A multi-objective optimization decision system for tunneling parameters of a mine tunneling machine

Abstract

The application discloses a kind of multi-objective optimization decision system of mine heading machine heading parameter, it is related to mine intelligent heading control technical field, obtains the cutting motor current in the preset test cutting stroke of target heading face, cutting arm vibration, pick temperature, gas concentration, advancing pressure and cutting head angle position, and is divided into operating data group according to cutting head angle position;Current ripple phase lag value, vibration frequency multiplication energy ratio, pick temperature rise residual value and gas delay release amount are calculated, and disturbance feature set is generated;The disturbance feature set is mapped to heading section partition grid, and the composite stratum response field including equivalent cuttable coefficient, mudification adhesion coefficient and pressure relief gas sensitive coefficient is established;Different heading parameter combinations are simulated based on the composite stratum response field, and combined with machine body deflection, gas instantaneous release risk, pick wear and unit energy consumption weight correction, output target heading parameter;The application can improve the safety and stability of heading parameter decision under composite abnormal stratum.

Description

Technical Field

[0001] This invention relates to the field of intelligent tunneling control technology in mining, specifically to a multi-objective optimization decision-making system for tunneling parameters of a mining tunneling machine. Background Technology

[0002] During the excavation of coal mine roadways, metal mine roadways, and complex coal-rock roadways, tunneling machines typically need to adjust excavation parameters such as cutter head rotation speed, advance speed, cutter arm swing speed, and spray volume based on the hardness of the coal and rock at the excavation face, water content, gas content, and equipment load. Existing methods for adjusting these parameters largely rely on operator experience, historical operating data, or single sensor thresholds. For example, they determine the load based on the cutter motor current, whether to slow down or stop the machine based on gas concentration, and whether the cutting is abnormal based on vibration amplitude. While these methods can handle conventional homogeneous coal-rock conditions, their ability to identify and make decisions regarding complex, short-scale, and concealed anomalous strata is insufficient.

[0003] Especially during deep mine excavation, thin layers of hard rock, cemented fissures, and sealed gas pockets may coexist in front of the target excavation face. These anomalous structures are small in spatial scale and may only be distributed in local areas of the excavation section, making it difficult for conventional advance detection or averaged geological parameters to reflect their true impact in a timely manner. When the cutting head enters this area, the thin layers of hard rock will cause periodic ripples and phase lag in the cutting motor current, the cemented fissures will cause abnormal temperature rise of the cutting teeth and increased cutting resistance, and the sealed gas pockets may produce delayed gas release after the cutting disturbance. The superposition of these multiple anomalous factors can easily lead to accelerated wear of the cutting teeth, increased machine sway, increased unit energy consumption, and an increased risk of instantaneous gas release. In severe cases, it can cause the excavator to jam, roof collapse, or shutdown due to excessive gas levels.

[0004] Existing multi-objective optimization methods typically prioritize parameters such as advance efficiency, energy consumption, or equipment load, lacking analysis of the temporal correlations between current, vibration, temperature, and gas response during the cutting cycle. They also fail to map local anomalies to different zones of the tunnel face. Therefore, existing methods struggle to distinguish between the combined disturbances generated by "high load on homogeneous hard rock" and "the combined effects of thin layers of hard rock, muddy fractures, and enclosed gas pockets," leading to optimized tunneling parameters that tend to favor efficiency targets and neglect the risks of delayed gas release and machine sway.

[0005] Therefore, there is an urgent need to provide a multi-objective optimization decision-making system for tunneling parameters of mining machines that can identify the characteristics of complex abnormal strata disturbances based on multi-source operating data within the trial cutting stroke, and make joint decisions based on factors such as advance efficiency, unit energy consumption, cutter wear, machine body sway, and the risk of instantaneous gas release. Summary of the Invention

[0006] The purpose of this invention is to provide a multi-objective optimization decision system for tunneling parameters of mining tunneling machines to address the shortcomings of the prior art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a multi-objective optimization decision-making system for tunneling parameters of a mining tunneling machine, comprising:

[0008] The data acquisition module is used to acquire the cutting motor current, cutting arm vibration, cutting tooth temperature, gas concentration, propulsion pressure and cutting head angle position of the tunneling machine within the preset test cutting stroke of the target tunneling face. The acquired data is divided into several operating data groups D according to the cutting head angle position.

[0009] The disturbance feature calculation module is used to calculate the current ripple phase lag value, vibration frequency energy ratio, cutting tooth temperature rise residual value and gas delayed release amount corresponding to each cutting cycle segment based on the running data group D, and generate a disturbance feature set F for characterizing the combined effect of thin-layer hard gangue, cemented cracks and closed gas sacs.

[0010] The composite formation response field construction module is used to map the disturbance feature set F to the partitioned grid of the tunnel section, obtain the equivalent cuttable coefficient, mud adhesion coefficient and depressurization and gas release sensitivity coefficient corresponding to each partitioned grid, and establish the composite formation response field G.

[0011] The tunneling parameter simulation and evaluation module is used to simulate different combinations of cutting head rotation speed, advance speed, cutting arm swing speed and spray volume based on the composite stratum response field G, and obtain the corresponding advance efficiency, unit energy consumption, cutting tooth wear, machine body sway and instantaneous gas release risk value for each combination.

[0012] The constraint elimination and weight correction module is used to eliminate parameter combinations that exceed the corresponding threshold for fuselage sway or instantaneous gas release risk value, and dynamically correct the unit energy consumption weight according to the cutter wear and mud adhesion coefficient to obtain the candidate parameter set P.

[0013] The multi-objective decision output module is used to select the parameter combination from the candidate parameter set P that satisfies the requirement that the advance efficiency is not lower than the preset efficiency lower limit and the multi-objective evaluation value is the smallest, as the target tunneling parameter output.

[0014] Preferably, the generation of the perturbation feature set F includes the following steps:

[0015] According to the cutting head angle position, the cutting motor current, cutting arm vibration, cutting tooth temperature and gas concentration and propulsion pressure in the operating data group D are aligned in the angle domain to obtain the multi-source response sequence under the same angle position;

[0016] Using the sudden change point of the propulsion pressure as the reference point of the load excitation, the current ripple phase lag value is extracted, and the vibration harmonic energy ratio is extracted based on the fundamental frequency of the cutting head rotation.

[0017] The residual value of the cutting tooth temperature rise is extracted based on the theoretical temperature rise value corresponding to the current and propulsion pressure, and the gas delay release is statistically calculated starting from the disturbance initiation angle position. The four are combined into a disturbance feature set F.

[0018] Preferably, the establishment of the composite formation response field G includes the following steps:

[0019] Based on the starting angle position, ending angle position, and cumulative advance displacement of the cutting cycle segment, the feature vectors in the disturbance feature set F are mapped to the partition grid of the tunnel section.

[0020] When the tooth tip sweep area of ​​the truncated periodic segment spans at least two partition grids, the feature vector is weighted and allocated according to the proportion of the tooth tip sweep area.

[0021] Based on the current ripple phase lag value, vibration frequency energy ratio, cutter temperature rise residual value, gas delay release amount and propulsion pressure in each partition grid, the three-dimensional response vector Gij=(Kij, Mij, Rij) is obtained.

[0022] Preferably, obtaining the ternary response vector Gij includes the following steps:

[0023] Based on the dimensionless current ripple phase lag value, vibration frequency doubling energy ratio and average propulsion pressure, the equivalent machinability coefficient Kij is generated.

[0024] Based on the dimensionless cumulative value of the residual temperature rise of the cutting tooth, the growth rate of the residual temperature rise of the cutting tooth, and the amplitude of the propulsion pressure fluctuation, the mud adhesion coefficient Mij is generated.

[0025] Based on the dimensionless gas delay release, current ripple phase lag, and gas release boundary enhancement of adjacent grid zones, the pressure relief gas release sensitivity coefficient Rij is generated.

[0026] Preferably, the simulation evaluation of tunneling parameters includes the following steps:

[0027] The composite stratum response field G is traversed in the order of the cut-off head position, and the equivalent cut-off load in each partition grid is calculated based on the equivalent cuttable coefficient Kij and the mud adhesion coefficient Mij.

[0028] The advance distance is reduced based on the equivalent machinability coefficient Kij to obtain the advance efficiency corresponding to each parameter combination, and the unit energy consumption is obtained based on the equivalent cutting load, advance speed and spray volume.

[0029] Based on the equivalent cutting load, mud adhesion coefficient Mij, depressurization and gas release sensitivity coefficient Rij, and the cutting head rotation speed, propulsion speed, cutting arm swing speed, and spray volume in each parameter combination, the cutting tooth wear, machine body sway, and instantaneous gas release risk values ​​are obtained.

[0030] Preferably, obtaining the candidate parameter set P includes the following steps:

[0031] Parameter combinations that exceed the fuselage yaw threshold or the instantaneous gas release risk value exceed the instantaneous gas release risk threshold will be eliminated.

[0032] Identify continuous mud-adhesion sections where the mud-adhesion coefficient Mij is continuously higher than the mud-adhesion discrimination value along the actual sweeping sequence of the cutting head, and compare the wear of the cutting teeth in the section with the adjacent non-mud-adhesion sections.

[0033] When the wear of the cutting teeth in the continuous mud-adhesion section exceeds the upper limit of normal formation wear fluctuation, the unit energy consumption weight of the corresponding parameter combination is increased, and a candidate parameter set P is formed based on the corrected multi-objective evaluation values.

[0034] Preferably, the output of the target tunneling parameters includes the following steps:

[0035] The lower limit of the cutting efficiency is determined based on the planned cutting depth of the target roadway, the cutting efficiency of normal strata, and the equivalent cuttable coefficient Kij in the response field of composite strata.

[0036] Select parameter combinations from the candidate parameter set P that have a cutting efficiency not lower than the lower limit of the cutting efficiency, and select the parameter combination with the smallest multi-objective evaluation value as the preferred parameter combination.

[0037] When the multi-objective evaluation values ​​of at least two parameter combinations are the same, they are sorted in order according to the instantaneous gas release risk value, machine body sway, cutting tooth wear, unit energy consumption and advance efficiency. The cutter head speed, advance speed, cutter arm swing speed and spray volume with the best ranking are output as target tunneling parameters.

[0038] Preferably, when there is no parameter combination in the candidate parameter set P with a cutting efficiency not lower than the lower limit of the cutting efficiency, the output of the target tunneling parameters further includes the following steps:

[0039] The parameter combination with the smallest instantaneous gas release risk value in the candidate parameter set P is selected as the benchmark parameter combination.

[0040] The corrected parameter combination is obtained by reducing the propulsion speed and the cutting arm swing speed in the reference parameter combination by 1 adjustable step and increasing the spray volume by 1 adjustable step.

[0041] When the combination of corrected parameters satisfies the constraints of fuselage sway and instantaneous gas release risk value, it is output as the target tunneling parameter;

[0042] If the constraints are not met even after reaching the minimum advance speed, minimum cutter arm swing speed, and maximum spray volume, the target tunneling parameters will not be output.

[0043] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0044] This invention differs from existing technologies that adjust parameters solely based on cutting motor current, gas concentration, or driver experience. Instead, it focuses on the actual cutting process within a preset trial cutting stroke, aligning current, vibration, temperature, gas concentration, and propulsion pressure according to the cutting head's angular position. Furthermore, it extracts current ripple phase lag, vibration frequency harmonic energy ratio, cutting tooth temperature rise residual, and delayed gas release, thus forming a disturbance feature set. This invention does not simply collect sensor data; instead, it establishes the load hysteresis, frequency harmonic impact, continuous frictional heating, and delayed gas release response when the cutting head cuts into local coal and rock as correlated features within the same cutting cycle. This allows for direct identification of complex anomalies caused by the combined effects of thin-layer hard rock, cemented fissures, and enclosed gas pockets. This solves the problem of existing technologies easily averaging local anomalies and struggling to distinguish between high loads in ordinary hard rock and small-scale complex disturbances. It enables subsequent parameter decisions to be adjusted based on the source of the anomaly, rather than passively slowing down and shutting down only after the anomaly has caused overload or increased gas levels.

[0045] Furthermore, unlike existing technologies that treat the tunnel face as a single, uniform working condition for optimization, this invention maps the disturbance feature set F to a partitioned grid of the tunnel section, establishing a composite formation response field G that includes the equivalent cutability coefficient, mud adhesion coefficient, and depressurization and gas release sensitivity coefficient. Based on this response field, it simulates the effects of different combinations of cutter head rotation speed, advance speed, cutter arm swing speed, and spray volume on advance efficiency, unit energy consumption, cutter tooth wear, machine sway, and instantaneous gas release risk value grid by grid. This invention allows parameter optimization to move beyond simply pursuing optimal efficiency or energy consumption; instead, it incorporates the ineffective energy consumption caused by localized difficult cutting, mud adhesion, machine sway caused by unilateral load, and the instantaneous gas release risk of enclosed gas chambers into the decision-making process simultaneously. Furthermore, by eliminating parameter combinations that exceed the sway and gas release thresholds, and dynamically correcting the unit energy consumption weight based on the wear of the cutting teeth and the mud adhesion coefficient, the probability of high-wear and high-ineffective cutting parameters being mistakenly selected can be reduced, so that the output target tunneling parameters have both executability, safety margin and continuous tunneling stability. Attached Figure Description

[0046] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0047] Figure 1 This is a flowchart of a multi-objective optimization decision-making system module for tunneling parameters of a mining tunneling machine according to the present invention.

[0048] Figure 2 This is a flowchart of the tunneling parameter simulation evaluation method of the present invention.

[0049] Figure 3 This is a flowchart of the method for outputting the target tunneling parameters of the present invention. Detailed Implementation

[0050] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0051] Example 1, please refer to Figure 1 and Figure 2 and Figure 3 As shown in this embodiment, a multi-objective optimization decision-making system for tunneling parameters of a mining tunneling machine includes:

[0052] The data acquisition module is used to acquire the cutting motor current, cutting arm vibration, cutting tooth temperature, gas concentration, propulsion pressure and cutting head position of the tunneling machine within the preset test cutting stroke of the target tunneling face. The acquired data is divided into several operating data groups D according to the cutting head position.

[0053] In one embodiment of the present invention, the data acquisition module is installed in the control system of the mining tunneling machine and is electrically connected to the cutting motor current sensor, the cutting arm vibration sensor, the cutting tooth temperature sensor, the gas concentration sensor, the propulsion cylinder pressure sensor, and the cutting head angle position encoder, respectively. It is used to synchronously acquire tunneling machine operating data within a preset test cutting stroke at the target tunneling face. The preset test cutting stroke can be set according to the tunnel cross-sectional dimensions and the scale of geological fluctuations, preferably 0.5m to 3m, to ensure that the acquired data can reflect the short-term disturbance characteristics before and after entering the target tunneling face.

[0054] Specifically, the data acquisition module triggers an acquisition command when the tunneling machine enters the preset test cutting stroke starting point, continuously acquiring the cutting motor current signal, cutting arm vibration signal, cutting tooth temperature signal, gas concentration signal, propulsion pressure signal, and cutting head angle position signal at a fixed sampling frequency. Among these, the cutting motor current characterizes the cutting load change, the cutting arm vibration characterizes the cutting impact and off-center load state, the cutting tooth temperature characterizes the degree of frictional heating, the gas concentration characterizes the gas response after cutting disturbance, the propulsion pressure characterizes the propulsion resistance change, and the cutting head angle position characterizes the current angle segment of the cutting trajectory. To ensure the timing consistency between different signals, the data acquisition module assigns a unified timestamp to each sensor signal and aligns them according to a unified clock.

[0055] Furthermore, the data acquisition module divides the aligned multi-source data into several operational data groups D based on the cutting head completing a single angle cycle or reaching a preset angle interval. Each operational data group D includes the cutting motor current sequence, cutting arm vibration sequence, cutting tooth temperature sequence, gas concentration sequence, and propulsion pressure sequence within the corresponding angle position interval, so that each operational data group D corresponds to an actual cutting area in the tunnel section. Preferably, an operational data group D is generated when the cutting head angle position crosses a preset angle step, wherein the preset angle step can be set to 5° to 30°.

[0056] The disturbance feature calculation module is used to calculate the current ripple phase lag value, vibration frequency energy ratio, cutting tooth temperature rise residual value and gas delayed release amount corresponding to each cutting cycle segment based on the running data group D, and generate a disturbance feature set F to characterize the combined effect of thin-layer hard gangue, cemented cracks and closed gas sacs.

[0057] Specifically, each set of operational data D corresponds to a continuous angular domain interval at the cutting head angle position. Within the same set of operational data D, using the cutting head angle position as a unified benchmark, the cutting motor current, cutting arm vibration, cutting tooth temperature, gas concentration, and propulsion pressure are aligned on the time axis and angle axis to ensure that each signal has a one-to-one correspondence at the same angular position.

[0058] It should be noted that the reason for using the cutting head corner position as the alignment benchmark is that the cutting head's entry, exit, and re-entry into the coal and rock all occur repeatedly with the corner position. The load characteristics at the same corner position are comparable, which can reduce the time scale error caused by fluctuations in propulsion speed and small fluctuations in rotation speed, thereby improving the stability of subsequent feature extraction.

[0059] The acquisition of the current ripple phase lag value includes: firstly, extracting the segmented motor current sequence and propulsion pressure sequence from the operating data group D, and resampling by angular position to obtain the angular domain current sequence and angular domain propulsion pressure sequence. Then, the angular domain current sequence undergoes DC component removal and local baseline subtraction to retain the current ripple component reflecting periodic load changes. Further, within each segmented period, the angular position corresponding to the current ripple peak value is identified, and simultaneously, within the same segmented period, the local maximum point of the absolute value of the first-order difference of the propulsion pressure is identified as the load excitation reference point. The difference between the angular position of the current ripple peak value and the angular position of the load excitation reference point is recorded as the angular domain phase difference, and then combined with the instantaneous speed of the segmented period to convert it into a time quantity, thus obtaining the current ripple phase lag value.

[0060] It should be noted that the abrupt change point of the propulsion pressure is used as a reference because the response of the propulsion pressure to the change in the cutting resistance of coal and rock precedes the full establishment of the motor current ripple in the local abnormal coal and rock. Especially in the thin layer of hard rock interlayer, after the cutting tooth contacts the hard interlayer, it will first show the propulsion being blocked, and then the current ripple peak position will shift backward. Therefore, this phase lag value can more sensitively distinguish between uniform hard rock loading and local thin layer hard rock embedding loading.

[0061] To obtain the vibration harmonic energy ratio, the vibration sequence of the cutting arm corresponding to the operating data group D is first extracted, and the reference cutting frequency of the cutting cycle segment is determined based on the rate of change of the cutting head angular position. Then, the vibration sequence is synchronously expanded in the angular domain, arranging the vibration response at equal angles according to the cutting head rotation period. Next, frequency band decomposition is performed on each cutting cycle segment to obtain the fundamental frequency energy, second harmonic energy, and third harmonic energy. The second harmonic energy and the third harmonic energy are added together and then divided by the fundamental frequency energy to obtain the vibration harmonic energy ratio.

[0062] The reason for using the superimposed energy of the second and third harmonics instead of the total vibration amplitude alone is that the impact of thin-layer hard rock boundaries and localized off-center loading is more likely to form energy accumulation in the harmonic band, while uniform coal and rock cutting is mainly dominated by the fundamental frequency, and the two have significant differences in spectral distribution. The larger the vibration harmonic energy ratio, the more obvious the non-uniform impact within the cutting cycle segment, and the more sensitive it is to thin-layer hard rock boundaries and localized off-center loading.

[0063] To obtain the residual value of the cutting tooth temperature rise, the cutting tooth temperature sequence, cutting motor current sequence, and feed pressure sequence are first extracted from the operating data group D. A heat load response relationship is then established within the same cutting cycle segment. This relationship uses the cutting motor current to reflect the cutting work and the feed pressure to reflect the degree of contact compression, with historical test cutting data under normal moisture conditions serving as the calibration basis. Specifically, samples formed by the same equipment in areas without obvious cementitious cracks are selected to establish a corresponding curve between "current-feed pressure-theoretical temperature rise." The actual current and feed pressure of the current cutting cycle segment are then substituted into this corresponding curve to obtain the theoretical temperature rise value. Subtracting the theoretical temperature rise value from the actual measured temperature rise value yields the residual value of the cutting tooth temperature rise.

[0064] If the residual value remains positive and increases in magnitude, it indicates that there is additional frictional heat generation under the same work and extrusion conditions, generally corresponding to poor adhesion, encapsulation, and slag removal caused by cementitious cracks. Using the residual form instead of directly using the absolute value of temperature rise can eliminate the influence of natural temperature rise variations under different load conditions, making the residual value of the cutting tooth temperature rise more suitable for characterizing abnormal adhesion friction.

[0065] The acquisition of the delayed gas release includes: first, extracting the gas concentration sequence from the operational data set D, and then determining the disturbance initiation angle position of the cut-off period segment by combining the current ripple phase lag value and the vibration frequency harmonic energy ratio. Next, using the time corresponding to the disturbance initiation angle position as the starting point, the gas concentration increment is statistically analyzed along a subsequent preset observation window to form a gas response curve.

[0066] In this embodiment, the preset observation window can be one to three cutting cycles, preferably two cutting cycles. The reason for choosing two cutting cycles is that if the observation window is less than one cutting cycle, the delayed release of the gas chamber after disturbance has not yet fully manifested; if the observation window is greater than three cutting cycles, subsequent propulsion and ventilation disturbances will be superimposed, reducing the correspondence to the current cutting cycle segment. The integral area within the observation window where the gas concentration is higher than the initial reference concentration is defined as the delayed gas release. Using the integral area instead of a single-point peak value can simultaneously reflect the release intensity and duration, making it more suitable for identifying the delayed release phenomenon of the gas chamber after cutting disturbance.

[0067] After obtaining the current ripple phase lag value, vibration frequency energy ratio, cutting tooth temperature rise residual value, and gas delayed release amount, the four are combined into a feature vector according to the order of the cutting cycle segments, and all feature vectors belonging to the same operating data group D are summarized to generate the disturbance feature set F.

[0068] It should be noted that, to ensure comparability between different dimensions, the four types of features were dimensionless. Dimensionlessness can be achieved using interval normalization based on historical test samples, where the upper and lower boundaries of each feature are determined by the statistical quantile values ​​of the same type of equipment in normal and abnormal formations. Using statistical quantile values ​​instead of extreme values ​​is to reduce the stretching effect of individual abnormal noise samples on the feature scale. In the dimensionless disturbance feature set F, the current ripple phase lag value mainly characterizes the hysteretic loading characteristics caused by thin-layer hard rock; the vibration overtone energy ratio mainly characterizes the non-uniform impact characteristics caused by the thin-layer hard rock boundary and local off-center loading; the cutting tooth temperature rise residual value mainly characterizes the continuous adhesive friction characteristics caused by cemented fractures; and the delayed gas release mainly characterizes the hysteretic gas release characteristics formed after the closed gas chamber is disturbed.

[0069] The composite formation response field construction module is used to map the disturbance feature set F to the partitioned grid of the tunnel section, obtain the equivalent cuttable coefficient, mud adhesion coefficient and depressurization and gas release sensitivity coefficient corresponding to each partitioned grid, and establish the composite formation response field G.

[0070] Specifically, the tunneling section is divided into several angular partitions along the direction of the cutting head's angular position, and the angular step size of the angular partitions is consistent with the angular position cutting step size of the running data group D; the preset test cutting stroke is divided into several radial partitions along the advancing direction, and the width of the radial partitions is taken as the average advancing displacement of the cutting head within one cutting cycle.

[0071] For any eigenvector in the disturbance feature set F, determine the starting angle position, ending angle position, and cumulative advance displacement at the end of the corresponding cutting cycle segment. Determine its angular partition based on the starting and ending angle positions, and its radial partition based on the cumulative advance displacement. If the tooth tip sweep area of ​​the cutting cycle segment falls into only one partition grid, then the entire eigenvector is assigned to that partition grid. If the tooth tip sweep area spans more than two partition grids, then a weighted allocation is performed according to the proportion of the tooth tip sweep area in each partition grid. Thus, each partition grid yields corresponding current ripple phase hysteresis, vibration frequency harmonic energy ratio, cutting tooth temperature rise residual, gas delay release, and advance pressure data.

[0072] Before calculating the coefficients, statistical analysis is performed on similar data falling within the same grid partition to obtain the mean current ripple phase lag (A1), mean vibration frequency harmonic energy ratio (A2), cumulative residual value of cutting tooth temperature rise (A3), delayed gas release (A4), and mean propulsion pressure (A5) for that grid partition. If multiple cutting cycle segments exist within the same grid partition, the differences between adjacent cutting cycle segments are calculated in chronological order to obtain the gradient of each feature. The gradient is used to preserve the abrupt change information corresponding to the thin-layer hard rock boundary, cemented fracture boundary, and closed gas sac boundary.

[0073] Dimensionless processing was performed on A1, A2, A3, A4, and A5. The lower boundary of the dimensionless processing was taken as the 5th percentile of the test cutting samples in normal strata, and the upper boundary was taken as the 95th percentile of the test cutting samples in abnormal strata. When test cutting samples in abnormal strata were unavailable, the upper boundary was taken as the mean of the test cutting samples in normal strata plus three times the standard deviation. The above boundaries were derived from test cutting data obtained by the same tunneling machine under the same cutter head structure, the same sensor arrangement, and similar roadway cross-section conditions.

[0074] It should be noted that the 5th and 95th percentile values ​​are used to eliminate the influence of individual noise points or instantaneous sensor drift on the normalization scale; the mean plus three times the standard deviation is used as a substitute upper boundary to ensure that an feasible feature scale can still be determined based on the normal fluctuation range when outlier samples are lacking. Dimensionless values ​​are all limited to between 0 and 1.

[0075] The equivalent machinability coefficient is calculated based on the dimensionless current ripple phase lag value B1, the vibration frequency harmonic energy ratio B2, and the average propulsion pressure B5. First, the local cutting resistance index H is calculated according to H=[(1+B1)(1+B2)(1+B5)-1] / 7, where B1 reflects the load build-up lag caused by the thin layer of hard rock, B2 reflects the non-uniform impact at the hard-soft interface, and B5 reflects the actual propulsion resistance. Then, the equivalent machinability coefficient K is obtained according to K=1-H. Since B1, B2, and B5 are all between 0 and 1, H and K are also between 0 and 1. The closer K is to 0, the stronger the cutting resistance and the worse the cutting continuity of the grid section; the closer K is to 1, the easier the grid section is to cut.

[0076] The mud adhesion coefficient is calculated based on the dimensionless cumulative residual value of the cutter temperature rise B3, the growth rate of the residual temperature rise C3, and the amplitude of the advance pressure fluctuation C5. The growth rate of the residual temperature rise C3 is obtained by dividing the difference in the cumulative residual temperature rise of the cutter temperature rise of adjacent cutting cycle segments within the same grid by the corresponding time interval; the amplitude of the advance pressure fluctuation C5 is obtained by dimensionlessizing the difference between the maximum and minimum advance pressure values ​​within the same grid. If a grid segment satisfies the condition that the increase in the cumulative residual temperature rise of the cutter temperature rise is greater than the standard deviation of the residual temperature rise of normal formation test samples in two consecutive cutting cycle segments, then the grid segment is considered to have a continuous heating characteristic. Subsequently, the mud adhesion coefficient M is calculated according to M = 0.5B3 + 0.3C3 + 0.2C5. The larger M is, the more pronounced the adhesion of the cutter temperature rise caused by mud-bearing fractures, the obstruction of slag discharge, and the continuous friction.

[0077] The pressure relief and gas release sensitivity coefficient is calculated based on the dimensionless delayed gas release amount B4, the current ripple phase lag value B1, and the gas release boundary enhancement amount C4. The delayed gas release amount is retrospectively assigned to the corresponding grid cell according to the disturbance initiation angle position, not according to the time of occurrence of the gas concentration peak; this processing method can avoid mismatch of gas release position due to gas diffusion lag. The gas release boundary enhancement amount C4 is obtained by comparing the difference in delayed gas release amounts between adjacent angular and radial zones. Its judgment value is the larger of the 90th percentile value of the difference between all adjacent zones in the current tunneling section and the mean value of the difference of normal stratum test cut samples plus twice the standard deviation.

[0078] In this embodiment, the reason for using this judgment value is that the 90th percentile value is used to screen out a few abrupt boundary changes in the current cross section, and the mean plus twice the standard deviation is used to constrain the normal fluctuation range. Taking the larger of the two values ​​can reduce the probability that small fluctuations in the homogeneous layer are misjudged as gas release boundaries.

[0079] The pressure relief and gas release sensitivity coefficient R is then calculated using the formula R = 0.6B4 + 0.25B1 + 0.15C4. B4 directly characterizes the release scale after the sealed gas chamber is disturbed, B1 reflects the pressure relief hysteresis caused by local structural density, and C4 reflects the intensity of the local abrupt change at the gas release boundary. A larger R indicates a higher probability of instantaneous gas release from the corresponding grid section under subsequent truncation perturbations.

[0080] After obtaining the equivalent cutability coefficient K, mud adhesion coefficient M, and depressurization and gas release sensitivity coefficient R for each zone grid, the three are combined into a ternary response vector Gij = (Kij, Mij, Rij) according to the same zone grid index, where i represents the angular zone number and j represents the radial zone number. The ternary response vectors of all zone grids together constitute the composite formation response field G. For zone grids that do not fall into the eigenvector, distance-weighted padding is performed using adjacent zone grids that have eigenvectors; if adjacent zone grids also have no eigenvectors, the ternary response vector corresponding to the normal formation test cut sample is taken as the initial value of that zone grid.

[0081] The tunneling parameter simulation and evaluation module is used to simulate different combinations of cutting head rotation speed, advance speed, cutting arm swing speed and spray volume based on the composite stratum response field G, and obtain the corresponding advance efficiency, unit energy consumption, cutting tooth wear, machine body sway and instantaneous gas release risk value for each combination.

[0082] In one embodiment of the present invention, after establishing the composite formation response field G, simulation calculations are performed for different combinations of cutter head rotation speed, advance speed, cutter arm swing speed, and spray volume. Each grid cell in the composite formation response field G is represented as Gij = (Kij, Mij, Rij), where Kij is the equivalent abrasive coefficient, Mij is the mud adhesion coefficient, Rij is the depressurization and gas release sensitivity coefficient, i represents the angular partition number, and j represents the radial partition number.

[0083] First, determine the combination of tunneling parameters to be evaluated. Each parameter combination is represented as X = (n, v, s, q), where n is the cutting head rotation speed, v is the propulsion speed, s is the cutting arm swing speed, and q is the spray volume. The value ranges of n, v, s, and q are jointly determined by the rated range on the tunneling machine nameplate, the allowable operating range of the hydraulic actuators, and the allowable range of the mine safety regulations; the value step size is determined according to the minimum stable adjustment amount of the corresponding actuator. For example, the step size of the cutting head rotation speed is the minimum rotation speed change that the frequency converter can stably maintain, and the step size of the propulsion speed is the minimum resolvable speed change of the propulsion cylinder when propulsing at a low speed. This setting ensures that the parameter combination obtained from the simulation can be actually executed, avoiding theoretically optimal but unadjustable results in the field.

[0084] For any parameter combination X, the equivalent cut load Lij in each grid of the composite formation response field G is calculated by traversing the grid in the order of the cut head position. The equivalent cut load Lij is calculated according to the following formula: Where L0 is the normal formation calibration load, which is determined by the average load of the cutting motor during trial cutting in normal formation by the same tunneling machine; v0 is the normal formation calibration advance speed; and n0 is the normal formation calibration cutting head speed. 1 / Kij is used to reflect the cutting difficulty in local areas, and 1+Mij is used to reflect the additional resistance caused by mud adhesion.

[0085] In this embodiment, the advance efficiency E1 corresponding to the parameter combination X is calculated. The advance efficiency E1 is calculated as the ratio of the effective advance distance to the simulation time within a preset simulation time. The effective advance distance is not directly equal to v multiplied by time, but is reduced according to the equivalent machinability coefficient Kij of each partition grid. The calculation formula is: E1=Σ(v·Δtij·Kij) / T, where Δtij is the time required for the cutting head to sweep through the i-th and j-th partition grids, and T is the complete simulation time. After this processing, the difficult-to-cut area will not be incorrectly calculated as having the same advance contribution as the easy-to-cut area.

[0086] Further calculate the unit energy consumption E2. Unit energy consumption E2 is the sum of the cutting motor energy consumption, propulsion hydraulic energy consumption, and spray energy consumption during the simulation time, divided by the effective propulsion distance. The calculation expression for cutting motor energy consumption is: EC=Σ[2π·n·Tij·Δtij / 60], where n is the cutting head rotation speed, propulsion hydraulic energy consumption is calculated by accumulating the product of the equivalent propulsion pressure and the propulsion speed v, and spray energy consumption is calculated by accumulating the pump power corresponding to the spray volume q. Using unit energy consumption instead of total energy consumption is to eliminate comparison deviations caused by different actual advance distances due to different parameter combinations.

[0087] Further calculate the wear amount W of the cutting teeth. The wear amount W of the cutting teeth is determined by the equivalent cutting load Lij, the mud adhesion coefficient Mij, and the cutting head rotation speed n. In the calculation, the wear amount W of the cutting teeth in each grid is expressed as: W=Σ[a·Lij·Δtij+b·Mij·n·Δtij], where a is the load wear calibration coefficient and b is the adhesive friction wear calibration coefficient.

[0088] It should be noted that coefficients a and b are determined according to the measurement method of the cutting tooth wear W. When W is expressed as the cutting edge wear width, the unit of a is mm / (kN·min), and the unit of b is mm / r; when W is expressed as mass loss, the unit of a is g / (kN·min), and the unit of b is g / r. The selection of a is based on the cutting tooth wear records in normal water-bearing strata without obvious mud adhesion, and is used to characterize the basic mechanical wear generated under the continuous action of the equivalent cutting load Lij; the selection of b is based on the cutting tooth adhesion, slag removal obstruction and continuous friction wear records in cement-containing fractured strata, and is used to characterize the additional wear caused by the mud adhesion coefficient Mij and the number of cutting head rotations. Specifically, historical test cut records under the same type of cutting tooth, the same cutting head structure, and the same wear detection method are selected. The measured tooth wear W0, the equivalent cumulative cutting load Σ(Lij·Δtij), and the cumulative adhesion rotation Σ(Mij·n·Δtij) for each test cut are statistically analyzed. Non-negative least squares fitting is then performed according to W0=aΣ(Lij·Δtij)+bΣ(Mij·n·Δtij) to obtain a and b. If samples containing cemented fractures are lacking in the initial stage, a is first fitted with b=0 in normal formation samples. After obtaining samples with clay adhesion, a is fixed and b is fitted. Through this method, a corresponds to the basic wear caused by high-load cutting, and b corresponds to the additional wear caused by clay adhesion. This allows the tooth wear W to distinguish between "high-load wear in hard rock" and "clay adhesion friction wear," thereby improving the accuracy of tooth life prediction in subsequent parameter combination evaluations.

[0089] Further calculations are performed on the machine body yaw rate Y. The yaw rate Y is calculated based on the equivalent cutting load difference between adjacent left-right or top-bottom grid sections. For the same angular section, the Lij difference between its adjacent radial sections and symmetrical angular sections is statistically analyzed, and the difference is multiplied by the stress arm and accumulated to obtain the yaw moment; then, it is converted into the yaw rate based on the tunneling machine's support width and overall mass. This calculation reflects the machine body attitude deviation caused by a sudden increase in load on one side when the cutting head enters a local thin layer of hard rock or a hard-soft interface area.

[0090] Finally, the instantaneous gas release risk value Rgas is calculated. The instantaneous gas release risk value is determined based on the depressurization and gas release sensitivity coefficient Rij, the propulsion speed v, the cutting arm swing speed s, and the spray volume q, and the calculation formula is: Where vmax is the maximum allowable stable advance speed of the current tunneling machine, smax is the maximum allowable stable swing speed of the current cutting arm, and qmax is the maximum stable spray volume of the spray device. qmax is determined by the rated pressure of the spray pump and the nozzle flow rate curve. v and s are introduced because the greater the advance speed and the swing speed of the cutting arm, the greater the volume of disturbed coal and rock per unit time; 1-q / qmax is introduced because increasing the spray volume can reduce the heat generation during cutting and the degree of coal body crushing and pulverization, thereby reducing the evaluation value of instantaneous release risk.

[0091] In this embodiment, through the above calculations, each parameter combination X yields the corresponding advance efficiency E1, unit energy consumption E2, cutter wear W, fuselage sway Y, and instantaneous gas release risk value Rgas. These results correspond grid-by-grid to Kij, Mij, and Rij in the composite formation response field G, reflecting the combined impact of different local anomaly regions on tunneling efficiency, energy consumption, wear, stability, and gas safety. This provides a repeatable data foundation for subsequent parameter combination elimination, weight correction, and multi-objective evaluation.

[0092] The constraint elimination and weight correction module is used to eliminate parameter combinations that exceed the corresponding threshold for fuselage sway or instantaneous gas release risk value, and dynamically corrects the unit energy consumption weight based on the wear of the cutting teeth and the mud adhesion coefficient to obtain the candidate parameter set P.

[0093] Specifically, for any parameter combination X = (n, v, s, q), the corresponding machine sway Y and instantaneous gas release risk value Rgas are extracted. The machine sway threshold Ymax is determined jointly based on the equipment safety limit and the current test-cut stability level. Specifically: first, the maximum permissible lateral attitude deviation in the tunneling machine's instruction manual or factory test record is obtained as the equipment safety limit; then, the mean and standard deviation of the machine sway are calculated from the test-cut data in normal strata, and the mean plus twice the standard deviation is used as the current reference value for stable roadway operation; finally, the smaller value between the equipment safety limit and the stable operation reference value is taken as Ymax. This determination method can avoid the risk identification lag caused by simply using the equipment limit, and can also avoid misjudging normal test-cut fluctuations as instability.

[0094] The instantaneous gas release risk threshold Rmax is determined based on the gas alarm margin and ventilation dilution capacity. Specifically: First, the gas alarm concentration in the target roadway and the background gas concentration before the test cut are obtained, and the difference between the two is taken as the allowable concentration increment; then, the dilution capacity per unit time is determined based on the measured air volume and roadway cross-sectional area in the test cut area; subsequently, using test cut records of normal formations and known gas-releasing formations, a one-to-one correspondence is established between the instantaneous gas release risk value Rgas and the measured gas concentration increment, and the Rgas value corresponding to the allowable concentration increment is taken as the alarm margin boundary; at the same time, the mean value of Rgas in the normal formation test cut is calculated plus twice the standard deviation, which is taken as the background risk upper limit; finally, the smaller value between the alarm margin boundary and the background risk upper limit is taken as Rmax. In this way, Rmax is not directly replaced by the gas alarm concentration, but is converted into a threshold with the same evaluation scale as Rgas, thereby ensuring that the elimination judgment can be repeated.

[0095] Then, a judgment is made on all parameter combinations:

[0096] When the fuselage yaw rate Y corresponding to the parameter combination is greater than Ymax, or the instantaneous gas release risk value Rgas is greater than Rmax, the parameter combination shall be removed.

[0097] When Y is not greater than Ymax and Rgas is not greater than Rmax, this parameter combination is retained.

[0098] If all parameter combinations are eliminated, the propulsion speed v and the cutting arm swing speed s will be reduced by 1 executable adjustment step, the spray volume q will be increased by 1 executable adjustment step, and the simulation calculation and safety constraint judgment will be performed again.

[0099] If no parameter combination meets the safety constraints after reaching the minimum advance speed, minimum cutter arm swing speed, and maximum spray volume, then the target tunneling parameters will not be output.

[0100] This approach avoids providing tunneling parameters when the risks of gas release or hull sway are unacceptable.

[0101] After completing the safety constraint removal, the mud adhesion coefficient Mij in the composite formation response field G is read along the actual sweeping sequence of the cutting head, and the continuous mud adhesion section is identified.

[0102] The mud adhesion discrimination value is determined based on the test cut samples of normal strata, taking the mean mud adhesion coefficient of normal strata plus 2 times the standard deviation; when samples containing cemented fractures have been obtained, and the 25th percentile value of the mud adhesion coefficient in the samples containing cemented fractures is higher than the discrimination value, the middle value of the two is taken as the new mud adhesion discrimination value.

[0103] It should be noted that this value method is adopted in order to identify abnormal samples based on the upper limit of normal fluctuations when there are no abnormal samples, and to use weak muddy samples to correct the discrimination boundary when there are abnormal samples, thereby reducing the probability of missing weak muddy adhesion sections.

[0104] When at least two consecutive partitioned grids appear along the sweeping sequence, and the mud adhesion coefficient Mij of each partitioned grid is higher than the mud adhesion discrimination value, the set of consecutive partitioned grids is determined as a continuous mud adhesion segment. The requirement of at least two consecutive partitioned grids is used to eliminate misjudgments caused by single-point temperature drift or single-blow pressure fluctuations. For each continuous mud adhesion segment, the cutter wear amount per unit sweeping time within the segment is calculated and compared with the adjacent non-mud adhesion segment. Adjacent non-mud adhesion segments are preferentially selected from partitioned grids within one cutting cycle before and after the continuous mud adhesion segment, and whose mud adhesion coefficient is not higher than the discrimination value; if they exist before and after, the average of the results before and after is taken; if they exist on only one side, the result on that side is taken; if they do not exist on either side, the average cutter wear amount per unit sweeping time in normal formation test samples is used as the comparison benchmark. The wear deviation degree of the continuous mud adhesion segment is thus obtained.

[0105] When the wear deviation does not exceed the average wear of the cutting teeth per unit sweep time plus twice the standard deviation in normal formation test samples, the unit energy consumption weight remains unchanged; when the wear deviation exceeds this upper limit, the unit energy consumption weight of this parameter combination is increased. The increase is determined jointly by the wear deviation and the proportion of the continuous mud-bonded section to the entire sweep path: the greater the wear deviation and the higher the proportion of the continuous mud-bonded section, the greater the increase in the unit energy consumption weight; the increased unit energy consumption weight shall not exceed 1.5 times the original unit energy consumption weight.

[0106] It should be noted that the reason for setting an upper limit is that if the weight of unit energy consumption is too high, the evaluation results will be overly biased towards energy conservation and ignore the efficiency of advance and safety and stability. Limiting it to within 1.5 times can amplify the impact of ineffective energy consumption caused by mud adhesion, while maintaining the balance of multi-objective evaluation.

[0107] In this embodiment, after weight correction, the feed efficiency, unit energy consumption, cutter wear, fuselage sway, and instantaneous gas release risk value are normalized. Specifically, unit energy consumption, cutter wear, fuselage sway, and instantaneous gas release risk value are normalized in the direction that smaller values ​​are better; feed efficiency is reverse-normalized in the direction that larger values ​​are better, so that each indicator is converted into an evaluation quantity where smaller values ​​are better. Then, the multi-objective evaluation value of each parameter combination is recalculated using the corrected unit energy consumption weight, and sorted in ascending order of multi-objective evaluation value.

[0108] Finally, a candidate parameter set P is formed based on the ranking results. The number of candidate parameters is 30% of the remaining parameter combinations after eliminating safety constraints, rounded up; if there are fewer than 3 remaining parameter combinations, all are retained; if there are more than 30 remaining parameter combinations, a maximum of 10 are retained. 30% is used to retain parameter combinations with significant evaluation advantages, with at least 3 retained for on-site comparison of low, medium, and high levels of conservatism, and a maximum of 10 retained to avoid frequent switching and repeated trial runs. The resulting candidate parameter set P simultaneously satisfies the safety constraints of fuselage sway and instantaneous gas release, and can reduce the probability of incorrectly selecting parameter combinations with high wear and high ineffective energy consumption in the mud adhesion zone.

[0109] The multi-objective decision output module is used to select the parameter combination from the candidate parameter set P that satisfies the requirement that the advance efficiency is not lower than the preset efficiency lower limit and the multi-objective evaluation value is the smallest, as the target tunneling parameter output.

[0110] First, determine the lower limit of the advance efficiency. The lower limit is determined based on the target tunnel excavation plan requirements, the advance efficiency under normal strata test cutting, and the equivalent cutability coefficient in the current composite strata response field G. Specifically, first, obtain the planned advance for a single shift in the target tunnel and convert it into the minimum advance efficiency per unit time; then, calculate the average advance efficiency under normal strata test cutting conditions; subsequently, reduce the average advance efficiency of normal strata by the average equivalent cutability coefficient of all grid cells in the composite strata response field G to obtain the current allowable efficiency reference value; finally, take the smaller value between the minimum advance efficiency per unit time and the current allowable efficiency reference value as the lower limit of the advance efficiency. The reason for determining the lower limit of the advance efficiency in this way is that if only the planned advance is used, it may impose excessively high requirements when thin layers of hard rock, cemented fractures, and closed gas pockets coexist; if only the reduced efficiency of the current strata is used, it may lead to excessively low excavation efficiency. Therefore, taking the smaller value of the two can balance the production task and the current cutability of the strata.

[0111] Then, the parameter combinations in the candidate parameter set P are screened for efficiency. The advance efficiency corresponding to each parameter combination is compared with the lower limit of advance efficiency, and parameter combinations with advance efficiency not lower than the lower limit of advance efficiency are retained to form an effective decision set. If the effective decision set is not empty, the effective decision set is sorted in ascending order of multi-objective evaluation value, and the parameter combination with the smallest multi-objective evaluation value is selected as the preferred parameter combination.

[0112] If two or more parameter combinations have the same multi-objective evaluation value, they are compared step by step in the following order: instantaneous gas release risk value, machine sway, cutting tooth wear, unit energy consumption, and advance efficiency. Specifically, the value with the lower instantaneous gas release risk value is selected first; if they are still the same, the value with the lower machine sway is selected; if they are still the same, the value with the lower cutting tooth wear is selected; if they are still the same, the value with the lower unit energy consumption is selected; if they are still the same, the value with the higher advance efficiency is selected. The reason for adopting this order is that the instantaneous gas release risk after the gas chamber is disturbed is a priority safety factor; machine sway directly affects equipment stability; cutting tooth wear affects the reliability of continuous tunneling; unit energy consumption affects economy; and advance efficiency is used as the last comparison factor after meeting the lower limit, thus avoiding a reduction in safety margin due to simply pursuing efficiency.

[0113] If the effective decision set is empty, the parameter combination with the smallest multi-objective evaluation value in the candidate parameter set P is not directly output. Instead, a conservative correction is performed. During conservative correction, the parameter combination with the smallest instantaneous gas release risk value in the candidate parameter set P is taken as the baseline parameter combination. Its advance speed v is reduced by one executable adjustment step, the cutter arm swing speed s is reduced by one executable adjustment step, and the spray volume q is increased by one executable adjustment step. The cutter head rotation speed n is kept at the value corresponding to the baseline parameter combination. Then, the advance efficiency, unit energy consumption, cutter tooth wear, machine yaw, and instantaneous gas release risk value of the corrected parameter combination are recalculated. If the corrected parameter combination satisfies the constraints of machine yaw and instantaneous gas release risk value, it is output as the target tunneling parameter. If the constraints are still not met, correction in the same direction continues until the minimum advance speed, minimum cutter arm swing speed, and maximum spray volume are reached. If the safety constraints are still not met after reaching the above boundaries, the target tunneling parameter is not output, and a pause tunneling prompt is generated. This processing can avoid outputting unsafe parameters in order to meet the efficiency lower limit when the risk of abnormal formations is high.

[0114] After determining the optimal or modified parameter combination, the executability of the cutting head rotation speed n, propulsion speed v, cutting arm swing speed s, and spray volume q is further calibrated. Executability calibration involves adjusting each parameter to the nearest executable setting or adjustment step size that the corresponding actuator can stably maintain, and then confirming that the calibrated parameters still meet the equipment's rated range and safety constraints. If the calibration results in a risk value for instantaneous gas release or the machine body sway exceeding the corresponding threshold, the calibration result is discarded, and the next best parameter combination that meets the constraints is selected.

[0115] The final output target tunneling parameters include cutter head rotation speed, advance speed, cutter arm swing speed and spray volume, and at the same time record the corresponding advance efficiency, unit energy consumption, cutter tooth wear, machine body sway and instantaneous gas release risk value.

[0116] In this embodiment, the above processing ensures that the target tunneling parameters are not only optimal in multi-objective evaluation, but also meet the minimum advance requirements, equipment feasibility requirements, and gas and hull stability and safety constraints under the current geological conditions.

[0117] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A multi-objective optimization decision-making system for tunneling parameters of a mining tunneling machine, characterized in that, include: The data acquisition module is used to acquire the cutting motor current, cutting arm vibration, cutting tooth temperature, gas concentration, propulsion pressure and cutting head angle position of the tunneling machine within the preset test cutting stroke of the target tunneling face. The acquired data is divided into several operating data groups D according to the cutting head angle position. The disturbance feature calculation module is used to calculate the current ripple phase lag value, vibration frequency energy ratio, cutting tooth temperature rise residual value and gas delayed release amount corresponding to each cutting cycle segment based on the running data group D, and generate a disturbance feature set F for characterizing the combined effect of thin-layer hard gangue, cemented cracks and closed gas sacs. The composite formation response field construction module is used to map the disturbance feature set F to a partitioned grid of the tunnel section, obtain the equivalent cutability coefficient, mud adhesion coefficient, and depressurization and gas release sensitivity coefficient corresponding to each partitioned grid, and establish the composite formation response field G, including: Based on the starting angle position, ending angle position, and cumulative advance displacement of the cutting cycle segment, the feature vectors in the disturbance feature set F are mapped to the partition grid of the tunnel section. When the tooth tip sweep area of ​​the truncated periodic segment spans at least two partition grids, the feature vector is weighted and allocated according to the proportion of the tooth tip sweep area. Based on the current ripple phase lag value, vibration frequency harmonic energy ratio, cutter temperature rise residual value, gas delayed release amount, and propulsion pressure within each grid zone, a ternary response vector Gij = (Kij, Mij, Rij) is obtained, where i represents the angular partition number and j represents the radial partition number. The acquisition of the ternary response vector Gij includes the following steps: The dimensionless current ripple phase lag value within the partitioned grid is denoted as B1, the dimensionless vibration frequency energy ratio as B2, the dimensionless cumulative residual value of the cutting tooth temperature rise as B3, the dimensionless gas delayed release amount as B4, and the dimensionless average propulsion pressure as B5. The local cutting resistance index H is calculated according to H=[(1+B1)(1+B2)(1+B5)-1] / 7, and the equivalent machinability coefficient is obtained according to K=1-H. The temperature rise residual growth rate C3 of the cutting teeth is obtained based on the cumulative value of the temperature rise residual of the cutting teeth in adjacent cutting cycle segments within the same grid. The dimensionless propulsion pressure fluctuation amplitude C5 is obtained based on the difference between the maximum and minimum propulsion pressure within the same grid. The mud adhesion coefficient is obtained according to M = 0.5B3 + 0.3C3 + 0.2C5. The gas release boundary enhancement amount C4 is obtained based on the difference in gas delay release amount between adjacent angular zones and adjacent radial zones, and the pressure relief gas release sensitivity coefficient is obtained according to R = 0.6B4 + 0.25B1 + 0.15C4. The tunneling parameter simulation and evaluation module is used to simulate different combinations of cutter head rotation speed, advance speed, cutter arm swing speed, and spray volume based on the composite stratum response field G. It obtains the corresponding advance efficiency, unit energy consumption, cutter tooth wear, machine sway, and instantaneous gas release risk values ​​for each combination, including: The combination of tunneling parameters to be evaluated is represented as X = (n, v, s, q), where n is the cutting head rotation speed, v is the advance speed, s is the cutting arm swing speed, and q is the spray volume. Following the order of the cutting head position, the composite stratum response field G is traversed. Based on the equivalent cuttable coefficient Kij, the mud adhesion coefficient Mij, and the combination of tunneling parameters for each grid cell, the equivalent cutting load Lij for the corresponding grid cell is calculated. The expression is as follows: Where L0 is the normal formation calibration load, v0 is the normal formation calibration advance speed, and n0 is the normal formation calibration cutter head rotation speed; The advance distance is reduced based on the equivalent machinability coefficient Kij of each partition grid, and the advance efficiency E1 is calculated according to E1=Σ(v·Δtij·Kij) / T, where Δtij is the time for the cutting head to sweep through the i-th and j-th partition grids, and T is the complete simulation time. The unit energy consumption is obtained by dividing the sum of the cutting motor energy consumption, propulsion hydraulic energy consumption, and spray energy consumption during the simulation time by the effective propulsion distance. The calculation expression for the cutting motor energy consumption EC is: EC=Σ[2π·n·Tij·Δtij / 60]; the propulsion hydraulic energy consumption is calculated by accumulating the product of the equivalent value of the propulsion pressure and the propulsion speed v; and the spray energy consumption is calculated by accumulating the pump power corresponding to the spray volume q. The wear amount W of the cutting teeth is calculated based on the equivalent cutting load Lij, the mud adhesion coefficient Mij, and the cutting head rotation speed n corresponding to each partition grid. The wear amount of the cutting teeth is obtained according to W=Σ[a·Lij·Δtij+b·Mij·n·Δtij], where a is the load wear calibration coefficient and b is the adhesion friction wear calibration coefficient. The sway Y of the machine body is calculated based on the equivalent cutting load difference between adjacent left and right or upper and lower grid sections; for the same angular section, the Lij difference between its adjacent radial sections and symmetrical angular sections is calculated separately, and the difference is multiplied by the stress arm and then accumulated to obtain the sway torque; then the sway is converted into the sway amount based on the support width of the tunneling machine and the overall mass of the machine. The instantaneous gas release risk value is determined based on the depressurization and gas release sensitivity coefficient Rij corresponding to each grid zone, as well as the propulsion speed v, cutter arm swing speed s, and spray volume q in the parameter combination. The expression is: Where vmax is the maximum stable advance speed allowed by the current tunneling machine, smax is the maximum stable swing speed allowed by the current cutting arm, and qmax is the maximum stable spray volume of the spray device. qmax is determined by the rated pressure of the spray pump and the nozzle flow rate curve. The constraint elimination and weight correction module is used to eliminate parameter combinations where the fuselage sway or instantaneous gas release risk value exceeds the corresponding threshold, and dynamically corrects the unit energy consumption weight based on the cutter wear and mud adhesion coefficient to obtain the candidate parameter set P, including: Parameter combinations that exceed the fuselage yaw threshold or the instantaneous gas release risk value exceed the instantaneous gas release risk threshold will be eliminated. Identify continuous mud-adhesion sections where the mud-adhesion coefficient Mij is continuously higher than the mud-adhesion discrimination value along the actual sweeping sequence of the cutting head, and compare the wear of the cutting teeth in the section with the adjacent non-mud-adhesion sections. When the wear deviation does not exceed the average wear of the cutting teeth per unit sweep time plus twice the standard deviation in normal formation test samples, the unit energy consumption weight remains unchanged; when the wear deviation exceeds the upper limit, the unit energy consumption weight of this parameter combination is increased. And a candidate parameter set P is formed by sorting the corrected multi-objective evaluation values; The multi-objective decision output module is used to select the parameter combination from the candidate parameter set P that satisfies the requirement that the advance efficiency is not lower than the preset efficiency lower limit and the multi-objective evaluation value is the smallest as the target tunneling parameter output; The method for obtaining the multi-objective evaluation values ​​is as follows: the feed efficiency, unit energy consumption, cutting tooth wear, fuselage sway, and instantaneous gas release risk value are normalized respectively. Specifically, the unit energy consumption, cutting tooth wear, fuselage sway, and instantaneous gas release risk value are normalized in the direction that the smaller the value, the better the evaluation; the feed efficiency is reversed and normalized in the direction that the larger the value, the better the evaluation, so that each indicator is converted into an evaluation quantity where the smaller the value, the better. The multi-objective evaluation value of each parameter combination is calculated based on the normalized values ​​and their corrected unit energy consumption weights.

2. The multi-objective optimization decision-making system for tunneling parameters of a mining tunneling machine according to claim 1, characterized in that, The generation of the perturbation feature set F includes the following steps: According to the cutting head angle position, the cutting motor current, cutting arm vibration, cutting tooth temperature and gas concentration and propulsion pressure in the operating data group D are aligned in the angle domain to obtain the multi-source response sequence under the same angle position; Using the sudden change point of the propulsion pressure as the reference point of the load excitation, the current ripple phase lag value is extracted, and the vibration harmonic energy ratio is extracted based on the fundamental frequency of the cutting head rotation. The residual value of the cutting tooth temperature rise is extracted based on the theoretical temperature rise value corresponding to the current and propulsion pressure, and the gas delay release is statistically calculated starting from the disturbance initiation angle position. The four are combined into a disturbance feature set F.

3. The multi-objective optimization decision-making system for tunneling parameters of a mining tunneling machine according to claim 1, characterized in that, The output of the target tunneling parameters includes the following steps: The lower limit of the cutting efficiency is determined based on the planned cutting depth of the target roadway, the cutting efficiency of normal strata, and the equivalent cuttable coefficient Kij in the response field of composite strata. Select parameter combinations from the candidate parameter set P that have a cutting efficiency not lower than the lower limit of the cutting efficiency, and select the parameter combination with the smallest multi-objective evaluation value as the preferred parameter combination. When the multi-objective evaluation values ​​of at least two parameter combinations are the same, they are sorted in order according to the instantaneous gas release risk value, machine body sway, cutting tooth wear, unit energy consumption and advance efficiency. The cutter head speed, advance speed, cutter arm swing speed and spray volume with the best ranking are output as target tunneling parameters.

4. The multi-objective optimization decision-making system for tunneling parameters of a mining tunneling machine according to claim 3, characterized in that, When there is no parameter combination in the candidate parameter set P with a cutting efficiency not lower than the lower limit of the cutting efficiency, the output of the target tunneling parameters further includes the following steps: The parameter combination with the smallest instantaneous gas release risk value in the candidate parameter set P is selected as the benchmark parameter combination. The corrected parameter combination is obtained by reducing the propulsion speed and the cutting arm swing speed in the reference parameter combination by 1 adjustable step and increasing the spray volume by 1 adjustable step. When the combination of corrected parameters satisfies the constraints of fuselage sway and instantaneous gas release risk value, it is output as the target tunneling parameter; If the constraints are not met even after reaching the minimum advance speed, minimum cutter arm swing speed, and maximum spray volume, the target tunneling parameters will not be output.

Images