A method and system for autonomous positioning of coal mining machine based on inertial navigation system

Abstract

The present invention discloses an autonomous positioning method and system for a coal mining machine based on an inertial navigation device, and specifically relates to the field of positioning monitoring technology. By combining a path dependency analysis index and a dynamic change rate of decision time, the navigation and path planning capabilities of a graph search algorithm in a complex mine environment are effectively improved; by real-time monitoring and evaluation of algorithm performance, potential failure modes of the algorithm are quickly identified and responded to, and the adaptability and generalization capabilities of the algorithm are significantly enhanced; monitoring of a high path dependency index helps determine whether the algorithm is overly dependent on specific environmental features, and dynamic monitoring of decision time ensures that the algorithm can maintain a consistent response speed under various operating conditions; once a problem is discovered, an immediate rollback and adjustment mechanism can quickly restore system stability, reduce potential operational interruptions, and enable the coal mining machine to maintain efficient and safe operation in a constantly changing mine environment.

Description

Technical Field

[0001] The present invention relates to the technical field of positioning and monitoring, and more specifically, to an autonomous positioning method and system for a coal mining machine based on an inertial navigation device. Background Art

[0002] In mining operations, automated and remotely controlled coal mining machines have become an important technology to improve efficiency and safety. Coal mining machines need to accurately navigate in complex mine environments to perform mining operations. Traditional navigation methods rely on GPS or other external signals, however, these methods are often limited in underground mines because GPS signals cannot penetrate the surface. In addition, the dynamic changes in the environment inside the mine, such as channel collapse or the emergence of new excavation areas, increase the complexity of navigation.

[0003] Traditional algorithms are overly dependent on specific types of mine layouts or conditions. When faced with atypical or unexpected environmental variables, the algorithm's path planning may not be accurate enough or even fail. The lack of real-time evaluation and prediction of possible failure modes of the algorithm limits the ability to respond to potential problems in a timely manner.

[0004] In order to solve the above problems, a technical solution is now provided. Summary of the invention

[0005] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides an autonomous positioning method and system for a coal mining machine based on an inertial navigation system to solve the problems raised in the above-mentioned background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A method for autonomous positioning of a coal mining machine based on an inertial navigation system comprises the following steps:

[0008] S1: Use laser scanning and inertial navigation data fusion technology to capture the spatial changes inside the mine in real time, build and update the three-dimensional model of the mine in real time;

[0009] S2: Build a graph search algorithm that adapts to narrow spaces, introduce the spatial constraints unique to mines into the graph search algorithm, and optimize the path calculation of the coal mining machine;

[0010] S3: Repeatedly run the graph search algorithm in the simulation environment to adjust the environmental variables, use the graph edit distance to compare the path similarity of different results, apply principal component analysis and cluster analysis to extract key features, and generate a path dependency analysis index;

[0011] S4: During the operation of the graph search algorithm, the time of each decision point is recorded, the change rate of the decision time is calculated using the variational method, and the volatility model is combined with the time series prediction model to generate the dynamic change rate of the decision time;

[0012] S5: Based on the path dependency analysis index and the dynamic change rate of decision time, determine whether the graph search algorithm has potential failure modes;

[0013] S6: When the graph search algorithm has a potential failure mode, the graph search algorithm rollback and adjustment process is immediately triggered.

[0014] In a preferred embodiment, S1 specifically includes:

[0015] S101: Install laser scanning equipment on the coal mining machine;

[0016] S102: Synchronize the inertial navigation system with the laser scanning device;

[0017] S103: Apply Kalman filter algorithm to fuse laser scanning data and inertial navigation data in real time;

[0018] S104: constructing a real-time three-dimensional model of the mine on a computing platform using the fused data;

[0019] S105: Regularly update the three-dimensional model according to new scanning data.

[0020] In a preferred embodiment, S2 specifically includes:

[0021] S201: Analyze the geometry of the mine and identify mine-specific spatial constraints:

[0022] The three-dimensional data of the interior of the mine is obtained through laser scanning technology; the three-dimensional point cloud processing algorithm is used to extract planes and identify features of the collected point cloud data, filter noise and extract key geometric structures in the mine; the processing results of the point cloud data are verified using ground measured data;

[0023] S202: Select a graph search algorithm suitable for narrow spaces:

[0024] Based on the narrow space characteristics of the mine, the A* algorithm is selected and customized; the heuristic function expression is: ;

[0025] is the value of the heuristic function, are the coordinates of the target node, For the current node The coordinates of is the weight coefficient, Is the current node spatial constraints;

[0026] S203: Optimize the path of the coal mining machine by dynamic weight adjustment in the narrow space of the mine:

[0027] Evaluate the surrounding environment of the coal mining machine in real time and adjust the weight value. The formula for dynamic adjustment is: ;

[0028] is the initial weight, It's time The weight adjustment amount on ;

[0029] S204: Perform algorithm testing by simulating a mine environment, and continuously iterate and optimize path calculation performance based on feedback data.

[0030] In a preferred embodiment, S3 is specifically:

[0031] After each environment adjustment, run the graph search algorithm and record the resulting paths;

[0032] The path generated by each run is represented as a sequence of nodes, each node corresponding to a navigation point of the coal mining machine;

[0033] Graph edit distance calculation: ;in, and Respectively represent the paths generated by the two runs, represents the graph edit distance;

[0034] All path similarity data are used as input to perform principal component analysis to extract the most important change patterns in the path selection process;

[0035] Construct a path similarity matrix, reduce the high-dimensional data to principal components through principal component analysis, and identify the main change directions in path selection;

[0036] Use K-means clustering algorithm to classify the principal component features corresponding to different paths;

[0037] Input the data after principal component analysis and set the initial number of cluster centers; apply the K-means algorithm to iterate until the cluster centers are stable;

[0038] Based on the principal components extracted by principal component analysis and the clustering results, the path dependency analysis index is calculated and is defined as: ;in, is the path dependency analysis index;

[0039] is the number of cluster centers, It is The number of paths contained in a cluster, is the total number of paths, is the number of principal components considered in each cluster, It is The first The variance of the principal components.

[0040] In a preferred embodiment, S4 is specifically:

[0041] Collect and record the timestamp of each decision point during the execution of the graph search algorithm ,in represents the index of the decision point;

[0042] Store the timestamp data of all decision points in an array ,in is the total number of decision points;

[0043] Calculate the rate of change of decision time by using the variational method: ;in, It is The time rate of change of decision points; and are the timestamps of adjacent decision points, respectively; is the unit time interval;

[0044] Calculate the time rate of change between adjacent decision points by using the variational method ;

[0045] Apply the volatility model: ;in, represents the volatility of the decision time change rate, It is The time rate of change, is the average of all rates of change;

[0046] Applying time series forecasting models: ;in, is the time rate of change of the predicted next decision point, and is the regression coefficient, is the current time rate of change, is the residual term;

[0047] Generate decision time dynamic rate of change: ;in, is the dynamic rate of change of decision time; is the predicted time rate of change.

[0048] In a preferred embodiment, S5 is specifically:

[0049] set up The upper threshold ;set up The lower threshold ;

[0050] set up The upper threshold ;

[0051] Calculate the and ;

[0052] like , or , then it is determined that the graph search algorithm has a potential failure mode; otherwise, it is determined that the graph search algorithm does not have a potential failure mode.

[0053] On the other hand, the present invention provides an autonomous positioning system for a coal mining machine based on an inertial navigation system, comprising a mine model building module, a path calculation optimization module, a path dependency analysis module, a decision change analysis module, a failure mode judgment module, and a failure mode decision module;

[0054] Mine model building module: uses laser scanning and inertial navigation data fusion technology to capture the spatial changes inside the mine in real time, build and update the three-dimensional model of the mine in real time;

[0055] Path calculation optimization module: Build a graph search algorithm that adapts to narrow spaces, introduce mine-specific spatial constraints into the graph search algorithm, and optimize the path calculation of the coal mining machine;

[0056] Path dependency analysis module: repeatedly run the graph search algorithm in a simulation environment to adjust environmental variables, use graph edit distance to compare the path similarities of different results, apply principal component analysis and cluster analysis to extract key features, and generate a path dependency analysis index;

[0057] Decision change analysis module: records the time of each decision point during the operation of the graph search algorithm, calculates the change rate of decision time using the variational method, and applies the volatility model combined with the time series prediction model to generate the dynamic change rate of decision time;

[0058] Failure mode judgment module: Based on the path dependency analysis index and the dynamic change rate of decision time, it is judged whether the graph search algorithm has potential failure modes;

[0059] Failure mode decision module: When a graph search algorithm has a potential failure mode, it immediately triggers the graph search algorithm rollback and adjustment process.

[0060] The technical effects and advantages of the coal mining machine autonomous positioning method and system based on inertial navigation system of the present invention are as follows:

[0061] The solution of the present invention effectively improves the navigation and path planning capabilities of the graph search algorithm in a complex mine environment by combining the path dependency analysis index and the dynamic change rate of the decision time. By real-time monitoring and evaluating the algorithm performance, the solution can quickly identify and respond to the potential failure modes of the algorithm, significantly enhancing the adaptability and generalization of the algorithm; monitoring of the high path dependency index helps determine whether the algorithm is overly dependent on specific environmental characteristics, while dynamic monitoring of the decision time ensures that the algorithm can maintain a consistent response speed under various operating conditions. In addition, once a problem is discovered, an immediate fallback and adjustment mechanism can quickly restore system stability and reduce potential operational interruptions. These advantages enable coal mining machines to maintain efficient and safe operation in a constantly changing mine environment, thereby improving production efficiency and reducing the risk of accidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 A schematic diagram of an autonomous positioning method for a coal mining machine based on an inertial navigation system according to the present invention;

[0063] Figure 2 The present invention is a schematic structural diagram of an autonomous positioning system for a coal mining machine based on an inertial navigation system. DETAILED DESCRIPTION

[0064] 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.

[0065] Example 1

[0066] Figure 1 The present invention provides a method for autonomous positioning of a coal mining machine based on an inertial navigation system, which comprises the following steps:

[0067] S1: Use laser scanning and inertial navigation data fusion technology to capture the spatial changes inside the mine in real time, build and update the three-dimensional model of the mine in real time.

[0068] S2: Build a graph search algorithm that adapts to narrow spaces, introduce mine-specific spatial constraints into the graph search algorithm, and optimize the path calculation of the coal mining machine.

[0069] S3: Repeatedly run the graph search algorithm in a simulation environment to adjust environmental variables, use graph edit distance to compare the path similarities of different results, apply principal component analysis and cluster analysis to extract key features, and generate a path dependency analysis index.

[0070] S4: During the operation of the graph search algorithm, the time of each decision point is recorded, the rate of change of the decision time is calculated using the variational method, and the volatility model is combined with the time series prediction model to generate the dynamic rate of change of the decision time.

[0071] S5: Based on the path dependency analysis index and the dynamic change rate of decision time, determine whether the graph search algorithm has potential failure modes.

[0072] S6: When the graph search algorithm has a potential failure mode, the graph search algorithm rollback and adjustment process is immediately triggered.

[0073] S1 specifically includes:

[0074] S101: Installing laser scanning equipment on a coal mining machine to capture spatial details of the mine:

[0075] High-precision laser scanning equipment is installed at key locations of the coal mining machine, such as the top or front of the machine. This equipment should have the characteristics of high resolution and wide scanning range to capture the spatial details inside the mine.

[0076] The scanning device model of choice can be Lidar X2, which has a scanning frequency of 100Hz and can provide accurate distance measurement and object recognition in low light conditions.

[0077] The scanner is fixed at the top center of the coal mining machine to ensure a 360-degree unobstructed view.

[0078] On-site calibration is performed to ensure that the equipment's measurements accurately match the actual dimensions of the mine.

[0079] S102: Synchronize the inertial navigation system with the laser scanning device to obtain position and orientation data:

[0080] An optional Honeywell HG1700 AG58 inertial measurement unit combines a gyroscope and accelerometer to provide highly accurate orientation and acceleration data.

[0081] Time synchronization techniques are used to ensure that the data collected from the laser scanner and the inertial measurement unit have the same time stamp.

[0082] Use CAN bus interface for data transmission to ensure real-time data transmission and reduce delays.

[0083] S103: Apply Kalman filter algorithm to fuse laser scanning data and inertial navigation data in real time:

[0084] In the autonomous positioning system of coal mining machines, the key is to accurately fuse laser scanning and inertial navigation data. The goal is to reduce the impact of individual sensor errors and improve the accuracy of position estimation. The Kalman filter algorithm was selected because it can effectively process noisy sensor data and optimize the estimation results in real time.

[0085] The Kalman filter algorithm is based on the following dynamic system model:

[0086] The state equation is: .

[0087] : Indicates the time step The system state vector, An index used to represent discrete time steps or sequences.

[0088] : Indicates the time step The system state vector includes the position, speed and other dynamic parameters of the coal mining machine.

[0089] : The state transfer matrix is ​​used to predict the state of the next time step from the current state. This matrix depends on the dynamic characteristics of the system. For example, if the system model assumes that the position and velocity change linearly, then Will design accordingly.

[0090] : The control input vector includes any external inputs that may affect the system state, such as drive commands or environmental forces.

[0091] : Control input matrix, control input Mapped into state space. Its structure depends on how the control input affects the state vector.

[0092] : Process noise vector, reflecting the uncertainty and external disturbance in model prediction. It is usually assumed is Gaussian noise.

[0093] The observation equation is: .

[0094] : At time step The observation vector is provided by actual sensor measurements, which may include distance and angle data obtained from a laser scanning device.

[0095] : The observation matrix is ​​used to map the state vector to the observation space. Its design is based on how sensors extract information from the system state.

[0096] : Observation noise vector, reflecting the random error in the observation, also assumed to be Gaussian noise.

[0097] Use historical data and simulated test data to adjust and optimize , and matrix to ensure optimal performance in the actual operating environment.

[0098] Here, , and The specific values ​​of the matrix need to be set according to the characteristics of the laser scanning and inertial navigation equipment and the environmental characteristics of the mine. For example, Matrices often represent the natural evolution of state and are essential for updating position and velocity. The matrix depends on the geometric layout of the observation equipment and the type of sensors.

[0099] By comparing the algorithm output with the actual measurement value, the error of the position estimation is calculated, and statistical indicators such as the sum of squared errors (SSE) and mean square error (MSE) are used to evaluate the algorithm performance.

[0100] S104: Use the fused data to build a real-time 3D model of the mine on the computing platform:

[0101] The laser scan data were processed using Autodesk Recap software and combined with the inertial navigation data after fusion via Kalman filtering.

[0102] After data processing, the model is rendered on an NVIDIA Quadro P6000 graphics processing unit to support high-complexity real-time computing.

[0103] The real-time update frequency is set to once every 10 minutes, and the optimization algorithm ensures that the update speed is improved without sacrificing accuracy.

[0104] S105: Regularly update the 3D model based on new scan data:

[0105] When a new data point differs from the existing model by more than a preset threshold (e.g. 5 cm), an update is automatically triggered.

[0106] Regular verification using laser interferometry ensures the accuracy of the model remains consistent with the real world.

[0107] Deploy data analysis tools to monitor the efficiency and accuracy of update operations to ensure that each update meets the established performance standards.

[0108] S2 specifically includes:

[0109] S201: Analyze the mine geometry and identify mine-specific spatial constraints such as channel width and height restrictions:

[0110] Laser scanning technology is used to obtain three-dimensional data of the mine interior. This data includes the mine's corridor width, ceiling height, and other geometric features that affect the navigation of coal mining machines.

[0111] A 3D point cloud processing algorithm (such as the RANSAC algorithm) is used to perform plane extraction and feature recognition on the collected point cloud data to filter out noise and extract key geometric structures in the mine. RANSAC was chosen because of its high efficiency and robustness in processing large-scale point cloud data.

[0112] The threshold parameter in the RANSAC algorithm was set to 0.05 m, which was chosen based on the minimum identifiable feature size within the mine and the accuracy of the laser scanner. This parameter is used to determine the accuracy of the plane fitting in the point cloud data.

[0113] The processing results of the point cloud data are verified using ground measured data to ensure that the extracted geometric features accurately reflect the actual conditions of the mine, with the error controlled within 2 cm.

[0114] S202: Select a graph search algorithm suitable for narrow spaces (such as A or D Lite) and embed mine space constraints into the framework of the graph search algorithm:

[0115] Based on the narrow space characteristics of the mine, the A* algorithm was selected and customized to adapt to the specific constraints within the mine.

[0116] The heuristic function in the A* algorithm is customized to take into account the spatial constraints of the mine. The heuristic function expression is: .

[0117] : The value of the heuristic function, indicating the The estimated cost to the target node includes the Euclidean distance and the additional cost due to environmental constraints.

[0118] : The coordinates of the target node are the coordinates of the final position that the coal mining machine needs to reach. With these coordinates, the algorithm can calculate the straight-line distance from the current node to the target node.

[0119] : Current node The coordinates represent the coordinates of the current location of the coal mining machine. These coordinates are continuously updated during the path planning process to calculate the path of the coal mining machine from the current node to the target node.

[0120] : Weight coefficient, used to balance the impact of the straight-line distance of the path and the environmental constraints on path planning. It determines the importance of environmental constraints in path planning. It is determined through simulation experiments and historical data analysis. The initial value can be set to 0.3, but it can be dynamically adjusted to adapt to different environmental changes. The weight setting is based on the complexity and safety requirements of the mine, with the aim of optimizing the calculation efficiency of the path while ensuring safety.

[0121] : Current node The spatial constraints are obtained by analyzing the geometric structure of the mine and identifying specific spatial constraints. They reflect the impact of the specific environment in the mine on path selection, such as narrow passages, low areas, etc., and are usually represented by an additional path cost.

[0122] The performance of the new algorithm was verified by comparing the path selection effects of the traditional A* algorithm and the customized algorithm. The performance of different heuristic functions was compared using data from a mine simulation environment to ensure that the selected function can effectively reduce collisions and dangerous situations in path planning in actual applications.

[0123] S203: Optimize the path of the coal mining machine by dynamic weight adjustment in the narrow space of the mine:

[0124] In order to improve the path planning flexibility of the coal mining machine in the mine, a dynamic weight adjustment mechanism is introduced to respond to environmental changes in real time.

[0125] Design a dynamic weight adjustment module based on machine learning to evaluate the surrounding environment of the coal mining machine in real time (such as the emergence of new obstacles or path restricted areas) and adjust the weight value. The core formula of dynamic adjustment is: .

[0126] : Initial weight, which represents the basic weight value when there is no impact of environmental changes. It is used to set an initial weight parameter in the early stage of path planning.

[0127] :time The weight adjustment amount on the path is adjusted according to real-time environmental changes (such as the appearance of new obstacles or path-restricted areas), dynamically adjusting the focus of path planning.

[0128] Conduct dynamic weight adjustment tests in a simulated environment, record the impact of different adjustment strategies on path planning, and select the weight adjustment strategy that best improves path safety and efficiency.

[0129] S204: Algorithm testing is performed by simulating a mine environment, and path calculation performance is continuously optimized based on feedback data:

[0130] By simulating the mine environment, the path planning algorithm designed in the previous steps is tested and optimized based on the test results.

[0131] A high-fidelity simulation platform was used to reproduce typical environments and obstacle arrangements in a mine. The simulation tests included various scenarios such as different mine structures, coal mining machine operating speeds, and the random appearance of obstacles.

[0132] The following evaluation indicators are set: path length, calculation time, obstacle avoidance success rate, and safety of the coal mining machine (such as the number of collisions). These indicators are set based on the actual needs and safety standards of mine operations.

[0133] According to the test results, the algorithm is iteratively optimized by adjusting the heuristic function parameters, dynamic weight adjustment strategy, etc. After each optimization, retest and record the improvement until the expected performance target is achieved.

[0134] Detailed data for each test was recorded, including the number of visits to each node, path selection results, and the response time of the coal mining machine under different environments. Statistical analysis methods, such as regression analysis and principal component analysis, were used to identify the key factors affecting path planning performance and further optimize the algorithm accordingly.

[0135] S3 is specifically:

[0136] On a dedicated high-fidelity mine simulation platform, the operation scenarios of coal mining machines in the mine are reproduced. Environmental variables include changes in channel width, obstacle locations, mine height, humidity changes, etc. These variables are adjusted randomly or according to predetermined rules by the simulation system to test the performance of the algorithm in different environments.

[0137] After each environment adjustment, the graph search algorithm was run and the paths it generated were recorded. To ensure the adequacy of the data, the algorithm was run no less than 100 times for each variable combination.

[0138] The path generated by each run is represented as a sequence of nodes, each node corresponding to a navigation point of the coal mining machine. The representation of the path can be achieved by serializing coordinates or node indices.

[0139] Graph edit distance calculation: ;in, and Respectively represent the paths generated by the two runs, represents the graph edit distance (reflecting the difference between the paths generated by the two runs); a smaller graph edit distance indicates a high similarity of the paths, whereas a larger one indicates a large difference.

[0140] All path similarity data (i.e., the graph edit distance between the path generated in each run and other paths) were used as input to perform principal component analysis (PCA) to extract the most dominant patterns of variation in the path selection process.

[0141] A path similarity matrix is ​​constructed, where each element represents the edit distance between two paths. These high-dimensional data are then reduced to two or three principal components (PCs) through PCA to identify the main change directions in path selection.

[0142] The main parameter of PCA is the number of principal components selected (usually the first few principal components that explain at least 90% of the data variance are selected). These principal components provide the main characteristics of different path selection strategies and can reveal the dependencies of the algorithms under different environments.

[0143] The K-means clustering algorithm is used to classify the principal component features corresponding to different paths. Each cluster center represents a typical path selection mode.

[0144] Input the data after principal component analysis and set the initial number of cluster centers (usually 3-5, the specific value is determined according to the distribution of the data and the preliminary test results). Then, apply the K-means algorithm to iterate until the cluster centers are stable. The effect of clustering is evaluated by the Silhouette score to select the best number of clusters.

[0145] The selection of the number of initial cluster centers is based on cluster stability and the diversity of path patterns, with the goal of determining the main path patterns generated by the algorithm under different conditions.

[0146] Based on the principal components extracted by PCA and the clustering results, the path dependency analysis index is calculated and is defined as: ;in, is the path dependency analysis index.

[0147] The path dependency analysis index is used to evaluate the consistency of the algorithm's path selection in different environments. A higher path dependency analysis index indicates a stronger dependence on path selection, a more concentrated path selection, and a lower adaptability; it indicates that the algorithm tends to generate similar paths when encountering similar environmental conditions, which may indicate that the algorithm is too dependent on specific environmental characteristics or initial conditions.

[0148] A lower path dependence analysis index indicates that the path selection is more widely distributed and has higher adaptability. It can flexibly adjust its path selection in different environments, showing better adaptability and robustness.

[0149] : The number of cluster centers indicates the number of main categories of path selection patterns, that is, the number of typical patterns of different path selections determined in the cluster analysis.

[0150] : No. The number of paths contained in the cluster indicates the number of paths in the The number of paths around a cluster center that are classified into the same category.

[0151] : The total number of paths, that is, the total number of all generated paths during the entire path analysis process.

[0152] : The number of principal components considered in each cluster, i.e., the number of principal components extracted in the principal component analysis (PCA) to explain the variation in path selection within the cluster.

[0153] : No. The first The variance of the principal component indicates that in the principal component analysis, the first The first The larger the variance, the more significant the impact of the principal component on the path selection pattern.

[0154] After the simulation environment test, the algorithm was deployed in a real mine environment to verify its actual operation effect. The path selection results in the field test were recorded, and the path dependency analysis index in the actual operation was calculated to ensure consistency with the simulation results. It is particularly suitable for path planning optimization in complex mine environments. By identifying and analyzing path dependencies, it can provide a basis for further optimizing the path planning algorithm and improve the autonomous navigation capability of coal mining machines in complex environments.

[0155] S4 is specifically:

[0156] During the execution of the graph search algorithm, collect and record the timestamp of each decision point ,in Indicates the index of the decision point. Each time the algorithm makes a path selection, the current system time is recorded , these timestamp data will be used for subsequent change rate calculations.

[0157] Store the timestamp data of all decision points in an array ,in is the total number of decision points.

[0158] Calculate the rate of change of decision time by using the variational method: ;in, It is The time rate of change of decision points; and are the timestamps of adjacent decision points, respectively; is the unit time interval, usually set to 1 second to normalize the rate of change.

[0159] Calculate the time rate of change between adjacent decision points by using the variational method , it is possible to identify fluctuations in the response time of the algorithm during its execution.

[0160] Apply the volatility model: ;in, represents the volatility of the decision time change rate, It is The time rate of change, is the average of all rates of change.

[0161] A volatility model (such as the ARCH model) is used to calculate the volatility of the decision time change rate, reflecting the stability of the algorithm under different conditions. The reason for choosing the ARCH model is that it can effectively handle the volatility in time series data.

[0162] Applying time series forecasting models: ;in, is the time rate of change of the predicted next decision point, and is the regression coefficient, is the current time rate of change, is the residual term.

[0163] and Used in the regression part of the ARIMA model to determine the impact of the current time rate of change on the future time rate of change.

[0164] The ARIMA model is used to predict the future decision-making time change rate. The model can predict future change trends based on historical data. The ARIMA model combines the autoregressive part (AR), the difference part (I), and the moving average part (MA), and is suitable for processing time series data with trend and seasonality.

[0165] Generate decision time dynamic rate of change: ;in, It is the dynamic change rate of decision time, which reflects the difference between the actual change rate and the predicted change rate and serves as an indicator for evaluating the stability of the algorithm.

[0166] is the predicted time rate of change, predicted by the ARIMA model.

[0167] By calculating in different simulation environments , verify the response speed and stability of the algorithm in practical applications.

[0168] Higher Indicates that the response time of the algorithm varies greatly at different decision points, indicating that the algorithm may have problems in processing time efficiency, which may lead to failure in scenarios that require fast response. This indicates that the algorithm's time response at each decision point is relatively consistent and can reliably respond within the predetermined time, showing high stability and predictability.

[0169] S5 is as follows:

[0170] Evaluate the path dependency analysis index of graph search algorithms in a simulation environment by setting specific thresholds and the dynamic rate of change of decision time , to determine whether the algorithm has potential failure modes.

[0171] set up The upper threshold ,For example, .

[0172] set up The lower threshold ,For example, .

[0173] A larger value usually indicates that the algorithm has good environmental adaptability and can find effective paths under various conditions. The values ​​are usually positive, indicating that the algorithm can flexibly adapt to different environmental conditions. Abnormally low to near zero, which may mean that the algorithm is too random or lacks consistency in practical applications. Usually not a problem.

[0174] Among them, the lower threshold is close to 0 and is used to distinguish Abnormally low to near zero conditions.

[0175] if Greater than upper threshold , it indicates that the algorithm path selection is too dependent on specific environmental characteristics; if Less than the lower threshold , it may mean that the algorithm is too random or lacks consistency in practical applications.

[0176] set up The upper threshold ,For example, .

[0177] if Greater than upper threshold , indicating that the algorithm has inconsistent temporal responses at different decision points, which may lead to performance issues in practical applications.

[0178] The graph search algorithm is run in a designed simulated mine environment, and a sufficient amount of path selection and decision time data is recorded.

[0179] Use statistical software or custom scripts to calculate the and , and collect these data for further analysis.

[0180] like , or , then it is determined that the graph search algorithm has a potential failure mode; otherwise, it is determined that the graph search algorithm does not have a potential failure mode.

[0181] By setting specific thresholds to evaluate the path dependency analysis index and decision time dynamic change rate of the graph search algorithm, potential failure modes can be effectively identified and prevented, ensuring the stability and reliability of the algorithm in practical applications. This evaluation method enables algorithm developers to clarify the performance boundaries of the algorithm based on quantitative data, thereby optimizing the algorithm design and improving its ability to adapt to different environments. For example, high This suggests that the algorithm may be overly dependent on specific environmental features and needs to be adjusted to enhance generalization ability. This indicates that the algorithm may have fluctuations in time response and needs to be improved to ensure consistent response. Through these specific threshold judgments, the adjustment of the algorithm is more targeted, which can solve problems in a targeted manner, reduce the risk of failure in key operations, and improve overall performance. In addition, systematically recording and analyzing data can also help the development team accumulate experience and continuously improve the algorithm to make it more robust and efficient.

[0182] Among them, common types of failure modes include:

[0183] Inadequate performance: The algorithm cannot effectively solve the problem or achieve the expected performance indicators. For example, a graph search algorithm may not be able to find the optimal path in a reasonable time, resulting in inefficiency.

[0184] Poor adaptability: The system or algorithm cannot effectively adjust when encountering a changing environment or new conditions. For example, the path planning algorithm cannot adapt to different environmental conditions, resulting in the inability to find an effective path.

[0185] Stability issues: The system or algorithm exhibits inconsistent behavior during operation. For example, a large dynamic change rate in decision time may cause unstable system response and affect real-time performance.

[0186] Safety hazards: There are potential safety risks in the system or algorithm, which may cause the system to lose control or cause an accident. For example, the failure of the path planning algorithm of the coal mining machine in the mine may cause equipment collision or other safety accidents.

[0187] S6 is specifically:

[0188] Temporarily roll back the graph search algorithm to a previously stable algorithm version to maintain the continuity and stability of system operation, especially in critical applications.

[0189] After identifying a potential failure mode, the system will automatically roll back to the last stable algorithm version if feasible. This measure ensures that the main functions of the system are not affected and continue to provide services to users while the problem is being fixed. Several verified stable versions are preset as alternatives to ensure quick rollback to any stable state.

[0190] Analyze the specific reasons that lead to potential failure modes of graph search algorithms, whether they are caused by specific scenario configurations, data processing errors, or algorithm logic defects.

[0191] For too high , explore increasing the diversity of the algorithm, introducing additional randomization parameters or more complex decision logic to reduce dependence on specific environmental characteristics.

[0192] For too low or too high , optimize the data processing and decision time management in the algorithm, improve the time efficiency and consistency of path decisions.

[0193] Thoroughly test the adjusted algorithm in a safe test environment to evaluate its performance improvements and any newly introduced issues. Once the test confirms that it is correct, gradually redeploy the improved algorithm to the production environment and continue to monitor its performance.

[0194] Record the situation and resolution process of each failure for future reference, learn from each problem and continuously optimize the response strategy. Continuously adjust monitoring and threshold settings based on actual operation feedback to optimize the accuracy of fault detection and response speed.

[0195] Example 2

[0196] The difference between Example 2 of the present invention and Example 1 is that this example introduces an autonomous positioning system for a coal mining machine based on an inertial navigation system.

[0197] Figure 2A structural schematic diagram of a coal mining machine autonomous positioning system based on an inertial navigation system of the present invention is given, which includes a mine model construction module, a path calculation optimization module, a path dependence analysis module, a decision change analysis module, a failure mode judgment module and a failure mode decision module.

[0198] Mine model construction module: Use laser scanning and inertial navigation data fusion technology to capture the spatial changes inside the mine in real time, build and update the three-dimensional model of the mine in real time.

[0199] Path calculation optimization module: Build a graph search algorithm that adapts to narrow spaces, introduce mine-specific spatial constraints into the graph search algorithm, and optimize the path calculation of the coal mining machine.

[0200] Path dependency analysis module: repeatedly run the graph search algorithm in a simulation environment to adjust environmental variables, use graph edit distance to compare the path similarities of different results, apply principal component analysis and cluster analysis to extract key features, and generate a path dependency analysis index.

[0201] Decision change analysis module: During the operation of the graph search algorithm, the time of each decision point is recorded, the change rate of decision time is calculated using the variational method, and the volatility model is combined with the time series prediction model to generate the dynamic change rate of decision time.

[0202] Failure mode judgment module: Based on the path dependency analysis index and the dynamic change rate of decision time, it is judged whether the graph search algorithm has potential failure modes.

[0203] Failure mode decision module: When a graph search algorithm has a potential failure mode, it immediately triggers the graph search algorithm rollback and adjustment process.

[0204] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters and thresholds in the formula are set by technicians in this field according to actual conditions.

[0205] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium, or may be transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from one website, computer, server or data center to another website, computer, server or data center by wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server or data center that contains one or more available media sets. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium may be a solid-state hard disk.

[0206] Those of ordinary skill in the art will appreciate that the modules and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0207] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and modules described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0208] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the modules is only a logical function division. There may be other division methods in actual implementation, such as 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 interfaces, indirect coupling or communication connection of devices or modules, which can be electrical, mechanical or other forms.

[0209] 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, and 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.

[0210] In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0211] If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage media include: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, and other media that can store program codes.

[0212] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0213] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A coal mining machine path planning method based on an inertial navigation system, characterized in that: The steps include: S1: Use laser scanning and inertial navigation data fusion technology to capture the spatial changes inside the mine in real time, build and update the three-dimensional model of the mine in real time; S2: Build a graph search algorithm that adapts to narrow spaces, introduce the spatial constraints unique to mines into the graph search algorithm, and optimize the path calculation of the coal mining machine; S3: Repeatedly run the graph search algorithm in the simulation environment to adjust the environmental variables, use the graph edit distance to compare the path similarity of different results, apply principal component analysis and cluster analysis to extract key features, and generate a path dependency analysis index; S4: During the operation of the graph search algorithm, the time of each decision point is recorded, the change rate of the decision time is calculated using the variational method, and the volatility model is combined with the time series prediction model to generate the dynamic change rate of the decision time; S5: Based on the path dependency analysis index and the dynamic change rate of decision time, determine whether the graph search algorithm has potential failure modes; S6: When the graph search algorithm has a potential failure mode, immediately trigger the graph search algorithm rollback and adjustment process; S1 specifically includes: S101: Install laser scanning equipment on the coal mining machine; S102: Synchronize the inertial navigation system with the laser scanning device; S103: Apply Kalman filter algorithm to fuse laser scanning data and inertial navigation data in real time; S104: constructing a real-time three-dimensional model of the mine on a computing platform using the fused data; S105: Regularly update the three-dimensional model according to new scanning data.

2. A coal mining machine path planning method based on an inertial navigation system according to claim 1, characterized in that: S2 specifically includes: S201: Analyze the geometry of the mine and identify mine-specific spatial constraints: The three-dimensional data of the interior of the mine is obtained through laser scanning technology; the three-dimensional point cloud processing algorithm is used to extract planes and identify features of the collected point cloud data, filter noise and extract key geometric structures in the mine; the processing results of the point cloud data are verified using ground measured data; S202: Select a graph search algorithm suitable for narrow spaces: Based on the narrow space characteristics of the mine, the A* algorithm was selected and customized; The heuristic function expression is: h(n) is the value of the heuristic function, x goal ,y goal is the coordinate of the target node, x n ,y n is the coordinate of the current node n, w is the weight coefficient, and constraint(n) is the spatial constraint of the current node n; S203: Optimize the path of the coal mining machine by dynamic weight adjustment in the narrow space of the mine: Evaluate the surrounding environment of the coal mining machine in real time and adjust the weight value. The formula for dynamic adjustment is: w=w0+Δw(t); w0 is the initial weight, Δw(t) is the weight adjustment at time t; S204: Perform algorithm testing by simulating a mine environment, and continuously iterate and optimize path calculation performance based on feedback data.

3. A coal mining machine path planning method based on an inertial navigation system according to claim 2, characterized in that: S3 is specifically: After each environment adjustment, run the graph search algorithm and record the resulting paths; The path generated by each run is represented as a sequence of nodes, each node corresponding to a navigation point of the coal mining machine; Graph edit distance calculation: d(P1, P2) = min(Insertions+Deletions+Substitutions); where P1 and P2 represent the paths generated by the two runs respectively, and d(P1, P2) represents the graph edit distance; All path similarity data are used as input to perform principal component analysis to extract the most important change patterns in the path selection process; Construct a path similarity matrix, reduce the high-dimensional data to principal components through principal component analysis, and identify the main change directions in path selection; Use K-means clustering algorithm to classify the principal component features corresponding to different paths; Input the data after principal component analysis and set the initial number of cluster centers; apply the K-means algorithm to iterate until the cluster centers are stable; Based on the principal components extracted by principal component analysis and the clustering results, the path dependency analysis index is calculated and is defined as: Among them, D i is the path dependency analysis index; f is the number of cluster centers, b j is the number of paths contained in the jth cluster, N is the total number of paths, m is the number of principal components considered in each cluster, Var(PC j，a ) is the variance of the ath principal component in the jth cluster.

4. A coal mining machine path planning method based on an inertial navigation system according to claim 3, characterized in that: S4 is specifically: Collect and record the timestamp s of each decision point during the execution of the graph search algorithm i , where i represents the index of the decision point; The timestamp data of all decision points are stored in an array S = {s1, s2, ..., s y }, where y is the total number of decision points; Calculate the rate of change of decision time by using the variational method: Among them, Δτ i is the time rate of change of the ith decision point; s i+1 and i are the timestamps of adjacent decision points; Δs is the unit time interval; Calculate the time change rate Δτ between adjacent decision points by using the variational method i ; Apply the volatility model: in, represents the volatility of the decision time change rate, Δτ i is the ith time rate of change, is the average of all rates of change; Applying time series forecasting models: in, is the time rate of change of the predicted next decision point, α and β are regression coefficients, and τ i is the current time rate of change, ∈ i is the residual term; Generate decision time dynamic rate of change: Among them, δ τ is the dynamic rate of change of decision time; is the predicted time rate of change.

5. A coal mining machine path planning method based on an inertial navigation system according to claim 4, characterized in that: S5 is as follows: Setting D i The upper threshold D i，max ; Set D i The lower threshold D i，min ; Setting δ τ The upper threshold δ τ，max ; Calculate D for each test i and δ τ ; If D i >D i，max , D i <D i，min or δ τ >δ τ，max , then it is determined that the graph search algorithm has a potential failure mode; otherwise, it is determined that the graph search algorithm does not have a potential failure mode.

6. An autonomous positioning system for a coal mining machine based on an inertial navigation system, used to implement a path planning method for a coal mining machine based on an inertial navigation system as described in any one of claims 1 to 5, characterized in that: It includes a mine model building module, a path calculation optimization module, a path dependency analysis module, a decision change analysis module, a failure mode judgment module and a failure mode decision module; Mine model building module: uses laser scanning and inertial navigation data fusion technology to capture the spatial changes inside the mine in real time, build and update the three-dimensional model of the mine in real time; Path calculation optimization module: Build a graph search algorithm that adapts to narrow spaces, introduce mine-specific spatial constraints into the graph search algorithm, and optimize the path calculation of the coal mining machine; Path dependency analysis module: repeatedly run the graph search algorithm in a simulation environment to adjust environmental variables, use graph edit distance to compare the path similarities of different results, apply principal component analysis and cluster analysis to extract key features, and generate a path dependency analysis index; Decision change analysis module: records the time of each decision point during the operation of the graph search algorithm, calculates the change rate of decision time using the variational method, and applies the volatility model combined with the time series prediction model to generate the dynamic change rate of decision time; Failure mode judgment module: Based on the path dependency analysis index and the dynamic change rate of decision time, it is judged whether the graph search algorithm has potential failure modes; Failure mode decision module: When a graph search algorithm has a potential failure mode, it immediately triggers the graph search algorithm rollback and adjustment process.

Images