A tunneling machine propulsion correction control method and medium

Abstract

The application discloses a kind of tunneling machine propulsion deviation correction control method and medium, it is related to underground tunneling equipment control technical field, method includes: obtaining predetermined working path establishes calibration path posture;Activation dual-domain perception module data acquisition establishes perception dataset;Establish position attitude error vector and path deviation trend vector;Position attitude error vector and predetermined working path are used as input data, execute first deviation correction control parameter based on adaptive posture correction channel generation;After compensation by prediction posture compensation channel, generate second deviation correction control parameter.The application solves the technical problem that tunnel construction, tunneling machine propulsion cannot obtain posture and path deviation information comprehensively due to traditional deviation correction means, leading to deviation correction not timely or overcorrection, difficult to meet the precise propulsion demand, reaches the technical effect that precise deviation correction is met in tunneling machine propulsion, the precise propulsion demand of tunneling machine in tunnel construction.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of underground tunneling equipment control, and in particular to a tunneling machine propulsion deviation control method and medium. BACKGROUND

[0002] In tunnel construction, accurate propulsion of the tunneling machine is crucial to the quality and safety of the project, and path deviation during propulsion can lead to decreased construction efficiency, increased project cost, and even safety hazards. In the prior art, the main method for tunneling machine propulsion deviation correction is manual observation and single sensor monitoring. These methods can play a certain role in simple geological conditions and short-distance construction, but as tunnel construction develops towards long distances and complex geological environments, traditional deviation correction techniques have obvious limitations. Due to the complex and variable tunnel geology and the special construction environment, traditional methods cannot fully obtain the posture and path deviation information of the tunneling machine, leading to untimely or excessive deviation correction, which makes it difficult to meet the demand for accurate propulsion of the tunneling machine in tunnel construction. SUMMARY

[0003] The present application solves the technical problem of the inability of traditional deviation correction methods to fully obtain posture and path deviation information during tunneling machine propulsion in tunnel construction, leading to untimely or excessive deviation correction, which makes it difficult to meet the demand for accurate propulsion.

[0004] To solve the above technical problems, the present application provides a tunneling machine propulsion deviation control method and medium.

[0005] In a first aspect, the present application provides a tunneling machine propulsion deviation control method, wherein the method comprises: interactively obtaining a predetermined working path of the tunneling machine, and establishing a calibration path posture according to the predetermined working path; activating a dual-domain perception module, performing perception data collection, and establishing a perception data set, the perception data set including a real-time domain perception data set and a prediction domain perception data set; using the real-time domain perception data set to perform error checking based on the calibration path posture, and establishing a position and posture error vector; performing prediction perception verification of the prediction domain perception data set in the perception data set through the predetermined working path, and establishing a path deviation trend vector; taking the position and posture error vector and the predetermined working path as input data, performing simulated posture correction based on an adaptive posture correction channel, and establishing a first deviation correction parameter, the first deviation correction parameter mapping fitted position and posture data; synchronizing the fitted position and posture data and the path deviation trend vector to a prediction posture compensation channel, and establishing a position compensation parameter; generating a second deviation correction parameter after compensating the first deviation correction parameter through the position compensation parameter.

[0006] In a second aspect, the present application also provides a computer readable storage medium storing a computer program, which, when executed by a processor, implements the tunneling machine propulsion deviation correction control method.

[0007] The present application proposes one or more technical solutions, at least having the following technical effects:

[0008] The present application obtains the predetermined working path of the tunneling machine through interaction and establishes the calibration path posture, activates the dual-domain perception module to collect real-time domain and prediction domain perception data, uses real-time data to verify errors to establish a position and posture error vector, verifies prediction data through the predetermined path to establish a path deviation trend vector, generates a first deviation correction control parameter through the adaptive posture correction channel, and then combines the position compensation parameter of the prediction posture compensation channel for compensation to generate a second deviation correction control parameter, thereby realizing accurate deviation correction during the propulsion process of the tunneling machine, making the propulsion control of the tunneling machine in tunnel construction more accurate and reliable, and achieving the technical effect of accurate deviation correction during the propulsion of the tunneling machine and meeting the accurate propulsion requirements of the tunneling machine in tunnel construction. BRIEF DESCRIPTION OF DRAWINGS

[0009] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without any creative effort.

[0010] Figure 1 is a flowchart of a tunneling machine propulsion deviation correction control method provided by an embodiment of the present application.

[0011] Figure 2 is a flowchart of simulating posture correction based on an adaptive posture correction channel in a tunneling machine propulsion deviation correction control method provided by an embodiment of the present application. DETAILED DESCRIPTION

[0012] The present application provides a tunneling machine propulsion deviation correction control method and medium, which is used to solve the technical problem that in tunnel construction, the posture and path deviation information cannot be comprehensively obtained by traditional deviation correction means when the tunneling machine is propelled, which leads to untimely or excessive deviation correction and makes it difficult to meet the accurate propulsion requirements.

[0013] The technical solutions in the embodiments of the present application will be described clearly and completely in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without any creative effort are within the scope of protection of the present application.

[0014] It is to be understood that any variations of the terms "comprising" and "including" or any other analogous terms are intended to cover a non-exclusive inclusion, for example, a process, method, system, product or device that comprises a list of steps or elements is not necessarily limited to those steps or elements that are expressly listed, but can include additional steps or elements that are not expressly listed or inherent to such process, method, product or device.

[0015] Embodiment one, as shown in a tunneling machine propulsion deviation control method, wherein the method comprises: Figure 1

[0016] Step A100: interactively obtaining a predetermined working path of the tunneling machine, and establishing a calibration path pose according to the predetermined working path.

[0017] In the embodiments of the present application, the tunneling machine is a device used in tunneling and other underground engineering construction, and is the object of propulsion deviation control in the method. The calibration path pose is a parameter set containing position and attitude reference established based on the predetermined working path obtained by interactive acquisition.

[0018] Specifically, the predetermined working path of the tunneling machine is usually obtained based on the pre-generated design drawings of the tunnel engineering, and contains key parameters such as the starting point, the ending point, the trend of each section, the slope, the curvature, etc. of the tunnel. When this path is interactively obtained, a person skilled in the art can input these design parameters through the human-computer interaction interface of the control system, such as a tunnel straight line segment length of 500 meters, a starting point coordinate (X0, Y0, Z0), an ending point coordinate (X500, Y500, Z500), a curve segment curvature radius of 300 meters, a slope of 2°, etc. The system receives and stores these parameters as the original data of the predetermined working path.

[0019] Based on these original data, the system will further process to establish the calibration path pose. First, the predetermined working path is discretized, and the continuous path is decomposed into path points with an interval of 0.3 meters, and each path point corresponds to a set of three-dimensional coordinates (Xi, Yi, Zi). Subsequently, combined with the structural parameters of the tunneling machine, an attitude reference value is configured for each path point, including the heading angle, the pitch angle and the roll angle. For example, the heading angle of the straight line segment path point is uniformly set to 30°, the pitch angle matching the slope is set to 2°, and the roll angle is set to 0°; the heading angle of the curve segment path point is adjusted gradually according to the curvature, and is increased by 0.057° every 0.3 meters, according to the calculation of the curvature radius, 180 / π / 300×0.3≈0.057°, to ensure that the attitude reference is consistent with the path trend.

[0020] These discretized path point coordinates and corresponding attitude reference values jointly constitute the calibration path pose, and each data point is accurate to three decimal places, providing a clear reference for error checking of subsequent real-time domain perception data.​

[0021] Through the above steps, the engineering designed predetermined path is converted into a specific, calibrated path pose containing position and attitude parameters, which provides a precise reference for attitude correction during the propulsion of the roadheader, and ensures the accuracy of subsequent error detection and deviation correction control.

[0022] Step A200: activate the dual-domain perception module, perform perception data acquisition, and establish a perception data set, which includes a real-time domain perception data set and a prediction domain perception data set.

[0023] In the embodiments of the present application, the dual-domain perception module is a perception unit including a real-time domain perception module and a prediction domain perception module, which is used to activate and then perform perception data acquisition. The real-time domain refers to the field of collecting current state data of the roadheader through the real-time domain perception module, and the collected data includes acceleration data, angular velocity data, image point cloud data, and turning angle change data. The prediction domain refers to the field of collecting environment data related to the future path through the prediction domain perception module.

[0024] Optionally, the real-time domain perception module in the dual-domain perception module is activated, which collects acceleration data, angular velocity data, image point cloud data, and turning angle change data through IMU sensors, structured light sensors, and laser gyroscope sensors to establish a real-time domain perception data set, and the specific steps are described in detail in A210-A230. At the same time, the prediction domain perception module in the dual-domain perception module is activated to construct a prediction domain perception data set. The real-time domain perception data set and the prediction domain perception data set are combined to establish a complete perception data set, and the specific steps are described in detail in A241-A243.

[0025] Step A300: use the real-time domain perception data set to perform error checking based on the calibrated path pose, and establish a position and attitude error vector.

[0026] In an embodiment of the present application, first, the current position and attitude information of the roadheader is extracted from the real-time domain perception data set. For example, the image point cloud data collected by the structured light sensor is reconstructed in three dimensions to obtain the current three-dimensional coordinates (Xt, Yt, Zt) of the roadheader; combined with the acceleration data of the IMU sensor and the angular velocity and turning angle change data of the laser gyroscope sensor, the current heading angle αt, pitch angle βt, and roll angle γt are obtained through integral operation, which together constitute the real-time position and attitude parameters of the roadheader.

[0027] Then, the extracted real-time position and attitude parameters are compared with the corresponding parameters in the calibration path attitude one by one. In the calibration path attitude, the path point corresponding to the current tunneling progress is provided with a starting point coordinate (X0, Y0, Z0) and a reference angle (a0, b0, g0). The position deviation is calculated: Ax = Xt-X0, Ay = Yt-Y0, Az = Zt-Z0; the attitude deviation is calculated: a = at-a0, b = bt-b0, g = gt-g0, which quantifies the deviation degree of the real-time state from the reference state.

[0028] Then, the position deviation and the attitude deviation are integrated according to the preset dimension to form a position and attitude error vector. The vector has (Ax, Ay, Az, a, b, g) as components, each component corresponds to the deviation size in a specific direction, for example, Ax = 0.05m represents that the X direction deviates from the reference by 0.05m, a = 0.2° represents that the heading angle deviates from the reference by 0.2°, the modulus value of the vector reflects the overall deviation degree, and the direction reflects the deviation trend.

[0029] By comparing the current position and attitude extracted from the real-time domain perception data with the calibration path attitude, calculating the deviation and integrating, a position and attitude error vector is established, which provides a quantitative deviation basis for subsequent deviation correction control based on the vector.

[0030] Step A400: performing predictive perception verification on the predictive domain perception data set in the perception data set through the predetermined working path, and establishing a path deviation trend vector.

[0031] Specifically, key parameters related to the geological environment in front of the tunneling machine are extracted from the predictive domain perception data set, including stratum density, rock interface depth and geological structure distribution within 50 meters in front, such as the position of small fissures. These data are collected and analyzed by a geological radar sensor, reflecting the geological conditions that the tunneling machine will face when advancing along the predetermined working path.

[0032] Then, the predictive perception verification is performed based on the design parameters of the predetermined working path. The predetermined working path contains preset information about the geological environment along the line, for example, it is expected to pass through homogeneous sandstone in this 50m section, the stratum density is stable at 2.3g / cm³, there is no obvious rock interface and structural fissure. The actual geological parameters in the predictive domain perception data set are compared with the preset information one by one, and the deviation values are calculated, for example, the actual stratum density minimum value is 0.1g / cm³ lower than the preset value, there are 2 rock interfaces that are not identified in the predetermined path, and the crack distribution causes a 20% decrease in local disturbance resistance.

[0033] Based on the above deviation analysis results, the potential impact of these geological deviations on the tunneling machine path is evaluated. For example, a lower density stratum can cause uneven resistance to the tunneling machine's advance, which is expected to result in a lateral deviation of 0.015 meters for every 10 meters of advance; a hardness difference at the rock stratum interface can cause a deviation of 0.08° in the pitch angle. These effects are quantified as trend parameters, including lateral deviation rate, longitudinal slope change rate, heading angle deflection rate, etc.

[0034] Finally, the quantified trend parameters are integrated by spatial dimension to form a path deviation trend vector. The components of this vector include lateral deviation trend, positive value indicating deviation to the right, negative value indicating deviation to the left; longitudinal slope trend, positive value indicating steeper slope, negative value indicating gentler slope; heading angle change trend, positive value indicating clockwise deflection, negative value indicating counterclockwise deflection, thus fully reflecting the direction and extent of possible path deviation of the tunneling machine along the predetermined path in the future.

[0035] By validating the prediction domain perception dataset with the predetermined working path and quantifying the impact of geological deviations on the path, the path deviation trend vector is established, providing a quantitative basis for predicting the future path deviation of the tunneling machine.

[0036] Step A500: input the position and attitude error vector and the predetermined working path as input data, perform simulated attitude correction based on an adaptive attitude correction channel, and establish first deviation correction control parameters, which map fitted position and attitude data.

[0037] In the embodiments of the present application, the adaptive attitude correction channel is a functional channel for performing simulated attitude correction. The first deviation correction control parameters are deviation correction control parameters established based on multiple rounds of repair optimization results, which are integrated by iteratively optimized key control parameters. These parameters are the optimal solution obtained by combining the predetermined working path and the position and attitude error vector under the goal of balancing transient equipment damage, path disturbance repair, and compensation distance.

[0038] Optionally, the adaptive attitude correction channel is initialized by obtaining geological stratum data, a deviation correction objective function containing transient equipment damage balancing term, path disturbance repair balancing term, and compensation distance balancing term is configured, and multiple rounds of repair optimization are performed on the position and attitude error vector with the predetermined working path as the follow-up target to establish the first deviation correction control parameters. The specific steps are described in detail in A510-A530.

[0039] Step A600: synchronize the fitted position and attitude data and the path deviation trend vector to the prediction attitude compensation channel to establish position compensation parameters.

[0040] In the embodiments of the present application, the prediction attitude compensation channel is a functional channel for processing fitted position and attitude data and path deviation trend vectors to establish position compensation parameters.

[0041] Optionally, after the key frame sequence is established by calling the key frame extraction sub-channel to extract the key frame of the fitted position attitude data, the trend trajectory stream is generated and mapped to the sequence, and the key frame offset trend is analyzed by calling the attitude processing layer to establish the position compensation parameter, the specific steps are described in detail in A610-A630.

[0042] Step A700: After the first deviation correction control parameter is compensated by the position compensation parameter, a second deviation correction control parameter is generated.

[0043] Optionally, the first deviation correction control parameter is a basic control parameter obtained by multiple rounds of repair optimization in the adaptive attitude correction channel, including lateral deviation adjustment value, heading angle correction parameter, etc., and the fitted position attitude data mapped thereby reflects the expected attitude of the roadheader after adjustment according to the parameter. The position compensation parameter is generated based on the fitted position attitude data and the path offset trend vector by the prediction attitude compensation channel, and is used to compensate the offset value of the path offset, which covers the additional adjustment amount of the key frame position.

[0044] When compensating, the position compensation parameter is matched with the first deviation correction control parameter according to the corresponding matching of the propulsion position. For example, at the key frame position of the path turning, the lateral compensation amount in the position compensation parameter is added to the lateral deviation adjustment value of the first deviation correction control parameter; at the key frame position of the speed change, the corresponding heading compensation amount is integrated into the heading angle correction parameter of the first parameter, so that the adjustment is more in line with the trend of path offset.

[0045] After compensation, the integrated parameters are reasonably checked to ensure that the adjustment range is within the preset safety range, avoiding exceeding the attitude adjustment capability of the device. After checking, these compensated and integrated parameters are arranged to form the second deviation correction control parameter, which retains the repair logic of the first deviation correction control parameter for the current attitude error, and integrates the pre-judgment compensation for the future path offset.

[0046] By compensating and integrating the first deviation correction control parameter with the position compensation parameter, the second deviation correction control parameter generated takes into account both the current attitude correction and the path offset trend prediction, improving the accuracy and foresight of the roadheader propulsion deviation control.

[0047] Further, as shown in Figure 2 , the method provided by the embodiment of the present application comprises:

[0048] A510: Obtain geological stratum data, and initialize the adaptive attitude correction channel according to the geological stratum data.

[0049] A520: configure a correction target function in the adaptive posture correction channel, the correction target function is a balance target function, the balance term of the correction target function includes a transient device damage balance term, a path disturbance repair balance term, and a compensation distance balance term.

[0050] A530: taking the predetermined working path as a following target, performing multi-round repair optimization of the position and posture error vector based on the correction target function, and establishing first correction control parameters according to the multi-round repair optimization result.

[0051] In the embodiments of the present application, the transient device damage refers to the short-term and instantaneous damage that may be caused to the device due to instantaneous posture adjustment or force change during the correction process of the heading machine, which is one of the factors that need to be balanced in the correction target function. The path disturbance refers to the interference or fluctuation generated by the deviation of the actual advancing path of the heading machine from the predetermined working path, which needs to be repaired through the correction mechanism and is one of the contents to which the balance term of the correction target function is directed. The compensation distance refers to the distance involved in the correction adjustment needed to make the heading machine return to the predetermined working path from the deviation state, which is one of the key parameters that need to be balanced in the correction target function.

[0052] Specifically, first, geological stratum data including stratum type, disturbance resistance, and rebound modulus parameters are obtained, and a parameter template containing correction weight, adjustment amplitude threshold, and path recovery sensitivity is configured in the adaptive posture correction channel according to these data to complete the initialization of the channel, and the specific steps are described in detail in A511-A512.

[0053] Next, the correction target function in the adaptive posture correction channel is configured, and the quantification standards of the three balance terms are first determined. For the transient device damage balance term, the instantaneous force data of the key components of the heading machine, such as the hydraulic cylinder and the cutter, are monitored, and the damage coefficient is set: when the hydraulic system pressure is within the safe range of 10-15 MPa, the damage coefficient is 0.2; when it exceeds 15 MPa to the vulnerable range, the coefficient is linearly increased to 0.8, quantifying the short-term damage risk of the device.

[0054] The path disturbance repair balance term takes the cumulative value of the deviation between the actual path and the predetermined path as an index: when the deviation is slight disturbance of 0-0.05 meters, the repair weight is set to 0.3; when the deviation is moderate disturbance of 0.05-0.1 meters, the weight is increased to 0.6; and when the deviation is severe disturbance of more than 0.1 meters, the weight is increased to 0.9, to preferentially ensure the repair intensity for large deviations.

[0055] The compensation distance balance term takes the straight-line distance moved for correction as a parameter: when the distance is within the reasonable range of 0.1-0.3 meters, the balance coefficient is 0.5; when the distance is less than 0.1 meter, the coefficient is reduced to 0.3, which may not be completely repaired; and when the distance is greater than 0.3 meters, the coefficient is reduced to 0.2, to avoid excessive compensation or insufficient compensation.

[0056] The three balance terms are integrated into a balance objective function by weighted summation, and the formula is F=0.3 D+0.5 P+0.2 L, the weight is adjusted according to geology, and can be set to 0.3, 0.4, 0.3 in hard strata, wherein D is a transient device damage value, P is a path disturbance value, and L is a compensation distance value. Through historical correction data training, when the function value F is minimized, it is usually ≤0.3, and the corresponding parameters can balance the three indicators at the same time.

[0057] Then, an initial parameter set is configured according to the position and attitude error vector and the predetermined working path, the strategy fitness of the initial parameter set is analyzed by using the correction target function, and a fitness value is established. After a random factor is configured, the parameter set is iteratively updated in combination with the fitness value, and continuous iteration is performed to complete multiple rounds of repair optimization, and specific steps are described in detail in A531-A535.

[0058] After the multiple rounds of repair optimization are completed, the key control parameters optimized through iteration are extracted from the final converged parameter set. These parameters are the optimal solution obtained by balancing the transient device damage, path disturbance repair and compensation distance under the combination of the predetermined working path and the position and attitude error vector, including the lateral correction amount, the heading angle correction value and the longitudinal slope adjustment coefficient. These parameters are integrated according to the control logic to form the first correction control parameter, which directly corresponds to the attitude adjustment instruction of the roadheader and guides the device to perform targeted attitude correction during the advancing process.

[0059] Meanwhile, the fitting position and attitude data mapped by the first correction control parameter are the expected attitude data of the roadheader simulated and calculated based on these control parameters, including the three-dimensional coordinates after correction, the pitch angle, the roll angle and the heading angle, etc. These data reflect the position and attitude state of the roadheader that should theoretically be achieved after adjustment by the first correction control parameter, and provide a reference for subsequent calculation of the position compensation parameter of the predicted attitude compensation channel.

[0060] By configuring the balance objective function including the three balance terms of transient device damage, path disturbance repair and compensation distance, the quantitative balance of multiple constraints in the correction process is realized, and an explicit optimization target is provided for subsequent multiple rounds of optimization based on the function.

[0061] Further, the step A530 in the method provided in the embodiment of the application comprises:

[0062] A531: An initial parameter set is configured according to the position and attitude error vector and the predetermined working path.

[0063] A532: The strategy fitness of the initial parameter set is analyzed by using the correction target function, and a strategy fitness value is established.

[0064] A533: After the random factor is configured, the initial parameter set is iteratively updated using the random factor and the policy fitness value, and an iterative update result is established.

[0065] A534: After the initial parameter set is updated using the iterative update result, one round of iterative update is completed.

[0066] A535: The continuous update iteration of the initial parameter set is performed, and multiple rounds of repair optimization are completed.

[0067] In the embodiments of the application, the policy fitness is obtained by analyzing the initial parameter set by the correction target function, and is an index reflecting the adaptation degree of the parameter set in repairing position and attitude error and following the predetermined working path. The policy fitness value is quantified, and the smaller the value is, the higher the fitness is. The random factor is a random numerical value configured during the iterative update of the initial parameter set, which is used to adjust the update amplitude and direction of the parameter in combination with the policy fitness value, to help expand the optimization range and avoid the parameter iteration from falling into local optimum.

[0068] Optionally, the initial parameter set is configured according to the position and attitude error vector and the predetermined working path, and the parameter set contains key parameters such as lateral correction amount, longitudinal slope adjustment value, and heading angle correction coefficient. For example, if the position and attitude error vector shows that the lateral offset is 0.08 meters and the heading angle deviation is 0.3°, and the predetermined working path is a curve with a curvature radius of 500 meters, the initial parameter set can be set to a lateral correction amount of 0.01 meters per propulsion cycle, a heading angle correction coefficient of 0.05° per cycle, and a longitudinal slope consistent with the predetermined path.

[0069] Then, the policy fitness of the initial parameter set is analyzed by using the correction target function, and the fitness value is calculated. Assuming that the parameter set of the hard stratum is 0.25, the equipment stress is within the safety range, the path disturbance repair balance item value is 0.3, the deviation repair efficiency is moderate, and the compensation distance balance item value is 0.2, the compensation distance is reasonable, through the target function F=0.3×0.25+0.4×0.3+0.3×0.2=0.075+0.12+0.06=0.255, the policy fitness value is 0.255, and the smaller the value is, the higher the fitness is.

[0070] Then, a random factor is configured, such as a value range of 0.02-0.05, for example, a random factor of 0.03 is selected. The initial parameter set is iteratively updated in combination with the random factor and the strategy fitness value: for example, when the strategy fitness value is 0.255, the adaptability is medium, the random factor of 0.03 is selected, the lateral correction amount in the initial parameter set is adjusted by 3% on the basis of 0.01 meters to 0.0103 meters, and the heading angle correction coefficient is adjusted by 3% on the basis of 0.05° to 0.0515°. While ensuring the effectiveness of the parameter update direction, it avoids excessive fluctuations affecting the stability of the optimization.

[0071] Further, the configuration of the random factor needs to be combined with the accuracy requirement and optimization efficiency of the roadheader parameter adjustment, and the value range is obtained based on historical correction data and parameter sensitivity analysis. For example, by statistically analyzing the parameter adjustment range of the past 100 successful correction cases, it is found that when the single adjustment proportion of parameters such as the lateral correction amount and the heading angle correction coefficient is 2%-5%, the effectiveness of parameter update can be ensured, and the strategy fitness value can be prevented from rising sharply due to excessive amplitude, that is, the adaptability is reduced. Therefore, the value range of the random factor is set to 0.02-0.05. When configuring specifically, the current strategy fitness value is dynamically selected: if the strategy fitness value is high, such as 0.3 and above, the adaptability is poor, a larger random factor of 0.04-0.05 is selected to expand the parameter exploration range to quickly improve the adaptability; if the strategy fitness value is low, such as 0.2 and below, the adaptability is good, a smaller random factor of 0.02-0.03 is selected to achieve fine adjustment of the parameters to stabilize the adaptability.

[0072] Next, the iteratively updated result obtained by adjusting the random factor and the strategy fitness value, that is, the new parameter set containing the lateral correction amount of 0.0103 meters / cycle and the heading angle correction coefficient of 0.0515° / cycle, replaces the original initial parameter set, that is, the lateral correction amount of 0.01 meters / cycle and the heading angle correction coefficient of 0.05° / cycle, to complete a round of iterative update.

[0073] Subsequently, the new parameter set is re-analyzed for strategy fitness based on the correction target function, assuming that the calculated strategy fitness value is 0.248, which is lower than the strategy fitness value of the initial parameter set of 0.255. Since the smaller the strategy fitness value, the better the adaptability of the parameter set in balancing the transient equipment damage, path disturbance repair, and compensation distance, this result shows that the adjustment of the parameter set in combination with the random factor and the strategy fitness value can more effectively promote the optimization of the parameters in the direction of adapting to the predetermined working path and the repair position and attitude error, and the adjustment direction is effective.

[0074] Finally, the optimal fitness parameter after each iteration is added to the historical memory pool, and after the preset threshold number of iterations, the historical memory pool is activated and trend analysis is performed to extract trend control factors, which are used to constrain subsequent iterations to complete multiple rounds of repair optimization. The specific steps are described in detail in A535-1-A535-3.

[0075] By configuring the initial parameters, analyzing the fitness, updating with random factors, and completing one round of iteration according to the above steps, a foundation is provided for subsequent continuous iteration optimization, gradually improving the matching degree of the parameter set and the current position and attitude error and the predetermined path.

[0076] Further, the method provided in the embodiments of the present application includes the following steps A535:

[0077] A535-1: After each round of iteration is completed, the optimal fitness parameter in the current parameter set is added to the historical memory pool.

[0078] A535-2: When the number of iterations meets the preset threshold, the historical memory pool is activated, and trend analysis of the historical memory pool is performed to extract trend control factors.

[0079] A535-3: The trend control factors are used to constrain the optimization of subsequent iterations to complete multiple rounds of repair optimization.

[0080] In the embodiments of the present application, the historical memory pool is a storage unit for storing the optimal fitness parameter in the current parameter set after each round of iteration is completed.

[0081] Specifically, after each round of iteration is updated, the parameter combination with the smallest strategy fitness value is selected from the current parameter set as the optimal fitness parameter. For example, after a certain round of iteration, the parameter set contains 5 groups of parameters, and their strategy fitness values are 0.28, 0.25, 0.31, 0.23, and 0.26, respectively. The parameter corresponding to 0.23, i.e., the lateral correction amount 0.012 meters / cycle and the heading angle correction coefficient 0.04° / cycle, is determined as the optimal fitness parameter. The specific values of the parameter and the corresponding fitness value are stored in the historical memory pool. The memory pool records the optimal parameters of the first 20 rounds in order according to the iteration order.

[0082] Then, when the number of iterations reaches a preset threshold, such as 30 rounds, the historical memory pool is activated and the stored optimal fitness parameters are analyzed for trends. By calculating the change in parameters between adjacent rounds, it is found that the lateral correction amount gradually decreases from the initial 0.018 meters per cycle to 0.012 meters per cycle, with an average reduction of 0.003 meters per 5 rounds; the heading angle correction coefficient decreases from 0.07° / cycle to 0.04° / cycle, with an average reduction of 0.006° per 5 rounds. Based on these change patterns, trend control factors are extracted, including the convergence rate of the lateral correction amount (0.003 meters / 5 rounds), the stable threshold of the heading angle correction coefficient (≤0.04° / cycle), and the maximum fluctuation amplitude of parameter adjustment (±0.002 meters / cycle).

[0083] After that, the extracted trend control factors are applied to the subsequent iteration process to constrain parameter updates. For example, the trend control factor limits the lateral correction amount to no more than 0.002 meters per round, and when the parameter approaches the stable threshold, such as the heading angle correction coefficient of 0.04° / cycle, the adjustment amplitude is further reduced to 0.001° / cycle, avoiding large fluctuations of the parameter around the optimal value. At the same time, if the parameter update direction of a certain round of iteration is opposite to the convergence direction indicated by the trend control factor, such as the lateral correction amount suddenly increasing to 0.015 meters / cycle, a secondary check is triggered to recalculate the strategy fitness value to correct the update direction.

[0084] The above process is continuously executed, allowing the parameter set to gradually converge to the global optimal solution under the constraint of the trend control factor, completing multiple rounds of repair optimization.

[0085] By storing the optimal parameters each round, triggering trend analysis based on thresholds to extract control factors and constrain subsequent iterations, stable convergence of the parameter set to the optimal solution is achieved, and multiple rounds of repair optimization are completed.

[0086] Further, the step A510 in the method provided by the embodiment of the present application comprises:

[0087] A511: The geological stratum data comprises stratum type, disturbance resistance, and rebound modulus parameters.

[0088] A512: Configure a parameter template in the adaptive attitude correction channel according to the geological stratum data, wherein the parameter template comprises a correction weight, an adjustment amplitude threshold, and a path recovery sensitivity.

[0089] Specifically, first, geological stratum data is obtained, and a person skilled in the art can obtain stratum types, disturbance resistance, and rebound modulus parameters by integrating information collected by geological survey reports, drilling data, and geological radar sensors in a prediction domain perception module. For example, it is detected that a 20-meter road section ahead is a sandstone stratum, the disturbance resistance value of which is 85, the value range is 0-100, and the higher the value, the less likely the stratum is disturbed, and the rebound modulus is 32 MPa; the next 30 meters is a shale stratum, the disturbance resistance of which is 60, and the rebound modulus is 21 MPa.

[0090] Next, an adaptive attitude correction channel is constructed, which includes a data input layer, a parameter mapping layer, and an output layer. The input layer receives geological stratum data, the parameter mapping layer has an associated model of geological parameters and correction parameters built-in, and the output layer outputs a prepared parameter template. When training the channel, historical construction data is used as a sample, which includes geological parameters corresponding to different geological conditions such as sandstone and shale, and the optimal parameter template verified by practice under the condition, such as correction weight, adjustment amplitude threshold, and path recovery sensitivity. The correlation coefficient of the parameter mapping layer is adjusted through iterative training, so that the deviation of the parameter template output by the channel when inputting new geological data from the historical optimal template is controlled within 8%, and the channel training is completed.

[0091] Then, the parameter template is configured according to the obtained geological stratum data, and is mapped according to the associated model after the channel training. For example, for a sandstone stratum, the correction weight is set to 0.7, because a hard stratum responds more directly to correction instructions, a higher weight needs to be given, the adjustment amplitude threshold is set to ±0.06 meters to avoid overloading of equipment caused by strong correction, and the path recovery sensitivity is set to 0.8 to quickly respond to deviations to maintain the path; for a shale stratum, the correction weight is set to 0.5, the soft stratum needs to be adjusted gently, the weight is reduced, the adjustment amplitude threshold is set to ±0.03 meters to reduce disturbance to the stratum, and the path recovery sensitivity is set to 0.6 to gradually recover the path to adapt to the characteristics of the stratum.

[0092] By obtaining geological stratum data, the adaptive attitude correction channel constructed and trained maps and configures a parameter template that adapts to the current geological conditions, completes channel initialization, and provides basic parameters matched with geological characteristics for subsequent correction control based on the channel.

[0093] Further, step A600 in the method provided in the embodiments of the present application includes:

[0094] A610: calling a key frame extraction sub-channel in the prediction attitude compensation channel, performing key frame extraction of the fitted position attitude data according to the key frame extraction sub-channel, and establishing a key frame sequence, wherein the key features of the key frame extraction include spatial path turning features and speed change features.

[0095] A620: After the trend trajectory stream is generated based on the path offset trend vector, the alignment mapping of the trend trajectory stream and the key frame sequence is performed.

[0096] A630: The pose processing layer in the predicted pose compensation channel is called to perform key frame offset trend influence analysis on each key frame in the key frame sequence according to the alignment mapping result, and to establish a position compensation parameter.

[0097] Specifically, first, a predicted pose compensation channel is constructed, including two core modules of a key frame extraction sub-channel and a pose processing layer. The key frame extraction sub-channel is built-in with a spatial feature detector and a speed feature analyzer. The spatial feature detector is used to identify the turning position of the path curvature change exceeding 5° in the fitted position and pose data, and the speed feature analyzer is used to capture the moment when the tunneling speed fluctuation exceeds 0.1 m / s; the pose processing layer is equipped with an offset influence evaluation model, which takes the key frame pose parameters and the path offset trend vector as input, and outputs a single frame offset influence coefficient.

[0098] When training the channel, 100 groups of sample data in the history construction are used, which includes fitted position and pose data, corresponding path offset trend vector and the best position compensation parameter verified by practice. By adjusting the turning recognition threshold of the spatial feature detector, such as adjusting the initial 3° to 5°, the key frame extraction accuracy is improved, the fluctuation sensitivity of the speed feature analyzer is optimized, and when 0.1 m / s is set as the threshold, the speed feature capture error is ≤0.02 m / s, and the weight coefficient of the offset influence evaluation model is iteratively trained, so that the deviation between the predicted compensation parameter and the actual value is ≤0.005 m, and the channel training is completed.

[0099] Then, the key frame extraction sub-channel in the predicted pose compensation channel is called to extract key frames from the fitted position and pose data. The fitted position and pose data includes the simulated pose parameters of the roadheader along the predetermined path. For example, in the 0-100 meter propulsion range, the path has a turning point at 30 meters where the curvature radius changes from 500 meters to 300 meters, i.e. the spatial path turning feature, and the tunneling speed decreases from 0.5 m / s to 0.3 m / s at 60 meters, i.e. the speed change feature. The key frame extraction sub-channel marks the pose data at 30 meters and 60 meters as key frames according to the trained threshold, and forms a key frame sequence together with the pose parameters of the previous and subsequent 5 meters. Each key frame records the three-dimensional coordinates, heading angle and speed value of the corresponding position.

[0100] Then, a trend trajectory stream is generated based on a path offset trend vector, the vector containing a lateral offset trend +0.015 m / 10 m and a longitudinal slope trend -0.08° / 10 m, and the trend trajectory stream generated therefrom presents an offset cumulative effect along the advancing direction, such as 0.015 m of lateral offset at 10 m, 0.03 m of cumulative offset at 20 m, and 0.045 m of cumulative offset at 30 m. Alignment mapping of the trend trajectory stream and the key frame sequence is performed, and the key frame at 30 m in the key frame sequence is associated with the offset data at the 30 m position in the trend trajectory stream, i.e., 0.045 m of lateral offset and -0.24° of slope, and the key frame at 60 m is associated with the offset data at the 60 m position in the trajectory stream, i.e., 0.09 m of lateral offset and -0.48° of slope, so as to ensure that each key frame is accurately matched with the offset trend at the same position in the trajectory stream.

[0101] Subsequently, a pose processing layer in the predicted pose compensation channel is called to analyze the offset trend influence of each key frame according to the alignment mapping result. For example, the key frame at 30 m is affected by the lateral offset trend due to the path turning, and the predicted pose is expected to additionally generate 0.008 m of lateral deviation; the key frame at 60 m is affected by the longitudinal slope trend due to the speed reduction, and the pitch angle may additionally deviate by 0.03°. The pose processing layer calculates the compensation amount of each key frame through an offset influence evaluation model, and the key frame at 30 m needs to be compensated by -0.008 m in the lateral direction to offset the deviation, and the key frame at 60 m needs to be compensated by +0.03° in the pitch angle, and these compensation amounts are integrated in the advancing order to form position compensation parameters, which include the lateral compensation value, the pitch angle compensation value and the heading angle compensation value corresponding to each 10 m of advancement.

[0102] By constructing and training the predicted pose compensation channel, extracting the key frame sequence, aligning the trend trajectory stream and analyzing the offset influence, the position compensation parameters are established, which provide accurate quantitative basis for compensating the first deviation correction control parameters using the parameters.

[0103] Further, the step A800 in the method provided by the embodiments of the present application comprises:

[0104] A810: after the functional advancement task is performed by using the second deviation correction control parameter, collecting pose feedback data in the execution stage.

[0105] A820: synchronizing the pose feedback data and the fitted position pose data to a feedback comparison module to perform pose difference analysis and establish a deviation feedback vector.

[0106] A830: performing channel compensation update of the adaptive pose correction channel by using the deviation feedback vector.

[0107] In the embodiments of the present application, the feedback comparison module is a functional module for receiving pose feedback data and fitted position pose data, which performs pose difference analysis by synchronizing the two types of data.

[0108] In one embodiment, after the second deviation correction control parameter is generated, the parameter is used to guide the roadheader to perform the function advancing task. During the advancing process, the posture feedback data of the execution stage is collected by the sensors of the real-time domain perception module, which reflects the position, angle and other posture information of the actual operation of the roadheader.

[0109] Then, the collected posture feedback data and the fitted position posture data mapped by the first deviation correction control parameter are transmitted to the feedback comparison module. The feedback comparison module analyzes the posture difference between the two, compares the deviation between the actual posture and the expected posture at the same advancing position, integrates the deviations by dimension, and forms a deviation feedback vector, which directly reflects the gap between the actual execution and the theoretical expectation.

[0110] Then, the deviation feedback vector is used to update the channel compensation of the adaptive posture correction channel. According to the deviation information in the vector, the channel adjusts the internal parameter template (such as correction weight, adjustment amplitude threshold) and the balance term weight of the deviation correction objective function, so that the correction logic of the channel is more suitable for the actual working condition, and the accuracy of subsequent simulated posture correction is improved.

[0111] By collecting posture feedback data, analyzing differences to establish a deviation feedback vector, and updating an adaptive posture correction channel, dynamic optimization of the deviation correction control process is achieved, and the adaptability and accuracy of the deviation correction control of the roadheader are enhanced.

[0112] Further, the method provided in the embodiment of the application comprises the following steps A200:

[0113] A210: activate the real-time domain perception module in the dual-domain perception module, wherein the perception sensors of the real-time domain perception module include IMU sensors, structured light sensors and laser gyro sensors.

[0114] A220: perform data collection using the real-time domain perception module to establish a real-time domain perception data set, wherein the real-time domain perception data set includes acceleration data, angular velocity data, image point cloud data and rotation angle change data.

[0115] A230: establish a perception data set according to the real-time domain perception data set.

[0116] Optionally, the state perception of the tunneling machine often relies on a single type of sensor, with limited data dimensions, making it difficult to fully reflect the real-time state. First, activate the real-time domain perception module in the dual-domain perception module. The IMU sensor equipped in this module can capture three-dimensional acceleration data of the tunneling machine in real time, with a sampling frequency of 100 Hz and a data range of -10 to 10 m / s²; the laser gyroscope sensor outputs angular velocity data at a frequency of 50 Hz, with an accuracy of 0.01 rad / s, synchronously records the change in angle, and the resolution is 0.001°; the structured light sensor generates a frame of image point cloud data every 0.5 seconds, containing millions of three-dimensional coordinate points of the tunneling machine and the relative position of the surrounding environment.

[0117] Next, using the cooperative work of these sensors, the real-time domain perception module continuously collects data: the IMU sensor updates acceleration information every 10 milliseconds, reflecting the instantaneous motion acceleration of the tunneling machine; the laser gyroscope synchronously outputs the corresponding angular velocity and change in angle, recording the rotation state of the machine body; the structured light sensor generates high-density point clouds by projecting structured light and receiving reflected signals, presenting the relative position relationship between the tunneling machine and the surrounding rock of the tunnel. These data are integrated to form a real-time domain perception dataset containing time stamps, sensor identifiers, and corresponding measurement values.

[0118] At the same time, activate the prediction domain perception module in the dual-domain perception module, which is equipped with a geological radar sensor to scan the geological environment in front and collect data such as stratum density and rock interface depth, etc. After analysis, a prediction domain perception dataset is formed, and the specific steps are described in detail in A241-A243. The sensors and collected data sets of the dual-domain perception module are shown in Table 1.

[0119] Finally, the real-time domain perception dataset and the prediction domain perception dataset are timestamped and standardized, and finally a complete perception dataset is established, which contains not only real-time data reflecting the current state of the tunneling machine, but also prediction data reflecting the geological environment in front.

[0120] Through the dual-domain perception module, real-time domain and prediction domain data are collected and integrated to establish a perception dataset containing real-time state and prediction information, providing comprehensive data support for subsequent position and attitude error correction and path deviation trend analysis.

[0121] Table 1: Correspondence table of sensors and collected data sets of the dual-domain perception module

[0122]

[0123] Further, the method provided in the embodiments of the present application comprises the following steps:

[0124] A241: activate the prediction domain perception module in the dual-domain perception module, the prediction domain perception module comprising a ground penetrating radar sensor.

[0125] A242: perform data collection using the prediction domain perception module to establish a prediction domain perception dataset.

[0126] A243: establish a perception dataset from the prediction domain perception dataset and the real-time domain perception dataset.

[0127] Optionally, a prediction domain perception module in the dual-domain perception module is activated, which carries a ground penetrating radar sensor. The sensor detects the geological environment within 30 meters in front of the heading machine at a scanning interval of 5 meters. The ground penetrating radar transmits high-frequency electromagnetic waves and receives reflection signals from different stratum interfaces. After signal processing, data such as stratum density, stratum interface depth, stratum hardness gradient, and position coordinates of small faults are analyzed.

[0128] Then, the analyzed data are indexed according to collection time and corresponding heading position to form a prediction domain perception dataset. The dataset contains a sequence of geological parameters in each detection range, for example, in the 10-15 meter interval, the stratum density decreases from 2.4 g / cm³ to 2.2 g / cm³, and there is a 0.5 meter thick weak interlayer. These data reflect the changes in geological conditions that the heading machine will encounter when advancing along the predetermined path.

[0129] Then, the prediction domain perception dataset is integrated with the existing real-time domain perception dataset. Through timestamp matching, the data segments of acceleration, angular velocity, etc. corresponding to the prediction domain detection position in the real-time domain are associated with the prediction geological data in this interval, and then standardized according to a unified data format, such as a three-dimensional coordinate-geological parameter-time matrix form, to finally form a complete perception dataset.

[0130] By activating the prediction domain perception module to collect and process geological data, and combining the real-time domain perception dataset to integrate and establish the perception dataset, prediction data containing front geological information are provided for subsequent path deviation trend analysis, which together with the real-time data constitute a comprehensive perception basis.

[0131] In summary, the heading machine advancing deviation control method provided by the embodiments of the present application has the following technical effects:

[0132] The application obtains the predetermined working path of the tunneling machine through interaction and establishes a calibration path posture, activates a dual-domain perception module to collect perception data, establishes a position posture error vector and a path deviation trend vector through error checking and prediction verification, generates a correction control parameter through an adaptive posture correction channel and a prediction posture compensation channel, updates the correction channel in combination with a feedback comparison module, thereby realizing accurate correction control in the tunneling machine advancing process, making the tunneling machine advancing path more consistent with the predetermined working path, improving the accuracy and stability of the tunneling operation, and achieving accurate correction when the tunneling machine advances, meeting the accurate advancing requirements of the tunneling machine in tunnel construction.

[0133] In the second embodiment, the application further provides a computer readable storage medium, which can be used to store software programs, computer executable programs and modules, such as program instructions / modules corresponding to the tunneling machine advancing correction control method in the embodiments of the application, so as to realize the above-mentioned tunneling machine advancing correction control method.

[0134] It should be understood that the embodiments and the above description disclosed by the application can enable those skilled in the art to implement the application. Meanwhile, the application is not limited to the above-mentioned part of the embodiments, and it should be understood that those skilled in the art can still modify the technical solutions recorded in the above-mentioned embodiments or make equivalent replacement for some technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the application, and should be included in the protection scope of the application.

Claims

1. A method for propulsion deviation correction control of a heading machine, characterized by, The method comprises: interactively obtaining a predetermined working path of the heading machine, and establishing a calibration path pose according to the predetermined working path; activating a dual-domain perception module, performing perception data acquisition, and establishing a perception data set, wherein the perception data set comprises a real-time domain perception data set and a prediction domain perception data set; performing error checking based on the calibration path pose by using the real-time domain perception data set, and establishing a position and pose error vector; performing prediction perception verification of the prediction domain perception data set in the perception data set by using the predetermined working path, and establishing a path deviation trend vector; taking the position and pose error vector and the predetermined working path as input data, performing analog pose correction based on an adaptive pose correction channel, establishing a first deviation correction control parameter, and mapping the first deviation correction control parameter with fitting position and pose data; synchronizing the fitting position and pose data and the path deviation trend vector to a prediction pose compensation channel, and establishing a position compensation parameter; generating a second deviation correction control parameter after compensating the first deviation correction control parameter by using the position compensation parameter; taking the position and pose error vector and the predetermined working path as input data, performing analog pose correction based on an adaptive pose correction channel, comprising: obtaining geological stratum data, and initializing the adaptive pose correction channel according to the geological stratum data; configuring a deviation correction objective function in the adaptive pose correction channel, wherein the deviation correction objective function is a balance objective function, and a balance term of the deviation correction objective function comprises a transient equipment damage balance term, a path disturbance repair balance term, and a compensation distance balance term; taking the predetermined working path as a following target, performing multi-round repair optimization of the position and pose error vector based on the deviation correction objective function, and establishing a first deviation correction control parameter according to a multi-round repair optimization result; after generating the second deviation correction control parameter, comprising: after executing a functional propulsion task by using the second deviation correction control parameter, collecting pose feedback data in an execution stage; synchronizing the pose feedback data and the fitting position and pose data to a feedback comparison module, performing pose difference analysis, and establishing a deviation feedback vector; performing channel compensation update of the adaptive pose correction channel by using the deviation feedback vector.

2. The propulsion deviation correction control method of a tunneling machine according to claim 1, characterized in that, the taking of the predetermined working path as a following target, the performing of multi-round repair optimization of the position and pose error vector based on the deviation correction objective function, comprising: configuring an initial parameter set according to the position and pose error vector and the predetermined working path; performing strategy fitness analysis of the initial parameter set by using the deviation correction objective function, and establishing a strategy fitness value; after configuring a random factor, performing iterative update of the initial parameter set by using the random factor and the strategy fitness value, and establishing an iterative update result; after updating the initial parameter set by using the iterative update result, completing one round of iterative update; performing continuous update iteration of the initial parameter set, and completing multi-round repair optimization.

3. The propulsion deviation correction control method of a tunneling machine according to claim 2, characterized in that, the performing of continuous update iteration of the initial parameter set, and the completing of multi-round repair optimization, comprising: after completing each round of iteration, adding an optimal fitness parameter in a current parameter set to a historical memory pool; When the number of iterations meets a preset threshold, a history memory pool is activated, and trend analysis of the history memory pool is performed to extract a trend regulation factor; The trend regulation factor is used for optimization constraint of a subsequent iteration process to complete multi-round repair optimization.

4. The propulsion deviation correction control method of a tunneling machine according to claim 1, characterized in that, The geological stratum data is acquired, and initialization of a self-adaptive attitude correction channel is performed according to the geological stratum data, including: The geological stratum data includes stratum type, disturbance resistance and rebound modulus parameters; A parameter template in the self-adaptive attitude correction channel is configured according to the geological stratum data, and the parameter template includes correction weight, adjustment amplitude threshold and path recovery sensitivity.

5. The propulsion deviation correction control method of a tunneling machine as claimed in claim 1, characterized in that, The fitted position attitude data and the path offset trend vector are synchronized to a prediction attitude compensation channel to establish position compensation parameters, including: A key frame extraction sub-channel in the prediction attitude compensation channel is called, key frame extraction of the fitted position attitude data is performed according to the key frame extraction sub-channel to establish a key frame sequence, and key features of the key frame extraction include spatial path turning features and speed change features; After a trend trajectory stream is generated based on the path offset trend vector, alignment mapping of the trend trajectory stream and the key frame sequence is performed; An attitude processing layer in the prediction attitude compensation channel performs offset trend influence analysis of each key frame in the key frame sequence according to the alignment mapping result to establish the position compensation parameters.

6. The propulsion deviation correction control method of a tunneling machine as claimed in claim 1, characterized in that, The dual-domain perception module is activated to perform perception data acquisition and establish a perception data set, including: A real-time domain perception module in the dual-domain perception module is activated, and perception sensors of the real-time domain perception module include IMU sensors, structured light sensors and laser gyro sensors; Data acquisition is performed by using the real-time domain perception module to establish a real-time domain perception data set, and the real-time domain perception data set includes acceleration data, angular velocity data, image point cloud data and turning angle change data; The perception data set is established according to the real-time domain perception data set.

7. The propulsion deviation correction control method of a heading machine according to claim 6, wherein The perception data set is established according to the real-time domain perception data set, and the method further includes: A prediction domain perception module in the dual-domain perception module is activated, and the prediction domain perception module includes a geological radar sensor; Data acquisition is performed by using the prediction domain perception module to establish a prediction domain perception data set; The perception data set is established according to the prediction domain perception data set and the real-time domain perception data set.

8. A computer-readable storage medium having stored thereon a computer program, characterized in that, The program is executed by the processor to implement the tunneling machine propulsion deviation correction control method in any one of claims 1 to 7.

Images