An algorithm for precise adjustment of track panel for double-block ballastless track

Abstract

The present invention discloses a fine adjustment algorithm for a track panel of a double-block ballastless track, belonging to the technical field of track panel fine adjustment, and comprising the steps: S1: establishing a physical model and a virtual adjustment system for a pair of legs of the track panel; S2: establishing a motion planner model; S3: randomly initializing the state; S4: continuously outputting new actions through a reinforcement learning algorithm; S5: repeating S4 until convergence; S6: obtaining the initial state and the real-time state of the inclination angle of the cross beam of the on-site control system and the deflection angle of the lead screw through an inclination sensor; S7: calculating the moving distance of each connecting rod on the straight line where the lead screw is located, converting the rail kinematic data into the kinematic data of each lead screw through a lead screw motion planner, and controlling the track panel to complete the adjustment; S8: repeating S6 to S7 until the rail reaches the allowable error range of the target point. The present invention can perform accurate super-elevation adjustment and horizontal adjustment of the track panel according to the rail parameters, with simple calculation and few steps.

Description

Technical Field

[0001] The invention belongs to the technical field of laying double-block ballastless tracks, and specifically relates to a track-laying fine-tuning algorithm for double-block ballastless tracks. Background Technique

[0002] The double-block ballastless track structure has currently become the main track structure for high-speed railways. The precision adjustment of the track-laying during the laying of high-speed rails will directly affect the operation safety of high-speed trains.

[0003] In the field of railway construction, multiple technologies have been applied to solve the problem of efficient adjustment of track-laying positioning. For example, a driving device is introduced during the track-laying adjustment process to achieve the adjustment of the track-laying height. This technology achieves precise control of the track-laying height by adjusting the joint changes of the screw rod. However, this method has limitations in the adjustment direction and lacks flexibility, which to a certain extent restricts its application scope and efficiency. In addition, there is also a method that realizes the adjustment of the track-laying position in multiple directions by using motors for adjustment item by item and comparing positions. This technology can theoretically provide higher flexibility and a wider adjustment range, but in the actual operation process, multiple steps of adjustment are required, which not only increases the complexity of the operation but also leads to an extension of the construction period, thus affecting the overall construction efficiency. Summary of the Invention

[0004] Aiming at the above problems existing in the prior art, the purpose of the present invention is to provide a track-laying fine-tuning algorithm for double-block ballastless tracks, which can perform precise super-elevation adjustment and horizontal adjustment of the track-laying according to the rail parameters, meeting the requirements of precise adjustment of the track-laying for double-block ballastless tracks. The calculation is simple, the steps are few, reducing the complexity of track-laying adjustment and improving the construction efficiency of track-laying. By introducing a reinforcement learning algorithm, it can adapt to different construction environment changes during track-laying, reduce the on-site commissioning cost, and improve the laying precision and efficiency.

[0005] In order to achieve the above purpose, the technical solution adopted by the present invention is as follows:

[0006] A track-laying fine-tuning algorithm for double-block ballastless tracks includes the following steps:

[0007] S1: Establish a physical model of a pair of legs of the track-laying, and construct a virtual adjustment system based on the physical model. Among them, the physical model of the legs includes a cross beam and two rails. The two rails are arranged in parallel and supported on the cross beam. The virtual adjustment system includes four connecting rods and four screw rods. Each end of the cross beam is connected to two connecting rods. The four screw rods respectively drive the linear reciprocating motion of one end of the four connecting rods, and the four connecting rods drive the cross beam hinged to the other end thereof to rotate and / or move;

[0008] S2: Establish a leg screw motion planner model, where the planner model includes the motion target position and motion speed;

[0009] S3: Randomly initialize the starting position state of each connecting rod on the screw and the starting deflection state of the connecting rod, and randomize the environmental observation data;

[0010] S4: Through the reinforcement learning algorithm, select the rail motion trajectory according to the environmental observation data and reward prediction. Use the screw motion planner to convert the rail kinematic data into motor control instructions for controlling the screw motion, drive the leg adjustment, and update the reward data based on the new environmental observation data using the reward function, and continue to output new actions;

[0011] S5: Repeat S4, iteratively update the policy learning and value estimation network and store the experience in the experience pool until convergence, end the training and save the policy as the rail motion trajectory;

[0012] S6: Obtain the initial state and real-time state of the crossbeam tilt angle and screw deflection angle of the on-site control system through the tilt sensor;

[0013] S7: Calculate the moving distance of each connecting rod on the straight line where the screw is located, convert the rail kinematic data into the kinematic data of each screw through the screw motion planner, and control the track panel to complete the adjustment;

[0014] S8: Repeat S6 to S7 until the error range allowed for the rail to reach the target point.

[0015] As a further improvement of the above technical solution.

[0016] In step S1, the four connecting rods are the right first connecting rod, the right second connecting rod, the left first connecting rod, and the left second connecting rod respectively. The four screws are arranged in parallel at intervals, and are the left first screw, the left second screw, the right second screw, and the right first screw arranged in sequence in one direction. The left first connecting rod, the left second connecting rod, the right first connecting rod, and the right second connecting rod are respectively connected to the left first screw, the left second screw, the right first screw, and the right second screw. The upper end of the right first connecting rod is connected to the right first screw and the lower end is connected to the crossbeam. The lower end of the right second connecting rod is connected to the right second screw and the upper end is connected to the crossbeam. The upper end of the left first connecting rod is connected to the left first screw and the lower end is connected to the crossbeam. The lower end of the left second connecting rod is connected to the left second screw and the upper end is connected to the crossbeam. There is a center line. The left first screw and the right first screw are symmetric with respect to this center line, the left second screw and the right second screw are also symmetric with respect to this center line, the left first connecting rod and the right first connecting rod are symmetric with respect to this center line, and the left second connecting rod and the right second connecting rod are also symmetric with respect to this center line.

[0017] The environmental observation data during the adjustment of the track panel includes: sensor parameters, device status, and target parameters. Among them, the sensor parameters include: the distance L1 between the left first screw rod and the right first screw rod, the distance L2 between the left second screw rod and the right second screw rod, the distance L3 between the two rails, the height difference between the two rails, the offset distance, and the tilt angle. Among them, the offset distance refers to the distance that the center point between the two rails deviates from the center line, and the tilt angle refers to the angle between the cross beam and the horizontal plane; the device status includes: the displacement and speed of each connecting rod on the straight line where the corresponding screw rod is located, and the deflection angle of the connecting rod; the target parameters include: the target superelevation value and the target offset distance, where the superelevation value refers to the height difference between the two rails.

[0018] Step S4 includes the following steps:

[0019] S4-1: Initialization state: Obtain the initial state feature vector of the track panel, including sensor parameters, device status, and target parameters;

[0020] S4-2: Action generation: Generate an action according to the current state through the current network policy function in policy learning, where the action includes the target position of the rail and the motion speed instruction;

[0021] S4-3: Action execution and observation: Send the action to the screw rod motion planner, convert it into a motor control execution instruction, and obtain a new state;

[0022] S4-4: Experience storage: Store the five-tuple composed of the current state, action, immediate reward, new state, and termination flag into the experience replay pool, where the immediate reward is calculated by the reward function, reflecting the adjustment accuracy, and the termination flag is used to judge whether the target error range is reached;

[0023] S4-5: Network training: Sample multiple samples from the experience replay pool;

[0024] S4-6: Target network synchronization: Copy the current network parameters of value estimation or policy learning to the target network every C steps, where C represents the update frequency of the target network parameters;

[0025] S4-7: Loop termination condition: If the new state is a termination state, that is, the track panel reaches the target accuracy, end the current training round; otherwise, return to step S4-2 to continue iteration.

[0026] Step S7 includes the following steps:

[0027] S7-1: Perform superelevation adjustment, and the superelevation value changes after adjustment;

[0028] S7-2: Perform horizontal adjustment with superelevation, and the center point between the two rails deviates after adjustment.

[0029] In step S7-1, calculate the tilt angle that the crossbeam needs to be adjusted according to the superelevation value to be adjusted and the distance between the two rails, and calculate the displacements that each connecting rod needs to complete on the corresponding lead screw according to the relationship between the linear movement speeds of the two connecting rods on the same side of the center line.

[0030] Step S7-2 includes the following steps:

[0031] S7-2-1: Calculate the deflection angles of each connecting rod after superelevation adjustment;

[0032] S7-2-2: Calculate the displacements that each connecting rod needs to complete on the corresponding lead screw according to the target deflection angles, target offset distances of each connecting rod, and the deflection angles obtained in step S7-2-1.

[0033] In step S7-1:

[0034] ;

[0035] ;

[0036] ;

[0037] ;

[0038] ;

[0039] ;

[0040] Where: A is the superelevation value to be adjusted, is the tilt angle of the crossbeam to be adjusted, is the movement speed of the left first lead screw, is the movement speed of the right first lead screw, is the movement speed of the left second lead screw, is the movement speed of the right second lead screw, is the displacement that the left first lead screw and the right first lead screw need to move, is the displacement that the left second lead screw and the right second lead screw need to move, and T is the time for superelevation adjustment.

[0041] In step S7-2-2:

[0042] ;

[0043] ;

[0044] Where i = 1, 2, 3, 4, represents the length of each connecting rod, represents the displacement of each connecting rod on the corresponding lead screw, Indicates the deflection angles of each connecting rod relative to the initial state after the horizontal adjustment with superelevation is completed. Indicates the offset distance of the center point between the two rails. Indicates the deflection angle relative to the state after superelevation adjustment after the horizontal adjustment with superelevation.

[0045] The beneficial effects of the present invention are as follows:

[0046] (1) It can accurately adjust the superelevation and horizontal of the track panel according to the rail parameters, meeting the precise adjustment requirements of the double-block ballastless track panel.

[0047] (2) The calculation is simple and the steps are few, reducing the complexity of track panel adjustment and improving the construction efficiency of track panel laying.

[0048] (3) By introducing the reinforcement learning algorithm, it can adapt to the changes in different construction environments of track panel laying, reduce the on-site commissioning cost, and improve the laying accuracy and efficiency. Brief Description of the Drawings

[0049] Figure 1 is the flow chart of the track panel precise adjustment algorithm of the present invention.

[0050] Figure 2 is the schematic diagram of the mechanism before track panel adjustment.

[0051] Figure 3 is the schematic diagram of the mechanism after superelevation adjustment of the track panel.

[0052] Figure 4 is the schematic diagram of the mechanism after horizontal adjustment with superelevation of the track panel.

[0053] Reference numerals: 1, cross beam; 2, rail; ab, right first connecting rod; cd, right second connecting rod; ef, left first connecting rod; gh, left second connecting rod; 11, left lead screw; 12, left second lead screw; 21, right first lead screw; 22, right second lead screw. Detailed Description of the Preferred Embodiments

[0054] The following will describe in detail the specific embodiments of the present invention with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only for the purpose of illustration and explanation of the present invention, and are not intended to limit the present invention.

[0055] For ease of description, spatial relative terms, such as "above", "over", "on the upper surface", "upper", etc., may be used herein to describe the spatial positional relationship of one device or feature to other devices or features as shown in the figures. It should be understood that the spatial relative terms are intended to encompass different orientations in use or operation in addition to the orientation depicted in the figures. For example, if the device in the figures is inverted, a device described as "above" or "over" other devices or structures will then be positioned "below" or "under" the other devices or structures. Thus, the exemplary term "above" can include both orientations of "above" and "below". The device may also be positioned in other different ways (rotated 90 degrees or in other orientations), and corresponding interpretations of the spatial relative descriptions used herein will be made.

[0056] An alignment and adjustment algorithm for a track panel of a double-block ballastless track, as Figure 1 shown, includes the following steps:

[0057] S1: Establish a physical model of a pair of legs of the track panel, simplify the structural analysis, and construct a virtual adjustment system based on the physical model to simulate the influence of the movement of the lead screws (see the description below) on the track parameters; wherein, the legs are the unit structures that make up the track panel.

[0058] The physical model of the legs includes a cross beam 1 and two rails 2. The two rails 2 are arranged in parallel and supported on the cross beam 1, and the length of the rail 2 is substantially perpendicular to the length of the cross beam 1.

[0059] The virtual adjustment system includes four connecting rods and four lead screws. The four connecting rods are respectively connected to the four lead screws. One end of each connecting rod is hinged to the cross beam 1, and the other end is connected to the lead screw and can reciprocate on the straight line where the corresponding lead screw is located. Specifically, the lead screw can have two movement modes: One, the lead screw can only reciprocate along its own straight line to drive the synchronous movement of one end of the connecting rod connected thereto, and the motor drives the lead screw to move linearly. Two, the lead screw itself can only rotate, and one end of the connecting rod is screwed onto the lead screw. When the lead screw rotates, one end of the connecting rod connected thereto moves linearly along the length direction of the lead screw, and the motor drives the lead screw to rotate. For ease of description, hereinafter, the displacement and speed of one end of the connecting rod connected to the lead screw moving on the straight line where the lead screw is located are referred to as the displacement and speed of the lead screw.

[0060] The four connecting rods are respectively the right first connecting rod ab, the right second connecting rod cd, the left first connecting rod ef, and the left second connecting rod gh. The four lead screws are arranged in parallel at intervals, and are respectively the left first lead screw 11, the left second lead screw 12, the right second lead screw 22, and the right first lead screw 21 arranged in sequence in one direction. There is a center line, and the left first lead screw 11 and the right first lead screw 21 are symmetric with respect to this center line, and the left second lead screw 12 and the right second lead screw 22 are also symmetric with respect to this center line.

[0061] The two end points of the right first connecting rod ab are point a and point b respectively. Among them, point b is hinged at one end of the cross beam 1, and point a is hinged on the right first lead screw 21, so that the right first connecting rod ab can move up and down and can deflect relative to the right first lead screw 21, and point b is located below point a.

[0062] The two end points of the right second connecting rod cd are point c and point d respectively. Among them, point d is hinged at one end of the cross beam 1, and point c is hinged on the right second lead screw 22, so that the right second connecting rod cd can move up and down and can deflect relative to the right second lead screw 22, and point c is located below point d.

[0063] The two end points of the left first connecting rod ef are point e and point f respectively. Among them, point f is hinged at one end of the cross beam 1, and point e is hinged on the left first lead screw 11, so that the left first connecting rod ef can move up and down and can deflect relative to the left first lead screw 11, and point f is located below point e.

[0064] The two end points of the left second connecting rod gh are point g and point h respectively. Among them, point g is hinged at one end of the cross beam 1, and point h is hinged on the left second lead screw 12, so that the left second connecting rod gh can move up and down and can deflect relative to the left second lead screw 12, and point h is located below point g.

[0065] The left end of the cross beam 1 is connected to point f of the left first connecting rod ef and point g of the left second connecting rod gh, and the right end is connected to point d of the right second connecting rod cd and point b of the right first connecting rod ab. Points f, g, d, and b are on a straight line, and this straight line is parallel to the length direction of the cross beam 1. When the length direction of the cross beam 1 is perpendicular to the length directions of the respective lead screws, the straight lines where the right first connecting rod ab, the right second connecting rod cd, the left first connecting rod ef, and the left second connecting rod gh are located are respectively parallel to the right first lead screw 21, the right second lead screw 22, the left first lead screw 11, and the left second lead screw 12. The distance from point f to the center line is equal to the distance from point b to the center line, and the distance from point g to the center line is equal to the distance from point d to the center line.

[0066] Based on the above structure, each lead screw can drive the corresponding connecting rod to move, and the connecting rod drives the cross beam to tilt and / or move horizontally.

[0067] The physical model of the support leg and the virtual adjustment system constitute the mechanism of this solution.

[0068] S2: Establish a motion planner model for the support leg lead screw, including the target position and motion speed of the lead screw.

[0069] The models established in steps S1 and S2 are based on the model-based reinforcement learning method, using the existing models for learning and decision-making, and modeling the rail alignment adjustment process as a Markov decision process (MDP).

[0070] During the adjustment process of the track panel, the state space S, which is the environmental observation data, includes: sensor parameters, device status, and target parameters, where:

[0071] The sensor parameters include: the distance L1 between the left first screw rod 11 and the right first screw rod 21, the distance L2 between the left second screw rod 12 and the right second screw rod 22, the distance L3 between the two rails 2, the height difference A between the two rails 2, the offset distance ε (the distance that the center point between the two rails 2 deviates from the center line), and the tilt angle θ (the angle between the cross beam 1 and the horizontal plane);

[0072] The device status includes: the screw rod displacements s1, s2, the speed , , and the deflection angle of the connecting rod ~ ;

[0073] The target parameters include: the target superelevation value , the target offset .

[0074] S3: Randomly initialize the starting position state of the screw rod and the starting deflection state of the connecting rod, and randomize the environmental observation data. By introducing a parameter randomization strategy, enhance the generalization ability of the reinforcement learning model to the real construction environment, and effectively solve the problem of the gap between the simulation environment and the engineering reality.

[0075] S4: Through the reinforcement learning (RL) algorithm, select the rail movement trajectory according to the environmental observation data and the reward prediction. Use the screw rod motion planner to convert the rail kinematic data into motor control instructions to drive the leg adjustment, and update the reward data based on the new environmental observation data using the reward function, and continue to continuously output new actions.

[0076] Among them, the reinforcement learning algorithm specifically uses the Deep Deterministic Policy Gradient (DDPG) algorithm as the benchmark algorithm, and uses the experience replay mechanism. During the interaction process between the agent and the environment, store information such as the state, action, reward, and next state experienced in a replay buffer, and then randomly sample a small batch of data from the experience pool for sampling training.

[0077] In step S4, adopt the Actor-Critic architecture, separate the policy learning (Actor) and value estimation (Critic), and use deep neural networks to approximate the policy function and the value function. The specific steps are as follows:

[0078] S4-1: Initialize the state: Obtain the initial state feature vector of the track panel , including sensor parameters: , the screw rod state: displacement , speed , , connecting rod deflection angle , target parameter , .

[0079] S4-2: Action generation: Generate an action through the current network policy function in policy learning and according to the state to generate an action , where the action includes the target position of the rail and the motion speed command.

[0080] S4-3: Action execution and observation: Send the action to the lead screw motion planner, convert it into a motor control execution instruction, and obtain: the new state , that is, the updated track panel parameters, the immediate reward , where is calculated by the reward function, reflecting the adjustment accuracy, the termination flag is_end, and is_end is used to judge whether the target error range is reached.

[0081] S4-4: Experience storage: Store the five-tuple ) into the experience replay pool .

[0082] S4-5: Network training: Sample N samples from , where j represents the index of the j-th sample randomly sampled from the experience replay pool. Update according to the following rules:

[0083] Critic network update:

[0084] Calculate the target value :

[0085]

[0086] where Q′ is the Critic target network, ω′ is the parameter of the Critic target network, is the discount factor, generally taking 0.9~0.99, represents the next state of the j-th sample, represents the feature extraction of the next state , is the policy function of the Actor target network.

[0087] Update the current parameters of the Critic by minimizing the mean square error loss . represents the current state of the j-th sample, ​Indicates the feature extraction of the current state Q is the current network

[0088] Actor network update:

[0089] Using policy gradient Update the current network parameters of the Actor where represents the gradient of the policy parameter and represents the gradient of the action a represents the policy gradient is the network policy function generated according to the state

[0090] S4-6: Target network synchronization: Every C steps, copy the current network parameters of the Critic or Actor to the target network (where C represents the update frequency of the target network parameters)

[0091] S4-7: Loop termination condition: If is the termination state, that is, the track panel reaches the target accuracy, end the current training round; otherwise, return to step S4-2 to continue the iteration

[0092] S5: Repeat S4, iteratively update the Actor and Critic networks and store the experiences in the experience pool until convergence, end the training and save the policy as the rail movement trajectory

[0093] S6, obtain the initial state and real-time state of the inclination angle θ of the crossbeam 1 of the on-site control system and the deflection angle of the lead screw ~ through the inclination sensor, and obtain the parameters required for adjusting the elevation and horizontal from the existing state to the final state of the rail through the track inspection trolley. Including the required offset distance of the rail center and Figure 2 the distance L1 between the left first lead screw 11 and the right first lead screw 21 in

[0094] ; the distance L2 between the left second lead screw 12 and the right second lead screw 22; the distance L3 between the vertices of the two rails 2; the height difference between the left and right rails 2 of the track, that is, the superelevation value of the rail. Use the above data obtained by the track inspection trolley and the inclination sensor as the environmental data of the on-site control system, load the pre-trained model in S5 into the on-site control system, update the experience pool according to the environmental data and optimize the policy online, save the optimal policy, and use the saved policy to select the rail movement trajectory​

[0095] Among them, the horizontal adjustment with superelevation refers to the horizontal adjustment because the superelevation value also changes during the horizontal adjustment process.

[0096] In step S7, controlling the track panel to complete the superelevation adjustment and the horizontal adjustment with superelevation includes the following steps:

[0097] S7-1: Calculate the moving distances of the left and right lead screws to complete the superelevation adjustment, and the superelevation value changes after the adjustment:

[0098] In this step, in this example, it is assumed that Figure 2 is the initial state. At this time, the crossbeam 1 is parallel to the horizontal plane, and the superelevation value is 0. The target state is Figure 3 , at this time, the left rail 2 is higher than the right rail 2, and the superelevation value is A, that is, the superelevation value that needs to be adjusted for the two rails 2 is A mm. It is necessary to adjust the four lead screws of the track panel to achieve the superelevation adjustment. First, calculate the inclination angle that the crossbeam 1 needs to be adjusted:

[0099] ;

[0100] Among them, A is the superelevation value of the left and right rails, is the distance between the vertices of the two rails.

[0101] When the track panel makes a central deflection around the center line, the speed planning needs to meet the following regulations:

[0102] ;

[0103] Among them is the moving speed of the left first lead screw 11, is the moving speed of the left second lead screw 12, is the moving speed of the right first lead screw 21, is the moving speed of the right second lead screw 22, is the distance between the left first lead screw 11 and the right first lead screw 21, and L2 is the distance between the left second lead screw 12 and the right second lead screw 22.

[0104] The displacements that the left first lead screw 11 and the right first lead screw 21 need to move are:

[0105] ;

[0106] Among them, T is the time for superelevation adjustment, and the moving directions of the left first lead screw 11 and the right first lead screw 21 are opposite.

[0107] The displacements that the left second lead screw 12 and the right second lead screw 22 need to move are:

[0108] ;

[0109] Among them, the moving directions of the second left lead screw 12 and the second right lead screw 22 are opposite.

[0110] After completing the ultra-high adjustment, the track panel mechanism is as Figure 3 shown.

[0111] S7-2: Calculate the moving distances of each connecting rod on the straight line where the lead screw is located for horizontal adjustment with ultra-high. This step includes the following steps:

[0112] S7-2-1: Calculate the deflection angles of the connecting rods after ultra-high adjustment: Calculate the deflection angles of the first right connecting rod ab, the second right connecting rod cd, the first left connecting rod ef, and the second left connecting rod gh after completing the ultra-high adjustment, which are respectively , , , .

[0113] Find : From Figure 3 it can be known that:

[0114] ;

[0115] ... ;

[0116] Therefore:

[0117] ;

[0118] Among them, is the perpendicular distance from point b on the first right connecting rod ab to the first right lead screw 21, represents the distance from point b on the first right connecting rod ab to the track panel center point Z, and point Z is the midpoint of the straight line bf. is the deflection angle of the first right connecting rod ab after completing the ultra-high adjustment, represents the length of the first right connecting rod ab. The straight line Zp is parallel to the horizontal plane, and the straight line bp is perpendicular to the straight line Zp.

[0119] Find : From Figure 3 it can be known that:

[0120] ;

[0121] ;

[0122] Therefore:

[0123] ;

[0124] Among them, is the perpendicular distance from point d on the second right connecting rod cd to the second right lead screw 22, It represents the distance from point d in the right second link cd to the center point Z of the track. To complete the deflection angle of the right second link cd after super-high adjustment, Point t is a point on the right second screw rod 22, and line dt is perpendicular to the right second screw rod 22.

[0125] beg : where point i is a point on the left thread 11, and line fi is perpendicular to the left thread 11. Since the left connecting rod ef and the right connecting rod ab are of equal length, and the vertical distance from point f in the left connecting rod ef to the left thread 11 is equal to the vertical distance from point b in the right connecting rod ab to the right thread 21, we can obtain the following equation:

[0126] ;

[0127] beg , where point k is a point on the left second screw rod 12, and line gk is perpendicular to the left second screw rod 12. Since the left second connecting rod gh and the right second connecting rod cd are of equal length, and the vertical distance from point g in the left second connecting rod gh to the left second screw rod 12 is equal to the vertical distance from point d in the right second connecting rod cd to the right second screw rod 22. Similarly, we can get:

[0128] ;

[0129] S7-2-2: Perform horizontal adjustment with superelevation. After adjustment, the center points of the two rails are offset:

[0130] like Figure 3 As shown, the relative superelevation value has been adjusted. In this embodiment, assuming that the center position of the rail is to the left, the rail needs to be moved to the right as a whole along the figure. The distance of the translation is Now the right connecting rod ab needs to move upwards along the right thread rod 21. After completion, the vertical distance H between the right rail 2 and the ground remains unchanged.

[0131] Calculate the distance D1, D2, D3, D4 that each connecting rod moves on the straight line where the screw rod is located, as well as the connecting rod deflection angle during horizontal adjustment 、 、 、 :

[0132] like Figure 4 To complete the track mechanism diagram with superelevation horizontal adjustment, for the right connecting rod ab:

[0133]

[0134] Solving equations (1) and (2), we can get:

[0135] ;

[0136] ;

[0137] Similarly, it can be generalized to obtain the moving distances D1, D2, D3, and D4 of the right first connecting rod ab, the right second connecting rod cd, the left first connecting rod ef, and the left second connecting rod gh on each lead screw respectively, as well as the deflection angles during horizontal adjustment , , , :

[0138] ;

[0139] ;

[0140] where i = 1, 2, 3, 4. represents the lengths of the connecting rods ab, cd, ef, and gh, represents the deflection angles of the connecting rods after completing the horizontal adjustment with super elevation, represents the offset distance of the center of the rail 2.

[0141] S8. Repeat S6 to S7 until the rail reaches the allowable error range of the target point.

[0142] Finally, it is necessary to state here that the above embodiments are only used to further illustrate the technical solutions of the present invention and should not be construed as limiting the protection scope of the present invention. Some non-essential improvements and adjustments made by those skilled in the art based on the above content of the present invention all fall within the protection scope of the present invention.

Claims

1. A track panel fine adjustment algorithm for double-block ballastless track, characterized in that The method includes the following steps: S1: Establish a physical model of a pair of legs of a track panel, and construct a virtual adjustment system based on the physical model. The physical model of the legs includes a cross beam (1) and two steel rails (2). The two steel rails (2) are arranged in parallel and supported on the cross beam (1). The virtual adjustment system includes four connecting rods and four lead screws. Two connecting rods are connected to each end of the cross beam (1). The four lead screws drive one end of the four connecting rods to move linearly and reciprocally, and the four connecting rods drive the cross beam (1) hinged to the other end thereof to rotate and / or move; S2: Establish a motion planner model for the leg lead screws. The planner model includes a motion target position and a motion speed; S3: Randomly initialize the starting position state of each connecting rod on the lead screw and the starting deflection state of the connecting rod, and randomize the environmental observation data; S4: Through a reinforcement learning algorithm, select the motion trajectory of the steel rail according to the environmental observation data and reward prediction. Use the lead screw motion planner to convert the kinematic data of the steel rail into a motor control command for controlling the motion of the lead screw, drive the leg to adjust, and update the reward data based on the new environmental observation data using the reward function, and continue to output new actions; S5: Repeat step S4, iteratively update the policy learning and value estimation network, and store the experience in the experience pool until convergence, end the training, and save the policy as the motion trajectory of the steel rail; S6: Obtain the initial state and real-time state of the inclination angle of the cross beam of the on-site control system and the deflection angle of the lead screw through an inclination sensor; S7: Calculate the moving distance of each connecting rod on the straight line where the lead screw is located. Use the lead screw motion planner to convert the kinematic data of the steel rail into the kinematic data of each lead screw, and control the track panel to complete the adjustment; S8: Repeat steps S6 to S7 until the steel rail reaches the allowable error range of the target point.

2. The track panel fine-tuning algorithm according to claim 1, wherein: In step S1, the four connecting rods are respectively the right first connecting rod (ab), the right second connecting rod (cd), the left first connecting rod (ef), and the left second connecting rod (gh). The four lead screws are arranged in parallel at intervals and are respectively the left first lead screw (11), the left second lead screw (12), the right second lead screw (22), and the right first lead screw (21) arranged in sequence in one direction. The left first connecting rod (ef), the left second connecting rod (gh), the right first connecting rod (ab), and the right second connecting rod (cd) are respectively connected to the left first lead screw (11), the left second lead screw (12), the right first lead screw (21), and the right second lead screw (22). The upper end of the right first connecting rod (ab) is connected to the right first lead screw (21), and the lower end is connected to the cross beam (1). The lower end of the right second connecting rod (cd) is connected to the right second lead screw (22), and the upper end is connected to the cross beam (1). The upper end of the left first connecting rod (ef) is connected to the left first lead screw (11), and the lower end is connected to the cross beam (1). The lower end of the left second connecting rod (gh) is connected to the left second lead screw (12), and the upper end is connected to the cross beam (1). There is a center line. The left first lead screw (11) and the right first lead screw (21) are symmetric with respect to this center line. The left second lead screw (12) and the right second lead screw (22) are also symmetric with respect to this center line. The left first connecting rod (ef) and the right first connecting rod (ab) are symmetric with respect to this center line. The left second connecting rod (gh) and the right second connecting rod (cd) are also symmetric with respect to this center line.

3. The track panel fine adjustment algorithm according to claim 2, characterized in that: Environmental observation data during the adjustment process of the track panel include: sensor parameters, device status, and target parameters. Among them, the sensor parameters include: the distance between the left first screw rod (11) and the right first screw rod (21). The distance L2 between the left second screw rod (12) and the right second screw rod (22), the distance L3 between the two rails (2), the height difference between the two rails (2), the offset distance, and the tilt angle. Among them, the offset distance refers to the distance by which the center point between the two rails (2) deviates from the center line, and the tilt angle refers to the angle between the cross beam (1) and the horizontal plane; the device status includes: the displacement and speed of each connecting rod on the straight line where the corresponding screw rod is located, and the deflection angle of the connecting rod; the target parameters include: the target superelevation value and the target offset distance, where the superelevation value refers to the height difference between the two rails (2).

4. The track panel fine-tuning algorithm according to claim 3, wherein: Step S4 includes the following steps: S4-1: Initialization state: Obtain the initial state feature vector of the track panel, including sensor parameters, device status, and target parameters; S4-2: Action generation: Generate an action according to the current state through the current network policy function in policy learning, where the action includes the target position of the rail and the motion speed command; S4-3: Action execution and observation: Send the action to the lead screw motion planner, convert it into a motor control execution command, and obtain a new state; S4-4: Experience storage: Store the five-tuple composed of the current state, action, immediate reward, new state, and termination flag in the experience replay pool, where the immediate reward is calculated by the reward function, reflecting the adjustment accuracy, and the termination flag is used to judge whether the target error range is reached; S4-5: Network training: Sample multiple samples from the experience replay pool; S4-6: Target network synchronization: Copy the current network parameters of value estimation or policy learning to the target network every C steps, where C represents the update frequency of the target network parameters; S4-7: Loop termination condition: If the new state is a termination state, that is, the track panel reaches the target accuracy, end the current training round; otherwise, return to step S4-2 to continue iteration.

5. The track panel fine-tuning algorithm according to claim 3, characterized in that: Step S7 includes the following steps: S7-1: Perform superelevation adjustment, and the superelevation value changes after adjustment; S7-2: Perform horizontal adjustment with superelevation, and the center point between the two rails shifts after adjustment.

6. The track lining fine adjustment algorithm according to claim 5, wherein: In step S7-1, calculate the tilt angle that the cross beam (1) needs to be adjusted according to the superelevation value to be adjusted and the distance between the two rails, and calculate the displacement that each connecting rod needs to complete on the corresponding lead screw according to the relationship between the linear movement speeds of the two connecting rods on the same side of the center line.

7. The track-laying fine-tuning algorithm according to claim 5, characterized in that: Step S7-2 includes the following steps: S7-2-1: Calculate the deflection angles of each connecting rod after superelevation adjustment; S7-2-2: Calculate the displacement that each connecting rod needs to complete on the corresponding lead screw according to the target deflection angle, target offset distance of each connecting rod, and the deflection angle obtained in step S7-2-1.

8. The track panel fine adjustment algorithm according to claim 6, wherein: In step S7-1, ； ； ； ； ； ； Where: A is the ultra-high value to be adjusted, is the tilt angle of the cross beam (1) to be adjusted, is the moving speed of the left lead screw (11), is the moving speed of the right lead screw (21), is the moving speed of the left second lead screw (12), is the moving speed of the right second lead screw (22), is the displacement that the left lead screw (11) and the right lead screw (21) need to move, is the displacement that the left second lead screw (12) and the right second lead screw (22) need to move, and T is the time for ultra-high adjustment.

9. The track panel fine adjustment algorithm according to claim 7, wherein: In step S7-2-2, ； ； wherein i = 1, 2, 3, 4, represents the length of each connecting rod, represents the displacement of each connecting rod on the corresponding lead screw, represents the deflection angle of each connecting rod relative to the initial state after the horizontal adjustment with superelevation is completed, represents the offset distance of the center point between two rails (2), represents the deflection angle relative to the state after superelevation adjustment after the horizontal adjustment with superelevation.

Images