Underground coal mine positioning method based on LS and ABC-SVM

Abstract

The invention discloses a coal mine underground positioning method based on LS and ABC-SVM, and relates to the technical field of positioning. The method comprises the following steps: S1, obtaining a UWB positioning coordinate data set through an experiment; s2, performing primary positioning on the mobile node by using an LS algorithm; s3, performing normalization processing on the positioned mobile node coordinate data, and dividing a processed data set into a training set and a test set according to a proportion of 7: 3; s4, constructing an SVM model; s5, optimizing the SVM model by using an artificial bee colony optimization algorithm; s6, inputting test set UWB data to test the ABC-SVM model; according to the underground coal mine positioning method based on the LS and the ABC-SVM, error correction is carried out on the result obtained after preliminary positioning through the ABC-SVM model, the underground coal mine non-line-of-sight environment positioning precision can be effectively improved, and powerful support is provided for safety management of underground coal mine locomotives and personnel.

Description

Technical Field

[0001] The present invention relates to the field of positioning technology, and in particular to a coal mine underground positioning method based on LS and ABC-SVM. Background Art

[0002] In order to effectively improve the safety of the overall production of coal mines, it is necessary to accurately locate the locomotives and personnel in the coal mine during the coal mine operation. Once a safety accident occurs, they can be rescued based on the positioning coordinate information, thereby reducing the casualties in the mine. Due to the complex non-line-of-sight environment in coal mines, general positioning technology is not applicable;

[0003] UWB positioning technology has the characteristics of high ranging accuracy, fast transmission rate, strong penetration and good scalability. Currently, common UWB positioning algorithms include TOA algorithm, TDOA algorithm, AOA algorithm, RSSI algorithm, Kalman filter, and least squares positioning algorithm. Among them, the least squares positioning algorithm is widely used, but in the complex non-line-of-sight (NLOS) environment in coal mines, this positioning algorithm has a large positioning error.

[0004] In response to the above problems, the inventors proposed a coal mine underground positioning method based on LS and ABC-SVM to solve the above problems. Summary of the Invention

[0005] In order to solve the problem of large positioning error of UWB positioning algorithm, the purpose of the present invention is to provide a coal mine underground positioning method based on LS and ABC-SVM.

[0006] To solve the above technical problems, the present invention adopts the following technical solution: a coal mine underground positioning method based on LS and ABC-SVM, comprising the following steps:

[0007] S1: Obtain UWB positioning coordinate dataset through experiments;

[0008] S2: Use LS algorithm to perform initial positioning of the mobile node;

[0009] S3: Normalize the coordinate data of the mobile node after positioning, and divide the processed data set into a training set and a test set in a ratio of 7:3;

[0010] S4: Build SVM model;

[0011] S5: Optimize the SVM model using the artificial bee colony optimization algorithm;

[0012] S6: Input the test set UWB data to test the ABC-SVM model.

[0013] Preferably, in S2, the LS algorithm is used to perform initial positioning of the mobile node, as follows:

[0014] S21: Assume that the coordinates of the four base stations A, B, C, and D are (x1, y1, z1), (x2, y2, z2), (x3, y3, z3), and (x4, y4, z4) respectively. The coordinates of the mobile node are (x, y, z), and the distances to each base station are d1, d2, d3, and d4 respectively.

[0015] Using the definition of distance, we can establish an equation for each base station:

[0016]

[0017] Where i = 1, 2, 3, 4;

[0018] S22: To solve the least squares problem, construct a linear system of equations:

[0019]

[0020] After simplification with the first base station as a reference, the equation can be converted into the following form:

[0021]

[0022] These equations can be combined into matrix form:

[0023] A·X=b Formula (4)

[0024] Where A is a coefficient matrix containing the differences in the coordinates of each base station; X is the mobile node coordinate vector; and b is a vector consisting of the measured distance and the base station coordinates.

[0025] S23: Finally, the coordinates of the moving point are solved by the least squares method:

[0026]

[0027] Q is the weight matrix, which can be set to the identity matrix or according to the confidence of the measurement.

[0028] Preferably, in S3, the UWB data is divided into a training set and a test set in a ratio of 7:3;

[0029] Preferably, in S4, an SVM model is constructed as follows:

[0030] S41: Define the regression fitting function as:

[0031] f(x i )=w T x i +b formula (5)

[0032] w represents the input sample x i The weight vector, f(x i ) is the output, b represents the offset;

[0033] S42: Introducing slack variables ξ i and Used to coordinate fitting errors, find the optimal w and b, and then solve the optimal regression function:

[0034]

[0035] st is the constraint condition, C is the regularization coefficient, Construct the Lagrangian function and transform it into the dual form:

[0036]

[0037] x i ,x j is the support vector of the original space, is the support vector in the mapping space, and its constraints are:

[0038]

[0039] S43: When the data is nonlinear training data, the optimization objective function can be expressed by the kernel function. Formula (8) introduces the kernel function K(x i ,x j ) can be transformed into:

[0040]

[0041] Correspondingly, the regression function changes to:

[0042]

[0043] The structural characteristics of support vector machine regression mainly lie in the introduction of slack variables and kernel functions so that more samples can meet the constraints during the regression fitting process.

[0044] Preferably, in S5, the SVM model is optimized using an artificial bee colony optimization algorithm, specifically as follows:

[0045] S51: The artificial bee colony is divided into collecting bees, observing bees and scout bees. At the beginning of the algorithm, a population containing multiple food sources is randomly initialized. The quality of the food source represents the fitness of the solution. The location of the food source is represented by x i , the fitness is f(x i ), each bee updates the location of the food source using the following formula:

[0046]

[0047] where x i,j is the value of the i-th food source in the j-th dimension. ij is a random number generated by uniform distribution between [-1,1]. k,j is the location of another food source selected randomly, the new food source Evaluated for fitness if Better than f(x i,j ), then update the original position

[0048] S52: Observer bees select food sources for further exploration based on the performance of the bees. The selection probability can be calculated according to the fitness:

[0049]

[0050] Where N is the number of food sources, the selected food source will be updated again, and the update method is similar to the bee-collecting stage;

[0051] S53: When a food source has not been updated by scout bees or observer bees within several rounds, it is considered exhausted. The scout bees will abandon this location and randomly explore new food sources:

[0052] x i,j =min+rand×(max-min) Formula (15)

[0053] Among them, min and max are the search space boundaries of the problem. The artificial bee colony algorithm will continue to repeat the stages of collecting bees, observing bees and scout bees until the stopping condition is met.

[0054] Preferably, in S6, the test set UWB data is input to test the ABC-SVM model, and error analysis is performed to ensure accuracy.

[0055] Compared with the prior art, the present invention has the following beneficial effects:

[0056] In the present invention, a coal mine underground positioning method based on LS and ABC-SVM is adopted, and the least squares method LS and artificial bee colony algorithm are combined with UWB technology. First, the least squares method is used for preliminary positioning, and then the artificial bee colony (ABC) algorithm is used to optimize the support vector machine (SVM) to obtain the ABC-SVM model. Finally, the ABC-SVM model is used to correct the error of the result after preliminary positioning. This method can effectively improve the positioning accuracy in the non-line-of-sight environment of coal mines, and provide strong support for the safety management of locomotives and personnel in coal mines. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0058] Figure 1 This is a flow chart of a coal mine underground positioning method based on LS and ABC-SVM in the present invention. DETAILED DESCRIPTION

[0059] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0060] Example: Figure 1 As shown, the present invention provides a coal mine underground positioning method based on LS and ABC-SVM, comprising the following steps:

[0061] S1: Obtain UWB positioning coordinate dataset through experiments;

[0062] S2: Use LS algorithm to perform initial positioning of the mobile node;

[0063] S3: Normalize the coordinate data of the mobile node after positioning, and divide the processed data set into a training set and a test set in a ratio of 7:3;

[0064] S4: Build SVM model;

[0065] S5: Optimize the SVM model using the artificial bee colony optimization algorithm;

[0066] S6: Input the test set UWB data to test the ABC-SVM model.

[0067] Furthermore, in S1, a DW-type sensor and an STM32 microcontroller were selected to simulate the complex non-linear line-of-sight environment in a coal mine, and a UWB positioning system was constructed to obtain a data set through experiments.

[0068] Furthermore, in S2, UWB data is input, including the distance between the tag and the base station, the base station coordinates, and the tag coordinates. The LS algorithm is then used to perform initial positioning of the mobile node, and the positioned mobile node coordinate data is normalized.

[0069] Furthermore, in S3, the processed data is divided into a training set and a test set in a ratio of 7:3;

[0070] Furthermore, in S4, an SVM model is constructed as follows:

[0071] S41: Define the regression fitting function as:

[0072] f(x i )=w T x i +b formula (5)

[0073] w represents the input sample x i The weight vector, f(x i ) is the output, b represents the offset;

[0074] S42: Introducing slack variables ξ i and Used to coordinate fitting errors, find the optimal w and b, and then solve the optimal regression function:

[0075]

[0076] st is the constraint condition, C is the regularization coefficient, Construct the Lagrangian function and transform it into the dual form:

[0077]

[0078] x i ,x j is the support vector of the original space, is the support vector in the mapping space, and its constraints are:

[0079]

[0080] S43: When the data is nonlinear training data, the optimization objective function can be expressed by the kernel function. Formula (8) introduces the kernel function K(x i ,x j ) can be transformed into:

[0081]

[0082] Correspondingly, the regression function changes to:

[0083]

[0084] The structural characteristics of support vector machine regression are mainly the introduction of slack variables and kernel functions, so that more samples can meet the constraints in the regression fitting process;

[0085] Furthermore, in S5, the SVM model is optimized using the artificial bee colony optimization algorithm, as follows:

[0086] S51: Initialize SVM parameters, set the kernel function and fitness function;

[0087] S52: Set the relevant parameters of the ABC algorithm, including the bee colony size, the maximum number of iterations, etc.

[0088] S53: Initialize the bee colony. The bee colony selects the best solution based on the fitness function for local search and optimization. The observer bees select a better solution based on the solution of the sampling bees and continue the search. The scout bees randomly generate new solutions to avoid local optimality.

[0089] S54: Determine whether the iteration end condition is met. If so, output the optimal solution of the penalty coefficient C and the kernel function g. Otherwise, return to step S53.

[0090] S55: Assign the optimal solution to SVM and perform predictive analysis on the sample data in the test set.

[0091] Example 1:

[0092] In step S2, the LS algorithm is used to perform the initial positioning of the mobile node. This algorithm can effectively handle noise and errors by minimizing the residual sum of squares, thereby improving positioning accuracy. In addition, the calculation process of the LS algorithm is relatively simple, which is suitable for fast calculation and real-time positioning needs. It has strong adaptability and can process a variety of data sources, which enables it to perform well in different application scenarios. In wireless positioning, the LS algorithm can also reduce the impact of multipath effects by fusing multiple measurements, thereby further improving positioning accuracy. At the same time, the high efficiency of the LS algorithm enables it to meet real-time positioning needs and perform well in dynamic environments. It has good scalability and can be easily combined with other positioning algorithms to achieve higher positioning accuracy and robustness. Using the LS algorithm for initial positioning lays a good foundation for subsequent fine-tuning and optimization, which helps to improve the performance of the entire positioning system;

[0093] Example 2:

[0094] In step S4, the SVM model is constructed as follows:

[0095] S41: Define the regression fitting function as:

[0096] f(x i )=w T x i +b formula (5)

[0097] w represents the input sample x iThe weight vector, f(x i ) is the output, and b represents the offset.

[0098] S42: Introducing slack variables ξ i and Used to coordinate fitting errors, find the optimal w and b, and then solve the optimal regression function:

[0099]

[0100] st is the constraint condition, C is the regularization coefficient, Construct the Lagrangian function and transform it into the dual form:

[0101]

[0102] x i ,x j is the support vector of the original space, is the support vector in the mapping space, and its constraints are:

[0103]

[0104] S43: When the data is nonlinear training data, the optimization objective function can be expressed by the kernel function. Formula (8) introduces the kernel function K(x i ,x j ) can be transformed into:

[0105]

[0106] Correspondingly, the regression function changes to:

[0107]

[0108] Example 3:

[0109] In step S5, the SVM model is optimized using an artificial bee colony optimization algorithm, as follows:

[0110] S51: The artificial bee colony is divided into collecting bees, observing bees and scout bees. At the beginning of the algorithm, a population containing multiple food sources is randomly initialized. The quality of the food source represents the fitness of the solution. The location of the food source is set to x i , the fitness is f(x i ). Each bee updates the location of the food source using the following formula:

[0111]

[0112] where x i,j is the value of the i-th food source in the j-th dimension. ijis a random number generated by uniform distribution between [-1,1]. k,j is the location of another food source chosen randomly. Evaluated for fitness if Better than f(x i,j ), then update the original position

[0113] S52: The observation bee selects a food source for further exploration based on the performance of the bee collector. The selection probability can be calculated based on the fitness:

[0114]

[0115] Where N is the number of food sources. The selected food source will be updated again, and the update method is similar to the bee-collecting stage;

[0116] S53: When a food source has not been updated by scout bees or observer bees within several rounds, it is considered exhausted. The scout bees will abandon this location and randomly explore new food sources:

[0117] x i,j =min+rand×(max-min) Formula (15)

[0118] Where min and max are the search space boundaries of the problem. The artificial bee colony algorithm will continue to repeat the stages of collecting bees, observing bees, and scouting bees until the stopping condition is met.

[0119] Working principle: Coal mine underground positioning method based on LS and ABC-SVM. UWB technology has the characteristics of high ranging accuracy, fast transmission rate, strong penetration and good ductility. It can effectively accurately locate the non-line-of-sight environment in underground coal mines and obtain positioning coordinate information. This information helps to understand the location of personnel and locomotives in the coal mine, and can effectively improve the overall safety of coal mine production during coal mine operations. Secondly, the least squares method (LS) can quickly obtain the initial positioning value. On this basis, the artificial bee colony algorithm is added to optimize the parameters of the support vector machine, which can make positioning more accurate, thereby greatly improving the safety of underground coal mine work.

[0120] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. A coal mine underground positioning method based on LS and ABC-SVM, characterized in that: The following steps are involved: S1: Obtain UWB positioning coordinate dataset through experiments; S2: Use LS algorithm to perform initial positioning of the mobile node; S3: Normalize the coordinate data of the mobile node after positioning, and divide the processed data set into a training set and a test set in a ratio of 7:3; S4: Build SVM model; S5: Optimize the SVM model using the artificial bee colony optimization algorithm; S6: Input the test set UWB data to test the ABC-SVM model.

2. The coal mine underground positioning method based on LS and ABC-SVM according to claim 1, characterized in that: In S2, the LS algorithm is used to perform initial positioning of the mobile node, and the least squares method is used for data fitting and analysis. The specific process is as follows: First, the coordinates of the four base stations and the distance between the mobile node and each base station are set as unknowns. The relationship between coordinates and distances is obtained according to the distance formula. The equations are transformed into a matrix form, and then the coordinates of the mobile node are solved using the least squares algorithm.

3. The coal mine underground positioning method based on LS and ABC-SVM according to claim 1, characterized in that: In S4, by building an SVM model, we can distinguish the decision boundaries of different categories of data while maintaining the generalization ability of the model. The specific process is as follows: First, determine the initialization parameters of the SVM model, then select the kernel function and fitness function, and use the training set samples to train the SVM model.

4. The coal mine underground positioning method based on LS and ABC-SVM according to claim 1, characterized in that: In S5, the optimal parameters of the SVM model are obtained through ABC algorithm optimization; The ABC algorithm optimization includes initializing the bee colony, the bee collection optimization stage, the observation bee optimization stage, and the scout bee optimization stage; In the bee-collecting optimization stage, bees perform search optimization locally; In the observation bee optimization stage, the observation bees perform selective search optimization based on the information; In the scout bee optimization stage, global search optimization is performed on the area where the optimal solution has not been found until the iteration conditions are met.

5. The coal mine underground positioning method based on LS and ABC-SVM according to claim 1, characterized in that: In S6, in the output stage, the trained model is used to perform positioning prediction on the test set data, and the prediction results are analyzed for errors to ensure the accuracy of the model.

Images