Coal mine spontaneous combustion and water inrush disaster coordinated regulation method and system

Abstract

The application discloses a coal mine spontaneous combustion and water inrush disaster collaborative regulation method and system, relates to the technical field of coal mine safety, and comprises the following steps: obtaining tide level time series data and roadway monitoring data of a target object; performing mutual consistent scoring on the tide level time series data and the roadway monitoring data to generate a consistent scoring sequence; performing consistent scoring driven adaptive phase tracking on the tide level time series data to generate a phase window label, and performing grouping and phase window projection on the roadway monitoring data according to the phase window label to generate a hydraulic potential sequence. According to the scheme, the tide level time series data is subjected to consistent scoring driven adaptive phase tracking to generate a phase window label and project to generate a hydraulic potential sequence, the tidal period is stably mapped to a phase window, and the roadway response is projected to a hydraulic intensity representation, so that the runoff influence is phase-positioned and insensitive to phase jump.

Description

Technical Field

[0001] This invention relates to the field of coal mine safety technology, specifically to a method and system for the coordinated control of spontaneous combustion and water inrush disasters in coal mines. Background Technology

[0002] During coal mining, seepage channels and ventilation channels in underground goafs, fracture zones, and surrounding rock of roadways often coexist and are coupled with each other. Especially in coastal coal mines, the external tidal water level has a periodic driving effect on the hydraulic boundary of the near-shore aquifer. The rise and fall of the tide will cause periodic changes in the pore water pressure and seepage direction underground, thereby inducing periodic fluctuations in the intensity of water inflow. It may also change the leakage pattern and oxygen supply conditions of the fracture zone at certain times, thus affecting the evolution process of the high-temperature zone of spontaneous combustion of coal.

[0003] In existing technologies, coal spontaneous combustion disasters typically employ threshold alarms based on monitoring parameters such as roadway temperature, carbon monoxide concentration, and oxygen concentration, along with trend analysis and nitrogen inerting, sealing of leaks, and pressure equalization ventilation. Water inrush disasters typically employ threshold alarms based on monitoring parameters such as inrush volume and pore water pressure, along with drainage and pressure relief, and grouting for water plugging. The advantage of these solutions lies in their relatively mature engineering implementation paths. However, in coastal coal mines, the external tidal cycle drives groundwater runoff to fluctuate periodically. The above control solutions mostly rely on inrush flow rate or pore pressure amplitude thresholds for triggering, lacking a comprehensive consideration of the tidal cycle in relation to the wellbore... The current method of establishing a stable correspondence between monitoring responses makes it difficult to conduct comparable analysis of risks on different dates at the same tidal cycle position. Consequently, in the absence of phase reference, it is difficult to separate the tidal-driven hydraulic response components and the spontaneous combustion-related oxygen trend components in the tunnel monitoring data. This leads to the overlap of common mode changes caused by water level cycle fluctuations and spontaneous combustion trends, causing mutual interference between water inrush judgment and spontaneous combustion judgment. Furthermore, due to the above overlap, it is difficult to quantify the impact of drainage behavior on oxygen supply changes, and it is difficult to identify certain phase windows where drainage is more likely to trigger enhanced oxygen supply, thus making it difficult to implement fine-grained control. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for the coordinated control of spontaneous combustion and water inrush disasters in coal mines, so as to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the technical solution of the present invention is as follows: In a first aspect, this invention discloses a method for the coordinated control of spontaneous combustion and water inrush disasters in coal mines, applicable to the coordinated control of spontaneous combustion and water inrush disasters in coastal coal mines, comprising the following steps: Acquire tidal level time series data and tunnel monitoring data for the target object; The tidal time series data is used to perform mutual verification and consistency scoring on the roadway monitoring data to generate a consistent score sequence; Adaptive phase tracking driven by consistent scoring is performed on the tidal time series data to generate phase window labels. Based on the phase window labels, the tunnel monitoring data is grouped and projected within the phase window to generate a hydraulic potential sequence. The roadway monitoring data, the consistent scoring sequence, and the hydraulic potential sequence are subjected to in-direction component separation processing to generate an oxygen potential sequence. Then, phase differential gain construction is performed on the hydraulic potential sequence and the oxygen potential sequence to generate an oxygen-inducing gain sequence. A quantile threshold self-calibration gating is performed on the oxygen-inducing gain sequence and the phase window label to generate a gating window set. Then, a sequence graph search with gating constraints is performed on the gating window set, the oxygen potential sequence, and the oxygen-inducing gain sequence to generate a sequence scheme table. The phase-bearing water-blocking curtain parameter generation process is performed on the sequence scheme table, the gate window set, the tidal time series data, and the oxygenation gain sequence to generate the control result.

[0006] Secondly, this invention discloses a coordinated control system for spontaneous combustion and water inrush disasters in coal mines, comprising: The data acquisition module is used to acquire the tidal time series data and roadway monitoring data of the target object; The consistency evaluation module is used to perform mutual verification consistency scoring on the tidal time series data and the roadway monitoring data, and generate a consistency score sequence. The hydraulic analysis module is used to perform consistent scoring-driven adaptive phase tracking on the tidal time series data, generate phase window labels, and perform grouping and phase window projection on the tunnel monitoring data according to the phase window labels to generate a hydraulic potential sequence. The gain processing module is used to perform in-direction component separation processing on the roadway monitoring data, the consistent scoring sequence and the hydraulic potential sequence to generate the oxygen potential sequence, and to perform phase difference gain construction on the hydraulic potential sequence and the oxygen potential sequence to generate the oxygen-inducing gain sequence. The sequence scheme generation module is used to perform quantile threshold self-calibration gating on the oxygen-inducing gain sequence and the phase window label to generate a gating window set, and to perform a sequence graph search with gating constraints on the gating window set, the oxygen potential sequence and the oxygen-inducing gain sequence to generate a sequence scheme table; The regulation result generation module is used to perform phase-bearing water-blocking curtain parameter generation processing on the sequence scheme table, the gate window set, the tide time series data, and the oxygenation gain sequence to generate regulation results.

[0007] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This scheme generates phase window labels by performing consistent scoring-driven adaptive phase tracking on tidal time series data and projects them to generate hydraulic potential sequences. It maps the tidal cycle to a stable phase window and projects the roadway response to a hydraulic intensity characterization, so that the runoff impact is phase-localized and insensitive to phase jumps. It performs common-mode and differential-mode separation on roadway monitoring data, consistent scoring sequences, and hydraulic potential sequences to generate oxygen potential sequences and constructs oxygen-inducing gain sequences. It separates the common-mode impact in the same direction as hydraulics and highlights the slow trend of oxygen change. At the same time, it characterizes the degree of oxygen amplification caused by unit hydraulic change, so that the risk of water control causing fires can be quantified. It generates a set of gated windows by self-calibrating gating of oxygen-inducing gain sequences and phase window labels and generates a sequence scheme table by sequence graph search. It adaptively divides prohibited, restricted, and suitable discharge according to the phase window and selects the inertial blocking and drainage sequence accordingly, so that the timing and sequence of control are constrained to the low negative effect window.

[0008] 2. This scheme reduces the proportion of low-confidence samples affecting the monitoring results by weighting and correcting the roadway monitoring data according to the consistent scoring sequence within the confidence window set. This makes the retained data more concentrated in representing the true roadway response. Projecting the weighted monitoring data in the same direction along the hydraulic potential sequence can extract the common mode response consistent with tidal-driven runoff and clarify the contribution boundary of hydraulic coupling components. Subtracting the same-direction projection result from the weighted monitoring data can weaken the masking of the hydraulic common mode on the identification of oxygen and fuel changes and improve the separation degree of spontaneous combustion-related slow-change information. Performing single-axis mapping on the differential mode monitoring data according to the sampling order and summarizing it generates an oxygen and fuel potential sequence, which can form continuous and comparable oxygen and fuel characterization results and improve the consistency of subsequent phase gating judgment. Attached Figure Description

[0009] The disclosure of this invention is illustrated with reference to the accompanying drawings. It should be understood that the drawings are for illustrative purposes only and are not intended to limit the scope of protection of this invention. In the drawings, the same reference numerals are used to refer to the same parts. Wherein: Figure 1 A flowchart illustrating the steps of the method for coordinated control of spontaneous combustion and water inrush disasters in coal mines provided by the present invention; Figure 2 A schematic diagram of the process for generating phase window labels provided by the present invention; Figure 3 A schematic diagram of the process for generating hydraulic potential sequences provided by the present invention; Figure 4 This is a schematic diagram of the process for generating the oxygen potential sequence provided by the present invention. Figure 5 A schematic diagram of the module functions of the coal mine spontaneous combustion and water inrush disaster coordinated control system provided by the present invention. Detailed Implementation

[0010] It is readily understood that, based on the technical solution of this invention, those skilled in the art can propose various interchangeable structural methods and implementations without altering the essential spirit of the invention. Therefore, the following detailed embodiments and accompanying drawings are merely illustrative examples of the technical solution of this invention and should not be considered as the entirety of the invention or as limitations or restrictions on the technical solution of this invention.

[0011] Application Overview: In traditional coastal coal mining, external tidal levels periodically drive the hydraulic boundary of near-shore aquifers. Tidal fluctuations cause periodic changes in pore water pressure and seepage direction underground, inducing periodic fluctuations in water inrush intensity and potentially altering the leakage pattern and oxygen supply conditions of fracture zones. Existing technologies for controlling coal spontaneous combustion and water inrush hazards primarily rely on threshold triggers based on monitoring data, lacking methods to establish a stable correlation between tidal cycles and underground monitoring responses. This makes it difficult to conduct comparable analyses of risks on different dates at the same tidal cycle location. Consequently, the tidal-driven hydraulic response components and the combustion oxygen trend components related to spontaneous combustion are difficult to separate in roadway monitoring data, leading to mutual interference between water inrush and spontaneous combustion assessments. Furthermore, due to the aforementioned overlap, the impact of drainage behavior on oxygen supply changes is difficult to quantify, making it difficult to identify certain phase windows where drainage is more likely to trigger enhanced oxygen supply.

[0012] For example, in the monitoring scenario of the goaf area of ​​a coastal coal mine, the tide level completes a rise and fall cycle every 12.4 hours. The underground roadway monitoring system records that the carbon monoxide concentration and water inflow increase synchronously during the rising tide phase. The increase in carbon monoxide concentration may be due to the spontaneous combustion process of coal or the change in the leakage pattern caused by the tide. However, due to the lack of phase reference, it is difficult to distinguish the hydraulic response and the oxygen combustion trend components. Furthermore, after implementing drainage operations within a specific tide phase window, the oxygen supply conditions do not change significantly; while in another phase window, the same drainage volume leads to enhanced oxygen supply conditions. However, existing technology cannot identify the specific phase position of this difference, thus affecting the targeted implementation of control measures.

[0013] If the above problems are not addressed, the aliasing effect will persist, reducing the reliability of the disaster early warning system and potentially leading to false alarms or missed alarms. For example, misjudging tidal-driven water inrush as a precursor to spontaneous combustion may trigger unnecessary nitrogen inerting operations, disrupting normal production processes, or it may overlook the true risk of spontaneous combustion, as the tides mask the oxygen combustion trend, increasing the probability of a fire. Furthermore, the difficulty in quantifying the impact of drainage on oxygen supply may lead to drainage actions being carried out in the wrong phase window, unintentionally enhancing oxygen supply conditions, accelerating the coal spontaneous combustion process, and thus increasing the risk of disaster.

[0014] After introducing the basic concept of the present invention, the embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0015] Example 1: Please see Figure 1 A method for coordinated control of spontaneous combustion and water inrush disasters in coal mines, applied to the coordinated control of spontaneous combustion and water inrush disasters in coastal coal mines, includes the following steps: Acquire tidal level time series data and tunnel monitoring data for the target object; Perform cross-validation and consistency scoring on the tidal level time series data and the roadway monitoring data to generate a consistent score sequence; Adaptive phase tracking driven by consistent scoring is performed on the tidal time series data to generate phase window labels. Based on the phase window labels, the roadway monitoring data is grouped and projected within the phase window to generate a hydraulic potential sequence. The roadway monitoring data, consistent scoring sequence, and hydraulic potential sequence are subjected to in-direction component separation processing to generate the oxygen potential sequence. The hydraulic potential sequence and the oxygen potential sequence are subjected to phase difference gain construction to generate the oxygen-inducing gain sequence. A quantile threshold self-calibration gating is performed on the oxygen induction gain sequence and phase window label to generate a gating window set. Then, a sequence graph search with gating constraints is performed on the gating window set, the oxygen potential sequence, and the oxygen induction gain sequence to generate a sequence scheme table. The phase-bearing water-blocking curtain parameter generation process is performed on the sequence scheme table, gate window set, tidal time series data and oxygenation gain sequence to generate the control results.

[0016] Among them, tidal time series data refers to the water level height sequence data that changes continuously over time due to changes in the water level of external water bodies in the coastal coal mining area; Roadway monitoring data refers to a multi-parameter time series data set that is directly collected in the underground roadway space of the Binhai Coal Mine and arranged according to a unified time benchmark; Mutual verification consistency scoring refers to the process of quantifying the credibility of the coupling between tide level drive and roadway response by aligning the tide level time series data and roadway monitoring data in time and dividing them into sliding windows under the same sampling time reference, based on the directional consistency, amplitude smoothness consistency and isolated peak suppression between the tide level change and the roadway monitoring response within each sliding window. Consistent scoring sequence refers to the quantitative result of the degree of consistency between tide level time series data and roadway monitoring data at the same time scale; Phase window labels are structured identifiers used to map a continuous time axis to discrete phase intervals; Hydraulic potential sequence refers to a single-valued time series obtained by robustly projecting roadway monitoring data within a phase window within a credible set, used to characterize the intensity of the hydraulic response of a roadway driven by tidal level. The process of separating co-directional components refers to the process of distinguishing co-directional response components from roadway monitoring data based on the tidal-driven hydraulic response direction characterized by the hydraulic potential sequence, and retaining the remaining response components that are distinct from the co-directional response components. Oxygen potential series refers to time series data that is more sensitive to the slowly changing oxygen supply trend, after the common-mode hydraulic response that changes in the same direction as the tidal phase is extracted from the roadway monitoring data in the phase window coordinate system. Oxygen-inducing gain sequence refers to time series data used to characterize the intensity of oxygen combustion trend changes corresponding to a unit change in hydraulic action under specific tidal phase conditions; Self-calibrated quantile threshold gating refers to a gating process that adaptively determines the gating threshold of a corresponding phase window based on the numerical distribution characteristics of the oxygen induction gain sequence in each phase window within the phase interval defined by the phase window label, and classifies the permissibility and intensity level of the drainage action of the phase window according to the distribution state of the oxygen induction gain relative to the gating threshold. The gate control window set refers to a structured data set formed after classifying the permissibility and intensity level of drainage actions for each phase window under a unified phase window label coordinate system; A sequence scheme table refers to a set of structured phase modulation sequence data generated using phase window labels as the unique index primary key. The control result refers to the set of structured control data that is uniformly organized and output based on phase coordinates after completing the multi-stage correlation processing of tide time series data and roadway monitoring data.

[0017] This scheme performs cross-validation and consistency scoring on roadway monitoring data using tidal time-series data and generates a consistency scoring sequence. This filters out distorted responses and highlights true coupling segments, improving the accuracy of subsequent phase identification. Adaptive phase tracking driven by consistency scoring is performed on tidal time-series data, and a hydraulic potential sequence is generated by combining phase window projection. This stably extracts tidal-driven runoff intensity, improving the ability to characterize periodic water hazards. Common-mode and differential-mode separation is performed on roadway monitoring data, consistency scoring sequences, and hydraulic potential sequences to construct an oxygen-inducing gain sequence. This separates spontaneous combustion-related changes from hydraulic common-mode disturbances, improving the accuracy of negative effect identification. A quantile threshold self-calibration gating is performed on the oxygen-inducing gain sequence and phase window labels, and a sequence scheme table is generated by combining sequence graph search. This forms an action sequence adapted to phase changes, improving the continuity of control. Phase-bearing water-blocking curtain parameters are generated and processed on the sequence scheme table, gating window set, tidal time-series data, and oxygen-inducing gain sequence, generating control results. This allows one-time grouting to adapt to periodic tidal impacts, improving long-term sealing stability.

[0018] The above describes the complete scheme for the coordinated control of spontaneous combustion and water inrush disasters in coal mines. The following section describes the acquisition of tidal level time-series data and roadway monitoring data for the target object, specifically including: Tidal level time series data of the target object are obtained by water level gauge; tidal level time series data includes but is not limited to sampling time, water level value, direction of water level change, magnitude of water level change, and water level fluctuation relationship between adjacent sampling times; The monitoring data of the target object is obtained by monitoring sensors; the monitoring data includes, but is not limited to, sampling time, temperature information, carbon monoxide concentration information, oxygen concentration information, water flow change information in the tunnel, pressure change information in the tunnel, and seepage response information of surrounding rock or fissures. Perform cross-validation and consistency scoring on the tidal level time series data and the roadway monitoring data to generate a consistent score sequence: The tidal time series data and the roadway monitoring data are aligned according to the same sampling order to obtain a pair of sample sets. Then, the pair of sample sets are segmented according to a sliding window of fixed length. Within each sliding window, the directional consistency amount, smoothing consistency amount, and isolation suppression amount are calculated for the tidal time series data segment and the roadway monitoring data segment, and then fused to obtain the window score. The calculation process of the directional consistency amount is as follows: the difference between adjacent sample values ​​of the tidal time series data segment and the roadway monitoring data segment is calculated to obtain a difference sequence. Then, the positive and negative signs of the difference sequence are converted into a directional sequence. The number of times the two directional sequences have the same sign at corresponding positions is counted and divided by the length of the directional sequence within the window to obtain the directional consistency amount. The process of calculating the smoothing uniformity is as follows: take the absolute value of the two difference sequences to obtain the amplitude sequence, calculate the average value of the amplitude sequence and take the absolute value of the difference between the two as the amplitude difference, then divide the amplitude difference by the sum of the average values ​​of the two amplitude sequences to obtain the difference ratio, and finally subtract the difference ratio from 1 to obtain the smoothing uniformity. The closer the amplitude change is to the smoothing uniformity, the larger the smoothing uniformity is. The process of calculating the isolated suppression amount is as follows: find the maximum value of the amplitude sequence of the roadway monitoring data segment and the average value of the remaining amplitudes. Calculate the average value of the remaining amplitudes and divide it by the maximum value to obtain the isolated suppression amount, so that the amount is reduced when there is an isolated peak in the monitoring segment. The sliding window score is obtained by multiplying the direction consistency value, the smooth consistency value, and the isolation suppression value. The scores are then arranged in chronological order according to the sliding window time, forming a consistency score sequence.

[0019] The above describes the acquisition of tidal level time-series data and roadway monitoring data for the target object. The following describes the application of consistent scoring-driven adaptive phase tracking to the tidal level time-series data to generate phase window labels. Please refer to [reference needed]. Figure 2 , Figure 2 This is a schematic diagram of the process for generating phase window labels provided in an embodiment of this application. Generating phase window labels specifically includes: The consensus score sequence is sorted in descending order and the difference is calculated to generate a difference sequence. The difference sequence is then processed to locate the maximum difference and truncate it to generate a set of confidence windows. Within the set of confidence windows, local rise and fall directions and morphological reference analysis are performed on the tide time series data to generate observation phase data. Then, based on the consistent scoring sequence, the observation phase data is weighted and the phase intervals are divided equally to generate phase window labels.

[0020] Among them, the difference sequence refers to the structural distribution change characterization data generated based on the internal sorting structure of the consistent rating sequence, which is used to characterize the degree of abrupt change between adjacent rating values ​​in the rating distribution; A trustworthy window set refers to a window-level data set generated from a consistent score sequence after sorting, differencing, and maximum transition truncation. Observational phase data refers to continuous phase observations generated after local morphological analysis of tide time series data within a set of confidence windows, used to characterize the current periodic position of the tide.

[0021] The above content will be described in detail below: The consensus score sequence is sorted in descending order and the difference is calculated to generate a difference sequence. The difference sequence is then processed to locate the maximum difference and truncate it, generating a set of confidence windows. The consistent score sequence is rearranged in descending order of value to obtain a sorted score sequence. The first consistent score value after sorting is the maximum value among all consistent scores, and each subsequent consistent score value is no greater than the previous one. The difference between each pair of adjacent consistent scores is calculated by subtracting the next consistent score value from the previous one, resulting in multiple adjacent differences. These differences are then arranged in the order of their positions in the sorted score sequence to form a difference sequence. For example, the first difference value is equal to the first sorted score value minus the second sorted score value, the second difference value is equal to the second sorted score value minus the third sorted score value, and so on. The remaining difference values ​​are obtained in the same way. Compare the magnitudes of all difference values ​​in the difference sequence to determine the largest difference value, and determine the position of the largest difference value in the difference sequence as the cutoff position. The cutoff position corresponds to the boundary where the change in the score values ​​between the two segments in the sorted score sequence is most obvious. Using the cutoff position as the cutoff basis, the sliding windows to which the score values ​​in the sorted score sequence belong before the cutoff position and the part in front of the cutoff position belong are divided into a set of confidence windows. Within the set of confidence windows, local rise and fall directions and morphological reference analyses are performed on the tide time series data to generate observed phase data. Then, based on the consistent scoring sequence, weighted corrections and equal phase interval divisions are applied to the observed phase data to generate phase window labels. The tide time series data segments corresponding to each confidence window are extracted in chronological order, and local rise and fall direction and morphological reference analysis is performed on each segment to generate observation phase data. The specific processing method is as follows: First, the adjacent tide time series data in the current confidence window are subtracted. The items with a value greater than zero in the subtraction result are recorded as rising items and the items with a value less than zero are recorded as falling items. Then, the number of rising items and the number of falling items are counted respectively, and the rise and fall direction values ​​are calculated. The rise and fall direction values ​​are obtained by subtracting the number of falling items from the number of rising items and then dividing by the sum of the number of rising items and the number of falling items. The window mean is obtained by summing the tide time series data within the same confidence window and dividing by the number of tide time series data within the window. The window mean is then subtracted from each tide time series data to obtain the deviation sequence. The product of adjacent terms in the deviation sequence is counted to obtain the sign flip number. The sign flip number is then divided by the length of the deviation sequence within the window to obtain the morphological inflection value. The window width is obtained by taking the difference between the maximum and minimum tide levels within the current reliable window. The morphological discrepancy is then obtained by dividing the sum of the absolute values ​​of the differences between each tide level time series data point and the window mean by the number of tide level time series data points within the window. This is used to obtain the rising and falling direction values. morphological transition value Window width and morphological discrete value Subsequently, morphological reference fusion processing is performed on the four data points to generate observation phase data. The specific calculation formula is as follows: ; In the formula, and Indicates the fusion weight coefficient. and Represents the normalization coefficient and the stability coefficient. Represents the arctangent function in the four quadrants. This represents the phase normalization function; all the data above have been normalized during the calculation. The fusion weight coefficients are generated by comparing the fluctuation levels of the rise / fall direction values ​​and the morphological terms within the confidence window set: First, rise / fall direction values ​​are extracted window by window within the confidence window set, and the morphological inflection value, window amplitude value, and morphological dispersion value are multiplied window by window to obtain the morphological term; then, the variances of the rise / fall direction values ​​and the morphological terms across all confidence windows are calculated; then, the reciprocals of their respective variances are taken as the initial weights of the components, giving smaller weights to components with large fluctuations and poor stability, and larger weights to components with small fluctuations and high stability; finally, the two initial weights are normalized to fix their relative proportions, and these are used as the fusion weight coefficients corresponding to the rise / fall direction values. Fusion weight coefficients corresponding to morphological terms ; The normalization coefficient is generated by the overall order of magnitude of the numerators within the set of confidence windows: after the fusion weight coefficients are determined, the absolute value of the numerator obtained by weighting the rise / fall direction value and the shape term with the fusion weights is calculated window by window, and then the absolute value of this numerator is averaged over all confidence windows to obtain the typical order of magnitude of the numerator; this typical order of magnitude is used as the normalization coefficient. ; The stability coefficient is generated by the relative relationship between the maximum amplitude of the vertical direction value and the normalization coefficient: the maximum absolute value of the vertical direction value is taken within the confidence window set, and then the normalization coefficient is divided by the maximum absolute value of the vertical direction value to obtain the stability coefficient. ; Within the set of confidence windows, determine the maximum and minimum consensus scores. Then, subtract the minimum consensus score from the current confidence window's consensus score and divide by the maximum and minimum consensus scores to obtain the current confidence window's score weight. Multiply the observed phase data by this score weight to obtain the weighted phase value. Simultaneously, multiply the observed phase data by 1 and subtract the score weight to obtain the retained phase value. Finally, add the weighted phase value and the retained phase value to obtain the corrected phase value. Sort all corrected phase values, take the difference between the largest and smallest phase values ​​after sorting as the total phase span, and then divide the total phase span evenly to obtain multiple phase intervals that are connected end to end. Each phase interval corresponds to a unique interval number. Then, match the corrected phase value of each confidence window with each phase interval one by one to determine the interval in which it belongs, and output the interval number as the phase window label of the confidence window.

[0022] This scheme generates a difference sequence by sorting the consistent score sequence in descending order and calculating the difference. It then performs maximum difference location and truncation on the difference sequence to generate a set of confidence windows. This clearly separates windows with significantly high coupling consistency from disturbed windows in the score distribution, reducing the pulling bias of low-confidence samples on the phase tracking benchmark. Within the set of confidence windows, local rise and fall directions and morphological reference analysis are performed on the tide time series data to generate observed phase data. This transforms local tide fluctuations into a stable continuous phase representation, improving the temporal coherence of phase identification. Based on the consistent score sequence, the observed phase data undergoes weighted correction and equal phase interval division to generate phase window labels. This suppresses phase jumps caused by abnormal fluctuations and unifies the phase segmentation scale, improving the consistency and comparability of subsequent phase window discrimination.

[0023] The above describes the application of consistent scoring-driven adaptive phase tracking to tidal time-series data to generate phase window labels. The following describes the grouping and in-phase projection of roadway monitoring data based on these phase window labels to generate a hydraulic potential sequence. Please refer to [link / reference]. Figure 3 , Figure 3This is a schematic diagram of the process for generating a hydraulic potential sequence provided in an embodiment of this application. Generating the hydraulic potential sequence specifically includes: Within the set of confidence windows, the tunnel monitoring data are grouped according to the phase window labels, and the grouping results are weighted and summed according to the consistent scoring sequence and the observed phase data to generate a hydraulic projection vector. The grouping results are subjected to inner product projection with the hydraulic projection vector and then summarized to generate a hydraulic potential sequence.

[0024] Among them, the grouping result refers to the classification and aggregation of monitoring samples that meet the set of confidence windows under the same sampling order coordinate system, based on their corresponding phase window label numbers, to form a data structure with the phase window number as the main index and the set of monitoring sample vectors as the content; The hydraulic projection vector is a direction vector used to extract the response component in the same direction as the tidal drive within the phase window.

[0025] The above content will be described in detail below: Within the set of confidence windows, the tunnel monitoring data are grouped according to the phase window labels, and the grouping results are weighted and summed based on the consistent scoring sequence and the observed phase data to generate a hydraulic projection vector. Within the set of trustworthy windows, the roadway monitoring data is grouped using the phase window label as the grouping index. Specifically, each piece of roadway monitoring data is assigned to the same phase window group according to its corresponding phase window label to form multiple phase window groups, and the phase window group is output as the grouping result. For each phase window group, two types of weights are generated for weighting. The first type of weight comes from the consistent score sequence. The minimum and maximum values ​​of the consistent score sequence within the phase window group are taken, and each consistent score value is normalized. The normalized score is calculated by subtracting the minimum consistent score value within the group from the current score value, and then dividing by the difference between the maximum and minimum consistent score values ​​within the group, thereby obtaining the score weight corresponding to each roadway monitoring data within the phase window group. The second type of weight comes from the observed phase data. Phase consistency mapping is performed on the observed phase data corresponding to each roadway monitoring data in the phase window group. The phase consistency mapping process is to obtain the phase weight by cosine mapping the observed phase data. The phase weight is used to characterize the degree of fit between the observed phase and the central phase of the phase window group, and the sample with a higher degree of fit has a larger weight. The above two types of weights are fused. The calculation process of the fusion weight is to multiply the score weight and the phase weight to obtain the sample fusion weight of each roadway monitoring data. Then, the roadway monitoring data in the phase window group are weighted and summed to generate a hydraulic projection vector. The calculation process is to multiply the monitoring vector of each roadway monitoring data in the phase window group with its corresponding sample fusion weight to obtain a weighted monitoring vector. All weighted monitoring vectors are added item by item according to the same vector dimension to obtain a weighted sum vector. Then, the weighted sum vector is normalized according to its vector magnitude to obtain the hydraulic projection vector. Perform an inner product projection operation on the grouping results and the hydraulic projection vectors, and summarize them to generate a hydraulic potential sequence. The specific calculation formula is as follows: ; In the formula, Indicates the first The tunnel monitoring data in the group results of each sample Indicates the relationship with the first The hydraulic projection vectors corresponding to the group results of each sample are all normalized during the calculation.

[0026] This scheme groups roadway monitoring data within a set of reliable windows based on phase window labels. It then performs a weighted summation of the grouping results based on the consistency score sequence and observed phase data to generate a hydraulic projection vector. This merges monitoring responses within the same phase window under a unified phase benchmark. Furthermore, it utilizes the correlation between consistency strength and phase to highlight the dominant tidal-driven response, weaken stray disturbances and out-of-phase responses, and improve the accuracy of the projection direction in representing hydraulic changes. Finally, it performs an inner product projection operation on the grouping results and the hydraulic projection vector, summarizing them to generate a hydraulic potential sequence. This compresses the multidimensional monitoring response into a continuous and comparable single sequence, enhancing the distinguishability and continuity of hydraulic response intensity between different phase windows.

[0027] The above describes the grouping and phase window projection of roadway monitoring data based on phase window labels to generate a hydraulic potential sequence. The following describes the in-phase component separation processing of roadway monitoring data, consistency score sequences, and hydraulic potential sequences to generate an oxygen-fuel potential sequence. Please refer to [link / reference]. Figure 4 , Figure 4 This is a schematic diagram of the process for generating an oxygen potential sequence provided in an embodiment of this application. Generating the oxygen potential sequence specifically includes: Within the set of confidence windows, the tunnel monitoring data is weighted and corrected according to the consistent scoring sequence to obtain weighted monitoring data, and the weighted monitoring data is projected in the same direction along the hydraulic potential sequence. The same-direction projection results are subtracted from the weighted monitoring data to obtain the differential monitoring data. The differential monitoring data are then subjected to single-axis mapping in the sampling order and summarized to generate the oxygen potential sequence.

[0028] Among them, weighted monitoring data refers to the monitoring vector sequence formed after performing consistency weighting correction on the sample dimension of the roadway monitoring data within the set of confidence windows, with the consistent scoring sequence as the weight reference; The co-projection result refers to the vector result obtained by mapping the components of the weighted monitoring data in the reference direction with the hydraulic response direction corresponding to the hydraulic potential sequence as the reference direction within the set of confidence windows. Differential monitoring data refers to the set of differential components that are orthogonal or weakly correlated with the direction of hydraulic potential, remaining after deducting the common-mode components in the weighted monitoring data that change in the same direction as the hydraulic potential sequence within the confidence window set.

[0029] The above content will be described in detail below: Within the confidence window set, the tunnel monitoring data is weighted and corrected according to the consistent scoring sequence to obtain weighted monitoring data, which is then projected in the same direction along the hydraulic potential sequence. The roadway monitoring data and the consistent score sequence are matched with the window index of the confidence window set to ensure that each monitoring sampling point in each confidence window has a unique corresponding consistent score value. Then, the roadway monitoring data is weighted and corrected to generate weighted monitoring data. The calculation process is as follows: for any sampling point in each confidence window, the roadway monitoring data value of the sampling point is recorded as the monitoring value, and the consistent score value corresponding to the sampling point is recorded as the score value. The monitoring value is scaled by the score weight to obtain the weighted monitoring value. The weighted monitoring value is obtained by multiplying the monitoring value by the score weight. All weighted monitoring values ​​in the confidence window set are summarized and output as weighted monitoring data in the original sampling order. Using a confidence window as a unit, adjacent differences are taken from the hydraulic potential sequence within the confidence window, and the difference sign is used as the same-direction reference sign. Then, adjacent differences are calculated from the weighted monitoring values ​​of each sampling point within the confidence window, and the monitoring difference sign is obtained. When the monitoring difference sign is consistent with the same-direction reference sign, the weighted monitoring value of the sampling point is retained as the same-direction projection value. When the two are inconsistent, the opposite number of the weighted monitoring value of the sampling point is taken as the same-direction projection value, so that the sampling point is consistent with the hydraulic potential sequence in the same-direction projection space. Finally, all the same-direction projection values ​​in the confidence window set are summarized and output as the same-direction projection result according to the sampling order. The co-projection results are subtracted from the weighted monitoring data to obtain the differential monitoring data. The differential monitoring data are then subjected to single-axis mapping in the sampling order and summarized to generate the oxygen potential sequence. For each sampling location, a corresponding component subtraction process is performed, that is, each component in the weighted monitoring data of the sampling location is subtracted from the corresponding component in the same direction projection result of the same sampling location to obtain the residual component group of the sampling location, and the residual component group is determined as the differential mode monitoring data of the sampling location. The residual total for each sampling location is obtained by summing the differential monitoring data at each sampling location. Then, all differential monitoring data are summed according to component location to obtain the cumulative residual for each component. All cumulative residuals are arranged into a single-axis directional quantity group in the original component order. The single-axis directional quantity group is subjected to scale unification processing, specifically by taking the square root of the sum of the squares of the cumulative residuals of each component to obtain the total directional scale. Then, the cumulative residual of each component is divided by the total directional scale to obtain the unit directional value corresponding to each component. All unit directional values ​​together constitute the single-axis mapping reference. Based on this, single-axis mapping is performed on the differential monitoring data of each sampling location. Specifically, each residual component of the sampling location is multiplied by the corresponding unit directional value in the single-axis mapping reference, and all product results are summed to obtain the single-axis mapping value of the sampling location. The single-axis mapping value is the oxygen potential value of the sampling location. The calculation process is to multiply each residual component by the corresponding unit directional value one by one and then sum them. Finally, all oxygen potential values ​​are arranged and summarized according to the sampling order to generate an oxygen potential sequence.

[0030] This scheme obtains weighted monitoring data by weighting and correcting the roadway monitoring data within a set of reliable windows based on a consistent scoring sequence. This allows monitoring segments with higher reliability to occupy a higher proportion in subsequent processing, while reducing the interference of fluctuating anomaly segments on the overall characterization results. This improves the stability of the monitoring data in representing the true changes in the downhole state. By projecting the weighted monitoring data in the same direction along the hydraulic potential sequence, common-mode components consistent with hydraulic disturbances can be extracted from the weighted monitoring data, providing a concentrated characterization of tidal-driven runoff response and a clear reference for subsequent separation of non-hydraulic dominant components. Subtracting the same-direction projection result from the weighted monitoring data yields differential-mode monitoring data, which effectively isolates response components jointly driven by hydraulic changes. This allows the remaining data to more prominently reflect independent change trends related to spontaneous combustion evolution, thereby reducing the confusion of tidal disturbances in disaster identification. Finally, single-axis mapping is performed on the differential-mode monitoring data according to the sampling order and the data is summarized to generate a combustion oxygen potential sequence. This transforms discrete differential-mode information into continuous and comparable trend quantities, providing a stable data foundation for the unified characterization of combustion oxygen change states.

[0031] The above describes the process of performing co-directional component separation on roadway monitoring data, consistent scoring sequences, and hydraulic potential sequences to generate an oxygen-fuel potential sequence. The following describes the process of performing phase differential gain construction on the hydraulic potential and oxygen-fuel potential sequences to generate an oxygen-inducing gain sequence, specifically including: Based on the phase window labels, the hydraulic potential sequence and the oxygen-fuel potential sequence are segmented and adjacent difference operations are performed to generate the hydraulic difference sequence and the oxygen-fuel difference sequence respectively. The absolute value of the hydraulic difference sequence is taken and the mean is calculated. The oxygen-fuel difference sequence is then normalized to the same scale based on the mean calculation result. The smallest positive value after taking the absolute value of the hydraulic difference sequence is extracted as the non-zero term. The result of the same-scale normalization is divided by the sum of the hydraulic difference sequence and the non-zero term, and then summarized to generate the oxygenation gain sequence.

[0032] Among them, the hydraulic differential sequence refers to the sequence formed by performing differential operations on the hydraulic potential values ​​at two adjacent sampling times in the hydraulic potential sequence under the same sampling order; The oxygen-fuel differential sequence refers to sequence data used to characterize the magnitude and direction of change in oxygen-fuel potential between adjacent sampling times; The mean calculation result refers to the scalar result obtained by averaging the set of absolute difference values ​​obtained by taking the absolute values ​​of each difference term in the hydraulic difference sequence within the same phase window; Same-scale normalization results refer to the standardized results obtained by mapping the numerical amplitude of the oxygen-fuel differential sequence to the same dimension and order of magnitude range as the hydraulic differential sequence under phase window segmentation conditions.

[0033] The above content will be described in detail below: Based on the phase window labels, adjacent difference operations are performed on the hydraulic potential sequence and the oxygen-fuel potential sequence in segments, respectively, to generate the hydraulic difference sequence and the oxygen-fuel difference sequence: The correspondence between phase window labels, hydraulic potential sequence and oxygen fuel potential sequence in the same sampling order is used as the basis for data processing. Continuous potential values ​​with the same sampling order and the same phase window label are grouped into the same phase window segment, thereby obtaining the hydraulic segment set and the oxygen fuel segment set. Within the hydraulic segment set, two adjacent hydraulic potential values ​​are extracted according to the original sampling order. The hydraulic potential value at the next sampling position is subtracted from the hydraulic potential value at the previous sampling position, and the difference is used as the hydraulic difference value corresponding to the adjacent position. All hydraulic difference values ​​are arranged according to the original order within each phase window segment to form a hydraulic difference sequence. Within the oxygen-fuel segment set, the same processing method as the hydraulic segment set is adopted. The oxygen potential values ​​of two adjacent samples are extracted according to the original sampling order. The oxygen potential value of the sampled position is subtracted from the oxygen potential value of the previous sampled position, and the difference is used as the oxygen difference value corresponding to the adjacent position. All oxygen difference values ​​are arranged according to the original order within each phase window segment to form an oxygen difference sequence. The absolute values ​​of the hydraulic difference sequences are taken, and the mean is calculated. Based on the mean calculation results, the oxygen-fuel difference sequences are then normalized using the same scale. After taking the absolute value of the hydraulic difference sequence, the mean is calculated. Each difference value of the oxygen-fuel difference sequence is taken as the normalized object. The difference value is divided by the mean calculation result to obtain the same-scale normalization result of the corresponding term. The smallest positive value after taking the absolute value of the hydraulic difference sequence is extracted as the non-zero term. The result of the same-scale normalization is divided by the sum of the hydraulic difference sequence and the non-zero term, and then summarized to generate the oxygenation gain sequence.

[0034] This scheme performs adjacent difference operations on the hydraulic potential sequence and the oxygen-fuel potential sequence in segments according to the phase window labels, generating corresponding hydraulic difference sequences and oxygen-fuel difference sequences. This can transform static quantities within different phase windows into dynamic quantities, eliminating the interference of the original amplitude superposition on the phase correlation determination. After taking the absolute value of the hydraulic difference sequence, the mean is calculated, and the oxygen-fuel difference sequence is normalized to the same scale based on the mean calculation result. This can unify the dimensions and fluctuation scale of the two types of difference data, improve the consistency of gain representation, extract the minimum positive value after taking the absolute value of the hydraulic difference sequence as the non-zero term, and divide the same-scale normalization result by the sum of the hydraulic difference sequence and the non-zero term before summing to generate the oxygen-inducing gain sequence. This can avoid denominator instability and highlight the oxygen-fuel increment corresponding to a unit hydraulic change.

[0035] The above describes the phase differential gain construction performed on the hydraulic potential sequence and the oxygen potential sequence to generate the oxygen-inducing gain sequence. The following describes the quantile threshold self-calibration gating performed on the oxygen-inducing gain sequence and the phase window labels to generate a set of gating windows, specifically including: The oxygen-inducing gain sequences are grouped according to the phase window labels to obtain the gain subsequences of each phase window. The gain subsequences are then sorted in descending order and subjected to adjacent difference operations to generate the gain difference sequence. In the gain difference sequence, the index with the largest difference is located as the transition position data. Based on the transition position data, the descending gain subsequence is located and the median is extracted to generate the corresponding phase window gating threshold table. The gain subsequences of each phase window are compared with the corresponding gating threshold table and classified according to their proportions to generate a set of gating windows.

[0036] Among them, the gain subsequence refers to the finite ordered numerical sequence of all oxygen induction gain sample values ​​corresponding to the same phase window number, arranged in the original time order or the sorted order. Gain difference sequence refers to the sequence data obtained by performing a difference operation on the gain values ​​of two adjacent sorting positions after the gain subsequence corresponding to a certain phase window has been sorted in descending order; Transition position data refers to the structured positioning data corresponding to the position with the largest difference between two adjacent sorted values ​​in the sorted sequence obtained after sorting the corresponding gain subsequences in descending order within the same phase window. A gating threshold table is a phase window-level data table generated by structurally representing the oxygen induction gain distribution in the phase window dimension.

[0037] The above content will be described in detail below: The oxygen-inducing gain sequences are grouped according to phase window labels to obtain the gain subsequences for each phase window. These subsequences are then sorted in descending order and subjected to adjacent difference operations to generate the gain difference sequence. Based on the phase window label, the oxygen induction gain sequence is merged according to the phase window to which the sampling point belongs: For any phase window, all sampling point numbers identified as that phase window in the phase window label are traversed, and the corresponding oxygen induction gain values ​​are extracted from the oxygen induction gain sequence according to these numbers and the gain subsequence of that phase window is formed according to the extraction order, so as to obtain the gain subsequence corresponding to each phase window. For each phase window, a descending order operation is performed on the gain subsequence to generate a sorted subsequence. Specifically, all gain values ​​in the gain subsequence of the phase window are rearranged in descending order, and the descending sorted subsequence is output. Adjacent difference operation is performed on the sorted subsequence to generate a gain difference sequence. Specifically, the difference is calculated item by item for each adjacent item in the sorted subsequence, the difference value at the corresponding position is obtained by subtracting the next item from the previous item, and all difference values ​​are arranged in the order of calculation to form a gain difference sequence. In the gain difference sequence, the index with the largest difference is located as the transition position data. Based on the transition position data, the descending gain subsequence is located and the median is extracted to generate the corresponding phase window gating threshold table. The gain difference sequence is subjected to maximum difference location processing. Specifically, all difference values ​​are compared one by one and the index of the gain difference value with the largest value is taken as the transition position data. The transition position data represents the boundary position of the transition from the high value dense area to the low value flat area within the gain subsequence. After obtaining the transition position data, the descending gain subsequence is subjected to location processing according to the index. That is, the gain value from the first gain value to the first transition position is determined as the front-end location sequence, and the gain value after the transition position is determined as the back-end location sequence. The front-end location sequence represents the high gain concentrated area, and the back-end location sequence represents the normal gain area. The median extraction process is performed on the preceding positioning sequence. Specifically, the number of elements in the preceding positioning sequence is first counted. If the number of elements is odd, the gain value at the middle position is taken as the median threshold data. The middle position is obtained by adding 1 to the number of elements and dividing by 2. If the number of elements is even, the average of the two middle gain values ​​is taken as the median threshold data. The last gain value of the preceding positioning sequence is also extracted as the transition threshold data. Finally, the phase window number, transition position data, transition threshold data and median threshold data are combined according to the same entry to generate the gating threshold table corresponding to the phase window. The gain subsequences of each phase window are compared with the corresponding gating threshold table and categorized according to their proportions to generate a set of gating windows: Within the same phase window, each gain value of the phase window is compared with the transition threshold data and the median threshold data to form a high-order comparison result sequence and a median comparison result sequence. When a gain value is greater than or equal to the transition threshold data, it is recorded as a high-order hit value in the high-order comparison result sequence. When a gain value is less than the transition threshold data but greater than or equal to the median threshold data, it is recorded as a median hit value in the median comparison result sequence. When a gain value is less than the median threshold data, it is recorded as a low-order hit value. After obtaining the high-order, median, and low-order hit values, the three types of hit values ​​are accumulated to obtain the high-order hit count, median hit count, and low-order hit count. These three are then divided by the total data volume of the phase window gain subsequence to calculate the high-order percentage, median percentage, and low-order percentage. The phase window is categorized into three categories: high-position percentage (high-position hit count divided by total data volume), median percentage (median hit count divided by total data volume), and low-position percentage (low-position hit count divided by total data volume). After obtaining these three percentages, they are compared within the same window. The largest percentage is taken as the dominant percentage for that phase window. The classification result of the phase window is then output based on the category to which the dominant percentage belongs. Specifically, a "no-displacement window" flag is output when the dominant percentage corresponds to a high-position percentage, a "limited-displacement window" flag is output when the dominant percentage corresponds to a median percentage, and a "suitable-displacement window" flag is output when the dominant percentage corresponds to a low-position percentage. Finally, the classification results of all phase windows are summarized by phase window number and assigned to the "no-displacement window" subset, the "limited-displacement window" subset, and the "suitable-displacement window" subset, resulting in a gated window set containing these three subsets.

[0038] This scheme groups the oxygen induction gain sequences according to phase window labels to obtain the gain subsequences of each phase window. It then performs descending ordering and adjacent difference operations on the gain subsequences to generate gain difference sequences. This concentrates the oxygen induction gain changes within different phase windows onto the same comparison scale, highlighting the distribution breakpoints between high-gain samples and ordinary samples, and improving the clarity of risk stratification within the phase window. The scheme locates the sequence with the largest difference in the gain difference sequence as the transition position data. Based on the transition position data, it locates and extracts the median of the descending gain subsequences to generate a gating threshold table for the corresponding phase window. This allows the threshold to adaptively change with the actual gain distribution of each phase window, reducing gating bias caused by fixed thresholds. The scheme compares the gain subsequences of each phase window with the corresponding gating threshold table and categorizes them according to their proportions to generate a set of gating windows, which can stably distinguish between prohibited, restricted, and suitable drainage windows.

[0039] The above describes the quantile threshold self-calibration gating of the oxygen-inducing gain sequence and phase window labels to generate a set of gated windows. The following describes a sequence graph search with gating constraints performed on the set of gated windows, the oxygen potential sequence, and the oxygen-inducing gain sequence to generate a sequence scheme table, specifically including: Perform window position enumeration constraints on the phase window labels and gated window set to generate an action candidate table; Intra-window dominant gain mapping is performed on the hydraulic potential sequence, the oxygen potential sequence, and the oxygen inducement gain sequence to generate an action gain table. Then, gain quantile cost quantization is performed on the action gain table to generate an action cost table. The candidate action table, action benefit table, and action cost table are evaluated by candidate reduction scoring to generate an action scoring table. Then, the action scoring table is subjected to adjacent window consistency constraints based on the phase window labels to generate a sequence scheme table.

[0040] Among them, the action candidate table refers to a two-dimensional enumeration constraint table with the row index of each phase window number in the phase window label and the column index of the standard action set; The action benefit table is a two-dimensional structured data table with phase window labels as the organizational index and three types of candidate actions as columns; The action cost table is a multidimensional data table constructed with phase window labels as the organizing index and action type as the column item. The action scoring table is a two-dimensional data structure with phase window labels as row indexes and action types as column indexes.

[0041] The above content will be described in detail below: Window position enumeration constraints are applied to the phase window labels and gated window sets to generate an action candidate table: The phase window labels are subjected to window expansion processing. Specifically, all different numbers are extracted according to the order of appearance of the phase window numbers to form a phase window position sequence. Then, the phase window position sequence is sorted to preserve the order and remove duplicates to obtain a window position base table in which each window position appears only once. The calculation process is as follows: the phase window numbers of adjacent sampling positions are compared side by side, and the position where the number changes is recorded as the starting point of the new window position. The window position base table is formed by summarizing the numbers corresponding to the starting points of each new window position. Then, the gated window set is subjected to gate expansion processing. The gated categories are matched with the phase window numbers to obtain a gated correspondence table. The calculation process is as follows: the phase window numbers of the prohibited, restricted, and suitable windows in the gated window set are extracted respectively. Then, the phase window numbers are paired with the corresponding gated categories to form gated correspondence items. All gated correspondence items are arranged in the order of phase window numbers to form a gated correspondence table. The window position enumeration constraint processing is performed, specifically by registering each window position in the window position base table with the corresponding gate control item with the same number in the gate control correspondence table to generate a window position constraint table. The calculation process is as follows: using the window position number as the association key, the phase window number of each window position is concatenated with the gate control category one by one to obtain a ternary correspondence consisting of window position number - phase window number - gate control category. Further, based on the window position constraint table, the action mapping processing is performed to constrain and associate the predefined inertia action, blocking action, and drainage action with the gate control category to generate the action allowable set for each window position. The action allowable set for the prohibited drainage window consists of inertia action and blocking action, the action allowable set for the limited drainage window consists of inertia action, blocking action, and limited strong drainage action, and the action allowable set for the suitable drainage window consists of inertia action, blocking action, and drainage action. The calculation process of the action allowable set is as follows: the gating category is converted into an action open identifier, the action open identifier is then screened and combined with the standard set of three types of actions to obtain the candidate action item that is uniquely corresponding to each window position, and finally the window position number, phase window number, gating category, and candidate action item of all windows are summarized in window position order to generate an action candidate table. Intra-window dominant gain mapping is performed on the hydraulic potential sequence, oxygen potential sequence, and oxygen-inducing gain sequence to generate an action gain table. Then, gain quantization is performed on the action gain table to generate an action cost table. The absolute values ​​of each hydraulic potential segment within the current phase window are taken and summed to obtain the hydraulic cumulative amount. The calculation process involves summing the absolute values ​​of all hydraulic potential values ​​within the window to obtain a single value. Then, the absolute values ​​of each oxygen potential segment within the current phase window are taken and summed to obtain the oxygen cumulative amount. The calculation process involves summing the absolute values ​​of all oxygen potential values ​​within the window to obtain a single value. Finally, the absolute values ​​of each oxygen induction gain segment within the current phase window are taken and summed to obtain the gain cumulative amount. The calculation process involves summing the absolute values ​​of all oxygen induction gain values ​​within the window to obtain a single value. The total amount within the window is obtained by adding the cumulative hydraulic amount, the cumulative oxygen amount, and the cumulative gain amount. The calculation process is to directly sum the three to obtain a single value, and then divide the cumulative hydraulic amount by the total amount within the window to obtain the hydraulic percentage, divide the cumulative oxygen amount by the total amount within the window to obtain the oxygen percentage, and divide the cumulative gain amount by the total amount within the window to obtain the gain percentage. The dominant benefit mapping within the execution window generates drainage benefit value, inerting benefit value, and closure benefit value respectively. The drainage benefit value is calculated by multiplying the hydraulic proportion by one and subtracting the gain proportion, which represents the drainage benefit intensity when the hydraulic component is dominant and the oxygen induction gain is relatively low. The inerting benefit value is calculated by multiplying the combustion oxygen proportion by the gain proportion, which represents the inerting benefit intensity when the combustion oxygen component and the oxygen induction component are simultaneously dominant. The closure benefit value is calculated by first adding the hydraulic proportion and the combustion oxygen proportion and then dividing by two to obtain the coupling proportion, and then multiplying the coupling proportion by the gain proportion, which represents the closure benefit intensity when the hydraulic and combustion oxygen components have a high degree of presence and the oxygen induction gain participation is high. The drainage, inerting, and closure benefits corresponding to each phase window are arranged by window number to generate an action benefit table. Gain quantification is then performed on this table. Specifically, the oxygen-inducing gain segments corresponding to the current phase window are first sorted by value from smallest to largest to form a gain ranking sequence within the window. Then, the total number of samples for the oxygen-inducing gain segments of the current phase window is counted. Subsequently, the drainage, inerting, and closure benefits of the current phase window in the action benefit table are taken as benefit reference values, and positions within the gain ranking sequence that are not greater than the corresponding benefit reference value are located. The calculation process involves counting the positions within the ranking sequence that are not greater than the corresponding benefit reference value. The number of gain reference values ​​is divided by the total number of samples to obtain the gain quantile value of the corresponding action. Then, the cost value of the corresponding action is obtained by subtracting the gain quantile value of the drainage benefit value from the gain quantile value of the drainage benefit value, the cost value of the inertia benefit value from the gain quantile value of the inertia benefit value, and the cost value of the blocking benefit value from the gain quantile value of the blocking benefit value. In this way, the position of the benefit value in the gain distribution within the window is directly converted into the cost value. Finally, the drainage cost value, inertia cost value, and blocking cost value corresponding to each phase window are arranged according to the window number to generate an action cost table. The action candidate table, action benefit table, and action cost table are evaluated by candidate reduction scoring to generate an action score table. Then, based on the phase window labels, the action score table is subjected to adjacent window consistency constraints to generate a sequence scheme table. Obtain the action candidate table, action benefit table, action cost table, and phase window labels, and align the fields with the same phase window number and the same action number to form a data matrix corresponding to each window and each action. The action candidate list is subjected to candidate reduction processing, specifically as follows: For all action candidates in each phase window, first count the total number of candidates in that phase window to obtain the total number of candidates, then calculate the ratio of the candidate identifier value of a single action candidate to the total number of candidates to obtain the initial proportion value of that action candidate, then sum all the initial proportion values ​​in the same phase window and divide the single initial proportion value by the sum to obtain the candidate normalization value, then calculate the candidate discrete value based on the average value of all candidate normalization values ​​in the same phase window, the candidate discrete value is the absolute value of the difference between the single candidate normalization value and the average value, then sum all the candidate discrete values ​​and divide the single candidate discrete value by the sum to obtain the candidate reduction coefficient; The action reward table undergoes reward standardization, specifically: Within each phase window, all action reward values ​​are extracted, the maximum and minimum rewards within that phase window are calculated, and the minimum reward value for each individual action is subtracted from the minimum reward value and then divided by the difference between the maximum and minimum rewards to obtain a normalized reward value. Simultaneously, the action cost table undergoes cost standardization, specifically: Within each phase window, all action cost values ​​are extracted, the maximum and minimum costs within that phase window are calculated, and the minimum cost value for each individual action is subtracted from the minimum cost value and then divided by the difference between the maximum and minimum costs to obtain a normalized cost value. Based on these, candidate reduction coefficients, normalized reward values, and normalized cost values ​​within the same phase window are standardized. The candidate reduction scoring process is as follows: First, the candidate reduction coefficient is multiplied by the normalized value of the revenue to obtain the candidate revenue amount. Then, the candidate reduction coefficient is multiplied by the normalized value of the cost to obtain the candidate cost amount. Then, the candidate revenue amount is subtracted from the candidate cost amount to obtain the original score value of the corresponding action within the phase window. Further, window-by-window normalization processing is performed on all the original score values. That is, all the original score values ​​are extracted in each phase window, the maximum value and the minimum value of the original score are obtained, and the minimum value of the original score is subtracted from the single original score value and then divided by the difference between the maximum value and the minimum value of the original score to obtain the action score value. All action score values ​​are arranged according to the phase window number and the action number to generate an action score table. The action scoring table is processed according to the phase window labels to achieve neighboring window consistency constraints. Specifically, the action scoring values ​​of two adjacent phase windows are extracted according to the order of the phase window labels. The absolute value of the difference between the scores of the same action in the previous phase window and the next phase window is calculated to obtain the action neighboring window difference value. Then, the neighboring window differences of all actions in the same phase window are summed, and the neighboring window difference value of a single action is divided by the sum to obtain the action neighboring window difference ratio. Then, the corresponding action neighboring window difference ratio is subtracted from the score value of a single action in the phase window to obtain the neighboring window constraint score value. The neighboring window constraint score value indicates the degree of consistency between the action and the action distribution of adjacent phase windows while maintaining the scoring advantage of the current window. Then, all neighboring window constraint score values ​​in each phase window are sorted in descending order, and the action number corresponding to the first position in the sort is taken as the target action number of the phase window. The target action number is combined with the corresponding phase window label to obtain the single window action item. Finally, all single window action items are concatenated according to the order of the phase window labels to generate a sequence scheme table.

[0042] This scheme generates an action candidate table by enumerating window positions on phase window labels and gating window sets. This allows for the early elimination of action types incompatible with the current window gating state, ensuring consistency between candidate actions and phase window positions. This reduces invalid search volume in subsequent sequence generation and improves the standardization of action placement. It performs intra-window dominant benefit mapping on hydraulic potential sequences, oxygen potential sequences, and oxygen-inducing gain sequences to generate an action benefit table. Furthermore, it quantifies the gain quantile cost of the action benefit table to generate an action cost table. This transforms the advantages and burdens of actions within different windows into a comparable and unified expression, enhancing the clarity of action priority judgment. Finally, it performs candidate reduction scoring on the action candidate table, action benefit table, and action cost table to generate an action score table. Based on the phase window labels, it imposes adjacent window consistency constraints on the action score table to generate a sequence scheme table. This reduces abrupt action changes between adjacent windows, making the output sequence more continuous and stable.

[0043] The above describes a sequence graph search for gating constraints on the gated window set, oxygen potential sequence, and oxygen induction gain sequence to generate a sequence scheme table. The following describes the generation of phase-bearing water-blocking curtain parameters on the sequence scheme table, gated window set, tidal time series data, and oxygen induction gain sequence to generate the control results, specifically including: Phase bearing capacity mapping is performed on the observation phase data and hydraulic potential sequence corresponding to the tidal time series data to generate a phase bearing capacity table. The phase bearing capacity table and oxygenation gain sequence are then encapsulated by quantile mapping to generate a grouting curtain parameter package. The phase window label, gated window set, sequence scheme table and grouting curtain parameter package are integrated into a phase window encapsulation to generate control results.

[0044] Among them, the phase load capacity table refers to a phase-level load capacity characteristic data table constructed with phase window labels as the index unit; The grouting curtain parameter package refers to a data structure used to describe the load-bearing layer structure and material grade combination of a one-piece molded water-blocking curtain under various phase window conditions.

[0045] The above content will be described in detail below: Phase bearing capacity mapping is performed on the observed phase data and hydraulic potential sequence corresponding to the tidal time series data to generate a phase bearing capacity table. The phase bearing capacity table is then screened for limited bearing capacity to generate peak bearing capacity phase data. The peak bearing capacity phase data and oxygen-inducing gain sequence are encapsulated in a locked three-layer parameter package to generate the grouting curtain parameter package: Phase carrying capacity mapping is performed on the observed phase data and hydraulic potential sequence to generate a phase carrying capacity table. Specifically, the observed phase data is divided into phase groups for all sampling times, the phase range is divided into equal phase segments, and each phase segment is assigned a phase segment number. Then, the carrying capacity value is calculated by taking the hydraulic potential sequence segment corresponding to each phase segment. The carrying capacity value is obtained by multiplying the mean hydraulic potential value and the hydraulic potential fluctuation within the phase segment. The mean hydraulic potential value is the arithmetic mean of all hydraulic potential values ​​within the phase segment, and the hydraulic potential fluctuation is the average of the absolute values ​​of the differences between each hydraulic potential value and the mean hydraulic potential within the phase segment. This realizes the phase carrying capacity mapping that binds the phase position to the hydraulic scour intensity, and the phase carrying capacity table is obtained by summarizing each phase segment number and its carrying capacity value. Subsequently, a restricted load screening was performed on the phase load capacity table to generate peak load phase data. Specifically, the load capacity values ​​in the phase load capacity table were arranged in descending order to form a load capacity sorting sequence, and the difference between adjacent elements in the load capacity sorting sequence was calculated to form a load capacity difference sequence. The maximum transition position in the load capacity difference sequence was located as the restricted cutoff position, and the set of phase segments before the restricted cutoff position was determined as the candidate high load capacity set to avoid insufficient differentiation due to full selection. At the same time, the load capacity density of each phase segment in the candidate high load capacity set was calculated. The load capacity density was obtained by the ratio of the load capacity value of the phase segment to the median of the load capacity values ​​in the candidate high load capacity set. The phase segment with the highest load capacity density in the candidate high load capacity set was selected as the peak load capacity phase segment, and its phase segment number and corresponding observation phase range were output to form the peak load capacity phase data. A three-layer parameter encapsulation method is applied to the peak bearing capacity phase data and the oxygen-inducing gain sequence to generate a grouting curtain parameter package. Specifically, the grouting window is locked using the peak bearing capacity phase data, and the gain segment corresponding to the grouting window is extracted from the oxygen-inducing gain sequence. The gain level of the gain segment is calculated, and the gain level is obtained by the ratio of the average gain of the gain segment to the average gain of the total gain. Based on this, a three-layer parameter set is generated and encapsulated into a grouting curtain parameter package. The first layer is the outer scour layer parameter, which is determined by the bearing capacity value and gain level of the peak bearing capacity phase segment. The larger the bearing capacity value and the larger the gain level, the higher the viscosity level and the shorter the gel time level of the outer scour layer, so as to shorten the turbulent time of the grout in the high scour window. The second layer consists of parameters for the inner dense layer. These parameters are determined by the average carrying capacity and gain level of adjacent phase segments of the peak carrying capacity phase segment. The average carrying capacity of adjacent phase segments is obtained by the arithmetic mean of the carrying capacity values ​​of the left and right adjacent phase segments of the peak carrying capacity phase segment. The higher the average carrying capacity of adjacent phase segments, the higher the diffusion coverage level of the inner dense layer, in order to form a continuous low-permeability core. The third layer consists of the orifice sealing layer parameters, which are determined by the gain fluctuation of the gain segment and the phase range width of the peak carrying phase segment. The gain fluctuation is the average of the absolute values ​​of the differences between each gain value in the gain segment and the mean of the gain segment. The phase range width is the angular span of the phase segment corresponding to the peak carrying phase data. The larger the gain fluctuation or the larger the phase range width, the higher the sealing strength level of the orifice sealing layer is to reduce the risk of flow around the boundary. The three-layer parameter set, the phase segment number of the grouting window, and the observation phase range are locked and encapsulated to output the grouting curtain parameter package.

[0046] The phase window label, gated window set, sequence scheme table and grouting curtain parameter package are integrated into a phase window encapsulation to generate control results.

[0047] This scheme maps the observed phase data and hydraulic potential sequence corresponding to the tidal time series data to generate a phase carrying capacity table. This establishes a quantitative correspondence between the tidal cycle position and the downhole hydraulic response intensity, thereby accurately identifying phase windows with high carrying capacity. The phase carrying capacity table and oxygen induction gain sequence are hierarchically encapsulated through quantile mapping to generate a grouting curtain parameter package. This allows for simultaneous differentiation between water-blocking carrying capacity requirements and oxygen induction risk levels, resulting in more precise matching between curtain layers and material parameters. The phase window labels, gate window sets, sequence scheme tables, and grouting curtain parameter package are integrated into a phase window encapsulation to generate control results. This unifies the constraint sequence of actions, drainage permissibility, and grouting parameters, thereby improving the temporal consistency and execution stability of the control results.

[0048] The above describes the generation of phase-bearing water-blocking curtain parameters from the sequence scheme table, gate window set, tidal time series data, and oxygenation gain sequence, generating the control results. The following describes the phase-bearing mapping of the observed phase data and hydraulic potential sequence corresponding to the tidal time series data, generating a phase-bearing table, and then performing quantile mapping hierarchical encapsulation on the phase-bearing table and the oxygenation gain sequence to generate the grouting curtain parameter package, specifically including: Phase bearing capacity mapping is performed on the observed phase data and hydraulic potential sequence corresponding to the tidal time series data to generate a phase bearing capacity table. The phase bearing capacity table is then screened for limited bearing capacity to generate peak bearing capacity phase data. The peak bearing phase data and oxygen-inducing gain sequence are encapsulated in a locked three-layer parameter package to generate the grouting curtain parameter package.

[0049] Peak carrying capacity phase data refers to a phase structured data unit obtained by mapping observed phase data and hydraulic potential sequence in a phase coordinate system. It is used to characterize the local extreme value of hydraulic scour carrying capacity within a unit phase window and satisfies the restricted screening rules.

[0050] This part has already been described in detail above, so I will not repeat it here.

[0051] This scheme maps the observed phase data and hydraulic potential sequence corresponding to the tidal time series data to generate a phase bearing capacity table. This establishes a quantitative correspondence between the tidal phase position and the downhole hydraulic scour intensity, improving the accuracy of identifying bearing differences in different phase windows. The phase bearing capacity table is screened for limited bearing capacity to generate peak bearing capacity phase data. This allows for the stable locking of the key phase with the strongest scour and most suitable for plugging design from multiple phase windows, reducing the selection bias of the grouting window. The peak bearing capacity phase data and oxygen-inducing gain sequence are encapsulated in a locked three-layer parameter package to generate a grouting curtain parameter package. This package can simultaneously meet the requirements of scour resistance, tight plugging, and orifice sealing, improving the continuous adaptability of a single grouting to periodic tidal disturbances.

[0052] Example 2: Please see Figure 5 A coal mine spontaneous combustion and water inrush disaster coordinated control system, including: The data acquisition module is used to acquire the tidal time series data and roadway monitoring data of the target object; The consistency assessment module is used to perform cross-verification consistency scoring on tidal time series data and roadway monitoring data, and generate a consistency score sequence. The hydraulic analysis module is used to perform consistent scoring-driven adaptive phase tracking on tidal time series data, generate phase window labels, and perform grouping and phase window projection on roadway monitoring data based on the phase window labels to generate hydraulic potential sequences. The gain processing module is used to perform in-direction component separation processing on the roadway monitoring data, consistent scoring sequence and hydraulic potential sequence to generate the oxygen potential sequence, and to perform phase difference gain construction on the hydraulic potential sequence and the oxygen potential sequence to generate the oxygen-inducing gain sequence. The sequence scheme generation module is used to perform quantile threshold self-calibration gating on the oxygen induction gain sequence and phase window labels, generate a set of gating windows, and perform a sequence graph search with gating constraints on the set of gating windows, the oxygen potential sequence, and the oxygen induction gain sequence to generate a sequence scheme table. The regulation result generation module is used to perform phase-bearing water-blocking curtain parameter generation processing on the sequence scheme table, gate window set, tidal time series data and oxygenation gain sequence to generate regulation results.

[0053] This embodiment has the same technical effects as Embodiment 1.

[0054] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. The data mentioned in this application have undergone normalization and other preprocessing to unify dimensions during formula calculations.

[0055] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for synergistically regulating coal spontaneous combustion and water inrush disasters, applied to synergistically regulate coal spontaneous combustion and water inrush disasters in a coastal coal mine, characterized in that, Includes the following steps: Acquire tidal level time series data and tunnel monitoring data for the target object; The tidal time series data includes sampling time, water level value, direction of water level change, magnitude of water level change, and the relationship between water level fluctuations between adjacent sampling times. The tunnel monitoring data includes sampling time, temperature information, carbon monoxide concentration information, oxygen concentration information, water flow change information in the tunnel, pressure change information in the tunnel, and seepage response information of surrounding rock or fissures. The tidal time series data is used to perform a cross-validation consistency scoring on the roadway monitoring data to generate a consistency score sequence. The cross-validation consistency scoring includes aligning the tidal time series data and the roadway monitoring data in the same sampling order to obtain a pair of sample sets, dividing the pair of sample sets into segments by sliding windows of fixed length, calculating the directional consistency amount, smooth consistency amount, and isolation suppression amount for the tidal time series data segment and the roadway monitoring data segment in each sliding window, multiplying the directional consistency amount, the smooth consistency amount, and the isolation suppression amount to obtain the score value of the sliding window, and arranging the score values ​​in the time order of the sliding window to obtain a consistency score sequence. Adaptive phase tracking driven by consistent scoring is performed on the tidal time series data to generate phase window labels. Based on the phase window labels, the tunnel monitoring data is grouped and projected within the phase window to generate a hydraulic potential sequence. Specifically, performing consistent scoring-driven adaptive phase tracking on the tide time series data to generate phase window labels includes: The consistent scoring sequence is sorted in descending order and the difference is calculated to generate a difference sequence. The difference sequence is then subjected to maximum difference location and truncation to generate a set of confidence windows. Within the set of confidence windows, the tide time series data is subjected to local rise and fall direction and shape reference analysis to generate observation phase data. Then, the observation phase data is weighted and divided into equal phase intervals according to the consensus score sequence to generate phase window labels. The process of grouping and projecting the tunnel monitoring data within the phase window according to the phase window label to generate a hydraulic potential sequence specifically includes: Within the set of trustworthy windows, the tunnel monitoring data are grouped according to the phase window labels, and the grouping results are weighted and summed according to the consistent scoring sequence and the observed phase data to generate a hydraulic projection vector. The grouping results are subjected to an inner product projection operation with the hydraulic projection vector and then summarized to generate a hydraulic potential sequence. The roadway monitoring data, the consistent scoring sequence, and the hydraulic potential sequence are subjected to in-direction component separation processing to generate an oxygen potential sequence. Phase differential gain construction is then performed on the hydraulic potential sequence and the oxygen potential sequence to generate an oxygen-inducing gain sequence. A quantile threshold self-calibration gating is performed on the oxygen-inducing gain sequence and the phase window label to generate a gating window set. Then, a sequence graph search with gating constraints is performed on the gating window set, the oxygen potential sequence, and the oxygen-inducing gain sequence to generate a sequence scheme table. The phase-bearing water-blocking curtain parameter generation process is performed on the sequence scheme table, the gate window set, the tide time series data, and the oxygenation gain sequence to generate the control result.

2. The coal spontaneous combustion and water inrush disaster synergic regulation and control method according to claim 1, characterized in that: Performing co-directional component separation processing on the tunnel monitoring data, the consistent scoring sequence, and the hydraulic potential sequence to generate the oxygen combustion potential sequence specifically includes: Within the set of confidence windows, the tunnel monitoring data is weighted and corrected according to the consistent scoring sequence to obtain weighted monitoring data, and the weighted monitoring data is projected in the same direction along the hydraulic potential sequence. The same-direction projection results are subtracted from the weighted monitoring data to obtain differential monitoring data. The differential monitoring data are then subjected to single-axis mapping in the sampling order and summarized to generate an oxygen potential sequence.

3. The coal spontaneous combustion and water inrush disaster synergic regulation method according to claim 1, characterized in that: Performing phase differential gain construction on the hydraulic potential sequence and the oxygen potential sequence to generate the oxygen-inducing gain sequence specifically includes: According to the phase window label, the hydraulic potential sequence and the oxygen-fuel potential sequence are segmented and adjacent difference operations are performed respectively to generate hydraulic difference sequences and oxygen-fuel difference sequences. The absolute value of the hydraulic difference sequence is taken and the mean is calculated. The oxygen-fuel difference sequence is then normalized to the same scale based on the mean calculation result. The minimum positive value after taking the absolute value of the hydraulic difference sequence is extracted as the non-zero term. The result of the same-scale normalization is divided by the sum of the hydraulic difference sequence and the non-zero term, and then summarized to generate the oxygenation gain sequence.

4. The coal spontaneous combustion and water inrush disaster synergic regulation method according to claim 1, characterized in that: Performing quantile threshold self-calibration gating on the oxygen-inducing gain sequence and the phase window label to generate a gating window set specifically includes: The oxygen-inducing gain sequence is grouped according to the phase window label to obtain the gain subsequence of each phase window. The gain subsequence is then sorted in descending order and subjected to adjacent difference operations to generate the gain difference sequence. The sequence number with the largest difference in the gain difference sequence is used as the transition position data, and the gain subsequence after descending order is located and the median is extracted based on the transition position data to generate a gating threshold table for the corresponding phase window. The gain subsequences of each phase window are compared with the corresponding gating threshold table and classified according to their proportions to generate a set of gating windows.

5. The coal spontaneous combustion and water inrush disaster synergic regulation method according to claim 1, characterized in that: Performing a sequence graph search with gating constraints on the gated window set, the oxygen potential sequence, and the oxygen induction gain sequence to generate a sequence scheme table specifically includes: Window position enumeration constraints are applied to the phase window labels and the gated window set to generate an action candidate table; The hydraulic potential sequence, the oxygen potential sequence, and the oxygen-inducing gain sequence are mapped to dominance within a window to generate an action gain table. The action gain table is then quantized by gain quantile cost to generate an action cost table. The action candidate table, the action benefit table, and the action cost table are evaluated by candidate reduction scoring to generate an action score table. Then, the action score table is subjected to adjacent window consistency constraints based on the phase window labels to generate a sequence scheme table.

6. The coal spontaneous combustion and water inrush disaster synergic regulation method according to claim 1, characterized in that: The phase-bearing water-blocking curtain parameter generation process is performed on the sequence scheme table, the gate window set, the tidal time series data, and the oxygenation gain sequence to generate the regulation results, specifically including: Phase carrying capacity mapping is performed on the observed phase data and hydraulic potential sequence corresponding to the tidal time series data to generate a phase carrying capacity table. The phase carrying capacity table and the oxygenation gain sequence are then encapsulated by quantile mapping to generate a grouting curtain parameter package. The phase window label, the gated window set, the sequence scheme table, and the grouting curtain parameter package are integrated into a phase window encapsulation to generate the control result.

7. The coal spontaneous combustion and water inrush disaster synergic regulation method according to claim 6, characterized in that: Phase carrying capacity mapping is performed on the observed phase data and hydraulic potential sequence corresponding to the tidal time series data to generate a phase carrying capacity table. Furthermore, the phase carrying capacity table and the oxygenation gain sequence are encapsulated through quantile mapping to generate a grouting curtain parameter package, specifically including: Phase carrying capacity mapping is performed on the observed phase data and the hydraulic potential sequence corresponding to the tidal time series data to generate a phase carrying capacity table, and the phase carrying capacity table is screened for limited carrying capacity to generate peak carrying capacity phase data. The peak bearing phase data and the oxygen-inducing gain sequence are encapsulated in a locked three-layer parameter package to generate a grouting curtain parameter package.

8. A coal mine spontaneous combustion and water inrush disaster coordinated control system, characterized in that, include: The data acquisition module is used to acquire the tidal time series data and roadway monitoring data of the target object; The tidal time series data includes sampling time, water level value, direction of water level change, magnitude of water level change, and the relationship between water level fluctuations between adjacent sampling times. The tunnel monitoring data includes sampling time, temperature information, carbon monoxide concentration information, oxygen concentration information, water flow change information in the tunnel, pressure change information in the tunnel, and seepage response information of surrounding rock or fissures. The consistency evaluation module is used to perform mutual verification consistency scoring on the tide level time series data and the roadway monitoring data to generate a consistency score sequence. The mutual verification consistency scoring includes aligning the tide level time series data and the roadway monitoring data according to the same sampling order to obtain a pair of sample sets, dividing the pair of sample sets into segments according to a sliding window of fixed length, calculating the directional consistency amount, smooth consistency amount, and isolation suppression amount for the tide level time series data segment and the roadway monitoring data segment respectively within each sliding window, multiplying the directional consistency amount, the smooth consistency amount, and the isolation suppression amount to obtain the score value of the sliding window, and arranging the score values ​​according to the time order of the sliding window to obtain a consistency score sequence. The hydraulic analysis module is used to perform consistent scoring-driven adaptive phase tracking on the tidal time series data, generate phase window labels, and perform grouping and phase window projection on the tunnel monitoring data according to the phase window labels to generate a hydraulic potential sequence. Specifically, performing consistent scoring-driven adaptive phase tracking on the tide time series data to generate phase window labels includes: The consistent scoring sequence is sorted in descending order and the difference is calculated to generate a difference sequence. The difference sequence is then subjected to maximum difference location and truncation to generate a set of confidence windows. Within the set of confidence windows, the tide time series data is subjected to local rise and fall direction and shape reference analysis to generate observation phase data. Then, the observation phase data is weighted and divided into equal phase intervals according to the consensus score sequence to generate phase window labels. The process of grouping and projecting the tunnel monitoring data within the phase window according to the phase window label to generate a hydraulic potential sequence specifically includes: Within the set of trustworthy windows, the tunnel monitoring data are grouped according to the phase window labels, and the grouping results are weighted and summed according to the consistent scoring sequence and the observed phase data to generate a hydraulic projection vector. The grouping results are subjected to an inner product projection operation with the hydraulic projection vector and then summarized to generate a hydraulic potential sequence. The gain processing module is used to perform in-direction component separation processing on the roadway monitoring data, the consistent scoring sequence and the hydraulic potential sequence to generate the oxygen potential sequence, and to perform phase difference gain construction on the hydraulic potential sequence and the oxygen potential sequence to generate the oxygen-inducing gain sequence. The sequence scheme generation module is used to perform quantile threshold self-calibration gating on the oxygen-inducing gain sequence and the phase window label to generate a gating window set, and to perform a sequence graph search with gating constraints on the gating window set, the oxygen potential sequence and the oxygen-inducing gain sequence to generate a sequence scheme table; The regulation result generation module is used to perform phase-bearing water-blocking curtain parameter generation processing on the sequence scheme table, the gate window set, the tide time series data, and the oxygenation gain sequence to generate regulation results.

Images