Attitude prediction model construction method, attitude prediction method, equipment and medium

Abstract

The invention discloses an attitude prediction model construction method, an attitude prediction method, equipment and a medium, and relates to the technical field of shield tunneling, and the technical scheme is characterized in that historical shield parameters of a shield tunneling machine under a historical time sequence are acquired; performing dimension reduction screening on the historical shield parameters to obtain shield parameters after dimension reduction; a multi-objective optimization function enabling the difference between the horizontal positions and the vertical positions of the front end and the rear end of the shield to be minimum is constructed, the shield parameters after dimension reduction are combined, the multi-objective optimization function is solved through a non-dominated genetic algorithm, and the optimal shield parameters are obtained; and training a pre-constructed long-short term memory neural network by using the optimal shield parameters, and obtaining a posture prediction model capable of being used for predicting the posture of the shield tunneling machine after the training is completed. The problem that an attitude prediction model provided in the prior art is insufficient in calculation efficiency and prediction precision is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of shield tunneling, and more specifically, it relates to a method for constructing an attitude prediction model, an attitude prediction method, a device, and a medium. Background Art

[0002] With the rapid development of urbanization, the demand for subway tunnel projects within cities and between cities is also increasing day by day. As a commonly used device for subway tunnel excavation, during the tunneling process of a shield machine, due to some unpredictable factors such as blade eccentric loading, formation changes, and the interference of the shield's own weight, it often causes the deviation of the shield machine's attitude position, making the shield machine move in a snake-like manner in the tunnel. Therefore, when the deviation exceeds the allowable range, it is necessary to promptly adjust the attitude position of the shield machine to make it return to the preset track axis.

[0003] Currently, the prior art has adopted genetic algorithms and principal component analysis methods, combined with long short-term memory neural models to achieve multi-dimensional influencing factor optimization and shield attitude prediction. However, this prior art has deficiencies in calculation efficiency and prediction accuracy. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for constructing an attitude prediction model, an attitude prediction method, a device, and a medium, which solves the problems of insufficient calculation efficiency and prediction accuracy of the attitude prediction model provided by the prior art.

[0005] The above technical purpose of the present invention is achieved through the following technical solutions: The first aspect of the present invention provides a method for constructing an attitude prediction model, the method comprising: Obtaining historical shield parameters of the shield machine in a historical time series; Performing dimensionality reduction screening on the historical shield parameters to obtain the shield parameters after dimensionality reduction; Constructing a multi-objective optimization function that minimizes the difference between the horizontal position and the vertical position of the front end and the rear end of the shield, and combining the shield parameters after dimensionality reduction, using a non-dominated genetic algorithm to solve the multi-objective optimization function to obtain the optimal shield parameters; Training a pre-constructed long short-term memory neural network using the optimal shield parameters, and after the training is completed, obtaining an attitude prediction model that can be used for predicting the attitude of the shield machine.

[0006] In one implementation, the historical shield parameters include position parameters of the shield attitude and shield operation parameters.

[0007] In one implementation, before performing dimensionality reduction screening on the historical shield parameters to obtain the shield parameters after dimensionality reduction, the method further comprises: using the K-S parameter verification method to perform a normality verification on the historical shield parameters to eliminate outlier data in the historical shield parameters.

[0008] In one implementation, the principal component analysis method is used to perform dimensionality reduction screening on the historical shield parameters to obtain the shield parameters after dimensionality reduction.

[0009] In one implementation, before using the non-dominated genetic algorithm to solve the multi-objective optimization function to obtain the optimal shield parameters, the method further includes: using the ant colony algorithm to adjust the search strategy of the non-dominated genetic algorithm.

[0010] In one implementation, the improved ant colony algorithm refers to the pheromone global update rule that performs probabilistic search according to the state transition rule in each loop, enhances the pheromone of the optimal search path obtained in each loop, and performs local update when constructing the search path, and performs a local update on the search path after each loop.

[0011] In one implementation, the expression of the state transition rule is , where J represents the set of nodes that can be directly reached by the node and are not in the sequence of nodes that have been visited, is the node i to the node j pheromone, is the node i to the node j heuristic function, is the pheromone incentive factor, β is the expected heuristic factor, q represents a random number, q0 represents a random number from 0 to 1, and S represents the probability calculated by the probability calculation function; The expression of the pheromone global update rule is: , where C represents the path length selected by the ant, represents the pheromone global evaporation factor, represents the amount of pheromone released by the ant on the edge formed by nodes i and j; The expression of the pheromone local update rule is , where represents the pheromone local evaporation factor, represents the chaos factor generated by the chaos mapping function, k represents the current loop count, K represents the maximum loop count, represents the initial value of the pheromone.

[0012] The second aspect of the present invention provides a posture prediction method, and the method includes: Obtain the shield parameters of the shield machine under the current time series; Input the shield parameters into the attitude prediction model constructed by the attitude prediction model construction method provided in the first aspect of the present invention, and output the attitude prediction result of the shield machine in the next time series.

[0013] The third aspect of the present invention provides an electronic device, including a memory and a processor; The memory is used to store a computer program, and the computer program includes program instructions; The processor is used to execute the program instructions so that the electronic device executes an attitude prediction model construction method provided in the first aspect of the present invention and an attitude prediction method provided in the second aspect of the present invention.

[0014] The fourth aspect of the present invention provides a computer program product containing program instructions. When the program instructions are run on an electronic device, the electronic device is enabled to execute an attitude prediction model construction method provided in the first aspect of the present invention and an attitude prediction method provided in the second aspect of the present invention.

[0015] Compared with the prior art, the present invention has the following beneficial effects: In an attitude prediction model construction method provided by the present invention, the improved ant colony algorithm is used to adjust the search strategy of the non-dominated genetic algorithm to adjust the search strategy of the non-dominated genetic algorithm, so that the non-dominated genetic algorithm can accelerate the solution efficiency and accuracy of the multi-objective optimization function with the smallest difference in the horizontal and vertical positions between the front and rear ends of the shield. Then, based on the optimal shield parameters, the pre-constructed long short-term memory neural network is trained. When the training is completed, an attitude prediction model that can be used for shield machine attitude prediction can be obtained. Since the adjusted non-dominated genetic algorithm improves the solution accuracy, the prediction accuracy of the attitude prediction model trained based on the solved optimal shield parameters is also correspondingly improved, thus solving the problems of insufficient calculation efficiency and prediction accuracy in the attitude prediction method provided by the prior art. Description of the Drawings

[0016] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, form a part of this application, and do not limit the embodiments of the present invention. In the drawings: Figure 1 It is a schematic flow chart of an attitude prediction model construction method provided by an embodiment of the present invention; Figure 2 It is a schematic flow chart of an attitude prediction method provided by an embodiment of the present invention. Detailed Embodiments

[0017] To make the objectives, technical solutions, and advantages of the present invention more clear and understandable, the present invention will be further described in detail below with reference to embodiments and the accompanying drawings. The illustrative embodiments of the present invention and the description thereof are only used to explain the present invention and do not limit the present invention.

[0018] It should be noted that the term "comprising" or "may comprise" that can be used in various embodiments of the present application indicates the presence of the claimed functions, operations, or elements, and does not limit the addition of one or more functions, operations, or elements. In addition, as used in various embodiments of the present application, the terms "comprising", "having", and their cognates are only intended to represent specific features, numbers, steps, operations, elements, components, or combinations of the foregoing items, and should not be construed as precluding the existence or addition of one or more other features, numbers, steps, operations, elements, components, or combinations of the foregoing items.

[0019] It should be understood that terms such as "first" and "second" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance or implicitly indicating the quantity of the indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include one or more of such features. In the description of the present invention, "a plurality of" means two or more unless otherwise specifically defined.

[0020] Please refer to Figure 1 , Figure 1 which is a schematic flowchart of a method for constructing an attitude prediction model provided by an embodiment of the present invention. As Figure 1 shown, the method includes: S101, obtaining historical shield parameters of the shield machine in a historical time series.

[0021] In this embodiment, the historical shield parameters include position parameters of the shield attitude and shield operation parameters. For example, the position parameters of the shield attitude include displacement parameters, angle parameters, tangent mileage, and three-dimensional coordinates, etc. Among them, the displacement parameters can be divided into the horizontal deviation and vertical deviation at the front end of the shield, and the horizontal deviation and vertical deviation at the rear end of the shield. Among them, the horizontal deviation at the front end of the shield refers to the distance by which the head of the shield machine deviates from the tunnel design axis in the horizontal direction, and the vertical deviation at the front end of the shield refers to the distance by which the head of the shield machine deviates from the tunnel design axis in the vertical direction.

[0022] The angular parameters include pitch angle, azimuth angle, roll angle, etc. The pitch angle refers to the angle between the central axis of the shield machine and the horizontal plane, reflecting the inclination degree of the shield machine in the vertical direction and determining the slope of the tunnel. Azimuth angle: That is, the yaw angle, which refers to the angle between the projection of the shield machine's tunneling direction on the horizontal plane and the designed axis of the tunnel, reflecting the deviation direction of the shield machine in the horizontal plane. Roll angle: Also known as the torsion angle or yaw angle, it refers to the angle by which the shield machine rotates around its own central axis, mainly caused by the reaction force of the shield cutterhead, reflecting the rotation state of the shield machine.

[0023] The cutting mileage refers to the mileage position where the cutting part of the shield machine is located, which can help determine the longitudinal position of the shield machine in the tunnel.

[0024] The three-dimensional coordinates are the three-dimensional coordinate positions of the center of the shield machine's cutterhead or key parts, which can accurately determine the specific position of the shield machine in space.

[0025] The shield operation parameters include tunneling speed, propulsion cylinder pressure, propulsion cylinder stroke difference, cutterhead rotation speed, cutterhead torque, etc. Each parameter described in this embodiment is a common working parameter of the shield machine, and will not be specifically described one by one in this embodiment.

[0026] Sensors are arranged on the shield machine to record the shield attitude position parameters and shield operation parameters. Taking each ring as a recording sample, more than 1800 rings are recorded in total. The statistical parameters are shown in Table 1. Among them, parameters 1-14 are used as input indicators, and parameters 15-18 are used as output indicators. The position parameter coordinates of the shield attitude take the intersection of a designed axis and a cross-section as the origin. The horizontal direction is negative when deviating to the left and positive when deviating to the right. The vertical direction is negative when deviating downwards and positive when deviating upwards.

[0027] Table 1

[0028] S102. Perform dimensionality reduction screening on the historical shield parameters to obtain the shield parameters after dimensionality reduction.

[0029] Specifically, the principal component analysis method is used to perform dimensionality reduction screening on the historical shield parameters to obtain the shield parameters after dimensionality reduction.

[0030] Use the PCA principal component analysis method to analyze and screen the 18 parameters provided above. The specific implementation is as follows: Perform standardization processing on the original data, adjust the mean of each variable to 0 and the variance to 1 to eliminate the influence of the dimension and order of magnitude between different variables.

[0031] According to the standardized data, calculate the covariance matrix between variables.

[0032] Perform eigen decomposition on the covariance matrix to obtain the eigenvalues and the corresponding eigenvectors. The eigenvalues represent the variance magnitudes of the principal components, and the eigenvectors represent the directions of the principal components.

[0033] According to the magnitudes of the eigenvalues, select the top k larger eigenvalues and their corresponding eigenvectors as the principal components.

[0034] Project the original data onto the selected principal components to obtain the principal component scores.

[0035] Analyze and screen the 18 parameters through the principal component scores, and finally select a total of 17 characteristic parameters, including 13 input features and 4 output features, as the sample set for the long short-term memory neural network training. S103. Construct a multi-objective optimization function that minimizes the differences in the horizontal and vertical positions between the front and rear ends of the shield. Combine the dimension-reduced shield parameters, and use the non-dominated genetic algorithm to solve the multi-objective optimization function to obtain the optimal shield parameters.

[0036] In this embodiment, the expression of the multi-objective optimization function that minimizes the differences in the horizontal and vertical positions between the front and rear ends of the shield is: , where represents the predicted value of the i-th solution of objective n (n takes 1 - 4), is the true value of objective n, that is, the coordinate values of the horizontal and vertical positions of the front and rear ends of the shield. In other words, it is the difference between the front-end horizontal position, the front-end vertical position, the rear-end horizontal position, and the rear-end vertical position. Take the minimum value of S as the optimal shield attitude parameter for calculation.

[0037] The non-dominated genetic algorithm uses the third-generation non-dominated genetic algorithm. The Non-dominated Sorting Genetic Algorithm III (NSGA-III) is an improved version of NSGA-II (hereinafter referred to as the genetic algorithm), mainly dealing with problems with four or more objectives (multi-objective optimization problems, Many-Objective Optimization) (Xu et al., 2025). When the problem involves four or more objectives, the main difficulties include the non-dominance problem of the population, the low efficiency of the recombination operation, the difficulty in representing the high-dimensional objective surface, the high calculation cost of the performance index, and the increased difficulty in visualization. The core improvement of NSGA-III lies in its selection operation, especially the maintenance of the diversity among the population members, which is achieved through the provision and dynamic update mechanism of the reference points. Its calculation formula is as follows:

[0038]

[0039]

[0040]

[0041] Where: zi represents the extreme point; D represents the spatial dimension; a i represents the intercept of the hyperplane with the target axis; f i (x) represents individual x in the i-th population i The i-th objective function value of ; w i Represents the weight vector; grid x Represents the position index of individual x in the grid; grid i Represents the position index of grid i; Represents the degree of grid crowding; d is the unit length of the grid; floor() represents rounding down.

[0042] The above formula is used to explore multi-dimensional space, the grid congestion is dynamically adjusted according to decision preferences, and the weighting method is used to screen the Pareto optimal solution in the multi-dimensional space to achieve optimal consideration of multiple objectives.

[0043] The above embodiment can determine the solution set of the Pareto front of the four optimization objectives of optimizing the front and rear ends of the shield in the horizontal / vertical direction, but it corresponds to multiple groups of operating parameter combinations and has certain differences, so this embodiment adopts a multi-objective optimization function, brings the combination of each shield operating parameter into the prediction model, outputs the results of the four index parameters, and then performs an absolute value operation of the difference with the actual value measured on site, and finally sums them up to minimize the total difference S of the four objectives. The processing of historical data in the previous article is mainly in the model training part so that it can solve the corresponding solution set, and then combined with the actual project, the four absolute values ​​can be weighted to select the required optimal solution from the obtained results.

[0044] In one implementation, before using a non-dominated genetic algorithm to solve the multi-objective optimization function and obtain optimal shield parameters, based on the steps described in this embodiment, the method further includes: using an ant colony algorithm to adjust the search strategy of the non-dominated genetic algorithm.

[0045] The ant colony algorithm provided in this embodiment is an improved ant colony optimization (IACO), which improves the parameters of its heuristic function based on the traditional ant colony optimization (ACO), and can improve the computational efficiency of the overall algorithm (Moissa et al., 2021). Its basic process is as follows: Set the basic grid parameters and generate a certain number of ant populations within the grid. Various features will attract the ant populations to move as "food". During the process of searching for "food", each ant releases pheromones on its path, which affects the path selection of the next ant. Finally, the shortest path is selected by recording the overall path distance. Its calculation formula is as follows: ; where η ij represents a node i to node j 's heuristic function; E j and E S represent the energy levels of the corresponding nodes; N i represents the set of neighboring nodes of node i ;

[0046] The improved ant colony algorithm refers to the pheromone global update rule that performs probabilistic search according to the state transition rule in each loop, enhances the pheromone of the optimal search path obtained in each loop, and performs local update when constructing the search path, and performs a local update of the pheromone on the search path once after each loop.

[0047] Specifically, in the ant colony algorithm, ants perform probabilistic search on paths according to the random proportion rule. In the improved ant colony algorithm, ants use the state transition rule for search. This rule can not only store the information already obtained, but also perform a biased search on the actual problem.

[0048] In the ant colony algorithm, all ants perform the pheromone global update rule, which will cause ants not to quickly find the shortest path, further reducing the search efficiency of the algorithm. In the improved ant colony algorithm, not all ants perform global update. Instead, after each loop, only the pheromone of the optimal path is enhanced, and the pheromone of other non-optimal paths does not need to be enhanced. The pheromone on these paths will gradually volatilize, resulting in a decrease in concentration, and finally causing more ants to choose the optimal path, greatly improving the search efficiency of the algorithm and correspondingly shortening the search time.

[0049] In the ant colony algorithm, when a loop is performed, a local update is carried out on all paths. In the improved ant colony algorithm, ants not only perform local updates while constructing paths, but also perform an additional local update on the paths after each loop. This can make the paths that have been visited become less attractive when being visited by ants, indirectly making ants more inclined to search for unvisited paths, so that ants will not converge to the same path. Without local updates, all ants will search within the narrow area of the previously found optimal path, thus preventing further search for better solutions. It can be seen that the introduction of the pheromone local update rule will effectively enhance the optimization performance of the ant colony algorithm.

[0050] Specifically, the expression of the state transition rule is , where J represents the set of nodes that can be directly reached by the node and are not in the sequence of nodes that have been visited, is the pheromone from node i to node j , is the heuristic function from node i to node j , is the pheromone incentive factor, β is the expected heuristic factor, q represents a random number, q0 represents a random number between 0 and 1, and S represents the probability calculated by the probability calculation function.

[0051] Specifically, q0 is a very important control parameter introduced in the improved ant colony algorithm. In the state transition rule of the improved ant colony algorithm, the probability that an ant selects the current optimal moving direction is q0. At the same time, the ant searches each edge with a probability of (1 - q0) in a biased manner. By adjusting q0, the balance between "exploitation" and "biased exploration" can be effectively adjusted, thus determining whether the algorithm should explore new areas or stay near the current optimal path in the next step. That is, the magnitude of the parameter q0 determines the relative importance between using prior knowledge and exploring new paths.

[0052] The expression of the pheromone global update rule is: , where C represents the length of the path selected by the ant, represents the pheromone global evaporation factor, represents the amount of pheromone released by the ant on the edge formed by node i and node j.

[0053] Specifically, in the improved ant colony algorithm, only the ants on the currently optimal path (the ants that have constructed the shortest path from the start of the algorithm to the current iteration) are allowed to release pheromones, that is, enhance the pheromones on this path with an intensity of Q. This strategy, combined with the state transition rule, greatly enhances the guiding nature of the algorithm search. After all ants have constructed their paths in each iteration, the global pheromone update rule is used. Whether it is the evaporation or release of pheromones, it is only carried out on the edges belonging to the so-far optimal path, which is very different from the basic ant colony algorithm. Because in the basic ant colony algorithm, pheromone updates are performed on all edges, and the computational complexity is O(n 2 ); while in the improved ant colony algorithm, the computational complexity of pheromone update is O(n), which significantly reduces the search complexity and can effectively improve the solution efficiency of the non-dominated genetic algorithm.

[0054] The expression of the local pheromone update rule is , where represents the local pheromone evaporation factor, represents the chaos factor generated by the chaos mapping function, k represents the current loop count, K represents the maximum loop count, represents the initial value of the pheromone.

[0055] Specifically, for the local update rule, in this embodiment, a chaos factor generated by the chaos mapping function is introduced to determine the local pheromone evaporation factor, so as to enhance the diversity of local search, ensure the update performance of pheromones, and thus reduce the probability that ants choose the path affected by the local pheromone update rule.

[0056] S104. Use the optimal shield parameters to train the pre-constructed long short-term memory neural network. After the training is completed, an attitude prediction model that can be used for shield machine attitude prediction is obtained.

[0057] In this embodiment, the gates in the long short-term memory neural network include an input gate, a forget gate, and an output gate, and their calculation formulas are as follows: , where: f t is the t output of the forget gate at time σ is the activation function; W f and U f are the weight vectors of the forget gate; x t is the t input vector of the memory unit at time h t-1 is the t output vector of the memory unit at time -1; bf Denotes the bias vector of the forget gate.

[0058]

[0059]

[0060] , where: i t Is the output of the input gate; tanh Is the activation function; W i , W c , U i , U c Are the input gate weight vectors; C t Is t The state information at the moment; C’ t Is t The candidate information at the moment; b i And b c Are the bias vectors of the input gate.

[0061]

[0062] , where: o t Is t The output of the output gate at the moment; W o And U o Are the output gate weight vectors; b o Is the bias vector of the output gate; h t Is the value of the memory cell at t The moment.

[0063] The method of training the pre-constructed long short-term memory neural network with optimal shield parameters is common knowledge and will not be described in this embodiment.

[0064] Based on the above description, the present invention uses the improved ant colony algorithm to adjust the search strategy of the non-dominated genetic algorithm, so as to adjust the search strategy of the non-dominated genetic algorithm, thereby accelerating the solution efficiency and accuracy of the multi-objective optimization function with the smallest difference in the horizontal and vertical positions between the front and rear ends of the shield tunneling machine. Then, based on the optimal shield parameters, the pre-constructed long short-term memory neural network is trained. When the training is completed, an attitude prediction model that can be used for shield tunneling machine attitude prediction can be obtained. Since the adjusted non-dominated genetic algorithm improves the solution accuracy, the prediction accuracy of the attitude prediction model trained based on the solved optimal shield parameters is also correspondingly improved, thus solving the problems of insufficient calculation efficiency and prediction accuracy in the attitude prediction method provided by the prior art.

[0065] In one embodiment, before performing step S102, the model construction method described above further includes: using the K-S parameter verification method to perform a normality verification on the historical shield parameters to eliminate the outlier data in the historical shield parameters.

[0066] Specifically, use SPSS software to calculate the K-S statistic for each parameter, and the calculated values are all greater than 0.05, which means that under the 95% confidence interval, the parameters conform to the normal distribution. When the parameter samples conform to or approximately conform to the normal distribution, the 3 σ sigma rule can be used to eliminate the outlier data, so as to ensure the accuracy of the historical shield parameters.

[0067] Please refer to Figure 2 , Figure 2 which is a schematic flowchart of an attitude prediction method provided by an embodiment of the present invention. As Figure 2 shown, the method includes: S201, obtaining the shield parameters of the shield tunneling machine at the current time series; S202, inputting the shield parameters into the attitude prediction model constructed by the attitude prediction model construction method described in the above embodiment, and outputting the attitude prediction result of the shield tunneling machine at the next time series.

[0068] In this embodiment, the present invention uses an improved ant colony algorithm to adjust the search strategy of the non-dominated genetic algorithm, so as to adjust the search strategy of the non-dominated genetic algorithm, thereby accelerating the solution efficiency and accuracy of the multi-objective optimization function with the smallest difference in the horizontal and vertical positions between the front and rear ends of the shield tunneling machine. Then, based on the optimal shield parameters, the pre-constructed long short-term memory neural network is trained. When the training is completed, an attitude prediction model that can be used for shield tunneling machine attitude prediction can be obtained. Since the adjusted non-dominated genetic algorithm improves the accuracy of solving the multi-objective optimization function, the prediction accuracy of the attitude prediction model trained based on the solved optimal shield parameters is also correspondingly improved, thus solving the problems of insufficient calculation efficiency and prediction accuracy in the attitude prediction method provided by the prior art.

[0069] The embodiment of the present invention also provides an electronic device. Among them, the electronic device includes a processor, a memory, a communication interface, and at least one communication bus for connecting the processor, the memory, and the communication interface. The memory includes, but is not limited to, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (PROM), or a portable read-only memory (CD-ROM), and the memory is used for relevant instructions and data.

[0070] The communication interface is used to receive and send data. The processor can be one or more CPUs. In the case where the processor is a single CPU, the CPU can be a single-core CPU or a multi-core CPU. The processor in the electronic device is used to read one or more programs stored in the memory and perform the following operations: obtaining the historical shield parameters of the shield tunneling machine in the historical time series; performing dimensionality reduction screening on the historical shield parameters to obtain the shield parameters after dimensionality reduction; constructing a multi-objective optimization function that minimizes the difference in the horizontal and vertical positions between the front and rear ends of the shield tunneling machine, and combining the shield parameters after dimensionality reduction, using the non-dominated genetic algorithm to solve the multi-objective optimization function to obtain the optimal shield parameters; using the optimal shield parameters to train the pre-constructed long short-term memory neural network, and after the training is completed, obtaining an attitude prediction model that can be used for shield tunneling machine attitude prediction; and also performing the following operations: obtaining the shield parameters of the shield tunneling machine in the current time series; inputting the shield parameters into the attitude prediction model, and outputting the attitude prediction result of the shield tunneling machine in the next time series.

[0071] It should be noted that the specific implementation of each operation can be the corresponding description of the method embodiment shown above Figure 1 The electronic device can be used to execute an attitude prediction model construction method and an attitude prediction method in the method embodiments of the present application above, and will not be specifically described herein.

[0072] Embodiments of the present invention further provide a computer-readable storage medium, which is a memory device in a computer device and is used to store programs and data. It can be understood that the computer-readable storage medium here can include both the built-in storage medium in the computer device and, of course, the extended storage medium supported by the computer device. The computer-readable storage medium provides a storage space, and this storage space stores the operating system of the terminal. Moreover, in this storage space, one or more instructions suitable for being loaded and executed by a processor are also stored, and these instructions can be one or more computer programs (including program codes). It should be noted that the computer-readable storage medium here can be a high-speed RAM memory or a non-volatile memory, such as at least one disk memory. One or more instructions stored in the computer-readable storage medium can be loaded and executed by the processor to implement the corresponding steps of a method for constructing an attitude prediction model and a method for attitude prediction in the above embodiments. Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can be implemented in the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can be implemented in the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memories, CD-ROMs, optical memories, etc.) containing computer-usable program codes.

[0073] Embodiments of the present invention further provide a computer program product containing program instructions. The computer program product can be software or a program product containing program instructions that can run on a computing device or be stored in any available medium. When the computer program product runs on at least one electronic device, at least one electronic device is caused to execute a method for constructing an attitude prediction model and a method for attitude prediction.

[0074] The specific embodiments described above further elaborate on the purpose, technical solutions, and beneficial effects of the present invention. It should be understood that the above description is only the specific embodiments of the present invention and is not used to limit the protection scope of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for constructing a pose prediction model, characterized in that the method Including: Obtain the historical shield parameters of the shield machine in the historical time series; Perform dimensionality reduction screening on the historical shield parameters to obtain the shield parameters after dimensionality reduction; Construct a multi-objective optimization function that minimizes the differences in the horizontal and vertical positions between the front and rear ends of the shield, and combine the shield parameters after dimensionality reduction. Use the non-dominated genetic algorithm to solve the multi-objective optimization function to obtain the optimal shield parameters; Use the optimal shield parameters to train a pre-constructed long short-term memory neural network. After the training is completed, obtain an attitude prediction model that can be used for shield machine attitude prediction.

2. The method for constructing a posture prediction model according to claim 1, wherein The historical shield parameters include the position parameters of the shield attitude and the shield operation parameters.

3. A method for constructing a posture prediction model according to claim 1, characterized in that Before performing dimensionality reduction screening on the historical shield parameters to obtain the shield parameters after dimensionality reduction, the method further includes: using the K-S parameter verification method to perform normality verification on the historical shield parameters to eliminate the outlier data in the historical shield parameters.

4. A method for constructing an attitude prediction model according to claim 1, characterized in that, Use the principal component analysis method to perform dimensionality reduction screening on the historical shield parameters to obtain the shield parameters after dimensionality reduction.

5. A method for constructing an attitude prediction model according to claim 1, characterized in that, Before using the non-dominated genetic algorithm to solve the multi-objective optimization function to obtain the optimal shield parameters, the method further includes: using an improved ant colony algorithm to adjust the search strategy of the non-dominated genetic algorithm.

6. The method for constructing an attitude prediction model according to claim 5, wherein The improved ant colony algorithm refers to the pheromone global update rule that performs probabilistic search according to the state transition rule in each loop process, enhances the pheromone of the optimal search path obtained in each loop, and performs local update when constructing the search path, and performs a local update on the search path once after each loop.

7. A method for constructing an attitude prediction model according to claim 6, characterized in that, The expression of the state transition rule is , where J represents the set of nodes that can be directly reached by the node and are not in the sequence of nodes that have been visited, is the pheromone from node i to node j . is the heuristic function from node i to node j . is the pheromone incentive factor, β is the expected heuristic factor, q represents a random number, q0 represents a random number from 0 to 1, and S represents the probability calculated by the probability calculation function; The expression of the global update rule of pheromone is as follows: , where C represents the path length selected by the ant, represents the global evaporation factor of pheromone, represents the amount of pheromone released by the ant on the edge formed by nodes i and j; The expression of the pheromone local update rule is , where represents the pheromone local evaporation factor, represents the chaotic factor generated by the chaotic mapping function, k represents the current iteration number, K represents the maximum iteration number, represents the initial value of the pheromone.

8. A posture prediction method, characterized in that the method Including: Obtain the shield parameters of the shield machine in the current time series; Input the shield parameters into the attitude prediction model constructed by the method for constructing an attitude prediction model according to any one of claims 1 to 7, and output the attitude prediction result of the shield machine in the next time series.

9. An electronic device, characterized in that, Including a memory and a processor; The memory is used to store a computer program, and the computer program includes program instructions; The processor is used to execute the program instructions so that the electronic device executes the method for constructing an attitude prediction model according to any one of claims 1 to 7 and the method for an attitude prediction according to claim 8.

10. A computer program product comprising program instructions, characterized in that, When the program instructions are run by the electronic device, the electronic device is caused to execute the method for constructing an attitude prediction model according to any one of claims 1 to 7 and the method for an attitude prediction according to claim 8.

Images