Mine intelligent laser sensing multi-parameter fusion monitoring platform

Abstract

The application discloses a mine intelligent laser sensing multi-parameter fusion monitoring platform and relates to the technical field of multi-information fusion. The application comprises the following steps: calculating the change trend and the correlation structure between the explicit monitoring parameters, deducing the implicit monitoring parameters which cannot be directly measured, and forming an extended parameter set together with the explicit monitoring parameters; dividing the extended parameter set into multiple conflict type groups according to the correlation consistency inside the extended parameter set; respectively executing corresponding fusion strategies on different conflict type groups to generate an initial fusion result, and constructing a missing degree atlas based on the residual conflicts exposed in the fusion process; and generating targeted compensation collection instructions according to the missing degree distribution based on the missing degree atlas to obtain compensation collection data. The application re-quantizes the remaining inconsistencies after the initial fusion, constructs a missing degree atlas, and upgrades the monitoring from passive fusion to an explainable, compensable and closed-loop optimized high-reliability fusion mode.

Description

Technical Field

[0001] This invention relates to the field of multi-information fusion technology, and in particular to an intelligent laser sensing multi-parameter fusion monitoring platform for mines. Background Technology

[0002] With the development of intelligent mining, multi-source sensing devices such as lidar, fiber optic sensors, deformation monitoring, and environmental parameter sensing are widely deployed underground, enabling monitoring systems to simultaneously acquire multimodal data and fuse them on the same platform. Existing technologies typically employ synchronous calibration, model compensation, and weighted fusion to address deviations in the fusion process of heterogeneous multi-source data, achieving a certain degree of consistency in the fusion results. However, in complex underground environments, factors such as sensor obstruction, dust disturbance at the working face, and abrupt changes in reflection conditions can still lead to unresolved data conflicts after fusion, such as temporal mismatches, abnormal spatial distributions, or distorted dependency chains.

[0003] While existing technologies can resolve relatively obvious conflicts in multi-source sensor data through preliminary data fusion processing, they often struggle to effectively identify and address persistent, subtle data inconsistencies after the initial fusion. Specifically: in the temporal dimension, microsecond-level timestamp misalignments may exist between different monitoring parameters; while these minor differences don't affect overall trend judgment, they weaken the accuracy of correlation analysis. Spatially, abrupt changes in data gradients may occur in certain local areas, disrupting spatial continuity even before reaching alarm thresholds. Regarding parameter logic, monitoring indicators that should maintain stable causal relationships may exhibit imperceptible transmission lags or response distortions. More critically, existing technologies lack a systematic assessment mechanism for these residual conflicts, failing to accurately quantify the severity of various data gaps or precisely locate the boundaries of problem areas in a three-dimensional spatial coordinate system. This lack of capability directly leads to a lack of clear objective guidance in subsequent data compensation and acquisition work, often resulting in uniform encryption or indiscriminate supplementary measurements, wasting monitoring resources and failing to effectively improve data quality. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides an intelligent laser sensing multi-parameter fusion monitoring platform for mines to solve the problem of insufficient targeted compensation acquisition caused by the inability to accurately locate areas with missing data.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] This invention provides an intelligent laser sensing multi-parameter fusion monitoring platform for mines, comprising:

[0008] The acquisition module collects laser sensing parameters and environmental perception parameters, performs data normalization processing, and outputs explicit monitoring parameters.

[0009] The derivation module calculates the changing trends and correlation structures among explicit monitoring parameters, derives implicit monitoring parameters that cannot be directly measured, and forms an extended parameter set together with the explicit monitoring parameters.

[0010] The parsing module divides the extended parameter set into multiple conflict type groups based on the consistency of associations within the extended parameter set.

[0011] The missing value generation module executes the corresponding fusion strategy for different conflict type groups, generates the initial fusion result, and constructs the missing value map based on the residual conflicts exposed during the fusion process.

[0012] The compensation module, based on the missing value map, generates targeted compensation acquisition instructions according to the missing value distribution to obtain compensation acquisition data;

[0013] The re-fusion module updates the extended parameter set based on the compensated collection data, executes the fusion strategy to generate a comprehensive monitoring parameter set, and outputs early warning information based on the comprehensive monitoring parameter set.

[0014] As a preferred embodiment of the intelligent laser sensing multi-parameter fusion monitoring platform for mines described in this invention, the method for deriving the implicit monitoring parameters that cannot be directly measured includes:

[0015] Based on continuous measurement sequences of explicit monitoring parameters, their dynamic characteristics in the time dimension are obtained as the trend of change; the dynamic characteristics include the rate of change, the direction of change, and the magnitude of change.

[0016] Based on the distribution relationship of explicit monitoring parameters in the spatial dimension, the cooperative change characteristics and dependencies among different parameters are analyzed to obtain the correlation structure;

[0017] Based on the changing trend, the order of different explicit monitoring parameters in the time series changes is quantified by calculating the maximum response time of the time-delay correlation. A leading ranking is formed. The explicit monitoring parameter at the top of the leading ranking is determined as the dominant direction. Causal inference is used to explore the causal transmission path between different parameters in the correlation structure.

[0018] Under the constraint of the dominant direction, the dominant parameter is used as the driving variable for state prediction. Based on the causal transmission path, the state estimation compensation is performed on the dependent variable parameter with transmission lag and external disturbance, and the implicit monitoring parameter that cannot be directly measured is derived.

[0019] As a preferred embodiment of the intelligent laser sensing multi-parameter fusion monitoring platform for mines described in this invention, the method for forming the extended parameter set includes:

[0020] The implicit monitoring parameters and the explicit monitoring parameters are aligned in data space to form a unified multidimensional parameter space;

[0021] Within the multidimensional parameter space, implicit monitoring parameters are assigned dynamic weights, which are obtained based on the confidence level of explicit monitoring parameters and the stability of the causal transmission path when deriving implicit monitoring parameters.

[0022] Based on dynamic weights, implicit monitoring parameters and explicit monitoring parameters are weighted and concatenated to generate an extended parameter set.

[0023] As a preferred embodiment of the intelligent laser sensing multi-parameter fusion monitoring platform for mines described in this invention, the method for dividing the extended parameter set into multiple conflict type groups includes:

[0024] Based on the time synchronization degree of any pair of extended parameters in the extended parameter set; the time synchronization degree is used to measure the alignment degree and consistency of the change trend of the extended parameter pair on the timestamp; when the time synchronization degree is lower than the preset synchronization threshold, the extended parameter pair is marked as a time conflict pair;

[0025] Obtain the spatial gradient difference of spatially adjacent extended parameter pairs in the extended parameter set; when the spatial gradient difference exceeds a preset reasonable range, the extended parameter pair is marked as a spatial conflict pair;

[0026] Based on the dependency structure within the extended parameter set, the dependency consistency between extended parameters on the same dependency chain is calculated; when the dependency consistency does not meet the preset logical conditions, the relevant parameters are marked as dependency conflict pairs.

[0027] All time conflict pairs, space conflict pairs, and dependency conflict pairs are clustered according to conflict type to form time conflict groups, space conflict groups, and dependency conflict groups, which are multiple conflict type groups.

[0028] As a preferred embodiment of the intelligent laser sensing multi-parameter fusion monitoring platform for mines described in this invention, the method for generating the initial fusion result includes:

[0029] Sequence alignment is performed on the extended parameter pairs within the time conflict group, and the time confidence is calculated based on the noise level of each extended parameter after alignment; weighted fusion is performed according to the time confidence to obtain the time consistency fusion result;

[0030] Based on the spatial gradient differences of the extended parameter pairs within the spatial conflict group, data defect regions are identified; supplementary data is generated from spatially adjacent normal extended parameters through spatial reconstruction to correct the data defect regions; and spatial consistency fusion is performed on the corrected extended parameter pairs to obtain the spatial consistency fusion result.

[0031] Based on the causal transmission path within the dependency conflict group, causal distortion parameters are identified; reasonable values ​​of the distortion parameters are predicted based on the causal transmission path, and the reasonable values ​​are used to correct the causal distortion parameters; the parameters that satisfy the consistency of the causal transmission path after correction are fused to obtain the dependency consistency fusion result.

[0032] The initial fusion result is generated by integrating the results of multiple conflict type groups after fusion using corresponding strategies.

[0033] As a preferred embodiment of the intelligent laser sensing multi-parameter fusion monitoring platform for mines described in this invention, the method for constructing a missing degree map based on residual conflicts exposed during the fusion process includes:

[0034] Based on the initial fusion results, the time synchronization degree, spatial gradient difference and dependency consistency of each extended parameter are recalculated; and compared with the preset synchronization threshold, preset reasonable range and preset logical conditions. Extended parameter pairs that do not meet the corresponding conditions are identified as residual conflict pairs, and their conflict type and conflict intensity are recorded.

[0035] Based on the conflict type of the residual conflict pairs, their conflict intensity is quantified as temporal missing degree, spatial missing degree, or dependency missing degree, respectively.

[0036] The missing values ​​of all residual conflict pairs are mapped and normalized according to their spatial location to generate a missing value map that reflects the spatial distribution of residual conflicts.

[0037] As a preferred embodiment of the intelligent laser sensing multi-parameter fusion monitoring platform for mines described in this invention, the method for obtaining compensated acquisition data includes:

[0038] By spatially clustering the missing values ​​in the missing value map, regions with significant missing values ​​can be identified.

[0039] For regions with significant missing values, a set of compensation acquisition instructions is generated for the region and the corresponding missing parameters based on the dominant missing value type and spatial distribution characteristics, in order to obtain compensation acquisition data.

[0040] As a preferred embodiment of the intelligent laser sensing multi-parameter fusion monitoring platform for mines described in this invention, the method for generating the comprehensive monitoring parameter set includes:

[0041] After the compensation-collected data is processed through data normalization, it is used as a new explicit monitoring parameter, which together with the original extended parameter set constitutes an updated extended parameter set.

[0042] For the updated extended parameter set, the conflict type groups are re-divided, and the corresponding fusion strategy is executed for each conflict type group to generate a comprehensive monitoring parameter set.

[0043] As a preferred embodiment of the intelligent laser sensing multi-parameter fusion monitoring platform for mines described in this invention, the method for outputting early warning information includes:

[0044] Inconsistency analysis of the comprehensive monitoring parameter set was conducted to identify anomalous time segments, anomalous spatial regions, and distorted dependency links, which served as risk characterization indicators.

[0045] Based on the magnitude of changes in risk characterization indicators and their impact range in the comprehensive monitoring parameter set, the potential risk level is assessed; and based on the potential risk level and its corresponding spatial distribution, early warning information is generated.

[0046] As a preferred embodiment of the intelligent laser sensing multi-parameter fusion monitoring platform for mines described in this invention, the method for correcting data defect areas includes:

[0047] A spatial neighborhood is established centered on the data defect area, and abnormal parameters that have been marked as spatial conflict pairs are filtered out from it to obtain the normal extended parameter set;

[0048] Spatial interpolation calculations are performed on the normal extended parameter set to generate supplementary data covering the defective areas of the data.

[0049] Replace the original outliers in the data defect area with supplementary data.

[0050] The beneficial effects of this invention are as follows: By analyzing the temporal patterns, spatial correlations, and causal relationships of explicit monitoring parameters, implicit monitoring parameters that cannot be directly measured are derived, expanding monitoring information from visible quantities to potential structures and enhancing the ability to describe complex mineral and rock conditions. Furthermore, after initial fusion, remaining inconsistencies are requantified in terms of time, space, and dependency dimensions to construct a missing value map, thereby intuitively locating weak monitoring areas. The compensation acquisition and refusion mechanism implemented based on the missing value map can continuously correct the parameter system, upgrading monitoring from passive fusion to a highly reliable fusion mode that is interpretable, compensable, and capable of closed-loop optimization. Attached Figure Description

[0051] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 This is a schematic diagram of the intelligent laser sensing multi-parameter fusion monitoring platform for mines in this invention.

[0053] Figure 2 This is a flowchart illustrating the derivation of the latent detection parameters in this invention.

[0054] Figure 3 This is a flowchart of the process for generating the initial fusion result in this invention.

[0055] Figure 4 This is a flowchart of the process for constructing the missing value map in this invention. Detailed Implementation

[0056] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0057] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0058] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0059] Reference Figure 1 , Figure 2 , Figure 3 and Figure 4 This is one embodiment of the present invention, which provides an intelligent laser sensing multi-parameter fusion monitoring platform for mines, comprising the following steps:

[0060] Methods for deriving latent monitoring parameters that cannot be directly measured include:

[0061] Based on continuous measurement sequences of explicit monitoring parameters, their dynamic characteristics in the time dimension are obtained as the trend of change; the dynamic characteristics include the rate of change, the direction of change, and the magnitude of change.

[0062] It should be noted that laser sensing parameters related to the rock and ore structure are obtained from laser scanning equipment, laser ranging devices, or deformation sensors at the mine site, while environmental parameters are obtained from environmental sensing devices such as temperature, humidity, air pressure, vibration, and wind speed. For raw data from different devices, sampling frequencies, and units of measurement, a unified preprocessing process is performed, including timestamp alignment, noise reduction, outlier removal, unit normalization, and format standardization. The standardized multi-source data is then organized into structured explicit monitoring parameters.

[0063] Furthermore, the explicit monitoring parameters are arranged according to their corresponding timestamps to form a sequence with a temporal relationship. The differences between adjacent measurement points in this sequence are then calculated to reflect the increase or decrease of the explicit monitoring parameters within adjacent time intervals. By accumulating the differences across multiple adjacent time intervals, the overall direction of change of the explicit monitoring parameters over a longer time period can be obtained.

[0064] By comparing the magnitude of the differences between adjacent measurement points, the rate of change of the explicit monitoring parameter can be characterized; and by comparing the difference between the maximum and minimum values ​​of the measurement sequence within a given time window, the magnitude of the change of the explicit monitoring parameter can be characterized. The above processing, based on timestamp sorting, calculation of differences between adjacent measurement points, and time window statistics, enables the continuous measurement sequence of the explicit monitoring parameter to exhibit its dynamic characteristics in the time dimension, thereby revealing the changing trend of the explicit monitoring parameter.

[0065] Based on the distribution relationship of explicit monitoring parameters in the spatial dimension, the cooperative change characteristics and dependencies between different parameters are analyzed to obtain the correlation structure.

[0066] It should be noted that each explicit monitoring parameter is bound to its corresponding spatial coordinates, so that the explicit monitoring parameters present a clear spatial distribution organization in a unified spatial coordinate system; and based on the spatial distance and positional adjacency between explicit monitoring parameters, explicit monitoring parameters that are close to each other are grouped into local spatial sets.

[0067] Furthermore, by combining the time series of explicit monitoring parameters, the explicit monitoring parameters in the local spatial set are compared in terms of trend, direction of change, and magnitude of change within a common time period, thereby identifying parameter combinations that change synchronously, follow each other, or have a clear sequential relationship.

[0068] Simultaneously, based on the spatial numerical differences and relative distribution of explicit monitoring parameters, continuously changing regions, abruptly changing regions, and slowly changing regions are identified; explicit monitoring parameters with stable cooperative change characteristics and clear dependencies are represented by connecting lines to form cooperative change lines, so that the spatial location of explicit monitoring parameters and dependency lines together constitute an association structure.

[0069] Based on the changing trend, the order of different explicit monitoring parameters in the time series changes is quantified by calculating the maximum response time of the time-delay correlation. The explicit monitoring parameter at the top of the leading order is identified as the dominant direction. Causal inference is used to explore the causal transmission path between different parameters in the correlation structure.

[0070] It should be noted that, based on the changing trends and correlation structures, the explicit monitoring parameters are compared in chronological order. By observing the sequence of the rate, direction, and magnitude of change of each explicit monitoring parameter within the same time window, the earliest time point at which a significant change occurs is determined. The stability of this time point is then repeatedly confirmed within multiple time windows. Based on the order of the initial response times, the explicit monitoring parameters are arranged into a leading order from earliest to latest.

[0071] Based on the leading ranking, the explicit monitoring parameter that ranks first in the leading ranking and changes earliest in multiple trend windows is determined as the dominant direction, so that the dominant direction can be used as the starting point of the trend in subsequent analysis.

[0072] Furthermore, by combining the coordinated change lines between the dominant direction and the recorded explicit monitoring parameters in the associated structure, the trend response is compared and matched between the explicit monitoring parameters connected by the lines in order of their leading position. By comparing the consistency of the change direction of the explicit monitoring parameters, the following of the change magnitude, and the sequential relationship of the time when the trend appears, the change chain propagating from the dominant direction to other explicit monitoring parameters can be identified. For example, when the explicit monitoring parameter in the dominant direction first shows an upward trend, and the explicit monitoring parameters at adjacent positions show a smaller but consistent upward trend in the subsequent time window, the two can be considered to have a continuous change relationship.

[0073] Subsequently, the chain of changes is gradually extended along the connection of the related structures, so that the primacy and subsequent response relationship between the explicit monitoring parameters is presented in a path manner, thereby forming a causal transmission path.

[0074] Under the constraint of the dominant direction, the dominant parameter is used as the driving variable for state prediction. Based on the causal transmission path, the state estimation compensation is performed on the dependent variable parameter with transmission lag and external disturbance, and the implicit monitoring parameter that cannot be directly measured is derived.

[0075] It should be noted that, based on the dominant parameter as the driving variable, and relying on the historical response patterns of the dominant parameter and each dependent variable parameter recorded in the correlation structure, historical response samples are extracted within the same time scale and overlapping time windows to form a response template; using the current time series of the dominant parameter as input, a matching and fitting method of local time series extrapolation and historical response template is used to infer the expected state of each dependent variable parameter in the future or missing time period. During the extrapolation process, the identified time lag characteristics and sequence are considered to correct the response time, thereby obtaining a set of dependent variable parameter prediction sequences consistent with the causal transmission path;

[0076] Based on the predicted sequence of dependent variable parameters, the deviation between the predicted and observed values ​​is calculated by comparing them with the observed parameters. The deviation is then used as a state estimation compensation amount after being co-verified with environmental perception parameters to eliminate obvious external disturbances. The compensation amount is superimposed on the observed values ​​or used as a substitute for missing responses to form implicit monitoring parameters for expressing physical quantities that cannot be directly measured. Subsequently, the consistency of the compensation results is checked in the direction of the associated structural path. If the compensation introduces new inconsistencies, the prediction and compensation are iteratively adjusted until the compensated dependent variable parameters show continuity and consistency in the causal transmission path, thus completing the derivation of the implicit monitoring parameters.

[0077] Methods for forming an extended parameter set include:

[0078] The implicit monitoring parameters are aligned with the explicit monitoring parameters in the data space to form a unified multidimensional parameter space.

[0079] It should be noted that the latent monitoring parameters and the explicit monitoring parameters are matched item by item according to their corresponding timestamps, spatial coordinates and physical meanings, so that parameters from different sources have a consistent way of expression under the same time reference and the same spatial reference.

[0080] Within a multidimensional parameter space, implicit monitoring parameters are assigned dynamic weights. These dynamic weights are obtained based on the confidence level of the explicit monitoring parameters and the stability of the causal transmission path used in deriving the implicit monitoring parameters.

[0081] It should be noted that the continuity of the direction, rate, and magnitude of change of explicit monitoring parameters across multiple overlapping time windows is checked. If these characteristics remain stable across different time windows—for example, exhibiting a pattern of slow change followed by rapid change in consecutive windows—then the explicit monitoring parameter is considered to have high trend stability. Simultaneously, the consistency of the causal transmission path across different time windows is statistically analyzed to determine whether it maintains the same sequence and connection structure. For example, if the path consistently transmits from parameter A to parameter B and then to parameter C across multiple windows, the degree of repetition reflects the stability of the causal transmission path. Subsequently, the trend stability of the explicit monitoring parameter and the repetitive stability of the causal transmission path are normalized and fused to form the dynamic weight of the implicit monitoring parameter.

[0082] Based on dynamic weights, implicit monitoring parameters and explicit monitoring parameters are weighted and concatenated to generate an extended parameter set.

[0083] It should be noted that, based on the dynamic weights of the latent monitoring parameters, the weights of the latent monitoring parameters are adjusted so that they can participate in the expression according to their reliability in subsequent splicing. Then, under a unified time reference, spatial coordinates, and physical meaning correspondence, the dynamically weighted latent monitoring parameters are inserted into the arrangement structure of the explicit monitoring parameters in their corresponding positions, so that the two types of parameters remain continuous, corresponding, and readable in the same multidimensional parameter space. After the insertion is completed, the explicit and latent monitoring parameters are combined into an extended parameter set.

[0084] Methods for dividing the extended parameter set into multiple conflict type groups include:

[0085] Based on the time synchronization degree of any extended parameter pair in the extended parameter set; the time synchronization degree is used to measure the alignment degree and consistency of the change trend of the extended parameter pair on the timestamp; when the time synchronization degree is lower than the preset synchronization threshold, the extended parameter pair is marked as a time conflict pair.

[0086] It should be noted that the extended parameters are arranged in order of timestamps. Within multiple overlapping time windows, it is checked whether the two extended parameters show trend changes simultaneously within the same time window, whether they reach significant points of change in magnitude within similar timestamps, and whether the direction of change remains consistent in most windows. Then, the repetition ratio of the three types of performance is used as a measure of synchronicity. For example, if eight out of ten overlapping windows show simultaneous changes in the same direction, the time synchronicity can be exemplarily recorded as 0.8.

[0087] Furthermore, the time synchronization values ​​of a large number of extended parameter pairs within the extended parameter set are statistically analyzed across multiple historical windows to observe their approximate distribution range. For example, the synchronization rate is typically concentrated between 0.6 and 0.9. Subsequently, based on the sensitivity requirements of conflict detection, boundary values ​​that can effectively distinguish between good synchronization and abnormal synchronization are selected within this concentrated distribution range. For example, in the sample case, 0.7 can be used as a preset synchronization threshold, so that extended parameter pairs below 0.7 are marked as time conflict pairs.

[0088] Obtain the spatial gradient difference of spatially adjacent extended parameter pairs in the extended parameter set; when the spatial gradient difference exceeds a preset reasonable range, mark the extended parameter pair as a spatial conflict pair.

[0089] It should be noted that the method for obtaining the spatial gradient difference of spatially adjacent extended parameter pairs in the extended parameter set is as follows: bind each extended parameter to its spatial coordinates, and filter extended parameter pairs whose spatial distance is within a preset proximity threshold; for each pair of spatially adjacent extended parameter pairs, calculate the ratio of the difference in their numerical changes to the spatial Euclidean distance in multiple consecutive time windows to obtain the instantaneous spatial gradient of each window, and then calculate the standard deviation of these instantaneous spatial gradients, and quantify the standard deviation as the spatial gradient difference of the pair of extended parameters;

[0090] Extract the extended parameter set within the historical period of the stable state, calculate the spatial gradient difference of all spatially adjacent extended parameter pairs in the historical extended parameter set; perform statistical analysis on these historical spatial gradient difference values, and determine their concentrated distribution interval, for example, by calculating the percentile to obtain the interval [-0.3, 0.3] from the 5th to the 95th percentile, which is determined as the preset reasonable range.

[0091] Based on the dependency structure within the extended parameter set, the dependency consistency between extended parameters on the same dependency chain is calculated; when the dependency consistency does not meet the preset logical conditions, the relevant parameters are marked as dependency conflict pairs.

[0092] It should be noted that the established causal transmission paths are extracted from the association structure, and each causal transmission path is a dependency chain. For each dependency chain, within multiple consecutive time windows, it is checked whether the direction of change of the preceding extended parameters and the subsequent extended parameters are consistent and whether the magnitude of change shows a reasonable following relationship. The results of checking the consistency of direction and the following relationship of magnitude are comprehensively quantified. For example, if nine out of ten time windows meet the requirements, the dependency consistency is recorded as 0.9.

[0093] When dependency consistency does not meet the preset logical conditions, the relevant parameters are marked as dependency conflict pairs. The preset logical conditions are the minimum thresholds set by analyzing the dependency consistency performance of each dependency chain under historical stable conditions. For example, the 10th percentile of the historical dependency consistency value of 0.75 is set as the preset logical conditions. If the dependency consistency of a dependency chain is lower than 0.75, the relevant extended parameters on that chain are marked as dependency conflicts.

[0094] All time conflict pairs, space conflict pairs, and dependency conflict pairs are clustered according to conflict type to form time conflict groups, space conflict groups, and dependency conflict groups, which are multiple conflict type groups.

[0095] It should be noted that time-related conflicts are clustered according to their time periods, with those occurring in consecutive or adjacent time periods grouped into time-related conflict groups; spatial conflicts are clustered according to their spatial location, with those located in adjacent locations or belonging to the same local region grouped into spatial conflict groups; and dependency conflicts are clustered according to their causal transmission paths, with those belonging to or adjacent dependency chains grouped into dependency conflict groups. Ultimately, this results in three conflict type groups that are concentrated in time, space, and dependency structure.

[0096] Methods for generating initial fusion results include:

[0097] Sequence alignment is performed on the extended parameter pairs within the time conflict group, and the time confidence is calculated based on the noise level of each extended parameter after alignment. Weighted fusion is then performed according to the time confidence to obtain the time consistency fusion result.

[0098] It should be noted that the extended parameters with time conflicts are matched by a sliding window on the time axis, and their timestamp sequences are adjusted by a dynamic time warping algorithm so that the key points of the change trends of the two extended parameters are optimally aligned on the time axis.

[0099] It should be noted that the expression for calculating time confidence is:

[0100] ;

[0101] in, It is time confidence. It is the noise variance of this extended parameter over multiple time windows. It represents the number of trend change points that were successfully matched after sequence alignment. It represents the number of points of change in the overall trend;

[0102] The expression for weighted fusion based on time confidence is:

[0103] ;

[0104] in, It is a time-consistent fusion result. It is the number of extended parameters participating in the fusion. It is an index variable of the extended parameters that participate in the fusion. It is the first Time confidence of each extended parameter It is the first The numerical values ​​of each extended parameter.

[0105] Based on the spatial gradient differences of the extended parameter pairs within the spatial conflict group, data defect areas are identified; supplementary data is generated from spatially adjacent normal extended parameters through spatial reconstruction to correct the data defect areas; and spatial consistency fusion is performed on the corrected extended parameter pairs to obtain the spatial consistency fusion result.

[0106] It should be noted that, based on the spatial gradient difference of the extended parameter pairs within the spatial conflict group, the spatial locations corresponding to the extended parameter pairs whose spatial gradient differences exceed the preset reasonable range are marked as outliers, and the continuous spatial range composed of multiple adjacent outliers is identified as a data defect area.

[0107] A spatial neighborhood is established centered on the data defect area, and anomalous parameters that have been marked as spatial conflict pairs are filtered out to obtain a normal extended parameter set; spatial interpolation is performed on the normal extended parameter set to generate supplementary data covering the data defect area; the supplementary data is used to replace the original outliers in the data defect area.

[0108] For the modified extended parameter pairs, the matching degree between their spatial gradient difference and the preset reasonable range is calculated as the spatial confidence level; a weighted average is calculated based on the spatial confidence level of each extended parameter, and the weighted average result is used as the spatial consistency fusion result of the region.

[0109] Based on the causal transmission path within the dependency conflict group, causal distortion parameters are identified; reasonable values ​​of the distortion parameters are predicted based on the causal transmission path, and the reasonable values ​​are used to correct the causal distortion parameters; the parameters that satisfy the consistency of the causal transmission path after correction are fused to obtain the dependency consistency fusion result.

[0110] It should be noted that within the dependency conflict group, based on the established causal transmission path, the consistency of the change direction and the following relationship of the change magnitude of each extended parameter on the path are checked one by one. Those extended parameters that continuously violate the expected change law of the causal transmission path and have significantly low dependency consistency in multiple consecutive time windows are identified as causal distortion parameters.

[0111] Using the extended parameter that is in a normal state and preceding the distorted parameter in the causal transmission path as input, and combining it with the historical response template of the causal transmission path, the reasonable value that the distorted parameter should have at the current moment is calculated through local temporal extrapolation and template matching methods. Then, the observed value of the causal distorted parameter is directly replaced with this reasonable value to complete the correction of the causal distorted parameter. For the parameters that satisfy the consistency of the causal transmission path after correction, that is, the parameters whose change direction and magnitude are consistent with the expected relationship of the causal transmission path, the arithmetic mean method is used for fusion calculation, and the fusion calculation result is used as the dependency consistency fusion result.

[0112] It should be noted that the expression for fusion calculation using the arithmetic mean method is as follows:

[0113] ;

[0114] in, It depends on the consistency fusion result. It is the number of extended parameters participating in the fusion. It is the sequence number of the extended parameter. It is the first The values ​​of the extended parameters that, after modification, satisfy the consistency of the causal transmission path.

[0115] The initial fusion result is generated by integrating the results of multiple conflict type groups after fusion using corresponding strategies.

[0116] Specifically, the temporal consistency fusion result, spatial consistency fusion result, and dependency consistency fusion result are registered according to a unified spatial coordinate benchmark to establish a spatial mapping relationship between the three types of fusion results. When multiple types of fusion results exist simultaneously in the same spatial location, the final value is selected according to the priority rule (temporal consistency fusion result takes precedence over spatial consistency fusion result, and spatial consistency fusion result takes precedence over dependency consistency fusion result). For spatial locations where a value exists only in a single type of fusion result, the value of that single fusion result is directly adopted. Through spatial registration and priority selection, the three types of fusion results are integrated to form a unified initial fusion result.

[0117] Methods for constructing missing degree maps based on residual conflicts exposed during the fusion process include:

[0118] Existing mine monitoring systems lack effective assessment of residual conflicts after data fusion, failing to identify potential temporal asynchrony, spatial discontinuity, and logical inconsistencies in the fusion results. The system directly uses preliminary fusion results for early warning, ignoring underlying data contradictions, leading to undetectable monitoring blind spots, affecting the accuracy of early warnings, and creating potential safety hazards.

[0119] Based on the initial fusion results, the time synchronization degree, spatial gradient difference and dependency consistency of each extended parameter are recalculated; and compared with the preset synchronization threshold, preset reasonable range and preset logical conditions. Extended parameter pairs that do not meet the corresponding conditions are identified as residual conflict pairs, and their conflict type and conflict intensity are recorded.

[0120] It should be noted that the same calculation method used when generating the extended parameter set is employed to recalculate the temporal synchronization, spatial gradient difference, and dependency consistency of each parameter in the initial fusion result. Subsequently, the recalculated temporal synchronization is compared with a preset synchronization threshold, the recalculated spatial gradient difference is compared with a preset reasonable range, and the recalculated dependency consistency is compared with a preset logical condition. Extended parameter pairs with temporal synchronization below the preset synchronization threshold are identified as temporal residual conflict pairs, extended parameter pairs with spatial gradient differences exceeding the preset reasonable range are identified as spatial residual conflict pairs, and extended parameter pairs with dependency consistency not meeting the preset logical condition are identified as dependency residual conflict pairs. Finally, the conflict type (temporal, spatial, or dependency) and conflict intensity of each set of residual conflict pairs are recorded, where the conflict intensity is quantified as the absolute value of the deviation between the corresponding indicator (temporal synchronization, spatial gradient difference, or dependency consistency) and the corresponding threshold (preset synchronization threshold, preset reasonable range boundary, or preset logical condition).

[0121] Based on the conflict type of the residual conflict pairs, their conflict intensity is quantified as temporal missingness, spatial missingness, or dependency missingness, respectively.

[0122] It should be noted that for extended parameter pairs identified as time residual conflict pairs, their recorded conflict strength (i.e., The absolute value of the deviation between the time synchronization degree and the preset synchronization threshold is linearly normalized and mapped to the [0,1] interval. The resulting value is quantified as the time missing degree. The larger the time missing degree value, the more severe the synchronization missing degree of the extended parameter pair in the time dimension. For extended parameter pairs identified as spatial residual conflict pairs, their recorded conflict intensity (i.e., the absolute value of the deviation between the spatial gradient difference and the preset reasonable range boundary) is normalized by range and mapped to the [0,1] interval. The resulting value is quantified as the spatial missing degree. The larger the spatial missing degree value, the more severe the continuity missing degree of the extended parameter pair in the spatial dimension. For extended parameter pairs identified as dependency residual conflict pairs, their recorded conflict intensity (i.e., the absolute value of the deviation between the dependency consistency and the preset logical condition) is normalized by maximum and minimum values ​​and mapped to the [0,1] interval. The resulting value is quantified as the dependency missing degree. The larger the dependency missing degree value, the more severe the logical consistency missing degree of the extended parameter pair in the dependency structure.

[0123] The missing values ​​of all residual conflict pairs are mapped and normalized according to their spatial location to generate a missing value map that reflects the spatial distribution of residual conflicts.

[0124] It should be noted that a spatial coordinate grid system corresponding to the mine monitoring area is established, and the spatial coordinates of each residual conflict pair are mapped to the corresponding grid cell. For each grid cell, the temporal missingness, spatial missingness, and dependency missingness of all residual conflict pairs contained therein are weighted and summed, where the weights are determined by the conflict intensity of each residual conflict pair, thereby obtaining the comprehensive temporal missingness, comprehensive spatial missingness, and comprehensive dependency missingness of the grid cell.

[0125] The range normalization process was applied to the three comprehensive missing values ​​of all grid cells in the entire mining area, and their numerical ranges were uniformly mapped to the [0,1] interval. The normalized comprehensive temporal missing value, comprehensive spatial missing value, and comprehensive dependency missing value were assigned to the red, green, and blue color channels, respectively, and the final color display of each grid cell was generated by RGB color synthesis technology.

[0126] All grid cells are arranged according to their spatial location to form a complete missingness map. This missingness map intuitively reflects the degree of missingness at different spatial locations through color intensity. Red areas indicate time-dominant missingness, green areas indicate space-dominant missingness, and blue areas indicate dependent missingness. The darker the color, the more severe the overall missingness in that area.

[0127] By constructing a missing data map, precise visualization and localization of residual conflicts after data fusion were achieved. This map clearly shows the distribution of missing data in the mining area across time, space, and logical dimensions, providing clear guidance for targeted data compensation, significantly improving the integrity and reliability of the monitoring system and effectively supporting mine safety decision-making.

[0128] Methods for obtaining compensation collection data include:

[0129] By spatially clustering the missing values ​​in the missing value map, regions with significant missing values ​​can be identified.

[0130] Specifically, the comprehensive temporal missingness, comprehensive spatial missingness, and comprehensive dependency missingness values ​​of all grid cells in the missingness map are extracted to form a multidimensional missingness feature vector. By finding high-density regions and separating low-density regions, grid cells that are spatially adjacent and have similar missingness features are grouped into the same cluster. Clusters whose missingness values ​​at their centers exceed a preset significance threshold are selected. For example, clusters with a comprehensive missingness greater than 0.7 are identified as significant regions, and the spatial regions covered by these clusters are identified as significant missingness regions.

[0131] For regions with significant missing values, a set of compensation acquisition instructions is generated for the region and the corresponding missing parameters based on the dominant missing value type and spatial distribution characteristics, in order to obtain compensation acquisition data.

[0132] It should be noted that by analyzing the relative magnitudes of the combined temporal, spatial, and dependency missing values ​​within each significant missing value region, the missing value type with the largest value is identified as the dominant missing value type. Subsequently, corresponding compensation acquisition instructions are generated based on the dominant missing value type. For regions dominated by temporal missing values, instructions to increase the sampling frequency are generated; for regions dominated by spatial missing values, instructions to increase spatial resolution are generated; and for regions dominated by dependency missing values, instructions to strengthen the synchronous acquisition of correlation parameters are generated. At the same time, the spatial distribution characteristics of the significant missing value regions are combined to determine the specific spatial coordinate range that needs to be focused on acquisition. Finally, these instructions are integrated to form a compensation acquisition instruction set for the significant missing value region, which guides the field equipment to acquire more targeted compensation acquisition data.

[0133] Methods for generating integrated monitoring parameter sets include:

[0134] After the compensation-collected data is processed through data normalization, it is used as a new explicit monitoring parameter, which together with the original extended parameter set constitutes an updated extended parameter set.

[0135] Specifically, the compensation data is first processed using the same standardized procedures as the initial explicit monitoring parameters, including timestamp alignment, noise reduction, outlier removal, dimension normalization, and format standardization. Then, the standardized compensation data is marked as the newly added explicit monitoring parameters, and these new explicit monitoring parameters are spatially matched and time-series aligned with the original extended parameter set according to a unified spatial coordinate and time reference. Finally, the data is stitched together to form an updated extended parameter set that includes both the original extended parameters and the newly added explicit monitoring parameters.

[0136] For the updated extended parameter set, the conflict type groups are re-divided, and the corresponding fusion strategy is executed for each conflict type group to generate a comprehensive monitoring parameter set.

[0137] Specifically, using the same partitioning method as the initial extended parameter set, and based on the temporal synchronization, spatial gradient difference, and dependency consistency of each parameter in the updated extended parameter set, temporal conflict pairs, spatial conflict pairs, and dependency conflict pairs are re-identified and clustered according to conflict type to form new temporal conflict groups, spatial conflict groups, and dependency conflict groups. Subsequently, corresponding fusion strategies are applied to these new conflict type groups respectively: sequence alignment and time confidence-based weighted fusion are performed on the new temporal conflict groups; data defect area correction and spatial consistency fusion are performed on the new spatial conflict groups; and causal distortion parameter correction and dependency consistency fusion are performed on the new dependency conflict groups. Finally, the three types of fusion results are integrated according to a unified spatial coordinate benchmark to generate a comprehensive monitoring parameter set.

[0138] Methods for outputting early warning information include:

[0139] Inconsistency analysis of the comprehensive monitoring parameter set was conducted to identify anomalous time segments, anomalous spatial regions, and distorted dependency links, which served as risk characterization indicators.

[0140] It should be noted that by using a sliding time window to detect abrupt changes and abnormal fluctuations in the parameter sequences of the comprehensive monitoring parameter set, time periods in which the time synchronization degree continuously falls below the preset synchronization threshold are marked as time abnormal segments.

[0141] Based on spatial gradient analysis, adjacent monitoring points whose parameter values ​​change drastically are located, and continuous areas where the spatial gradient difference continues to exceed the preset reasonable range are identified as spatial anomaly areas.

[0142] Traverse all causal transmission paths, check the dependency consistency between parameters on each path, and mark the transmission links whose dependency consistency does not meet the preset logical conditions as dependency distortion links; use the identified time anomaly segments, spatial anomaly regions and dependency distortion links as risk characterization indicators reflecting the risk status of the system.

[0143] Based on the magnitude of changes in risk characterization indicators and their impact range in the comprehensive monitoring parameter set, the potential risk level is assessed; and based on the potential risk level and its corresponding spatial distribution, early warning information is generated.

[0144] Specifically, quantify the magnitude and scope of change of each risk characterization indicator;

[0145] For time-abnormal segments, the absolute value of the deviation between its time synchronization degree and the preset synchronization threshold is quantified as the change amplitude, and the ratio of the number of extended parameters involved in the segment to the total number of parameters in the comprehensive monitoring parameter set is quantified as the range of influence.

[0146] For areas with spatial anomalies, the absolute value of the deviation between the spatial gradient difference and the preset reasonable range boundary is quantified as the change amplitude, and the ratio of the area of ​​this area to the total area of ​​the monitored area is quantified as the range of influence.

[0147] For distorted dependency links, the absolute value of the deviation between its dependency consistency and the preset logical conditions is quantified as the magnitude of change, and the ratio of the number of causal transmission paths affected by the link to the total number of paths is quantified as the scope of influence.

[0148] Multiply the magnitude of change of each risk indicator by its range of influence to obtain the comprehensive risk value of that risk indicator; set a threshold for the comprehensive risk value, for example: classify comprehensive risk values ​​below 0.3 as low potential risk level, between 0.3 and 0.7 as medium potential risk level, and above 0.7 as high potential risk level.

[0149] By combining the spatial distribution information of various risk characterization indicators, early warning information is generated that includes the potential risk level, specific spatial location, and corresponding risk type (temporal anomaly, spatial anomaly, or dependency distortion).

[0150] Methods for correcting data defect areas include:

[0151] A spatial neighborhood is established centered on the data defect area, and abnormal parameters that have been marked as spatial conflict pairs are filtered out from it to obtain the normal extended parameter set.

[0152] It should be noted that, with the geometric center of the data defect area as the center and a preset neighborhood radius (exemplarily set to 3 times the average spacing between monitoring points) as the distance, a circular spatial neighborhood is defined; all extended parameters within this spatial neighborhood are traversed and compared with the recorded list of spatial conflict pairs, and all abnormal parameters marked as spatial conflict pairs are filtered out; the remaining normal extended parameters that are not marked as abnormal are collected to form a set of normal extended parameters.

[0153] Spatial interpolation calculations are performed on the normal extended parameter set to generate supplementary data covering the defective areas of the data.

[0154] Replace the original outliers in the data defect area with supplementary data.

[0155] It should be noted that the Kriging spatial interpolation algorithm is used, with the spatial coordinates of each normal expansion parameter in the normal expansion parameter set as the position input and the corresponding parameter value as the attribute input.

[0156] A spatial variogram model is established based on the input coordinates and values. This spatial variogram model quantifies the autocorrelation characteristics of the normal expansion parameter in space by calculating the semivariogram values ​​at different distances and fits the optimal variogram curve.

[0157] Based on the established spatial variogram model, the optimal linear unbiased estimate of each interpolation point in the data defect region is calculated through the Kriging equation system. This calculation process comprehensively considers the spatial weight distribution of the surrounding normal expansion parameters. A continuous supplementary data surface covering the entire data defect region is generated through calculation. This supplementary data surface maintains consistency with the statistical relationship and spatial variation trend of the surrounding normal expansion parameters in space.

[0158] The supplementary data surface is precisely superimposed on the spatial range of the data defect area. The interpolation results of the supplementary data surface at the corresponding positions are used to replace the original extended parameter values ​​marked as abnormal in the data defect area one by one, thus completing the correction of the data defect area.

[0159] In summary, this invention, by analyzing the temporal patterns, spatial correlations, and causal relationships of explicit monitoring parameters, derives implicit monitoring parameters that cannot be directly measured, thus expanding monitoring information from visible quantities to potential structures and enhancing the ability to describe complex mineral and rock conditions. Furthermore, after initial fusion, it requantifies remaining inconsistencies in terms of time, space, and dependency dimensions to construct a missing value map, thereby intuitively locating weak monitoring areas. The compensation acquisition and refusion mechanism implemented based on the missing value map can continuously correct the parameter system, upgrading monitoring from passive fusion to a highly reliable fusion mode that is interpretable, compensable, and capable of closed-loop optimization.

[0160] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A multi-parameter fusion monitoring platform for intelligent laser sensing in mining, characterized in that, include: The acquisition module collects laser sensing parameters and environmental perception parameters, performs data normalization processing, and outputs explicit monitoring parameters. The derivation module calculates the changing trends and correlation structures among explicit monitoring parameters, derives implicit monitoring parameters that cannot be directly measured, and forms an extended parameter set together with the explicit monitoring parameters. The parsing module divides the extended parameter set into multiple conflict type groups based on the consistency of associations within the extended parameter set. The missing value generation module executes the corresponding fusion strategy for different conflict type groups, generates the initial fusion result, and constructs the missing value map based on the residual conflicts exposed during the fusion process. The compensation module, based on the missing value map, generates targeted compensation acquisition instructions according to the missing value distribution to obtain compensation acquisition data; The re-fusion module updates the extended parameter set based on the compensated collection data, executes the fusion strategy to generate a comprehensive monitoring parameter set, and outputs early warning information based on the comprehensive monitoring parameter set.

2. The intelligent laser sensing multi-parameter fusion monitoring platform for mines as described in claim 1, characterized in that, The methods for deriving latent monitoring parameters that cannot be directly measured include: Based on continuous measurement sequences of explicit monitoring parameters, their dynamic characteristics in the time dimension are obtained as the trend of change; the dynamic characteristics include the rate of change, the direction of change, and the magnitude of change. Based on the distribution relationship of explicit monitoring parameters in the spatial dimension, the cooperative change characteristics and dependencies among different parameters are analyzed to obtain the correlation structure; Based on the changing trend, the order of different explicit monitoring parameters in the time series changes is quantified by calculating the maximum response time of the time-delay correlation. A leading ranking is formed. The explicit monitoring parameter at the top of the leading ranking is determined as the dominant direction. Causal inference is used to explore the causal transmission path between different parameters in the correlation structure. Under the constraint of the dominant direction, the dominant parameter is used as the driving variable for state prediction. Based on the causal transmission path, the state estimation compensation is performed on the dependent variable parameter with transmission lag and external disturbance, and the implicit monitoring parameter that cannot be directly measured is derived.

3. The intelligent laser sensing multi-parameter fusion monitoring platform for mines as described in claim 2, characterized in that, The method for forming the extended parameter set includes: The implicit monitoring parameters and the explicit monitoring parameters are aligned in data space to form a unified multidimensional parameter space; Within the multidimensional parameter space, implicit monitoring parameters are assigned dynamic weights, which are obtained based on the confidence level of explicit monitoring parameters and the stability of the causal transmission path when deriving implicit monitoring parameters. Based on dynamic weights, implicit monitoring parameters and explicit monitoring parameters are weighted and concatenated to generate an extended parameter set.

4. The intelligent laser sensing multi-parameter fusion monitoring platform for mines as described in claim 3, characterized in that, The method for dividing the extended parameter set into multiple conflict type groups includes: Based on the time synchronization degree of any pair of extended parameters in the extended parameter set; the time synchronization degree is used to measure the alignment degree and consistency of the change trend of the extended parameter pair on the timestamp; when the time synchronization degree is lower than the preset synchronization threshold, the extended parameter pair is marked as a time conflict pair; Obtain the spatial gradient difference of spatially adjacent extended parameter pairs in the extended parameter set; when the spatial gradient difference exceeds a preset reasonable range, the extended parameter pair is marked as a spatial conflict pair; Based on the dependency structure within the extended parameter set, the dependency consistency between extended parameters on the same dependency chain is calculated; when the dependency consistency does not meet the preset logical conditions, the relevant parameters are marked as dependency conflict pairs. All time conflict pairs, space conflict pairs, and dependency conflict pairs are clustered according to conflict type to form time conflict groups, space conflict groups, and dependency conflict groups, which are multiple conflict type groups.

5. The intelligent laser sensing multi-parameter fusion monitoring platform for mines as described in claim 4, characterized in that, The method for generating the initial fusion result includes: Sequence alignment is performed on the extended parameter pairs within the time conflict group, and the time confidence is calculated based on the noise level of each extended parameter after alignment; weighted fusion is performed according to the time confidence to obtain the time consistency fusion result; Based on the spatial gradient differences of the extended parameter pairs within the spatial conflict group, data defect regions are identified; supplementary data is generated from spatially adjacent normal extended parameters through spatial reconstruction to correct the data defect regions; and spatial consistency fusion is performed on the corrected extended parameter pairs to obtain the spatial consistency fusion result. Based on the causal transmission path within the dependency conflict group, causal distortion parameters are identified; reasonable values ​​of the distortion parameters are predicted based on the causal transmission path, and the reasonable values ​​are used to correct the causal distortion parameters; the parameters that satisfy the consistency of the causal transmission path after correction are fused to obtain the dependency consistency fusion result. The initial fusion result is generated by integrating the results of multiple conflict type groups after fusion using corresponding strategies.

6. The intelligent laser sensing multi-parameter fusion monitoring platform for mines as described in claim 5, characterized in that, The method for constructing a missing degree map based on residual conflicts exposed during the fusion process includes: Based on the initial fusion results, the time synchronization degree, spatial gradient difference and dependency consistency of each extended parameter are recalculated; and compared with the preset synchronization threshold, preset reasonable range and preset logical conditions. Extended parameter pairs that do not meet the corresponding conditions are identified as residual conflict pairs, and their conflict type and conflict intensity are recorded. Based on the conflict type of the residual conflict pairs, their conflict intensity is quantified as temporal missing degree, spatial missing degree, or dependency missing degree, respectively. The missing values ​​of all residual conflict pairs are mapped and normalized according to their spatial location to generate a missing value map that reflects the spatial distribution of residual conflicts.

7. The intelligent laser sensing multi-parameter fusion monitoring platform for mines as described in claim 6, characterized in that, The method for obtaining the compensation collection data includes: By spatially clustering the missing values ​​in the missing value map, regions with significant missing values ​​can be identified. For regions with significant missing values, a set of compensation acquisition instructions is generated for the region and the corresponding missing parameters based on the dominant missing value type and spatial distribution characteristics, in order to obtain compensation acquisition data.

8. The intelligent laser sensing multi-parameter fusion monitoring platform for mines as described in claim 7, characterized in that, The method for generating the comprehensive monitoring parameter set includes: After the compensation-collected data is processed through data normalization, it is used as a new explicit monitoring parameter, which together with the original extended parameter set constitutes an updated extended parameter set. For the updated extended parameter set, the conflict type groups are re-divided, and the corresponding fusion strategy is executed for each conflict type group to generate a comprehensive monitoring parameter set.

9. The intelligent laser sensing multi-parameter fusion monitoring platform for mines as described in claim 5, characterized in that, The method for outputting early warning information includes: Inconsistency analysis of the comprehensive monitoring parameter set was conducted to identify anomalous time segments, anomalous spatial regions, and distorted dependency links, which served as risk characterization indicators. Based on the magnitude of changes in risk characterization indicators and their impact range in the comprehensive monitoring parameter set, the potential risk level is assessed; and based on the potential risk level and its corresponding spatial distribution, early warning information is generated.

10. The intelligent laser sensing multi-parameter fusion monitoring platform for mines as described in claim 6, characterized in that, The method for correcting data defect areas includes: A spatial neighborhood is established centered on the data defect area, and abnormal parameters that have been marked as spatial conflict pairs are filtered out from it to obtain the normal extended parameter set; Spatial interpolation calculations are performed on the normal extended parameter set to generate supplementary data covering the defective areas of the data. Replace the original outliers in the data defect area with supplementary data.

Images