Method for constructing a characterization model of factors influencing the motion stability of a heavy robot in a well

Abstract

The present application belongs to the technical field of downhole robots, and particularly relates to a method for constructing a characterization model of factors affecting the motion stability of a downhole heavy-load robot, which is used to solve the problem that the research on the stability of robots in the prior art does not consider the influence of multiple source factors and is difficult to be applied to downhole heavy-load robots. The method comprises the following steps: S0, obtaining multiple source coupled internal and external factors; S1, classifying the multiple source coupled internal and external factors; S2, quantifying and obtaining the influence weight according to the fuzzy analytic hierarchy process; S3, constructing a primary model for characterizing the factors affecting the motion stability of the downhole heavy-load robot; S4, distinguishing the parameter sensitivity; S5, fusing and identifying the key parameters; and S6, constructing a characterization model of factors affecting the motion stability of the downhole heavy-load robot. Compared with the prior art, the present application constructs a characterization model of factors affecting the motion stability close to the actual working conditions of the downhole heavy-load robot, and provides a reliable basis and support for the research on the motion stability of the downhole heavy-load robot.

Description

Technical Field

[0001] This invention relates to the field of downhole robot technology, and in particular to a method for constructing a characterization model of factors affecting the motion stability of a heavy-duty downhole robot. Background Technology

[0002] To meet the demands of economic development for coal production capacity, intelligent transformation has become an inevitable trend for the high-quality development of the coal industry, and the widespread application of underground robots in coal mines has become a key breakthrough in this transformation. As crucial equipment in complex and high-risk underground working conditions, heavy-duty underground robots are used to perform heavy-duty operations such as handling and hoisting in unstructured underground roadways. Their motion stability has a direct and crucial impact on the safety and efficiency of underground operations. Robot motion stability refers to the robot's ability to resist external interference and maintain a predetermined motion state during movement, and it is one of the core indicators for evaluating robot performance. The Zero Moment Point (ZMP) theory, due to its clear physical meaning and simple calculation, is widely used as a criterion for judging robot motion stability.

[0003] The stability of heavy-duty underground robots is influenced not only by their own structure but also by the complex external working space and environment. In actual operation, the working space and environment of heavy-duty underground robots are extremely complex, primarily manifested in the following ways: First, the underground roadway floor is often a mixture of soft, slippery, muddy, and uneven materials such as coal slurry, gravel, and groundwater, creating a complex road surface that can cause adhesion during movement and work, easily leading to sideslip and rollover. Second, due to the demands of coal mining and construction conditions, most roadways have numerous intersections, turns, and slope variations between 5 and 30 degrees, increasing the difficulty of stable operation for heavy-duty underground robots. Third, the underground environment generally maintains high temperatures of 35-40℃, humidity consistently above 80%, low illumination, and dust concentrations as high as 10-50 mg / m³. 3 This increases the interference and difficulty of the interaction between the perception system and environmental information of the underground heavy-duty robot, thus affecting the timeliness of the robot's feedback control and consequently its stability. Furthermore, the motion stability of the underground heavy-duty robot is also affected by the coordination and control between its internal mechanical, hydraulic, and electrical systems. Current research on robot stability often focuses on analyzing a single factor. However, during distributed heavy-duty operations, the stability of underground heavy-duty robots is easily affected by these multiple factors, leading to skidding and overturning. Therefore, current research on robot stability is difficult to apply to underground heavy-duty robots. Summary of the Invention

[0004] In view of this, it is necessary to provide a method for constructing a characterization model of the motion stability influencing factors of downhole heavy-duty robots. This method can integrate the complex and constrained working conditions of downhole heavy-duty robots during downhole operations, as well as the comprehensive impact of the coupling effects between their mechanical, hydraulic, and electrical systems on the motion stability of the downhole heavy-duty robot system. This will construct a characterization model of the motion stability influencing factors that closely reflects the actual working conditions of downhole heavy-duty robots, accurately reflect the influencing factors of motion stability in downhole heavy-duty robots, and provide a basis for improving the motion stability and reliability of downhole heavy-duty robots.

[0005] A method for constructing a characterization model of factors affecting the motion stability of a downhole heavy-duty robot includes the following steps:

[0006] S0. Obtain the internal and external factors of multi-source coupling, specifically:

[0007] Analyze and obtain the multi-source coupled internal and external factors affecting the motion stability of downhole heavy-duty robots;

[0008] S1. Categorize the internal and external factors of multi-source coupling, specifically as follows:

[0009] S10. Classify the above-mentioned multi-source coupling internal and external factors to obtain four types of coupling factors, including environment-mechanical coupling B1, mechanical body coupling B2, perception-environment coupling B3 and mechanical-electrical-hydraulic coupling B4.

[0010] S11. The above four types of coupling factors are further refined to obtain the corresponding coupling sub-factors;

[0011] S2. The influence weights are quantified and obtained using the fuzzy hierarchical analysis method, specifically as follows:

[0012] Based on the above four types of coupling factors and their corresponding sub-factors, the influence weights of the four types of coupling factors on the motion stability of the downhole heavy-duty robot, and the influence weights of the corresponding sub-factors on the motion stability of the downhole heavy-duty robot are quantified and obtained by fuzzy hierarchical analysis.

[0013] S3. Construct a preliminary model to characterize the factors affecting the motion stability of downhole heavy-duty robots;

[0014] Based on the influence weights of four types of coupling factors on the motion stability of downhole heavy-duty robots, a primary model representing the influencing factors of motion stability of downhole heavy-duty robots is constructed.

[0015] S4. Distinguish parameter sensitivity, specifically:

[0016] S40. The influence weights of the four types of coupling factors on the motion stability of the downhole heavy-duty robot are ranked, and the top two coupling factors in terms of influence weight are obtained, namely environmental-mechanical coupling B1 and mechanical-electrical-hydraulic coupling B4.

[0017] S41. Specifically identify the influence of the coupling sub-factors corresponding to the environment-mechanical coupling B1 on the motion stability of the downhole heavy-duty robot, and obtain the key parameters that are highly sensitive to the motion stability of the downhole heavy-duty robot among the coupling sub-factors corresponding to the environment-mechanical coupling B1.

[0018] S42. Distinguish the influence of the coupling sub-factors corresponding to the mechanical-electrical-hydraulic coupling B4 on the motion stability of the downhole heavy-duty robot, and obtain the key parameters of the coupling sub-factors corresponding to the mechanical-electrical-hydraulic coupling B4 that are highly sensitive to the motion stability of the downhole heavy-duty robot.

[0019] S5. Integrate and identify key parameters, specifically:

[0020] A hybrid identification algorithm combining particle swarm optimization (PSO) and least squares support vector machine (LSSVM) is adopted. Based on a pre-built simulation model, the key parameters that are highly sensitive to the motion stability of the downhole heavy-duty robot are identified by fusing the coupling sub-factors corresponding to the environment-mechanical coupling B1 and the coupling sub-factors corresponding to the mechanical-electrical-hydraulic coupling B4. The optimal values ​​of the key parameters affecting the motion stability of the downhole heavy-duty robot are obtained.

[0021] S6. Construct a characterization model for the factors affecting the motion stability of downhole heavy-duty robots, specifically:

[0022] Based on the optimal values ​​of key parameters affecting the motion stability of downhole heavy-duty robots and the preliminary model representing the factors influencing the motion stability of downhole heavy-duty robots, a model representing the factors influencing the motion stability of downhole heavy-duty robots is constructed.

[0023] Preferably, in step S11, the coupling sub-factors corresponding to the environmental-mechanical coupling B1 include the indentation coupling between the track and the bottom plate C11, the shear stress coupling between the track and the bottom plate C12, and the coupling between the tunnel inclination angle and the overturning moment of the fuselage C13.

[0024] The coupling sub-factors corresponding to mechanical body coupling B2 include the serial structural coupling between the walking mechanism and the fuselage C21, the geometric coupling between the fuselage lifting and translation C22, the attitude coupling between the actuator arm and the fuselage C23, and the structural stiffness coupling of the actuator arm joint C24.

[0025] The coupling sub-factors corresponding to the perception-environment coupling B3 include the imaging coupling of dust and visual sensors C31, the coupling of low illumination and laser sensing signals C32, the coupling of narrow space and sensing signals C33, and the noise coupling of electromagnetic environment and sensing signals C34.

[0026] The coupling factors corresponding to the electromechanical-hydraulic coupling B4 include the response coupling between electrical control signals and hydraulic actuation C41, the efficiency coupling between hydraulic energy and mechanical energy conversion C42, and the closed-loop coupling between mechanical motion state and electrical feedback C43.

[0027] Preferably, in step S2, the quantification and acquisition of influence weights based on fuzzy hierarchical analysis specifically includes the following steps:

[0028] S20. Set analysis objectives:

[0029] The objective of the analysis is to quantify the weights of the influence of multiple coupled internal and external factors on the motion stability of downhole heavy-duty robots.

[0030] S21. Construct a three-layer structure consisting of target layer A, criterion layer B, and sub-criterion layer C, specifically including:

[0031] Based on the analysis objectives, target layer A is constructed as follows: weighted analysis of the influence of multi-source coupling factors on the motion stability of downhole robots;

[0032] Based on the four types of coupling factors, the criterion layer B is constructed as follows: environment-mechanical coupling B1, mechanical body coupling B2, perception-environment coupling B3, and mechanical-electrical-hydraulic coupling B4.

[0033] Based on the corresponding coupling sub-factors of the four types of coupling factors, the sub-criteria layer C corresponding to criteria layer B is constructed as follows:

[0034] B1: Indentation coupling between track and floor plate C11, Shear stress coupling between track and floor plate C12, Coupling between tunnel inclination angle and overturning moment of fuselage C13;

[0035] B2: Series structural coupling between the walking mechanism and the fuselage C21, geometric coupling of fuselage lifting and translation C22, attitude coupling between the actuator arm and the fuselage C23, structural stiffness coupling of the actuator arm joint C24.

[0036] B3: Dust coupling with visual sensor imaging C31, Low illumination coupling with laser sensing signal C32, Confined space coupling with sensing signal C33, Electromagnetic environment coupling with sensing signal noise C34;

[0037] B4: Response coupling between electrical control signals and hydraulic actuation; C41: Efficiency coupling between hydraulic energy and mechanical energy conversion; C42: Closed-loop coupling between mechanical motion state and electrical feedback; C43:

[0038] S22. Establish the fuzzy judgment matrix, specifically as follows:

[0039] Based on the target layer A, criterion layer B, and sub-criterion layer C, domain experts are invited to construct the fuzzy judgment matrix of criterion layer B on target layer A and the fuzzy judgment matrix of sub-criterion layer C on criterion layer B using triangular fuzzy numbers.

[0040] S23. Obtain the influence weight, specifically:

[0041] Based on the fuzzy judgment matrix, the influence weight of criterion layer B on target layer A and the influence weight of sub-criterion layer C on criterion layer B are calculated using the fuzzy hierarchical analysis method.

[0042] Based on the influence weight of criterion layer B on target layer A and the influence weight of sub-criterion layer C on criterion layer B, the influence weight of sub-criterion layer C on target layer A is obtained;

[0043] Based on the correspondence between criterion layer B and the four types of coupling factors, and the correspondence between sub-criterion layer C and the coupling sub-factors, the influence weights of the four types of coupling factors on the motion stability of the downhole heavy-duty robot, and the influence weights of the corresponding coupling sub-factors on the motion stability of the downhole heavy-duty robot are obtained.

[0044] Preferably, in step S23, the influence weight of the sub-criteria layer C on the target layer A is obtained by multiplying the influence weight of the criteria layer B on the target layer A and the influence weight of the sub-criteria layer C on the criteria layer B.

[0045] Preferably, in step S41, the key parameter among the coupling sub-factors corresponding to the environment-mechanical coupling B1 that is highly sensitive to the motion stability of the downhole heavy-duty robot is:

[0046] The key parameter in the indentation coupling C11 between the track and the bottom plate is: soil cohesive deformation modulus k. c Soil frictional deformation modulus k φ The key parameters in the coupling of soil deformation index n and shear stress C12 between the track and the bottom plate are: shear strength parameter τ(s) and shear displacement coefficient K. s ;

[0047] The key parameter in the coupling between the tunnel inclination angle and the fuselage overturning moment C13 is: the inclination angle coupling coefficient K. θ ;

[0048] In step S42, the key parameter among the coupling sub-factors corresponding to the electromechanical-hydraulic coupling B4 that is highly sensitive to the motion stability of the downhole heavy-duty robot is:

[0049] The key parameter in the response coupling between electrical control signals and hydraulic actuation C41 is the electro-hydraulic response coefficient K. eh Response delay time t d ;

[0050] The key parameter in the efficiency coupling of hydraulic and mechanical energy conversion C42 is: hydraulic-mechanical conversion efficiency η. hm Mechanical transmission efficiency η m .

[0051] Preferably, in step S5, the hybrid identification algorithm combining particle swarm optimization (PSO) and least squares support vector machine (LSSVM) is used to fuse and identify key parameters highly sensitive to the motion stability of the downhole heavy-duty robot among the coupling sub-factors corresponding to the environment-mechanical coupling and the coupling sub-factors corresponding to the mechanical-electrical-hydraulic coupling, to obtain the optimal values ​​of the key parameters affecting the motion stability of the downhole heavy-duty robot, including the following steps:

[0052] S50. Obtain the dataset, specifically:

[0053] Based on theoretical analysis and simulation, the reasonable range and initial values ​​of the key parameters to be fused and identified are determined.

[0054] Sampling is performed within a reasonable range of values ​​for the key parameters to generate multiple sets of input combinations for the key parameters;

[0055] For each set of key parameter input combinations, the simulation is run in a pre-built simulation model to obtain the corresponding system output;

[0056] Data set D is formed based on the input combination of key parameters and the corresponding system output;

[0057] Divide dataset D into a training set and a validation set;

[0058] S51. Train the LSSVM model, specifically as follows:

[0059] A nonlinear mapping model from key parameter input to system output is constructed using LSSVM, as shown in Equation (1), where x is the key parameter input and y(x) is the system output. Here, b represents the support vector coefficients, b is the bias term, and N is the number of data sampling points.

[0060] Formula (1)

[0061] The PSO algorithm and training set are used to train the LSSVM model, and the optimal hyperparameters (γ*, σ*) of LSSVM and the trained LSSVM model are obtained.

[0062] S52. Perform global optimization iterations on key parameters, specifically as follows:

[0063] Construct the objective function J according to formula (2), where, Let be the normalized total weight of the sub-criteria layer C to which the i-th key parameter belongs with respect to the target layer A. and These are the measured value and the model predicted value of the i-th key parameter at the j-th time, respectively; n is the total number of parameters to be identified; and N is the number of data sampling points.

[0064] Formula (2)

[0065] The trained LSSVM model is invoked, and multiple sets of key parameters from the training set are used as input to obtain the model's predicted values.

[0066] Calculate the objective function J according to formula (2);

[0067] The objective function is minimized using PSO, so that each particle position represents a combination of key parameter inputs;

[0068] Update the optimal position of the individual and the optimal position of the group based on the particle fitness value, i.e., the J value, iteratively adjust the particle velocity and position, gradually approach the global optimal key parameter combination, and find and obtain the key parameter combination that minimizes the J value.

[0069] S53, Local fine-grained optimization, specifically:

[0070] Based on the influence weight of sub-criteria layer C on target layer A obtained in step S23, five local iterations are added to optimize the top five key parameters with the highest total weight.

[0071] When J < Alternatively, when the maximum number of iterations is reached, the iteration stops, and the optimal combination of key parameters is output.

[0072] Preferably, in step S51, the step of training the LSSVM model using the PSO algorithm and training set to obtain the optimal hyperparameters (γ*, σ*) of the LSSVM and the trained LSSVM model specifically involves:

[0073] Set PSO parameters: particle swarm size, learning factor, inertia weight w, and maximum number of iterations;

[0074] The LSSVM model is set up as follows: the kernel function is the radial basis function. The range of kernel parameters and regularization parameters γ and σ is initially determined through cross-validation. Each particle position in PSO represents a set of (γ, σ).

[0075] For each particle, the LSSVM model is trained using its corresponding (γ, σ) training set data to obtain the optimal hyperparameters (γ*, σ*) of the LSSVM and the trained LSSVM model.

[0076] Preferably, in step S50, the step of running the simulation in a pre-built simulation model for each set of key parameter combinations to obtain the corresponding system output specifically involves:

[0077] Construct a rigid-flexible coupled dynamic simulation model of a downhole heavy-duty robot in RecurDyn;

[0078] A mechanical-electro-hydraulic coupling simulation model of a downhole heavy-duty robot was built in Matlab / Simulink;

[0079] In the rigid-flexible coupling dynamics simulation model of the downhole heavy-duty robot, k is loaded into each set of key parameter combinations. c k φ , n, τ(s), K s and K θ Perform typical operating conditions and obtain the corresponding first system output, including ZMP position coordinates, fuselage roll angle and fuselage pitch angle;

[0080] In the electro-hydraulic coupling simulation model of the downhole heavy-duty robot, K is loaded into each set of key parameters. eh t d η hm and η m Perform typical working conditions and obtain the corresponding second system output, including the steady-state error of the joint angle of the actuator arm, the relative speed error, the anti-slip coefficient and the anti-rollover coefficient;

[0081] The key parameters of each set of key parameters are combined with the corresponding first system output and second system output to obtain the system output corresponding to each set of key parameters, including ZMP position coordinates, fuselage roll angle, fuselage pitch angle, steady-state error of actuator joint angle, relative speed error, anti-skid coefficient and anti-rollover coefficient.

[0082] Preferably, in step S3, the expression for the primary model characterizing the factors affecting the motion stability of the downhole heavy-duty robot is as follows: As shown in formula (3), in the formula, The weighting of the impact of environmental-mechanical coupling B1 on the motion stability of the downhole heavy-duty robot. The weight of the influence of mechanical body coupling B2 on the motion stability of the downhole heavy-duty robot is given. The weighting of the impact of perception-environment coupling B3 on the motion stability of the downhole heavy-duty robot. Weights for the impact of electromechanical-hydraulic coupling B4 on the motion stability of downhole heavy-duty robots:

[0083] Formula (3)

[0084] The aforementioned method for constructing a characterization model of factors affecting the motion stability of downhole heavy-duty robots involves the following steps: First, by analyzing the working space and environment of the downhole heavy-duty robot, as well as its electromechanical-hydraulic characteristics, four core coupling factors affecting its motion stability are identified. Second, using fuzzy hierarchical analysis, the influence weights of these four coupling factors on the motion stability of the downhole heavy-duty robot are scientifically quantified, and a primary characterization model of these factors is constructed based on these weights. Then, the two coupling factors with the largest influence weights on the motion stability of the downhole heavy-duty robot, namely environment-mechanical coupling B1 and electromechanical-hydraulic coupling B4, are selected. Parameter sensitivity discrimination is performed to obtain the key parameters that are highly sensitive to the motion stability of the downhole heavy-duty robot among the corresponding coupling sub-factors of B1 and B4. Finally, a particle swarm optimization algorithm (PSO) and least squares support vector optimization are used. A hybrid identification algorithm combining machine learning and LSSVM is used. Based on a pre-built simulation model, the key parameters highly sensitive to the motion stability of downhole heavy-duty robots in the corresponding coupling sub-factors B1 and B4 are fused and identified to obtain the optimal values ​​of the key parameters affecting the motion stability of downhole heavy-duty robots. Finally, based on the optimal values ​​of the key parameters affecting the motion stability of downhole heavy-duty robots and the primary model representing the factors affecting the motion stability of downhole heavy-duty robots, a model representing the factors affecting the motion stability of downhole heavy-duty robots is constructed. Compared with the prior art, this invention integrates the complex and constrained working conditions of downhole heavy-duty robots during downhole operations, as well as the comprehensive impact of the coupling effects between its own mechanical system, hydraulic system, and electrical system on the motion stability of downhole heavy-duty robots. It constructs a model representing the factors affecting the motion stability of downhole heavy-duty robots that closely reflects the actual working conditions of downhole heavy-duty robots, providing a reliable basis and support for the study of the motion stability of downhole heavy-duty robots. Attached Figure Description

[0085] Figure 1 This is a flowchart illustrating the method for constructing a characterization model of factors affecting the motion stability of a downhole heavy-duty robot in a specific embodiment of the present invention.

[0086] Figure 2 This is a schematic diagram of the theoretical spatial steady-state ZMP distribution of the downhole heavy-duty robot in a specific embodiment of the present invention.

[0087] Figure 3 This refers to the change in the fuselage pitch angle over time during the hill-climbing phase in the prototype simulation experiment of this invention.

[0088] Figure 4 This refers to the roll angle change during the lateral extension of the execution arm in the prototype simulation experiment of this invention.

[0089] Figure 5This study compares the predicted ZMP coordinates with the measured ZMP coordinates in the X direction using a characterization model of factors affecting the motion stability of a heavy-duty downhole robot during an in-situ test.

[0090] Figure 6 This study compares the predicted ZMP coordinates with the measured ZMP coordinates in the Y direction using a characterization model of factors affecting the motion stability of a heavy-duty downhole robot during an in-situ test.

[0091] Figure 7 This is a curve comparing the predicted and measured values ​​of the machine roll angle during hoisting operations in an underground field test.

[0092] Figure 8 This is a curve comparing the predicted and measured values ​​of the fuselage pitch angle during the hoisting operation in the underground field test. Detailed Implementation

[0093] The technical solutions and effects of the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0094] Please refer to Figure 1 A method for constructing a characterization model of factors affecting the motion stability of a downhole heavy-duty robot includes the following steps:

[0095] S0. Obtain the internal and external factors of multi-source coupling, specifically:

[0096] Analyze and obtain the multi-source coupled internal and external factors affecting the motion stability of downhole heavy-duty robots;

[0097] S1. Categorize the internal and external factors of multi-source coupling, specifically as follows:

[0098] S10. Classify the above-mentioned multi-source coupling internal and external factors to obtain four types of coupling factors, including environment-mechanical coupling B1, mechanical body coupling B2, perception-environment coupling B3 and mechanical-electrical-hydraulic coupling B4.

[0099] S11. The above four types of coupling factors are further refined to obtain the corresponding coupling sub-factors;

[0100] S2. The influence weights are quantified and obtained using the fuzzy hierarchical analysis method, specifically as follows:

[0101] Based on the above four types of coupling factors and their corresponding sub-factors, the influence weights of the four types of coupling factors on the motion stability of the downhole heavy-duty robot, and the influence weights of the corresponding sub-factors on the motion stability of the downhole heavy-duty robot are quantified and obtained by fuzzy hierarchical analysis.

[0102] S3. Construct a preliminary model to characterize the factors affecting the motion stability of downhole heavy-duty robots;

[0103] Based on the influence weights of four types of coupling factors on the motion stability of downhole heavy-duty robots, a primary model representing the influencing factors of motion stability of downhole heavy-duty robots is constructed.

[0104] S4. Distinguish parameter sensitivity, specifically:

[0105] S40. The influence weights of the four types of coupling factors on the motion stability of the downhole heavy-duty robot are ranked, and the top two coupling factors in terms of influence weight are obtained, namely environmental-mechanical coupling B1 and mechanical-electrical-hydraulic coupling B4.

[0106] S41. Specifically identify the influence of the coupling sub-factors corresponding to the environment-mechanical coupling B1 on the motion stability of the downhole heavy-duty robot, and obtain the key parameters that are highly sensitive to the motion stability of the downhole heavy-duty robot among the coupling sub-factors corresponding to the environment-mechanical coupling B1.

[0107] S42. Distinguish the influence of the coupling sub-factors corresponding to the mechanical-electrical-hydraulic coupling B4 on the motion stability of the downhole heavy-duty robot, and obtain the key parameters of the coupling sub-factors corresponding to the mechanical-electrical-hydraulic coupling B4 that are highly sensitive to the motion stability of the downhole heavy-duty robot.

[0108] S5. Integrate and identify key parameters, specifically:

[0109] A hybrid identification algorithm combining particle swarm optimization (PSO) and least squares support vector machine (LSSVM) is adopted. Based on a pre-built simulation model, the key parameters that are highly sensitive to the motion stability of the downhole heavy-duty robot are identified by fusing the coupling sub-factors corresponding to the environment-mechanical coupling B1 and the coupling sub-factors corresponding to the mechanical-electrical-hydraulic coupling B4. The optimal values ​​of the key parameters affecting the motion stability of the downhole heavy-duty robot are obtained.

[0110] S6. Construct a characterization model for the factors affecting the motion stability of downhole heavy-duty robots, specifically:

[0111] Based on the optimal values ​​of key parameters affecting the motion stability of downhole heavy-duty robots and the preliminary model representing the factors influencing the motion stability of downhole heavy-duty robots, a model representing the factors influencing the motion stability of downhole heavy-duty robots is constructed.

[0112] In this embodiment, the parameters of the downhole heavy-duty robot used are shown in Table 1:

[0113] Table 1

[0114] In this embodiment, in step S0, the multi-source coupled internal and external factors affecting the motion stability of the downhole heavy-duty robot are analyzed and obtained. The specific process is as follows:

[0115] First, based on the Zero Moment Point (ZMP) theory, a crucial indicator for determining the dynamic stability of a robot, if the robot's actual ZMP falls within the effective contact range, the robot can maintain both static and dynamic stability. Based on the parameters of the aforementioned heavy-duty downhole robot, the theoretical steady-state ZMP centroid spatial coordinates of the heavy-duty downhole robot are analyzed. , and As shown in formula (4), where, To perform arm joint motion angles within the longitudinal motion plane, The effective contact length of the dual-track walking mechanism with a bottom plate. The effective working length of the integration module. These represent the effective working lengths of each joint of the actuator arm, with i taking values ​​of 1, 2, and 3; H represents the body height of the downhole heavy-duty robot. For the quality of the tracked walking mechanism, For the mass of the lifting, translation and rotation mechanisms in the integrated platform module, The total mass of the integration module, For the quality of each section of the actuator arm turntable, The total mass of the boom turntable;

[0116] Formula (4)

[0117] Based on the actual motion of the downhole heavy-duty robot, its main working planes are the longitudinal YOZ plane and the transverse XOY plane. For ease of analysis, its steady-state ZMP is decomposed into the YOZ and XOY planes. Considering the actual load of the downhole heavy-duty robot and the actual condition of the walking platform, the steady-state ZMP in the longitudinal YOZ plane can be obtained, i.e. And the steady-state ZMP in the transverse XOY plane, i.e. As shown in formula (5), in the formula, The overall weight of the downhole heavy-duty robot under heavy-duty working conditions. To determine the load weight of the actuator arm, This refers to the actual effective working length of the integration module under heavy load conditions. The actual effective working arm length of the actuator:

[0118] Formula (5)

[0119] According to formula (5), substituting the parameters of the downhole heavy-duty robot, its actual ZMP will transform from a constant space on the transverse and longitudinal circular planes to a distribution on the planes of the same direction. As the dynamic center, On a spatial ellipsoid with dynamic radius, such as Figure 2 As shown;

[0120] analyze Figure 2 It can be seen that under load conditions, if the ZMP of the downhole heavy-duty robot is located at... Figure 2 Within the ellipsoid shown, the underground heavy-duty robot can operate normally and stably; otherwise, it may tip over. The position and size of the ellipsoid dynamically adjust due to factors such as load mass, changes in the robot's dynamic motion parameters, dynamic workload, and frequent adjustments to motion parameters. Analyzing the steady-state ZMP spatial distribution of the underground heavy-duty robot theoretically ensures its steady-state operation and obstacle crossing on rough terrain. However, in actual operation, factors such as the unstructured floor of the underground roadway, frequently changing external loads, and the robot's own structural coupling alter the ground pressure distribution, thus affecting the ZMP center position. The hysteresis of the electro-hydraulic coupling response introduces delays in joint movement, causing the zero-torque point to deviate from the theoretical trajectory. Therefore, its actual steady-state ZMP will not be limited to the theoretical trajectory. Figure 2 Random changes occur on the ellipsoid;

[0121] Therefore, it is necessary to analyze and obtain the multi-source coupled internal and external factors affecting the motion stability of the underground heavy-duty robot. Secondly, it is necessary to consider the underground space environment of the underground heavy-duty robot, that is, the external factors that affect its motion stability: the underground roadway is usually long and winding, with a width of 3-5 meters and a height of 2.5-4 meters. It usually adopts a fixed cross section, such as a trapezoidal, rectangular or semi-circular arched narrow restricted working space; the working roadway floor is mostly a complex road condition with a mixture of coal slurry, sand, groundwater and other materials, which is soft, slippery, muddy and uneven. During walking and working, it will be attached, which can easily cause the robot to slip and roll over. In addition, the roadway has a slope change, which increases the difficulty of the robot to work stably; in order to accurately establish the characterization model of the motion stability of the underground heavy-duty robot, it is necessary to consider the nonlinear resistance factor introduced by the indentation between the track walking mechanism and the roadway floor (considering that the underground heavy-duty robot has a large self-weight, the influence of track internal resistance, rolling resistance and steering resistance is ignored), as shown in formula (6), where, The positive pressure during the load-bearing process of the downhole heavy-duty robot. and These represent the soil cohesive deformation modulus and the soil friction deformation modulus, respectively; n is the soil deformation index; B is the width of the track on one side; and z is the deformation of the contact tunnel floor. ;

[0122] Formula (6)

[0123] The analysis identified multiple coupled internal and external factors affecting the motion stability of downhole heavy-duty robots. Furthermore, the working environment of downhole heavy-duty robots needs to be considered. Downhole environments typically maintain high temperatures of 35-40℃, humidity consistently above 80%, low illumination, and dust concentrations as high as 10-50 mg / m³. 3 This increases the interference and difficulty when the downhole robot's perception system interacts with environmental information, thereby affecting the timeliness of the downhole robot's feedback control and its stability.

[0124] The multi-source coupled internal and external factors affecting the motion stability of the downhole heavy-duty robot were analyzed and obtained. Finally, the influence of the robot's own structure also needs to be considered. By conducting a force analysis on the downhole heavy-duty robot and analyzing the connection form and motion transmission mode between its modules, the following results can be obtained:

[0125] 1) Coupling relationship analysis of the walking module, specifically:

[0126] The dual-track walking module enables the underground heavy-duty robot to walk by driving a hydraulic motor. It consists of an electric drive unit, a hydraulic execution unit, and a mechanical transmission unit. The coupling relationship is reflected in the conversion and dynamic feedback of "electrical signal-hydraulic energy-mechanical energy".

[0127] 2) Analysis of the coupling relationship between fuselage modules, specifically:

[0128] The underground heavy-duty robot's body module consists of a body and a lifting and transfer platform. The lifting mechanism within the platform can control the lifting and lowering of the actuator arm along the body's height direction by ±220mm; the translation mechanism can also control the lifting and lowering of the actuator arm along the body's height direction by ±220mm; the translation mechanism can control the translation of the actuator arm along the body's length direction by ±220mm; and the rotation mechanism can control the actuator arm's free rotation within a range of ±180°. A linear hydraulic actuator (hydraulic cylinder) controls the lifting, lowering, and translation of the lifting and transfer platform, while a rotary hydraulic actuator (hydraulic motor) controls its rotation.

[0129] 3) Analysis of coupling relationships between the execution arm modules, specifically:

[0130] The downhole heavy-duty robot arm adopts a series-parallel hybrid configuration, forming three motion control loops together with the hydraulic drive system. Among them, the linear hydraulic actuators in each loop control the pitching motion of the arm. The drive hydraulic cylinders of the three motion control loops are coupled with the arm structure to achieve coupled motion output of the robot.

[0131] In this embodiment, firstly, by analyzing the workspace and environment of the downhole heavy-duty robot, as well as the robot's own electromechanical-hydraulic characteristics, four core coupling factors affecting the motion stability of the downhole heavy-duty robot were identified. Secondly, using fuzzy hierarchical analysis, the influence weights of the four coupling factors on the motion stability of the downhole heavy-duty robot were scientifically quantified, and a primary characterization model of the influencing factors on the motion stability of the downhole heavy-duty robot was constructed based on the influence weights of the four coupling factors. Then, the two coupling factors with the largest influence weights on the motion stability of the downhole heavy-duty robot, namely environment-mechanical coupling B1 and electromechanical-hydraulic coupling B4, were selected, and parameter sensitivity discrimination was performed to obtain the key parameters that are highly sensitive to the motion stability of the downhole heavy-duty robot in the corresponding coupling sub-factors of B1 and B4. Finally, a combination of particle swarm optimization algorithm (PSO) and least squares support vector machine algorithm (LSSVM) was used. This invention employs a hybrid identification algorithm and, based on a pre-built simulation model, fuses and identifies key parameters highly sensitive to the motion stability of downhole heavy-duty robots among the corresponding coupling sub-factors B1 and B4, obtaining optimal values ​​for these key parameters affecting the motion stability of the downhole heavy-duty robot. Finally, based on the optimal values ​​of the key parameters affecting the motion stability of the downhole heavy-duty robot and the primary model representing the influencing factors of the robot's motion stability, a model representing the influencing factors of the robot's motion stability is constructed. Compared with existing technologies, this invention integrates and considers the complex and constrained working conditions of downhole heavy-duty robots during downhole operations, as well as the comprehensive impact of the coupling effects between its mechanical system, hydraulic system, and electrical system on the motion stability of the downhole heavy-duty robot. It constructs a model representing the influencing factors of motion stability that closely reflects the actual working conditions of downhole heavy-duty robots, providing a reliable basis and support for the study of the motion stability of downhole heavy-duty robots.

[0132] Furthermore, in step S11, the coupling sub-factors corresponding to the environmental-mechanical coupling B1 include the indentation coupling between the track and the bottom plate C11, the shear stress coupling between the track and the bottom plate C12, and the coupling between the roadway inclination angle and the overturning moment of the fuselage C13.

[0133] The coupling sub-factors corresponding to mechanical body coupling B2 include the serial structural coupling between the walking mechanism and the fuselage C21, the geometric coupling between the fuselage lifting and translation C22, the attitude coupling between the actuator arm and the fuselage C23, and the structural stiffness coupling of the actuator arm joint C24.

[0134] The coupling sub-factors corresponding to the perception-environment coupling B3 include the imaging coupling of dust and visual sensors C31, the coupling of low illumination and laser sensing signals C32, the coupling of narrow space and sensing signals C33, and the noise coupling of electromagnetic environment and sensing signals C34.

[0135] The coupling factors corresponding to the electromechanical-hydraulic coupling B4 include the response coupling between electrical control signals and hydraulic actuation C41, the efficiency coupling between hydraulic energy and mechanical energy conversion C42, and the closed-loop coupling between mechanical motion state and electrical feedback C43.

[0136] Further, in step S2, the quantification and acquisition of influence weights based on fuzzy hierarchical analysis specifically includes the following steps:

[0137] S20. Set analysis objectives:

[0138] The objective of the analysis is to quantify the weights of the influence of multiple coupled internal and external factors on the motion stability of downhole heavy-duty robots.

[0139] S21. Construct a three-layer structure consisting of target layer A, criterion layer B, and sub-criterion layer C, specifically including:

[0140] Based on the analysis objectives, target layer A is constructed as follows: weighted analysis of the influence of multi-source coupling factors on the motion stability of downhole robots;

[0141] Based on the four types of coupling factors, the criterion layer B is constructed as follows: environment-mechanical coupling B1, mechanical body coupling B2, perception-environment coupling B3, and mechanical-electrical-hydraulic coupling B4.

[0142] Based on the corresponding coupling sub-factors of the four types of coupling factors, the sub-criteria layer C corresponding to criteria layer B is constructed as follows:

[0143] B1: Indentation coupling between track and floor plate C11, Shear stress coupling between track and floor plate C12, Coupling between tunnel inclination angle and overturning moment of fuselage C13;

[0144] B2: Series structural coupling between the walking mechanism and the fuselage C21, geometric coupling of fuselage lifting and translation C22, attitude coupling between the actuator arm and the fuselage C23, structural stiffness coupling of the actuator arm joint C24.

[0145] B3: Dust coupling with visual sensor imaging C31, Low illumination coupling with laser sensing signal C32, Confined space coupling with sensing signal C33, Electromagnetic environment coupling with sensing signal noise C34;

[0146] B4: Response coupling between electrical control signals and hydraulic actuation; C41: Efficiency coupling between hydraulic energy and mechanical energy conversion; C42: Closed-loop coupling between mechanical motion state and electrical feedback; C43:

[0147] S22. Establish the fuzzy judgment matrix, specifically as follows:

[0148] Based on the target layer A, criterion layer B, and sub-criterion layer C, domain experts are invited to construct the fuzzy judgment matrix of criterion layer B on target layer A and the fuzzy judgment matrix of sub-criterion layer C on criterion layer B using triangular fuzzy numbers.

[0149] S23. Obtain the influence weight, specifically:

[0150] Based on the fuzzy judgment matrix, the influence weight of criterion layer B on target layer A and the influence weight of sub-criterion layer C on criterion layer B are calculated using the fuzzy hierarchical analysis method.

[0151] Based on the influence weight of criterion layer B on target layer A and the influence weight of sub-criterion layer C on criterion layer B, the influence weight of sub-criterion layer C on target layer A is obtained;

[0152] Based on the correspondence between criterion layer B and the four types of coupling factors, and the correspondence between sub-criterion layer C and the coupling sub-factors, the influence weights of the four types of coupling factors on the motion stability of the downhole heavy-duty robot, and the influence weights of the corresponding coupling sub-factors on the motion stability of the downhole heavy-duty robot are obtained.

[0153] In this embodiment, domain experts were invited to use triangular fuzzy numbers (l, m, u) and, combined with the aforementioned theoretical analysis of multiple internal and external coupling factors, to construct fuzzy judgment matrices of criterion layer B on target layer A, as shown in Table 2. The fuzzy judgment matrices of each sub-criterion layer C on its respective criterion layer B are shown in Tables 3-6. The scale references in the tables are: equally important (1, 1, 1), slightly important (1, 2, 3), significantly important (3, 4, 5), strongly important (5, 6, 7), extremely important (7, 8, 9), and reciprocity is (1 / u, 1 / m, 1 / l).

[0154] Table 2

[0155] As shown in Table 2, B1 directly affects the adhesion between the track and the bottom plate and the anti-overturning ability, and has the most significant impact on the motion stability of the downhole heavy-duty robot. B4 determines the power output accuracy, and its importance is second.

[0156] Table 3

[0157] Table 3 is the fuzzy judgment matrix of sub-criteria layer C of B1 to B1. In Table 3, C11 directly affects the normal pressure distribution of the track through formula (6), which is the basis of adhesion performance under complex terrain conditions; C12 determines the traction force and affects the anti-skid capability; C13 is related to the overturning moment, but its influence is weaker than the former two.

[0158] Table 4

[0159] Table 4 shows the fuzzy judgment matrix of sub-criteria layer C for B2. In Table 4, C21 is the "basic connection" of the mechanical body, which directly determines the coordination of the movement of each module. For example, whether the track turning causes a sudden change in the body posture has the greatest impact on the overall stability, so it has the highest weight. C22 affects the local motion accuracy of the lifting transfer platform, and its impact on the overall stability is weaker than C21 and C23. C23 is that the load and movement of the actuator arm can easily cause the body center to shift, affecting the static stability. Its importance is less than C21 but higher than C23. C24 mainly affects the working accuracy of the actuator arm, and its indirect impact on the motion stability of the body is the smallest.

[0160] Table 5

[0161] Table 5 shows the fuzzy judgment matrix of sub-criteria layer C for B3. In Table 5, C31, the visual sensor, is the core of the environmental modeling of the downhole heavy-duty robot. The imaging distortion caused by dust is the most direct perceptual interference and has the greatest impact on control decisions. C32, the lidar, relies on light reflection. Low illumination will reduce its detection range and accuracy, affecting path planning. Its importance is second only to C31. C33 only affects ranging accuracy in confined spaces. Its applicable scenarios are relatively limited, and its importance is weaker than C31 and C32. C34, electromagnetic interference can be mitigated through filtering and has the lowest impact on the perception results.

[0162] Table 6

[0163] Table 6 shows the fuzzy judgment matrix of sub-criteria layer C for B4. In Table 6, C41, the dynamic process of the electrical control signal driving the hydraulic actuator, is the starting point of "electro-hydraulic-mechanical" coordination and directly determines whether the control command can be accurately converted into hydraulic action. Its response speed has the most significant impact on dynamic stability, hence it has the highest weight. C42, based on the efficiency of hydraulic energy conversion into mechanical energy, reflects whether the output force of the hydraulic motor or hydraulic cylinder can be effectively transmitted to the mechanical structure. Efficiency loss will lead to insufficient power and affect the robot's anti-interference ability. Its importance is less than C41 but higher than C43. C43 corresponds to the closed-loop control of the robot's "mechanical motion state - sensor detection - electrical control and adjustment". It is the guarantee link for stability and depends on the command transmission of C41 and the energy conversion of C42. If the former two fail, the feedback closed loop cannot maintain stability on its own, hence it has the lowest importance.

[0164] Based on the above analysis results, the weight ranking of criterion layer B to target layer A (BA) can be obtained, as shown in Table 7, and the weight of sub-criterion layer C to criterion layer B (CB). By multiplying the weight of sub-criterion layer C to criterion layer B and the weight of criterion layer B to target layer A, the influence weight of sub-criterion layer C on target layer A can be obtained, as shown in Table 8.

[0165] Formula (3)

[0166] Table 7

[0167] In this embodiment, the primary model expression for representing the factors affecting the motion stability of the downhole heavy-duty robot is as follows: As shown in formula (3), in the formula, The weighting of the impact of environmental-mechanical coupling B1 on the motion stability of the downhole heavy-duty robot. The weight of the influence of mechanical body coupling B2 on the motion stability of the downhole heavy-duty robot is given. The weighting of the impact of perception-environment coupling B3 on the motion stability of the downhole heavy-duty robot. The weights for the influence of electromechanical-hydraulic coupling B4 on the motion stability of the downhole heavy-duty robot are 0.35, 0.20, 0.17 and 0.28, respectively.

[0168] Table 8

[0169] Furthermore, in step S41, the key parameter among the coupling sub-factors corresponding to the environment-mechanical coupling B1 that is highly sensitive to the motion stability of the downhole heavy-duty robot is:

[0170] The key parameter in the indentation coupling C11 between the track and the bottom plate is: soil cohesive deformation modulus k. c Soil frictional deformation modulus k φ The key parameters in the coupling of soil deformation index n and shear stress C12 between the track and the bottom plate are: shear strength parameter τ(s) and shear displacement coefficient K. s ;

[0171] The key parameter in the coupling between the tunnel inclination angle and the fuselage overturning moment C13 is: the inclination angle coupling coefficient K. θ ;

[0172] In step S42, the key parameter among the coupling sub-factors corresponding to the electromechanical-hydraulic coupling B4 that is highly sensitive to the motion stability of the downhole heavy-duty robot is:

[0173] The key parameter in the response coupling between electrical control signals and hydraulic actuation C41 is the electro-hydraulic response coefficient K. eh Response delay time t d ;

[0174] The key parameter in the efficiency coupling of hydraulic and mechanical energy conversion C42 is: hydraulic-mechanical conversion efficiency η. hm Mechanical transmission efficiency η m .

[0175] In this embodiment, the key parameters that are highly sensitive to the motion stability of the downhole heavy-duty robot in the coupling sub-factors corresponding to B1 are analyzed and obtained, as follows:

[0176] 1) C11:

[0177] A scaled-down simulation model of a heavy-duty underground robot was built, with a typical working condition of carrying a 20kg load. Three road conditions simulating actual underground scenarios were selected, including a smooth cement road, a bumpy road, and a mixed soft road, to conduct simulation tests and test the changes in the center of gravity of the simulation model.

[0178] Analysis of the test results shows that after reaching a stable motion state, the fluctuation range of the robot's center of gravity is less than ±5mm on a smooth cement road surface, and the indentation is close to 0, indicating high stability. On a bumpy road surface, the fluctuation range of the robot's center of gravity reaches 5-28mm, and the indentation varies with the road surface's convexity / concavity. On a mixed soft road surface, the indentation of the robot's center of gravity is stable at 8-41mm, with the largest indentation. Therefore, the key parameter in C11 that is highly sensitive to the motion stability of the underground heavy-duty robot is the soil cohesive deformation modulus k. c Soil frictional deformation modulus k φ and soil deformation index n;

[0179] 2) C12:

[0180] First, a rigid-flexible coupled dynamic simulation model of the downhole heavy-duty robot was built in RecurDyn, and typical parameters were set, including track width B = 400 mm and soil cohesive deformation modulus k. c =3MPa, soil friction deformation modulus k φ =16MPa, driving speed 1.02m / s; Three road conditions were selected to simulate actual underground scenarios, including smooth cement road, bumpy road and mixed soft road, and simulation tests were carried out to test the acceleration of the simulation model;

[0181] Analysis of the test results shows that: on smooth cement roads, the acceleration remains stable near zero with small fluctuations, indicating minimal disturbance to the heavy-duty robot during operation, near-uniform motion, uniform ground pressure distribution, and nearly constant shear stress. On rough roads, the acceleration exhibits large, high-frequency oscillations with significant positive and negative peaks, indicating continuous and intense impact and vibration during operation. This reflects the periodic compression, shearing, and rebound process of the tracks on the uneven ground, resulting in significant plastic deformation of the ground. On mixed soft roads, the acceleration curve shows alternating periods of smooth and oscillating segments, corresponding to the transition process of the vehicle traveling between different stiffness road surfaces. This indicates significant compression and shear deformation of the ground by the tracks, with significant shear strength parameter τ(s) and shear displacement coefficient K. sIt has become a key parameter in C12 that is highly sensitive to the motion stability of the downhole heavy-duty robot.

[0182] 3) C13

[0183] First, simulation scenarios with three slopes were constructed in RecurDyn's rigid-flexible coupling dynamics simulation model for downhole heavy-duty robots: 5°, 15°, and 25°. The driving speed was set to 2.5 km / h, and the driving function was STEP(time, 1, 0, 4, 3, 0). The driving process was divided into three stages: ground-slope transition stage (speed decrease, acceleration fluctuation), stable climbing stage (speed / acceleration stable), and slope-ground transition stage (speed recovery, acceleration fluctuation). The simulation model was tested for changes in driving speed and vertical acceleration under different slopes, as well as ZMP coordinates.

[0184] Analysis of the test results shows that, under a 5° slope, the speed of the simulation model during the stable climbing phase is basically the same as that on flat ground, with a speed deviation of less than 0.05 m / s and an acceleration fluctuation range of ±0.2 m / s². 2 Within this range, the ZMP (Zero Motion Point) remained near the center of the support area, with a stability margin greater than 50 mm, indicating that the robot has good anti-tipping ability. At a 15° slope, the simulation model's speed decreased slightly during the stable climbing phase, with a deviation of approximately 0.1 m / s, and the acceleration fluctuation increased to ±0.3 m / s². 2 A brief impact occurred during the transition phase, with a peak acceleration of 0.4 m / s². 2 However, the ZMP remained within the support area, with a stability margin greater than 20mm, and there was no risk of overturning. At this slope, overturning moment coupling began to appear, but had not yet triggered instability. At a 25° slope, significant velocity slippage occurred during the stable climbing phase, with a deviation of 0.3-0.4 m / s, and the acceleration fluctuation range increased to ±0.5 m / s. 2 The impact intensifies during the transition phase, with a peak acceleration of 0.7 m / s². 2 At this point, ZMP has approached the rear boundary of the support area, and the minimum stability margin has dropped to 8mm, nearing the instability threshold. This phenomenon stems from the combined effect of overturning moment coupling and electro-hydraulic response delay. At steep slopes, the gravity component increases, while the hydraulic system's lag in response prevents timely compensation of the driving force, leading to velocity slippage and attitude oscillations. Therefore, in C13, the key parameter highly sensitive to the motion stability of the downhole heavy-duty robot is the tilt coupling coefficient K. θ Simultaneously, it was also revealed that the key parameter highly sensitive to the motion stability of the downhole heavy-duty robot in C41 is the electro-hydraulic response coefficient K. eh Response delay time t d .

[0185] In this embodiment, the key parameters that are highly sensitive to the motion stability of the downhole heavy-duty robot in the coupling sub-factors corresponding to B4 are analyzed and obtained, as follows:

[0186] First, a mechanical-electrical-hydraulic simulation model of the downhole heavy-duty robot was built in Matlab / Simulink, with the hydraulic system working pressure set to 21MPa, the load to 2T, and the roadway inclination angle to 8°.

[0187] Secondly, parameter sensitivity analysis was conducted using the controlled variable method, and the parameters were adjusted as follows:

[0188] Range of variables: electro-hydraulic response coefficient K eh ∈[0.8,1.2], response delay time t d ∈[0.02s,0.12s], hydraulic-mechanical conversion efficiency η hm ∈[0.65,0.95], mechanical transmission efficiency η m ∈[0.80,0.95];

[0189] 1) C41

[0190] Evaluation metrics: steady-state error of the arm joint angle, ZMP position prediction error, and slope travel speed slippage;

[0191] An adjustable step voltage signal is input to the proportional valve actuator, and the flow and pressure responses of the hydraulic cylinder are acquired through a high-precision flow sensor. The response curve is fitted based on a first-order inertial plus pure delay model to determine the electro-hydraulic response coefficient K. eh Response delay time t d ;

[0192] The analysis results show that: K eh When the value of K varies within the range of [0.8, 1.2], the steady-state error of the robot's arm joint angle shows a linear decreasing trend; eh For every 1% change, the steady-state error of the joint angle changes by 1.34%;

[0193] Response delay time t d It has a significant impact on ZMP prediction error, when t d When ∈ [0.02s, 0.06s], the ZMP prediction error remains at a low level, ≤55mm, indicating good robot stability; when t d If the time is ≥0.08s, the robot is likely to be unstable or close to being unstable. Therefore, at this time, the ZMP error begins to rise sharply to over 125mm.

[0194] K eh and t d The influence of joint angle steady-state error exhibits a three-dimensional planar relationship: when t d When <0.06s, Keh The change in t has little effect on the steady-state error of the joint angle; when t d When K >0.08s, eh The sensitivity is significantly enhanced, and the error response surface changes sharply. Considering both cross-sensitivities, high K... eh and high t d ZMP has a large error and the system is prone to oscillation; low K eh and high t d The system response is slow, and the steady-state error of the joint angles is large; therefore, the optimal combination is K. eh =1.0-1.05 and t d The system is in a stable state with a joint angle steady-state error of less than 0.06s.

[0195] The above analysis shows that the key parameter for the high motion stability of the downhole heavy-duty robot in C41 is the electro-hydraulic response coefficient K. eh Response delay time t d At the same time, it revealed K eh and t d Within a reasonable range.

[0196] 2) C42

[0197] C42 reflects the efficiency of hydraulic energy conversion into mechanical energy, with key parameters including the hydraulic-mechanical conversion efficiency η. hm Mechanical transmission efficiency η m

[0198] Evaluation metrics: anti-slip coefficient, anti-rollover coefficient, and actuator load trajectory tracking error;

[0199] The analysis results show that:

[0200] η hm Both the anti-skid coefficient and the anti-rollover coefficient show a high linear correlation, and the anti-skid and anti-rollover capabilities increase with η. hm Synchronous increase or decrease, when η hm The benefits are most significant when the value is increased to the 0.8-0.9 range, and this range is within a region with sufficient safety margin; η hm Insufficient driving force will lead to a decrease in driving force, thereby reducing the ability to resist skidding and rollover; therefore, a reasonable range of values ​​is provided for subsequent fusion and identification of key parameters;

[0201] η m It is a key parameter affecting the performance of the actuator arm; improving it can reduce the tracking error of the actuator arm's load trajectory and improve the actuator arm's tracking performance; when η m达到 In version 1.0, the overall system efficiency can approach 99%.

[0202] The above analysis shows that the key parameter for the high sensitivity of motion stability of the downhole heavy-duty robot in C42 is η.hm η m This also reveals their tendency in terms of value range.

[0203] Further, in step S5, the hybrid identification algorithm combining particle swarm optimization (PSO) and least squares support vector machine (LSSVM) is used to fuse and identify key parameters highly sensitive to the motion stability of the downhole heavy-duty robot among the coupling sub-factors corresponding to the environment-mechanical coupling and the coupling sub-factors corresponding to the mechanical-electrical-hydraulic coupling, to obtain the optimal values ​​of the key parameters affecting the motion stability of the downhole heavy-duty robot. This includes the following steps:

[0204] S50. Obtain the dataset, specifically:

[0205] Based on theoretical analysis and simulation, the reasonable range and initial values ​​of the key parameters to be fused and identified are determined.

[0206] Sampling is performed within a reasonable range of values ​​for the key parameters to generate multiple sets of input combinations for the key parameters;

[0207] For each set of key parameter input combinations, the simulation is run in a pre-built simulation model to obtain the corresponding system output;

[0208] Data set D is formed based on the input combination of key parameters and the corresponding system output;

[0209] Divide dataset D into a training set and a validation set;

[0210] S51. Train the LSSVM model, specifically as follows:

[0211] A nonlinear mapping model from key parameter input to system output is constructed using LSSVM, as shown in Equation (1), where x is the key parameter input and y(x) is the system output. Here, b represents the support vector coefficients, b is the bias term, and N is the number of data sampling points.

[0212] Formula (1)

[0213] The PSO algorithm and training set are used to train the LSSVM model, and the optimal hyperparameters (γ*, σ*) of LSSVM and the trained LSSVM model are obtained.

[0214] S52. Perform global optimization iterations on key parameters, specifically as follows:

[0215] Construct the objective function J according to formula (2), where, Let be the normalized total weight of the sub-criteria layer C to which the i-th key parameter belongs with respect to the target layer A. and These are the measured value and the model predicted value of the i-th key parameter at the j-th time, respectively; n is the total number of parameters to be identified; and N is the number of data sampling points.

[0216] Formula (2)

[0217] The trained LSSVM model is invoked, and multiple sets of key parameters from the training set are used as input to obtain the model's predicted values.

[0218] Calculate the objective function J according to formula (2);

[0219] The objective function is minimized using PSO, so that each particle position represents a combination of key parameter inputs;

[0220] Update the optimal position of the individual and the optimal position of the group based on the particle fitness value, i.e., the J value, iteratively adjust the particle velocity and position, gradually approach the global optimal key parameter combination, and find and obtain the key parameter combination that minimizes the J value.

[0221] S53, Local fine-grained optimization, specifically:

[0222] Based on the influence weight of sub-criteria layer C on target layer A obtained in step S23, five local iterations are added to optimize the top five key parameters with the highest total weight.

[0223] When J < Alternatively, when the maximum number of iterations is reached, the iteration stops, and the optimal combination of key parameters is output.

[0224] In this embodiment, the initial values ​​and ranges of the key parameters are based on empirical values ​​or the aforementioned simulations, as shown in Table 9; after fusion identification, the values ​​of the key parameters are shown in Table 9, which are the optimal values ​​of the key parameters.

[0225] By comparing the initial values ​​and the values ​​after fusion identification of the key parameters in Table 9, it can be seen that after fusion identification of the key parameters, the soil cohesive deformation modulus k c The identified value increased by 9.33% compared to the initial value, indicating that the actual cohesive characteristics of the underground floor are stronger than theoretically assumed, while the soil friction deformation modulus k φ The decrease of 3.63% indicates that the soil is more susceptible to shear deformation, while the decrease of 6% in n indicates that the actual base plate is more susceptible to shear deformation, exhibiting certain plastic characteristics; the shear strength parameter τ(s) increased by 8.40%, reflecting that the actual shear resistance of the base plate is better than the theoretical estimate; the electro-hydraulic response coefficient K eh The 7% increase indicates that the control gain of the electrical signal on the hydraulic flow in the actual system is slightly higher than the design value, and the response delay time t dThe significant 14% reduction reflects that the actual response speed of the hydraulic valves is faster than conservatively estimated; the hydraulic-mechanical conversion efficiency η hm Mechanical transmission efficiency η m The increases of 3.66% and 3.41% respectively indicate that the energy loss of the actual system is slightly better than theoretically expected;

[0226] Table 9

[0227] In this embodiment, premature convergence is avoided by utilizing the global search capability of PSO, while the nonlinear fitting capability of LSSVM enables fine adjustment of key parameters. This provides an efficient and accurate solution for identifying key parameters during the construction of a characterization model of factors affecting the motion stability of downhole heavy-duty robots.

[0228] Further, in step S51, the step of training the LSSVM model using the PSO algorithm and training set to obtain the optimal hyperparameters (γ*, σ*) of the LSSVM and the trained LSSVM model specifically involves:

[0229] Set PSO parameters: particle swarm size, learning factor, inertia weight w, and maximum number of iterations;

[0230] The LSSVM model is set up as follows: the kernel function is the radial basis function. The range of kernel parameters and regularization parameters γ and σ is initially determined through cross-validation. Each particle position in PSO represents a set of (γ, σ).

[0231] For each particle, the LSSVM model is trained using its corresponding (γ, σ) training set data to obtain the optimal hyperparameters (γ*, σ*) of the LSSVM and the trained LSSVM model.

[0232] In this embodiment, the particle swarm size M=50, the learning factor c1=c2=2.0, the inertia weight w decreases linearly from 0.9 to 0.4, and the maximum number of iterations Tmax=50; the regularization parameter γ and the regularization parameter σ are in the ranges γ∈

[10] . -2 10 3 ],σ∈[10 -3 ,10].

[0233] Model validation:

[0234] To evaluate the accuracy, reliability, and practicality of the characterization model for the motion stability influencing factors of downhole heavy-duty robots, the following experiments were conducted using prototype simulation and downhole working conditions to verify the predictive accuracy and adaptability of the characterization model for the core states of motion stability (including ZMP position, body posture, and sideslip / rollover tendency) of downhole heavy-duty robots under various working conditions such as different floor conditions, load changes, and tilt angle changes.

[0235] Prototype simulation experiment

[0236] Inner Mongolia Bixuan Coal Mining Machinery Co., Ltd. was selected to build an underground heavy-duty robot simulation test platform, a multi-source data acquisition system and a data processing terminal; the multi-source data acquisition system is used to collect track contact pressure, shear displacement, cylinder pressure, system flow, attitude angle, ZMP coordinate, steady-state joint angle, roadway inclination angle and floor humidity;

[0237] The specific experimental plan is as follows:

[0238] Operating conditions: Base conditions: coal slime (moisture content 90%), sand and gravel (particle diameter 5-10mm), soil (density 1.8g / cm³) 3 Load: 2T; Tunnel inclination angle: 20°; Movement status: tunnel travel, boom operation;

[0239] Evaluation indicators:

[0240] 1. Absolute error is used during ZMP position verification;

[0241] 2. Root mean square error (RMSE) is used for attitude stability verification.

[0242] Key point data from three characteristic motion stages—starting acceleration, uniform speed climbing, and heavy-load lifting operation after climbing—were compared. Table 10 shows a comparison of the ZMP trajectory predicted by the motion stability influencing factor characterization model of the downhole heavy-load robot with the measured ZMP trajectory in the X and Y directions. The changes and comparisons of the fuselage pitch angle over time during the climbing stage are also shown in Table 10. Figure 3 As shown in Table 11, the roll angle changes and comparisons during lateral extension of the actuator are as follows: Figure 4 As shown in Table 11;

[0243] Table 10

[0244] Table 11

[0245] As shown in Table 10, during the test, the ZMP trajectory predicted by the motion stability influencing factor characterization model of the downhole heavy-duty robot was highly consistent with the measured ZMP trajectory. The maximum absolute error was 4.2 mm and the average absolute error was 3.5 mm, both of which were much smaller than the stability margin boundary (±50 mm) of the track grounding area. This fully demonstrates that the motion stability influencing factor characterization model of the downhole heavy-duty robot constructed in this invention can be effectively used to track and predict the dynamic ZMP trajectory of the downhole heavy-duty robot.

[0246] Figure 3 In the middle, the blue solid line represents the pitch angle change predicted by the model characterizing the factors affecting the motion stability of the downhole heavy-duty robot, and the red dashed line represents the actual pitch angle change measured by the system. Figure 4 In the diagram, the solid blue line represents the roll angle change predicted by the model characterizing the factors affecting the motion stability of the downhole heavy-duty robot, and the dashed red line represents the actual roll angle change measured by the system. Figure 3 , 4 As shown in Table 11, the characterization model accurately predicted the pitch angle during hill start and the roll angle during lateral extension of the actuator arm, indicating that the characterization model of motion stability influencing factors of the downhole heavy-duty robot constructed in this invention can be used to effectively reflect the attitude disturbance of the downhole heavy-duty robot caused by load changes and terrain excitation.

[0247] Downhole field test:

[0248] A field industrial test was conducted in an auxiliary transport roadway of a coal mine in Ningxia. The roadway had a cross-section width of 3.5m and a height of 3.0m. The floor was an uneven surface composed of a mixture of coal slime and crushed gangue, with localized water accumulation. The ambient humidity was over 90%, and the dust concentration was greater than or equal to 20mg / m³. 3 The underground heavy-duty robot completed a 20m journey along a 20° slope and performed a 2T material hoisting task at a fixed point during the journey. This verified the accuracy of the motion stability influencing factor characterization model of the underground heavy-duty robot in predicting ZMP position and body posture in a real complex environment, and evaluated the early warning capability of the motion stability influencing factor characterization model of the underground heavy-duty robot for the risk of sideslip and rollover.

[0249] (1) Validation of ZMP location prediction

[0250] Three characteristic moments during the lifting operation were selected: initial boom extension (t=3s), boom under full load (t=8s), and boom rotation (t=12s). The ZMP coordinates predicted by the model representing the factors affecting the motion stability of the downhole heavy-duty robot were compared with the ZMP coordinates measured by the six-dimensional force sensor. The trajectories in the X and Y directions were compared as follows: Figure 5 and Figure 6As shown in the figure, the blue solid line represents the predicted value predicted by the model of the influence factors of motion stability of the downhole heavy-duty robot, the red dashed line represents the measured value, the blue dots represent the predicted value at three characteristic time points, and the red dots represent the measured value at three characteristic time points.

[0251] Through analysis Figure 5 and Figure 6 It can be seen that the ZMP trajectory predicted by the characterization model of motion stability influencing factors of the downhole heavy-duty robot is in high agreement with the measured ZMP trajectory, indicating that the characterization model of motion stability influencing factors of the downhole heavy-duty robot constructed using this invention can effectively track and predict the dynamic stability point of the downhole heavy-duty robot, meeting the requirements of practical engineering applications.

[0252] (2) Verification of fuselage attitude prediction

[0253] The curves comparing the predicted and measured values ​​of the hull roll angle and pitch angle during the hoisting operation are shown below. Figure 7 and Figure 8 As shown, the root mean square error (RMSE) of the roll angle prediction was 0.41°, which is 28.1% higher than the experimental result of 0.32°. The root mean square error (RMSE) of the pitch angle prediction was 0.36°, which is 28% higher than the experimental result of 0.28°. The maximum dynamic deviation occurred at the moment of boom extension and start-up, with a roll angle deviation of 0.8° and a pitch angle deviation of 0.7°.

[0254] (3) Stability early warning verification

[0255] During the experiment, the characterization model predicted the ZMP trajectory online based on real-time sensor data. When the predicted ZMP was less than 10 mm from the boundary of the stable region, an early warning was triggered. A total of 5 boundary approach events were recorded, as shown in Table 12.

[0256] Table 12

[0257] Table 12 shows the early warning performance indicators: average early warning time: 1.82s, minimum early warning advance: 1.2s, maximum early warning advance: 2.5s; side slip risk prediction: during the boom rotation process at t=8.5s, the characterization model predicted a side slip probability of 32%, and a slight side slip actually occurred, indicating that the early warning was effective; rollover risk prediction: no rollover warning was triggered throughout the entire process, and the actual operation was stable with no rollover trend.

[0258] The results of the underground field verification show that, in the harsh and unstructured environment of coal mine roadways, the prediction of the core stability state of the underground heavy-duty robot using the underground heavy-duty robot characterization model is highly consistent with the actual performance, and can provide a reliable basis for predicting the motion stability state of the robot.

[0259] The above-disclosed embodiments are merely preferred embodiments of the present invention and should not be construed as limiting the scope of the invention. Those skilled in the art will understand that implementing all or part of the above-described embodiments and making equivalent changes in accordance with the claims of the present invention are still within the scope of the invention.

Claims

1. A method for constructing a characterization model of factors affecting the motion stability of a downhole heavy-duty robot, characterized in that, Includes the following steps: S0. Obtain the internal and external factors of multi-source coupling, specifically: Analyze and obtain the multi-source coupled internal and external factors affecting the motion stability of downhole heavy-duty robots; S1. Categorize the internal and external factors of multi-source coupling, specifically as follows: S10. Classify the above-mentioned multi-source coupling internal and external factors to obtain four types of coupling factors, including environment-mechanical coupling B1, mechanical body coupling B2, perception-environment coupling B3 and mechanical-electrical-hydraulic coupling B4. S11. The above four types of coupling factors are further refined to obtain the corresponding coupling sub-factors; S2. The influence weights are quantified and obtained using the fuzzy hierarchical analysis method, specifically as follows: Based on the above four types of coupling factors and their corresponding sub-factors, the influence weights of the four types of coupling factors on the motion stability of the downhole heavy-duty robot, and the influence weights of the corresponding sub-factors on the motion stability of the downhole heavy-duty robot are quantified and obtained by fuzzy hierarchical analysis. S3. Construct a preliminary model to characterize the factors affecting the motion stability of downhole heavy-duty robots; Based on the influence weights of four types of coupling factors on the motion stability of downhole heavy-duty robots, a primary model representing the influencing factors of motion stability of downhole heavy-duty robots is constructed. S4. Distinguish parameter sensitivity, specifically: S40. The influence weights of the four types of coupling factors on the motion stability of the downhole heavy-duty robot are ranked, and the top two coupling factors in terms of influence weight are obtained, namely environmental-mechanical coupling B1 and mechanical-electrical-hydraulic coupling B4. S41. Specifically identify the influence of the coupling sub-factors corresponding to the environment-mechanical coupling B1 on the motion stability of the downhole heavy-duty robot, and obtain the key parameters that are highly sensitive to the motion stability of the downhole heavy-duty robot among the coupling sub-factors corresponding to the environment-mechanical coupling B1. S42. Distinguish the influence of the coupling sub-factors corresponding to the mechanical-electrical-hydraulic coupling B4 on the motion stability of the downhole heavy-duty robot, and obtain the key parameters of the coupling sub-factors corresponding to the mechanical-electrical-hydraulic coupling B4 that are highly sensitive to the motion stability of the downhole heavy-duty robot. S5. Integrate and identify key parameters, specifically: A hybrid identification algorithm combining particle swarm optimization (PSO) and least squares support vector machine (LSSVM) is adopted. Based on a pre-built simulation model, the key parameters that are highly sensitive to the motion stability of the downhole heavy-duty robot are identified by fusing the coupling sub-factors corresponding to the environment-mechanical coupling B1 and the coupling sub-factors corresponding to the mechanical-electrical-hydraulic coupling B4. The optimal values ​​of the key parameters affecting the motion stability of the downhole heavy-duty robot are obtained. S6. Construct a characterization model for the factors affecting the motion stability of downhole heavy-duty robots, specifically: Based on the optimal values ​​of key parameters affecting the motion stability of downhole heavy-duty robots and the preliminary model representing the factors influencing the motion stability of downhole heavy-duty robots, a model representing the factors influencing the motion stability of downhole heavy-duty robots is constructed.

2. The method for constructing a characterization model of factors affecting the motion stability of a downhole heavy-duty robot as described in claim 1, characterized in that: In step S11, the coupling sub-factors corresponding to the environmental-mechanical coupling B1 include the indentation coupling between the track and the bottom plate C11, the shear stress coupling between the track and the bottom plate C12, and the coupling between the roadway inclination angle and the overturning moment of the fuselage C13. The coupling sub-factors corresponding to mechanical body coupling B2 include the serial structural coupling between the walking mechanism and the fuselage C21, the geometric coupling between the fuselage lifting and translation C22, the attitude coupling between the actuator arm and the fuselage C23, and the structural stiffness coupling of the actuator arm joint C24. The coupling sub-factors corresponding to the perception-environment coupling B3 include the imaging coupling of dust and visual sensors C31, the coupling of low illumination and laser sensing signals C32, the coupling of narrow space and sensing signals C33, and the noise coupling of electromagnetic environment and sensing signals C34. The coupling factors corresponding to the electromechanical-hydraulic coupling B4 include the response coupling between electrical control signals and hydraulic actuation C41, the efficiency coupling between hydraulic energy and mechanical energy conversion C42, and the closed-loop coupling between mechanical motion state and electrical feedback C43.

3. The method for constructing a characterization model of factors affecting the motion stability of a downhole heavy-duty robot as described in claim 2, characterized in that, In step S2, the step of quantifying and obtaining the influence weights based on fuzzy hierarchical analysis specifically includes the following steps: S20. Set analysis objectives: The objective of the analysis is to quantify the weights of the influence of multiple coupled internal and external factors on the motion stability of downhole heavy-duty robots. S21. Construct a three-layer structure consisting of target layer A, criterion layer B, and sub-criterion layer C, specifically including: Based on the analysis objectives, target layer A is constructed as follows: weighted analysis of the influence of multi-source coupling factors on the motion stability of downhole robots; Based on the four types of coupling factors, the criterion layer B is constructed as follows: environment-mechanical coupling B1, mechanical body coupling B2, perception-environment coupling B3, and mechanical-electrical-hydraulic coupling B4. Based on the corresponding coupling sub-factors of the four types of coupling factors, the sub-criteria layer C corresponding to criteria layer B is constructed as follows: B1: Indentation coupling between track and floor plate C11, Shear stress coupling between track and floor plate C12, Coupling between tunnel inclination angle and overturning moment of fuselage C13; B2: Series structural coupling between the walking mechanism and the fuselage C21, geometric coupling of fuselage lifting and translation C22, attitude coupling between the actuator arm and the fuselage C23, structural stiffness coupling of the actuator arm joint C24. B3: Dust coupling with visual sensor imaging C31, Low illumination coupling with laser sensing signal C32, Confined space coupling with sensing signal C33, Electromagnetic environment coupling with sensing signal noise C34; B4: Response coupling between electrical control signals and hydraulic actuation; C41: Efficiency coupling between hydraulic energy and mechanical energy conversion; C42: Closed-loop coupling between mechanical motion state and electrical feedback; C43: S22. Establish the fuzzy judgment matrix, specifically as follows: Based on the target layer A, criterion layer B, and sub-criterion layer C, domain experts are invited to construct the fuzzy judgment matrix of criterion layer B on target layer A and the fuzzy judgment matrix of sub-criterion layer C on criterion layer B using triangular fuzzy numbers. S23. Obtain the influence weight, specifically: Based on the fuzzy judgment matrix, the influence weight of criterion layer B on target layer A and the influence weight of sub-criterion layer C on criterion layer B are calculated using the fuzzy hierarchical analysis method. Based on the influence weight of criterion layer B on target layer A and the influence weight of sub-criterion layer C on criterion layer B, the influence weight of sub-criterion layer C on target layer A is obtained; Based on the correspondence between criterion layer B and the four types of coupling factors, and the correspondence between sub-criterion layer C and the coupling sub-factors, the influence weights of the four types of coupling factors on the motion stability of the downhole heavy-duty robot, and the influence weights of the corresponding coupling sub-factors on the motion stability of the downhole heavy-duty robot are obtained.

4. The method for constructing a characterization model of factors affecting the motion stability of a downhole heavy-duty robot as described in claim 3, characterized in that, In step S23, the influence weight of the sub-criteria layer C on the target layer A is obtained by multiplying the influence weight of the criteria layer B on the target layer A and the influence weight of the sub-criteria layer C on the criteria layer B.

5. The method for constructing a characterization model of factors affecting the motion stability of a downhole heavy-duty robot as described in claim 2, characterized in that: In step S41, the key parameter among the coupling sub-factors corresponding to the environment-mechanical coupling B1 that is highly sensitive to the motion stability of the downhole heavy-duty robot is: The key parameter in the indentation coupling C11 between the track and the bottom plate is: soil cohesive deformation modulus k. c Soil frictional deformation modulus k φ The key parameters in the coupling of soil deformation index n and shear stress C12 between the track and the bottom plate are: shear strength parameter τ(s) and shear displacement coefficient K. s ; The key parameter in the coupling between the tunnel inclination angle and the fuselage overturning moment C13 is: the inclination angle coupling coefficient K. θ ; In step S42, the key parameter among the coupling sub-factors corresponding to the electromechanical-hydraulic coupling B4 that is highly sensitive to the motion stability of the downhole heavy-duty robot is: The key parameter in the response coupling between electrical control signals and hydraulic actuation C41 is the electro-hydraulic response coefficient K. eh Response delay time t d ; The key parameter in the efficiency coupling of hydraulic and mechanical energy conversion C42 is: hydraulic-mechanical conversion efficiency η. hm Mechanical transmission efficiency η m .

6. The method for constructing a characterization model of factors affecting the motion stability of a downhole heavy-duty robot as described in claim 5, characterized in that, In step S5, the hybrid identification algorithm, which combines the particle swarm optimization algorithm (PSO) and the least squares support vector machine (LSSVM) algorithm, is used to fuse and identify key parameters that are highly sensitive to the motion stability of the downhole heavy-duty robot in the coupling sub-factors corresponding to the environment-mechanical coupling and the coupling sub-factors corresponding to the mechanical-electrical-hydraulic coupling, to obtain the optimal values ​​of the key parameters affecting the motion stability of the downhole heavy-duty robot. This includes the following steps: S50. Obtain the dataset, specifically: Based on theoretical analysis and simulation, the reasonable range and initial values ​​of the key parameters to be fused and identified are determined. Sampling is performed within a reasonable range of values ​​for the key parameters to generate multiple sets of input combinations for the key parameters; For each set of key parameter input combinations, the simulation is run in a pre-built simulation model to obtain the corresponding system output; Data set D is formed based on the input combination of key parameters and the corresponding system output; Divide dataset D into a training set and a validation set; S51. Train the LSSVM model, specifically as follows: A nonlinear mapping model from key parameter inputs to system output is constructed using LSSVM, as shown in the formula: In the formula, x is the key parameter input, and y(x) is the system output. Here, b represents the support vector coefficients, b is the bias term, and N is the number of data sampling points. The PSO algorithm and training set are used to train the LSSVM model, and the optimal hyperparameters (γ*, σ*) of LSSVM and the trained LSSVM model are obtained. S52. Perform global optimization iterations on key parameters, specifically as follows: According to the formula Construct the objective function J, where, Let be the normalized total weight of the sub-criteria layer C to which the i-th key parameter belongs with respect to the target layer A. and These are the measured value and the model predicted value of the i-th key parameter at the j-th time, respectively; n is the total number of parameters to be identified; and N is the number of data sampling points. The trained LSSVM model is invoked, and multiple sets of key parameters from the training set are used as input to obtain the model's predicted values. According to the formula Calculate the objective function J; The objective function is minimized using PSO, so that each particle position represents a combination of key parameter inputs; Update the optimal position of the individual and the optimal position of the group based on the particle fitness value, i.e., the J value, iteratively adjust the particle velocity and position, gradually approach the global optimal key parameter combination, and find and obtain the key parameter combination that minimizes the J value. S53, Local fine-grained optimization, specifically: Based on the influence weight of sub-criteria layer C on target layer A obtained in step S23, five local iterations are added to optimize the top five key parameters with the highest total weight. When J < Alternatively, when the maximum number of iterations is reached, the iteration stops, and the optimal combination of key parameters is output.

7. The method for constructing a characterization model of factors affecting the motion stability of a downhole heavy-duty robot as described in claim 6, characterized in that, In step S51, the process of training the LSSVM model using the PSO algorithm and training set to obtain the optimal hyperparameters (γ*, σ*) of the LSSVM and the trained LSSVM model specifically involves: Set PSO parameters: particle swarm size, learning factor, inertia weight w, and maximum number of iterations; The LSSVM model is set up as follows: the kernel function is the radial basis function. The range of kernel parameters and regularization parameters γ and σ is initially determined through cross-validation. Each particle position in PSO represents a set of (γ, σ). For each particle, the LSSVM model is trained using its corresponding (γ, σ) training set data to obtain the optimal hyperparameters (γ*, σ*) of the LSSVM and the trained LSSVM model.

8. The method for constructing a characterization model of factors affecting the motion stability of a downhole heavy-duty robot as described in claim 6, characterized in that, In step S50, for each combination of key parameters, the simulation is run in a pre-built simulation model to obtain the corresponding system output, specifically as follows: Construct a rigid-flexible coupled dynamic simulation model of a downhole heavy-duty robot in RecurDyn; A mechanical-electro-hydraulic coupling simulation model of a downhole heavy-duty robot was built in Matlab / Simulink; In the rigid-flexible coupling dynamics simulation model of the downhole heavy-duty robot, k is loaded into each set of key parameter combinations. c k φ , n, τ(s), K s and K θ Perform typical operating conditions and obtain the corresponding first system output, including ZMP position coordinates, fuselage roll angle and fuselage pitch angle; In the electro-hydraulic coupling simulation model of the downhole heavy-duty robot, K is loaded into each set of key parameters. eh t d η hm and η m Perform typical working conditions and obtain the corresponding second system output, including the steady-state error of the joint angle of the actuator arm, the relative speed error, the anti-slip coefficient and the anti-rollover coefficient; By combining the key parameters of each set of key parameters with the corresponding first system output and second system output, the system output corresponding to each set of key parameters is obtained, including ZMP position coordinates, fuselage roll angle, fuselage pitch angle, steady-state error of actuator joint angle, relative speed error, anti-skid coefficient and anti-rollover coefficient.

9. The method for constructing a characterization model of factors affecting the motion stability of a downhole heavy-duty robot as described in claim 1, characterized in that, In step S3, the expression for the primary model representing the influencing factors of motion stability of the downhole heavy-duty robot is as follows: For example, in the formula: In the formula, The weighting of the impact of environmental-mechanical coupling B1 on the motion stability of the downhole heavy-duty robot. The weight of the influence of mechanical body coupling B2 on the motion stability of the downhole heavy-duty robot is given. The weighting of the impact of perception-environment coupling B3 on the motion stability of the downhole heavy-duty robot. The weight of the influence of the electromechanical-hydraulic coupling B4 on the motion stability of the downhole heavy-duty robot.

Images