Intelligent control method and system for tower crane

Abstract

The invention discloses an intelligent control method and system for a tower crane. The method comprises the following steps: acquiring acquired data and weather conditions from a data acquisition module on the tower crane; preprocessing the collected data according to weather conditions to obtain preprocessed data; acquiring the current position, the target position and the working environment information of the tower crane according to the preprocessed data; judging whether an obstacle exists in the working process according to the current position, the target position and the working environment information; when an obstacle exists, an improved ant algorithm is adopted to carry out obstacle avoidance planning to obtain an optimal working path; and controlling the tower crane to operate according to the planned optimal working path. According to the method, related information is obtained from the preprocessed data to perform optimal path planning, so that the accuracy of path planning can be improved, safety factors of surrounding points are added into an ant algorithm, and the safety and operation efficiency of intelligent control of the tower crane can be improved by constraint penalty terms of swinging of a flexible body and a steel wire rope of the tower crane.

Description

Technical Field

[0001] The present invention relates to the technical field of tower crane operation control, and particularly to a smart management and control method and system for tower cranes. Background Art

[0002] Adverse weather can cause many impacts and challenges to the operation of tower cranes. For example: In windy weather, the boom and jib of the tower crane will be affected by a large lateral wind force, resulting in increased swaying of the tower crane, making it difficult to position accurately and perform stable lifting; a sudden change in wind direction will cause a change in the force direction of the tower crane, and the slewing mechanism of the tower crane needs to be adjusted frequently to adapt to the new wind direction, increasing the risk of mechanical failures and also increasing the operation difficulty. A slight mistake may lead to a collision between the tower crane and surrounding buildings or other tower cranes. Heavy rain weather will obstruct the operator's line of sight, making it difficult to clearly observe the operating state of the tower crane, the position of the hook, and the surrounding environmental conditions. In extremely low visibility conditions, the operator may not be able to accurately judge the distance between the hook and buildings or nearby tower cranes, increasing the probability of collision. In foggy weather, the signals of lidar may be scattered by the fog, making it impossible for the tower crane to accurately obtain the distance and azimuth between itself and other obstacles, and the anti-collision system is difficult to work effectively. Snowy weather will form a large amount of snow on the surface of the tower crane structure, increasing the load of the tower crane. Excessive snow load may cause deformation or even collapse of the tower crane structure. Therefore, a smart management and control method for tower cranes is needed to cope with various adverse weather conditions so that the tower crane can still operate normally in adverse weather. Summary of the Invention

[0003] The purpose of the present invention is to provide a smart management and control method and system for tower cranes, so that the tower crane is not affected by adverse weather and can operate normally, and can ensure that the work progress is not affected by the weather.

[0004] The present invention is achieved through the following technical solutions:

[0005] In a first aspect, a smart management and control method for a tower crane provided by a first embodiment of the present invention includes the following steps:

[0006] Obtain the collected data and weather conditions from the data acquisition module on the tower crane, and the collected data includes point cloud data and images;

[0007] Preprocess the collected data according to the weather conditions to obtain preprocessed data;

[0008] Obtain the current position, target position, and working environment information of the tower crane according to the preprocessed data;

[0009] Judge whether there are obstacles during the working process according to the current position, target position, and working environment information;

[0010] If not, perform normal planning to obtain the optimal working path;

[0011] If it exists, use an improved ant algorithm for obstacle avoidance planning to obtain the optimal working path. In the improved ant algorithm, constraints such as the safety factor of surrounding points, the flexible body of the tower crane, and the swing of the steel wire rope are added for constraint;

[0012] Control the operation of the tower crane according to the planned optimal working path.

[0013] Furthermore, the specific method for preprocessing the collected data according to the weather conditions to obtain the preprocessed data includes:

[0014] Use a point cloud denoising algorithm to segment weather noise points in the distorted point cloud data and remove the weather noise points from the point cloud data in a mask manner to obtain the point cloud data after removing noise points.

[0015] Furthermore, the image includes a foggy day image. The specific method for preprocessing the foggy day image according to the weather conditions to obtain the preprocessed data further includes:

[0016] Divide the foggy day image into multiple patch images;

[0017] Find the minimum value of the dark channel of each pixel point in each patch image;

[0018] The generator judges the darkest pixel information in the local area of the image according to the minimum value, and converts the foggy image into a defogged image;

[0019] The discriminator compares the difference between the defogged image and the real fog-free image to judge whether to output the defogged image;

[0020] Output the defogged image to obtain the image after defogging.

[0021] Furthermore, the weather conditions include rainy days, the image includes rainy day images, and the specific method for preprocessing the rainy day images according to the weather conditions to obtain the preprocessed data includes:

[0022] Use 4 cascaded residual fast Fourier transform convolutional modules as the backbone network to extract rain streak information from the rainy day image to obtain the rain streak region;

[0023] Use the context interaction Transformer module to combine context information to perform image restoration on the rain streak region and output the restored rain-free image.

[0024] Furthermore, the image includes snowy day images. The specific method for preprocessing the snowy day images according to the weather conditions to obtain the preprocessed data includes:

[0025] The snow image is decomposed into a first high-frequency component and a first low-frequency component by using the dual-tree complex wavelet transform;

[0026] The first low-frequency component is further decomposed into a second high-frequency component and a second low-frequency component at the next level, and so on until the b-th level is decomposed, where b > 0;

[0027] The pre-defined high-frequency reconstruction network is applied to the high-frequency components at different levels to reconstruct the reconstructed high-frequency components;

[0028] The pre-defined low-frequency reconstruction network is applied to the low-frequency components at different levels to reconstruct the reconstructed low-frequency components;

[0029] The reconstructed low-frequency components and high-frequency components are fused by the inverse dual-tree complex wavelet transform to obtain the snow-free image after fusion.

[0030] Furthermore, the specific method for obstacle avoidance planning using the improved ant algorithm to obtain the optimal working path includes:

[0031] An environment model is constructed using the grid method according to the operation environment information, and the initial point and the target point are set corresponding to the current position and the target position;

[0032] The particle swarm optimization algorithm is used to optimize the parameters of the ant colony algorithm, and the best combination of each parameter is found according to the optimization result;

[0033] The pheromone value, the current iteration number and the maximum iteration number at the initial moment of path planning are set, and the initial point is added to the path point taboo table;

[0034] Search for the path. The ant colony starts from the initial point and selects the next path node in the adjacent grid according to the improved state transition probability. The constraint penalty terms of the safety factor of the surrounding points, the flexible body of the tower crane and the swing of the steel wire rope are added to the improved state transition probability;

[0035] After an ant completes a search, the pheromone of the passed path is updated;

[0036] Judge whether the ant reaches the target point;

[0037] If the target point is not reached, continue to return to search for the path;

[0038] If the target point is reached, the search is completed;

[0039] When all ants complete an iteration process, the global pheromone concentration except the optimal path is updated;

[0040] When the iteration number is reached, the planning is completed, and the node order corresponding to the global optimal path in the taboo table is recorded to obtain the optimal working path;

[0041] The motion paths of the joints of the tower crane hook are obtained by inverse kinematics according to the optimal working path.

[0042] Further, the formula for the improved state transition probability is as follows:

[0043]

[0044] Among them, P ij m (t) represents the probability that ant m moves from the current node to the next node at time t, and τ ij (t) represents the amount of pheromone residue on the path from node i to node j by the ant at time t, and ξ ij is the heuristic function, representing the expected degree of the ant's transfer from node i to node j. α and β are the relative importance parameters of the pheromone and the heuristic function respectively. allowed m represents the node selected by ant m for the next step at time t, and ω o is the number of grids occupied by obstacles near node j, and ω s is the number of all grids near node j. γ is the weight value of the safety factor of surrounding points, and the value of γ is [0, 1]. τ ik (t) represents the amount of pheromone residue on the path from node i to node k by the ant at time t, and ξ ik is the heuristic function, representing the expected degree of the ant's transfer from node i to node k. is the constraint fitness function considering the deformation of the flexible body and the swing of the wire rope when moving from node i to node j. ε is the weight value of the constraint fitness function. is the constraint fitness function considering the deformation of the flexible body and the swing of the wire rope when moving from node i to node k. d ij is the distance between node i and node j, and d jq is the distance from node j to the target point. δ is the weight value of the distance from the next node to the target point, and the value of δ is [0, 0.5]. χ ij represents the maximum deformation of the flexible body of the tower crane when moving from node i to node j, and χ max is the maximum allowable deformation of the flexible body of the tower crane. λ is the constraint penalty coefficient used to adjust the influence degree of the constraint condition. θ ij is the maximum swing angle of the wire rope representing when moving from node i to node j, and θ max is the maximum allowable swing angle of the wire rope. φ1 and φ2 are the weight coefficients of the flexible body deformation and the wire rope swing angle respectively, satisfying φ1 + φ2 = 1.

[0045] Further, the formula for updating the pheromone on the passed path is as follows:

[0046] τij (t + 1) = (1 - ρ)τ ij (t) + Δτ ij (t);

[0047]

[0048]

[0049] Where ρ is the evaporation coefficient of pheromone, 0 < ρ < 1, and Δτ ij (t) is the sum of the pheromone concentrations released by all ants on the path from node i to node j in this iteration, and Δτ ij m (t) is the pheromone concentration released by ant m on the path from node i to node j, and Q1 is a constant representing the local pheromone intensity.

[0050] Furthermore, the method further includes: judging the operating condition of the tower crane. If the tower crane reaches the target position safely and controllably, the process ends. If it does not reach the target position, the tower crane is controlled to continue running along the planned optimal path.

[0051] In a second aspect, a tower crane intelligent control and management system provided by another embodiment of the present invention is used to implement the tower crane intelligent control and management method described in the first embodiment, and includes: a data acquisition module, a preprocessing module, an information acquisition module, and a control module;

[0052] The data acquisition module is used to obtain the collected data and weather conditions from the data acquisition module on the tower crane, and the collected data includes point cloud data and images;

[0053] The preprocessing module is used to preprocess the collected data according to the weather conditions to obtain preprocessed data;

[0054] The information acquisition module obtains the current position, target position, and working environment information of the tower crane according to the preprocessed data;

[0055] The control module judges whether there are obstacles during the working process according to the current position, target position, and working environment information. If not, normal planning is performed to obtain the optimal working path. If so, obstacle avoidance planning is performed to obtain the optimal working path. Constraints on the safety factors of surrounding points, the flexible body of the tower crane, and the swing of the steel wire rope are added to the improved ant algorithm for constraint, and the tower crane is controlled to run according to the planned optimal working path.

[0056] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0057] A tower crane intelligent control method and system provided by an embodiment of the present invention preprocess the collected data according to weather conditions to obtain preprocessed data. The preprocessed data reduces the impact of weather on the collected data. Obtaining relevant information from the preprocessed data for optimal path planning can improve the accuracy of path planning. Thus, the control of the tower crane is not affected by bad weather and can still perform precise and efficient control according to the planned optimal path.

[0058] During the process of obstacle avoidance path planning, the heuristic function is improved by combining the safety factors of surrounding points, which increases the effectiveness of the ant algorithm, improves the efficiency of the algorithm, and speeds up the convergence speed. The ant algorithm is improved by combining the constraint conditions of the flexible body of the tower crane and the swing of the steel wire rope, guiding the ants to search for the optimal path that not only meets the flexible movement of the tower crane and the swing characteristics of the steel wire rope but also can avoid obstacles. Using the improved ant algorithm in the tower crane intelligent control method can effectively avoid safety problems and low operation efficiency caused by excessive deformation of the flexible body or excessive swing amplitude of the steel wire rope, and improve the safety and operation efficiency of the tower crane intelligent control. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] In order to more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the drawings required for the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as limiting the scope. For those of ordinary skill in the art, other related drawings can be obtained based on these drawings without creative efforts. In the drawings:

[0060] Figure 1 is a flowchart of a tower crane intelligent control method provided by the first embodiment of the present invention;

[0061] Figure 2 is a structural block diagram of a tower crane intelligent control system provided by another embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0062] To make the objectives, technical solutions, and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below in combination with the embodiments and the drawings. The illustrative embodiments and descriptions of the present invention are only used to explain the present invention and are not intended to limit the present invention.

[0063] As Figure 1 shown, a tower crane intelligent control method provided by the first embodiment of the present invention includes the following steps:

[0064] Obtain the collected data and weather conditions from the data collection module on the tower crane, where the collected data includes point cloud data and images;

[0065] Preprocess the collected data according to the weather conditions to obtain the preprocessed data;

[0066] Obtain the current position, target position and working environment information of the tower crane according to the preprocessed data;

[0067] Judge whether there are obstacles during the working process according to the current position, target position and working environment information;

[0068] If not, perform normal planning to obtain the optimal working path;

[0069] If there are obstacles, use the improved ant algorithm for obstacle avoidance planning to obtain the optimal working path. In the improved ant algorithm, constraints of surrounding point safety factors, flexible body of the tower crane and wire rope swing are added;

[0070] Control the operation of the tower crane according to the planned optimal working path.

[0071] The data acquisition module installed on the tower crane includes an industrial-grade high-definition camera, an optical camera, and a lidar. Cameras are installed on the tower crane to obtain visual images. Two industrial-grade high-definition cameras are installed at both ends of the tower crane boom, with a field of view angle of 120 degrees, a resolution of 2560×1920 pixels, a frame rate of 30fps, and a baseline distance of 1.5 meters between the two cameras, ensuring complete coverage of the construction area below the boom. Information is obtained through the optical camera by implementing optical-to-digital conversion. The lidar is used for the point cloud data corresponding to the surrounding environment of the tower crane. Point cloud is a data type with sparse distribution, disordered arrangement, and unstructured. After scanning and obtaining environmental information through various methods, an environmental modeling method is used to construct the working map of the tower crane and identify and process obstacles. Cameras are installed to obtain visual images, and information is obtained through the optical camera by implementing optical-to-digital conversion. After scanning and obtaining environmental information through various methods, an environmental modeling method is used to construct the working map of the tower crane and identify and process obstacles. Since the point cloud data collected by the lidar under bad weather is distorted to varying degrees due to scattering and refraction, which will affect the path planning of the tower crane. Therefore, in this embodiment, the collected data is preprocessed according to the weather conditions to obtain preprocessed data, improving the reliability of the data. According to the preprocessed data, the current position, target position, and working environment information of the tower crane are obtained. According to the current position, target position, and working environment information, it is judged whether there are obstacles during the working process. If not, normal planning is carried out to obtain the optimal working path; if so, an improved ant algorithm is used for obstacle avoidance planning to obtain the optimal working path. In the improved ant algorithm, constraint conditions such as the safety factor of surrounding points, the flexible body of the tower crane, and the swing of the wire rope are added to improve the safety of the planned path; the tower crane is controlled to operate according to the planned optimal working path.

[0072] In another embodiment of the present invention, the specific method for preprocessing the collected data according to the weather conditions to obtain preprocessed data includes:

[0073] The point cloud denoising algorithm is used to segment weather noise points in the distorted point cloud data and remove the weather noise points from the point cloud data in the form of a mask to obtain the point cloud data after denoising. The 3D DenoiseNet network structure is used for point cloud denoising. The 3D DenoiseNet network structure includes a point cloud data structuring module and a spatial KNN encoding module. A single-branch network is used for input and output, ensuring the segmentation accuracy of weather noise points. The running process of this algorithm is as follows: input the current frame of point cloud data, project the three-dimensional point cloud coordinates in the Cartesian coordinate system onto the spherical coordinate system; then perform KNN convolution on adjacent points in the three-dimensional space, and perform spatial feature encoding through Residual Block, Average Pooling, Residual Block, and Dropout; reduce the channel dimension and restore the spatial dimension through Pixel Shufle, splice the spatial features of the current stage with the previously encoded spatial features in the channel dimension in the Residual Block, and obtain the bad weather segmentation result through standard convolution and Sofmax; finally, remove the weather factor noise points and retain the valid points of the object and the environment in the spherical coordinate system through mask operation, and further convert them into three-dimensional points in the Cartesian coordinate system.

[0074] In another embodiment of the present invention, the image includes a foggy day image. The specific method for preprocessing the foggy day image according to the weather conditions to obtain the preprocessed data further includes:

[0075] Divide the foggy day image into multiple patch images;

[0076] Find the minimum value of the dark channel of each pixel point in each patch image;

[0077] The generator judges the darkest pixel information in the local area of the image according to the minimum value, and converts the foggy image into a defogged image;

[0078] The discriminator compares the differences between the defogged image and the real fog-free image to judge whether to output the defogged image;

[0079] Output the defogged image to obtain the image after defogging.

[0080] For example, for a foggy image, it is divided into multiple patch images. Within each patch image, find the minimum value of the dark channels (i.e., the minimum value among the three color channels) of each pixel point in the patch image. This minimum value can more accurately reflect the darkest pixel information in this local area. This can avoid the interference to the dark channel estimation caused by the presence of bright objects in a larger area in the traditional dark channel prior method and improve the accuracy of transmittance estimation. The generative adversarial network is used for defogging. The generative adversarial network consists of a generator and a discriminator. In the image defogging task, the role of the generator is to convert the foggy image into a fog-free image. The discriminator is used to distinguish the generated fog-free image from the real fog-free image. The generator is usually a convolutional neural network (CNN) structure. It receives the foggy image as input, and after a series of operations such as convolution, activation, and upsampling, it outputs the defogged image. The discriminator is also a CNN. It receives an image (which can be the image generated by the generator or the real fog-free image) as input, compares the difference between the defogged image and the original real fog-free image, and determines whether to output the defogged image. This method has good adaptability to different weather conditions and scenes. For a light fog scene, the dark channel prior part based on Patch Map may be able to work well because the transmittance is relatively high at this time and the dark channel information of the local area is relatively easy to extract. In a thick fog scene, the generative adversarial network part can better restore the structure and color information of the image through the learned complex mapping relationship, making up for the deficiency that the transmittance estimation of the dark channel prior method may not be accurate enough in the case of thick fog. This method can effectively remove the fog in the foggy image and eliminate the impact of fog on image information acquisition.

[0081] In another embodiment of the present invention, the weather condition includes rainy days, the image includes rainy-day images, and the specific method for preprocessing the rainy-day images according to the weather condition to obtain the preprocessed data includes:

[0082] Four cascaded residual fast Fourier transform convolutional modules are used as the backbone network to extract rain streak information from the rainy-day images to obtain the rain streak region;

[0083] The context interaction Transformer module is used to combine the context information to perform image restoration on the rain streak region and output the restored rain-free image.

[0084] When performing rain removal on rainy images, four cascaded residual fast Fourier transform convolutional modules are used as the backbone network to extract rain streak information from rainy images, providing low-frequency information for the context interaction Transformer module and capturing the interaction between long-term and short-term information. The context interaction Transformer module not only inherits the original self-attention ability of the Transformer module but also realizes the mining of context information combining dynamic and static. By considering the neighborhood information around the rain streaks, the de-rained image is made more consistent with the original clean rain-free image, and the network uses a multi-stage method to remove rainwater, using the output of each stage to better guide the rain removal work of the next stage. The work of each stage is to remove the rain streaks remaining from the previous stage. While removing the rain streaks, this method also combines the context information of the rain streak area of the image to perform image restoration work on the rain streak area, greatly improving the quality of the clean image after rain removal and eliminating the impact of rain streaks on image information acquisition.

[0085] In another embodiment of the present invention, the image includes a snowy image. The specific method for preprocessing the snowy image according to the weather conditions to obtain preprocessed data includes:

[0086] Using the dual-tree complex wavelet transform to decompose the snowy image into a first high-frequency component and a first low-frequency component;

[0087] Continuing to decompose the first low-frequency component into a second high-frequency component and a second low-frequency component at the next level, and so on until the b-th level is decomposed, where b > 0;

[0088] Applying a predefined high-frequency reconstruction network to reconstruct the high-frequency components at different levels to obtain the reconstructed high-frequency components;

[0089] Applying a predefined low-frequency reconstruction network to reconstruct the low-frequency components at different levels to obtain the reconstructed low-frequency components;

[0090] Fusing the reconstructed low-frequency components and high-frequency components through the inverse dual-tree complex wavelet transform to obtain a fused snow-free image.

[0091] The high-frequency reconstruction network is responsible for eliminating the influence of small-sized snow particles and restoring detail information, and the low-frequency reconstruction network is used to remove large-sized snow particles and restore structural information. The reconstructed low-frequency components and high-frequency components are fused through the inverse dual-tree complex wavelet transform at different levels. Through recursive decomposition and reconstruction, it helps the network better handle snow particles of different scales. At the same time, this network also proposes a contradiction channel loss. The more residual snow in the image, the greater the corresponding loss value. Therefore, this loss can be used to more thoroughly remove the snow particles in the image, turning the snowy image into a snow-free image and reducing the impact of the snowy image on image information acquisition.

[0092] By performing preprocessing using different data processing methods according to different weather conditions as described above, the data obtained after preprocessing can reduce the adverse effects of weather on the collected data and ensure the reliability of image quality.

[0093] Obtain the current position, target position, and working environment information of the tower crane based on the preprocessed data. To determine whether there are obstacles during the working process according to the current position, target position, and working environment information, specifically include:

[0094] Determining the starting point and ending point based on the current position and target position can be achieved using existing technologies. Extract and match feature points from the preprocessed data, construct the initial three-dimensional point cloud of the lower area of the tower arm, optimize and fuse the initial three-dimensional point cloud using the iterative closest point registration algorithm, and identify and extract the spatial contours of the tower crane, hook, boom, material area, and destination based on the semantic segmentation method. Map the feature points of the spatial contours to the working environment coordinate system through coordinate transformation, establish a passable area map within the movement range of the hook using the adaptive voxel grid division algorithm, and calibrate the initial obstacle positions through the region growing method. Based on the detection data of the millimeter-wave radars on both sides of the boom and the ultrasonic radar at the hook, identify the obstacle contours using the adaptive threshold segmentation algorithm.

[0095] In another embodiment of the present invention, the specific method for performing obstacle avoidance planning using an improved ant algorithm to obtain the optimal working path includes:

[0096] Construct an environmental model using the grid method according to the working environment information, and set the initial point and target point corresponding to the current position and target position. Divide the working environment space into grids of equal volume. The size of the grid determines the fineness of the environmental model. And the accuracy of the grid will affect the running speed of the ant colony algorithm. The higher the fineness of the grid division, the longer the time spent by the ant colony during the search process. Therefore, an appropriate accuracy should be selected when establishing the grid map. Each grid in the grid map represents a path point, and the obstacle-free area and the obstacle area can be distinguished by filling different colors in the grid.

[0097] Optimize the parameters of the ant colony algorithm using the particle swarm algorithm, and find the best combination of each parameter according to the optimization results.

[0098] Algorithm initialization: Set the pheromone value τ ij (0) = τ0, set the current iteration number N = 0 and the maximum iteration number N max . Construct a path point taboo table, and add the initial point S to the path point taboo table; the ant colony will start from the initial point S and begin to explore the task space.

[0099] Search path: Starting from the initial point, the ant colony selects the next path node in the adjacent grids according to the improved state transition probability, and constraint penalty terms for the safety factor of surrounding points, the flexible body of the tower crane, and the swing of the steel wire rope are added to the improved state transition probability. After obtaining the selection probabilities of each adjacent node of the ant using the improved state transition probability, the next node is selected by the roulette method.

[0100] Among them, the formula for the improved state transition probability is:

[0101]

[0102] Among them, P ij m (t) represents the probability that ant m moves from the current node to the next node at time t, τ ij (t) represents the amount of pheromone residue on the path from node i to node j by the ant at time t, ξ ij is the heuristic function, representing the expected degree of the ant's transfer from node i to node j. α and β are respectively the relative importance parameters of the pheromone and the heuristic function. allowed m represents the node selected by ant m for the next move at time t, ω o is the number of grids occupied by obstacles near node j, ω s is the number of all grids near node j. γ is the weight value of the safety factor of surrounding points, and the value of γ is [0,1]. τ ik (t) represents the amount of pheromone residue on the path from node i to node k by the ant at time t, ξ ik is the heuristic function, representing the expected degree of the ant's transfer from node i to node k, is the constraint fitness function considering the deformation of the flexible body and the swing of the steel wire rope when moving from node i to node j; is the constraint fitness function considering the deformation of the flexible body and the swing of the steel wire rope when moving from node i to node j; d ij is the distance between node i and node j, d jt is the distance from node j to the target point. δ is the weight value of the distance from the next node to the target point, and the value of δ is [0,0.5]. χ ij represents the maximum deformation amount generated by the flexible body of the tower crane when moving from node i to node j. χ max is the maximum allowable deformation amount of the flexible body of the tower crane. λ is the constraint penalty coefficient used to adjust the influence degree of the constraint conditions. θ ij represents the maximum swing angle of the steel wire rope when moving from node i to node j. θ max is the maximum allowable swing angle of the steel wire rope. φ1 and φ2 are respectively the weight coefficients of the deformation of the flexible body and the swing angle of the steel wire rope, and satisfy φ1 + φ2 = 1.

[0103] The improved state transition probability formula mentioned above can achieve the following functions by constraining the fitness function:

[0104] When the deformation of the flexible body or the swing angle of the wire rope caused by the path approaches the allowable limit, the value will approach 0, thereby reducing the probability of this path being selected;

[0105] When the path meets the constraint conditions, the value is close to 1 and has little impact on the state transition probability;

[0106] By adjusting parameters such as λ, φ1, and φ2, the relationship between the path length and the constraint conditions can be flexibly balanced.

[0107] By improving the ant algorithm by combining the constraint conditions of the flexible body and the wire rope swing of the tower crane, guiding the ants to search for the optimal path that not only meets the flexible movement of the tower crane and the wire rope swing characteristics but also can avoid obstacles. Using the improved ant algorithm in the intelligent control method of the tower crane can effectively avoid safety problems and low operation efficiency caused by excessive deformation of the flexible body or excessive swing amplitude of the wire rope, and improve the safety and operation efficiency of the intelligent control of the tower crane.

[0108] Local pheromone update: After the ant completes a search, the following method is used to update the pheromone of the path it has passed;

[0109] τ ij (t + 1)=(1 - ρ)τ ij (t)+Δτ ij (t);

[0110]

[0111]

[0112] Among them, ρ is the evaporation coefficient of the pheromone, 0 < ρ < 1, Δτ ij (t) is the sum of the pheromone concentrations released by all ants on the path from node i to node j in this iteration, Δτ ij m (t) is the pheromone concentration released by ant m on the path from node i to node j, Q1 is a constant representing the local pheromone intensity, d ij represents the distance on the path from node i to node j.

[0113] Judge whether the ant reaches the target point: Every time the ant makes a state transition, it needs to judge the current node. If it has not reached the target point, it returns to continue searching for the path; if it reaches the target point, it executes the next step.

[0114] Completion of search: When an ant reaches the target point T from the starting point, completing a path search, the path path(S, T) can be obtained according to the taboo table, and the length of this path is calculated. At this time, it is necessary to determine whether there are redundant inflection points. If there are, a taboo table for redundant inflection points is constructed, and the inflection point is temporarily included in the taboo search redundant inflection point set and retreated to the starting position of the broken line segment, and the search path step is returned; if not, the obtained path path(S, T) is compared with the path lengths obtained by other ants in this iteration to obtain the shortest path pathmin(N) in this iteration. The optimal path of this iteration is compared with the optimal paths of previous iterations. If the path of this iteration is shorter, then let pathmin = pathmin(N), and record the node order in the path point taboo table.

[0115] Global pheromone update: When all ants complete an iteration process, the first two formulas of local pheromone update are used to update the global pheromone concentration except for the optimal path. Among them, the global pheromone update adopts the ant-cycle model, and the formula is:

[0116]

[0117] where Q2 is a constant representing the total amount of pheromone released by an ant in one cycle; L m represents the length of the path passed by the m-th ant.

[0118] When the iteration times are reached, the planning is completed, and the node order in the taboo table corresponding to the global optimal path is recorded to obtain the best working path.

[0119] According to the best working path, the motion paths of the joints of the tower crane hook are obtained by inverse kinematics solution.

[0120] Through the above method steps, an obstacle avoidance path planning using an improved ant algorithm is realized. When improving the ant algorithm, the heuristic function is improved by combining the safety factors of surrounding points, which can increase the effectiveness of the ant algorithm, improve the efficiency of the algorithm, and accelerate the convergence speed. The ant algorithm is improved by combining the constraints of the flexible body of the tower crane and the swing of the steel wire rope, guiding the ants to search for the optimal path that not only meets the flexible motion of the tower crane and the swing characteristics of the steel wire rope but also can avoid obstacles.

[0121] In another embodiment of the present invention, the intelligent management and control method for tower cranes further includes: judging the operation condition of the tower crane. If the tower crane reaches the target position safely and controllably, the process ends. If it does not reach the target position, the tower crane is controlled to continue running along the planned optimal path. The method also detects and judges the operation condition of the tower crane, facilitating to know whether the operation condition of the tower crane is safe and controllable. If the tower crane reaches the target position safely and controllably, it ends normally. If it does not reach the target position, it continues to run along the planned optimal path. If the operation of the tower crane is abnormal, it ends abnormally and corresponding prompt information is given. In this way, it can be known in real time whether the tower crane is operating normally, and corresponding prompt information can be obtained in time in case of abnormality.

[0122] A kind of intelligent management and control method for tower cranes provided by an embodiment of the present invention preprocesses the collected data according to the weather condition to obtain preprocessed data. The preprocessed data reduces the influence of the weather on the collected data. Obtaining relevant information from the preprocessed data for optimal path planning can improve the accuracy of path planning. The management and control of the tower crane are not affected by bad weather, and it can still be accurately and efficiently managed and controlled according to the planned optimal path.

[0123] During the process of obstacle avoidance path planning, the heuristic function is improved by combining the safety factors of surrounding points, which increases the effectiveness of the ant algorithm, improves the efficiency of the algorithm, and speeds up the convergence speed. The ant algorithm is improved by combining the constraint conditions of the flexible body of the tower crane and the swing of the steel wire rope, guiding the ants to search for the optimal path that not only meets the flexible movement of the tower crane and the swing characteristics of the steel wire rope but also can avoid obstacles. Using the improved ant algorithm in the intelligent management and control method of tower cranes can effectively avoid safety problems and low operation efficiency caused by excessive deformation of the flexible body or excessive swing amplitude of the steel wire rope, and improve the safety and operation efficiency of the intelligent management and control of tower cranes.

[0124] As Figure 2 shown, an intelligent management and control system for tower cranes provided by another embodiment of the present invention is used to implement the intelligent management and control method for tower cranes described in the above first embodiment, and includes: a data acquisition module, a preprocessing module, an information acquisition module, and a control module;

[0125] The data acquisition module is used to obtain the collected data and the weather condition from the data acquisition module on the tower crane, and the collected data includes point cloud data and images;

[0126] The preprocessing module is used to preprocess the collected data according to the weather condition to obtain preprocessed data;

[0127] The information acquisition module obtains the current position, target position, and working environment information of the tower crane according to the preprocessed data;

[0128] The control module determines whether there are obstacles during the working process according to the current position, target position, and working environment information. If there are no obstacles, it performs normal planning to obtain the optimal working path. If there are obstacles, it performs obstacle avoidance planning to obtain the optimal working path, and controls the operation of the tower crane according to the planned optimal working path.

[0129] Among them, the execution process of each module can be executed according to the process steps of a tower crane intelligent management and control method provided in the first embodiment, and will not be elaborated one by one in this embodiment. This system and method are based on the same inventive concept and have the same beneficial effects, which will not be elaborated here.

[0130] The specific embodiments described above further elaborate on the purpose, technical solutions, and beneficial effects of the present invention. It should be understood that the above are only specific embodiments of the present invention and are not used to limit the protection scope of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A smart control method for tower cranes, characterized in that, Including the following steps: Obtain the collected data and weather conditions from the data acquisition module on the tower crane, where the collected data includes point cloud data and images; Preprocess the collected data according to the weather conditions to obtain preprocessed data; Obtain the current position, target position, and working environment information of the tower crane based on the preprocessed data; Judge whether there are obstacles during the working process according to the current position, target position, and working environment information; If not, perform normal planning to obtain the optimal working path; If there are obstacles, use an improved ant algorithm for obstacle avoidance planning to obtain the optimal working path, and constraints such as the safety factor of surrounding points, the flexible body of the tower crane, and the swing of the steel wire rope are added to the improved ant algorithm for constraint; Control the operation of the tower crane according to the planned optimal working path.

2. The intelligent control method for tower cranes according to claim 1, characterized in that, The specific method for preprocessing the collected data according to the weather conditions to obtain preprocessed data includes: Use a point cloud denoising algorithm to segment weather noise points from the distorted point cloud data and remove the weather noise points from the point cloud data in a mask manner to obtain denoised point cloud data.

3. The intelligent control method of the tower crane according to claim 2, wherein The weather conditions include foggy days, and the images include foggy day images. The specific method for preprocessing the foggy day images according to the weather conditions to obtain preprocessed data also includes: Divide the foggy day image into multiple patch images; Find the minimum value of the dark channel of each pixel point in each patch image; The generator judges the darkest pixel information in the local area of the image according to the minimum value and converts the foggy image into a defogged image; The discriminator compares the difference between the defogged image and the real fog-free image and judges whether to output the defogged image; Output the defogged image to obtain the defogged image.

4. The intelligent control method for tower cranes according to claim 2, characterized in that The weather conditions include rainy days, and the images include rainy day images. The specific method for preprocessing the rainy day images according to the weather conditions to obtain preprocessed data includes: Use 4 cascaded residual fast Fourier transform convolutional modules as the backbone network to extract rain streak information from the rainy day image to obtain the rain streak area; Use the context interaction Transformer module to combine context information to perform image restoration on the rain streak area and output the restored rain-free image.

5. The intelligent control method of the tower crane according to claim 2, characterized in that The weather conditions include snowy days, and the images include snowy day images. The specific method for preprocessing the snowy day images according to the weather conditions to obtain preprocessed data includes: Use the dual-tree complex wavelet transform to decompose the snowy day image into a first high-frequency component and a first low-frequency component; Continue to decompose the first low-frequency component into a second high-frequency component and a second low-frequency component at the next level, and so on until it is decomposed to the bth level, where b > 0; Apply a predefined high-frequency reconstruction network to reconstruct the high-frequency components at different levels to obtain reconstructed high-frequency components; Apply a predefined low-frequency reconstruction network to reconstruct the low-frequency components at different levels to obtain reconstructed low-frequency components; Fuse the reconstructed low-frequency components and high-frequency components through the inverse dual-tree complex wavelet transform to obtain a fused snow-free image.

6. The intelligent control method of the tower crane according to claim 1, wherein, The specific method for using the improved ant algorithm for obstacle avoidance planning to obtain the optimal working path includes: Construct an environmental model using the grid method according to the working environment information, and set the initial point and the target point corresponding to the current position and the target position respectively; Use the particle swarm optimization algorithm to optimize the parameters of the ant colony algorithm, and find the best combination of each parameter according to the optimization results; Set the pheromone value, the current iteration number and the maximum iteration number at the initial moment of path planning, and add the initial point to the path point tabu list; Search for a path. The ant colony starts from the initial point and selects the next path node in the adjacent grid according to the improved state transition probability. The constraints and penalties of the surrounding point safety factor, the flexible body of the tower crane and the rope swing are added to the improved state transition probability; After an ant completes a search, update the pheromone of the path passed; Judge whether the ant reaches the target point; If the target point is not reached, continue to return to search for the path; If the target point is reached, the search is completed; When all ants complete an iteration process, update the global pheromone concentration except for the optimal path; When the iteration number is reached, the planning is completed, and the node order in the tabu list corresponding to the global optimal path is recorded to obtain the best working path; Obtain the motion paths of the joints of the tower crane hook through inverse kinematics according to the best working path; 7. The intelligent control method of the tower crane according to claim 6, characterized in that, The formula for the improved state transition probability is: Among them, P ij m (t) represents the probability that ant m moves from the current node to the next node at time t, and τ ij (t) represents the amount of pheromone residue on the path from node i to node j by the ant at time t, and ξ ij is the heuristic function, representing the expected degree for the ant to transfer from node i to node j. α and β are respectively the relative importance parameters of the pheromone and the heuristic function. allowed m represents the node selected by ant m for the next move at time t, and ω o is the number of grid cells occupied by obstacles near node j, and ω s is the number of all grid cells near node j. γ is the weight value of the safety factor of surrounding points, and the value of γ is in [0, 1]. τ ik (t) represents the amount of pheromone residue on the path from node i to node k by the ant at time t, and ξ ik is the heuristic function, representing the expected degree for the ant to transfer from node i to node k. is the constraint fitness function considering the deformation of the flexible body and the swing of the wire rope when moving from node i to node j. ε is the weight value of the constraint fitness function. is the constraint fitness function considering the deformation of the flexible body and the swing of the wire rope when moving from node i to node k. d ij is the distance between node i and node j, and d jq is the distance from node j to the target point. δ is the weight value of the distance from the next node to the target point, and the value of δ is in [0, 0.5]. χ ij represents the maximum deformation amount generated by the flexible body of the tower crane when moving from node i to node j, and χ max is the maximum deformation amount allowed for the flexible body of the tower crane. λ is the constraint penalty coefficient, used to adjust the influence degree of the constraint conditions. θ ij represents the maximum swing angle of the wire rope when moving from node i to node j, and θ max is the maximum swing angle allowed for the wire rope. φ1 and φ2 are respectively the weight coefficients of the flexible body deformation and the wire rope swing angle, satisfying φ1 + φ2 = 1.

8. The intelligent control method of the tower crane according to claim 7, characterized in that The formula for updating the pheromone of the path passed is: τ ij (t + 1) = (1 - ρ)τ ij (t) + Δτ ij (t); Among them, ρ is the evaporation coefficient of pheromone, 0 < ρ < 1, Δτ ij (t) is the sum of the pheromone concentrations released by all ants on the path from node i to node j in this iteration, Δτ ij m (t) is the pheromone concentration released by ant m on the path from node i to node j, and Q1 is a constant representing the local pheromone intensity.

9. The intelligent control method of the tower crane according to claim 1, characterized in that The method further includes: judging the operation condition of the tower crane. If the tower crane reaches the target position safely and controllably, the process ends. If it does not reach the target position, control the tower crane to continue running according to the planned best path; 10. A smart control system for tower cranes, which is used to implement the smart control method for tower cranes described in any one of claims 1-9, characterized in that, Including: A data acquisition module, a preprocessing module, an information acquisition module and a control module; The data acquisition module is used to obtain the acquired data and the weather condition from the data acquisition module on the tower crane. The acquired data includes point cloud data and images; The preprocessing module is used to preprocess the acquired data according to the weather condition to obtain the preprocessed data; The information acquisition module obtains the current position, the target position and the working environment information of the tower crane according to the preprocessed data; The control module judges whether there are obstacles in the working process according to the current position, the target position and the working environment information. If not, perform normal planning to obtain the best working path. If so, perform obstacle avoidance planning to obtain the best working path. The constraints of the surrounding point safety factor, the flexible body of the tower crane and the wire rope swing are added to the improved ant algorithm for constraint, and control the tower crane to run according to the planned best working path;

Images