Data-driven Shield Attitude Automatic Control Method and System

Abstract

The present invention discloses a data-driven automatic shield attitude control method and system. The control system includes: a data reading module, a historical engineering data extraction module, a shield tunneling trajectory planning module, a shield tunneling trajectory optimization module, and a shield attitude control module. The data-driven automatic shield attitude control method and system proposed by the present invention can improve the rationality of the calculation results of the trajectory planning model and enhance the accuracy of the control effect of the attitude control model.

Description

Technical Field

[0001] The present invention belongs to the technical field of shield attitude control, and relates to a shield attitude control method, in particular to a data-driven shield attitude automatic control method and system. Background Art

[0002] The control effect of the shield tunneling attitude is an important factor affecting the quality of tunnel construction. In order to achieve safe and efficient tunnel construction, the automation level of the shield machine is continuously improved, and it is hoped that the shield machine can accurately track the tunnel design axis and automatically advance through information technology. However, due to the complexity and uncertainty of the surrounding environment of underground engineering, manual experience is still relied on for control at present. However, the experience of technicians is limited and uneven, the control quality of the shield attitude cannot be guaranteed, and the risk of tunnel construction still exists.

[0003] In view of this, there is an urgent need to design a new shield tunneling attitude control method to overcome at least some of the above defects existing in the existing shield tunneling attitude control methods. Summary of the Invention

[0004] The present invention provides a data-driven shield attitude automatic control method and system, which can solve the technical problems in the existing methods that they cannot cope with the learning of massive construction data, cannot comprehensively consider the influence of soil environment on the deviation correction trajectory and attitude control, cannot comprehensively evaluate the attitude control effect, the control method cannot automatically eliminate the causes of deviation, and is not operable in practice.

[0005] To solve the above technical problems, according to one aspect of the present invention, the following technical solution is adopted:

[0006] A data-driven shield attitude automatic control system, the control system includes:

[0007] A data reading module, used to read the construction parameters and shield operation data in the current construction state of the shield, and transmit the data to the shield tunneling trajectory planning module in a fixed data format;

[0008] A historical project data extraction module, used to extract the formed tunnel data, tunneling data, and shield data of historical projects from a massive historical project construction database respectively, and correspondingly construct a formed tunnel attitude target extractor, an excellent tunneling section selector, and a similar shield classifier according to the extracted data;

[0009] A shield tunneling trajectory planning module, used to obtain a global attitude control target that meets the requirements of the current working condition according to the construction parameters in the current working condition, and plan the shield tunneling trajectory based on the global attitude control target and the current working condition parameters;

[0010] The shield tunneling trajectory optimization module is used to evaluate the planned trajectory output by the shield tunneling trajectory planning module, optimize the planned tunneling trajectory according to the evaluation result of the evaluation function, and determine the local target of shield attitude control;

[0011] The shield attitude control module is used to calculate the set operation parameter values required by the shield propulsion system to reach the attitude control target; the attitude optimization controller is obtained by training based on the similar shield construction data output by the similar shield classifier.

[0012] As an implementation manner of the present invention, the control system further includes: a model retraining module, which is used to store the construction data generated during shield tunneling in the current tunnel project and trigger the retraining mechanisms of the motion trajectory planner and the attitude optimization controller; and store them in the attitude control cache library and the trajectory planning cache library respectively.

[0013] As an implementation manner of the present invention, the model retraining module includes an attitude control cache library, a trajectory planning cache library, and a model retraining trigger;

[0014] The attitude control cache library is used to store the soil layer distribution in the neighborhood of the position of the current construction shield, the shield attitude data, and the shield operation parameters; the shield attitude data, the shield operation parameters, and the corresponding soil layer distribution are stored in the attitude control cache library according to the dimensions required by the attitude optimization controller;

[0015] The trajectory planning cache library is used to store the excellent tunneling trajectory data in the current construction project and the corresponding soil layer distribution; the shield trajectory data and the corresponding soil layer distribution are stored in the trajectory planning cache library according to the dimensions required by the trajectory planner;

[0016] The model retraining trigger is used to judge the set trigger conditions in real time. When the retraining conditions are met, the model of the motion trajectory planner and the attitude controller is updated; this will effectively improve the adaptability of the control system to the current construction environment and effectively eliminate the control error of the control system.

[0017] As an implementation manner of the present invention, the data reading module includes a shield current working condition data packet construction module;

[0018] The shield current working condition data packet construction module is used to construct a shield current working condition data packet; the current working condition data packet includes vectors representing the shield specification parameters, shield performance parameters, shield attitude data, segment attitude data, tail clearance data, design axis data, current shield operation parameters, and the soil layer distribution in the neighborhood of the current position;

[0019] As an implementation manner of the present invention, the historical engineering data extraction module includes a massive historical project construction database, a formed tunnel attitude target extractor, an excellent tunneling section selector module, and a similar shield classifier module;

[0020] The massive historical project construction database is used to store massive historical project construction data;

[0021] The formed tunnel attitude target extractor is used to set the ideal position target after the initial assembly of the formed tunnel and output it to the attitude target setter as a consideration factor for setting the shield attitude control target; the formed tunnel attitude target extractor is used to take the historical engineering formed tunnel attitude data, the tunnel design axis coordinates, and the shield attitude data as the input of the data-driven model and output the ideal spatial coordinates after the initial assembly of the tunnel segments;

[0022] The excellent tunneling section selector is used to select excellent tunneling sections in the historical tunnel project for constructing the motion trajectory planner; the excellent tunneling section selector is used to take the deviation magnitude of the shield attitude, the disturbance of the shield tunneling to the soil body, the smoothness of the shield travel trajectory, and the construction standards required for each working condition or region as the constraint factors considered by the excellent tunneling section selector, and extract the sections that meet each constraint factor as the excellent tunneling sections in the historical tunnel project;

[0023] The similar shield classifier is used to find shield construction data with similar specifications, hydraulic performance, and thrust-shield attitude change relationship under a certain working condition for constructing the attitude optimization controller; the similar shield classifier is used to take the shield specification parameters, performance parameters, the soil layer distribution of the corresponding historical project, and the corresponding thrust-shield attitude change data as the classifier and use a data-driven method to construct the similar shield classifier.

[0024] As an implementation manner of the present invention, the shield tunneling trajectory planning module includes an attitude target setter and a motion trajectory planner;

[0025] The attitude control target setter is used to set the definition of the final ideal control target for the horizontal and elevation of the shield cutterhead and output it to the motion target planner as its planning target; the attitude control target setter is used to take the ideal spatial coordinates after the initial assembly of the tunnel segments, the tunnel design axis data, and the soil layer distribution of the neighborhood corresponding to the position of the shield as the input and the ideal control target of the shield attitude as the output, and use a data-driven method to construct the shield attitude control target setter;

[0026] The motion trajectory planner is used to plan the tunneling trajectory of the shield to reach a certain target in space; the motion trajectory planner is used to establish a motion trajectory planner by using a data-driven method, taking the segment attitude data, tail clearance data, designed axis data, soil layer distribution in the neighborhood of the current position, and the shield attitude in the current state as inputs and the shield attitude data in the next state as outputs, according to the construction data corresponding to excellent tunneling sections.

[0027] As an embodiment of the present invention, the shield tunneling trajectory optimization module includes a trajectory target optimizer;

[0028] The trajectory target optimizer is used to optimize the tunneling trajectory of the shield according to the motion performance of the shield under the current working conditions; the trajectory target optimizer is used to obtain the data after spatial curve fitting between the target attitude and the current shield attitude, evaluate the formed motion trajectory; when the evaluation score result is lower than the set threshold, search and optimize the neighborhood of the recommended target, update the attitude control targets of the shield cutterhead and the tail, and further optimize the tunneling trajectory of the shield to reach a certain space target.

[0029] As an embodiment of the present invention, the shield attitude control module includes an attitude optimization controller;

[0030] The attitude optimization controller is used to construct an attitude optimization controller by using a data-driven method, taking the shield attitude data in the current state, the shield target attitude corresponding to the next state, and the geological data in the neighborhood of the position where the shield is located as inputs and the propulsion operation parameters of the same type of shield as outputs; the attitude optimization controller is used to obtain the attitude control target, the geological conditions in the neighborhood corresponding to the position where the shield is located in the working condition data packet, the current shield operation parameters, the current shield attitude, the shield specification parameters, and the shield performance parameters, and obtain the operation parameters required for the shield to reach the target attitude; transmit the operation parameters into the corresponding memory of the shield machine PLC to execute and realize the automatic propulsion of the shield.

[0031] As an embodiment of the present invention, the trajectory target optimizer considers the smoothness, cutterhead deviation degree, tail deviation degree, attitude angle change amount, and turning curvature of the deviation correction trajectory formed after considering a given local target in its evaluation function. The evaluation function is:

[0032] C total (s) = C smoth (s) + C incision (s) + C tail (s) + C changdir (s) + C curvature (s)

[0033] C total is the cost function value of the trajectory;

[0034] Csmoth For the description of track smoothness; the sum of the squares of the first derivatives and the sum of the squares of the second derivatives at each point of the shield tunneling trajectory in the horizontal plane and the vertical plane;

[0035] C smoth (s)=k1Δβ(s′)+k2v(s″)

[0036] where Δβ is the angle between the tangents of the trajectories in the horizontal plane and the vertical plane and the shield tunneling direction; v is the set speed before shield tunneling, and the higher the tunneling speed, the higher the requirement for track smoothness; C incision (s) is the sum of the squares of the relative distances between the cutterhead trajectory and the designed trajectory at each point; C tail (s) is the sum of the squares of the relative distances between the shield tail trajectory and the designed trajectory at each point; C curvature (s) is the curvature value at each point of the two-plane trajectory.

[0037] A data-driven automatic control method for shield attitude, the control method includes:

[0038] (1) Construct a formed tunnel target extractor and an attitude target setter based on historical engineering formed tunnel data; among them, combine the tunnel design axis data and the shield attitude change trajectory to construct a formed tunnel attitude target extractor for automatically extracting the ideal control targets of the shield attitude in each construction section; according to the ideal control targets of the shield attitude, the tunnel design axis data, and combining the soil layer distribution in the neighborhood corresponding to the position of the shield, use a data-driven method to construct an attitude control target setter;

[0039] (2) Construct an excellent tunneling section selector and a motion trajectory planner based on historical engineering tunneling data; among them, consider the deviation of the shield attitude, the disturbance of the shield tunneling to the soil layer, and the smoothness of the shield traveling trajectory during the historical construction process as the constraint factors for the excellent tunneling section selector; for the tunneling data of the selected excellent tunneling sections, according to the shield attitude data, segment ring attitude data, shield tail clearance data, and design axis data therein, use a data-driven method to establish a motion trajectory planner;

[0040] (3) Construct a similar shield classifier and an attitude optimization controller based on the shield data in historical engineering; among them, use a data-driven method to construct a similar shield classifier according to the shield specification parameters and performance parameters; use a data-driven method to construct an attitude optimization controller according to the propulsion operation parameters of the same type of shield, the shield attitude change data, and the geological conditions in the neighborhood corresponding to the position of the shield;

[0041] (4) Construct a shield current working condition data packet, including vectors representing shield specification parameters, shield performance parameters, shield attitude data, segment ring attitude data, shield tail clearance data, design axis data, current shield operation parameters, and current soil layer distribution;

[0042] (5) Input the soil layer distribution and design axis data of the shield in the current working condition data packet into the attitude control target setter to obtain the control target under the current working condition;

[0043] (6) Input the control target, design axis information, current shield attitude information, segment attitude information, and tail gap information of the shield into the motion trajectory planner to obtain the target attitude for the next construction unit;

[0044] (7) Perform spatial curve fitting between the target attitude and the current shield attitude, input it into the trajectory target optimizer, and score the formed motion trajectory; when the evaluation score result is lower than the set threshold, search and optimize the neighborhood of the recommended target, and calculate the final attitude control target;

[0045] (8) Input the calculated attitude control target, geological conditions of the neighborhood corresponding to the position of the shield in the working condition data packet, current shield operation parameters, current shield attitude, shield specification parameters, and shield performance parameters into the attitude optimization controller to obtain the operation parameters required for the shield to reach the target attitude;

[0046] (9) Transmit the operation parameters into the corresponding memory of the shield machine PLC to execute the automatic propulsion of the shield;

[0047] (10) The actual control effect of the shield attitude and the surrounding construction environment information at the corresponding position are respectively stored in the attitude control cache library and the trajectory planning cache library according to the dimensions required by the attitude controller and the trajectory planner; the model retraining trigger will judge the trigger conditions in real time. When the conditions for retraining are met, the motion trajectory planner and the attitude controller are updated; improve the adaptability of the control system to the current construction environment and eliminate the control error of the control system.

[0048] The beneficial effects of the present invention are as follows: The data-driven shield attitude automatic control method and system proposed by the present invention provide a method framework for shield attitude control applicable to various data-driven technologies for trajectory planning problems and shield attitude control problems. The present invention can effectively utilize a large amount of historical construction data, decompose the shield tunneling attitude automatic control problem into three sub-problems: attitude control target setting, motion trajectory planning, and attitude control. Each unit is connected in series and parallel and integrated, clearly and effectively solving the shield deviation automatic control problem. At the same time, the present invention can quickly train a high-accuracy trajectory planning model and attitude control model, improve the rationality of the calculation results of the trajectory planning model, and improve the accuracy of the control effect of the attitude control model.

[0049] The framework of the present invention includes a trajectory target optimizer, which evaluates and searches for optimization of the results of the trajectory planning model, effectively preventing the data-driven model from falling into local optimization and resulting in unreasonable target postures. In addition, with the application of the shield tunneling machine in a certain project, this framework stores the new data continuously generated by the shield tunneling into the corresponding cache libraries respectively, and online updates the data-driven model under certain conditions. It has the ability of autonomous deep learning and can continuously improve the adaptability of the system to the current project. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 FIG. is a schematic diagram of the composition of the data-driven shield attitude automatic control system according to an embodiment of the present invention.

[0051] Figure 2 FIG. is a schematic diagram of the soil layer distribution around the shield tunneling machine according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0052] The preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0053] In order to further understand the present invention, the preferred implementation schemes of the present invention will be described below in conjunction with embodiments. However, it should be understood that these descriptions are only for further explaining the features and advantages of the present invention, rather than limiting the claims of the present invention.

[0054] The description of this part only focuses on several typical embodiments, and the present invention is not limited to the scope described in the embodiments. The mutual replacement of the same or similar prior art means and some technical features in the embodiments are also within the scope of description and protection of the present invention.

[0055] The expression of the steps in each embodiment in the specification is only for convenience of description, and the implementation manner of the present application is not limited by the order of step implementation. The "connection" in the specification includes both direct connection and indirect connection.

[0056] The present invention discloses a data-driven shield attitude automatic control system Figure 1 FIG. is a schematic diagram of the composition of the data-driven shield attitude automatic control system according to an embodiment of the present invention; please refer to Figure 1 , the control system includes: a data reading module 1, a historical project data extraction module 2, a shield tunneling trajectory planning module 3, a shield tunneling trajectory optimization module 4, and a shield attitude control module 5.

[0057] The data reading module 1 is used to read the construction parameters and shield operation data in the current construction state of the shield tunneling machine, and transmit the data to the shield tunneling trajectory planning module 3 in a fixed data format.

[0058] The historical engineering data extraction module 2 is used to extract historical formed tunnel data, historical tunneling data, and historical shield data from a massive historical project construction database, and respectively construct a formed tunnel attitude target extractor, an excellent tunneling section selector, and a similar shield classifier according to the extracted data.

[0059] The shield tunneling trajectory planning module 3 is used to obtain a global attitude control target that meets the requirements of the current working condition according to the construction parameters under the current working condition, and plan the shield tunneling trajectory based on the global attitude control target and the current working condition parameters.

[0060] The shield tunneling trajectory optimization module 4 is used to evaluate the planned trajectory output by the shield tunneling trajectory planning module, and optimize the planned tunneling trajectory according to the evaluation result of the evaluation function to determine the local target of shield attitude control.

[0061] The shield attitude control module 5 is used to calculate the set operation parameter values required for the shield propulsion system to reach the attitude control target; an attitude optimization controller is trained based on the similar shield construction data output by the similar shield classifier.

[0062] In an embodiment of the present invention, the control system further includes: a model retraining module 6, which is used to store the construction data generated during shield tunneling in the current tunnel project and trigger the retraining mechanisms of the motion trajectory planner and the attitude optimization controller; and store them in the attitude control cache library and the trajectory planning cache library respectively.

[0063] In an embodiment of the present invention, the data reading module 1 includes a shield current working condition data packet construction module 101; the shield current working condition data packet construction module 101 is used to construct a shield current working condition data packet; the current working condition data packet may include vectors representing shield specification parameters, shield performance parameters, shield attitude data, segment attitude data, tail gap data, design axis data, current shield operation parameters, and the soil layer distribution in the neighborhood of the current position. In an embodiment, the shield current working condition data packet construction module 101 is used to obtain the working condition data, various operation parameters, and status data of the shield currently located, integrate them into the data format required by each controller, and sequentially transmit them to each control module.

[0064] In an embodiment of the present invention, the historical engineering data extraction module 2 includes a massive historical project construction database 201, a formed tunnel attitude target extractor 202, an excellent tunneling section selector module 203, and a similar shield classifier module 204.

[0065] The massive historical project construction database 201 is used to store massive historical project construction data.

[0066] The formed tunnel attitude target extractor 202 is used to set the ideal position target after the initial assembly of the formed tunnel and output it to the attitude target setter as a consideration factor for shield attitude control target setting (in one embodiment, the formed tunnel target extractor 202 considers tunnel forming data, tunnel design axis data, and shield attitude change trajectory during the extraction process); the formed tunnel attitude target extractor 202 is used to take the historical project formed tunnel attitude data, tunnel design axis coordinates, and shield attitude data as the input of the data-driven model and output the ideal space coordinates after the initial assembly of the tunnel segments.

[0067] The excellent tunneling section selector 203 is used to select excellent tunneling sections in historical tunnel projects for constructing a motion trajectory planner; the excellent tunneling section selector 203 is used to take the deviation magnitude of the shield attitude during the historical construction process, the disturbance of the shield tunneling to the soil body, the smoothness of the shield travel trajectory, and the construction standards required for each working condition or region as the constraint factors considered by the excellent tunneling section selector, and extract the sections that meet each constraint factor as the excellent tunneling sections in the historical tunnel project.

[0068] The similar shield classifier 204 is used to find shield construction data with similar specifications, hydraulic performance, and thrust-shield attitude change relationships under a certain working condition for constructing an attitude optimization controller; the similar shield classifier 204 is used to take the shield specification parameters, performance parameters, corresponding historical project soil layer distribution, and corresponding thrust-shield attitude change data as those of the classifier, and use a data-driven method to construct the similar shield classifier. In one embodiment, considering the shield specification parameters and performance parameters, a data-driven method is used to construct the similar shield classifier 204.

[0069] In one embodiment of the present invention, the shield tunneling trajectory planning module 3 includes an attitude target setter 301 and a motion trajectory planner 302.

[0070] The attitude control target setter 301 is used to set the definition of the final ideal control target for the horizontal and elevation of the shield cutterhead and output it to the motion target planner as its planning target; in one embodiment, the ideal space coordinates after the initial assembly of the tunnel segments, the tunnel design axis data, and the soil layer distribution in the corresponding neighborhood of the shield position are used as the input, and the ideal control target of the shield attitude is used as the output, and a data-driven method is used to construct the shield attitude control target setter. In one embodiment, considering the ideal control target of the shield attitude, the tunnel design axis data, and combining the geological conditions in the corresponding neighborhood of the shield position, a data-driven method is used to establish the attitude control target setter 301.

[0071] The motion trajectory planner 302 is used to plan the tunneling trajectory of the shield to reach a certain target in space. The motion trajectory planner 302 is used to establish a motion trajectory planner using a data-driven method with the segment attitude data, tail clearance data, design axis data, soil layer distribution in the neighborhood of the current position, and shield attitude in the current state as inputs and the shield attitude data in the next state as outputs according to the construction data corresponding to excellent tunneling sections. In one embodiment, based on the tunneling data of excellent tunneling sections, a motion trajectory planner 302 is established using a data-driven method considering the shield attitude data, segment attitude data, tail clearance data, and design axis data.

[0072] In one embodiment of the present invention, the shield tunneling trajectory optimization module 4 includes a trajectory target optimizer 401; the trajectory target optimizer 401 is used to optimize the tunneling trajectory of the shield according to the motion performance of the shield under the current working conditions. The trajectory target optimizer 401 is used to obtain the data after spatial curve fitting between the target attitude and the current shield attitude, evaluate the formed motion trajectory; when the evaluation score result is lower than the set threshold, search and optimize the neighborhood of the recommended target, update the attitude control targets of the shield cutterhead and the tail, and further optimize the tunneling trajectory of the shield to reach a certain space target.

[0073] In one embodiment, the trajectory target optimizer 401 considers the smoothness, cutterhead deviation degree, tail deviation degree, attitude angle change amount, and turning curvature of the deviation correction trajectory formed after considering the given local target in its evaluation function. The evaluation function is:

[0074] C total (s)=C smoth (s)+C incision (s)+C tail (s)+C changdir (s)+C curvature (s)

[0075] C total is the cost function value of the trajectory; C smoth is the description of the trajectory smoothness; the sum of the squares of the first derivatives and the sum of the squares of the second derivatives at each point of the shield trajectory in the horizontal plane and the vertical plane;

[0076] C smoth (s)=k1Δβ(s′)+k2v(s″)

[0077] where Δβ is the angle between the tangent of the trajectory in the horizontal plane and the vertical plane and the shield tunneling direction; v is the speed setting before shield tunneling, and the higher the tunneling speed, the higher the requirement for the smoothness of the trajectory; C incision (s) is the sum of the squares of the relative distances between the cutterhead trajectory and the design trajectory at each point; C tail(s) is the sum of the squares of the relative distances between the shield tail trajectory and the designed trajectory at each point; C curvature (s) is the curvature value at each point of the two-plane trajectory.

[0078] In an embodiment of the present invention, the shield attitude control module 5 includes an attitude optimization controller 501; the attitude optimization controller 501 is used to take the shield attitude data in the current state, the shield target attitude corresponding to the next state, and the geological data in the neighborhood of the shield's location as inputs, and the propulsion operation parameters of similar shields as outputs, and uses a data-driven method to construct an attitude optimization controller. The attitude optimization controller 501 is used to obtain the attitude control target, the geological conditions in the neighborhood corresponding to the shield's location in the working condition data packet, the current shield operation parameters, the current shield attitude, the shield specification parameters, and the shield performance parameters, and obtain the operation parameters required for the shield to reach the target attitude; transfer the operation parameters into the corresponding memory of the shield machine PLC to execute and realize the automatic propulsion of the shield.

[0079] In an embodiment of the present invention, the model retraining module 6 includes an attitude control cache 601, a trajectory planning cache 602, and a model retraining trigger 603.

[0080] The attitude control cache 601 is used to store the soil layer distribution in the neighborhood of the current construction shield's location, the shield attitude data, and the shield operation parameters; the shield attitude data, the shield operation parameters, and the corresponding soil layer distribution are stored in the attitude control cache according to the dimensions required by the attitude optimization controller. The trajectory planning cache 602 is used to store the excellent tunneling trajectory data in the current construction project and the corresponding soil layer distribution; the shield trajectory data and the corresponding soil layer distribution are stored in the trajectory planning cache according to the dimensions required by the trajectory planner. The model retraining trigger 603 is used to judge the set trigger conditions in real time. When the retraining conditions are met, the motion trajectory planner and the attitude controller are retrained and the model is updated; this will effectively improve the adaptability of the control system to the current construction environment and effectively eliminate the control error of the control system.

[0081] In an embodiment, the attitude control cache 601 stores the shield operation parameter data, the attitude change data, and the surrounding soil layer conditions during the shield attitude adjustment process; the trajectory planning cache 602 stores the shield attitude change data, the segment attitude data, the shield tail clearance, and the surrounding soil layer conditions; the stored data dimensions are kept consistent with the dimensions required by the attitude control model and the trajectory planning model. The model retraining trigger 603 triggers the model to be retrained and the model is updated when the system operation reaches the model retraining condition.

[0082] The present invention also discloses a data-driven shield attitude automatic control method; please refer toFigure 1 , in an embodiment of the present invention, the method for automatically controlling the shield attitude of the present invention includes the following steps:

[0083]

Step S1

[0084] In this specific implementation, the following formula is used to calculate the attitude target of the formed tunnel:

[0085] G i =A i +T i +O i

[0086] Among them, i is the i-th ring; G is the attitude control target in the horizontal or elevation direction; A is the spatial position of the tunnel design axis in the horizontal or elevation direction; T is the deviation of the shield in the horizontal or elevation direction during the tunnel boring stage; O is the offset of the segment in the horizontal or elevation direction after the tunnel is formed.

[0087] Construct an attitude control target setter using a neural network according to the ideal control target of the shield attitude, the tunnel design axis data, and the geological conditions in the neighborhood corresponding to the position of the shield.

[0088] Construct an excellent tunneling section selector and a motion trajectory planner based on historical engineering tunneling data. Among them, the deviation of the shield attitude during the historical construction process, the disturbance of the shield tunneling to the soil body, and the smoothness of the shield travel trajectory are used as the constraint factors considered by the excellent tunneling section selector. For the tunneling data of the selected excellent tunneling sections, a motion trajectory planner is established using a long short-term memory network (LSTM) according to the shield attitude data, segment attitude data, tail clearance data, and design axis data therein.

[0089] Construct a similar shield classifier and an attitude optimization controller based on the shield data in historical engineering. Among them, a data-driven method is used to construct a similar shield classifier according to the shield specification parameters and performance parameters. An attitude optimization controller is constructed using a BP neural network according to the propulsion operation parameters of the same type of shield, the shield attitude change data, and the geological conditions in the neighborhood corresponding to the position of the shield.

[0090] In this specific implementation, the settings for relevant information are as follows:

[0091] Tunnel design axis information It can be the change amount of the spatial coordinates of the design axis per ring;

[0092] The shield attitude can be the horizontal deviation of the cutting edge, the horizontal deviation of the shield tail, the elevation deviation of the cutting edge, the elevation deviation of the shield tail, the slope angle, and the deflection angle;

[0093] Figure 2 It is a schematic diagram of the soil layer distribution around the shield in an embodiment of the present invention; please refer to Figure 2 , in an embodiment of the present invention, the soil layer distribution around the shield is where (m1, m2,... m k ) c are the relevant soil quality parameters of the c-th layer of soil, and the soil quality parameters are soil unit weight, natural water content, internal friction angle, and cohesion. r c represents the proportion of the c-th layer of soil at the current position, and the soil body area range is a cube including the current shield position, as Figure 2 shown.

[0094] Among them, the segment attitude data is used to describe the spatial information of the latest assembled segment, and it is the elevation deviation of the segment center, the horizontal deviation of the segment center, the up and down lead of the segment, and the left and right lead of the segment.

[0095] The shield specification parameters (S1, S2,... S i ) are the shield body length, the jack length, the cutter head diameter, and the shield body shape; the shield performance parameters (P1, P2,... P j ) are the maximum torque of the cutter head, the maximum thrust of the jack, and the minimum turning radius.

[0096] According to the national "Code for Construction and Acceptance of Shield Tunnels (GB 50446-2017)", the deviation ranges of the cutting edge and the shield tail in the excellent tunneling section are set within ±100 mm.

[0097] The disturbance of the shield to the soil is comprehensively represented by the single settlement amount and the cumulative settlement amount of the settlement monitoring points. The calculation method used in this scheme is:

[0098] S cal =(k1S singal +k2S total )×Δα×Δβ

[0099] In the formula, S cal is the disturbance of the shield to the soil, S singal is the single settlement value of the measuring point in the corresponding area of the shield, S total is the cumulative settlement value of the measuring point in the corresponding area of the shield, Δα is the change amount of the pitching angle of the shield in this section, and Δβ is the change amount of the deflection angle of the shield in this section. K1 and k2 can be set according to the key points of construction site control; in an embodiment, it is recommended to set k1 and k2 as follows: k1 = 0.5, k2 = 0.5.

[0100] The smoothness of the shield tunneling trajectory is comprehensively represented according to the smoothness of the cutting head and the tail shield trajectories on the horizontal and vertical planes. The calculation method used in this solution is as follows:

[0101] F cal = k1(f h ″ + f t ″) hor + k2(f h ″ + f t ″) ver

[0102] In the formula, F cal is the smoothness of the shield tunneling trajectory at a certain position, f h is the second derivative of the cutting head trajectory at a certain position, f t is the second derivative of the tail shield trajectory at a certain position, hor is the trajectory on the horizontal plane, and ver is the trajectory on the vertical plane. K1 and k2 can be set according to the key control points at the construction site; in one embodiment, it is recommended to set k1 and k2 as follows: k1 = 0.5, k2 = 0.5.

[0103]

Step S2

[0104] In this implementation, the current shield operation parameters are selected as the opening degrees of the jack oil pressure valves in each zone.

[0105]

Step S3

[0106]

Step S4

[0107]

Step S5

[0108] The trajectory target optimizer, whose evaluation function considers the smoothness of the deviation correction trajectory, the deviation degree of the cutting head, the deviation degree of the tail shield, the change amount of the attitude angle, and the turning curvature after a given attitude target. The trajectory evaluation function is:

[0109] C total (s) = Csmoth (s) + C incision (s) + C tail (s) + C changdir (s) + C curvature (s)

[0110] C total is the cost function value of the trajectory; C smoth is the description of the trajectory smoothness; the sum of the squares of the first derivatives and the sum of the squares of the second derivatives at each point of the shield tunneling trajectory in the horizontal and vertical planes;

[0111] C smoth (s) = k1Δβ(s′) + k2v(s″)

[0112] Δβ is the angle between the tangents of the horizontal and vertical plane trajectories and the shield tunneling direction; v is the set speed given before shield tunneling, and the higher the tunneling speed, the higher the requirement for the smoothness of the trajectory; C incision (s) is the sum of the squares of the relative distances between the cutting edge trajectory and the designed trajectory at each point; C tail (s) is the sum of the squares of the relative distances between the shield tail trajectory and the designed trajectory at each point; C curvature (s) is the curvature value at each point of the two - plane trajectory.

[0113]

Step S6

[0114]

Step S7

[0115]

Step S8

[0116]

Step S9

[0117]

Step S10

[0118] In summary, the data-driven shield attitude automatic control method and system proposed by the present invention provide a method framework for shield attitude control applicable to various data-driven technologies for the trajectory planning problem and the shield attitude control problem. The present invention can effectively utilize a large amount of historical construction data, decompose the shield tunneling attitude automatic control problem into three sub-problems: attitude control target setting, motion trajectory planning, and attitude control. Each unit is connected in series and parallel for fusion, clearly and effectively solving the shield deviation automatic control problem. At the same time, the present invention can quickly train a high-accuracy trajectory planning model and an attitude control model, improve the rationality of the calculation results of the trajectory planning model, and improve the accuracy of the control effect of the attitude control model.

[0119] The framework of the present invention includes a trajectory target optimizer, which evaluates and searches for optimization of the results of the trajectory planning model, effectively preventing the data-driven model from falling into local optimization and resulting in unreasonable target attitudes. In addition, with the application of the shield in a certain project, this framework stores the new data continuously generated during the shield propulsion into the corresponding cache libraries respectively, and updates the data-driven model online under certain conditions, having the ability of autonomous deep learning, and can continuously improve the adaptability of the system to the current project.

[0120] It should be noted that the present application can be implemented in software and / or a combination of software and hardware; for example, it can be implemented using an application-specific integrated circuit (ASIC), a general-purpose computer, or any other similar hardware device. In some embodiments, the software program of the present application can be executed by a processor to implement the above steps or functions. Similarly, the software program of the present application (including related data structures) can be stored in a computer-readable recording medium; for example, a RAM memory, a magnetic or optical drive, or a floppy disk and the like. In addition, some steps or functions of the present application can be implemented using hardware; for example, a circuit that cooperates with the processor to execute each step or function.

[0121] The technical features of the above-described embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered to be within the scope described in this specification.

[0122] The description and application of the present invention herein are illustrative and are not intended to limit the scope of the present invention to the above embodiments. The effects or advantages involved in the embodiments may not be reflected in the embodiments due to various interferences, and the description of the effects or advantages is not used to limit the embodiments. Modifications and changes to the disclosed embodiments are possible, and various components of substitution and equivalence of the embodiments are known to those of ordinary skill in the art. Those skilled in the art should clearly understand that the present invention can be implemented in other forms, structures, arrangements, proportions, and with other components, materials, and parts without departing from the spirit or essential characteristics of the present invention. Other modifications and changes can be made to the disclosed embodiments without departing from the scope and spirit of the present invention.

Claims

1. A data-driven automatic shield attitude control system, characterized in that The control system includes: A data reading module, which is used to read the construction parameters and shield operation data under the current working conditions of the shield, and transmit the data to the shield tunneling trajectory planning module in a fixed data format; A historical project data extraction module, which is used to extract historical project formed tunnel data, historical project tunneling data, and historical project shield data from the massive historical project construction database respectively, and correspondingly construct a formed tunnel attitude target extractor, an excellent tunneling section selector, and a similar shield classifier according to the extracted data; A shield tunneling trajectory planning module, which is used to obtain the overall attitude control target that meets the requirements of the current working conditions according to the construction parameters under the current working conditions, and plan the shield tunneling trajectory based on the overall attitude control target and the construction parameters under the current working conditions; A shield tunneling trajectory optimization module, which is used to evaluate the planned trajectory output by the shield tunneling trajectory planning module, and optimize the planned tunneling trajectory according to the evaluation result of the evaluation function to determine the local target of shield attitude control; A shield attitude control module, which is used to calculate the set operation parameter values required for the shield propulsion system to reach the attitude control target; and train an attitude optimization controller based on the similar shield construction data output by the similar shield classifier; The shield tunneling trajectory planning module includes an attitude control target setter and a motion trajectory planner; The attitude control target setter is used to set the definition of the final ideal control target for the horizontal and elevation of the shield cutterhead, and output it to the motion trajectory planner as the target for its planning; the attitude control target setter is used to use the ideal space coordinates after the initial assembly of the tunnel segments, the tunnel design axis data, and the soil layer distribution in the corresponding neighborhood of the shield's location as inputs, and the ideal control target of the shield attitude as the output, and construct the shield attitude control target setter using a data-driven method; The motion trajectory planner is used to plan the tunneling trajectory for the shield to reach a certain target in space; the motion trajectory planner is used to use the construction data corresponding to the excellent tunneling section, and use the segment attitude data, the tail gap data, the design axis data, the soil layer distribution in the current position neighborhood, and the shield attitude under the current state as inputs, and the shield attitude data in the next state as the output, and establish the motion trajectory planner using a data-driven method.

2. The data-driven automatic shield attitude control system according to claim 1, wherein: The control system further includes: a model retraining module, which is used to store the construction data generated during the shield tunneling in the current tunnel project and trigger the retraining mechanism of the motion trajectory planner and the attitude optimization controller; and store them in the attitude control cache library and the trajectory planning cache library respectively.

3. The data-driven automatic shield attitude control system according to claim 2, wherein: The model retraining module includes an attitude control cache library, a trajectory planning cache library, and a model retraining trigger; The attitude control cache library is used to store the soil layer distribution in the neighborhood of the current position of the shield tunneling machine under construction, the shield attitude data, and the shield operation parameters; the shield attitude data, the shield operation parameters, and the corresponding soil layer distribution are stored in the attitude control cache library according to the dimensions required by the attitude optimization controller; The trajectory planning cache library is used to store the excellent tunneling trajectory data in the current construction project and the corresponding soil layer distribution; the shield trajectory data and the corresponding soil layer distribution are stored in the trajectory planning cache library according to the dimensions required by the motion trajectory planner; The model retraining trigger is used to judge the set trigger conditions in real time. When the retraining conditions are met, the motion trajectory planner and the attitude control target setter are updated; this will effectively improve the adaptability of the control system to the current construction environment and effectively eliminate the control error of the control system.

4. The data-driven shield attitude automatic control system according to claim 1, wherein: The data reading module includes a shield current working condition data packet construction module; The shield current working condition data packet construction module is used to construct a shield current working condition data packet; the current working condition data packet includes vectors representing shield specification parameters, shield performance parameters, shield attitude data, segment attitude data, tail clearance data, design axis data, current shield operation parameters, and the soil layer distribution in the neighborhood of the current position.

5. The data-driven shield attitude automatic control system according to claim 1, wherein: The historical project data extraction module includes a massive historical project construction database, a formed tunnel attitude target extractor, an excellent tunneling section selector, and a similar shield classifier; The massive historical project construction database is used to store massive historical project construction data; The formed tunnel attitude target extractor is used to set the ideal position target after the initial assembly of the formed tunnel and output it to the attitude control target setter as a consideration factor for shield attitude control target setting; the formed tunnel attitude target extractor is used to use the historical project formed tunnel attitude data, the tunnel design axis coordinates, and the shield attitude data as the input of the data-driven model and output the ideal spatial coordinates after the initial assembly of the tunnel segments; The excellent tunneling section selector is used to select excellent tunneling sections in historical tunnel projects for constructing the motion trajectory planner; the excellent tunneling section selector is used to use the deviation of the shield attitude, the disturbance of the shield tunneling to the soil body, the smoothness of the shield travel trajectory, and the construction standards required for each working condition or region as the constraint factors considered by the excellent tunneling section selector, and extract the sections that meet each constraint factor as the excellent tunneling sections in historical tunnel projects; The similar shield classifier is used to find shield construction data with similar specifications, hydraulic performance, and thrust-shield attitude change relationships under a certain working condition, and is used to construct an attitude optimization controller. The similar shield classifier uses shield specification parameters, performance parameters, the corresponding historical engineering soil layer distribution, and the corresponding thrust-shield attitude change data as the input data of the classifier, and constructs a similar shield classifier using a data-driven method.

6. The data-driven shield attitude automatic control system according to claim 1, wherein: The shield tunneling trajectory optimization module includes a trajectory target optimizer; The trajectory target optimizer is used to optimize the tunneling trajectory of the shield according to the motion performance of the shield under the current working condition. The trajectory target optimizer is used to obtain the data after spatial curve fitting between the target attitude and the current shield attitude, and evaluate the formed motion trajectory; When the evaluation score result is lower than the set threshold, it searches and optimizes the neighborhood of the recommended target, updates the attitude control targets of the shield cutterhead and the shield tail, and further optimizes the tunneling trajectory of the shield to reach a certain spatial target.

7. The data-driven shield attitude automatic control system according to claim 1, wherein: The shield attitude control module includes an attitude optimization controller; The attitude optimization controller uses the shield attitude data in the current state, the corresponding shield target attitude in the next state, and the geological data in the neighborhood of the shield's location as inputs, and the shield's propulsion operation parameters as outputs, and constructs an attitude optimization controller using a data-driven method. The attitude optimization controller is used to obtain the geological conditions in the neighborhood corresponding to the shield's location in the attitude control target working condition data packet, the current shield operation parameters, the current shield attitude, the shield specification parameters, and the shield performance parameters, and obtain the operation parameters required for the shield to reach the target attitude; transmit the operation parameters into the corresponding memory of the shield machine PLC to execute and realize the automatic propulsion of the shield.

8. A data-driven automatic shield attitude control method, characterized in that The control method includes: (1) Construct a formed tunnel attitude target extractor and an attitude control target setter based on the historical engineering formed tunnel data. Among them, combine the tunnel design axis data and the shield attitude change trajectory to construct a formed tunnel attitude target extractor for automatically extracting the ideal control targets of the shield attitude in each construction section. According to the ideal control targets of the shield attitude, the tunnel design axis data, and the soil layer distribution in the neighborhood corresponding to the shield's location, use a data-driven method to construct an attitude control target setter; (2) Construct an excellent tunneling section selector and a motion trajectory planner based on the historical engineering tunneling data. Among them, the deviation of the shield attitude, the disturbance of the shield tunneling to the soil layer, and the smoothness of the shield driving trajectory during the historical construction process are used as the constraint factors considered by the excellent tunneling section selector. For the tunneling data of the selected excellent tunneling sections, according to the shield attitude data, segment ring attitude data, shield tail gap data, and design axis data among them, use a data-driven method to establish a motion trajectory planner; (3) Construct a similar shield tunneling machine classifier and an attitude optimization controller based on the shield tunneling machine data in historical projects; among them, construct a similar shield tunneling machine classifier using a data-driven method according to the shield tunneling machine specification parameters and performance parameters; construct an attitude optimization controller using a data-driven method according to the propulsion operation parameters of similar shield tunneling machines, the shield tunneling machine attitude change data, and the geological conditions of the neighborhood corresponding to the location of the shield tunneling machine. (4) Construct a shield tunneling machine current working condition data packet, including vectors representing the shield tunneling machine specification parameters, shield tunneling machine performance parameters, shield tunneling machine attitude data, segment attitude data, tail clearance data, design axis data, current shield tunneling machine operation parameters, and current soil layer distribution. (5) Input the soil layer distribution and design axis data of the shield tunneling machine in the current working condition data packet into the attitude control target setter to obtain the control target under the current working condition. (6) Input the control target, design axis information, current shield tunneling machine attitude information, segment attitude information, and tail clearance information into the motion trajectory planner to obtain the target attitude in the next construction unit. (7) Perform a spatial curve fitting between the target attitude and the current shield tunneling machine attitude, input it into the trajectory target optimizer, and score the formed motion trajectory; when the evaluation score result is lower than the set threshold, search and optimize the neighborhood of the recommended target, and calculate the final attitude control target. (8) Input the calculated attitude control target, the geological conditions of the neighborhood corresponding to the location of the shield tunneling machine in the working condition data packet, the current shield tunneling machine operation parameters, the current shield tunneling machine attitude, the shield tunneling machine specification parameters, and the shield tunneling machine performance parameters into the attitude optimization controller to obtain the operation parameters required for the shield tunneling machine to reach the target attitude. (9) Transmit the operation parameters to the corresponding memory of the shield tunneling machine PLC to execute the automatic propulsion of the shield tunneling machine. (10) The actual control effect of the shield tunneling machine attitude and the surrounding construction environment information at the corresponding location are respectively stored in the attitude control cache library and the trajectory planning cache library according to the required dimensions of the attitude control target setter and the motion trajectory planner; the model retraining trigger will continuously judge the trigger conditions, and when the retraining conditions are met, update the models of the motion trajectory planner and the attitude control target setter; improve the adaptability of the control system to the current construction environment and eliminate the control error of the control system.

Images