An intelligent cleaning robot for photovoltaic power station and navigation control system

Abstract

This invention discloses an intelligent cleaning robot for photovoltaic power plants and its navigation control system. The intelligent cleaning robot includes a robot body, a polarized light dust detection module, and a point cloud fusion obstacle avoidance module. The navigation control system includes the aforementioned intelligent cleaning robot, and further includes a dynamic dirt map construction module and an anti-interference path planning module. This invention can accurately detect dust even in strong reflective environments. Through a continuous dirt heat map of the entire photovoltaic power plant, the path planning algorithm prioritizes cleaning highly soiled areas; and the anti-interference path planning achieves a balance between cleaning efficiency and safety.

Description

Technical Field

[0001] This invention relates to the field of automatic control technology, specifically to an intelligent cleaning robot and navigation control system for photovoltaic power plants. Background Technology

[0002] As the global energy structure shifts towards a green and low-carbon model, solar photovoltaic (PV) power generation, as a clean and renewable energy source, has seen its installed capacity grow rapidly. PV power plants are typically built in sunny, open areas such as deserts, wastelands, hills, and large factory rooftops. These areas have harsh environments, making it easy for dust, sand, bird droppings, snow, and other pollutants to accumulate on the surface of PV modules. Numerous studies have shown that dirt and smudges on the module surface can severely lead to the "hot spot effect" and power generation efficiency loss. In some cases, the efficiency degradation can reach as high as 15%-25%, causing not only significant economic losses but also shortening the lifespan of the modules.

[0003] Currently, the cleaning methods for photovoltaic modules mainly rely on traditional manual cleaning, semi-automatic devices, and cleaning robots. Manual cleaning requires workers to use mops, high-pressure water guns, and other tools for rinsing and wiping, which presents numerous problems such as high labor intensity, low efficiency, long cleaning cycles, high water consumption, high costs, and safety risks associated with working at heights. The drawbacks of manual cleaning are particularly pronounced in large-scale ground-mounted power plants and distributed rooftop power plants. To overcome the shortcomings of manual cleaning, track-mounted or rail-guided cleaning equipment and early-generation cleaning robots have emerged on the market. These devices have alleviated the burden on manpower to some extent, but still have significant limitations. First, they have poor environmental adaptability: lacking intelligent sensing capabilities, they cannot effectively identify obstacles (such as cables, supports, and snow) or areas with varying degrees of dirt. When encountering obstacles, they can only stop, requiring manual intervention, and cannot achieve truly fully autonomous operation. Second, their cleaning strategies are limited, unable to achieve "on-demand cleaning" based on dirt distribution; they can only perform uniform cleaning along a preset path, resulting in wasted energy and time. Therefore, there is an urgent need in this field for a photovoltaic power plant cleaning robot and navigation control system capable of autonomous navigation and obstacle avoidance, and possessing intelligent cleaning decision-making capabilities. Summary of the Invention

[0004] This invention provides an intelligent cleaning robot for photovoltaic power plants, which includes a robot body, a polarized light dust detection module, and a point cloud fusion obstacle avoidance module.

[0005] The robot body can move on the surface of photovoltaic panels to clean them;

[0006] The polarized light dust detection module is used to calculate the optical thickness of dust at a specified coordinate point on the surface of a photovoltaic panel. It includes a light detection unit that detects the wavelength of incident light, a parallel polarized light receiving unit, a vertically polarized component light receiving unit, and a calculation unit.

[0007] The point cloud fusion obstacle avoidance module is used to enable the robot to avoid obstacles when it moves on the surface of photovoltaic modules. It includes a millimeter-wave radar unit, a visual recognition unit, and a fusion processing unit.

[0008] Furthermore, the computing unit calculates the specified coordinate points. Dust optical thickness at the location The specific calculation process is as follows:

[0009]

[0010] in, The incident light wavelength, The intensity of the parallel polarization component light. The intensity of the vertically polarized component light. Atmospheric attenuation coefficient, From the light detection unit to the coordinate point The distance.

[0011] 3. The intelligent cleaning robot for photovoltaic power plants according to claim 1, characterized in that the millimeter-wave radar and the visual recognition unit are respectively used to acquire radar point clouds and visual point clouds, and the fusion processing unit acquires a fused point cloud of radar point clouds and visual point clouds. The specific process includes the following calculations:

[0012]

[0013] in, For visual point clouds, For radar point clouds, This is the optimal transformation matrix;

[0014]

[0015] in, The optimal transformation matrix is... It is a non-rigid transformation matrix. for The non-rigid transformation matrix at the minimum value , For visual point clouds, For radar point clouds, The number of matching point pairs in the visual point cloud and radar point cloud. The index variable represents the index of the currently being processed. One point pair, For the first A visual point In order to be with the first A visual point The corresponding radar point index, That is, the corresponding radar point. The regularization coefficient is . For non-rigid transformation functions The gradient.

[0016] The present invention also relates to a navigation control system, which includes the above-mentioned intelligent cleaning robot for photovoltaic power plants. The system includes a dynamic dirt map construction module and an anti-interference path planning module.

[0017] The dynamic dirt map construction module is used to control the intelligent cleaning robot of the photovoltaic power station. On each photovoltaic panel to be cleaned, the optical thickness of dust at multiple discrete points is measured, and the optical thickness of dust measured at discrete points is converted into a continuous dirt heat map including all photovoltaic panels to be cleaned.

[0018] The anti-interference path planning module is used to integrate the degree of dirtiness, obstacle risk, and movement cost to plan a path, and control the photovoltaic power station intelligent cleaning robot to clean according to the planned path.

[0019] Furthermore, obtaining a continuous dirt heatmap includes the following calculations:

[0020]

[0021] in, The index of the photovoltaic panel is the first one. Line number A row of photovoltaic panels, This indicates the dirt level of the photovoltaic panel. Indicates the first The center coordinates of the photovoltaic panel For in position The optical thickness of dust at a given location is obtained by polarized light dust detection, where K represents the thickness of the dust at the th position. The total number of all measurement points on the photovoltaic panel. The index variable represents the first One measurement point, This is the bandwidth parameter of the Gaussian kernel function.

[0022] Furthermore, path planning specifically includes the following calculations:

[0023]

[0024] in, From the current location of the intelligent cleaning robot in the photovoltaic power station to the node The total path cost for each location, where each node corresponds to a photovoltaic panel to be cleaned, and the node coordinates are obtained through... That is, the first Line number The photovoltaic panels in the column are indexed, and the node with the lowest cost is selected and added to the planned path each time. From the starting position to the node The actual cost of moving a location For nodes coordinates These are the coordinates of the current position. For nodes The corresponding dirt value of the photovoltaic panel, This is a weighting coefficient for the degree of dirtiness. The weighting coefficient for obstacle risk, Point cloud coordinates of the obstacle. The radar detection confidence level for obstacles. For the merged point cloud, Take all obstacle nodes and select the pairs of nodes. The maximum risk contribution ensures that obstacle avoidance is triggered even for a single high-risk obstacle.

[0025] The technical solution of this invention uses polarized light dust detection to calculate the optical thickness of dust at specified coordinate points on the surface of a photovoltaic panel. It quantifies the degree of dust accumulation by utilizing the characteristics of polarized light, overcoming the limitations of traditional images' sensitivity to illumination, and accurately detecting dust even in environments with strong reflections. By fusing millimeter-wave radar and visual point clouds, combining high visual resolution with radar anti-interference capabilities, it generates an accurate all-weather environmental model, while using non-rigid transformations to adapt to changes in the position of moving objects. Through dynamic dirt map construction, the optical thickness of dust measured at discrete points is converted into a continuous dirt heat map of the entire photovoltaic power station, facilitating path planning algorithms to prioritize cleaning of highly dirty areas. Through anti-interference path planning, it achieves a balance between cleaning efficiency and safety by integrating three factors: degree of dirt, obstacle risk, and movement cost. Attached Figure Description

[0026] Figure 1 This is a structural block diagram of the intelligent cleaning robot for photovoltaic power plants according to the present invention.

[0027] Figure 2 This is a structural block diagram of the navigation control system of the present invention. Detailed Implementation

[0028] The present invention will now be further described with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention and should not be construed as limiting the scope of protection of the present invention. It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of the present invention.

[0029] Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular forms are intended to include the plural forms as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms “comprising” and / or “including” are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0030] Embodiment 1 of the present invention relates to an intelligent cleaning robot for photovoltaic power plants, which includes a robot body, a polarized light dust detection module, and a point cloud fusion obstacle avoidance module.

[0031] The robot body can be any type of robot body capable of moving on the surface of the photovoltaic panel to clean the photovoltaic panel. Since these are all existing technologies, this invention does not limit them.

[0032] The polarized dust detection module is used to calculate specified coordinate points on the surface of the photovoltaic panel. The dust optical thickness at a specified coordinate point (x, y) is calculated using a light detection unit for detecting the incident light wavelength, a parallel polarized light receiving unit, a vertically polarized component light receiving unit, and a calculation unit. This polarized light dust detection module utilizes the characteristics of polarized light to quantify the degree of dust accumulation, overcoming the limitations of traditional images' sensitivity to illumination and enabling accurate dust detection even in strongly reflective environments. The method is as follows:

[0033]

[0034] in, The incident light wavelength, The intensity of the parallel polarization component light. The intensity of the vertically polarized component light. Atmospheric attenuation coefficient, From the light detection unit to the coordinate point The distance.

[0035] The point cloud fusion obstacle avoidance module is used to achieve obstacle avoidance when the robot moves on the surface of the photovoltaic module. It includes a millimeter-wave radar unit, a visual recognition unit, and a fusion processing unit. By fusing millimeter-wave radar and visual point clouds, a fused point cloud is obtained for obstacle avoidance by the robot. Combining high-resolution vision and radar anti-interference capabilities, an all-weather accurate environmental model is generated, while non-rigid transformation is used to adapt to changes in the position of moving objects.

[0036] The millimeter-wave radar and visual recognition unit are used to acquire radar point clouds and visual point clouds, respectively, and the fusion processing unit acquires a fused point cloud of radar point clouds and visual point clouds. The specific process includes:

[0037]

[0038] in, For visual point clouds, For radar point clouds, This is the optimal transformation matrix.

[0039]

[0040] in, The optimal transformation matrix is... It is a non-rigid transformation matrix. for The non-rigid transformation matrix at the minimum value , For visual point clouds, For radar point clouds, The number of matching point pairs in the visual point cloud and radar point cloud. The index variable represents the index of the currently being processed. One point pair, For the first A visual point In order to be with the first A visual point The corresponding radar point index, That is, the corresponding radar point. The regularization coefficient is . For non-rigid transformation functions The gradient.

[0041] Embodiment 2 of the present invention relates to a navigation control system, which includes the intelligent cleaning robot for photovoltaic power plants described in Embodiment 1, and performs navigation control on it. The system includes a dynamic dirt map construction module and an anti-interference path planning module.

[0042] The dynamic dirt map construction module is used to control the intelligent cleaning robot of the photovoltaic power station. On each photovoltaic panel to be cleaned, it measures the optical thickness of dust at multiple discrete points and converts the measured dust optical thickness into a continuous dirt heat map including all photovoltaic panels to be cleaned. This allows the path planning algorithm to prioritize cleaning of highly soiled areas. Specifically, the calculations include the following:

[0043]

[0044] in, The index of the photovoltaic panel is the first one. Line number A row of photovoltaic panels, This indicates the dirt level of the photovoltaic panel. Indicates the first The center coordinates of the photovoltaic panel For in position The optical thickness of dust at a given location is obtained by polarized light dust detection, where K represents the thickness of the dust at the th position. The total number of all measurement points on the photovoltaic panel. The index variable represents the first One measurement point, This is the bandwidth parameter of the Gaussian kernel function.

[0045] The anti-interference path planning module is used to control the intelligent cleaning robot of the photovoltaic power station to clean according to the planned path. The path planning integrates the degree of dirt, obstacle risk, and movement cost to achieve a balance between cleaning efficiency and safety.

[0046]

[0047] in, From the current location of the intelligent cleaning robot in the photovoltaic power station to the node The total path cost for each location, where each node corresponds to a photovoltaic panel to be cleaned, and the node coordinates are obtained through... That is, the first Line number The photovoltaic panels in the column are indexed, and the node with the lowest cost is selected and added to the planned path each time. From the starting position to the node The actual cost of moving a location For nodes coordinates These are the coordinates of the current position. For nodes The corresponding dirt value of the photovoltaic panel, This is a weighting coefficient for the degree of dirtiness. The weighting coefficient for obstacle risk, Point cloud coordinates of the obstacle. The radar detection confidence level for obstacles. For the merged point cloud, Take all obstacle nodes and select the pairs of nodes. The maximum risk contribution ensures that obstacle avoidance is triggered even for a single high-risk obstacle.

[0048] The above description is merely a preferred embodiment of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A navigation control system, the navigation control system comprising a photovoltaic power station intelligent cleaning robot, characterized in that, The intelligent cleaning robot for photovoltaic power stations includes a robot body, a polarized light dust detection module, and a point cloud fusion obstacle avoidance module. The robot body can move on the surface of photovoltaic panels to clean them; The polarized light dust detection module is used to calculate the optical thickness of dust at a specified coordinate point on the surface of a photovoltaic panel. It includes a light detection unit that detects the wavelength of incident light, a parallel polarized light receiving unit, a vertically polarized component light receiving unit, and a calculation unit. The point cloud fusion obstacle avoidance module is used to enable the robot body to avoid obstacles when moving on the surface of photovoltaic modules. It includes a millimeter-wave radar unit, a visual recognition unit, and a fusion processing unit. The computing unit calculates the dust optical thickness at the specified coordinate point The computing unit calculates the dust optical thickness at the specified coordinate point The specific calculation process is as follows: ； in, The incident light wavelength, The intensity of the parallel polarization component light. The intensity of the vertically polarized component light. Atmospheric attenuation coefficient, From the light detection unit to the coordinate point The distance; The millimeter-wave radar and visual recognition unit are used to acquire radar point clouds and visual point clouds, respectively, and the fusion processing unit acquires a fused point cloud from the radar point cloud and the visual point cloud. The specific process includes the following calculations: ； in, For visual point clouds, For radar point clouds, This is the optimal transformation matrix; ； in, The optimal transformation matrix is... It is a non-rigid transformation matrix. for The non-rigid transformation matrix at the minimum value , For visual point clouds, For radar point clouds, The number of matching point pairs in the visual point cloud and radar point cloud. The index variable represents the index of the currently being processed. One point pair, For the first A visual point In order to be with the first A visual point The corresponding radar point index, That is, the corresponding radar point. The regularization coefficient is . Non-rigid transformation function The gradient; The system includes a dynamic dirt map construction module and an anti-interference path planning module; The dynamic dirt map construction module is used to control the intelligent cleaning robot of the photovoltaic power station. On each photovoltaic panel to be cleaned, the optical thickness of dust at multiple discrete points is measured, and the optical thickness of dust measured at discrete points is converted into a continuous dirt heat map including all photovoltaic panels to be cleaned. The anti-interference path planning module is used to integrate the degree of dirtiness, obstacle risk, and movement cost to plan a path, and control the photovoltaic power station intelligent cleaning robot to clean according to the planned path; Obtaining a continuous dirt heatmap involves the following calculations: ； in, The index of the photovoltaic panel is the first one. Line number A row of photovoltaic panels, This indicates the dirt level of the photovoltaic panel. Indicates the first The center coordinates of the photovoltaic panel For in position The optical thickness of dust at a given location is obtained by polarized light dust detection, where K represents the thickness of the dust at the th position. The total number of all measurement points on the photovoltaic panel. The index variable represents the first One measurement point, This is the bandwidth parameter of the Gaussian kernel function.

2. The navigation control system according to claim 1, characterized in that, Path planning specifically includes the following calculations: ； in, From the current location of the intelligent cleaning robot in the photovoltaic power station to the node The total path cost for each location, where each node corresponds to a photovoltaic panel to be cleaned, and the node coordinates are obtained through... That is, the first Line number The photovoltaic panels in the column are indexed, and the node with the lowest cost is selected and added to the planned path each time. From the starting position to the node The actual cost of moving a location For nodes coordinates These are the coordinates of the current position. For nodes The corresponding dirt value of the photovoltaic panel, This is a weighting coefficient for the degree of dirtiness. The weighting coefficient for obstacle risk, Point cloud coordinates of the obstacle. The radar detection confidence level for obstacles. For the merged point cloud, Take all obstacle nodes and select the pairs of nodes. The maximum risk contribution ensures that obstacle avoidance is triggered even for a single high-risk obstacle.

Images