An automatic green plant pruning and branch and leaf recycling system for landscape engineering

Abstract

The present invention belongs to the technical field of green plant pruning. The present invention discloses an automatic green plant pruning and branch and leaf recovery system for landscape engineering; it includes the following modules: a data acquisition module for real-time acquisition of parameter data of green plants in the landscape engineering; a health assessment module for evaluating the health score of green plants based on the acquired parameter data, and marking the area to be pruned based on the health score. The present invention sets up a health assessment module, integrates a laser radar, a multi-spectral camera, an infrared thermal imager, and a temperature and humidity sensor, synchronously collects branch and leaf density, leaf surface moisture content, temperature, and environmental data, outputs the health score of the green plants in the corresponding area to be pruned, and classifies and marks the green plants in the landscape engineering based on the health score. The classification marks are respectively low health area, medium health area, and high health area, and the comprehensiveness of the assessment is improved through multimodal data fusion.

Description

Technical Field

[0001] The present invention relates to the technical field of green plant pruning, and more particularly to an automatic green plant pruning and branch and leaf recycling system for landscape engineering. Background Art

[0002] In landscape engineering, regular pruning and maintenance of green plants are key to ensuring aesthetics and ecological health. Existing automated pruning equipment can obtain information such as branch and leaf shape and density through sensors, such as visual cameras, and judge pruning needs based on this information to see whether green plants need to be pruned and to what length.

[0003] However, existing automated pruning equipment mostly relies on a single sensor to determine pruning needs, and can only obtain branch and leaf morphology or density information. It cannot comprehensively evaluate the physiological status of plants, such as moisture content and disease. It only uses visual identification of branch and leaf density, ignoring abnormal leaf temperature caused by drought or insect pests, resulting in inaccurate pruning strategies. In addition, existing path planning is mostly based on fixed rules to optimize path length or coverage, and cannot dynamically adapt to plant health status. It does not consider the health priority of plants, resulting in high-density or diseased areas not being prioritized. In addition, when planning pruning paths, it is impossible to predict and adopt reasonable and energy-saving paths, resulting in increased energy consumption of the pruning robot arm. Summary of the Invention

[0004] In order to solve the problems in the background technology that the existing automated pruning equipment cannot comprehensively evaluate the physiological status of plants, cannot dynamically adapt to the health status of plants, and cannot predict and adopt reasonable and energy-saving paths, the present invention proposes an automatic green plant pruning and branch and leaf recovery system for landscape engineering.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a system for automatically pruning green plants and recycling branches and leaves for landscape engineering, comprising the following modules:

[0006] Data acquisition module, used to collect parameter data of green plants in landscape projects in real time;

[0007] The health assessment module is used to evaluate the health score of green plants based on the collected parameter data and mark the areas to be pruned according to the health score;

[0008] A path planning module, which is used to generate the shortest pruning paths within the landscape project based on the health scores of green plants;

[0009] The path optimization module is used to obtain historical energy consumption data and build an energy consumption prediction model, and use the energy consumption prediction model to optimize the shortest pruning path to obtain the optimized pruning path;

[0010] The execution module is used to prune branches and leaves along the optimized pruning path through the multi-degree-of-freedom pruning robot arm and collect the cut branches and leaves for recycling.

[0011] Furthermore, the parameter data includes the density of branches and leaves per unit area D 密度 , leaf spectral reflectance R 红外 , leaf surface temperature T 叶面 、Ambient temperature T 环境 and leaf surface moisture content H 含水 , by integrating laser radar to scan the green plant area, generate three-dimensional point cloud data, and calculate the branch and leaf density D 密度 , obtain the leaf surface spectral reflectance R through the multispectral camera 红外 , and inversely obtain the leaf surface moisture content H 含水 , obtain the thermal imaging image of the green plant through the infrared thermal imager, select the leaf area in the thermal image, and obtain the leaf surface temperature T 叶面 , obtain the ambient temperature T of the environment through the temperature and humidity sensor 环境 .

[0012] Further, obtain the branch and leaf density D 密度 The process includes:

[0013] Divide the 3D point cloud data into grid cells with a unit volume of V, and count the number of points N in each grid;

[0014]

[0015] Get leaf moisture content H 含水 The process includes:

[0016] Leaf spectral reflectance R 红外 Including R near infrared and R short wave infrared, leaf water content is negatively correlated with reflectance;

[0017]

[0018] Where a and b are calibration coefficients, which are determined by calibrating blade samples in the laboratory.

[0019] Furthermore, the corresponding health score P is obtained according to the parameter data:

[0020]

[0021] in:

[0022] D 标准 It is the preset standard density value of branches and leaves, which needs to be set according to the plant type;

[0023] H 阈值 is the drought warning threshold;

[0024] ω1, ω2 and ω3 are weight coefficients.

[0025] Furthermore, X and Y are set, and the health score P is compared with the set X and Y to obtain a to-be-pruned area, wherein the to-be-pruned area includes a low health area and a medium health area;

[0026] When P < X, it is marked as a low health area and needs to be pruned first;

[0027] When X≤P≤Y, it is marked as a moderately healthy area and requires selective pruning;

[0028] When P>Y, it is marked as a high health area and does not need to be pruned for the time being.

[0029] Furthermore, the process of generating the shortest pruning path in the low health area and obtaining the shortest distance in the medium health area in the path planning module includes:

[0030] A starting point and an end point are set within the landscape project, and a pruning robot arm is set at the starting point. Based on the comparison results of the health score P with X and Y, the centers of all low-health areas are connected in sequence from the starting point to generate the shortest pruning path, and the shortest moving distance of all medium-health areas from the shortest pruning path is obtained.

[0031] Furthermore, the process of acquiring historical energy consumption data and building an energy consumption prediction model in the path optimization module includes:

[0032] Factors that affect the pruning energy consumption of the pruning robot include: moving distance, branch and leaf density in the area to be pruned, branch diameter, branch moisture content, and aging and wear score of the pruning robot;

[0033] The diameter of the branches in the area to be pruned is determined through LiDAR point cloud analysis and deep learning fitting. The moisture content of the branches in the area to be pruned is determined through multispectral imaging and near-infrared spectral analysis. The aging and wear score of the pruning robot arm is determined through multi-sensor fusion and digital twin models.

[0034] The current I of the joint motor of the trimming robot arm is monitored by current sensors. Through current analysis, it is found that aging will cause increased friction and increased current.

[0035]

[0036] I 实测 Refers to the current actually measured by the current sensor, I 基准 Refers to the reference value of current under normal circumstances. For I 实测 with I 基准 The difference between

[0037] The vibration spectrum f of the trimming robot arm is collected through a vibration sensor, and abnormal frequencies, such as the bearing damage frequency band, are identified through vibration analysis.

[0038] Obtain the aging wear score S of the pruning robot arm;

[0039]

[0040] f 异常 Refers to the frequency band where the vibration spectrum is abnormal. and is the weight coefficient;

[0041] Obtain historical energy consumption data of the pruning robot arm in a single area to be pruned, the historical energy consumption data including the corresponding movement distance of the pruning robot arm when working on the single area to be pruned, the branch and foliage density of the single area to be pruned, the branch diameter of the single area to be pruned, the branch moisture content of the single area to be pruned, and the aging and wear score of the pruning robot arm and the corresponding historical pruning energy consumption;

[0042] An energy consumption prediction set is generated based on the corresponding movement distance of the pruning robot arm when working on different areas to be pruned, the branch and leaf density of the corresponding areas to be pruned, the branch diameter of the corresponding areas to be pruned, the branch moisture content of the corresponding areas to be pruned, the aging and wear score of the pruning robot arm, and the corresponding historical pruning energy consumption, and is divided into a training set and a test set;

[0043] A convolutional neural network was constructed, using the corresponding movement distance of the pruning robot arm in the training set when working on different pruning areas, the branch and leaf density of the corresponding pruning area, the branch diameter of the corresponding pruning area, the branch moisture content of the corresponding pruning area, and the aging and wear score of the pruning robot arm as the input data of the convolutional neural network, and the historical pruning energy consumption in the training set as the output data of the convolutional neural network.

[0044] The convolutional neural network is trained to obtain an initial convolutional neural network, and the initial convolutional neural network is verified using a test set, and the initial convolutional neural network with a preset test error threshold is output as an energy consumption prediction model.

[0045] Furthermore, the process of optimizing the shortest pruning path using the energy consumption prediction model to obtain the optimized pruning path in the path optimization module includes:

[0046] The shortest moving distance of each medium-health area from the shortest pruning path, the branch and leaf density of the medium-health area, the branch diameter of the medium-health area, the branch moisture content of the medium-health area, and the aging wear score of the pruning robot arm are input into the energy consumption prediction model to obtain the actual excess pruning energy required by the pruning robot arm to introduce each medium-health area based on the shortest pruning path;

[0047] The moving distance of the pruning robot arm from the current low-health area to the next adjacent low-health area, the branch and leaf density of the next low-health area, the branch diameter of the next low-health area, the branch moisture content of the next low-health area, and the aging wear score of the pruning robot arm are input into the energy consumption prediction model to obtain the actual minimum pruning energy required by the pruning robot arm on its local shortest pruning path;

[0048] The actual shortest pruning energy consumption refers to the energy consumption of the pruning robot arm working along the local shortest pruning path between two adjacent low-health areas. The actual excess pruning energy consumption refers to the energy consumption of working after adding another medium-health area between two adjacent low-health areas.

[0049] The actual excess pruning energy consumption is divided by the actual shortest pruning energy consumption to obtain the energy consumption ratio, and the energy consumption ratio is compared with the set energy consumption threshold. If the energy consumption ratio is less than the set energy consumption threshold, the corresponding medium-healthy area is included in the shortest pruning path. The actual excess pruning energy consumption and the actual shortest pruning energy consumption of all medium-healthy areas are predicted, and the corresponding energy consumption ratios are obtained respectively. All medium-healthy areas with energy consumption ratios less than the energy consumption threshold are included in the shortest pruning path to generate the final optimized pruning path of the pruning robot arm.

[0050] Furthermore, the process of pruning branches and leaves by using the multi-degree-of-freedom pruning robot arm and collecting the fallen branches and leaves for recycling includes:

[0051] According to the optimized final pruning path, the branches and leaves of the nodes are pruned by a robotic arm, the cut branches and leaves are collected by a negative pressure adsorption system, and a crushing and compression device is used for resource processing. The blade gap of the crushing and compression device can be adjusted to control the particle size of the crushed branches and leaves, and then the crushed branches and leaves are compressed and stored.

[0052] The technical effects and advantages of the automatic pruning and leaf recycling system for green plants used in landscape engineering of the present invention are as follows:

[0053] (1) By setting up a health assessment module, the laser radar, multispectral camera, infrared thermal imager and temperature and humidity sensor are integrated to synchronously collect branch and leaf density, leaf surface moisture content, temperature and environmental data, and output the health score of the green plants in the corresponding area to be pruned. The green plants in the landscape project are classified and marked according to the health score. The classification marks are low health area, medium health area and high health area respectively, and the comprehensiveness of the assessment is improved through multimodal data fusion.

[0054] (2) By setting up a path planning module and a path optimization module, the areas to be pruned are classified and graded according to their health scores, and low-health areas, medium-health areas and high-health areas are obtained. All low-health areas are connected and the initial shortest pruning path is generated. By constructing an energy consumption prediction model, the actual excess pruning energy consumption and the actual shortest pruning energy consumption of all medium-health areas are predicted, and the corresponding energy consumption ratios are obtained respectively. All medium-health areas with energy consumption ratios less than the energy consumption threshold are included in the shortest pruning path to generate the final optimized pruning path of the pruning robot arm. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 Schematic diagram of the system structure of the present invention. DETAILED DESCRIPTION

[0056] The following will provide a clear and complete description of the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0057] Reference Figure 1 , an automatic green plant pruning and branch and leaf recycling system for landscape engineering, including the following modules:

[0058] Data acquisition module, used to collect parameter data of green plants in landscape projects in real time;

[0059] The health assessment module is used to evaluate the health score of green plants based on the collected parameter data and mark the areas to be pruned according to the health score;

[0060] A path planning module, which is used to generate the shortest pruning paths within the landscape project based on the health scores of green plants;

[0061] The path optimization module is used to obtain historical energy consumption data and build an energy consumption prediction model, and use the energy consumption prediction model to optimize the shortest pruning path to obtain the optimized pruning path;

[0062] The execution module is used to prune branches and leaves along the optimized pruning path through the multi-degree-of-freedom pruning robot arm and collect the cut branches and leaves for recycling.

[0063] It should be further explained that, in the specific implementation process, the parameter data includes the density of branches and leaves per unit area D 密度 , leaf spectral reflectance R 红外 , leaf surface temperature T 叶面 、Ambient temperature T 环境 and leaf surface moisture content H 含水 ;

[0064] By integrating the laser radar to scan the green plant area, three-dimensional point cloud data is generated. Each point contains coordinate (x, y, z) information, and the branch and leaf density D is calculated. 密度 , point cloud density and branch density D 密度 Positive correlation;

[0065] The leaf surface spectral reflectance R is obtained by a multispectral camera 红外 , and inversely obtain the leaf surface moisture content H 含水 ;

[0066] Use a blackbody radiation source to calibrate the infrared camera to eliminate environmental radiation interference, obtain the thermal imaging image of the green plant through the infrared thermal imager, select the leaf area in the thermal image, and obtain the leaf surface temperature T 叶面 ;

[0067] Get the ambient temperature T of the environment through the temperature and humidity sensor 环境 ,The temperature and humidity sensor is installed on the top of the equipment and away from the heat source of the robotic arm to avoid local temperature interference.

[0068] It should be further explained that, in the specific implementation process, the density of branches and leaves D 密度 The process includes:

[0069] Divide the 3D point cloud data into grid cells with a unit volume of V (e.g., 1m³), and count the number of points N in each grid cell;

[0070]

[0071] Get leaf moisture content H 含水 The process includes:

[0072] Leaf spectral reflectance R 红外 include and , leaf surface moisture content is negatively correlated with reflectivity;

[0073]

[0074] Where a and b are calibration coefficients, which are determined by calibrating leaf samples in the laboratory;

[0075] Taking shrubs as an example, rose leaves were calibrated in the laboratory, with a being 0.85 and b being 12.3;

[0076] Taking trees as an example, camphor tree leaves were calibrated in the laboratory, with a being 1.02 and b being 8.7.

[0077] It should be further explained that, in the specific implementation process, the corresponding health score P is obtained according to the various parameter data:

[0078]

[0079] in:

[0080] D 标准 This is the preset standard density value of branches and leaves, which needs to be set according to the plant type. Take shrubs as an example: ;

[0081] H 阈值 The drought warning threshold is preset at 30%. Most landscape plants wilt when the moisture content is less than 30%, requiring emergency irrigation or pruning.

[0082] ω1, ω2, and ω3 are weight coefficients, and their initial values ​​are 0.4, 0.3, and 0.3, respectively. Historical operation data (branch and leaf density, moisture content, temperature difference) and manually evaluated health scores are collected, and ω1, ω2, and ω3 are obtained after training with a large amount of data.

[0083] It should be further explained that, in the specific implementation process, X and Y are set, and the health score P is compared with the set X and Y to obtain the area to be pruned, which includes the low health area and the medium health area:

[0084] Set X to 0.6 and Y to 0.8;

[0085] When P < 0.6, it is marked as a low health area and needs to be pruned first;

[0086] When 0.6≤P≤0.8, it is marked as a moderately healthy area and requires selective pruning;

[0087] When P>0.8, it is marked as a high health area and does not need to be trimmed temporarily;

[0088] Taking shrubs as an example, if the density of branches and leaves of shrubs is D 密度 >D 标准 , leaf surface moisture content H 含水 Reaching 40%, and there are insect pests on the leaf surface, resulting in leaf surface temperature T 叶面 >Ambient temperature T 环境 , the final P is less than 0.6, which requires priority pruning.

[0089] It should be further explained that, in a specific implementation process, the process of generating the shortest pruning path in the low health area and obtaining the shortest distance in the medium health area in the path planning module includes:

[0090] Set a starting point and an end point within the landscape project. The pruning robot arm is placed at the starting point. Based on the comparison of the health score P with X and Y, the center of all low-health areas is connected in sequence from the starting point. The connection method is based on the principle of proximity. Starting from the starting point, the center of the nearest low-health area is connected to the end point to generate the shortest pruning path.

[0091] Taking a single medium-healthy area C as an example, obtain two low-healthy areas that are closest to the center of the medium-healthy area C and are adjacent to each other on the shortest pruning path, marked as A and B. The length of the shortest path of the pruning robot arm moving from A to C and then to B is used as the shortest moving distance to add C to the shortest pruning path. The same method is used to obtain the shortest moving distance between each medium-healthy area and the shortest pruning path.

[0092] It should be further explained that, in a specific implementation process, the process of obtaining historical energy consumption data and building an energy consumption prediction model in the path optimization module includes:

[0093] Factors that affect the pruning energy consumption of the pruning robot include: moving distance, branch and leaf density in the area to be pruned, branch diameter, branch moisture content, and aging and wear score of the pruning robot;

[0094] The diameter of the branches in the area to be pruned is determined through LiDAR point cloud analysis and deep learning fitting. The moisture content of the branches in the area to be pruned is determined through multispectral imaging and near-infrared spectral analysis. The aging and wear score of the pruning robot arm is determined through multi-sensor fusion and digital twin models.

[0095] The current I of the joint motor of the trimming robot arm is monitored by current sensors. Through current analysis, it is found that aging will cause increased friction and increased current.

[0096]

[0097] I 实测 Refers to the current actually measured by the current sensor, I 基准 Refers to the reference value of current under normal circumstances. For I 实测 with I 基准 The difference between

[0098] The vibration spectrum f of the trimming robot arm is collected through a vibration sensor, and abnormal frequencies, such as the bearing damage frequency band, are identified through vibration analysis.

[0099] Obtain the aging wear score S of the pruning robot arm;

[0100]

[0101] f 异常 Refers to the frequency band where the vibration spectrum is abnormal. and is the weight coefficient, and its initial values ​​are 0.4 and 0.6 respectively. Collect historical data (f 异常 、 ) and manually evaluated aging wear scores, and ω4 and ω5 were obtained after training with a large amount of data;

[0102] Obtain historical energy consumption data of the pruning robot arm in a single area to be pruned, the historical energy consumption data including the corresponding movement distance of the pruning robot arm when working on the single area to be pruned, the branch and foliage density of the single area to be pruned, the branch diameter of the single area to be pruned, the branch moisture content of the single area to be pruned, and the aging and wear score of the pruning robot arm and the corresponding historical pruning energy consumption;

[0103] An energy consumption prediction set is generated based on the corresponding movement distance of the pruning robot arm when working on different areas to be pruned, the branch and leaf density of the corresponding areas to be pruned, the branch diameter of the corresponding areas to be pruned, the branch moisture content of the corresponding areas to be pruned, the aging and wear score of the pruning robot arm, and the corresponding historical pruning energy consumption, and is divided into a training set and a test set;

[0104] A convolutional neural network was constructed, using the corresponding movement distance of the pruning robot arm in the training set when working on different pruning areas, the branch and leaf density of the corresponding pruning area, the branch diameter of the corresponding pruning area, the branch moisture content of the corresponding pruning area, and the aging and wear score of the pruning robot arm as the input data of the convolutional neural network, and the historical pruning energy consumption in the training set as the output data of the convolutional neural network.

[0105] The convolutional neural network is trained to obtain an initial convolutional neural network, and the initial convolutional neural network is verified using a test set, and the initial convolutional neural network with a preset test error threshold is output as an energy consumption prediction model.

[0106] It should be further explained that, in a specific implementation process, the process of optimizing the shortest pruning path using the energy consumption prediction model in the path optimization module to obtain the optimized pruning path includes:

[0107] The shortest moving distance of each medium-health area from the shortest pruning path, the branch and leaf density of the medium-health area, the branch diameter of the medium-health area, the branch moisture content of the medium-health area, and the aging wear score of the pruning robot arm are input into the energy consumption prediction model to obtain the actual excess pruning energy required by the pruning robot arm to introduce each medium-health area based on the shortest pruning path;

[0108] The moving distance of the pruning robot arm from the current low-health area to the next adjacent low-health area, the branch and leaf density of the next low-health area, the branch diameter of the next low-health area, the branch moisture content of the next low-health area, and the aging wear score of the pruning robot arm are input into the energy consumption prediction model to obtain the actual minimum pruning energy required by the pruning robot arm on its local shortest pruning path;

[0109] The actual shortest pruning energy consumption refers to the energy consumption of the pruning robot arm working along the local shortest pruning path between two adjacent low-health areas. The actual excess pruning energy consumption refers to the energy consumption of working after adding another medium-health area between two adjacent low-health areas.

[0110] The actual excess pruning energy consumption is divided by the actual shortest pruning energy consumption to obtain an energy consumption ratio. The energy consumption ratio is compared with the set energy consumption threshold. If the energy consumption ratio is less than the set energy consumption threshold, the corresponding medium-healthy area is included in the shortest pruning path. The actual excess pruning energy consumption and the actual shortest pruning energy consumption of all medium-healthy areas are predicted, and the corresponding energy consumption ratios are obtained respectively. All medium-healthy areas with energy consumption ratios less than the energy consumption threshold are included in the shortest pruning path to generate the final optimized pruning path of the pruning robot arm.

[0111] In an embodiment of the present invention, by deriving the energy consumption ratio of the actual excess pruning energy consumption and the actual shortest pruning energy consumption of all medium-healthy areas, it is predicted whether the corresponding medium-healthy area will be included in the shortest pruning path, wherein the shortest pruning energy consumption is fixed, and the actual excess pruning energy consumption is mainly predicted. The actual excess pruning energy consumption is related to the movement distance of the pruning robot arm, the branch diameter of the medium-healthy area, the branch moisture content, and the aging wear score of the pruning robot arm;

[0112] The length of the moving distance of the pruning robot arm directly affects the actual excess pruning energy consumption. The longer the moving distance, the greater the actual excess pruning energy consumption and the greater the energy consumption ratio. The moving distance of the pruning robot arm is positively correlated with the energy consumption ratio.

[0113] The thickness of the branch diameter in the healthy area indirectly affects the actual excess pruning energy consumption. The thicker the branch diameter, the greater the actual excess pruning energy consumption and the greater the energy consumption ratio. The branch diameter is positively correlated with the energy consumption ratio.

[0114] The amount of water content in the branches and trunks in the healthy area indirectly affects the actual excess pruning energy consumption. The higher the water content in the branches and trunks, the greater the actual excess pruning energy consumption and the greater the energy consumption ratio. The water content in the branches and trunks is positively correlated with the energy consumption ratio.

[0115] The aging wear score of the pruning robot arm indirectly affects the actual excess pruning energy consumption. The larger the aging wear score, the greater the actual excess pruning energy consumption and the greater the energy consumption ratio. The aging wear score is positively correlated with the energy consumption ratio.

[0116] If the branches in a moderately healthy area are thicker in diameter and have a higher moisture content, the distance to reach the moderately healthy area is longer, and the aging wear score of the pruning robot arm is higher, the actual excess pruning energy consumption required is higher, the energy consumption ratio is higher, and exceeds the energy consumption threshold, then it is not recommended to include the moderately healthy area in the shortest pruning path;

[0117] Assume that the actual excess pruning energy consumption of a medium-healthy area is 990 joules, the energy consumption threshold is 1.25, and the actual shortest pruning energy consumption is 720 joules. The energy consumption ratio is 1.375, which is the actual excess pruning energy consumption of 990 joules divided by the actual shortest pruning energy consumption of 720 joules. The energy consumption ratio of 1.375 is greater than the energy consumption threshold of 1.25. Therefore, it is not recommended to include this medium-healthy area in the shortest pruning path.

[0118] It should be further explained that, in the specific implementation process, the process of pruning branches and leaves and collecting the cut branches and leaves for recycling and resource processing by the multi-degree-of-freedom pruning robot arm includes:

[0119] The branches and leaves of the nodes are pruned by a robotic arm according to the optimized path, and the cut branches and leaves are collected by a negative pressure adsorption system. The posture position and suction of the negative pressure adsorption system are adjusted to collect all the fallen branches and leaves, and a crushing and compression device is used for resource processing. The blade gap of the crushing and compression device can be adjusted to control the particle size of the crushed branches and leaves. The crushed branches and leaves are then compressed and stored, and the blade gap adjusts the particle size to adapt to different resource needs.

[0120] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

[0121] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A system for automatically pruning green plants and recycling branches and leaves for landscape engineering, characterized in that: Includes the following modules: Data acquisition module, used to collect parameter data of green plants in landscape projects in real time; The health assessment module is used to evaluate the health score of green plants based on the collected parameter data and mark the areas to be pruned according to the health score; A path planning module, which is used to generate the shortest pruning paths within the landscape project based on the health scores of green plants; The path optimization module is used to obtain historical energy consumption data and build an energy consumption prediction model, and use the energy consumption prediction model to optimize the shortest pruning path to obtain the optimized pruning path; The execution module is used to prune branches and leaves along the optimized pruning path through the multi-degree-of-freedom pruning robot arm and collect the cut branches and leaves for recycling; The process of optimizing the shortest pruning path using the energy consumption prediction model to obtain the optimized pruning path in the path optimization module includes: The shortest moving distance of each medium-health area from the shortest pruning path, the branch and leaf density of the medium-health area, the branch diameter of the medium-health area, the branch moisture content of the medium-health area, and the aging wear score of the pruning robot arm are input into the energy consumption prediction model to obtain the actual excess pruning energy required by the pruning robot arm to introduce each medium-health area based on the shortest pruning path; The moving distance of the pruning robot arm from the current low-health area to the next adjacent low-health area, the branch and leaf density of the next low-health area, the branch diameter of the next low-health area, the branch moisture content of the next low-health area, and the aging wear score of the pruning robot arm are input into the energy consumption prediction model to obtain the actual minimum pruning energy required by the pruning robot arm on its local shortest pruning path; The actual excess pruning energy consumption is divided by the actual shortest pruning energy consumption to obtain the energy consumption ratio, and the energy consumption ratio is compared with the set energy consumption threshold. If the energy consumption ratio is less than the set energy consumption threshold, the corresponding medium-healthy area is included in the shortest pruning path. The actual excess pruning energy consumption and the actual shortest pruning energy consumption of all medium-healthy areas are predicted, and the corresponding energy consumption ratios are obtained respectively. All medium-healthy areas with energy consumption ratios less than the energy consumption threshold are included in the shortest pruning path to generate the final optimized pruning path of the pruning robot arm.

2. The automatic pruning and leaf recycling system for landscape engineering according to claim 1 is characterized in that: The parameter data includes the density of branches and leaves per unit area D 密度 , leaf spectral reflectance R 红外 , leaf surface temperature T 叶面 、Ambient temperature T 环境 and leaf surface moisture content H 含水 , by integrating laser radar to scan the green plant area, generate three-dimensional point cloud data, and calculate the branch and leaf density D 密度 , obtain the leaf surface spectral reflectance R through the multispectral camera 红外 , and inversely obtain the leaf surface moisture content H 含水 , obtain the thermal imaging image of the green plant through the infrared thermal imager, select the leaf area in the thermal image, and obtain the leaf surface temperature T 叶面 , obtain the ambient temperature T of the environment through the temperature and humidity sensor 环境 .

3. The automatic pruning and leaf recycling system for landscape engineering according to claim 2 is characterized in that: Get the branch and leaf density D 密度 The process includes: Divide the 3D point cloud data into grid cells with a unit volume of V, and count the number of points N in each grid cell; Get leaf moisture content H 含水 The process includes: Leaf spectral reflectance R 红外 include and , leaf surface moisture content is negatively correlated with reflectivity; Where a and b are calibration coefficients, which are determined by calibrating blade samples in the laboratory.

4. The automatic pruning and leaf recycling system for landscape engineering according to claim 3 is characterized in that: Obtain the corresponding health score P based on various parameter data: in: D 标准 It is the preset standard density value of branches and leaves, which needs to be set according to the plant type; H 阈值 is the drought warning threshold; ω1, ω2 and ω3 are weight coefficients.

5. The automatic green plant pruning and branch and leaf recycling system for landscape engineering according to claim 4 is characterized in that: Set X and Y, and compare the health score P with the set X and Y to obtain a to-be-pruned area, where the to-be-pruned area includes a low health area and a medium health area; When P < X, it is marked as a low health area and needs to be pruned first; When X≤P≤Y, it is marked as a moderately healthy area and requires selective pruning; When P>Y, it is marked as a high health area and does not need to be pruned for the time being.

6. The automatic green plant pruning and branch and leaf recycling system for landscape engineering according to claim 5, characterized in that: The process of generating the shortest pruning path in the low health area and obtaining the shortest distance in the medium health area in the path planning module includes: A starting point and an end point are set within the landscape project, and a pruning robot arm is set at the starting point. Based on the comparison results of the health score P with X and Y, the centers of all low-health areas are connected in sequence from the starting point to generate the shortest pruning path, and the shortest moving distance of all medium-health areas from the shortest pruning path is obtained.

7. The automatic green plant pruning and branch and leaf recycling system for landscape engineering according to claim 6, characterized in that: The process of acquiring historical energy consumption data and building an energy consumption prediction model in the path optimization module includes: Factors that affect the pruning energy consumption of the pruning robot include: moving distance, branch and leaf density in the area to be pruned, branch diameter, branch moisture content, and aging and wear score of the pruning robot; The diameter of the branches in the area to be pruned is determined through LiDAR point cloud analysis and deep learning fitting. The moisture content of the branches in the area to be pruned is determined through multispectral imaging and near-infrared spectral analysis. The aging and wear score of the pruning robot arm is determined through multi-sensor fusion and digital twin models. The current I of the joint motor of the trimming robot arm is monitored by current sensors. Through current analysis, it is found that aging will cause increased friction and increased current. I 实测 Refers to the current actually measured by the current sensor, I 基准 Refers to the reference value of current under normal circumstances. For I 实测 with I 基准 The difference between The vibration spectrum f of the trimming robot arm is collected by a vibration sensor, and abnormal frequencies are identified through vibration analysis; Obtain the aging wear score S of the pruning robot arm; f 异常 Refers to the frequency band where the vibration spectrum is abnormal. and is the weight coefficient; Obtain historical energy consumption data of the pruning robot arm in a single area to be pruned, the historical energy consumption data including the corresponding movement distance of the pruning robot arm when working on the single area to be pruned, the branch and foliage density of the single area to be pruned, the branch diameter of the single area to be pruned, the branch moisture content of the single area to be pruned, and the aging and wear score of the pruning robot arm and the corresponding historical pruning energy consumption; An energy consumption prediction set is generated based on the corresponding movement distance of the pruning robot arm when working on different areas to be pruned, the branch and leaf density of the corresponding areas to be pruned, the branch diameter of the corresponding areas to be pruned, the branch moisture content of the corresponding areas to be pruned, the aging and wear score of the pruning robot arm, and the corresponding historical pruning energy consumption, and is divided into a training set and a test set; A convolutional neural network was constructed, using the corresponding movement distance of the pruning robot arm in the training set when working on different pruning areas, the branch and leaf density of the corresponding pruning area, the branch diameter of the corresponding pruning area, the branch moisture content of the corresponding pruning area, and the aging and wear score of the pruning robot arm as the input data of the convolutional neural network, and the historical pruning energy consumption in the training set as the output data of the convolutional neural network. The convolutional neural network is trained to obtain an initial convolutional neural network, and the initial convolutional neural network is verified using a test set, and the initial convolutional neural network with a preset test error threshold is output as an energy consumption prediction model.

8. The automatic green plant pruning and branch and leaf recycling system for landscape engineering according to claim 7, characterized in that: The actual shortest pruning energy consumption refers to the energy consumption of the pruning robot arm working along the local shortest pruning path between two adjacent low-health areas, and the actual excess pruning energy consumption refers to the energy consumption after adding another medium-health area between two adjacent low-health areas.

9. The automatic green plant pruning and branch and leaf recycling system for landscape engineering according to claim 8, characterized in that: The process of pruning branches and leaves, collecting fallen branches and leaves, and recycling them for resource processing by the multi-degree-of-freedom pruning robot arm includes: According to the optimized final pruning path, the branches and leaves of the nodes are pruned by a robotic arm, the cut branches and leaves are collected by a negative pressure adsorption system, and a crushing and compression device is used for resource processing. The blade gap of the crushing and compression device can be adjusted to control the particle size of the crushed branches and leaves, and then the crushed branches and leaves are compressed and stored.

Images