrete point sequence, its horizontal coordinate value is substituted into the functional relationship corresponding to the membership function curve for calculation. By reading the vertical coordinate value corresponding to the horizontal coordinate value on the membership function curve, the vertical coordinate value is the membership value at that discrete point. This operation is performed on each discrete point in the discrete point sequence to obtain a sequence of membership values ​​corresponding to the discrete point sequence. This process transforms the continuous membership function curve into a discrete membership value sequence. The discrete point sequence is set by those skilled in the art according to the actual situation.

[0118] Step S3043: Multiply the x-coordinate values ​​of all discrete points with their corresponding membership values ​​and sum them to obtain the static moment of the rule set. Sum the membership values ​​of all discrete points to obtain the total membership sum.

[0119] In this embodiment, the discrete point sequence and its corresponding membership value sequence are summed. The abscissa value of each discrete point is multiplied by the membership value corresponding to that point to obtain the weighted abscissa value of the discrete point. The weighted abscissa values ​​of all discrete points are summed to obtain the static moment corresponding to the rule set. The membership values ​​corresponding to all points in the discrete point sequence are summed to obtain the total membership value.

[0120] Step S3044: Based on the quotient of the sum of static moments and membership degrees, determine the abscissa value of the centroid corresponding to the centroid of the area enclosed by the membership function curve of the rule set and the coordinate system.

[0121] In this embodiment, the static moment is used as the dividend and the sum of membership degrees is used as the divisor to perform a division operation. The quotient obtained by the division operation geometrically represents the projection position of the centroid of the entire planar region enclosed by the membership function curve and the horizontal coordinate axis on the horizontal coordinate axis. The quotient is used as the horizontal coordinate value of the centroid determined by the rule set.

[0122] Step S3045: Input the horizontal coordinate value of the center of gravity into a preset linear scaling function to convert it into an adaptive suction parameter that controls the negative pressure generated by the vacuuming component. The conversion range is the minimum and maximum value of the physical control quantity of the vacuuming component.

[0123] In this embodiment, the abscissa value of the center of gravity is input into a preset linear scaling function for processing. The linear scaling function specifies that its input variable is the abscissa value of the center of gravity, the input domain is the minimum and maximum values ​​of the suction level domain, and its output variable is the adaptive suction parameter. The output value domain is the minimum and maximum values ​​of the physical control quantity of the vacuum component. The linear scaling function linearly maps the input domain to the output value domain through proportional scaling calculation. Specifically, the minimum suction level is subtracted from the abscissa value of the center of gravity, and divided by the difference between the maximum and minimum suction level values ​​to obtain a normalized scaling factor. The scaling factor is multiplied by the difference between the maximum and minimum physical control quantity of the vacuum component, and the product is added to the minimum physical control quantity of the vacuum component. Finally, the adaptive suction parameter used to control the negative pressure generation unit of the vacuum component is calculated. The linear scaling function is constructed based on the suction level range and the physical control quantity range of the negative pressure generation unit of the vacuum component. The minimum and maximum values ​​of the physical control quantity are determined based on the physical characteristics of the vacuum component driver.

[0124] Step S4: Based on the adaptive suction parameters, control the dust collection component of the photovoltaic cleaning robot to generate negative pressure in the internal cavity of the roller brush, and suck the dirt into the inside of the roller brush through the array of holes on the surface of the roller brush.

[0125] The specific steps of step S4 are as follows:

[0126] Step S401: By establishing a mapping relationship between adaptive suction parameters and vacuuming components, the adaptive suction parameters are converted into the target rotation speed value of the vacuuming components, and a corresponding target control signal is generated based on the target rotation speed value.

[0127] In this embodiment, according to a preset suction parameter speed mapping table, which defines the reference speed values ​​of the vacuum cleaner component drive motor corresponding to different numerical ranges of the adaptive suction parameter, the adaptive suction parameter is compared with each numerical range in the suction parameter speed mapping table to determine its specific range. The reference speed value corresponding to this range is read as the initial target speed value. Based on the electrical and mechanical characteristic parameters of the vacuum cleaner component drive motor, the amount of negative pressure change generated by a unit speed change is calculated. Combined with the expected negative pressure intensity represented by the adaptive suction parameter, the initial target speed value is corrected to finally determine the target speed value of the vacuum cleaner component. Based on the difference between the target speed value and the current actual speed value of the drive motor, as well as the motor's acceleration time constant, the corresponding pulse is calculated. The duty cycle change command of the width modulation signal is a target control signal used to control the motor to reach the target speed. This involves setting different speed values ​​for the drive motor of the vacuum component and recording the stable negative pressure value generated by the corresponding internal cavity of the roller brush. The measured negative pressure value is normalized to the theoretical range of the adaptive suction parameters, thereby fitting a data table showing the correspondence between the suction parameters and the drive motor speed values, and establishing a suction parameter speed mapping table. The electrical and mechanical characteristic parameters are determined based on the product data of the drive motor. The motor's acceleration time constant is obtained by performing a standard step response test on the drive motor of the vacuum component. The desired negative pressure intensity is obtained by mapping the adaptive suction parameters through a linear scaling relationship, which is expressed as a fixed conversion coefficient between the suction parameters and the physical negative pressure value.

[0128] Step S402: Output the target control signal to the dust collection component, and obtain the negative pressure deviation value based on the comparison between the actual negative pressure value of the internal cavity of the roller brush and the target negative pressure value determined by the adaptive suction parameters.

[0129] In this embodiment, a target control signal is sent to the controller of the vacuum component drive motor, which drives the motor to rotate and drive the centrifugal fan impeller. The actual negative pressure value of the vacuum component's internal cavity is measured by a negative pressure sensor installed in the cavity of the roller brush. Based on the preset ratio between the adaptive suction parameters and the desired negative pressure intensity, the target negative pressure value is calculated. The actual negative pressure value read in real time is subtracted from the calculated target negative pressure value to obtain the negative pressure deviation value at the current moment. This negative pressure deviation value represents the pressure state difference between the current vacuuming effect and the desired vacuuming effect. The ratio is set by those skilled in the art according to the actual situation.

[0130] Step S403: Perform proportional-integral control on the negative pressure deviation value, calculate the control compensation amount used to correct the target control signal, superimpose the control compensation amount with the target control signal to generate the adjusted real-time control signal, and adjust the dust collection component.

[0131] The specific steps of step S403 are as follows:

[0132] Step S4031: Set the sampling period according to the pressure response time constant of the internal cavity of the roller brush, and multiply the negative pressure deviation value of the current sampling period by the preset proportional coefficient to obtain the proportional adjustment component.

[0133] In this embodiment, the sampling period of the closed-loop control system is calculated by dividing the pressure response time constant of the internal cavity of the roller brush by an empirical coefficient. The latest negative pressure deviation value is acquired periodically according to the sampling period. The negative pressure deviation value acquired in the current sampling period is multiplied by a preset proportional coefficient used to adjust the response speed. The product is used as a proportional adjustment component for real-time correction of the deviation, calculated based on the pressure deviation at the current moment. The empirical coefficient and the proportional coefficient are set by those skilled in the art according to the actual situation.

[0134] Step S4032: Sum the negative pressure deviation values ​​of all sampling periods, multiply the sum by the preset integral coefficient to obtain the integral adjustment component, and add the proportional adjustment component and the integral adjustment component to obtain the control compensation amount.

[0135] In this embodiment, historical negative pressure deviation values ​​generated within several sampling periods are stored in chronological order. The number of stored values ​​is determined by a preset integral window length. The negative pressure deviation value of the current sampling period is summed with all historical negative pressure deviation values ​​to obtain a total value representing the historical cumulative deviation. This total value is multiplied by a preset integral coefficient used to eliminate steady-state error to obtain an integral adjustment component used to correct the historical cumulative deviation. The proportional adjustment component and the integral adjustment component are algebraically added to obtain a control compensation amount used to dynamically correct the operating state of the dust collection component. The integral coefficient and the integral window length are set by those skilled in the art according to the actual situation.

[0136] Step S4033: Algebraically superimpose the control compensation amount with the target control signal and send it to the vacuuming component for adjustment.

[0137] In this embodiment, the control compensation amount and the target control signal are algebraically superimposed. If the control compensation amount is positive, it is added to the target control signal; if the control compensation amount is negative, it is subtracted from the target control signal. This yields a real-time corrected dust collection component drive control signal. The corrected dust collection component drive control signal is then sent to the power drive interface of the dust collection component drive motor. The power drive interface adjusts the electrical power applied to the motor according to the signal changes, thereby changing the instantaneous torque and speed of the motor. This achieves dynamic and precise adjustment of the negative pressure state of the internal cavity of the roller brush, bringing it close to and stabilizing it at the target negative pressure value.

[0138] Step S404: The adjusted suction components bring the internal cavity of the roller brush to the target negative pressure value. Based on the pressure difference between the internal cavity of the roller brush and the external environment through the array of holes on the surface of the roller brush, dirt is sucked in.

[0139] In this embodiment, the adjusted suction component ensures that the actual negative pressure value of the internal cavity of the roller brush reaches and stabilizes at the target negative pressure value determined by the adaptive suction parameters. A stable pressure connection is formed between the internal cavity of the roller brush and the external photovoltaic panel environment through the array of holes on the roller brush surface. Since the pressure inside the roller brush is lower than the atmospheric pressure of the external environment, the resulting pressure difference creates an airflow pointing towards the inside of the roller brush at the entrance of each array of holes. The airflow applies fluid drag to the dirt particles attached to the photovoltaic panel surface and loosened by ultrasonic vibration. At the same time, the roller brush bristles assist in peeling off the remaining attached particles, overcoming the residual adhesion and gravity between the dirt particles and the panel surface, driving the dirt particles to detach from the photovoltaic panel surface and move with the airflow. The dirt particles enter the internal cavity of the roller brush through the array of holes on the roller brush surface with the airflow, thereby completing the transfer and collection of pollutants from the photovoltaic panel surface to the inside of the roller brush.

[0140] Step S5: Discharge the dirt inside the suction roller brush to the designated area.

[0141] In this embodiment, when the roller brush moves along the cleaning path to the designated discharge area at the edge of the photovoltaic panel, the dust discharge channel of the roller brush is connected, the dust suction component is closed, and the valve connecting the internal cavity of the roller brush and the dust discharge channel is opened, so that the pressure in the internal cavity of the roller brush is restored to a level close to that of the environment, and the dirt particles accumulated in the internal cavity of the roller brush are discharged to the designated collection point on the ground through the dust discharge channel.

[0142] Example 2

[0143] Please see Figure 3 One embodiment of the present invention is a photovoltaic cleaning robot control system for ultrasonic collaborative intelligent vacuuming, comprising a dirt sensing module, an ultrasonic dirt vibration module, a suction power decision module, a negative pressure adsorption module, and an emission module.

[0144] The dirt sensing module is used to collect dirt image information, photovoltaic panel parameters, and photovoltaic cleaning robot parameters. The photovoltaic cleaning robot parameters include the physical characteristics of the roller brush, effective working thickness, and pressure response time constant.

[0145] The ultrasonic cleaning module is used to calculate the ultrasonic vibration parameters acting on the roller brush of the photovoltaic cleaning robot based on the image information of dirt on the photovoltaic panel, so that it generates vibration and transmits it to the photovoltaic panel through the roller brush to clean the photovoltaic panel.

[0146] The suction decision module is used to acquire optical feedback data of the photovoltaic panel after cleaning, analyze the dirt removal status, and calculate adaptive suction parameters based on the dirt removal status.

[0147] The negative pressure adsorption module is used to control the dust collection component of the photovoltaic cleaning robot to generate negative pressure in the internal cavity of the roller brush according to the adaptive suction parameters, so as to draw dirt into the inside of the roller brush through the array of holes on the surface of the roller brush.

[0148] The discharge module is used to discharge the dirt inside the suction roller brush to a designated area.

[0149] In addition, the parts of the technical solutions provided in the embodiments of this application that are consistent with the implementation principles of the corresponding technical solutions in the prior art have not been described in detail, so as to avoid excessive elaboration.

[0150] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims. All of these forms are within the protection scope of the present invention.

Claims

1. A control method for a photovoltaic cleaning robot with ultrasonic-assisted intelligent vacuuming, characterized in that, include: Collect image information of dirt on photovoltaic panel surface, photovoltaic panel surface parameters and photovoltaic cleaning robot parameters, the photovoltaic cleaning robot parameters include the physical characteristics of roller brush, effective working thickness and pressure response time constant; The ultrasonic vibration parameters acting on the roller brush of the photovoltaic cleaning robot are calculated based on the image information of the dirt on the photovoltaic panel, so that the robot vibrates and transmits the vibration to the photovoltaic panel through the roller brush to clean the photovoltaic panel. Obtain optical feedback data of the photovoltaic panel after cleaning, analyze the dirt removal status, and calculate adaptive suction parameters based on the dirt removal status; Based on the adaptive suction parameters, the dust collection components of the photovoltaic cleaning robot are controlled to generate negative pressure in the internal cavity of the roller brush, which draws dirt into the roller brush through the array of holes on the surface of the roller brush. The contaminants sucked into the roller brush are discharged to a designated area.

2. The control method for a photovoltaic cleaning robot with ultrasonic collaborative intelligent vacuuming according to claim 1, characterized in that, The process of calculating ultrasonic vibration parameters acting on the roller brush of the photovoltaic cleaning robot based on the image information of dirt on the photovoltaic panel surface, causing it to vibrate and transmit the vibration to the photovoltaic panel surface via the roller brush, thereby cleaning the photovoltaic panel surface, includes: Extract the optical and spatial distribution features of the contaminated areas from the contaminated image information of the photovoltaic panel to determine the contamination type, coverage density, and local average thickness. Based on the type of contamination, coverage density, and local average thickness, a mapping relationship is established with the initial ultrasonic vibration frequency parameters and the initial ultrasonic vibration amplitude parameters to construct the initial ultrasonic vibration parameter set. Based on the photovoltaic panel parameters and the physical characteristics of the roller brush, an energy transfer attenuation model for ultrasonic vibration energy is constructed. The initial set of ultrasonic vibration parameters is used as the input to the energy transfer attenuation model to calculate the vibration energy. Based on the comparison results between the vibration energy and the preset energy threshold, the initial ultrasonic vibration frequency parameters and the initial ultrasonic vibration amplitude parameters are iteratively optimized to obtain the ultrasonic vibration parameter set. Based on the spatial distribution characteristics and the set of ultrasonic vibration parameters, the roller brush is driven to generate vibration and the vibration is transmitted to the photovoltaic panel surface for cleaning.

3. The control method for a photovoltaic cleaning robot with ultrasonic collaborative intelligent vacuuming according to claim 2, characterized in that, Based on the photovoltaic panel parameters and the physical characteristics of the roller brush, an energy transfer attenuation model for ultrasonic vibration energy is constructed. The initial set of ultrasonic vibration parameters is used as input to the energy transfer attenuation model to calculate the vibration energy, including: The photovoltaic panel parameters and the physical properties of the roller brush are standardized to generate basic input parameters. Based on the basic input parameters, the propagation speed and attenuation constant of ultrasonic vibration in the roller brush are calculated. Based on the propagation speed and attenuation constant, the energy transmittance function of ultrasonic vibration energy transferred from the roller brush to the photovoltaic panel is derived. Combined with the energy reflection coefficient and transmission coefficient of the photovoltaic panel surface, an energy transfer attenuation model is constructed. The initial ultrasonic vibration frequency parameters and initial ultrasonic vibration amplitude parameters are converted into initial vibration energy density and input into the energy transfer attenuation model. The remaining vibration energy density is calculated using the energy transmittance function and the attenuation constant. The remaining vibration energy density is spatially integrated over the surface of the photovoltaic panel to calculate the total vibration energy value acting on the entire area of ​​the photovoltaic panel as the vibration energy.

4. The control method for a photovoltaic cleaning robot with ultrasonic collaborative intelligent vacuuming according to claim 3, characterized in that, The process of converting the initial ultrasonic vibration frequency parameters and initial ultrasonic vibration amplitude parameters into initial vibration energy density and inputting them into the energy transfer attenuation model, and calculating the remaining vibration energy density using the energy transmittance function and attenuation constant, includes: Based on the physical properties of the roller brush and the initial ultrasonic vibration amplitude parameters, the maximum kinetic energy of the roller brush per unit volume during the vibration process is calculated, and the initial mechanical energy density value is calculated in combination with the initial ultrasonic vibration frequency parameters. Multiply the attenuation constant by the effective thickness of the roller brush, and then raise the product to the negative power of the natural exponential function to obtain the exponential attenuation factor during the process of ultrasonic vibration penetrating the thickness of the roller brush. The theoretical vibration energy density of the photovoltaic panel is obtained by multiplying the initial mechanical energy density value by the energy transmission ratio represented by the energy transmittance function. By multiplying the exponential decay factor by the theoretical vibration energy density, the residual vibration energy density formed on the surface of the photovoltaic panel after the ultrasonic vibration has passed through the internal attenuation and transmission loss of the roller brush is determined.

5. The control method for a photovoltaic cleaning robot with ultrasonic collaborative intelligent vacuuming according to claim 4, characterized in that, The process of acquiring optical feedback data from the cleaned photovoltaic panel, analyzing the dirt removal status, and calculating adaptive suction parameters based on the dirt removal status includes: Obtain optical feedback data of the photovoltaic panel after cleaning, detect the pixel area of ​​the residual dirt area in the photovoltaic panel, compare it with the photovoltaic panel before cleaning, and calculate the area residual ratio and grayscale change ratio of the dirt. The area residual ratio and grayscale change ratio are used as inputs for fuzzification, mapped to linguistic variables and membership functions are set respectively. By using the area residual ratio and gray scale change ratio as preconditions and the suction level as postconditions, reasoning rules are established, and the rule set is obtained by taking the largest value operation in combination with the membership function. The membership function curve of the constructed rule set is combined with the abscissa value corresponding to the centroid of the area enclosed by the preset coordinate system and mapped to the physical control range of the vacuuming component. The adaptive suction parameters are obtained through linear transformation.

6. The control method for a photovoltaic cleaning robot with ultrasonic collaborative intelligent vacuuming according to claim 5, characterized in that, The membership function curve of the constructed rule set, combined with the abscissa value corresponding to the centroid of the area enclosed by the preset coordinate system, is mapped to the physical control range of the vacuum component. Adaptive suction parameters are obtained through linear transformation, including: Based on the trigger strength of the reasoning rules in the rule set, the corresponding membership function is truncated, and the maximum value of the truncated membership function is taken on the horizontal axis of the preset coordinate system to generate the membership function curve. Set a discrete point sequence on the horizontal axis, discretize the membership function curve, calculate the membership value of the rule set at the discrete points, and transform the membership function curve into a membership value sequence. Multiply the x-coordinates of all discrete points by their corresponding membership values ​​and sum them to obtain the static moment of the rule set. Sum the membership values ​​of all discrete points to obtain the total membership. Based on the quotient of the static moment and the sum of the membership degrees, determine the abscissa value of the centroid corresponding to the centroid of the area enclosed by the membership function curve of the rule set and the coordinate system. The horizontal coordinate value of the center of gravity is input into a preset linear scaling function and converted into an adaptive suction parameter that controls the negative pressure generated by the vacuuming component. The conversion range is the minimum and maximum value of the physical control quantity of the vacuuming component.

7. The control method for a photovoltaic cleaning robot with ultrasonic collaborative intelligent vacuuming according to claim 6, characterized in that, The step of truncating the corresponding membership function based on the trigger strength of the inference rules in the rule set, and then taking the maximum value of the truncated membership function on the horizontal axis of a preset coordinate system to generate a membership function curve includes: Obtain the trigger strength and membership function corresponding to the triggered inference rule, truncate the membership function in the vertical direction, and the height of the upper boundary of the truncation is the trigger strength corresponding to the inference rule to obtain the local membership function; Set the local membership function to the horizontal axis, read the corresponding local membership function value for each point on the horizontal axis, and if a point on the horizontal axis is not within the range of the local membership function, the membership degree is zero. For each point on the horizontal axis, compare the membership values ​​of the local membership function at that point, select the maximum value as the membership value of that point, connect the maximum membership values ​​of all points, and generate the membership function curve.

8. The control method for a photovoltaic cleaning robot with ultrasonic collaborative intelligent vacuuming according to claim 7, characterized in that, The process of controlling the suction components of the photovoltaic cleaning robot according to adaptive suction parameters to create negative pressure in the internal cavity of the roller brush, drawing dirt into the roller brush through the array of holes on its surface, includes: By establishing a mapping relationship between adaptive suction parameters and vacuuming components, the adaptive suction parameters are converted into target rotation speed values ​​of the vacuuming components, and corresponding target control signals are generated based on the target rotation speed values. The target control signal is output to the dust collection component, and the negative pressure deviation value is obtained by comparing the actual negative pressure value of the internal cavity of the roller brush with the target negative pressure value determined by the adaptive suction parameters. The negative pressure deviation value is subjected to proportional-integral control to calculate the control compensation amount used to correct the target control signal. The control compensation amount is superimposed with the target control signal to generate the adjusted real-time control signal and adjust the dust collection component. The adjusted suction components bring the internal cavity of the roller brush to the target negative pressure value. Based on the pressure difference between the internal cavity of the roller brush and the external environment, formed by the array of holes on the surface of the roller brush, dirt is sucked in.

9. A control method for a photovoltaic cleaning robot with ultrasonic collaborative intelligent vacuuming according to claim 8, characterized in that, The process of performing proportional-integral control on the negative pressure deviation value, calculating the control compensation amount used to correct the target control signal, superimposing the control compensation amount on the target control signal to generate an adjusted real-time control signal, and adjusting the dust collection components includes: The sampling period is set according to the pressure response time constant of the internal cavity of the roller brush. The negative pressure deviation value of the current sampling period is multiplied by the preset proportional coefficient to obtain the proportional adjustment component. The negative pressure deviation values ​​of all sampling periods are summed up, and the sum is multiplied by the preset integral coefficient to obtain the integral adjustment component. The proportional adjustment component is added to the integral adjustment component to obtain the control compensation amount. The control compensation amount is algebraically superimposed with the target control signal and sent to the dust collection component for adjustment.

10. A control system for a photovoltaic cleaning robot with ultrasonic collaborative intelligent vacuuming, used to implement the control method for a photovoltaic cleaning robot with ultrasonic collaborative intelligent vacuuming as described in any one of claims 1-9, characterized in that, It includes a dirt sensing module, an ultrasonic dirt-vibration module, a suction power decision module, a negative pressure adsorption module, and an emission module: The dirt sensing module is used to collect dirt image information, photovoltaic panel parameters, and photovoltaic cleaning robot parameters. The photovoltaic cleaning robot parameters include the physical characteristics of the roller brush, effective working thickness, and pressure response time constant. The ultrasonic cleaning module is used to calculate the ultrasonic vibration parameters acting on the roller brush of the photovoltaic cleaning robot based on the image information of the dirt on the photovoltaic panel, so that it generates vibration and transmits it to the photovoltaic panel through the roller brush to clean the photovoltaic panel. The suction power decision module is used to acquire optical feedback data of the photovoltaic panel after cleaning, analyze the dirt removal status, and calculate adaptive suction power parameters based on the dirt removal status. The negative pressure adsorption module is used to control the dust collection component of the photovoltaic cleaning robot to generate negative pressure in the internal cavity of the roller brush according to the adaptive suction parameters, so as to draw dirt into the inside of the roller brush through the array of holes on the surface of the roller brush. The discharge module is used to discharge the dirt inside the suction roller brush to a designated area.

Images