Self-adaptive control method for cable dragging torque of electric carry-scraper

Abstract

The invention belongs to the technical field of carry-scraper control, and discloses a self-adaptive control method for cable dragging torque of an electric carry-scraper. Comprising the steps of collecting cable comprehensive data; preprocessing the cable comprehensive data to obtain a state code set; based on the state code set, utilizing an improved cell algorithm to predict and obtain an ideal dragging torque value of the cable; positioning a difference region based on the ideal drag torque value; constructing a directed graph model based on the difference region; based on the directed graph model, the control parameters of the electric carry-scraper are adaptively adjusted, so that the dragging torque of the cable gradually approaches to an ideal value, adaptive control is realized, the intelligent level, the accuracy and the stability of control are remarkably improved, and the working efficiency and the safety of electric carry-scraper equipment are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of scraper control, and more specifically, to an adaptive control method for the cable towing torque of an electric scraper. Background Art

[0002] In working environments such as open-pit mines and construction sites, electric scrapers are widely used in various excavation, loading and transportation operations; electric scrapers usually obtain the required power by towing cables, so the safe and reliable operation of the cables is crucial for the normal operation of the entire system.

[0003] The patent application with the publication number CN115744505A discloses a cable reel control system for an underground electric scraper, which relates to the technical field of cable retraction and extension of scrapers, and includes a state unit, a reel unit, a control unit, a brake unit, a cable laying unit and a cable winding unit; the state unit is used to judge the state of the scraper and send the state signal of the scraper to the control unit; the reel unit is used to judge the state of the reel and send the reel state signal to the control unit; the control unit includes a main controller, and the main controller is provided with control logic for processing the signals of the state unit and the reel unit and sending instructions to the brake unit, the cable laying unit and the cable winding unit; the brake unit is used to implement braking when the scraper fails; the cable laying unit is used to implement cable laying for the scraper; the cable winding unit is used to implement cable winding for the scraper; the main purpose is to improve the efficiency and economy of equipment use through the good collaborative work between the underground electric scraper and its cable reel.

[0004] However, the traditional control method still lacks real-time performance and accuracy; during the movement of the cable, state parameters such as its position, tension, and curvature will change at any time, and the existing methods cannot obtain these dynamic data in real time and accurately, resulting in a lag in control response, unable to adjust control parameters in time, and ultimately unable to accurately control the towing torque of the cable; this brings great potential safety hazards to the safe operation of the cable and also affects the working efficiency of the scraper; secondly, the existing control strategies lack adaptability and intelligence; the movement of the cable is affected by many complex factors, such as environmental temperature and humidity, and the changes of these factors will cause drastic changes in the movement state of the cable; but the existing control methods usually adopt fixed control rules or simple feedback control, and cannot autonomously adjust the control strategy according to the changes of working conditions and the environment, making it difficult to achieve optimal control and resulting in poor control effects.

[0005] In view of this, the present invention proposes an adaptive control method for the cable towing torque of an electric scraper to solve the above problems. Summary of the Invention

[0006] To overcome the above defects of the prior art and to achieve the above object, the present invention provides the following technical solution: An adaptive control method for the cable towing torque of an electric LHD, comprising: S1, collecting comprehensive cable data; and preprocessing the comprehensive cable data to obtain a set of status codes; S2, based on the set of status codes, using an improved cellular algorithm to predict the ideal towing torque value of the cable; S3, based on the ideal towing torque value, locating the difference region; constructing a directed graph model based on the difference region; S4, based on the directed graph model, adaptively adjusting the control parameters of the electric LHD.

[0007] Further, the comprehensive cable data includes cable physical parameter data, cable motion state data, and environmental data.

[0008] Further, the way of preprocessing the comprehensive cable data includes: Discretizing the sequence data in the comprehensive cable data into finite discrete state values; normalizing the cable physical parameter data, and also denoting the normalized cable physical parameter data as discrete state values; all discrete state values form a state set; Encoding all discrete state values to obtain corresponding status codes, and storing the status codes in a mapping table; all status codes form a set of status codes.

[0009] Further, the way of discretizing the sequence data in the comprehensive cable data into finite discrete state values includes: Performing normalization or normalization processing on the sequence data in the comprehensive cable data; obtaining the corresponding normalized data sequence, splitting data points from the corresponding normalized data sequence according to time points to obtain corresponding data points; all data points form a data set; predefined the number of state values; setting the discretization parameters, the parameters include the maximum neighborhood number m, the local minimization number, and the maximum number of iterations; Randomly selecting data points in the data set that are the same as the predefined number of state values as the initial state centers and assigning state labels; for each state center, finding the m nearest data points in the data set, where m is an integer greater than 1; these m data points are constructed as the neighborhood of the corresponding state center; Define the adaptive candidate number ; where is the preset initial candidate number, is the preset decay rate, is the current number of iterations; 、 and are hyperparameters; is the change in the overall error in the most recent k iterations; k is an integer greater than 1; is the ratio of the current iteration number to the maximum iteration number; is the sigmoid function; For each iteration, calculate the distance from each data point in the neighborhood to the corresponding state center point, and sort the data points in ascending order of distance; select the data points with the smallest distance as candidate center points; for each candidate center point, temporarily use the corresponding candidate center point as the new state center point; reassign the state labels of other data points in the neighborhood so that they belong to the state center point with the closest distance; and calculate the overall error; compare the overall errors corresponding to all candidate center points, and select the candidate center point with the smallest overall error as the new state center point; repeat until the overall error no longer decreases; According to the finally obtained state center points, discretize the sequence data into finite state values; each data point is assigned the state value corresponding to the state center point closest to it, and the state value of each data point is used as the discrete state value.

[0010] Further, the method of storing the state codes into the mapping table includes: Traverse the state set, count the frequency of occurrence of each discrete state value; sort the discrete state values in descending order of frequency; all nodes form a node set; construct an encoding tree based on the node set, regard all discrete state values as the leaf nodes of the encoding tree, and the frequency as the weight of the corresponding leaf node; all leaf nodes form a node set; starting from the root node of the encoding tree, divide the node set into two, obtaining a left subtree and a right subtree; encode the leaf nodes of the left subtree as 0 and the leaf nodes of the right subtree as 1; recursively perform the same splitting and encoding on the left subtree and the right subtree until all leaf nodes are encoded; traverse the encoding tree, obtain the encoding of each leaf node, and insert the (state code, encoding) as a key-value pair into the mapping table, where the mapping table is a hash table.

[0011] Further, the method of dividing the node set into two includes: Define an adaptive splitting threshold ; where is a preset maximum allowable threshold constant; is the total weight of the current node set; is the weight of the root node; is a control parameter; is the current level; is a level control parameter; and are adaptive control parameters; The amount by which the encoding length resulting from the current splitting method exceeds that of the optimal splitting Sort the leaf nodes in the current node set in descending order of weight to obtain the sorted node set; Initialize two empty sets, denoted as the left set and the right set, for storing the leaf nodes after splitting. Traverse the sorted node set from the beginning, and add the leaf nodes to the left set one by one until the sum of the weights of the left set exceeds half of the total weight of the current node set. Then add the remaining leaf nodes to the right set; Calculate the weight difference diff between all the leaf nodes in the left set and the right set; If diff is less than or equal to , then accept this splitting; If diff is greater than , then take out the last leaf node from the left set, add it to the right set, and recalculate the weight difference to determine if it is less than or equal to ; If it is still greater than , then take out the first leaf node from the right set and add it to the left set; Repeat until the weight difference between the two is less than or equal to , or there is only one leaf node left in the left set and the right set; Recursively perform splitting on the left set and the right set to construct the left subtree and the right subtree.

[0012] Furthermore, the prediction method for the ideal towing torque value includes: Map the motion state of the cable to a two-dimensional grid. Each grid cell in the two-dimensional grid is called a cell, and the state of each cell is determined by the state code in the state code set; Define the range of the neighbor cells of each cell as the cells adjacent to the cell itself in the up, down, left, and right directions and the cells in the diagonal directions; Define the update rule function for the cell state. The formula of the update rule function is: ; where is the state of cell at time , is the normalization constant, represents the new state that the cell may transfer to, is the state of cell at time , represents the set of states of the neighbor cells of cell at time , is the independent variable value when maximizing, represents the set of other factors affecting the state transition of cell ; is an energy function, representing the energy cost for the cell to transfer to ; Assign status codes to each cell in the two-dimensional grid from top to bottom or from bottom to top according to the order of time points; each cell checks the status of itself and its neighboring cells, and updates the status of all cells synchronously according to the update rule function to form a new status of the cells; continue for a number of time steps until a stable state or convergence is reached; count the number of cells in each status in the stable state and their distribution in the entire two-dimensional grid, denoted as the cell status distribution; establish a mapping relationship table between the status of the cells and the cable motion parameters; according to the obtained cell status distribution, look up the mapping relationship table to restore the distribution of the corresponding cable motion parameters; construct a physical model of the cable, and substitute the restored cable motion parameters into the physical model; calculate the ideal drag torque value required for the cable during movement.

[0013] Further, the method for locating the difference region includes: Discretize the movement space of the cable into a regular grid; according to the current cable motion parameters, determine the initial grid where the cable is located in the regular grid, and mark this grid as the current status; define an objective function, where the objective function is the difference between the actual drag torque value and the ideal drag torque value; traverse each grid in the regular grid, calculate the value of its corresponding objective function, and store the value of the objective function in the corresponding grid; preset a difference threshold, and use the grids with the objective function value lower than the difference threshold as the difference region grids, and all the difference region grids form the difference region.

[0014] Further, the method for constructing the directed graph model includes: Discretize the difference region, divide it into several discrete nodes, and each node represents a sub-region or status in the difference region; if the sub-regions corresponding to two nodes are adjacent, determine that there is a connection relationship between them, and connect directed edges between the corresponding nodes with the connection relationship, and the direction of the directed edge is two-way, that is, connect a directed edge in each of the two directions; define the weight of the directed edge as the adjustment amount required for the transfer between the corresponding nodes; combine all the nodes and the directed edges between them to form a complete directed graph model.

[0015] Further, the method for adaptively adjusting the control parameters of the electric LHD based on the directed graph model includes: In the constructed directed graph model, determine the corresponding initial state node according to the actual motion state of the current cable; associate a set of status parameters with each node of the directed graph model, indicating the motion state represented by the node; calculate the distance or similarity between the cable motion parameters corresponding to the ideal drag torque value and the status parameters associated with each node; select the node with the minimum distance or the highest similarity as the ideal state node. Starting from the initial state node and ending at the ideal state node, the Dijkstra algorithm or A* algorithm is used to search for the shortest path from the starting point to the ending point in the directed graph model. The adjustment amounts corresponding to each directed edge are extracted from the obtained shortest path, and these adjustment amounts are arranged in the order of the shortest path to form an adjustment sequence. Each adjustment amount in the adjustment sequence is further decomposed into the corresponding control parameter adjustment amounts; according to the decomposed control parameter adjustment amounts, corresponding control instructions are generated; the generated control instructions are sent to the corresponding actuators of the electric LHD; the actuators adjust the operating state of the electric LHD according to the control instructions.

[0016] The technical effects and advantages of an adaptive control method for the cable towing torque of an electric LHD according to the present invention: The present invention realizes the intelligence and self - adaptability of cable towing torque control. By integrating multi - source heterogeneous data and establishing an accurate cable motion model, the ideal towing torque value can be predicted in real time, providing a precise target for control. At the same time, the state transition relationship is described by a directed graph model, and the control strategy is autonomously optimized based on the graph search algorithm, realizing the adaptive adjustment of the control logic, greatly improving the intelligence level of control; significantly improving the control accuracy and stability; by pre - processing and integrating the original data, the quality and reliability of the data are improved; using modeling and prediction algorithms, the description accuracy of cable motion is improved; the directed graph model and graph search algorithm ensure the rationality and optimality of control adjustment, thereby improving the accuracy and stability of control; improving the intelligence level, accuracy and stability of control, and improving the working efficiency and safety of electric LHD equipment. Description of the Drawings

[0017] Figure 1 It is a schematic diagram of an adaptive control method for the cable towing torque of an electric LHD according to the present invention; Figure 2 It is a schematic diagram of an adaptive control system for the cable towing torque of an electric LHD according to the present invention. Detailed Embodiments

[0018] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0019] Embodiment 1 Please refer to Figure 1 As shown, an adaptive control method for the cable towing torque of an electric LHD in this embodiment includes: S1, collecting cable comprehensive data; and preprocessing the cable comprehensive data to obtain a status code set; S2. Based on the state code set, the ideal drag torque value of the cable is predicted using the improved cellular algorithm; S3, locating the difference area based on the ideal drag torque value; constructing a directed graph model based on the difference area; S4. Based on the directed graph model, the control parameters of the electric scraper are adaptively adjusted so that the cable drag torque gradually approaches the ideal value to achieve adaptive control.

[0020] The comprehensive cable data includes cable physical parameter data, cable movement status data and environmental data; the cable physical parameter data includes cable length, cable diameter, cable weight and cable material (conductor, insulation layer, sheath, etc.); the cable movement status data includes cable position, cable movement speed, cable curvature, cable tension, cable temperature, cable dragging speed and cable dragging angle; the environmental data includes ambient temperature and ambient humidity; the cable movement status data and environmental data are both continuous sequence data, and the cable physical parameter data are discrete data.

[0021] It should be noted that these data are acquired in real time and appropriately processed. At the same time, the acquisition of these data requires the coordinated work of multiple sensors.

[0022] The methods for preprocessing cable comprehensive data include: The collected original cable comprehensive data is cleaned up to remove obvious outliers, noise and missing values ​​to ensure the integrity and accuracy of the data and obtain preliminary data; the sequence data in the cable comprehensive data is discretized into finite discrete state values.

[0023] Specifically, the sequence data in the cable comprehensive data (such as cable motion state data and environmental data) is standardized or normalized so that data of different dimensions are converted into the same numerical range; the corresponding normalized data sequence is obtained, and the data points are divided from the corresponding normalized data sequence according to the time point, that is, all the data at a time point (cable motion state data and environmental data) are vector-level spliced ​​to obtain the corresponding data points; for example, in the normalized data sequence corresponding to the cable temperature and the cable dragging speed, a time point is selected, the cable temperature and the cable dragging speed corresponding to this time point are extracted, and vector-level splicing is performed to obtain the corresponding data points.

[0024] All data points form a data set; the number of predefined state values is determined in advance; discretization parameters are set, including the maximum number of neighbors m, the number of local minimizations, and the maximum number of iterations; randomly select from the data set the same number of data points as the predefined number of state values as the initial state centroids and assign state labels to them; for each state centroid, find the m nearest data points in the data set, where m is an integer greater than 1; these m data points are constructed as the neighborhood corresponding to the state centroid; the distance is represented by the cosine similarity between the vectors corresponding to the data points.

[0025] Define the adaptive number of candidates ; where is the preset initial number of candidates, is the preset decay rate, which controls the basic decay speed of the number of candidates, is the current iteration number; 、 and are hyperparameters used to control the range and change rate of the number of candidates; is the change amount of the overall error in the last k iterations; k is an integer greater than 1; is the ratio of the current iteration number to the maximum iteration number, ranging from [0, 1]; is the sigmoid function, which maps the input value to the range (0, 1).

[0026] For each iteration, calculate the distance from each data point in the neighborhood to the corresponding state centroid, and sort the data points in ascending order of distance; select the data points with the smallest distance as candidate centroids; for each candidate centroid, temporarily use the corresponding candidate centroid as the new state centroid; reassign the state labels of other data points in the neighborhood so that they belong to the state centroid with the closest distance; and calculate the overall error (the sum of the Euclidean distances from all data points in the neighborhood to their respective state centroids); compare the overall errors corresponding to all candidate centroids and select the candidate centroid with the smallest overall error as the new state centroid; repeat until the overall error no longer decreases.

[0027] According to the finally obtained state centroids, discretize the sequence data into a finite number of state values; each data point is assigned the state value corresponding to the state centroid closest to it; specifically, assign a unique state value to each state centroid, and the state value can be an integer, a string, or other identifiers; for the sequence data in the cable comprehensive data (such as cable movement state data and environmental data), each time point corresponds to a data point; use the state value of each data point as the discrete state value at that time point.

[0028] Standardize the cable physical parameter data, and record the standardized cable physical parameter data as discrete state values as well; all discrete state values form a state set; encode all discrete state values, assign a unique state code, and store the state code in a mapping table for efficient retrieval; the encoding is implemented based on hash encoding; all state codes form a state code set.

[0029] Specifically, traverse the state set, count the frequency of each discrete state value; sort the discrete state values in descending order of frequency; all nodes form a node set; construct an encoding tree based on the node set, regard all discrete state values as the leaf nodes of the encoding tree, and the frequency as the weight of the corresponding leaf node; all leaf nodes form a node set; starting from the root node of the encoding tree, divide the node set into two parts to obtain a left subtree and a right subtree; encode the leaf nodes of the left subtree as 0 and the leaf nodes of the right subtree as 1; recursively perform the same division and encoding on the left subtree and the right subtree until all leaf nodes are encoded.

[0030] The ways to divide the node set into two parts include: Define an adaptive segmentation threshold ; where is a preset maximum allowable threshold constant, adjusted according to the data distribution; is the total weight of the current node set; is the weight of the root node; is a control parameter, usually taking values in [0.5, 1], used to adjust the influence degree of the weight ratio on the threshold; is the current level; it should be noted that the level of the root node is 0; for any non-root node, its level is equal to the level of its parent node plus 1; the level where the leaf node is located is the depth or height of the encoding tree; the deeper the level, the longer the corresponding encoding length; because more 0 / 1 encoding bits are required to reach this node from the root node; is a level control parameter, usually taking values in [0.5, 1], adjusting the influence degree of the level on the threshold; and are adaptive control parameters, used to dynamically adjust the threshold according to the encoding length increment; is the amount by which the encoding length increases when the current segmentation method is compared with the optimal segmentation; specifically, the optimal segmentation refers to dividing the current node set into two parts so that the sum of the weights of the two parts is the closest, thereby generating the shortest encoding length.

[0031] Sort the leaf nodes in the current node set in descending order of weight to obtain the sorted node set; Initialize two empty sets, denoted as the left set and the right set, to store the leaf nodes after splitting. Traverse the sorted node set from the beginning, and add the leaf nodes to the left set one by one until the sum of the weights of the left set exceeds half of the total weight of the current node set. Then add the remaining leaf nodes to the right set; Calculate the weight difference diff between all the leaf nodes in the left set and the right set; If diff is less than or equal to , then accept this split; If diff is greater than , then take out the last leaf node from the left set, add it to the right set, and recalculate the weight difference to determine whether it is less than or equal to ; If it is still greater than , then take out the first leaf node from the right set and add it to the left set; Repeat until the weight difference between the two is less than or equal to , or there is only one leaf node left in the left set and the right set; Recursively perform splitting on the left set and the right set to construct the left subtree and the right subtree.

[0032] Relax the strict requirements for node set splitting to a certain extent, allowing the weight difference between the two parts to fluctuate within a controllable range; This flexibility can reduce the unnecessary number of splits, thereby reducing the depth of the coding tree, shortening the coding length, and improving the coding efficiency.

[0033] Traverse the coding tree to obtain the codes of each leaf node (discrete state values), and insert (state code, code) as key-value pairs into the mapping table. An efficient data structure such as a hash table can be used; To find the code corresponding to a certain state value, simply search in the mapping table using the state value as the key; The search time complexity is low and it is very efficient; Starting from the root node, according to the 0 and 1 bits in the code, traverse down along the path of the tree. When reaching the leaf node, the state code corresponding to the leaf node is the decoding result.

[0034] The prediction methods for the ideal drag torque value include: Map the motion state of the cable to a two-dimensional grid. Each grid cell in the two-dimensional grid is called a cell, and the state of each cell is determined by the state code in the state code set; Define the range of the neighbor cells of each cell as the cells adjacent to the cell itself in the up, down, left, and right directions and the cells in the diagonal directions.

[0035] Define the update rule function for the cell state. The formula of the update rule function is: ; Among them, is the state of cell at time , is the normalization constant, Represents the new state to which the cell may transfer, for the cell at time moment, represents the cell of the neighboring cells at time moment state set, is the independent variable value when maximizing. Through the argmax operation, the optimal one can be selected from all possible new states as the state of the cell at the next moment, represents the set of other factors that affect the state transition of the cell ; is an energy function, representing the energy cost for the cell to transfer to ; The energy function is a weighted sum of internal variables; The update rule function describes how the cell makes state transitions based on its own and neighboring cells' states.

[0036] Specifically, ; Among them, represents the th influencing factor on the state of the cell at time t; The influencing factors include: the tension value of the cable position where the cell is located at time moment, the curvature value of the cable position where the cell is located at time moment, the temperature value of the cable position where the cell is located at time moment, the movement speed of the cable position where the cell is located at time moment, the environmental temperature at time moment, the environmental humidity at time moment.

[0037] According to the order of time points, the state codes are assigned to each cell from top to bottom or from bottom to top in a two-dimensional grid; Each cell checks the states of itself and neighboring cells, and synchronously updates the states of all cells according to the update rule function to form the new state of the cells; Continue for several time steps until a stable state or convergence is reached.

[0038] Statistically, count the number of states of each type of cell in the steady state and its distribution in the entire two-dimensional grid, which is denoted as the cell state distribution; establish a mapping relationship table between the cell state (state code) and the cable motion parameters. The mapping relationship table can be obtained through theoretical analysis or machine learning. The mapping relationship table gives the parameter values such as the cable position, speed, and tension corresponding to different cell states; according to the obtained cell state distribution, look up the mapping relationship table to restore the distribution of the corresponding cable motion parameters; construct a physical model of the cable, considering factors such as the mass, stiffness, and damping of the cable, and substitute the restored cable motion parameters into the physical model; calculate the ideal drag torque value required for the cable during movement; specifically, apply the generalized Lagrangian equation, the finite element analysis theory method, or use a machine learning algorithm to fit the physical model from the data.

[0039] The methods for locating the difference region include: Discretize the cable's motion space into a regular grid, where each grid in the regular grid corresponds to a motion state of the cable in that region; the fineness of the grid depends on the required calculation accuracy and the available computing resources.

[0040] According to the current cable motion parameters, determine the initial grid where the cable is located, and mark this grid as the current state; define an objective function to evaluate the difference between the state corresponding to each grid and the ideal state. The objective function is the difference between the actual drag torque value and the ideal drag torque value; traverse each grid in the regular grid, calculate the value of its corresponding objective function, and store the value of the objective function in the corresponding grid; preset a difference threshold, and regard the grids with the objective function value lower than the difference threshold as the difference region grids, and all the difference region grids form the difference region.

[0041] The methods for constructing the directed graph model include: Discretize the difference region, divide it into several discrete nodes, and each node represents a sub-region or state in the difference region; if the sub-regions corresponding to two nodes are adjacent, then determine that there is a connection relationship between them, and connect directed edges between the corresponding nodes with the connection relationship. The direction of the directed edge is two-way, that is, connect a directed edge in each of the two directions; define the weight of the directed edge as the adjustment amount required for the transfer between the corresponding nodes; consider multiple factors, such as changes in the cable drag torque, changes in motion parameters such as cable position / speed / tension, etc., and the adjustment amount can be obtained by weighted summation of these factors; combine all the nodes and the directed edges between them to form a complete directed graph model, and the directed graph model describes the topological structure and state transition relationship within the difference region.

[0042] In the constructed directed graph model, according to the actual motion state of the current cable, determine the corresponding initial state node; associate a set of state parameters, such as position, velocity, tension, etc., with each node of the directed graph model to represent the motion state represented by the node; calculate the distance or similarity between the cable motion parameters corresponding to the ideal drag torque value and the state parameters associated with each node; select the node with the minimum distance or the highest similarity as the ideal state node; the calculation of distance or similarity can use methods such as Euclidean distance, cosine similarity, etc.

[0043] Taking the initial state node as the starting point and the ideal state node as the ending point, use an algorithm (such as Dijkstra algorithm or A* algorithm, etc.) to search for the shortest path from the starting point to the ending point in the directed graph model. The shortest path represents the minimum adjustment cost required from the current state to the ideal state; extract the adjustment amount corresponding to each directed edge from the obtained shortest path, arrange these adjustment amounts in the order of the shortest path to form an adjustment sequence, and further decompose each adjustment amount in the adjustment sequence into the corresponding control parameter adjustment amount; for example, decompose the adjustment amount of the cable drag torque into the adjustment amounts of control parameters such as the hydraulic system pressure and motor speed of the electric LHD; decomposition can be carried out using methods such as inverse analysis method, optimization algorithm or machine learning model, etc.; generate corresponding control instructions according to the decomposed control parameter adjustment amounts; the control instructions can be hierarchical control instructions such as the pressure setting value of the hydraulic system and the motor speed setting value; send the generated control instructions to the corresponding actuators of the electric LHD, such as the hydraulic system controller, motor controller, etc.; the actuators adjust the operating state of the electric LHD according to the control instructions; during the operation of the electric LHD, monitor the motion state of the cable in real time; compare the monitored actual motion state with the ideal state to judge whether the control target is met; if the ideal state is not reached, re-determine the current state node and repeat the next round of iterative adjustment until the control target is met.

[0044] Through the above steps, use the directed graph model to adaptively adjust the control parameters of the electric LHD, so that the drag torque of the cable gradually approaches the ideal value, and realize adaptive control.

[0045] In this embodiment, the intelligent and self-adaptive control of the cable dragging torque is realized. By integrating multi-source heterogeneous data and establishing an accurate cable motion model, the ideal dragging torque value can be predicted in real time, providing a precise target for control. At the same time, the state transition relationship is described by a directed graph model, and the control strategy is autonomously optimized based on the graph search algorithm, realizing the adaptive adjustment of the control logic, greatly improving the intelligent level of control; significantly improving the accuracy and stability of control; improving the quality and reliability of data by preprocessing and fusing the original data; improving the description accuracy of the cable motion by using modeling and prediction algorithms; the directed graph model and the graph search algorithm ensure the rationality and optimality of the control adjustment, thus improving the precision and stability of control; improving the intelligent level, precision and stability of control, and improving the working efficiency and safety of the electric LHD equipment.

[0046] Embodiment 2 Please refer to Figure 2 As shown, for the parts not described in detail in this embodiment, refer to the description content of Embodiment 1. A self-adaptive control system for the cable dragging torque of an electric LHD is provided, including: A data acquisition and processing module, used to acquire comprehensive cable data; and preprocess the comprehensive cable data to obtain a set of status codes; A prediction module, based on the set of status codes, uses an improved cell algorithm to predict the ideal dragging torque value of the cable; A positioning module, based on the ideal dragging torque value, locates the difference area; constructs a directed graph model based on the difference area; A self-adaptive adjustment module, based on the directed graph model, adaptively adjusts the control parameters of the electric LHD; each module is connected by wired and / or wireless means to realize data transmission between modules.

[0047] Embodiment 3 This embodiment publicly provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, it realizes the operation mode of the above-provided self-adaptive control method for the cable dragging torque of an electric LHD.

[0048] Since the electronic device introduced in this embodiment is the electronic device adopted for implementing an adaptive control method for the cable towing torque of an electric LHD in the embodiments of the present application, based on the adaptive control method for the cable towing torque of an electric LHD introduced in the embodiments of the present application, those skilled in the art can understand the specific implementation manners and various variations of the electronic device in this embodiment. Therefore, the specific implementation of how this electronic device implements the method in the embodiments of the present application will not be described in detail here. As long as those skilled in the art implement the electronic device adopted for the adaptive control method for the cable towing torque of an electric LHD in the embodiments of the present application, it falls within the scope of protection of the present application.

[0049] The above formulas are all dimensionless and take their numerical calculations. The formulas are obtained by collecting a large amount of data for software simulation to obtain a formula closest to the actual situation. The preset parameters and threshold selection in the formulas are set by those skilled in the art according to the actual situation.

[0050] The above is only the preferred implementation manner of the present invention. The protection scope of the present invention is not limited to the above embodiments. All technical solutions falling within the idea of the present invention belong to the protection scope of the present invention. It should be noted that for ordinary technical users in the technical field, several improvements and refinements made without departing from the principle of the present invention should also be regarded as within the protection scope of the present invention.

Claims

1. A method for adaptively controlling cable drag torque of an electric scraper, characterized in that: include: S1. Collect comprehensive cable data; And pre-process the cable comprehensive data to obtain a state code set; S2. Based on the state code set, the ideal drag torque value of the cable is predicted using the improved cellular algorithm; S3, locating the difference area based on the ideal drag torque value; Build a directed graph model based on the difference area; S4. Based on the directed graph model, the control parameters of the electric scraper are adaptively adjusted.

2. The method for adaptively controlling the cable drag torque of an electric scraper according to claim 1, characterized in that: The cable comprehensive data includes cable physical parameter data, cable movement status data and environmental data.

3. The method for adaptively controlling the cable drag torque of an electric scraper according to claim 2, characterized in that: The method of preprocessing the cable comprehensive data includes: Discretize the sequence data in the cable comprehensive data into finite discrete state values; standardize the cable physical parameter data, and record the obtained standardized cable physical parameter data as discrete state values; all discrete state values ​​constitute a state set; All discrete state values ​​are encoded to obtain corresponding state codes, and the state codes are stored in a mapping table; all state codes constitute a state code set.

4. The method for adaptively controlling the cable drag torque of an electric scraper according to claim 3, characterized in that: The method of discretizing the sequence data in the cable comprehensive data into finite discrete state values ​​includes: Standardize or normalize the sequence data in the cable comprehensive data; obtain the corresponding normalized data sequence, divide the data points from the corresponding normalized data sequence according to the time point, and obtain the corresponding data points; all data points constitute a data set; predefine the number of state values; set the discretization parameters, including the maximum neighborhood number m, the local minimization number and the maximum number of iterations; Randomly select data points with the same number as the pre-defined state values ​​from the data set as the initial state center points and assign state labels; for each state center point, find the m data points closest to it in the data set, where m is an integer greater than 1; these m data points are constructed as the neighborhood of the corresponding state center point; Define the number of adaptive candidates ;in, is the preset initial candidate number, is the preset decay rate, is the current iteration number; , and is a hyperparameter; is the change in the overall error in the last k iterations; k is an integer greater than 1; is the ratio of the current number of iterations to the maximum number of iterations; is the sigmoid function; For each iteration, calculate the distance from each data point in the neighborhood to the corresponding state center point, and sort the data points from small to large distance; select the one with the smallest distance data points as candidate center points; for each candidate center point, temporarily set the corresponding candidate center point as the new state center point; reallocate the state labels of other data points in the neighborhood so that they belong to the state center point closest to them; and calculate the overall error; compare the overall errors corresponding to all candidate center points, and select the candidate center point with the smallest overall error as the new state center point; repeat until the overall error no longer decreases; According to the final state center point, the sequence data is discretized into finite state values; each data point is assigned to the state value corresponding to its nearest state center point, and the state value of each data point is used as a discrete state value.

5. The method for adaptively controlling the cable drag torque of an electric scraper according to claim 4, characterized in that: The method of storing the status code in the mapping table includes: Traverse the state set and count the frequency of occurrence of each discrete state value; sort the discrete state values ​​from high to low according to the frequency of occurrence; all nodes constitute a node set; build a coding tree based on the node set, regard all discrete state values ​​as leaf nodes of the coding tree, and use the frequency as the weight of the corresponding leaf node; all leaf nodes constitute a node set; starting from the root node of the coding tree, divide the node set into two to obtain a left subtree and a right subtree; encode the leaf nodes of the left subtree as 0, and encode the leaf nodes of the right subtree as 1; recursively perform the same division and encoding on the left subtree and the right subtree until all leaf nodes are encoded; traverse the coding tree, obtain the code of each leaf node, and insert (state code, code) as a key-value pair into a mapping table, wherein the mapping table is a hash table.

6. The method for adaptively controlling the cable drag torque of an electric scraper according to claim 5, characterized in that: The method of dividing the node set into two includes: Defining adaptive segmentation threshold ;in, is the preset maximum allowed threshold constant; is the total weight of the current node set; is the weight of the root node; To control the parameters; is the current level; is the hierarchical control parameter; and is the adaptive control parameter; The amount by which the encoding length increases due to the current segmentation method compared to the optimal segmentation method; Sort the leaf nodes in the current node set by weight from large to small to obtain the sorted node set; initialize two empty sets, denoted as the left set and the right set, to store the split leaf nodes, traverse the sorted node set from the beginning, and add the leaf nodes to the left set one by one until the sum of the weights of the left set exceeds half of the total weight of the current node set, and then add the remaining leaf nodes to the right set; calculate the weight difference diff of all leaf nodes in the left set and the right set; if diff is less than or equal to , then accept the split; if diff is greater than , then take the last leaf node from the left set, add it to the right set, and recalculate the weight difference to determine whether it is less than or equal to If it is still greater than , then take the first leaf node from the right set and add it to the left set; repeat until the weight difference between the two is less than or equal to , or there is only one leaf node left in the left set and the right set; recursively perform partitioning on the left set and the right set to construct the left subtree and the right subtree.

7. The method for adaptively controlling the cable drag torque of an electric scraper according to claim 6, characterized in that: The prediction method of the ideal drag torque value includes: The motion state of the cable is mapped to a two-dimensional grid. Each grid unit in the two-dimensional grid is called a cell. The state of each cell is determined by the state code in the state code set. The range of neighbor cells of each cell is defined as the cells adjacent to the cell above, below, left, right, and diagonal directions. Define the update rule function of the cell state. The formula of the update rule function is: ;in, For cells In time The state of the moment, is the normalization constant, represents the new states that the cell may transition to, For cells In time The state of the moment, Represents cells Neighbor cells In time The state set at a moment, To maximize the value of the independent variable, Indicates that the cells are affected A collection of other factors that affect state transition; is an energy function that indicates the cell transfer The energy cost; According to the sequence of time points, the two-dimensional grid assigns state codes to each cell from top to bottom or from bottom to top; each cell checks the state of itself and neighboring cells, and synchronously updates the state of all cells according to the update rule function to form a new state of the cell; this continues for several time steps until a stable state or convergence is reached; the number of states of each cell in the stable state and its distribution in the entire two-dimensional grid are counted, recorded as the cell state distribution; a mapping relationship table between the cell state and the cable motion parameters is established; based on the obtained cell state distribution, the mapping relationship table is searched to restore the corresponding distribution of the cable motion parameters; a physical model of the cable is constructed, and the restored cable motion parameters are substituted into the physical model; the ideal drag torque value required for the cable during movement is calculated.

8. The method for adaptively controlling the cable drag torque of an electric scraper according to claim 7, characterized in that: The method of locating the difference area includes: The motion space of the cable is discretized into a regular grid; according to the current cable motion parameters, the initial grid of the cable in the regular grid is determined, and the grid is marked as the current state; the objective function is defined, and the objective function is the difference between the actual drag torque value and the ideal drag torque value; each grid in the regular grid is traversed, the corresponding objective function value is calculated, and the objective function value is stored in the corresponding grid; a difference threshold is preset, and the grids with objective function values ​​lower than the difference threshold are regarded as difference area grids, and all difference area grids constitute the difference area.

9. The method for adaptively controlling the cable drag torque of an electric scraper according to claim 8, characterized in that: The construction method of the directed graph model includes: The difference region is discretized and divided into several discrete nodes, each node represents a sub-region or state in the difference region; if the sub-regions corresponding to two nodes are adjacent, it is determined that there is a connection relationship between them, and a directed edge is connected between the corresponding nodes with a connection relationship. The direction of the directed edge is bidirectional, that is, one directed edge is connected in each direction; the weight of the directed edge is defined as the adjustment amount required for the transfer between the corresponding nodes; all nodes and the directed edges between them are combined to form a complete directed graph model.

10. The method for adaptively controlling cable drag torque of an electric scraper according to claim 9, characterized in that: The method of adaptively adjusting the control parameters of the electric scraper based on the directed graph model includes: In the constructed directed graph model, the corresponding initial state node is determined according to the actual motion state of the current cable; a set of state parameters is associated with each node of the directed graph model to indicate the motion state represented by the node; the distance or similarity between the cable motion parameters corresponding to the ideal drag torque value and the state parameters associated with each node is calculated; the node with the smallest distance or the highest similarity is selected as the ideal state node; Taking the initial state node as the starting point and the ideal state node as the end point, the Dijkstra algorithm or the A* algorithm is used to search the shortest path from the starting point to the end point in the directed graph model, and the adjustment amount corresponding to each directed edge is extracted from the shortest path obtained by the search, and these adjustment amounts are arranged in the order of the shortest path to form an adjustment sequence, and each adjustment amount in the adjustment sequence is further decomposed into a corresponding control parameter adjustment amount; according to each control parameter adjustment amount obtained by the decomposition, a corresponding control instruction is generated; the generated control instruction is sent to the corresponding actuator of the electric scraper; the actuator adjusts the operating state of the electric scraper according to the control instruction.

Images