A large-diameter borehole instant pressure relief effect discrimination method based on a Bayesian network

Abstract

The present application relates to the technical field of coal mine safety monitoring, and particularly relates to a large-diameter borehole instant pressure relief effect discrimination method based on a Bayesian network, comprising the following steps: obtaining drilling parameters and unit drilling length coal powder amount based on the drilling process of a large-diameter pressure relief borehole, and taking the plastic zone self-calibration reference section of the current construction borehole opening end as a reference; performing borehole reference processing on the drilling parameters and unit drilling length coal powder amount to obtain drilling self-calibration pressure relief observation description; inputting an interactive Bayesian network discrimination model based on the drilling self-calibration pressure relief observation description, and using the coal body stress ratio and broken ratio work ratio to form the condition prior of the instant pressure relief state. The present application establishes the borehole reference of the coal powder amount, coal body stress and broken ratio work based on the plastic zone self-calibration reference section of the opening end, and distinguishes high stress powder discharge, coal body broken powder discharge and real instant pressure relief powder discharge through the deviation of the expected coal powder amount ratio and the actual coal powder amount ratio.

Description

Technical Field

[0001] This invention relates to the field of coal mine safety monitoring technology, and in particular to a method for judging the real-time pressure relief effect of large-diameter boreholes based on Bayesian networks. Background Technology

[0002] Large-diameter borehole pressure relief is a core technical means to prevent and control deep coal and rock rock bursts. Large-diameter borehole pressure relief is not a single static process, but a dual time-effect process composed of immediate pressure relief due to construction disturbance and delayed pressure relief in the borehole space. Immediate pressure relief refers to the process in which, during the construction of large-diameter boreholes, the drill bit cuts, squeezes and crushes the coal body, coal dust is discharged, and the structural integrity of the coal body around the borehole is rapidly destroyed, leading to the transient release of local stress, a decrease in bearing capacity, and rapid dissipation of elastic energy. The traditional evaluation of pressure relief effect mainly relies on the national standard GB / T 25217.5-2010 "Methods for Measurement, Monitoring and Prevention of Rock Bursts - Part 5: Drill Cuttings Monitoring Method". This method relies on manual periodic or segmental weighing of the discharged dust, which has time lag and human interference, and cannot achieve real-time quantitative evaluation during the drilling process.

[0003] Current methods for evaluating the pressure relief effect of large-diameter pressure vessels have certain theoretical limitations:

[0004] Chinese Patent Publication No. CN114087018A proposes a precise pressure relief method for large-diameter pressure relief boreholes based on stress sensing. This method dynamically adjusts pressure relief parameters by real-time measurement of torque, rotational speed, thrust, feed rate, and coal dust quantity, calculating borehole stress index and drill cuttings index distribution curves. However, this method treats each monitored parameter as an independent variable, neglecting the conditional dependence between drill cuttings quantity, drilling specific power, and coal stress. Under complex geological conditions, factors such as changes in coal seam thickness and interbedded rock distribution can severely interfere with the coal dust removal index, causing the method to fail to distinguish between the formation background response and the actual pressure relief response, resulting in a high misjudgment rate.

[0005] Chinese Patent Publication No. CN116070820A proposes a monitoring device and pressure relief effect evaluation method for large-diameter boreholes based on laser ranging. This method collects borehole spatial deformation data, such as borehole collapse and closure length, collapse time, and collapse rate, after large-diameter borehole formation using laser ranging, and then evaluates the pressure relief effect using a fuzzy comprehensive evaluation system. However, this method evaluates the delayed pressure relief effect after borehole formation, and cannot achieve real-time evaluation during drilling. Furthermore, this method relies on single-dimensional post-drilling data on borehole deformation, and cannot integrate multi-source dynamic information during drilling, making it difficult to quickly respond to the real-time adjustment needs of on-site anti-scour construction.

[0006] In summary, existing methods for evaluating the pressure relief effect of large-diameter boreholes during drilling, due to the assumption that various observation indicators are independent under a given pressure relief state and the failure to integrate multi-source dynamic information during drilling, tend to erroneously determine that high natural dust discharge caused by high stress or fractured coal body is sufficient for immediate pressure relief, and conversely, erroneously determine that low natural dust discharge caused by low stress and intact coal body is insufficient for immediate pressure relief. This has become the main bottleneck restricting the accurate judgment of the immediate pressure relief effect of large-diameter boreholes in deep coal mines. Therefore, it is urgent to design a method for judging the immediate pressure relief effect of large-diameter boreholes based on Bayesian networks. Summary of the Invention

[0007] This invention provides a method for judging the instant pressure relief effect of large-diameter boreholes based on Bayesian networks.

[0008] A method for judging the instantaneous pressure relief effect of large-diameter boreholes based on Bayesian networks includes the following steps:

[0009] Based on the drilling process of large-diameter pressure relief boreholes, the drilling parameters and coal powder content per unit drilling length are obtained, and the self-calibration reference section of the plastic zone at the opening end of the current construction borehole is used as the reference.

[0010] The drilling parameters and the amount of coal powder per unit drilling length are benchmarked in the borehole to obtain the self-calibrated pressure relief observation description during drilling. The self-calibrated pressure relief observation description during drilling includes the ratio of coal powder amount, the ratio of coal body stress, the ratio of crushing specific power, and the residual information of coal powder response.

[0011] Based on the self-calibrated stress relief observation description input interactive Bayesian network discrimination model during drilling, the conditional prior of the instantaneous stress relief state is formed using the coal stress ratio and the crushing specific power ratio.

[0012] Based on the instantaneous depressurization state, coal stress ratio, and crushing specific power ratio, the coal powder quantity ratio and coal powder response residual information are jointly interpreted to obtain the instantaneous depressurization posterior discrimination distribution;

[0013] Based on the aforementioned immediate pressure relief posterior discriminant distribution, the immediate pressure relief state with the highest posterior probability is selected as the immediate pressure relief discriminant level for the corresponding drilling position;

[0014] The instant pressure relief discrimination level is matched with the confidence threshold, drilling depth location and field verification true value to output the instant pressure relief discrimination result of the borehole;

[0015] The interactive Bayesian network discriminant model is updated based on the on-site verification truth values.

[0016] The beneficial effects of this invention are:

[0017] This invention establishes a reference for the amount of coal powder, coal stress, and crushing specific work in the hole by setting up a self-calibration and coal powder residual correction mechanism in the hole end plastic zone self-calibration reference section. It also distinguishes between high stress-induced coal powder discharge, coal crushing-induced coal powder discharge, and real-time pressure relief coal powder discharge by the deviation between the expected coal powder ratio and the actual coal powder ratio.

[0018] This invention, through the setting of an interactive Bayesian discrimination and verification update mechanism, utilizes the coal stress state and coal-rock integrity state to form an immediate decompression state as a conditional prior, then jointly interprets the coal powder quantity response and combines it with residual correction to output a reliable posterior discrimination result. At the same time, it updates the conditional probability table and residual correction coefficient based on the true value of field verification. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only for this invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a diagram of the interactive Bayesian network structure of drilling parameters and pulverized coal quantity in this invention;

[0021] Figure 2 This is a flowchart of the method for judging the instant pressure relief effect of large-diameter boreholes based on Bayesian networks according to the present invention.

[0022] Figure 3 This is a schematic diagram of the probability cloud map of borehole number and borehole depth according to the present invention;

[0023] Figure 4 This is an architecture diagram of the interactive Bayesian network discriminant model of the present invention. Detailed Implementation

[0024] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. For some well-known technologies, those skilled in the art may also use other alternative methods to implement the invention. Moreover, the accompanying drawings are only for more specific description of the embodiments and are not intended to specifically limit the present invention.

[0025] like Figures 1-4 As shown, a method for judging the instantaneous pressure relief effect of large-diameter boreholes based on Bayesian networks includes the following steps:

[0026] S1. During the drilling process of the large-diameter pressure relief borehole, the drilling parameters and the amount of coal dust per unit drilling length are obtained. Using the self-calibration reference section of the plastic zone at the opening end of the current construction borehole as a reference, the drilling parameters and the amount of coal dust per unit drilling length are benchmarked in the borehole to obtain the self-calibrated pressure relief observation description during drilling. The self-calibrated pressure relief observation description during drilling includes the ratio of coal dust amount, the ratio of coal body stress, the ratio of crushing specific power, and the residual information of coal dust response.

[0027] A mining drilling rig-mounted monitoring system (MDS) is a construction process monitoring and data processing system installed on a mining drilling rig. Its function is not simply to record whether the drilling rig is working, but rather to continuously collect drilling parameters during large-diameter pressure-relief drilling operations. These parameters reflect the state of the drill bit cutting, compressing, and breaking the coal body, such as drilling torque, drilling speed, thrust, feed rate, and drilling depth. These parameters are then linked to corresponding depth positions at preset drilling length intervals. Simultaneously, in conjunction with a coal dust collection or metering unit, the amount of coal dust discharged within each drilling length interval is recorded. This forms the original observation data corresponding to the drilling position, drilling parameters, and coal dust quantity. Based on the relationship between parameters such as torque, speed, thrust, and feed rate, the MDS can invert or convert the stress state and specific work state of the coal body at the current drilling position, providing fundamental data for subsequent calculations of coal stress ratio, specific work ratio, coal dust quantity ratio, and coal dust response residual information.

[0028] S11, during the construction of large-diameter pressure relief boreholes, the drilling process is continuously monitored using a mining drilling rig-equipped monitoring system, and the drilling process is recorded in segments according to preset drilling length intervals; the preset drilling length intervals are determined based on the on-site drilling accuracy and coal powder metering accuracy, and are preferably 1m; for the first... The system continuously acquires drilling torque, rig speed, thrust, feed rate, and drilling depth within a preset drilling length interval, and simultaneously acquires the amount of coal dust discharged within the preset drilling length interval, thereby forming the original observation sequence for drilling operations.

[0029] The original observation sequence during drilling is represented as follows:

[0030] ;

[0031] in, This represents the original observation sequence during drilling. Indicates the drilling position number; This indicates the total number of preset drilling length intervals that have been recorded in the current drilling hole; Indicates the first The drilling depth position corresponding to a preset drilling length interval; Indicates the first Drilling torque within a preset drilling length interval; Indicates the first Drilling speed within a preset drilling length interval; Indicates the first The thrust within a preset drilling length interval; Indicates the first Feed rate within a preset drilling length interval; Indicates the first The amount of coal dust discharged within a preset drilling length interval.

[0032] When the preset drilling length interval is 1m This refers to the amount of coal dust discharged per unit drilling length. When the preset drilling length interval is not 1m, the amount of coal dust discharged within that interval is converted into the amount of coal dust discharged per unit drilling length. Specifically, the total amount of coal dust actually discharged within that interval is first matched with the actual drilling length corresponding to that interval, and then converted according to the average amount of coal dust discharged per unit length. That is, the total amount of coal dust discharged within that interval is divided by the drilling length of that interval to obtain the amount of coal dust converted to every 1m of drilling length.

[0033] S12, based on the original observation sequence during drilling, the drilling parameters at each drilling location are inverted and converted to obtain the coal stress value and crushing specific power value at the corresponding drilling location, and the amount of coal dust within the same preset drilling length interval is determined as the amount of coal dust per unit drilling length; specifically, the drilling monitoring system equipped with the mining drilling rig inverts the stress state of the coal at the current drilling location according to the correspondence between drilling torque, drilling rig rotation, thrust, feed rate and drilling depth to obtain the coal stress value; at the same time, based on the drilling energy consumed by the crushing of a unit volume of coal during drilling, the crushing specific power value is calculated; the coal stress value and crushing specific power value are expressed as follows:

[0034] ;

[0035] ;

[0036] in, Indicates the first Coal stress values ​​at each drilling location; Indicates the first The specific power value of the breaking work at each drilling location; This represents the calibration function for inverting coal stress values ​​from drilling parameters; This represents the calibration function for converting the breaking specific power value from drilling parameters.

[0037] The calibration function for coal stress values ​​obtained from drilling parameter inversion is a progressive structure consisting of a parameter preprocessing layer, a drilling resistance characterization layer, a depth correction layer, and a stress calibration layer. Specifically, the calibration function first performs a parameter preprocessing layer, a drilling resistance characterization layer, a depth correction layer, and a stress calibration layer. The drilling torque, rig speed, thrust, feed rate, and drilling depth at each drilling position are synchronized and aligned. Abnormalities such as idling, momentary impact from stuck drill, and sensor jumps are eliminated. Stable representative values ​​are taken for parameters within the same preset drilling length interval. Then, based on the relationship between thrust reflecting axial bit indentation resistance, drilling torque reflecting cutting resistance, and feed rate and rig speed reflecting coal breaking efficiency per unit time, a comprehensive characterization of drilling resistance is constructed. That is, when thrust and torque increase, feed rate decreases, and rig speed fluctuations intensify, it indicates increased resistance in the coal body encountered by the drill bit, typically corresponding to higher coal body stress or... A strong confining pressure constraint, combined with the drilling depth, is used to correct the comprehensive characterization of drilling resistance. This correction is used to eliminate non-stress effects caused by changes in depth, such as wellhead disturbance, increased length of the powder discharge channel, and accumulated friction of the drill rod. Finally, the corrected comprehensive characterization of drilling resistance is input into a pre-calibrated stress mapping relationship to obtain the stress value of the coal body at the corresponding drilling location. This calibration relationship is established based on known stress monitoring data, borehole stress gauge data, micro-vibration energy release verification data, or historical borehole construction samples. It is an equivalent inversion result that converts the changes in the mechanical response of the drilling rig into the stress state of the coal body at the current drilling location.

[0038] The calibration function for converting the breaking power value using drilling parameters is a progressive structure consisting of an input power calculation layer, a unit footage energy conversion layer, a breaking volume normalization layer, and an integrity direction unification layer. Specifically, it first uses the first... The drilling torque and rig speed at each drilling position determine the rotary cutting input energy, and the axial propulsion input energy is determined based on the thrust and feed rate. These two parts together reflect the mechanical energy consumed by the drill bit in breaking the coal body. This mechanical energy is then converted according to the corresponding preset drilling length intervals to obtain the drilling energy consumption per unit depth. Combined with the borehole diameter, drilling length, and the theoretical borehole volume formed by the broken coal body, the drilling energy consumption per unit depth is converted into the energy required to break a unit volume of coal body, thus obtaining the specific energy value for breaking. Finally, the physical direction of the specific energy value for breaking is unified, that is, the larger the specific energy value for breaking, the higher the energy required by the drill bit to break a unit volume of coal body, and the more intact or difficult the coal body is to break. The smaller the specific energy value for breaking, the more broken the coal body, the more developed the fractures, or the easier it is to remove dust.

[0039] The specific energy value of the crushing is used to characterize the difficulty of drilling into the crushed coal body; that is, the larger the specific energy value of the crushing is, the more intact the coal body is.

[0040] S13, a preset depth range extending into the hole from the opening end of the current construction borehole is selected as the self-calibration reference segment of the plastic zone; the self-calibration reference segment of the plastic zone is preferably the drilling depth range of 0m to 5m of the current construction borehole; since the self-calibration reference segment of the plastic zone can reflect the dust discharge, stress and fracturing characteristics of the local micro-environment where the current borehole is located, using the self-calibration reference segment of the plastic zone as the self-calibration reference in the borehole can reduce the influence of coal seam thickness variation, interbedded rock distribution, soft and hard interlayer and construction disturbance differences between different boreholes on the pressure relief judgment.

[0041] Within the self-calibration reference section of the plastic zone, the reference values ​​for coal powder quantity, coal stress, and crushing specific energy are determined based on the coal powder quantity per unit drilling length, coal stress value, and crushing specific energy value, respectively, and are expressed as follows:

[0042] ;

[0043] ;

[0044] ;

[0045] in, This indicates the reference value for the amount of pulverized coal within the self-calibration reference section of the plastic zone; This represents the reference value of coal stress within the self-calibration reference section of the plastic zone; This represents the reference value of the breaking specific work within the self-calibration reference section of the plastic zone; Indicates the drilling depth position The amount of pulverized coal per unit drilling length at the location; Indicates the drilling depth position The stress value of the coal body at that location; Indicates the drilling depth position The specific work value of the breakage at the location; This indicates median operations.

[0046] If represented by discrete drilling position numbers, the above reference values ​​are derived from the drilling positions corresponding to each drilling position within the self-calibration reference section of the plastic zone. , and We obtain the result by taking the median of each.

[0047] S14, the coal powder quantity per unit drilling length, coal stress value, and crushing specific power value at the current drilling position are respectively subjected to in-hole benchmarking processing with the coal powder quantity benchmark value, coal stress benchmark value, and crushing specific power benchmark value to obtain the coal powder quantity ratio, coal stress ratio, and crushing specific power ratio; for any borehole and any drilling position The following three ratios are constructed as follows:

[0048] ;

[0049] ;

[0050] ;

[0051] in, Indicates the first The ratio of coal powder content at each drilling location is used as a characterization layer observation variable; Indicates the first The coal stress ratio at each drilling location is used as the first conditional variable. Indicates the first The breakdown ratio at each drilling location is used as a second conditional variable. Indicates the first The amount of pulverized coal per unit drilling length at each drilling location; Indicates the first Coal stress values ​​at each drilling location; Indicates the first The specific power value of the breaking work at each drilling location; , and These represent the reference values ​​for pulverized coal quantity, coal body stress, and crushing specific work, respectively.

[0052] The powder discharge, stress, and specific energy of different drilling locations and depths are uniformly converted into relative states with respect to the self-calibration reference section of the plastic zone of this hole.

[0053] S15, using the coal body stress ratio and crushing specific work ratio, determine the expected coal powder quantity ratio under the current stress state and coal-rock mass integrity state, and perform deviation processing on the coal powder quantity ratio and the expected coal powder quantity ratio to obtain coal powder response residual information; the expected coal powder quantity ratio is used to represent the change in normal coal powder quantity caused by background stress and coal-rock mass structural conditions under the current coal body stress state and coal-rock mass integrity state, and the expected coal powder quantity ratio is expressed as:

[0054] ;

[0055] in, Indicates the first The expected coal powder ratio at each drilling location; This represents a fitting function that determines the desired coal powder ratio based on the coal body stress ratio and the crushing specific work ratio. Indicates the first The stress ratio of the coal body at each drilling location; Indicates the first The breaking power ratio at each drilling location.

[0056] The fitting function is a structure for constructing the background coal dust removal influence term, calibrating the influence coefficient, and outputting the expected coal dust ratio. Specifically, the coal stress ratio is first used as the high-stress coal dust removal influence term because the higher the coal stress at the drilling location, the more obvious the coal body is subjected to pressure fracturing and the more obvious the release of drilling disturbance. Even if sufficient immediate pressure relief is not achieved, a higher coal dust removal rate may still be generated. Then, the crushing specific energy ratio is converted into the coal body's fracturing ease influence term. Since a larger crushing specific energy ratio indicates that the coal body is more intact and more difficult to crush, the inverse vector of the crushing specific energy ratio is used to characterize the degree of coal body crushing or the ease of coal dust removal. The lower the ratio of crushing specific power, the more crushed the coal body, and the easier it is for the background dust discharge to increase. Then, using historical construction data of the working face, verified borehole data in the same area, or stability data of the current drilling section, the influence coefficients of coal stress ratio and crushing specific power ratio are fitted and calibrated so that the fitting function can reflect the normal dust discharge background of the current working face. Finally, the coal stress ratio and crushing specific power ratio at the current drilling position are substituted into the fitting function to obtain the expected coal dust ratio that should theoretically appear under the current stress state and coal and rock mass integrity state.

[0057] The deviation between the actual pulverized coal quantity ratio and the expected pulverized coal quantity ratio is calculated, and the pulverized coal response residual information is expressed as follows:

[0058] ;

[0059] in, Indicates the first residual information of pulverized coal response at each drilling location; Indicates the first The ratio of actual pulverized coal quantity at each drilling location; Indicates the first The expected coal powder ratio at each drilling location.

[0060] The residual information of the pulverized coal response is used to distinguish between the normal pulverized coal quantity change caused by coal body stress and coal-rock integrity and the pulverized coal quantity response caused by instantaneous pressure relief failure; when Significantly higher than This indicates that the dust removal response at the drilling location exceeds the normal dust removal level under the current stress and coal and rock mass integrity background, and there may be stronger drilling damage and immediate pressure relief response.

[0061] S2, the description of the self-calibrated pressure relief observation while drilling is input into the interactive Bayesian network discriminant model, the conditional prior of the instantaneous pressure relief state is formed by the coal stress ratio and the crushing specific power ratio, and the instantaneous pressure relief state, the coal stress ratio and the crushing specific power ratio are used together to explain the coal powder quantity ratio and the coal powder response residual information to obtain the instantaneous pressure relief posterior discriminant distribution;

[0062] The interactive Bayesian network discriminant model has a structure in which four types of input state nodes and one target state node interact. The input state nodes include coal stress state, coal-rock integrity state, coal powder response state, and residual response state. The target state node is the instantaneous pressure relief state. The coal stress state is mapped from the coal stress ratio, the coal-rock integrity state from the crushing specific work ratio, the coal powder response state from the coal powder ratio, and the residual response state from the residual information of the coal powder response. The instantaneous pressure relief state is divided into no significant pressure relief response, weak pressure relief response, moderate pressure relief response, and strong pressure relief response. The core connection of the network is that the stress state of the coal body and the integrity state of the coal-rock mass both point to the instantaneous decompression state, which is used to form the a priori conditions for the instantaneous decompression state; the instantaneous decompression state, the stress state of the coal body, and the integrity state of the coal-rock mass all point to the coal powder quantity response state, which is used to explain the coal powder quantity response observed under different stress and integrity backgrounds; the residual response state is used to correct the observation probability of the coal powder quantity response state, so that the positive residuals higher than the expected coal powder quantity ratio can enhance the explanatory weight of the medium or strong decompression response, and prevent the high discharge caused by high stress or natural coal body breakage from being directly mistaken for sufficient decompression.

[0063] The construction process of the interactive Bayesian network discriminant model follows the sequence of state discretization, node connection, conditional probability table configuration, and residual correction rule configuration. First, the continuous coal powder ratio, coal body stress ratio, crushing specific power ratio, and coal powder response residual information are mapped to discrete states according to preset thresholds or sample quantile intervals. For example, the coal body stress state is divided into low stress, normal stress, and high stress; the coal-rock mass integrity state is divided into crushed coal body, normal integrity, and intact coal body; the coal powder response state is divided into low coal powder response, normal coal powder response, and high coal powder response; and the residual response state is divided into negative residual, near-zero residual, and positive residual. Then, an instantaneous decompression state conditional probability table is established to express the conditions under different coal bodies. Under the combination of stress state and coal-rock mass integrity state, the prior probability of each instantaneous decompression state is obtained. Then, a coal powder quantity response conditional probability table is established to express the probability of a certain coal powder quantity response state occurring given an instantaneous decompression state, coal stress state, and coal-rock mass integrity state. Finally, a correction rule for the residual response state on the coal powder quantity response conditional probability is set. For example, the more obvious the positive residual, the more it indicates that the actual powder discharge exceeds the expected powder discharge level under the current stress and coal-rock mass integrity background. Therefore, the observation interpretation weight corresponding to the medium or strong decompression response is increased. The more obvious the negative residual, the lower the observation interpretation weight corresponding to the strong decompression response is. This forms an interactive Bayesian network discriminant model that can distinguish between background powder discharge and actual decompression powder discharge.

[0064] The interactive Bayesian network discriminant model was trained using the following steps: initial calibration, sample statistics, probability smoothing, and incremental updates in the field. In the initial stage, historical borehole data from the working face, verified borehole data from the same area, expert experience, and data from the current borehole's plastic zone self-calibration benchmark section were used to determine the mapping thresholds for each state and to provide initial values ​​for the conditional probability tables of instantaneous decompression state and pulverized coal quantity response. Subsequently, the coal stress state, coal-rock integrity state, pulverized coal quantity response state, residual response state, and field verification true values ​​corresponding to each verified drilling location were paired and statistically analyzed. The field verification true values ​​could be derived from microseismic energy release, stress changes at adjacent monitoring points, borehole wall deformation, borehole collapse and closure records, or... The results of the re-drilling verification were used to statistically analyze the frequency of occurrence of each instantaneous pressure relief state and pulverized coal quantity response state under different state combinations, and these frequencies were converted into conditional probabilities. To avoid the probability of a small sample state combination being zero, the conditional probability table was smoothed so that states that did not occur sufficiently but were likely to occur in engineering still retained a small probability. In subsequent construction, when the new instantaneous pressure relief judgment results of the borehole were verified on-site, the verification results were fed back to the model as incremental samples to update the conditional probability table of instantaneous pressure relief state, the conditional probability table of pulverized coal quantity response, and the residual correction rules, so that the model could gradually adapt to the instantaneous pressure relief judgment requirements under different working faces, different coal seam structures, different interbedded rock distributions, and different construction parameters.

[0065] S21, the coal powder ratio, coal body stress ratio, crushing specific power ratio and coal powder response residual information in the drilling self-calibration pressure relief observation description are respectively mapped to obtain the coal powder response state, coal body stress state, coal-rock mass integrity state and residual response state.

[0066] To enable continuous ratio variables to enter the interactive Bayesian network discriminant model for probabilistic inference, the coal powder ratio, coal body stress ratio, crushing specific power ratio, and coal powder response residual information are mapped to discrete states respectively; the coal body stress state is determined by the coal body stress ratio. The stress state can be classified into low-stress state, conventional stress state, and high-stress state; the integrity state of the coal and rock mass is determined by the ratio of crushing specific work. It is determined that the coal body can be divided into three states: broken coal body, conventional integrity, and intact coal body; the coal powder quantity response state is determined by the coal powder quantity ratio. It is determined that the response can be divided into low pulverized coal response, conventional pulverized coal response, and high pulverized coal response; the residual response state is determined by the pulverized coal response residual information. It is determined that the residual response can be divided into negative residual response, near-zero residual response, and positive residual response; the state mapping relationship is expressed as follows:

[0067] ;

[0068] ;

[0069] ;

[0070] ;

[0071] in, Indicates the first The coal powder quantity response status at each drilling location; Indicates the first Stress state of coal at each drilling location; Indicates the first The integrity status of the coal and rock mass at each drilling location; Indicates the first The residual response state at each drilling location; A state mapping function representing the ratio of pulverized coal quantity; A state mapping function representing the stress ratio of coal seams; The state mapping function represents the ratio of breakage to power. A state mapping function representing the residual information of pulverized coal response.

[0072] State mapping function of pulverized coal quantity ratio The structure includes a benchmark neighborhood judgment, a low response judgment, and a high response judgment. Specifically, the benchmark state of the coal powder ratio is used as the central judgment interval. Since the coal powder ratio is the relative amount of coal powder per unit drilling length at the current drilling position to the benchmark value of coal powder in the self-calibration benchmark section of the plastic zone, when the coal powder ratio is close to 1, it indicates that the coal powder discharge level at this position is basically consistent with the self-calibration benchmark section of the plastic zone of this hole, which can be mapped to a normal coal powder response state. When the coal powder ratio is lower than the lower limit of the normal interval, it indicates that the coal powder discharge per unit length at this position is lower than the benchmark coal powder discharge level of this hole, which can be mapped to a low coal powder response state. When the coal powder ratio is higher than the upper limit of the normal interval, it indicates that the coal powder discharge per unit length at this position is higher than the benchmark coal powder discharge level of this hole, which can be mapped to a high coal powder response state.

[0073] The lower and upper limits of the conventional range for the coal powder ratio are determined jointly based on the self-calibration benchmark section of the current borehole plastic zone and the distribution of the coal powder ratio in verified boreholes on the same working face. Specifically, the benchmark value of the coal powder in the self-calibration benchmark section of the plastic zone is first used as the center to make the benchmark center of the coal powder ratio close to 1. Then, the fluctuation range of the coal powder ratio in the stable coal powder discharge interval of the previous section of the borehole or in the normal drilling samples of the same working face is statistically analyzed, and abnormal points such as stuck drill, coal blockage, instantaneous idle rotation, and manual hole cleaning are eliminated. Finally, the lower and upper boundaries that most normal coal powder discharge samples fall into are taken as the lower and upper limits of the conventional range.

[0074] State mapping function of coal stress ratio The structure is designed for low-stress identification, conventional stress identification, and high-stress identification. Specifically, since the coal stress ratio is a relative quantity of the coal stress value at the current drilling position relative to the coal stress benchmark value in the self-calibration benchmark section of the plastic zone, when the coal stress ratio is close to 1, it indicates that the coal stress level at the current drilling position is basically consistent with the self-calibration benchmark section of the plastic zone at the opening end of the hole, and can be mapped to a conventional stress state; when the coal stress ratio is lower than the preset low-stress threshold, it indicates that the stress level at this position is relatively low, and can be mapped to a low-stress state; when the coal stress ratio is higher than the preset high-stress threshold, it indicates that there is strong confining pressure constraint or local high stress concentration at this position, and can be mapped to a high-stress state.

[0075] The low-stress threshold and high-stress threshold of the coal stress ratio are determined based on the coal stress benchmark value of the self-calibrated benchmark section in the plastic zone, the known stress monitoring results of the working face, and the distribution of coal stress values ​​obtained through drilling inversion. Specifically, the coal stress values ​​at each drilling position of the current borehole are first converted into coal stress ratios relative to the coal stress benchmark value, so that the coal stress ratio near the benchmark is close to 1. Then, by combining borehole stress gauges, microseismic energy release, adjacent area ground pressure manifestation records, or historical construction samples, low-stress stable sections, conventional stress sections, and high-stress concentration sections are identified. The boundary value between the conventional stress section and the low-stress stable section is determined as the low-stress threshold, and the boundary value between the conventional stress section and the high-stress concentration section is determined as the high-stress threshold.

[0076] State mapping function of the breaking ratio The structure for identifying fractured coal bodies, conventional integrity coal bodies, and intact coal bodies is as follows: First, the physical direction of the specific energy ratio of the fractured coal body is kept consistent. That is, the larger the specific energy ratio of the fractured coal body, the more intact and difficult to fracture the coal body is; the smaller the specific energy ratio of the fractured coal body, the more fractured the coal body is, the more developed the fissures are, and the easier it is to remove dust. When the specific energy ratio of the fractured coal body is close to 1, it means that the integrity of the coal and rock mass at the current drilling position is basically consistent with the self-calibration benchmark section of the plastic zone, and it can be mapped to the conventional integrity state. When the specific energy ratio of the fractured coal body is lower than the preset fractured threshold, it means that the energy required to break the coal per unit volume is too low, and the current coal body is relatively fractured or structurally weakened, and it can be mapped to the fractured coal body state. When the specific energy ratio of the fractured coal body is higher than the preset integrity threshold, it means that the energy required to break the coal per unit volume is too high, and the current coal body is relatively intact or difficult to fracture, and it can be mapped to the intact coal body state.

[0077] The fracturing and integrity thresholds for the fracturing specific energy ratio are determined based on the statistical distribution of the fracturing specific energy ratio and the verification results of the coal and rock mass structure in the field. Specifically, since a larger fracturing specific energy ratio indicates a more intact and difficult-to-fracture coal body, while a smaller fracturing specific energy ratio indicates a more fractured coal body with more developed fractures, the fracturing specific energy ratio benchmark value of the self-calibration benchmark section in the plastic zone is first used as the normalization benchmark to make the normal center of the fracturing specific energy ratio close to 1. Then, combined with borehole sampling, changes in powder discharge particle size, rock inclusion records, drilling vibration, stuck drill conditions, or borehole wall fracturing conditions, samples with significantly low fracturing specific energy ratios and corresponding coal bodies that are fractured, loose, and easy to discharge powder are classified as fractured coal body samples, while samples with significantly high fracturing specific energy ratios and corresponding coal bodies that are intact, difficult to fracture, and have high drilling resistance are classified as intact coal body samples. Finally, the boundary value between the normal integrity state and the fractured coal body state is determined as the fracturing threshold, and the boundary value between the normal integrity state and the intact coal body state is determined as the integrity threshold.

[0078] State mapping function of pulverized coal response residual information The structure for negative residual identification, near-zero residual identification, and positive residual identification is as follows: First, read the... The residual information of pulverized coal response at each drilling location represents the deviation between the actual pulverized coal quantity ratio and the expected pulverized coal quantity ratio corresponding to the current coal stress ratio and crushing specific power ratio. When the residual information of pulverized coal response is within a preset allowable range near zero, it indicates that the actual pulverized coal discharge level basically meets the background pulverized coal discharge level under the current stress state and coal-rock integrity state, and can be mapped to a near-zero residual response state. When the residual information of pulverized coal response is lower than the negative residual threshold, it indicates that the actual pulverized coal quantity ratio is lower than the expected pulverized coal quantity ratio, and can be mapped to a negative residual response state. When the residual information of pulverized coal response is higher than the positive residual threshold, it indicates that the actual pulverized coal quantity ratio is higher than the expected pulverized coal quantity ratio, and there is an additional pulverized coal discharge response exceeding the background pulverized coal discharge conditions, and can be mapped to a positive residual response state.

[0079] The permissible interval, negative residual threshold, and positive residual threshold of the pulverized coal response residual information are determined based on the normal fitting error range between the actual pulverized coal quantity ratio and the expected pulverized coal quantity ratio. Specifically, firstly, using historical construction data of the working face, verified borehole data in the same area, or current borehole front-end stability data, a fitting function is established to predict the expected pulverized coal quantity ratio based on the coal body stress ratio and the crushing specific power ratio. Then, the difference between the actual pulverized coal quantity ratio and the expected pulverized coal quantity ratio in the training samples is calculated to form a residual sample set. Next, the small fluctuation range around zero in the residual samples is determined as the permissible interval, indicating that the actual pulverized coal discharge basically conforms to the normal pulverized coal discharge level under the current stress and integrity background. The negative boundary value outside the lower boundary of the permissible interval is determined as the negative residual threshold, and the positive boundary value outside the upper boundary of the permissible interval is determined as the positive residual threshold.

[0080] Through state mapping processing, the continuously changing description of self-calibrating pressure relief observations while drilling can be converted into discrete state inputs that can be invoked by conditional probability tables.

[0081] S22, In the interactive Bayesian network discriminant model, an instantaneous pressure relief state is set, and the instantaneous pressure relief state is divided into no obvious pressure relief response, weak pressure relief response, moderate pressure relief response and strong pressure relief response.

[0082] The immediate pressure relief state is classified based on the degree of deviation of the coal powder quantity response at the drilling location from the stress background and the coal and rock mass integrity background, combined with the true values ​​verified in the field, rather than simply judging by the amount of coal powder. Specifically, the state with a coal powder quantity response that basically meets the background expectation, a coal powder response residual close to zero or negative, and no obvious stress release or structural damage is classified as no obvious pressure relief response; the state with an actual coal powder quantity slightly higher than the background expectation, a coal powder response residual showing a small positive change, and only slight stress release or local damage signs are classified as weak pressure relief response; the state with an actual coal powder quantity significantly higher than the background expectation, a stable positive coal powder response residual, and accompanied by relatively clear stress decrease, energy release, or hole-perimeter coal body damage characteristics is classified as moderate pressure relief response; and the state with an actual coal powder quantity significantly higher than the background expectation, a coal powder response residual that remains at a high positive level, and field verification showing a significant reduction in the hole-perimeter coal body bearing capacity, full release of elastic energy, or continuous fracturing pressure relief characteristics is classified as strong pressure relief response.

[0083] The immediate depressurization state is recorded as... , and defined as:

[0084] ;

[0085] in, Indicates an immediate depressurization status; This indicates no significant pressure relief response; Indicates a weak pressure relief response; Indicates a moderate pressure relief response; This indicates a strong pressure relief response.

[0086] In the interactive Bayesian network discriminant model, the instantaneous pressure relief state is not determined directly based on the coal powder quantity response state. Instead, the conditional priors of the instantaneous pressure relief state are first formed by utilizing the coal stress state and the coal-rock integrity state.

[0087] Based on the stress state of the coal body and the integrity status of coal and rock mass By querying the conditional probability table for immediate depressurization states, the conditional prior representations corresponding to different immediate depressurization states are obtained as follows:

[0088] ;

[0089] in, Indicates the first Under the constraints of the coal stress state and coal-rock integrity state at each drilling location, the immediate decompression state is as follows: The prior conditions; Indicates the first Similar to an immediate depressurization state, and ; Indicates the stress state of the coal body; This indicates the integrity state of the coal and rock mass; this conditional prior is used to express the prior probability that the drilling location is in different instantaneous pressure relief states under different combinations of coal stress states and coal and rock mass integrity states.

[0090] Table 1 Conditional Probability Table of Instantaneous Depressurization State

[0091]

[0092] When the stress is high and the coal body is incomplete, even if there is a high amount of pulverized coal, it cannot be directly assumed that the pressure relief effect is strong; when the stress is low and the coal body is intact, even if the amount of pulverized coal is low, it cannot be directly assumed that the pressure relief effect is poor, thus avoiding the traditional method's singular interpretation of the pulverized coal quantity index.

[0093] S23. Based on the instantaneous depressurization state, coal stress state, and coal-rock integrity state, query the coal powder quantity response conditional probability table to obtain the observation probability of the coal powder quantity response state occurring under the combined action of different instantaneous depressurization states, different coal stress states, and different coal-rock integrity states.

[0094] Table 2 Conditional Probability of Coal Pulverization Response

[0095]

[0096] The interactive Bayesian network discriminant model sets up a conditional probability table for the coal powder quantity response, which is used to express the probability of observing a certain coal powder quantity response state given the instantaneous decompression state, coal stress state, and coal-rock integrity state. The table is expressed as follows:

[0097] ;

[0098] in, Indicates that in the immediate depressurization state The stress state of the coal body is The integrity state of the coal and rock mass is as follows: Under these conditions, the coal powder quantity response state was observed. The probability of; Indicates the first The coal powder quantity response status at each drilling location.

[0099] The conditional probability table of pulverized coal quantity response allows the pulverized coal quantity response to participate in the instantaneous pressure relief effect discrimination as direct phenological evidence, while also being subject to the conditional constraints of the coal body stress state and the coal-rock mass integrity state. The interactive Bayesian network discrimination model no longer regards pulverized coal quantity, coal body stress, and coal-rock mass integrity as mutually independent variables, but expresses the joint effect of the three on the instantaneous pressure relief state discrimination through conditional probability relationships.

[0100] The joint probability decomposition of an interactive Bayesian network is expressed as:

[0101] ;

[0102] in, This represents the instantaneous decompression state under the constraints of the coal body stress state and the coal-rock mass integrity state. Response state with coal powder quantity Joint probability of co-occurrence;

[0103] This represents the observed probability of the coal powder quantity response state; This represents the prior conditions for the immediate depressurization state.

[0104] S24, the observation probability is corrected using the residual response state, so that the positive deviation of the actual coal powder ratio from the expected coal powder ratio increases the explanatory weight of the strong or moderate pressure relief response, and reduces the influence of the background powder discharge response caused by the high stress state or the incomplete state of the coal and rock mass on the misjudgment of the instantaneous pressure relief state, thus obtaining the corrected observation probability.

[0105] Specifically, the desired coal powder ratio is determined by the coal body stress ratio and the crushing specific work ratio, and further, the coal powder response residual information is obtained. The residual information of the pulverized coal response is used to distinguish between normal pulverized coal quantity changes caused by stress and the integrity of the coal and rock mass and pulverized coal quantity responses that may be caused by instantaneous pressure relief failure.

[0106] The observation interpretation weight refers to the degree to which the coal powder quantity response supports a certain instantaneous pressure relief state. Specifically, once the interactive Bayesian network has determined that the current drilling location is under a certain coal body stress state and coal-rock integrity state, to what extent should the actual observed coal powder quantity response support the existence of this instantaneous pressure relief state? The observation interpretation weight is a weighting coefficient that adjusts the reliability of the conditional probability of the coal powder quantity response. It reflects whether the currently observed coal powder quantity is more likely to originate from actual instantaneous pressure relief or from a high-stress background or a naturally fractured coal body background. For example, when the coal powder quantity response is significantly higher than the expected coal powder quantity and the corresponding positive residual is large, it indicates that the powder discharge phenomenon is more likely caused by instantaneous pressure relief. In this case, the observation interpretation weights for moderate and strong pressure relief responses are increased. Conversely, when the coal powder quantity is relatively high, but mainly caused by a high-stress state or natural coal body fracture, the observation interpretation weight for strong pressure relief responses is decreased.

[0107] First, using historical verified borehole data from the working face, the coal powder response state at each drilling location is statistically correlated with the instantaneous pressure relief state obtained from field verification. Then, the proportion of each instantaneous pressure relief state that is verified in the field is calculated under the same coal stress state, coal-rock integrity state, and coal powder response state. Finally, the occurrence proportion is normalized and used as the initial observation interpretation weight corresponding to each instantaneous pressure relief state.

[0108] When the residual response state A positive residual response indicates that the actual coal powder ratio is higher than the expected coal powder ratio under the current coal stress state and coal-rock integrity state, suggesting that there may be stronger drilling damage and immediate pressure relief response at this location. In this case, increasing the medium pressure relief response is recommended. Or strong pressure relief response The corresponding observation interpretation weights; when the residual response state When the residual response is near zero, it indicates that the actual coal powder quantity response basically matches the background coal powder discharge level. In this case, the original observation probability of the coal powder quantity response conditional probability table is maintained; when the residual response state... When the residual response is negative, it indicates that the actual coal powder ratio is lower than the expected coal powder ratio. In this case, the explanatory weight of the observation corresponding to the strong pressure relief response is reduced, while the explanatory weight of the observation with no obvious pressure relief response or weak pressure relief response is increased. The corrected observation probability is expressed as:

[0109] ;

[0110] in, This represents the observation probability after the residual response state correction; This indicates the observed probability of the coal powder quantity response before correction; Indicates the residual response state; Indicates the first Each drilling position is designed for immediate depressurization. The residual correction coefficient is calculated by first determining the residual information of the coal powder response at each drilling position based on historical construction data of the working face, verified borehole data in the same area, or stable data of the current drilling section. The residual response is then divided into negative residual response, near-zero residual response, and positive residual response according to a preset threshold. Subsequently, under each residual response state, the actual occurrence frequency of the four types of instantaneous pressure relief states is statistically analyzed and compared with the basic observation probability of the corresponding instantaneous pressure relief state without residual correction. The gain or attenuation amplitude of each instantaneous pressure relief state due to the residual response is calculated, and finally, the gain or attenuation amplitude is normalized to the residual correction coefficient.

[0111] It can avoid the mistaken judgment that the pressure relief is sufficient due to high natural discharge caused by high stress or incomplete coal body, and it can also avoid the mistaken judgment that the pressure relief is insufficient due to low natural discharge caused by low stress or intact coal body.

[0112] S25, the conditional prior and the corrected observation probability are jointly calculated, and the responses with no obvious pressure relief, weak pressure relief, moderate pressure relief and strong pressure relief are normalized respectively to obtain the posterior probability set corresponding to each instantaneous pressure relief state. The posterior probability set is determined as the instantaneous pressure relief posterior discriminant distribution.

[0113] For the For each drilling location, the joint probability term for the four types of instantaneous depressurization states is calculated as follows:

[0114] ;

[0115] in, Indicates the first Each drilling position corresponds to an immediate pressure relief state. The joint probability term;

[0116] Indicates the corrected observation probability; Indicates a priori conditions; .

[0117] The joint probability terms corresponding to the four types of immediate depressurization states are normalized to obtain the posterior probability representation for each immediate depressurization state as follows:

[0118] ;

[0119] in, Indicates the first The posterior probability of the instantaneous decompression state at a drilling location after observing the coal powder quantity response state, coal stress state, coal and rock mass integrity state, and residual response state. Indicates the first The drilling position corresponds to the first... The joint probability term for a state similar to immediate depressurization; It represents the sum of the joint probability terms of the four types of instantaneous depressurization states, i.e., the total probability normalization constant.

[0120] Obtain the The set of posterior probabilities for each drilling location is represented as:

[0121]

[0122] in, Indicates the first Immediate decompression post-hoc discriminant distribution at each drilling location.

[0123] The immediate post-test discriminant distribution reflects the probability that the current drilling location, under the combined influence of the coal powder quantity response, coal stress state, coal-rock integrity state, and residual response state, belongs to no obvious pressure relief response, weak pressure relief response, moderate pressure relief response, and strong pressure relief response, respectively.

[0124] S3, select the instant pressure relief state with the highest posterior probability as the instant pressure relief discrimination level of the corresponding drilling position according to the instant pressure relief posterior discrimination distribution, and match the instant pressure relief discrimination level with the confidence threshold, drilling depth position and field verification true value, output the borehole instant pressure relief discrimination result, and update the interactive Bayesian network discrimination model based on the field verification true value.

[0125] S31, read the... The posterior discriminant distribution of instantaneous pressure relief corresponding to each drilling location, wherein the instantaneous pressure relief posterior discriminant distribution is used to represent the posterior probability that the drilling location belongs to no obvious pressure relief response, weak pressure relief response, moderate pressure relief response, and strong pressure relief response, respectively, is expressed as follows:

[0126] ;

[0127] in, Indicates the first Immediate decompression post-hoc discriminant distribution at each drilling location; This indicates no significant pressure relief response; Indicates a weak pressure relief response; Indicates a moderate pressure relief response; Indicates a strong pressure relief response; Indicates the first The joint observation evidence corresponding to each drilling location includes the coal powder quantity response state, coal stress state, coal and rock mass integrity state, and residual response state.

[0128] By comparing the posterior probabilities of the four types of immediate depressurization states in the immediate depressurization posterior discriminant distribution, the immediate depressurization state with the highest posterior probability is determined as the candidate immediate depressurization state, expressed as:

[0129] ;

[0130] in, Indicates the first Candidate instantaneous depressurization status of each drilling location; Indicates the first Similar to an immediate depressurization state; ; This indicates the selection of the instantaneous depressurization state that maximizes the posterior probability.

[0131] The maximum posterior probability is determined as the i-th The decision credibility of each drilling location is expressed as:

[0132] ;

[0133] in, Indicates the first The credibility of the decision regarding the drilling location; This indicates that the maximum value is selected from the posterior probabilities corresponding to the four types of immediate depressurization states.

[0134] The higher the decision credibility, the better the interactive Bayesian network discriminant model supports the candidate instantaneous decompression state; the closer the decision credibility is to the case where the probabilities of each state are equally distributed, the more it indicates that the current observation evidence supports multiple instantaneous decompression states to a certain extent, and the higher the discrimination uncertainty.

[0135] S32, the first The decision confidence level at each drilling location is compared with a preset confidence threshold; the preset confidence threshold is denoted as... First, historical verified borehole data from the working face are collected. The immediate pressure relief posterior discriminant distribution at each drilling location is paired with the corresponding field verification true value. The number of correctly and incorrectly discriminated samples within different maximum posterior probability intervals is counted. Then, multiple candidate confidence thresholds are selected with a certain step size. For each candidate threshold, its corresponding discrimination accuracy, misclassification rate, and required verification ratio are calculated. When the maximum posterior probability reaches a certain candidate confidence threshold, if its corresponding discrimination accuracy meets the preset requirements and the misclassification rate meets the field anti-impact safety requirements, then the candidate value is determined as the preset confidence threshold. The preset confidence threshold is preferably 0.62.

[0136] When the decision confidence level reaches or exceeds a preset confidence threshold, the candidate instantaneous depressurization state is determined as the instantaneous depressurization discrimination level of the corresponding drilling position, represented as follows:

[0137] ;

[0138] in, Indicates the first Real-time pressure relief assessment level for each drilling location; Indicates the candidate's immediate depressurization status; Indicates the credibility of the decision; This indicates the preset confidence threshold.

[0139] When the decision credibility is lower than the preset credibility threshold, the candidate immediate depressurization state is not directly used as a deterministic judgment conclusion. Instead, the corresponding drilling position is marked as a position that needs to be verified.

[0140] ;

[0141] in, Indicates the first Each drilling location needs to be checked; This indicates that no review is required. This indicates that a review is required.

[0142] For locations requiring verification, retain the corresponding candidate immediate depressurization states, immediate depressurization posterior discrimination distribution, decision credibility, coal powder response residual information, and drilling depth location. This allows for further verification in conjunction with microseismic monitoring, stress monitoring, borehole deformation, or re-drilling results. By setting a credibility threshold, it is possible to avoid forcibly outputting deterministic results when the posterior probabilities of multiple immediate depressurization states are close, thereby improving the reliability of borehole immediate depressurization discrimination results.

[0143] For example, if the maximum posterior probability of a certain drilling location is 0.75, the preset confidence threshold is 0.62, and the maximum posterior probability corresponds to a moderate pressure relief response. Since 0.75 > 0.62, the instantaneous pressure relief judgment level of the drilling location is determined to be a medium pressure relief response, and no verification mark is set.

[0144] S33, the instant pressure relief discrimination level, decision credibility, instant pressure relief post-verification discrimination distribution, and verification mark are bound to the corresponding borehole number and drilling depth position to form a drilling position level discrimination record; The discrimination record corresponding to each drilling position is represented as follows:

[0145] ;

[0146] in, Indicates the first Records of drilling locations; Indicates the corresponding borehole number; Indicates the corresponding drilling depth position; Indicates the immediate pressure relief assessment level; Indicates the credibility of the decision; This indicates the immediate post-test discriminant distribution after depressurization; This indicates that a review is required.

[0147] Using borehole number as one dimension and drilling depth position as another dimension, the identification records of multiple drilling positions are written into a two-dimensional grid matrix of borehole number and drilling depth position; the two-dimensional grid matrix is ​​represented as follows:

[0148] ;

[0149] in, Represents a two-dimensional grid matrix; Indicates the borehole number is Drilling depth position is Grid cells; This indicates the total number of boreholes included in the assessment; This represents the total number of drilling depth locations divided according to the preset drilling length intervals; each grid cell is used to store the instantaneous pressure relief judgment level, decision confidence, instantaneous pressure relief post-judgment distribution, and verification mark for the corresponding location.

[0150] Based on the two-dimensional grid matrix, generate instant depressurization status classification results, posterior probability distribution results, and a list of locations requiring verification.

[0151] First, one coordinate dimension of the two-dimensional grid matrix is ​​determined according to the borehole number, and another coordinate dimension is determined according to the drilling depth. The instantaneous pressure relief discrimination level, decision confidence, instantaneous pressure relief posterior discrimination distribution, and verification markers stored in each grid cell are then read. Subsequently, based on the instantaneous pressure relief discrimination level corresponding to each grid cell, different classification labels are assigned to no obvious pressure relief response, weak pressure relief response, moderate pressure relief response, and strong pressure relief response. These are then arranged according to the borehole number and drilling depth to form the instantaneous pressure relief state classification results, displaying the changes in instantaneous pressure relief state along the depth direction for each borehole and the continuity of pressure relief between adjacent boreholes. Simultaneously, the maximum posterior probability is extracted from the instantaneous pressure relief posterior discrimination distribution of each grid cell, or four... The posterior probability corresponding to the instantaneous depressurization state is written into the probability matrix according to the original grid position. The posterior probability distribution result and probability cloud map are generated through hierarchical display or adjacent grid smoothing. The adjacent grid smoothing is only used to improve the display continuity and does not change the original posterior probability. Finally, the verification marks in the two-dimensional grid matrix are traversed, and the grid cells with valid verification marks are selected. Their borehole number, drilling depth position, candidate instantaneous depressurization state, decision confidence, instantaneous depressurization posterior discrimination distribution and coal powder response residual information are extracted. They are arranged in order of decision confidence from low to high or drilling depth position to form a verification position list. Thus, the state classification, probability magnitude and uncertain position are output as the borehole instantaneous depressurization discrimination result.

[0152] When forming a probability cloud map from the posterior probability distribution results, the immediate pressure relief posterior discriminant distribution corresponding to each grid cell in the two-dimensional grid matrix is ​​first read, and the maximum posterior probability is selected as the probability value of that grid cell. Alternatively, the posterior probabilities corresponding to no obvious pressure relief response, weak pressure relief response, moderate pressure relief response, and strong pressure relief response are selected to generate different categories of probability maps. Then, with the borehole number as the horizontal coordinate and the drilling depth position as the vertical coordinate, the posterior probability of each grid cell is mapped to the corresponding spatial position and divided into multiple probability levels according to the probability magnitude. Different colors or gray levels are assigned to different probability levels so that areas with higher posterior probabilities are displayed more clearly and areas with lower posterior probabilities are displayed less clearly. Subsequently, spatial continuity constraints are applied to adjacent grid cells for interpolation and smoothing to form a continuous transition in probability changes between boreholes and along the drilling depth direction, thereby eliminating probability jumps caused by discrete sampling. However, the smoothing process is only used to improve the display effect and does not change the original posterior probability value of each grid cell. Finally, the probability distribution of all grid cells is continuously drawn according to the spatial position of the borehole to form a probability cloud map reflecting the spatial distribution characteristics of the immediate pressure relief effect.

[0153] The instant pressure relief status classification results are used to display the pressure relief status of each drilling position according to different instant pressure relief discrimination levels; the posterior probability distribution results can be further used to form a probability cloud map to display the pressure relief probability changes at different boreholes and different drilling depths; the list of locations requiring verification is used to centrally list drilling positions with decision credibility lower than a preset credibility threshold; the instant pressure relief status classification results, the posterior probability distribution results, and the list of locations requiring verification together constitute the borehole instant pressure relief discrimination results.

[0154] Specifically, the immediate pressure relief status classification results are first used as the final discrimination conclusions for different drilling depths of each borehole. This reflects which immediate pressure relief state each drilling location belongs to: no obvious pressure relief response, weak pressure relief response, moderate pressure relief response, or strong pressure relief response. Then, the posterior probability distribution results are mapped one-to-one with the corresponding immediate pressure relief status classification results. A corresponding posterior probability and its spatial distribution information are added to each discrimination conclusion to characterize the credibility of the judgment and the probability change characteristics along the borehole and between different boreholes. Finally, the low-credibility drilling locations recorded in the list of locations requiring verification are associated with the above discrimination results. A verification requirement is added to the corresponding drilling location, and its candidate immediate pressure relief status, posterior probability distribution, and decision credibility information are retained, prompting on-site key verification. After integration, each drilling location simultaneously possesses an immediate pressure relief status, corresponding posterior probability information, and a verification requirement indicator. All drilling locations are uniformly organized according to borehole number and drilling depth, thus forming a complete borehole immediate pressure relief discrimination result. .

[0155] Using 1.0m as the spatial grid span, the instantaneous pressure relief discrimination level, decision confidence, and instantaneous pressure relief posterior discrimination distribution of each 1.0m drilling section are written into the corresponding grid cells, and a probability cloud map is formed by continuous interpolation or smoothing display of adjacent grids; the smoothing display is only used for result visualization and does not change the original posterior probability and instantaneous pressure relief discrimination level of each drilling position.

[0156] S34, Judging the immediate pressure relief result from the borehole Extract the instantaneous pressure relief judgment level, decision credibility, instantaneous pressure relief post-verification judgment distribution, and verification mark corresponding to each drilling depth position, obtain the field verification data corresponding to each drilling depth position, and determine the field verification true value based on the field verification data; the field verification true value is determined based on at least one of the following: microseismic monitoring released energy, coal stress change at adjacent monitoring positions, borehole wall deformation, borehole collapse or closure records, and re-drilling verification results.

[0157] Specifically, microseismic monitoring of released energy reflects the degree of elastic energy release in the coal and rock mass after drilling disturbance; changes in coal stress at adjacent monitoring locations reflect whether local stress has decreased after drilling; borehole wall deformation, collapse, or closure records reflect whether the surrounding coal structure has been significantly weakened or damaged; re-drilling verification results reflect whether a low-resistivity fractured zone has formed in the coal mass surrounding the original borehole; the above verification information is mapped according to preset verification rules to no obvious pressure relief response, weak pressure relief response, moderate pressure relief response, or strong pressure relief response to obtain the true values ​​for field verification.

[0158] ;

[0159] in, Indicates the first The true value of the field verification at each drilling location.

[0160] Specifically, the process involves first acquiring the microseismic monitoring energy release at the current drilling depth, the stress changes in the coal body at adjacent monitoring locations, borehole wall deformation, borehole collapse or closure records, and re-drilling verification results. These are then converted into corresponding verification levels according to preset verification rules. Higher microseismic monitoring energy release indicates more complete release of elastic energy in the coal and rock mass after drilling disturbance; a greater decrease in coal body stress at adjacent monitoring locations indicates more significant local stress relief; more significant borehole wall deformation or moderate collapse / closure indicates significant structural weakening of the surrounding coal body; and a decrease in drilling resistance, reduction in coal dust per unit drilling length, or more pronounced fracture zone formation during re-drilling indicates the formation of an effective stress relief zone around the original borehole. Subsequently, the above-mentioned... The verification level is comprehensively determined. When none of the verification indicators show obvious stress relief characteristics, the field verification true value is mapped to no obvious stress relief response. When only some verification indicators show slight stress release or local structural weakening, it is mapped to a weak stress relief response. When most verification indicators indicate that the coal body has undergone significant stress release and structural damage, it is mapped to a moderate stress relief response. When the micro-vibration energy release, the magnitude of coal body stress reduction, the degree of borehole wall structural damage, and the re-drilling verification results all indicate that a stable and continuous stress relief zone has been formed around the borehole, it is mapped to a strong stress relief response. If there are conflicts among the verification indicators, a comprehensive decision is made according to the pre-set verification priority or weighting rules to finally obtain the field verification true value for the corresponding drilling location.

[0161] The pre-defined verification rules are as follows: First, standardized evaluations are performed on the energy released by microseismic monitoring, stress changes in coal seams at adjacent monitoring locations, borehole wall deformation, borehole collapse or closure records, and re-drilling verification results. Each verification information item is divided into four verification levels: no obvious response, slight response, moderate response, and significant response. Then, based on the pre-established verification level correspondence, each verification level is mapped to the degree of support for no obvious pressure relief response, weak pressure relief response, moderate pressure relief response, and strong pressure relief response. The number of supports or the comprehensive weight of each instantaneous pressure relief state are counted. When most verification indicators jointly support the same instantaneous pressure relief state, that instantaneous pressure relief state is directly taken as the true value for field verification; otherwise... For different immediate pressure relief states corresponding to the same verification index, a comprehensive decision is made according to the preset priority. Among them, the energy released by microseismic monitoring and the stress change of coal at adjacent monitoring locations are verification information that directly reflects the pressure relief effect and have a higher priority than borehole wall deformation, borehole collapse or closure records. The re-drilling verification result has the highest priority as the final verification basis. When the re-drilling verification result is inconsistent with other verification information, the re-drilling verification result shall prevail. If the re-drilling verification result is missing, the field verification true value shall be determined according to the comprehensive support of each verification index, so as to ensure that the field verification true value can truly reflect the actual immediate pressure relief effect at the current drilling position and provide a reliable label for the incremental update of the interactive Bayesian network discrimination model.

[0162] The true values ​​verified on-site are matched with the instantaneous pressure relief discrimination levels. When the two are consistent, the corresponding drilling location is determined as a consistent discrimination sample; when the two are inconsistent, the corresponding drilling location is determined as a discrimination deviation sample.

[0163] ;

[0164] in, Indicates the first Discriminant matching markers for each drilling location; Indicates samples that are considered consistent; Indicates a sample with discrimination bias; Indicates the immediate pressure relief assessment level; This indicates the true value verified on-site.

[0165] For locations requiring verification, the on-site verification truth value is directly included in the incremental update sample after it is obtained. For locations that have already output a deterministic immediate depressurization judgment level, the on-site verification truth value is also used to test the judgment result of the interactive Bayesian network discrimination model, thereby avoiding the situation where only the correct samples are used to update the interactive Bayesian network discrimination model and the conditional probability bias reflected by the misjudged samples is ignored.

[0166] S35, based on the matching result of the borehole instant pressure relief discrimination result and the field verification true value, the discrimination consistent sample is... and discrimination bias samples Together, they serve as incremental update samples to update the instantaneous depressurization state conditional probability table, pulverized coal quantity response conditional probability table, and residual correction coefficients in the interactive Bayesian network discriminant model.

[0167] For the conditional probability table of immediate stress relief state, the occurrence frequency of the immediate stress relief state corresponding to each field verification true value is statistically analyzed under different combinations of coal stress state and coal-rock integrity state for newly added samples. Based on the original sample count and the new sample count, the conditional prior representation is recalculated as follows:

[0168]

[0169] in, This represents the conditional probability of the updated immediate depressurization state.

[0170] This indicates the stress state of the coal seam before the update. and the integrity status of coal and rock mass Immediate depressurization state The historical sample count is determined by first retrieving drilling location samples from the historical sample database up to the time of this model update. Each historical sample should have at least the coal stress ratio, fracturing power ratio, borehole number, drilling depth location, and field verification true value. Samples with missing field verification true values, mismatched drilling depth locations, abnormal drilling parameters, duplicate storage, or incomplete state mapping information are removed. Then, for each valid historical sample, the coal stress ratio and fracturing power ratio are subjected to state mapping consistent with the current model, mapping the coal stress ratio to one of the following: low stress state, normal stress state, or high stress state. The crushing specific power ratio is mapped to one of the following states: crushed coal body, conventional integrity, or intact coal body. Then, the true value of the field verification for the historical sample is determined by the energy released from microseismic monitoring, stress changes in coal body at adjacent monitoring locations, borehole wall deformation, borehole collapse or closure records, and re-drilling verification results. This true value is then classified as one of the following: no significant stress relief response, weak stress relief response, moderate stress relief response, or strong stress relief response. Sample combinations are then established based on the coal body stress state, coal-rock integrity state, and field verification true value. All valid historical samples are evaluated one by one. When the coal body stress state of a certain historical sample equals the specified... The integrity state of the coal and rock mass is equal to the specified And the on-site verification showed that the true value was equal to the specified instantaneous depressurization state. When a condition is not met, the historical sample count for that combination is incremented by one; if any condition is not met, the combination is not included.

[0171] This indicates the number of newly added samples. First, samples with missing field verification values, inconsistent drilling depths, incomplete descriptions of self-calibrated stress relief observations during drilling, and duplicate records are removed from the new samples. Then, a coal stress ratio state mapping function is applied to the coal stress ratio of each valid new sample to determine its coal stress state. A breakage specific power ratio state mapping function is applied to the breakage specific power ratio to determine its coal-rock integrity state. The actual instantaneous stress relief state is then determined using the obtained field verification values. Subsequently, the combination of coal stress state, coal-rock integrity state, and field verification values ​​is counted one by one. When a new sample is simultaneously in a specified coal stress state... , specified coal and rock mass integrity status Furthermore, its on-site verification showed that the true value was an immediate depressurization state. Each time, the corresponding sample count is incremented once. The cumulative count obtained after traversing all newly added valid samples is the number of newly added samples in this instance. .

[0172] The probability smoothing coefficient is used to avoid zero probability for small sample conditional combinations. First, several candidate smoothing coefficients are determined based on the number of samples for each state combination in the historically validated samples. Larger candidate values ​​are set when there are fewer samples and the state distribution is more uneven, and smaller candidate values ​​are set when there are more samples. Then, the historical samples are divided into parameter calibration samples and validation samples. The conditional probability table of instantaneous decompression state is updated using each candidate smoothing coefficient. The instantaneous decompression state discrimination accuracy, log loss, probability calibration error, and zero probability occurrence of the updated model are compared on the validation samples. The candidate value that can eliminate zero probability, does not cause excessive convergence of the probabilities of the four types of instantaneous decompression states, and has the smallest validation error is selected as the probability smoothing coefficient.

[0173] This represents the enumeration sequence number of the immediate depressurization state.

[0174] For the conditional probability table of pulverized coal quantity response, the occurrence frequency of low pulverized coal response, conventional pulverized coal response, and high pulverized coal response was statistically analyzed under different field verification true values, coal stress states, and coal-rock integrity states, and the corresponding observation probabilities of pulverized coal quantity response were recalculated:

[0175]

[0176] in, This represents the updated conditional probability of the pulverized coal quantity response;

[0177] Indicates the current coal powder quantity response status; This represents a set of coal powder quantity response states, which includes low coal powder response, normal coal powder response, and high coal powder response.

[0178] This indicates the number of historical samples corresponding to the state combination before the update. First, historical samples with missing coal powder ratio, coal body stress ratio, crushing specific power ratio, or field verification true values, as well as those with mismatched drilling depths and locations, and duplicate records, are removed. Then, according to the state mapping rules used in the current interactive Bayesian network discriminant model, the coal powder ratio of each historical sample is mapped to one of low coal powder response, normal coal powder response, or high coal powder response; the coal body stress ratio is mapped to one of low stress state, normal stress state, or high stress state; and the crushing specific power ratio is mapped to one of crushed coal body state, normal integrity state, or intact coal body state. Based on the field verification true value of the historical sample, its actual instantaneous decompression state is determined. Subsequently, the combinations of coal powder response state, instantaneous decompression state, coal body stress state, and coal-rock integrity state are counted one by one. When the coal powder response state of a certain historical sample is... The on-site verification showed that the true value was in an immediate depressurization state. The stress state of the coal body is the first The coal and rock mass is in the first state and its integrity is the second state. When a class state is reached, the corresponding count is incremented once. The cumulative count obtained after traversing all valid historical samples is the number of historical samples for the corresponding state combination before the update. .

[0179] This indicates the number of samples with newly added corresponding state combinations. First, samples with consistent and biased discrimination added within this update cycle are screened, and samples with incomplete information, missing field verification true values, inconsistent drilling depths, or those already counted in the previous update cycle are removed. Then, the coal powder ratio of each new sample is mapped to the coal powder response state, the coal stress ratio is mapped to the coal stress state, and the crushing specific work ratio is mapped to the coal-rock integrity state. The immediate decompression state is determined using the obtained field verification true values. When a new sample simultaneously satisfies the coal powder response state... The on-site verification showed that the true value was in an immediate depressurization state. The stress state of the coal body is the first The coal and rock mass is in the first state and its integrity is the second state. When a new state is reached, the count is incremented once for each new valid sample. The cumulative count after all new valid samples have been counted is the total number of new samples. .

[0180] Represents the set of response states to pulverized coal quantity. The state index used when traversing the response states of each coal powder quantity corresponds to low coal powder response, normal coal powder response, and high coal powder response, respectively.

[0181] For the residual correction coefficient, based on the proportion of occurrence of the true values ​​of each field verification under different residual response states, the gain or attenuation effect of positive residuals, near-zero residuals, and negative residuals on the observation probability of the four types of instantaneous pressure relief states is re-evaluated. If the newly added discrimination bias samples indicate that the support level of positive residuals for strong pressure relief responses is set too high, the residual correction coefficient corresponding to strong pressure relief responses is reduced. If the newly added samples indicate that positive residuals have a more stable correspondence with medium or strong pressure relief responses, their residual correction coefficients are increased accordingly. The updated residual correction coefficients still maintain the basic direction of the near-zero residual correspondence coefficient being close to 1, positive residuals enhancing the explanatory weight of medium and strong pressure relief responses, and negative residuals enhancing the explanatory weight of responses with no obvious pressure relief and weak pressure relief responses.

[0182] Specifically, firstly, the samples with consistent discrimination and samples with discriminative bias obtained in this study are included in the incremental update samples. The determined field verification true value is used as the label for the immediate depressurization state, instead of using the original output of the model's immediate depressurization discrimination level as the update basis. Then, the coal powder response residual information of each incremental update sample is read and mapped to negative residual response, near-zero residual response, or negative residual response. The observation probability of the coal powder quantity response, the original residual correction coefficient, and the field verification true value of the sample before residual correction are recorded. Sample groups are established according to the residual response state and the immediate depressurization state. For each group, historical samples are statistically analyzed. The proportion of true values ​​appearing in each field validation sample in the newly added samples is used to determine whether the current residual correction coefficient supports the corresponding immediate decompression state in accordance with the field validation results. For example, if the proportion of field validation results showing moderate or strong decompression responses in the positive residual response samples is higher than the support level expressed by the current model after correction, the residual correction coefficients corresponding to moderate and strong decompression responses should be appropriately increased. If the positive residual response samples are actually mostly no obvious decompression response or weak decompression response, it indicates that the original positive residual gain setting is too strong, and the residual correction coefficients corresponding to moderate and strong decompression responses should be reduced. The correction coefficients for high residual responses with no obvious pressure relief or weak pressure relief are specified. For negative residual responses, updates are performed in the opposite direction, adjusting the gain or attenuation of the response to no obvious pressure relief, weak pressure relief, moderate pressure relief, and strong pressure relief based on field verification results. For near-zero residual responses, the corresponding residual correction coefficients are gradually maintained or reduced to near 1 to avoid excessive changes to the conditional probability of the raw coal powder quantity response. During the update process, historical sample statistics and newly added sample statistics are merged according to the sample size or preset update weights. When the sample size is small, the single adjustment amplitude is reduced. When the sample size is sufficient and the validation results are stable, the adjustment range is increased, and upper and lower limits are set for the residual correction coefficients to prevent abrupt changes in coefficients caused by abnormal samples in a single batch. After adjusting the coefficients corresponding to each instantaneous depressurization state, the corrected observation probabilities are normalized to ensure that the relative probability relationship between the four types of instantaneous depressurization states is effective. Finally, the updated residual correction coefficients are used to recalculate the posterior discriminant distribution of the instantaneous depressurization of the incremental validation samples. When the discrimination accuracy and probability reliability are improved, the set of coefficients is written into the interactive Bayesian network discrimination model; otherwise, the update range is reduced or the original residual correction coefficients are retained.

[0183] After the update is completed, an updated interactive Bayesian network discriminant model is formed. The updated interactive Bayesian network discriminant model is then used to process the description of the self-calibration pressure relief observations at subsequent drilling locations to obtain new real-time pressure relief posterior discriminant distributions and real-time pressure relief discriminant results for the borehole. This allows the model to be continuously corrected as field verification samples accumulate.

[0184] This invention encompasses any substitutions, modifications, equivalent methods, and solutions made within the spirit and scope of this invention. To provide the public with a thorough understanding of this invention, specific details are described in detail in the following preferred embodiments; however, those skilled in the art will fully understand the invention even without these details. Furthermore, to avoid unnecessary misunderstanding of the essence of this invention, well-known methods, processes, procedures, components, and circuits are not described in detail.

[0185] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for judging the instant pressure relief effect of large-diameter boreholes based on Bayesian networks, characterized in that, Includes the following steps: Based on the drilling process of large-diameter pressure relief boreholes, the drilling parameters and coal powder content per unit drilling length are obtained, and the self-calibration reference section of the plastic zone at the opening end of the current construction borehole is used as the reference. The drilling parameters and the amount of coal powder per unit drilling length are benchmarked in the borehole to obtain the self-calibrated pressure relief observation description during drilling. The self-calibrated pressure relief observation description during drilling includes the ratio of coal powder amount, the ratio of coal body stress, the ratio of crushing specific power, and the residual information of coal powder response. Based on the self-calibrated stress relief observation description input interactive Bayesian network discrimination model during drilling, the conditional prior of the instantaneous stress relief state is formed using the coal stress ratio and the crushing specific power ratio. Based on the instantaneous depressurization state, coal stress ratio, and crushing specific power ratio, the coal powder quantity ratio and coal powder response residual information are jointly interpreted to obtain the instantaneous depressurization posterior discrimination distribution; Based on the aforementioned immediate pressure relief posterior discriminant distribution, the immediate pressure relief state with the highest posterior probability is selected as the immediate pressure relief discriminant level for the corresponding drilling position. The instant pressure relief discrimination level is matched with the confidence threshold, drilling depth location and field verification true value to output the instant pressure relief discrimination result of the borehole; The interactive Bayesian network discriminant model is updated based on the on-site verification truth value; The coal powder amount per unit drilling length, coal stress value, and crushing specific power value at the current drilling position are respectively benchmarked against the coal powder amount benchmark value, coal stress benchmark value, and crushing specific power benchmark value in the borehole to obtain the coal powder amount ratio, coal stress ratio, and crushing specific power ratio. The expected coal powder quantity ratio under the current stress state and coal-rock mass integrity state is determined using the coal body stress ratio and the crushing specific work ratio. The deviation between the current coal powder quantity ratio and the expected coal powder quantity ratio is then processed to obtain the coal powder response residual information, which is expressed as follows: ;in, Indicates the first residual information of pulverized coal response at each drilling location; Indicates the first The ratio of actual pulverized coal quantity at each drilling location; Indicates the first The expected coal powder ratio at each drilling location.

2. The method for judging the instant pressure relief effect of large-diameter boreholes based on Bayesian networks according to claim 1, characterized in that, During the construction of large-diameter pressure relief boreholes, drilling torque, rig speed, thrust, feed rate, and drilling depth are continuously acquired at preset drilling length intervals as drilling-while-drilling parameters. Simultaneously, the amount of coal dust discharged within each preset drilling length interval is also acquired, forming the original observation sequence for drilling-while-drilling, as shown below: ； in, This represents the original observation sequence during drilling. Indicates the drilling position number; This indicates the total number of preset drilling length intervals that have been recorded in the current drilling hole; Indicates the first The drilling depth position corresponding to a preset drilling length interval; Indicates the first Drilling torque within a preset drilling length interval; Indicates the first Drilling speed within a preset drilling length interval; Indicates the first The thrust within a preset drilling length interval; Indicates the first Feed rate within a preset drilling length interval; Indicates the first The amount of coal dust discharged within a preset drilling length interval.

3. The method for judging the instant pressure relief effect of large-diameter boreholes based on Bayesian networks according to claim 2, characterized in that, Based on the original observation sequence during drilling, the drilling parameters at each drilling location were inverted and converted to obtain the coal stress value and crushing specific work value at the corresponding drilling location, respectively expressed as: ； ； in, Indicates the first Coal stress values ​​at each drilling location; Indicates the first The specific power value of the breaking work at each drilling location; This represents the calibration function for inverting coal stress values ​​from drilling parameters; This represents the calibration function for converting the specific power of the crushing process from the parameters while drilling, and defines the amount of pulverized coal within the same preset drilling length interval as the amount of pulverized coal per unit drilling length. A preset depth range extending into the hole from the current borehole opening end is selected as the self-calibration reference segment of the plastic zone. Based on the coal powder content per unit drilling length, coal stress value, and crushing specific energy value within the self-calibration reference segment of the plastic zone, the reference values ​​for coal powder content, coal stress, and crushing specific energy are respectively determined as follows: ； ； ； in, This indicates the reference value for the amount of pulverized coal within the self-calibration reference section of the plastic zone; This represents the reference value of coal stress within the self-calibration reference section of the plastic zone; This represents the reference value of the breaking specific work within the self-calibration reference section of the plastic zone; Indicates the drilling depth position The amount of pulverized coal per unit drilling length at the location; Indicates the drilling depth position The stress value of the coal body at that location; Indicates the drilling depth position The specific work value of the breakage at the location; This indicates median operations.

4. The method for judging the instant pressure relief effect of large-diameter boreholes based on Bayesian networks according to claim 3, characterized in that, The coal powder ratio, coal body stress ratio, crushing specific power ratio, and coal powder response residual information in the drilling self-calibration pressure relief observation description are respectively mapped to obtain the coal powder response state, coal body stress state, coal-rock mass integrity state, and residual response state, which are respectively expressed as: ； ； ； ； in, Indicates the first The coal powder quantity response status at each drilling location; Indicates the first Stress state of coal at each drilling location; Indicates the first The integrity status of the coal and rock mass at each drilling location; Indicates the first The residual response state at each drilling location; A state mapping function representing the ratio of pulverized coal quantity; A state mapping function representing the stress ratio of coal seams; The state mapping function represents the ratio of breakage to power. A state mapping function representing the residual information of pulverized coal response.

5. The method for judging the instant pressure relief effect of large-diameter boreholes based on Bayesian networks according to claim 4, characterized in that, In the interactive Bayesian network discriminant model, an instantaneous pressure relief state is set, and the instantaneous pressure relief state is divided into no obvious pressure relief response, weak pressure relief response, moderate pressure relief response and strong pressure relief response. Based on the stress state of the coal body and the integrity state of the coal and rock mass, the conditional priors corresponding to different instantaneous pressure relief states are obtained by querying the conditional probability table of instantaneous pressure relief states. Based on the instantaneous decompression state, the coal body stress state, and the coal-rock mass integrity state, the probability of the coal powder quantity response under the combined effects of different instantaneous decompression states, different coal body stress states, and different coal-rock mass integrity states is obtained by querying the coal powder quantity response conditional probability table. ； in, Indicates that in the immediate depressurization state The stress state of the coal body is The integrity state of the coal and rock mass is as follows: Under these conditions, the coal powder quantity response state was observed. The probability of; Indicates the first The coal powder quantity response status at each drilling location; Indicates the first Stress state of coal at each drilling location; Indicates the first The integrity status of the coal and rock mass at each drilling location.

6. The method for judging the instant pressure relief effect of large-diameter boreholes based on Bayesian networks according to claim 5, characterized in that, The observation probability is corrected using the residual response state to enhance the explanatory weight of strong or moderate pressure relief responses by increasing the positive deviation of the actual coal powder ratio from the expected coal powder ratio, and to reduce the misjudgment of the immediate pressure relief state by the background powder discharge response caused by high stress or incomplete coal and rock mass. The corrected observation probability is expressed as follows: ； in, This represents the observation probability after the residual response state correction; This indicates the observed probability of the coal powder quantity response before correction; Indicates the residual response state; Indicates the first Each drilling position is designed for immediate depressurization. The residual correction factor; Indicates the first Similar to an immediate depressurization state; The conditional prior and the modified observation probability are jointly calculated, and the responses with no obvious pressure relief, weak pressure relief, moderate pressure relief and strong pressure relief are normalized respectively to obtain the posterior probability set corresponding to each instantaneous pressure relief state. The posterior probability set is determined as the instantaneous pressure relief posterior discriminant distribution.

7. The method for judging the instant pressure relief effect of large-diameter boreholes based on Bayesian networks according to claim 6, characterized in that, Read the posterior discriminant distribution of the instantaneous pressure relief at the corresponding drilling location, and compare the posterior probabilities corresponding to no obvious pressure relief response, weak pressure relief response, moderate pressure relief response, and strong pressure relief response, respectively. Determine the instantaneous pressure relief state with the highest posterior probability as the candidate instantaneous pressure relief state, and express the decision confidence of determining the corresponding drilling location with the highest posterior probability as follows: ； in, Indicates the first The credibility of the decision regarding the drilling location; This indicates that the maximum value is selected from the posterior probabilities corresponding to the four types of immediate depressurization states. Indicates the first Similar to an immediate depressurization state; ; Indicates the first Joint observational evidence corresponding to each drilling location.

8. The method for judging the instant pressure relief effect of large-diameter boreholes based on Bayesian networks according to claim 7, characterized in that, The decision credibility is compared with a preset credibility threshold; when the decision credibility reaches or exceeds the preset credibility threshold, the candidate instant decompression state is determined as the instant decompression discrimination level of the corresponding drilling position. When the decision confidence level is lower than the preset confidence level threshold, the corresponding drilling position is marked as a position that needs to be reviewed, and the candidate instant depressurization state and its instant depressurization posterior discrimination distribution are retained. The instant pressure relief discrimination level, the decision credibility, the instant pressure relief posterior discrimination distribution, and the verification mark are bound to the corresponding borehole number and drilling depth position. A two-dimensional grid matrix is ​​formed according to the borehole number and drilling depth position. Based on the two-dimensional grid matrix, the instant pressure relief state classification result, the posterior probability distribution result, and the verification position list are generated to obtain the borehole instant pressure relief discrimination result.

9. The method for judging the instant pressure relief effect of large-diameter boreholes based on Bayesian networks according to claim 8, characterized in that, Obtain the field verification true value corresponding to the drilling depth position, match the field verification true value with the instant pressure relief discrimination level, and determine the discrimination consistent sample and the discrimination deviation sample; wherein, the field verification true value is determined based on at least one of the following: microseismic monitoring energy release, coal stress change at adjacent monitoring positions, borehole wall deformation, borehole collapse or closure records, and re-drilling verification results; The consistent discrimination samples and the biased discrimination samples are used as incremental update samples to update the instantaneous pressure relief state conditional probability table, pulverized coal quantity response conditional probability table, and residual correction coefficient in the interactive Bayesian network discrimination model. The updated interactive Bayesian network discrimination model is then used to perform instantaneous pressure relief effect discrimination on the self-calibrated pressure relief observation description of subsequent drilling positions.

Images