Rock drilling mechanical arm tail end positioning compensation method and device for optimizing BP neural network based on GPS-SBOA algorithm

Abstract

The drilling mechanical arm tail end positioning compensation method for optimizing the BP neural network based on the GPS-SBOA algorithm comprises the steps that the tail end target posture and the tail end current posture of a drilling mechanical arm are obtained, and basic control parameters are generated according to the tail end target posture and the tail end current posture; and performing positioning compensation on the basic control parameters by using a pre-trained BP neural network to generate actual control parameters. According to the drilling mechanical arm tail end positioning compensation method and device for optimizing the BP neural network based on the GPS-SBOA algorithm, the BP neural network can be effectively prevented from falling into a local optimal solution in the training process, and then it is guaranteed that the BP neural network can conduct tail end positioning compensation on the drilling mechanical arm more accurately.

Description

Technical Field

[0001] The invention relates to the technical field of rock drilling mechanical arm control, in particular to a method and a device for positioning and compensating a rock drilling mechanical arm end based on a GPS-SBOA algorithm optimizing a BP neural network. Background Art

[0002] With the rapid development of modern engineering technology, large hydraulic rock drilling robots have been widely used in mining, tunneling, underground engineering and other fields. However, due to the influence of mechanical geometry error, drive system error, load change, control system error and environmental factors on the rock drilling robot arm, there is a certain error between the actual position and theoretical position of the end of the rock drilling robot arm.

[0003] Common error compensation methods mainly include model-based methods and non-model-based methods. The model-based method corrects the robot's kinematic model through the robot's kinematic parameter errors. This method can bring good results when the system is known and stable, but its modeling and calculation process is complicated and the compensation effect on non-geometric errors is poor. The non-model-based method realizes error compensation by identifying the error pattern and law according to real-time feedback information or historical data, without detailed modeling of the system. This method helps to improve the stability and accuracy of the system, but there are certain limitations in the actual application of rock drilling robot arm error compensation.

[0004] Among the many existing error compensation methods, the error compensation algorithm based on neural network has the advantages of high efficiency, adaptability and robustness, and can handle complex nonlinear relationships and improve the efficiency of error correction. However, BP neural network is prone to fall into local optimality during the training process, and the training speed is slow, which seriously affects its application effect in practical engineering. To solve this problem, many optimization algorithms for optimizing BP neural network have been born in the prior art. For example, optimization using genetic algorithm, optimization using back propagation algorithm, and optimization using particle swarm algorithm, etc. Although these optimization algorithms have improved the accuracy of robot positioning to a certain extent, there is still a disadvantage of low search efficiency, and the optimization effect needs to be further improved. Summary of the invention

[0005] In order to address the deficiencies in the prior art, the present invention provides a method and device for compensating the end positioning of a rock drilling robot arm based on a GPS-SBOA algorithm-optimized BP neural network, which can effectively prevent the BP neural network from falling into a local optimal solution during the training process, thereby ensuring that the BP neural network can more accurately perform end positioning compensation for the rock drilling robot arm.

[0006] In order to achieve the above object, the specific scheme adopted by the present invention is:

[0007] The positioning compensation method of the end of the rock drilling robot arm based on the optimization of the BP neural network by the GPS-SBOA algorithm includes the following steps: Acquire the end target posture and the end current posture of the rock drilling robot arm, and generate basic control parameters according to the end target posture and the end current posture; Use the pre-trained BP neural network to perform positioning compensation on the basic control parameters and generate actual control parameters; Control the movement of the rock drilling robot arm according to actual control parameters; Among them, the method of training the BP neural network includes: The initial population position of the Secretary bird optimization algorithm (SBOA) is generated based on the Good Point Set (GPS) strategy. The initial population position corresponds to the search space of the initial weights and biases of the BP neural network, and the Secretary bird optimization algorithm is executed to update the position of the individual Secretary bird to obtain the individual optimized position. Generate optimization parameters of BP neural network according to individual optimization positions, the optimization parameters include optimization weight parameters and optimization bias parameters; The BP neural network is trained according to the optimized parameters.

[0008] As a further optimization of the above-mentioned rock drilling robot arm end positioning compensation method based on GPS-SBOA algorithm to optimize BP neural network: the specific method of generating the initial parameters of the SBOA algorithm based on the best point set strategy includes: Construct a unit cube Hs in s-dimensional Euclidean space; Generate multiple candidate points in the unit cube Hs; The candidate points are screened based on the deviation between the candidate points to obtain multiple optimal points; Convert the best point to the initial population position of the SBOA algorithm.

[0009] As a further optimization of the above-mentioned rock drilling robot arm end positioning compensation method based on GPS-SBOA algorithm optimization BP neural network: the method of screening the candidate points based on the deviation between the candidate points is: Based on the preset deviation conditions, the best points are selected from all candidate points to form a good point set. Among them, n is the number of candidate points, k is, is the fractional part, and r={2cos(2πk / p),1≤k≤s}, p is the smallest prime number satisfying (p-3) / 2≥s; The deviation condition is: Deviation satisfy And c(r,ε)n -1+ε is a constant that depends only on r and ε, where ε is an arbitrary constant.

[0010] As a further optimization of the above-mentioned rock drilling robot arm end positioning compensation method based on GPS-SBOA algorithm to optimize BP neural network: after generating the good point set, the good points in the good point set are mapped to the solution space of the SBOA algorithm to generate the initial population position of the SBOA algorithm. The mapping method is: Among them, x i (j) is the initial population position of the SBOA algorithm, For the best point, lb j andub j They are the lower and upper bounds of the solution space of the SBOA algorithm respectively.

[0011] As a further optimization of the above-mentioned rock drilling robot arm end positioning compensation method based on GPS-SBOA algorithm to optimize BP neural network: the specific method of executing the secretary bird optimization algorithm to optimize the original position of the population includes: Setting a maximum number of optimization iterations, and determining a first optimization stage, a second optimization stage, and a third optimization stage based on the maximum number of optimization iterations; In the first optimization stage, the secretary bird individuals are transferred from their original positions to the first stage positions by differential evolution. In the second optimization stage, the secretary bird individual is transferred from the first stage position to the second stage position in the form of Brownian motion; in the third optimization stage, the secretary bird individual is transferred from the second stage position to the individual optimization position based on the Levy flight strategy.

[0012] As a further optimization of the above-mentioned rock drilling robot arm end positioning compensation method based on GPS-SBOA algorithm to optimize BP neural network: in the first optimization stage, the specific method of differential evolution of secretary bird individuals is as follows: in, is the value of its j-th dimension, x random_1 and x random_2 is the random candidate solution of the first optimization stage, R1 is an array of dimension 1×Dim randomly generated in the interval [0,1], and Dim is the dimension of the solution space of the SBOA algorithm.

[0013] As a further optimization of the above-mentioned rock drilling robot arm end positioning compensation method based on GPS-SBOA algorithm optimization BP neural network: In the second optimization stage, the specific method of making the secretary bird individual move in the Brownian motion is: Among them, x best is the current position of the secretary bird individual, RB=randn(1,Dim) is the Brownian motion formula, and randn(1,Dim) is an array of dimension 1×Dim randomly generated from a standard normal distribution.

[0014] As a further optimization of the above-mentioned rock drilling robot arm end positioning compensation method based on GPS-SBOA algorithm to optimize BP neural network: In the third optimization stage, the specific method of transferring the secretary bird individual based on Levy flight strategy is: Wherein, RL = 0.5 × Levy (Dim), Levy (Dim) is the Levy flight distribution function.

[0015] As a further optimization of the above-mentioned rock drilling robot arm end positioning compensation method based on GPS-SBOA algorithm to optimize BP neural network: the basic control parameters and the actual control parameters both include multiple joint values, and the joint values ​​correspond to the rotation joints or telescopic joints of the rock drilling robot arm.

[0016] The rock drilling arm end positioning compensation device is used to implement the above-mentioned rock drilling arm end positioning compensation method based on GPS-SBOA algorithm optimization BP neural network, and the device includes: A parameter acquisition module is used to obtain the end target posture and the end current posture of the rock drilling robot arm, and generate basic control parameters according to the end target posture and the end current posture; Model training module, used to train BP neural network; The central control module is used to use the pre-trained BP neural network to perform positioning compensation on the basic control parameters to generate actual control parameters, and control the movement of the rock drilling robot arm according to the actual control parameters.

[0017] Beneficial effects: The present invention firstly adopts the best point set strategy to initialize the position of the secretary bird individuals in the secretary bird optimization algorithm, so as to make the initial population distribution more uniform and accelerate the convergence speed of the algorithm; then uses GPS-SBOA to optimize the initial weights and biases of the BP neural network, effectively overcoming the defect of the BP neural network falling into the local optimum; finally, by establishing a mapping relationship between the joint value of the robotic arm and the terminal positioning error, error compensation is realized to improve the positioning accuracy of the rock drilling robotic arm. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 is a flow chart of the end compensation of the rock drilling robot arm according to the present invention; Figure 2 It is a flowchart for training BP neural network; Figure 3It is a flowchart of the GPS-SBOA algorithm; Figure 4 It is a schematic diagram of the structure of the BP neural network; Figure 5 It is a specific process diagram of terminal compensation; Figure 6 It is a result diagram of the experiment in the specific implementation mode. DETAILED DESCRIPTION

[0019] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0020] First, it is necessary to determine the type and specific structure of the rock drilling arm to be controlled. In this embodiment, the rock drilling arm of the rock drilling trolley model G3Zi produced by Henan Gengli Engineering Equipment Co., Ltd. is taken as the control object. The rock drilling arm is a 7-degree-of-freedom arm, which includes five rotating joints and two telescopic joints. The total length of the boom and the beam is 7835mm, the extendable amount of the boom is 1600mm, and the extendable amount of the beam is 1800mm.

[0021] Afterwards, the conventional DH method was used to model the rock drilling robot arm. The DH parameters of the rock drilling robot arm are shown in Table 1.

[0022] Table 1 DH parameter table of rock drilling robot arm

[0023] During the movement of the rock drilling robot arm, the joint link coordinate system {O i-1} and {O i The coordinates between them can be transformed by rotation or translation. The transformation matrix of adjacent coordinate systems is as follows: in, is the coordinate system {O i-1}In the coordinate system {O i}, Rot represents the rotation transformation and Trans represents the translation transformation.

[0024] By expanding the above transformation matrix, we can get the posture transformation matrix between adjacent coordinate systems:

[0025] Furthermore, the coordinate system is inserted into the corresponding parameters according to the homogeneous transformation matrix and multiplied in sequence to obtain the end position of the robotic arm: in, P=[p x ,p y ,p z ] T , R represents the end posture of the rock drilling robot arm, and P represents the end position of the rock drilling robot arm.

[0026] Based on the above analysis, the positioning error of the end of the rock drilling arm in each direction can be recorded as X error , Y error and Z error , the total position error can be expressed as

[0027] See also Figures 1 to 3 The present invention first provides a method for compensating the end positioning of a rock drilling robot arm based on a GPS-SBOA algorithm to optimize a BP neural network, including S1 to S3.

[0028] S1. Obtain the terminal target posture and the terminal current posture of the rock drilling robot arm, and generate basic control parameters according to the terminal target posture and the terminal current posture. The basic control parameters and the actual control parameters both include multiple joint values, and the joint values ​​correspond to the rotation joints or telescopic joints of the rock drilling robot arm. The terminal target posture and the terminal current posture are both composed of the above-mentioned DH parameters, which will not be repeated here.

[0029] S2. Use the pre-trained BP neural network to perform positioning compensation on the basic control parameters and generate actual control parameters.

[0030] S3. Control the movement of the rock drilling robot arm according to the actual control parameters.

[0031] The method of training the BP neural network includes T1 to T4.

[0032] T1. Generate the initial population position of the SBOA algorithm based on the good point set strategy. The population includes multiple individuals, and the positions of the individuals are determined in the initial population position.

[0033] The specific methods for generating the initial population position of the SBOA algorithm based on the good point set strategy include T11 to T14.

[0034] T11. Construct the unit cube Hs in s-dimensional Euclidean space.

[0035] T12. Generate multiple candidate points in the unit cube Hs.

[0036] T13. Screen the candidate points based on the deviation between the candidate points to obtain multiple optimal points. The method for screening the candidate points based on the deviation between the candidate points is: Based on the preset deviation conditions, the best points are selected from all candidate points to form a good point set. Among them, n is the number of candidate points, k is, is the fractional part, and r={2cos(2πk / p),1≤k≤s}, p is the smallest prime number satisfying (p-3) / 2≥s; Deviation condition is: Deviation satisfy And c(r,ε)n -1+ε is a constant that depends only on r and ε, where ε is an arbitrary constant.

[0037] T14, convert the good points into the initial population position of the SBOA algorithm. Specifically, after generating the good point set, the good points in the good point set are mapped to the solution space of the SBOA algorithm to generate the initial population position of the SBOA algorithm. The mapping method is: Among them, x i (j) is the initial population position of the SBOA algorithm, For the best point, lb j andub j They are the lower and upper bounds of the solution space of the SBOA algorithm respectively.

[0038] T2. Based on the initial population position of the secretary bird generated by the Secretary Bird Optimization Algorithm (SBOA) of the good point set strategy, the Secretary Bird Optimization Algorithm is executed to update the position of the individual secretary bird to obtain the individual optimized position. The specific method of executing the Secretary Bird Optimization Algorithm to optimize the original population position includes T21 to T24.

[0039] T21, set the maximum number of optimization iterations, and determine the first optimization stage, the second optimization stage and the third optimization stage based on the maximum number of optimization iterations. In this embodiment, the maximum number of optimization iterations is set to T, the first optimization stage is The second optimization stage is The third optimization stage is t is the current iteration number. In the first optimization stage, the position of the secretary bird is updated by simulating the process of the secretary bird searching for prey. In the second optimization stage, the position of the secretary bird is updated by simulating the process of the secretary bird consuming prey. In the third optimization stage, the position of the secretary bird is updated by simulating the process of the secretary bird attacking prey. By continuously updating the position of the secretary bird, the initial parameters of the BP neural network corresponding to the secretary bird gradually approach the optimal solution in the solution space, thus completing the optimization of the BP neural network.

[0040] T22. In the first optimization stage, the secretary bird individuals are transferred from their original positions to the first stage positions by differential evolution. In the first optimization stage, the specific method of differential evolution of the secretary bird individuals is as follows: in, is the value of its j-th dimension, x random_1 and x random_2 is the random candidate solution of the first optimization stage, R1 is an array of dimension 1×Dim randomly generated in the interval [0,1], and Dim is the dimension of the solution space of the BP neural network. Differential evolution uses the differences between secretary bird individuals to generate new solutions, which enhances the diversity and global search capabilities of the algorithm.

[0041] T23. In the second optimization stage, the secretary bird individual is moved from the first stage position to the second stage position in the manner of Brownian motion. In the second optimization stage, the specific method of making the secretary bird individual move in the manner of Brownian motion is: Among them, x best is the current position of the secretary bird, RB = randn (1, Dim) is the Brownian motion formula, and randn (1, Dim) is an array of dimension 1 × Dim randomly generated from a standard normal distribution. The introduction of Levy flight strategy enhances the global search capability of the algorithm and improves the convergence accuracy of the algorithm.

[0042] T24. In the third optimization stage, the secretary bird individual is transferred from the second stage position to the individual optimization position based on the Levy flight strategy. In the third optimization stage, the specific method of transferring the secretary bird individual based on the Levy flight strategy is: Wherein, RL = 0.5 × Levy (Dim), Levy (Dim) is the Levy flight distribution function.

[0043] In addition, the escape strategies of individual secretary birds can also be simulated, which mainly include two methods: environmental camouflage and flight escape, as follows: where r i =0.5, R2 means randomly generating an array of dimension (1×Dim) from a normal distribution, x random represents a random candidate solution for the current iteration, and K represents a random selection of an integer 1 or 2.

[0044] In other embodiments of the present invention, the lengths of the three optimization stages can also be dynamically set or cross-set. For example, when the lengths of the three optimization stages are dynamically set, in the first optimization stage and the second optimization stage, the optimization results are tested according to the updated first stage position and second stage position. If the test passes, the corresponding optimization stage is ended and the next optimization stage is entered. The test of the first stage position and the second stage position can be performed based on the position change of the secretary bird individual. If the range passed by the secretary bird individual in the process of transferring from the initial position to the first stage position or from the first stage position to the second stage position is less than the preset threshold, it can be considered that the optimization effect has not met expectations, and the optimization is continued at this time, that is, the first stage position or the second stage position is continued to be updated. When the cross-setting method is adopted, the total number of iterations T can be evenly divided into multiple first sections, each first section is evenly divided into three second sections, and the first optimization stage, the second optimization stage and the third optimization stage are correspondingly performed in the second section.

[0045] T3. Generate optimization parameters of the BP neural network according to the individual optimization position. The optimization parameters include optimization weight parameters and optimization bias parameters.

[0046] T4. Train the BP neural network according to the optimized parameters.

[0047] See also Figure 4 In this embodiment, the BP neural network consists of an input layer, a hidden layer and an output layer, and each layer of neurons is connected by weights. The input of the BP neural network is 7 joint values, and the output is the error of the end position corresponding to this set of joint values ​​in the three directions of XYZ, so the number of input nodes of the BP neural network is 7, and the number of output layer nodes is 3. The hidden layer is one layer and contains 15 neurons; when the tansig function is used as the activation function. During the training process, the number of network training iterations is 1500 times, and the learning rate is set to 0.01. Since the total number of weights and biases of the BP neural network is 168, the population individual dimension dim is set to 168 in GPS-SBOA, and the number of secretary bird individuals in the population is set to 50.

[0048] In order to find the global optimal solution, the mean square error (MSE) between the predicted value and the actual value is used as the fitness function. A smaller MSE value means a smaller gap between the predicted value and the true value, indicating a more accurate and better solution.

[0049] See also Figure 5 The process of compensating the positioning error of the rock drilling robot arm is as follows: first determine the joint values ​​of the rock drilling robot arm to be compensated. Using the mapping relationship established by the BP neural network, the end positioning error E corresponding to the joint value is calculated. aim , which is the required compensation amount. Then the theoretical position of the end is calculated through the forward kinematics of the rock drilling robot arm Finally, add the compensation amount to this theoretical position to get the actual position of the end, that is,

[0050] In order to verify the effect of the present invention, 3500 sets of sample data were selected, each set containing the values ​​of the 7 joints of the robot arm and their corresponding end position errors. These sample data cover the entire working space of the rock drilling robot arm. 85% of the sample data were selected for training, and 15% of the data, i.e. 525 sets of data, were used for network prediction. Afterwards, experiments were conducted, and the experimental results are shown in Figure 2. Figure 6 shown.

[0051] exist Figure 6 In the figure, it can be seen from part a that when the number of samples is less than 3500, the maximum prediction deviation gradually decreases with the increase of the number of samples. However, once the number of samples exceeds 3500, the maximum prediction deviation tends to stabilize and finally converges to 36.078mm. It can be seen from part b that the average prediction deviation also shows a similar trend, that is, it first decreases and then stabilizes with the increase of the number of samples. When the number of samples exceeds 3500, the average prediction deviation stabilizes at about 28.208mm. According to this analysis, when the number of samples reaches 3500, the prediction error of the trained BP neural network begins to stabilize at a lower level, and further increasing the number of samples will not significantly improve the model performance. At this time, increasing the number of samples will lead to an increase in the cost of data processing and training time, while the effect on improving model performance is relatively limited.

[0052] The present invention further provides a rock drilling robot arm end positioning compensation device, which is used to implement the above-mentioned rock drilling robot arm end positioning compensation method based on GPS-SBOA algorithm optimization BP neural network, and the device includes a parameter acquisition module, a model training module and a central control module.

[0053] The parameter acquisition module is used to obtain the terminal target posture and the terminal current posture of the rock drilling robot arm, and generate basic control parameters according to the terminal target posture and the terminal current posture.

[0054] Model training module, used to train BP neural network.

[0055] The central control module is used to use the pre-trained BP neural network to perform positioning compensation on the basic control parameters to generate actual control parameters, and control the movement of the rock drilling robot arm according to the actual control parameters.

[0056] Those skilled in the art can clearly understand that, for the convenience and simplicity of description, the specific working process of the system and device described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here. In the several embodiments provided in the present disclosure, it should be understood that the disclosed system, device and method can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the modules is only a logical function division. There may be other division methods in actual implementation. For example, multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, indirect coupling or communication connection of devices or modules, which can be electrical, mechanical or other forms.

[0057] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical modules, that is, they may be located in one place or distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0058] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for compensating the end positioning of a rock drilling robot arm based on the optimization of the BP neural network using the GPS-SBOA algorithm, characterized in that: The steps include: Acquire the end target posture and the end current posture of the rock drilling robot arm, and generate basic control parameters according to the end target posture and the end current posture; Use the pre-trained BP neural network to perform positioning compensation on the basic control parameters and generate actual control parameters; Control the movement of the rock drilling robot arm according to actual control parameters; Among them, the method of training the BP neural network includes: The initial population position of the Secretary bird optimization algorithm (SBOA) is generated based on the Good Point Set (GPS) strategy. The initial population position corresponds to the search space of the initial weights and biases of the BP neural network, and the Secretary bird optimization algorithm is executed to update the position of the individual Secretary bird to obtain the individual optimized position. Generate optimization parameters of BP neural network according to individual optimization positions, the optimization parameters include optimization weight parameters and optimization bias parameters; The BP neural network is trained according to the optimized parameters.

2. The method for compensating the end positioning of a rock drilling robot arm based on the optimization of the BP neural network by the GPS-SBOA algorithm according to claim 1, characterized in that: The specific method of generating the initial population position of the SBOA algorithm based on the good point set strategy includes: Construct a unit cube Hs in s-dimensional Euclidean space; Generate multiple candidate points in the unit cube Hs; The candidate points are screened based on the deviation between the candidate points to obtain multiple optimal points; Convert the best point to the initial population position of the SBOA algorithm.

3. The method for compensating the end positioning of a rock drilling robot arm based on the optimization of the BP neural network by the GPS-SBOA algorithm according to claim 2, characterized in that: The method for screening the candidate points based on the deviation between the candidate points is: Based on the preset deviation conditions, the best points are selected from all candidate points to form a good point set. Among them, n is the number of candidate points, k is, is the fractional part, and r={2cos(2πk / p),1≤k≤s}, p is the smallest prime number satisfying (p-3) / 2≥s; The deviation condition is: Deviation satisfy And c(r,ε)n -1+ε is a constant that depends only on r and ε, where ε is an arbitrary constant.

4. The method for compensating the end positioning of a rock drilling robot arm based on the optimization of the BP neural network by the GPS-SBOA algorithm according to claim 3, characterized in that: After generating the good point set, the good points in the good point set are mapped to the solution space of the SBOA algorithm to generate the initial parameters of the SBOA algorithm. The mapping method is: Among them, x i (j) is the initial population position in the SBOA algorithm, For the best point, lb j andub j They are the lower and upper bounds of the solution space of the SBOA algorithm respectively.

5. The method for compensating the end positioning of a rock drilling robot arm based on the optimization of the BP neural network by the GPS-SBOA algorithm according to claim 1, characterized in that: The specific methods of implementing the secretary bird optimization algorithm to optimize the original position of the population include: Setting a maximum number of optimization iterations, and determining a first optimization stage, a second optimization stage, and a third optimization stage based on the maximum number of optimization iterations; In the first optimization stage, the secretary bird individuals are transferred from their original positions to the first stage positions by differential evolution. In the second optimization stage, the secretary bird individual is moved from the position in the first stage to the position in the second stage in the manner of Brownian motion; In the third optimization stage, the secretary bird individual is transferred from the second stage position to the individual optimized position based on Levy's flight strategy.

6. The method for compensating the end positioning of a rock drilling robot arm based on the optimization of the BP neural network by the GPS-SBOA algorithm according to claim 5, characterized in that: In the first optimization stage, the specific method of differential evolution of secretary bird individuals is as follows: in, is the value of its j-th dimension, x random_1 and x random_2 is the random candidate solution of the first optimization stage, R1 is an array of dimension 1×Dim randomly generated in the interval [0,1], and Dim is the dimension of the solution space of the SBOA algorithm.

7. The method for compensating the end positioning of a rock drilling robot arm based on the optimization of the BP neural network by the GPS-SBOA algorithm according to claim 5, characterized in that: In the second optimization stage, the specific method of making the secretary bird individual move in the Brownian motion is: Among them, x best is the current position of the secretary bird individual, RB=randn(1,Dim) is the Brownian motion formula, and randn(1,Dim) is an array of dimension 1×Dim randomly generated from a standard normal distribution.

8. The method for compensating the end positioning of a rock drilling robot arm based on the optimization of the BP neural network by the GPS-SBOA algorithm according to claim 5, characterized in that: In the third optimization stage, the specific method of transferring the secretary bird individuals based on the Levy flight strategy is: Wherein, RL = 0.5 × Levy (Dim), Levy (Dim) is the Levy flight distribution function.

9. The method for compensating the end positioning of a rock drilling robot arm based on the optimization of the BP neural network by the GPS-SBOA algorithm according to claim 1, characterized in that: Both the basic control parameters and the actual control parameters include a plurality of joint values, and the joint values ​​correspond to the rotary joints or telescopic joints of the rock drilling robot arm.

10. A rock drilling arm end positioning compensation device, used to implement the rock drilling arm end positioning compensation method based on GPS-SBOA algorithm optimization BP neural network as described in any one of claims 1 to 9, characterized in that: The device comprises: A parameter acquisition module is used to obtain the end target posture and the end current posture of the rock drilling robot arm, and generate basic control parameters according to the end target posture and the end current posture; Model training module, used to train BP neural network; The central control module is used to use the pre-trained BP neural network to perform positioning compensation on the basic control parameters to generate actual control parameters, and control the movement of the rock drilling robot arm according to the actual control parameters.

Images