Mine master-slave machine inspection path planning method based on space-time modeling

Abstract

The invention discloses a space-time modeling-based mine primary and secondary aircraft inspection path planning method. According to the method, aiming at the actual problem that inspection paths need to be frequently adjusted due to staging advance of mining and dynamic evolution of geological conditions, smooth switching of the inspection paths in different mining stages is realized by constructing a staging space-time model and detecting transition signals and utilizing interpolation fusion and an A * algorithm; meanwhile, mining progress and geological environment change data are collected in real time, response requirements are judged through differential analysis, the model is dynamically updated through a genetic algorithm, and a path is re-planned; further extracting targeted optimization elements in the path, and generating an optimized master-slave collaborative inspection scheme through weighted fusion of continuous inspection requirements; finally, after continuity and pertinence verification, execution parameters are continuously adjusted in combination with real-time feedback and reinforcement learning, a safe and efficient space-time stratified dynamic inspection path with high adaptability and complete coverage is formed, and the operation efficiency of mine unmanned aerial vehicle inspection and the system reliability are remarkably improved.

Description

Technical Field

[0001] This invention belongs to the field of intelligent inspection technology, and in particular relates to a method for planning inspection paths of mine mother and daughter machines based on spatiotemporal modeling. Background Technology

[0002] As mining operations move towards large-scale and intelligent development, the demand for automation and precision in inspection operations is becoming increasingly prominent. The mother-daughter drone collaborative inspection system, as a novel operational mode, integrates the wide-range cruising capabilities of the mother drone with the flexible and precise detection advantages of the daughter drone. However, its inspection path planning faces multiple challenges, including dynamic changes in the mining environment, phased advancement of mining stages, and constraints related to the collaboration of heterogeneous drones. Existing path planning methods are mostly based on static or pre-set environmental models, making it difficult to adapt to the spatiotemporal dynamic characteristics such as the shifting mining boundaries and evolving geological conditions. This often leads to breaks in the inspection path during stage transitions or rapid failure after environmental disturbances, failing to guarantee the continuity, integrity, and real-time responsiveness of the inspection.

[0003] Current technologies primarily rely on classical graph search algorithms or offline optimization models. While these can find locally optimal paths in fixed scenarios, they generally lack the ability to uniformly model spatiotemporal evolution processes. Specifically, existing solutions struggle to organically integrate the spatial topology and temporal accessibility attributes of different mining phases, and cannot quickly transform real-time collected environmental change data (such as mining progress and rock displacement) into a basis for path adjustment. Furthermore, for the unique challenges of range heterogeneity, task priority differences, and collaborative node placement inherent in mother-daughter aircraft systems, existing methods often employ simplified unified planning strategies. This fails to achieve a refined distinction and efficient connection between the high-frequency operating areas of the daughter aircraft and the stable flight areas of the mother aircraft, resulting in limited overall inspection efficiency and uneven resource allocation.

[0004] In summary, mine inspection path planning urgently requires an innovative method that can deeply integrate spatiotemporal dynamic information, support smooth stage transitions, respond to environmental disturbances in real time, and finely coordinate heterogeneous UAV operations. This method needs to construct a dynamically evolving spatiotemporal model as the planning foundation, combine it with intelligent optimization algorithms to achieve online path adjustment and optimization, and ultimately form a highly adaptable, comprehensive, safe, and efficient intelligent inspection solution. This will overcome current technical bottlenecks and improve the overall intelligence level and operational reliability of mine inspection systems. Summary of the Invention

[0005] This invention proposes a mine machine inspection path planning method based on spatiotemporal modeling to solve the problems existing in the prior art.

[0006] To achieve the above objectives, this invention provides a method for mine mother-daughter machine inspection path planning based on spatiotemporal modeling, the method comprising:

[0007] Acquire spatial range data of the current mining phase, extract corresponding boundary coordinates and time cycle information from the spatiotemporal database, construct a phased spatiotemporal model, decompose the space into multiple sub-regions through grid partitioning and label the time accessibility attributes;

[0008] Based on the phased spatiotemporal model, phased transition signals are detected. If a transition signal is detected, the spatial range data of the next phase is obtained and fused with the current phase through an interpolation calculation method to determine the spatiotemporal conversion mechanism.

[0009] Transition path parameters are extracted from the spatiotemporal transformation mechanism, and the A* algorithm is used for optimization calculation to generate transition inspection paths.

[0010] Real-time data on changes in the mining environment are collected, a spatiotemporal trajectory model of the changes is constructed, and the model is compared with the current stage spatiotemporal model using differential analysis to determine the response requirements to environmental changes.

[0011] If there is a need to respond to environmental changes, the spatial topology and temporal reachability attributes of the phased spatiotemporal model are updated, and the path is recalculated using a genetic algorithm to obtain a dynamically adjusted inspection path.

[0012] Targeted optimization elements are extracted from the dynamically adjusted inspection path, and continuous inspection requirements are integrated through a weighted summation method to generate an optimized mother-daughter machine inspection path scheme.

[0013] The optimized scheme is continuously and specifically verified. After the verification is successful, it is imported into the mother-daughter control system. The execution parameters are adjusted cyclically in combination with real-time feedback to obtain the final spatiotemporal hierarchical planning path.

[0014] Optionally, the construction of the phased spatiotemporal model includes:

[0015] Retrieve vector polygon boundary coordinate sequences and time period span data from spatiotemporal databases to generate phased basic data objects;

[0016] Based on the phased basic data object, a three-dimensional spatial bounding box is constructed and discretized into an initialized set of spatial meshes;

[0017] Perform geometric inclusion detection on the spatial grid set to determine the set of effective mining sub-regions located within the boundary coordinate sequence;

[0018] Calculate the time accessibility attribute value of the effective mining sub-region set based on the time period span data, and generate a phased spatiotemporal model.

[0019] Optionally, the determination of the spatiotemporal transformation mechanism includes:

[0020] The time series data is processed by a phased spatiotemporal model, and the presence of transition signals is analyzed using detection rules. If a transition signal is detected, the time point of the signal and related characteristics are recorded.

[0021] Extract the spatial range data of the current period and the next period from the spatiotemporal database;

[0022] Interpolation calculation methods are used to process the two sets of spatial range data to generate spatial transition data for the transition stage;

[0023] By combining the temporal characteristics of the transition signals, the spatiotemporal transformation path of the transition phase is constructed, and key nodes and changing trends are identified.

[0024] A time-space transition mechanism for the transition phase is generated based on the aforementioned time-space transition path.

[0025] Optionally, the generation of the transition inspection path includes:

[0026] Extract transformation rule data from the spatiotemporal transformation mechanism database, filter rules related to the inspection path, and obtain a transformation rule set;

[0027] Extract transition path parameters from the set of transformation rules to determine the set of key transition path parameters;

[0028] The A* algorithm is used to optimize the set of parameters for the critical transition path, resulting in an optimized combination of path parameters.

[0029] An initial inspection path scheme is constructed based on the combination of path parameters, and the path switching points are smoothed to generate a smooth switching path scheme.

[0030] By combining the path planning constraints, the smooth switching path scheme is locally adjusted to obtain the transition inspection path.

[0031] Optionally, the requirement to determine environmental changes and respond includes:

[0032] Real-time acquisition of mining progress data and geological condition evolution data to form the original environmental change sequence;

[0033] Construct a spatiotemporal trajectory model of environmental changes based on the original environmental change sequence;

[0034] The difference between the spatiotemporal trajectory model of environmental change and the current periodic spatiotemporal model is calculated using the difference analysis method to obtain the difference distribution sequence;

[0035] If there are differential points in the differential distribution sequence that exceed a preset threshold, then the corresponding spatiotemporal location is marked as a region of significant change;

[0036] Extract a subset of geological conditions and mining progress data from the significantly changed areas, and perform clustering processing using a time series clustering algorithm to obtain change pattern clusters;

[0037] The response requirement level is determined based on the degree of matching between the change pattern cluster and the phased spatiotemporal model.

[0038] Optionally, obtaining the dynamically adjusted inspection path includes:

[0039] The deviation between real-time data of external environmental parameters and pre-stored benchmark values ​​is detected, and an update signal is triggered if the deviation exceeds a preset threshold.

[0040] The latest environmental data is obtained based on the update signal and loaded into the phased spatiotemporal model;

[0041] Update the spatial topology and temporal reachability attributes in the phased spatiotemporal model;

[0042] A genetic algorithm is used to perform path optimization calculations on the updated phased spatiotemporal model to obtain dynamically adjusted inspection paths.

[0043] Optionally, the optimized inspection path scheme for the mother-daughter unit includes:

[0044] Features of discrete path segments are extracted from dynamically adjusted inspection paths to construct targeted optimization element vectors;

[0045] The continuity constraints and task priorities are quantified into a continuous inspection requirement matrix;

[0046] The targeted optimization element vector and the continuous inspection requirement matrix are weighted and summed to obtain a comprehensive fusion score;

[0047] The integrated scoring marks the high-frequency operating area of ​​the sub-machine and the stable navigation area of ​​the mother machine;

[0048] By combining the sub-aircraft range limit calculation with the collaborative node, the high-frequency operation area of ​​the sub-aircraft and the stable navigation area of ​​the mother aircraft are connected to generate an optimized inspection path scheme for the sub-aircraft and mother aircraft.

[0049] Optionally, the path to obtain the final spatiotemporal hierarchical planning path includes:

[0050] The optimized inspection path scheme for the mother and daughter units was continuously verified and specifically verified.

[0051] Import the verified path scheme into the host-daughter control system;

[0052] Based on real-time feedback data, the deviation of path execution parameters is extracted, and the execution parameters are adjusted using reinforcement learning algorithms.

[0053] The path is continuously optimized through feedback loops until the path deviation converges, thus determining the final spatiotemporal hierarchical planning path.

[0054] Compared with the prior art, the present invention has the following advantages and technical effects:

[0055] This invention achieves dual adaptation of mine inspection paths to smooth transitions between mining stages and real-time environmental evolution by constructing a dynamic, phased spatiotemporal model. A combination of the A* algorithm and genetic algorithm ensures path continuity during stage transitions and global optimization during dynamic adjustments. By integrating targeted optimization elements and continuous inspection requirements, the division of labor and coordination in the collaborative operation of the mother and daughter machines are refined, forming a spatiotemporally layered adaptive path scheme. Finally, through real-time feedback and reinforcement learning mechanisms, the inspection path is continuously optimized during execution, significantly improving the coverage integrity, operational safety, and overall system efficiency of mine inspections. Attached Figure Description

[0056] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0057] Figure 1 This is a flowchart of a method according to an embodiment of the present invention. Detailed Implementation

[0058] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0059] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0060] Example 1

[0061] like Figure 1 As shown, this embodiment provides a method for mine mother-daughter machine inspection path planning based on spatiotemporal modeling. The method includes:

[0062] Acquire spatial range data of the current mining phase, extract corresponding boundary coordinates and time cycle information from the spatiotemporal database, construct a phased spatiotemporal model, decompose the space into multiple sub-regions through grid partitioning and label the time accessibility attributes;

[0063] Based on the periodic spatiotemporal model, the periodic transition signal is detected. If the transition signal is detected, the spatial range data of the next period is obtained and fused with the current period through interpolation calculation method to determine the spatiotemporal transformation mechanism.

[0064] Transition path parameters are extracted from the spatiotemporal transformation mechanism, and the A* algorithm is used for optimization calculation to generate transition inspection paths.

[0065] Real-time data on changes in the mining environment are collected, a spatiotemporal trajectory model of the changes is constructed, and the model is compared with the current stage spatiotemporal model using differential analysis to determine the response requirements to environmental changes.

[0066] If there is a need to respond to environmental changes, the spatial topology and temporal reachability attributes of the phased spatiotemporal model are updated, and the path is recalculated using a genetic algorithm to obtain a dynamically adjusted inspection path.

[0067] Targeted optimization elements are extracted from the dynamically adjusted inspection path, and continuous inspection requirements are integrated through a weighted summation method to generate an optimized mother-daughter machine inspection path scheme.

[0068] The optimized scheme is continuously and specifically verified. After the verification is successful, it is imported into the master-slave control system. The execution parameters are adjusted cyclically in combination with real-time feedback to obtain the final spatiotemporal hierarchical planning path.

[0069] Specifically, the following steps are included:

[0070] S101. Obtain the spatial range data corresponding to the current mining phase, extract the boundary coordinates and time cycle information of the phase from the pre-established spatiotemporal database, construct an independent phased spatiotemporal model for the extracted spatial range data, decompose the spatial range into multiple sub-regions through the grid division method and label the time accessibility attributes to obtain the phased spatiotemporal model.

[0071] Specifically, the process involves retrieving vector polygon boundary coordinate sequences and time period span data from a spatiotemporal database to generate phased basic data objects; constructing a three-dimensional spatial bounding box based on the phased basic data objects and discretizing it into an initialized spatial mesh set; performing geometric inclusion detection on the initialized spatial mesh set to determine a set of effective mining sub-regions located within the range of the vector polygon boundary coordinate sequences; and calculating the time accessibility attribute value of the effective mining sub-regions based on the time period span data to generate a phased spatiotemporal model.

[0072] In this embodiment, retrieving the vector polygon boundary coordinate sequence and time period span data from the spatiotemporal database is the first step in generating the phased spatiotemporal model. The vector polygon boundary coordinate sequence is usually stored in the form of an ordered list of latitude and longitude points. For example, the boundary of a mining area is formed by 20 points closing clockwise. The time period span data records the effective start and end times of the polygon, such as from January 1, 2020 to December 31, 2025. After extraction, a phased basic data object can be formed, containing spatial boundaries and time attributes, which facilitates subsequent unified processing.

[0073] Specifically, when constructing a three-dimensional bounding box based on phased basic data objects, the minimum bounding rectangle is first calculated from the polygon boundary coordinates, and then a certain range is extended in the height direction to form an axis-aligned bounding box (AABB).

[0074] For example, if the minimum x-axis of the mining area polygon in the planar projection is 108.5, the maximum x-axis is 109.2, the minimum y-axis is 34.1, and the maximum y-axis is 34.8, then the bottom surface of the bounding box is this rectangle, and the height can be set from 0 meters at sea level to the maximum depth of the mining area -800 meters, thus obtaining a three-dimensional box. This bounding box can quickly determine the overall spatial range and effectively reduce the amount of subsequent calculations.

[0075] For example, when discretizing the 3D bounding box to obtain an initialized spatial mesh set, a regular voxel partitioning method can be used. The bounding box is divided into mesh units of 100 meters, 100 meters, and 50 meters in the x, y, and z directions, respectively, forming thousands of small cubic meshes. This discretization transforms the originally continuous space into an enumerable set of units, facilitating parallel processing and storage.

[0076] In this embodiment, a geometric inclusion test is performed on the initialized spatial grid set to determine the set of valid mining sub-regions located within the boundaries of the vector polygon. During the test, either the ray casting method or the number of turns method can be used to determine whether the center point of each grid falls within the polygon.

[0077] For example, after planar projection, if the center point of the grid is inside the boundary polygon of the mining area and the z-coordinate is within the effective depth range, then the grid is marked as a valid mining sub-region. This filtering process retains only the grids that truly belong to the mining area, greatly reducing interference from invalid regions and improving model accuracy.

[0078] For example, when calculating the time accessibility attribute value of an effective mining sub-region set based on time period span data, the time span can be converted into mining days or stage number.

[0079] Specifically, if the time span is 5 years, it can be evenly divided into 5 annual stages. Each valid grid inherits the accessibility value of 1 for that stage, indicating that mining is possible in that year; if a stage is suspended due to policy, the corresponding value is 0. This time accessibility attribute gives each grid dynamic characteristics, supporting subsequent mining progress simulation.

[0080] It should be noted that the final generated phased spatiotemporal model is a data structure composed of all effective mining sub-region grids and their temporal accessibility attributes. This model can intuitively express the spatial availability of a mining area in different time periods. For example, when querying the exploitable volume in 2023, it is only necessary to summarize the total volume of grids with an accessibility of 1 for that year. This phased spatiotemporal model significantly improves the accuracy of mine planning, supports applications such as dynamic assessment of resource reserves, optimization of mining sequences, and prediction of environmental impacts, effectively reduces the risk of blind mining, and improves resource utilization.

[0081] S102. Detect phase transition signals based on the obtained phased spatiotemporal model. If a phase transition signal is detected, obtain the spatial range data of the next phase from the spatiotemporal database. Merge the current phased spatiotemporal model and the spatial range data of the next phase using an interpolation calculation method to determine the spatiotemporal conversion mechanism of the transition phase.

[0082] Specifically, the input time-series data is processed using a phased spatiotemporal model. Pre-established detection rules are used to analyze the presence of transition signals. If a transition signal is detected, the time point of its occurrence and related characteristics are recorded, yielding a preliminary judgment result. Based on this preliminary judgment, spatial range data corresponding to the current phase and the next phase are extracted from the spatiotemporal database. The boundaries and coverage of the two sets of data are obtained, determining the spatial data set to be fused. For this set, interpolation methods are used to process the spatial range data of the current and next phases, generating spatial transition data for the transition stage and obtaining a smooth spatial range change result. By analyzing the spatial range change result and combining it with the time point characteristics of the transition signal, a spatiotemporal transition path for the transition stage is constructed, identifying key nodes and trends. Based on the key nodes of the spatiotemporal transition path, the dynamic change process from the current phase to the next phase is analyzed, generating a spatiotemporal transition mechanism for the transition stage and obtaining a complete description of the transition rules.

[0083] To process the input time-series data using a phased spatiotemporal model, it is first necessary to understand that time-series data here refers to monitoring sequences related to mining activities, such as daily or hourly equipment operating parameters, production records, or geological sensor data.

[0084] In this embodiment, the time-series data can be a continuously collected sequence of vibration signals from the mining area. The model scans these signals point by point to identify abnormal fluctuations. Pre-established detection rules are used to analyze whether transition signals exist. These detection rules are typically based on a combination of threshold and pattern matching.

[0085] Specifically, the rule can be set to determine a transitional signal when the signal amplitude exceeds 1.5 times the historical average within a short period of time and is accompanied by changes in frequency components.

[0086] In this embodiment, assuming the current phase is the 5th mining phase, a sudden and sustained high-frequency disturbance is detected in the signal at time t=180. The system immediately records this time point, along with the signal's peak value, duration, and other characteristics, forming a preliminary judgment result for the transition signal. This record provides a precise time anchor for subsequent fusion, effectively avoiding the computational burden caused by blindly processing the entire time series. Based on the preliminary judgment result of the transition signal, spatial range data for the current phase and the next phase are extracted from the spatiotemporal database, such as extracting the polygonal boundary coordinate sequence of the 5th phase and the corresponding sequence of the 6th phase.

[0087] It should be noted that the coverage areas of the two sets of data may overlap or extend. By comparing boundary points, the spatial data set to be merged can be quickly determined, typically including the sub-regions that are about to be phased out in the current phase and the sub-regions to be added in the next phase. This targeted extraction significantly improves the efficiency of data processing, focusing only on the changed parts rather than the entire mining area. For the spatial data set to be merged, an interpolation calculation method is used to generate spatial transition data for the transition phase.

[0088] For example, linear interpolation or inverse distance weighted interpolation is preferably used to gradually deform the boundary points in the time dimension, so that the boundary of the 5th period smoothly transitions to the boundary of the 6th period.

[0089] Specifically, if the coordinates of a vertex at the boundary of stage 5 are (1200, 850), and the corresponding coordinates for stage 6 expand to (1350, 920), then intermediate coordinates are generated daily over 30 days after the transition signal occurs, ultimately resulting in a continuous spatial range change. This smoothing process ensures that the model does not experience abrupt changes during stage switching, which is beneficial to the stability of subsequent dynamic simulations. By analyzing the spatial range change results and combining them with the temporal characteristics of the transition signal, a spatiotemporal transition path for the transition phase is constructed.

[0090] For example, in the above embodiment, starting from t=180 days, the key nodes of the path may include the boundary expansion starting point, the intermediate maximum deformation point, and the final stabilization point, with the change trend showing a gradual advancement from the inside out. This path construction clearly depicts the movement trajectory of the mining frontier, facilitating the prediction of potential risk areas. Based on the key nodes of the spatiotemporal transition path, the dynamic change process from the current phase to the next phase is analyzed, ultimately generating the spatiotemporal transition mechanism for the transition phase.

[0091] For example, the mechanism description could include "within 30 days of the transition signal being triggered, the eastern boundary will expand outward by 150 meters, accompanied by a gradual change in the grid time accessibility attribute from 0 to 1." This complete transition rule description not only provides precise guidance for production scheduling but also significantly improves the continuity of mining operations and resource utilization, avoiding potential equipment idleness or safety hazards that could result from phased hard switching.

[0092] S103. Extract transition path parameters from the determined spatiotemporal transformation mechanism, use the A algorithm to optimize the extracted transition path parameters, generate a smooth switching inspection path scheme, and obtain the transition inspection path.

[0093] Transition rule data is obtained from the spatiotemporal transition mechanism database. This data is initially screened to remove invalid rules that do not meet the inspection path requirements, resulting in a pre-screened set of transition rules. For this set, parameter data related to the transition path is extracted. The parameter data is then structured and analyzed to determine the correlation between parameters, identifying the key transition path parameter set. Based on this set, the A algorithm is used to optimize the parameters, generating a smooth transition path optimization result, resulting in the optimized path parameter combination. Using this optimized path parameter combination, an initial inspection path plan is constructed. If discontinuities exist at path switching points in the initial plan, these points are smoothed to determine if they meet the preset switching effect standard, resulting in a smooth transition path plan. For this smooth transition path plan, constraint data for path planning is obtained. If conflicts exist between the constraint data and the path plan, the path plan is locally adjusted to determine the final inspection path planning result. Based on the final inspection path planning result, complete transition inspection path data is generated. Data validation tools are used to check the completeness and consistency of the path data, resulting in the final executable transition inspection path.

[0094] For example, when retrieving conversion rule data from the spatiotemporal conversion mechanism database, preliminary screening can be performed based on the applicable scenarios of the rules and the basic requirements of the inspection path. Assuming the database stores 100 conversion rules, 30 of them may not match the spatiotemporal characteristics of the current inspection path, such as mismatched time range or spatial coverage. These rules will be removed, leaving 70 rules to form the preliminary set of conversion rules. This process ensures the relevance of subsequent analysis.

[0095] For example, when extracting parameter data related to the transition path from the initially screened set of transition rules, attention can be paid to information such as the coordinates and timestamps of the path's starting point, ending point, and intermediate switching points. Suppose a rule specifies transition path parameters from region A to region B, including a path length of 5 kilometers and a switching time of 10 minutes. By structuring this data and analyzing the correlations between parameters, such as the ratio of time to distance, the key transition path parameter set can be determined, such as the optimal combination of switching time and path length.

[0096] For example, when using Algorithm A to optimize parameters, key parameters can be input into the algorithm model to generate a smooth path optimization result. Assuming there are abrupt changes in the path switching points before optimization, after optimization, by adjusting the coordinates and time distribution of the switching points, a continuous combination of path parameters can be formed. For instance, the switching time can be smoothly adjusted from 10 minutes to 12 minutes, and the path length can be fine-tuned to 5.2 kilometers, ensuring a more natural switching process.

[0097] For example, after constructing the initial inspection path plan, if it is found that the switching points are not continuous, such as a certain path jumping directly from area A to area C, ignoring the intermediate area B, intermediate transition points can be inserted through smoothing to adjust the path trajectory and determine whether the switching effect standard is met, such as controlling the switching time error within 1 minute, and finally obtaining a smooth switching path plan.

[0098] For example, when obtaining constraint data for a smooth switching path plan, if the constraint includes that the inspection time must not exceed 30 minutes, and the current plan takes 35 minutes in total, then the path plan needs to be adjusted locally, such as shortening the inspection range of a certain non-critical path, and finally determining the inspection path planning result to ensure that it meets all constraints.

[0099] For example, after generating complete transition inspection path data, data validation tools can be used to check the completeness and consistency of the path data. If a segment of coordinate points is missing in the path data, the tool will automatically identify and prompt for completion, ensuring that the final executable transition inspection path is complete and guarantees the smooth implementation of the inspection task. These steps collectively improve the reliability of path planning.

[0100] S104. Obtain mining environment change data, collect mining progress and geological condition evolution information in real time from the spatiotemporal database, construct a spatiotemporal trajectory model of change based on the collected environmental change data, compare the difference between the spatiotemporal trajectory model of change and the current staged spatiotemporal model through differential analysis, and determine the environmental change response requirements.

[0101] Real-time data on mining progress and geological condition evolution are collected from a spatiotemporal database to obtain the original environmental change sequence. A spatiotemporal trajectory model of environmental change is constructed based on this sequence, resulting in a standardized spatiotemporal trajectory representation. Difference values ​​between the standardized spatiotemporal trajectory representation and the current staged spatiotemporal model are calculated using differential analysis to obtain a difference distribution sequence. If any difference points in the difference distribution sequence exceed a preset threshold, their corresponding spatiotemporal locations are marked as significant change regions, resulting in a significant change set. Subsets of geological condition evolution data and mining progress data are extracted from this significant change set to obtain associated change subsequences. A time series clustering algorithm is used to cluster these associated change subsequences, resulting in change pattern clusters. Based on the matching degree between the change pattern clusters and the staged spatiotemporal model, the response requirement level of each cluster is determined, yielding the response requirement classification results.

[0102] For example, in actual mining environments, real-time acquisition of mining progress data and geological condition evolution data from spatiotemporal databases is particularly important. Assuming an open-pit coal mine scenario, daily mining progress data, such as a daily extraction volume of 5000 tons, and geological condition data, such as changes in rock hardness and groundwater level fluctuations, are collected through sensors and monitoring equipment. This data forms the original environmental change sequence, laying the foundation for subsequent analysis. By processing this data, a spatiotemporal trajectory model of environmental change can be constructed. For instance, mining progress and geological condition changes can be plotted as dynamic trajectory maps along time and space dimensions, visually reflecting the changing trends of a certain area over the past 30 days.

[0103] For example, in constructing a standardized spatiotemporal trajectory representation, the collected data can be normalized, unifying the time unit to hours and the spatial unit to meters, ensuring the comparability of data from different sources. Then, using differential analysis, the difference between the standardized trajectory and the current phased spatiotemporal model is calculated. Suppose the current phased model predicts a rock hardness of level 5 in a certain area, while actual monitoring data shows level 7, resulting in a difference of 2. If a preset threshold of 1.5 is set, this area is marked as a significantly changed region. This method can quickly locate potential anomalies during mining, providing a basis for subsequent adjustments.

[0104] For example, after extracting the set of significant changes, the processing of subsets of geological condition evolution data and mining progress data can focus on data within areas of significant change. For instance, in a marked area, geological conditions might show increased rock fractures, while mining progress might show a decrease in daily production to 3000 tons, below average. By extracting associated change subsequences, a potential link between increased fractures and decreased mining efficiency can be discovered. Processing these subsequences using time series clustering algorithms may yield three change pattern clusters: rapidly decreasing clusters, slowly changing clusters, and stable clusters, each corresponding to different response requirement levels.

[0105] For example, in analyzing the matching degree between change pattern clusters and phased spatiotemporal models, assuming that the matching degree of rapidly declining clusters is only 30%, far below the preset standard of 70%, they are classified as high response requirement level. This means that mining strategies need to be adjusted first in this area, such as increasing equipment investment or optimizing the mining sequence. For stable clusters with higher matching degrees, the existing plan can be maintained to reduce resource waste. This classification method helps to rationally allocate resources and improve overall mining efficiency.

[0106] For example, in practical implementation, the response requirement classification results can further guide on-site management. For instance, in areas with high response requirements, a professional team can be assigned to conduct real-time monitoring, analyzing the causes of changes based on historical data to ensure mining safety. In areas with low response requirements, the monitoring frequency can be reduced, saving costs. This tiered response approach not only improves the targeting of management but also effectively reduces potential risks and ensures the stability of the mining process.

[0107] S105. If the environmental change response requirements are determined, the spatial topology and temporal reachability attributes of the phased spatiotemporal model are updated. A genetic algorithm is used to recalculate the path of the updated phased spatiotemporal model to obtain the dynamically adjusted inspection path.

[0108] Specifically, the deviation between real-time data of external environmental parameters and pre-stored baseline values ​​is detected. If the deviation exceeds a preset threshold, an environmental change is identified, triggering an update signal. Based on the update signal, the latest environmental data is acquired and loaded into the phased spatiotemporal model. The spatial topology in the phased spatiotemporal model is updated, and the connection status between nodes is re-established. Based on the updated spatial topology, the temporal reachability attributes between each node are recalculated. A genetic algorithm is used to perform path optimization calculations on the phased spatiotemporal model, which includes the latest spatial topology and temporal reachability attributes, to obtain the adjusted inspection path.

[0109] For example, in the field of mine environmental monitoring, deviation analysis of real-time data detection of external environmental parameters from pre-stored baseline values ​​can be performed by using sensor networks to collect parameters such as temperature, humidity, and gas concentration in real time and compare them with historical baseline values. Assuming the baseline value is 25 degrees Celsius and the real-time data is 30 degrees Celsius, the deviation is 5 degrees Celsius. If the preset threshold is 3 degrees Celsius, this is considered an environmental change, triggering an update signal. This method can promptly capture environmental anomalies, ensuring timely adjustments to the subsequent model.

[0110] For example, regarding the step of acquiring the latest environmental data and loading it into the phased spatiotemporal model after triggering an update signal, sensor nodes deployed in different areas of the mine can collect the latest geological subsidence and air quality data and map them into the phased spatiotemporal model. Suppose that the subsidence data in a certain area increases from 0.2 meters to 0.5 meters, the system will update this change in the model. This data loading method helps maintain the synchronization between the model and the actual environment.

[0111] For example, when updating the spatial topology in the phased spatiotemporal model, the spatial connectivity between monitoring points within the mine can be re-analyzed. If a passage is blocked due to geological changes, the system will update the topology, mark the passage as inaccessible, and replan the connection paths between monitoring points. This update mechanism reflects the true spatial structure of the mining environment, providing a reliable basis for subsequent path planning.

[0112] For example, in the process of recalculating the time-accessibility attributes between nodes, the time required to travel from monitoring point A to monitoring point B can be estimated based on the updated topology. Suppose the original path takes 10 minutes, but due to a passage blockage requiring a detour, the time increases to 15 minutes; the system will record this change. This calculation method provides time-dimensional data support for inspection path optimization.

[0113] For example, when using a genetic algorithm to perform path optimization calculations on a phased spatiotemporal model, multiple inspection paths can be simulated, and the shortest or safest path can be selected iteratively. Assuming there are 5 monitoring points in the mine, the system uses the algorithm to select a path with a total time of 40 minutes, saving 10 minutes compared to the original path. This optimization method can improve inspection efficiency while reducing personnel exposure time in high-risk areas.

[0114] For example, regarding the generation of adjusted inspection routes, the route information can be distributed to inspection equipment or personnel based on the optimization results. Assuming the new route avoids areas with subsidence risk, the system will generate an inspection sequence from point A through point C to point B, and mark any risk warnings. This route adjustment method ensures the safety of inspection work while guaranteeing coverage of all critical monitoring areas. Through these multifaceted implementation methods, the importance of the complete process from environmental parameter detection to route optimization in mine environmental management can be seen. The detailed processing of each step provides strong support for improving monitoring accuracy and response efficiency, while also ensuring the safety of mine operations.

[0115] S106. Extract targeted optimization elements from the dynamically adjusted inspection path, and integrate the continuous inspection requirements and targeted optimization elements through a weighted summation method to generate an optimized mother-daughter machine inspection path scheme.

[0116] Specifically, discrete path segments in the dynamically adjusted inspection path are obtained, and features of the discrete path segments are extracted to construct a targeted optimization element vector. Continuity constraints and task priorities are transformed into a continuous inspection requirement quantification matrix. The targeted optimization element vector and the continuous inspection requirement quantification matrix are weighted and summed to obtain a comprehensive fusion score. The high-frequency operating area of ​​the slave aircraft and the stable navigation area of ​​the mother aircraft are marked according to the comprehensive fusion score. Coordination nodes are calculated in conjunction with the slave aircraft's range limitations to connect the high-frequency operating area of ​​the slave aircraft and the stable navigation area of ​​the mother aircraft, generating an optimized slave-mother aircraft inspection path scheme.

[0117] For example, in scenarios involving dynamically adjusted inspection paths, identifying discrete path segments is a fundamental step in path optimization. Discrete path segments refer to discontinuous parts of the inspection path that are broken down due to environmental changes or task requirements. Suppose that in an industrial park, the inspection path is divided into multiple independent segments due to temporary obstacles or equipment malfunctions. It is necessary to identify the location and length of these segments; for example, one segment might be 200 meters long, and another 150 meters. By analyzing the spatial distribution and connectivity characteristics of these segments, features such as the curvature of the path segments and obstacle density can be extracted to construct targeted optimization feature vectors for subsequent path optimization.

[0118] For example, the conversion between continuity constraints and task priorities can be quantified into a matrix to reflect the urgency of inspection tasks and the need for path connectivity. Assuming there are three task points with high, medium, and low priorities, their corresponding quantification values ​​can be set to 3, 2, and 1. Continuity constraints are weighted according to the distance and difficulty of traversing path segments; for example, a closer path might have a weight of 0.8, while a farther path might have a weight of 0.2. By weighted summation, the targeted optimization element vector is combined with the continuity inspection requirement quantification matrix to obtain a comprehensive score used to evaluate the feasibility of the path. This scoring mechanism helps the system quickly determine which path segments need to be prioritized and which can be postponed.

[0119] For example, in the process of marking high-frequency operation areas for slave aircraft and stable navigation areas for mother aircraft, comprehensive fusion scoring plays a crucial role. Suppose the scoring indicates that a certain area has dense tasks and complex paths, suitable for frequent slave aircraft operations, then it is marked as a high-frequency operation area, such as a densely populated equipment area within an industrial park; while areas with lower scores and flatter paths are marked as stable navigation areas for mother aircraft, such as main roads on the outskirts of the park. This division helps to rationally allocate inspection tasks between slave and mother aircraft, improving overall efficiency.

[0120] For example, calculating the coordination node based on the sub-aircraft's range limit is to connect the two areas mentioned above. Assuming the sub-aircraft's single-trip range is limited to 5 kilometers, and the distance between the high-frequency operation area and the stable flight area is 4 kilometers, a coordination node can be set up between the two areas as a transit / resupply point or mission switching point for the sub-aircraft. By planning the location of the coordination node, such as selecting an intersection near the midpoint between the two areas, an optimized inspection route plan for both the sub-aircraft and the mother aircraft can be generated, ensuring that the sub-aircraft completes its mission within its range while the mother aircraft operates efficiently in the stable area. This approach improves the coordination and coverage of inspections.

[0121] For example, in practical applications, feature extraction for discrete path segments can be combined with dynamic environmental data, such as real-time wind speed or ground slipperiness, to further refine and optimize the feature vectors. For instance, if higher wind speeds increase the difficulty of traversing a certain path, the system will lower the priority of that path segment and replan a safer route. This flexible adjustment ensures the smooth execution of inspection tasks while reducing the risk of equipment damage. Overall, the above methods provide comprehensive support for the optimization of dynamic inspection paths through multi-dimensional analysis and quantitative methods.

[0122] S107. Verify the continuity and relevance of the optimized mother-daughter unit inspection path scheme. If the verification is successful, import the scheme into the mother-daughter unit control system and adjust the path execution parameters in real time through feedback to obtain the final spatiotemporal hierarchical planning path.

[0123] Specifically, the process involves obtaining an optimized inspection path scheme for the mother-daughter unit, performing continuity verification on the path scheme to determine if there are any interruptions between path nodes, marking the path scheme as invalid if interruptions exist, and obtaining a continuity verification result. From the continuity verification result, a valid path scheme is obtained, and a targeted verification is performed on the valid path scheme to determine if the path coverage matches the preset inspection area. If they match, the targeted verification is considered successful, and a targeted verification result is obtained. From the targeted verification result, the verified path scheme is obtained and imported into the mother-daughter unit control system, obtaining the path data of the imported control system. Path execution is initiated from the imported control system path data, and real-time feedback data during the execution process is obtained, resulting in a real-time feedback sequence. Path execution parameter deviations are extracted from the real-time feedback sequence, and a reinforcement learning algorithm is used to adjust the path execution parameters. It is determined whether the deviation exceeds a preset threshold; if it does, the execution parameters are updated, resulting in the adjusted execution parameters. A feedback loop is applied to the adjusted execution parameters, continuously obtaining updated real-time feedback data, and determining whether the path deviation converges in the feedback loop. If convergence occurs, a spatiotemporal hierarchical planning path is determined, resulting in the final spatiotemporal hierarchical planning path. The final spatiotemporal hierarchical planning path is output to the control system. The execution data of the hierarchical planning path in multiple time periods is obtained. The consistency of the hierarchical structure is judged by rules. If the consistency is consistent, the spatiotemporal hierarchical planning path is locked, and the locked planning path is obtained.

[0124] For example, continuity verification is a crucial step in optimizing the inspection path for a mother-daughter unit. The purpose of continuity verification is to ensure that there are no interruptions between path nodes, thus guaranteeing the integrity of the inspection task. Suppose that in an inspection scenario in an industrial park, the mother-daughter unit needs to cover multiple equipment locations. If two nodes in the path planning cannot be directly connected due to obstacles or excessive distance, the system will mark the path plan as invalid. This verification method can prevent task interruptions during the inspection process, thereby improving the overall reliability of the inspection.

[0125] For example, targeted verification focuses on whether the route coverage conforms to the preset inspection area. Suppose there are 10 critical equipment locations in an industrial park, and the preset inspection area requires coverage of all locations. If a route only covers 8 locations, the system will determine that the targeted verification fails. This verification method ensures the comprehensiveness of the inspection task and avoids missing critical areas.

[0126] For example, after the path planning is implemented in the mother-daughter air transport system, the acquisition of real-time feedback data becomes particularly important. Imagine that when the mother-daughter air transport system is performing an inspection task, the system provides position and status data every 5 minutes, forming a real-time feedback sequence. By analyzing this data, it can be discovered that the daughter air transport system moves too slowly in certain areas, possibly because the path planning did not consider terrain factors. This real-time feedback mechanism helps to identify problems promptly and make adjustments.

[0127] For example, the application of reinforcement learning algorithms to adjust path execution parameter deviations is an innovative aspect. Suppose the actual flight path of the sub-machine deviates from the planned path by 10%, exceeding a preset threshold of 5%, the system will adjust the sub-machine's speed or path direction using a learning algorithm, generating adjusted execution parameters. This dynamic adjustment method can effectively cope with unexpected situations during inspections and improve the stability of mission execution.

[0128] For example, the continuous application of feedback loops and the determination of path deviation convergence are core to ensuring the feasibility of spatiotemporal hierarchical planning paths. Assuming that after multiple adjustments, the path deviation gradually decreases from an initial 10% to 2%, the system determines that the deviation has converged and ultimately determines the spatiotemporal hierarchical planning path. This approach ensures the consistency of path planning across multiple time periods.

[0129] For example, multi-time period execution data analysis and consistency judgment of hierarchical path planning are crucial steps in determining the final path. Suppose the system tests path execution data in three different time periods and finds that the hierarchical structure remains consistent—for example, the slave unit is responsible for dense area inspections while the mother unit is responsible for long-distance navigation—then the planned path is ultimately determined. This consistency verification ensures the reliability of the path scheme in practical applications, providing stable support for long-term inspection tasks.

[0130] The above are merely preferred embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for planning inspection paths of mine mother-daughter machines based on spatiotemporal modeling, characterized in that, Includes the following steps: Acquire spatial range data of the current mining phase, extract corresponding boundary coordinates and time cycle information from the spatiotemporal database, construct a phased spatiotemporal model, decompose the space into multiple sub-regions through grid partitioning and label the time accessibility attributes; Based on the phased spatiotemporal model, phased transition signals are detected. If a transition signal is detected, the spatial range data of the next phase is obtained and fused with the current phase through an interpolation calculation method to determine the spatiotemporal conversion mechanism. Transition path parameters are extracted from the spatiotemporal transformation mechanism, and the A* algorithm is used for optimization calculation to generate transition inspection paths. Real-time data on changes in the mining environment are collected, a spatiotemporal trajectory model of the changes is constructed, and the model is compared with the current stage spatiotemporal model using differential analysis to determine the response requirements to environmental changes. If there is a need to respond to environmental changes, the spatial topology and temporal reachability attributes of the phased spatiotemporal model are updated, and the path is recalculated using a genetic algorithm to obtain a dynamically adjusted inspection path. Targeted optimization elements are extracted from the dynamically adjusted inspection path, and continuous inspection requirements are integrated through a weighted summation method to generate an optimized mother-daughter machine inspection path scheme. The optimized scheme is continuously and specifically verified. After the verification is successful, it is imported into the master-slave control system. The execution parameters are adjusted cyclically in combination with real-time feedback to obtain the final spatiotemporal hierarchical planning path.

2. The method according to claim 1, characterized in that, The construction of the phased spatiotemporal model includes: Retrieve vector polygon boundary coordinate sequences and time period span data from spatiotemporal databases to generate phased basic data objects; Based on the phased basic data object, a three-dimensional spatial bounding box is constructed and discretized into an initialized set of spatial meshes; Perform geometric inclusion detection on the spatial grid set to determine the set of effective mining sub-regions located within the boundary coordinate sequence; Calculate the time accessibility attribute value of the effective mining sub-region set based on the time period span data, and generate a phased spatiotemporal model.

3. The method according to claim 1, characterized in that, The determined spatiotemporal transformation mechanism includes: The time series data is processed by a phased spatiotemporal model, and the presence of transition signals is analyzed using detection rules. If a transition signal is detected, the time point of the signal and related characteristics are recorded. Extract the spatial range data of the current period and the next period from the spatiotemporal database; Interpolation calculation methods are used to process the two sets of spatial range data to generate spatial transition data for the transition stage; By combining the temporal characteristics of the transition signals, the spatiotemporal transformation path of the transition phase is constructed, and key nodes and changing trends are identified. A time-space transition mechanism for the transition phase is generated based on the aforementioned time-space transition path.

4. The method according to claim 1, characterized in that, The generated transition inspection path includes: Extract transformation rule data from the spatiotemporal transformation mechanism database, filter rules related to the inspection path, and obtain a transformation rule set; Extract transition path parameters from the set of transformation rules to determine the set of key transition path parameters; The A* algorithm is used to optimize the set of parameters for the critical transition path, resulting in an optimized combination of path parameters. An initial inspection path scheme is constructed based on the combination of path parameters, and the path switching points are smoothed to generate a smooth switching path scheme. By combining the path planning constraints, the smooth switching path scheme is locally adjusted to obtain the transition inspection path.

5. The method according to claim 1, characterized in that, The requirements for judging and responding to environmental changes include: Real-time acquisition of mining progress data and geological condition evolution data to form the original environmental change sequence; Construct a spatiotemporal trajectory model of environmental changes based on the original environmental change sequence; The difference between the spatiotemporal trajectory model of environmental change and the current periodic spatiotemporal model is calculated using the difference analysis method to obtain the difference distribution sequence; If there are differential points in the differential distribution sequence that exceed a preset threshold, then the corresponding spatiotemporal location is marked as a region of significant change; Extract a subset of geological conditions and mining progress data from the significantly changed areas, and perform clustering using a time series clustering algorithm to obtain change pattern clusters; The response requirement level is determined based on the degree of matching between the change pattern cluster and the phased spatiotemporal model.

6. The method according to claim 1, characterized in that, The dynamically adjusted inspection path includes: The deviation between real-time data of external environmental parameters and pre-stored benchmark values ​​is detected, and an update signal is triggered if the deviation exceeds a preset threshold. The latest environmental data is obtained based on the update signal and loaded into the phased spatiotemporal model; Update the spatial topology and temporal reachability attributes in the phased spatiotemporal model; A genetic algorithm is used to perform path optimization calculations on the updated phased spatiotemporal model to obtain dynamically adjusted inspection paths.

7. The method according to claim 1, characterized in that, The optimized inspection path scheme for the mother-daughter unit includes: Features of discrete path segments are extracted from dynamically adjusted inspection paths to construct targeted optimization element vectors; The continuity constraints and task priorities are quantified into a continuous inspection requirement matrix; The targeted optimization element vector and the continuous inspection requirement matrix are weighted and summed to obtain a comprehensive fusion score; The integrated scoring marks the high-frequency operating area of ​​the sub-machine and the stable navigation area of ​​the mother machine; By combining the sub-aircraft range limit calculation with the collaborative node, the high-frequency operation area of ​​the sub-aircraft and the stable navigation area of ​​the mother aircraft are connected to generate an optimized inspection path scheme for the sub-aircraft and mother aircraft.

8. The method according to claim 1, characterized in that, The final spatiotemporal hierarchical planning path includes: The optimized inspection path scheme for the mother and daughter units was continuously verified and specifically verified. Import the verified path scheme into the host-daughter control system; Based on real-time feedback data, the deviation of path execution parameters is extracted, and the execution parameters are adjusted using reinforcement learning algorithms. The path is continuously optimized through feedback loops until the path deviation converges, thus determining the final spatiotemporal hierarchical planning path.

Images