Optimization method for spatial layout of armored vehicle power system considering disassembly and assembly operation influence

Abstract

The application discloses a method for optimizing the spatial layout of an armored vehicle power system considering the influence of disassembly and assembly operations, and relates to the technical field of armored vehicle power systems, and comprises the following steps: step S1, constructing a virtual maintenance tool, a maintenance object and a human body model of a virtual maintenance personnel in a computer through three-dimensional modeling software; step S2, setting layout optimization parameters and setting a system layout optimization target; step S3, constructing a system layout optimization model; and step S4, setting a solution algorithm of the system layout optimization model constructed in step S3. The application realizes the optimization of the spatial layout of the armored vehicle power system by optimizing the spatial efficiency and the maintenance efficiency.

Description

Technical Field

[0001] This invention relates to the field of armored vehicle power system technology, and in particular to a method for optimizing the spatial layout of armored vehicle power systems considering the impact of disassembly and assembly operations. Background Technology

[0002] The main components of an armored vehicle's power system include the engine, transmission system, cooling system, fuel system, intake and exhaust system, and electrical auxiliary system. In existing armored vehicles, the engine and transmission system are located in the center, while the cooling system, fuel system, intake and exhaust system, and electrical auxiliary system are placed on either side of the hull. In this layout, maintenance tasks are relatively concentrated on one side, resulting in low time utilization. Therefore, a more rational spatial layout is needed to reduce maintenance time.

[0003] Traditional layout optimization methods include genetic algorithms, simulated annealing algorithms, and tabu search algorithms. However, under complex constraints, these methods are prone to getting stuck in local optima, having excessively long convergence times, or consuming too much memory. Summary of the Invention

[0004] In view of the above-mentioned shortcomings and deficiencies of the prior art, the present invention provides a method for optimizing the spatial layout of the armored vehicle power system considering the impact of disassembly and assembly operations. By synergistically optimizing space efficiency and maintenance effectiveness, the spatial layout of the armored vehicle power system is optimized.

[0005] To achieve the above objectives, the main technical solutions adopted by the present invention include:

[0006] This invention provides a method for optimizing the spatial layout of an armored vehicle's power system, taking into account the impact of disassembly and assembly operations, comprising the following steps:

[0007] Step S1: Construct virtual maintenance tools, maintenance objects, and virtual maintenance personnel human body models in the computer using 3D modeling software, import them into the 3D rendering engine, and use the state machine function provided by the 3D rendering engine to control the motion of the imported virtual maintenance personnel human body models for switching maintenance operations.

[0008] Step S2: Set layout optimization parameters, including cumulative space value and maintenance time statistics; set system layout optimization objectives, including maximum maintenance efficiency and maximum space efficiency. Maximum maintenance efficiency means the shortest disassembly and assembly preparation time and maintenance time, and maximum space efficiency means the minimum space occupied by the equipment.

[0009] Step S3: Construct the system layout optimization model as follows:

[0010] ;

[0011] Where: α is the dynamic weighting coefficient of the spatial cumulative value, and β is the dynamic weighting coefficient of the maintenance time statistics. For the exploration factor, N is the number of simulations for the parent node, and n is the number of simulations for the current node; the cumulative space value is the total space occupied by the optimized equipment, and the maintenance time statistics value is the time required for the layout of the vehicle body to be virtually disassembled and repaired according to the maintenance steps specified in the maintenance manual;

[0012] Step S4: Set the solution algorithm for the system layout optimization model constructed in step S3, including the following steps:

[0013] Step S401: Starting from the root node, select child nodes according to the following model:

[0014] ;

[0015] Step S402: Use an ε-greedy strategy to select branches with potentially high node value;

[0016] Step S403: Starting from the newly generated child nodes, simulate the complete layout, complete the placement of the remaining components, and calculate the final node value according to the formula in step S3.

[0017] Step S404: Reverse update node attributes;

[0018] Propagate backward along the search path, updating the statistical attributes of all nodes according to the following rules:

[0019] Spatial cumulative value: ;in, For learning rate, This is the average spatial cumulative value;

[0020] Repair time statistics: ; As the attenuation factor, This is a statistical value representing the average repair time.

[0021] Step S405: If the convergence condition is met, output the optimized layout:

[0022] Set the maximum number of iterations or the convergence threshold for the layout evaluation index, and solve for the optimal layout result. The formula for setting the dynamic convergence threshold is as follows:

[0023] ;

[0024] Where t is the current iteration number.

[0025] Furthermore, in step S1, the human body model of the virtual maintenance personnel is made using the Humaonid skeletal standard.

[0026] Furthermore, in step S2, the highest maintenance efficiency is prioritized, followed by the highest space efficiency.

[0027] Furthermore, in step S401, the cumulative spatial value is based on the ratio of the component envelope volume to the total space of the layout container, and a normalized spatial coverage index is used.

[0028] Furthermore, in step S401, the maintenance time statistics are converted into time cost by summing the component disassembly and assembly path length estimated by path planning simulation and the maintenance time calculated by model method.

[0029] Furthermore, in step S402, the optimal value node is selected with probability 1-ε, and new branches are randomly explored with probability ε, where ε decays with the number of iterations. During random exploration, branches with potentially high node value are selected first.

[0030] The beneficial effects of this invention are:

[0031] This invention utilizes the Monte Carlo Tree Search (MCTS) algorithm to achieve synergistic optimization of spatial efficiency and maintenance effectiveness, while simultaneously considering search efficiency and solution quality, thus finding the optimal solution for spatial layout more quickly.

[0032] This invention achieves collaborative optimization through a reward function, which is a weighted sum of two objectives: space utilization and maintenance. The algorithm automatically finds the layout that maximizes the overall score, naturally achieving collaboration. During the simulation, maintenance efficiency is evaluated in real time, ensuring that each evaluated layout considers both space and maintenance dimensions simultaneously.

[0033] To find the optimal solution more quickly, MCTS does not search all possible layouts evenly. Instead, it intelligently concentrates computational resources on more promising regions based on historical information. Even if it terminates early, MCTS can still return the best solution found so far, making it ideal for time-sensitive applications. The multiple simulations in MCTS are independent of each other, allowing for convenient parallel computation and greatly accelerating the search process. Detailed Implementation

[0034] To better explain and facilitate understanding of the present invention, the present invention will be described in detail below through specific embodiments.

[0035] This invention provides a method for optimizing the spatial layout of an armored vehicle's power system, taking into account the impact of disassembly and assembly operations, comprising the following steps:

[0036] Step S1: Construct virtual repair tools, repair objects, and virtual repair personnel human models in a computer using 3D modeling software such as 3ds Max or Maya; the repair objects can be digital prototypes. The virtual repair personnel's human model is created using the Humaoniid skeleton standard. Import the constructed model into a 3D rendering engine, specifically Unity3D. Use the state machine function provided by Unity3D to control the motion of the imported virtual repair personnel's human model for switching actions during repair operations.

[0037] Step S2: Set layout optimization parameters, including cumulative space value and maintenance time statistics; set system layout optimization objectives, including maximum maintenance efficiency and maximum space efficiency. Maximum maintenance efficiency means the shortest disassembly and assembly preparation time and maintenance time, and maximum space efficiency means the minimum space occupied by the equipment.

[0038] Specifically, a requirements analysis is conducted on the layout optimization system, analyzing the design requirements for system layout optimization from two aspects: structural characteristic design and maintainability design. Structural characteristics mainly revolve around the size and weight of layable components in the power system, while maintainability focuses on the size of the maintenance space, the type of maintenance tools, the timing of disassembly and assembly sequences, and maintenance path planning. Simultaneously, the design objectives of system layout optimization are clarified: first, a space efficiency objective, guided by minimizing the actual space occupied by the layout scheme, calculating the actual occupied space and calculating the cumulative space value; second, a maintenance efficiency objective, with optimization of disassembly and assembly preparation time as the core, quantifying the time cost of maintenance operations and calculating maintenance time. The layout optimization problem is transformed into a space efficiency and maintenance time efficiency problem. More specifically, the shortest disassembly and assembly preparation time is prioritized while simultaneously minimizing the space occupied.

[0039] Step S3: Construct the system layout optimization model as follows:

[0040] ;

[0041] The model uses a reward function, where α is the dynamic weighting coefficient of the spatial cumulative value, and β is the dynamic weighting coefficient of the maintenance time statistics. For the exploration factor, N is the number of simulations of the parent node, and n is the number of simulations of the current node; the cumulative space value is the sum of the space occupied by the optimized equipment, and the maintenance time statistics value is the time taken for the layout of the vehicle body to be virtually disassembled and repaired according to the maintenance steps specified in the maintenance manual.

[0042] Specifically, the core objective of constructing a system layout optimization model is to improve space utilization and maintenance efficiency. First, the size and weight data of the layable components are extracted, and the layout space is constrained based on the size data. For example, directional bounding boxes (OBB) are used to create component bounding boxes for space collision detection. Second, the layable components are laid out sequentially in the vehicle body space. The laid-out vehicle body is then virtually disassembled and repaired according to the maintenance steps specified in the maintenance manual, and the maintenance time is statistically analyzed.

[0043] Step S4: Set the solution algorithm for the system layout optimization model constructed in Step S3 to achieve efficient solution of the system optimization layout model in Step S3. Solving the layout optimization model oriented towards the parallel design requirements of the system is essentially a two-stage optimization process. On the one hand, since the optimization process of maintainability design will find an optimal result for each feasible solution of the layout optimization problem, it will seriously affect the efficiency of model solution. On the other hand, since the layout optimization problem is based on the premise that the disassembly sequence is known, and the disassembly preparation process between different steps is independent of each other, the optimization process of part of the maintainability design can still be executed in the case of incomplete layout. Based on the above characteristics, combined with the multi-objective characteristics of the system layout optimization model, the layout solution process is transformed into an ordered addition process. By making full use of Monte Carlo tree search technology, a solution technique suitable for system layout optimization problems is designed to achieve efficient solution of the system optimization layout model.

[0044] Includes the following steps:

[0045] Step S401: Starting from the root node, select child nodes according to the following model:

[0046] .

[0047] First, the problem is modeled and the structure mapping of Monte Carlo Tree Search (MCTS) is completed. MCTS searches for the optimal layout result through a cyclical iterative process of four stages: selection, expansion, simulation, and backtracking. The layout optimization problem is transformed into a Monte Carlo Tree Search framework, defining each tree node as the current layout state, i.e., the set of laid-out components and their positions / poses. Child nodes represent component placement operations, selecting unplaced components and determining their positions. The problem is decomposed based on the characteristics of the two stages. In the initial layout stage, a basic space allocation framework is constructed guided by the priority of the disassembly sequence. In the incremental optimization stage, maintainability evaluation is dynamically embedded in some layout states, and the component placement order is adjusted through a backtracking mechanism.

[0048] Subsequently, a multi-objective guided node selection strategy was formulated, combining a dual objective function of space utilization and maintenance time. The weighted selection function, through multi-objective quantification fusion, dynamic weight adjustment, and constraint-sensitive response, achieved synergistic optimization of space efficiency and maintenance effectiveness, while also considering search efficiency and solution quality. Therefore, a weighted selection function was chosen for the multi-objective guided node selection strategy. In system layout optimization, to integrate the dual objectives of space utilization and maintenance efficiency, the upper confidence bound UCB formula is extended as follows:

[0049] .

[0050] Specifically, the node attributes and update strategies in MCTS are defined. First, component attributes are quantified. The spatial cumulative value is calculated based on the ratio of the component envelope volume to the total space of the layout container, using a normalized space coverage index. The maintenance time statistics are calculated by path planning simulation, such as the A* algorithm, to estimate the component disassembly and assembly path length and the maintenance time calculated by the model method, and then convert it into time cost. Simulation count: Record the frequency of node access to balance exploration and utilization. Prioritize access to layout branches with high space utilization, short maintenance time and insufficient exploration.

[0051] Then comes the dynamic update rules. In the backtracking phase, node attributes are updated along the search path, the space accumulation value is accumulated using a rolling time-domain optimization method, and the maintenance time statistics are updated using an exponentially weighted moving average. The weight is adaptive: the α / β ratio is dynamically adjusted according to the current layout completion. Initially, the focus is on space utilization, and later the optimization of maintenance time is gradually strengthened.

[0052] More specifically, α is set to 0.7, β to 0.3, γ to the default value of 1.5, and γ is dynamically adjusted as the search stalls. If there is no improvement after 5 rounds, it will increase by 20%. N is the number of simulations for the parent node, and n is the number of simulations for the current node.

[0053] Step S402: Use an ε-greedy strategy to select branches with potentially high node value.

[0054] Specifically, after completing the layout of branches with high access space utilization, short maintenance time, and insufficient exploration using the UCB algorithm, an ε-greedy strategy is adopted. The optimal value node is selected with a probability of 1-ε, and new branches are explored randomly with a probability of ε. Initially, ε=0.2, and ε decays with the number of iterations. ε←ε×0.95 every 100 iterations. During random exploration, branches with potentially high node value are selected first. The core of the ε-greedy strategy is to use the optimal branch with a high probability and explore new branches with a low probability, so as to find the branch with the highest average return.

[0055] The next steps are the expansion and simulation phases. In the expansion phase, when the number of unlaid components exceeds a threshold, a feasible candidate set (components that satisfy geometric interference constraints) is generated based on the current layout state. Then, a fast simulation strategy is implemented, using a lightweight geometric approximation algorithm (such as axis-aligned bounding boxes) to quickly estimate the layout envelope volume and predict the space occupancy. Finally, based on the constructed assembly / disassembly path probability model, the time cost is predicted by regression based on historical simulation data, thus estimating the maintenance time.

[0056] Step S403: Starting from the newly generated child nodes, simulate the complete layout, complete the placement of the remaining components, and calculate the final node value according to the formula in step S3; during this process, α is gradually adjusted from 0.7 to 0.3, β is gradually adjusted from 0.3 to 0.7, and the sum is kept to be 1.

[0057] Step S404: Reverse update node attributes;

[0058] Finally, backtracking and convergence determination are performed by propagating backward along the search path to update the statistical attributes of all nodes, providing a more reliable basis for subsequent selection stages.

[0059] The update rules are as follows:

[0060] Spatial cumulative value: ;in, The learning rate is η = 0.1. This is the average spatial cumulative value;

[0061] Repair time statistics: ; The attenuation factor is λ = 0.9. This is a statistical value for average repair time.

[0062] Step S405: If the convergence condition is met, output the optimized layout:

[0063] Set the maximum number of iterations or the convergence threshold for the layout evaluation index, and solve for the optimal layout result. The formula for setting the dynamic convergence threshold is as follows:

[0064] ;

[0065] Where t is the current iteration number.

[0066] Based on the above steps, a layout optimization problem-solving algorithm based on Monte Carlo tree search was designed. The entire solution process follows the flow of "selection-expansion-simulation-backtracking" to search for the optimal layout result, thus achieving efficient solution of the system optimization layout model.

[0067] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Any modifications, alterations, substitutions, and variations made by those skilled in the art to the above embodiments are within the scope of the present invention.

Claims

1. A method for optimizing the spatial layout of a powertrain system of an armored vehicle taking into account the impact of disassembly operations, characterized in that, Comprise the following steps: Step S1, in the computer, the virtual maintenance tool, the maintenance object, the human body model of the virtual maintenance personnel is constructed by three-dimensional modeling software, is introduced to 3D rendering engine, the state machine function provided by 3D rendering engine is used to control the action of the imported human body model of virtual maintenance personnel, for the action switching of maintenance operation; Step S2, set layout optimization parameters, including space cumulative value, maintenance time statistics value; Set system layout optimization target, including the highest maintenance efficiency, the highest space efficiency, wherein the highest maintenance efficiency is the shortest disassembly preparation time and maintenance time, and the highest space efficiency is the minimization of equipment occupied space; Step S3, the system layout optimization model is constructed as follows: ； Wherein: α is the dynamic weight coefficient of space cumulative value, β is the dynamic weight coefficient of maintenance time statistical value, N is the number of simulation times of the parent node, n is the number of simulation times of the current node; the space cumulative value is the sum of the sizes of the occupied spaces of the optimized equipment, and the maintenance time statistical value is the virtual disassembly and maintenance time of the layout vehicle body according to the maintenance steps specified in the maintenance manual. Step S4, the solving algorithm of the system layout optimization model constructed in step S3 is set, comprising the following steps: Step S401, starting from the root node, select the subnode according to the following model: ； Step S402, adopt ε-greedy strategy, select potential high node value branch; Step S403, starting from the new subnode generated by extension, simulate complete layout, complete the placement of remaining components, calculate the final node value according to the formula in step S3; Step S404, update node attributes reversely; Along the search path, update the statistical properties of all nodes, and the update rule is as follows: Spatial cumulative value: ; wherein, is a learning rate, is an average spatial cumulative value; Maintenance time statistics value: ; is an attenuation factor, is the average maintenance time statistics value; Step S405, meet the convergence condition, output the optimized layout: Set the maximum iteration number or the convergence threshold of layout scheme evaluation index, solve the optimal layout result, and the setting formula of dynamic convergence threshold is as follows: ； Wherein, t is the current iteration number.

2. The method for optimization of spatial layout of armored vehicle power system taking into account the impact of disassembly operations according to claim 1, characterized in that: In the step S1, the human body model of virtual maintenance personnel is made by Humaonid skeleton standard.

3. The method for optimization of spatial layout of armored vehicle power system taking into account the impact of disassembly operations according to claim 1, characterized in that: In the step S2, the highest maintenance efficiency is set preferentially, and the highest space efficiency is set secondly.

4. The method for optimization of spatial layout of armored vehicle power system taking into account the impact of disassembly operations according to claim 1, characterized in that: In the step S401, the space cumulative value is the ratio based on component envelope volume and total space of layout container, and the normalized space coverage index is adopted.

5. The method for optimization of spatial layout of armored vehicle power system taking into account the impact of disassembly operations according to claim 1, characterized in that: In the step S401, the maintenance time statistics value is converted into time cost by estimating the component disassembly path length and calculating the sum of maintenance time through path planning simulation.

6. The method for optimization of spatial layout of armored vehicle power system taking into account the impact of disassembly operations according to claim 1, characterized in that: In the step S402, the optimal value node is selected with probability 1-ε, and the new branch is randomly explored with probability ε, and ε decays with iteration number, and when randomly exploring, the branch with potential high node value is selected preferentially.

Images