A stainer tree 3D pipeline automatic planning method based on a non-uniform network

Abstract

This invention discloses an automatic 3D pipeline planning method based on Steiner trees using non-uniform networks, relating to the field of new energy. By establishing a mathematical model to define the three-dimensional layout space, multi-level components, interfaces, and their logical connections, the flow channel plate pipeline planning problem is transformed into a constrained optimization problem across scales. In the solution process, a non-uniform quadtree model is used to achieve strong subdivision of interface neighborhoods and sparse sampling of open areas, balancing representation accuracy and computational efficiency. Subsequently, the wiring sequence is reconstructed based on a priority evaluation function, and an improved A algorithm is coupled with the Steiner tree model to achieve dynamic path merging by identifying the optimal branch points of the existing network. This invention achieves global optimization of 100% connectivity and minimum flow resistance performance while satisfying physical constraints.

Description

Technical Field

[0001] This invention relates to the field of new energy, and in particular to an automatic pipeline planning method for 3D pipelines based on Steiner trees with non-uniform networks. Background Technology

[0002] The integrated thermal management system for new energy vehicles is the core component for maintaining the power battery, motor, and electronic control system at their optimal operating temperatures. Its level of integration directly affects the vehicle's range and safety. The flow channel plate, as the physical carrier of this system, needs to achieve precise layout and efficient interconnection of dozens of components such as water pumps, valve bodies, and heat exchangers within a limited space.

[0003] However, existing automated flow channel planning methods still face significant technical bottlenecks in practical applications. Firstly, in terms of spatial representation, traditional methods often employ uniform grid discretization. Given the unique "cross-scale" characteristics of thermal management modules—the need for layouts that combine millimeter-level micro-interfaces with centimeter-level large-size panels—refining the grid to improve accuracy leads to a geometric increase in node size, causing a "computational explosion." Conversely, using a coarse grid makes it difficult to meet the precise interface docking constraints, failing to guarantee a 100% connectivity success rate. Secondly, in terms of path planning logic, existing algorithms are mostly limited to point-to-point independent optimization, lacking a systematic decision-making mechanism for merging multiple pipeline topologies. This results in a lack of effective merging between pipelines, generating significant spatial redundancy and increasing the difficulty of system layout and manufacturing complexity. Finally, limited by the lack of efficient global optimization search tools, existing solutions often stop at obtaining a "feasible solution," failing to achieve deep decoupling and optimal balance between reducing total system flow resistance, minimizing fluid pressure loss, and improving space utilization.

[0004] Therefore, those skilled in the art are dedicated to developing an automatic pipeline planning method based on Steiner trees in non-uniform networks for 3D pipelines. Developing an automatic planning method capable of adaptively representing complex cross-scale spaces, efficiently scheduling multi-level design tasks, and achieving dynamic optimization of flow channel topology has significant engineering application value for overcoming the performance bottlenecks in integrated design of thermal management systems, reducing R&D cycles and manufacturing costs, and solving key technical challenges urgently needing to be addressed in the field of automated design of integrated modules for new energy vehicles. Summary of the Invention

[0005] In view of the above-mentioned deficiencies of the prior art, the technical problem to be solved by the present invention is to adaptively represent complex cross-scale space, efficiently schedule multi-level design tasks, and realize automatic planning for dynamic optimization of flow channel topology.

[0006] To achieve the above objectives, this invention provides an automatic pipeline planning method for 3D pipelines based on a non-uniform network using Steiner trees, including establishing routing priorities for core components using functional levels and coupling improved A... The algorithm and Steiner tree model are used to dynamically absorb pipelines into the existing network and optimize branch points.

[0007] Furthermore, this includes task priority scheduling, constructing a priority evaluation function, establishing a timing sequence of core exits first, then general connections and shortest paths at the same level first, performing task serialization, and avoiding deadlocks and interference.

[0008] Furthermore, this includes improved A based on Steiner trees. The optimization process defines existing pipelines as a dynamic target set, guides new paths to the network based on Steiner tree logic to perform dynamic adsorption, and automatically establishes branch points at the lowest cost to optimize the topology.

[0009] Furthermore, a non-uniform quadtree representation framework is adopted, which adaptively switches between a forced subdivision strategy for the interface neighborhood or a sparse sampling strategy for open areas based on geometric features.

[0010] Furthermore, the non-uniform quadtree performs recursive subdivision in the interface neighborhood and sparse sampling in the open area.

[0011] Furthermore, this includes task timing scheduling and dynamic topology optimization.

[0012] Furthermore, it includes the following steps: Step 1: Extract features from the layout, calculate the span of each path using Euclidean metric, and simultaneously identify the physical attributes of the interface entrances and exits, mapping them to energy level factors that characterize the constraint strength. Step 2: Construct a priority evaluation function based on functional level, enforce the core exit first, core entry second, and shortest path first strategy at the same level, and realize the serialization and reorganization of global routing tasks; Step 3: Discretize the planning space using a non-uniform quadtree, perform forced subdivision in the interface neighborhood to ensure precise docking, and perform sparse sampling in the open area to compress the node size. Step 4: Optimize according to the task sequence, and improve A. The algorithm defines the existing pipeline as a dynamic target set, combines Steiner tree topology logic to detect the existing network in real time, automatically identifies and establishes the optimal branch point at the point where the cost function is minimized, guides new paths to perform dynamic adsorption to the existing network, and outputs the solution that minimizes the total flow resistance.

[0013] Furthermore, in step 2, by establishing a priority evaluation function, the wiring task is transformed into a task flow with ordered energy levels.

[0014] Furthermore, in step 3, the spatial representation of the non-uniform quadtree reduces the dimensionality of the three-dimensional layout solution space into a series of two-dimensional topological planes with physical hierarchical constraints, and performs environmental discretization and feature mapping through multi-scale recursive subdivision.

[0015] Furthermore, in step 4, the incremental adaptive evolution of the flow channel system is carried out based on point set-driven Steiner tree topology merging and a strategic circuit breaker mechanism is introduced.

[0016] Addressing the dynamic topology merging problem, this invention utilizes functional levels to establish routing priorities for core components and couples improved A... The algorithm and Steiner tree model enable dynamic absorption of pipelines into the existing network and optimization of branch points. This invention includes: 1. Task priority scheduling: Constructing a priority evaluation function to establish a time sequence of "core exits first, then general connections" and "shortest path first" for peers, achieving task serialization to avoid deadlock and interference. 2. Improved A based on Steiner tree. Optimization: Define existing pipelines as a dynamic target set, guide new paths to the network based on Steiner tree logic to perform dynamic adsorption, and automatically establish branch points at the lowest cost to optimize the topology.

[0017] For spatial adaptive representation, this invention employs a non-uniform quadtree representation framework, adaptively switching between a forced subdivision strategy for the interface neighborhood and a sparse sampling strategy for open areas based on geometric features. The non-uniform quadtree of this invention performs recursive subdivision in the interface neighborhood to ensure precise docking, and performs sparse sampling in open areas to compress the search space.

[0018] Compared with the prior art, the present invention has the following obvious substantive features and significant advantages: 1. This invention effectively avoids geometric interference and wiring conflicts of multiple pipelines in narrow spaces, reduces redundant fluid paths through dynamic topology merging, minimizes the total flow resistance of all channels, and optimizes the total flow resistance performance of the system.

[0019] 2. This invention achieves efficient compression of the three-dimensional solution space, reducing the global number of nodes to be searched and improving pipeline planning efficiency while ensuring 100% interface connectivity accuracy.

[0020] The following will further explain the concept, specific structure, and technical effects of the present invention in conjunction with the accompanying drawings, so as to fully understand the purpose, features, and effects of the present invention. Attached Figure Description

[0021] Figure 1 This is a general framework diagram of a preferred embodiment of the present invention; Figure 2 This is a schematic diagram of a non-uniform quadtree spatial partitioning. Figure 3 This is a schematic diagram of node topology extraction based on a mesh. Figure 4 This is a schematic diagram of the end-to-end isolated pipeline planning results; Figure 5 This is a schematic diagram of the pipeline adsorption and merging results based on Steiner trees; Figure 6 This is a comparison chart of search efficiency for the pipeline planning task in the basic task (Case A); Figure 7 This is a comparison chart of search efficiency for pipeline planning tasks in high-density integration tasks (Case B); Figure 8 This is a top-down view of the 3D pipeline planning results in a high-density integrated environment, as shown in Case B. Figure 9 This is a 3D pipeline planning result diagram (isoaxial view) in the high-density integrated environment of Case B. Detailed Implementation

[0022] The following description, with reference to the accompanying drawings, illustrates several preferred embodiments of the present invention to make its technical content clearer and easier to understand. The present invention can be embodied in many different forms, and the scope of protection of the present invention is not limited to the embodiments mentioned herein.

[0023] In the accompanying drawings, components with the same structure are indicated by the same numerical designation, and components with similar structures or functions are indicated by similar numerical designations. The dimensions and thicknesses of each component shown in the drawings are arbitrary, and the present invention does not limit the dimensions and thicknesses of each component. To make the illustrations clearer, the thickness of some components has been appropriately exaggerated in the drawings.

[0024] This invention provides an automatic pipeline planning method based on the merging of non-uniform quadtrees and Steiner trees. By establishing a mathematical model to define a three-dimensional layout space, multi-level components, interfaces, and their logical connections, the flow channel plate pipeline planning problem is transformed into a constrained optimization problem across scales. In the solution process, a non-uniform quadtree model is used to achieve strong subdivision of interface neighborhoods and sparse sampling of open areas, balancing representation accuracy and computational efficiency. Subsequently, the routing timing is reconstructed based on a priority evaluation function, and an improved A... The algorithm, based on the Steiner tree model, dynamically merges paths by identifying the optimal branch points in an existing network. This invention achieves global optimization of 100% connectivity and minimum flow resistance while satisfying physical constraints.

[0025] I. Establishment of the flow channel model

[0026] (1) Symbol description: A collection of all components The shape of component m Components A collection of all interfaces, for each component Includes several interfaces Components The One interface, The collection of all pipeline connections Component The Interfaces and components The Interface connection,

[0027] (2) Problem parameters: Representation Component The The relative coordinates of the component center of each interface : indicates a component The appearance can be represented by a set of coordinates of the component's shape. Component interface radius Pipeline connection radius Minimum clearance between heat pipes and other components or pipes. Width and height of the flow channel plate (3) Decision variables: Representation Component The absolute coordinates of the geometric center on the flow channel plate, Representation Component interface Absolute coordinates on the flow channel plate, Representation Component The Interfaces and components The The route connected by each interface

[0028] (4) Optimization objective:

[0029] in Indicates from component The Interfaces and components The The flow resistance on the path of each interface.

[0030] (5) Pipeline constraints: ① Panel size constraints: All components must be laid out on the panel; ② Fixed component constraints: Some components are fixed in specific positions on the panel; ③ Non-interference of components: Each component occupies a specific volume space. In principle, these components must meet the non-interference condition in physical layout, that is, they cannot overlap in space, so as to ensure that all components can be arranged on the flow channel plate.

[0031] ,

[0032] (Some components, due to their specific space-saving installation methods (such as the WCC which can be rotated to be outside the panel), do not need to consider interference between their shape and other components.)

[0033] ④ Irreversibility of flow channels: Any newly planned flow channel must not cross other component interfaces other than its own source interface and target interface, nor may it cross other planned flow channels.

[0034] (Except for the starting point and the end point)

[0035] ⑤ Flow channel plate layer constraint: The maximum number of flow channel plates limits the vertical space that the flow channel design can span. During the process of searching for feasible connecting flow channels layer by layer, it is essential to strictly ensure that the number of layers containing the flow channel does not exceed the maximum number of layers of the flow channel plate to guarantee the effectiveness and feasibility of the design.

[0036] ⑥ Thermal Pipe Spacing Constraints: To ensure the stability and independence of fluid temperatures within pipes with different temperature properties, thermal pipe design must adhere to strict spacing requirements. Thermal pipes must maintain a certain safe distance from all surrounding flow channels and component interfaces. This distance is determined based on heat conduction theory, fluid dynamics principles, and practical engineering experience.

[0037] ⑦ Pump flow channel layer constraint: The number of flow channels at the pump outlet must be higher than or equal to the number of flow channels at the pump inlet.

[0038] ⑧ Constraint on the number of flow channels in a one-way valve: The number of flow channels at the outlet of a one-way valve shall not be less than the number of flow channels at the inlet of the one-way valve.

[0039] II. Pipeline Optimization Design

[0040] (1) Overall framework design

[0041] like Figure 1 As shown, the pipeline planning problem is a typical strongly NP-hard problem. The main challenge lies in the high coupling effect between pipeline routing and space occupancy under cross-scale constraints: the physical distribution of component interfaces determines the initial state of the solution space, while topology merging between multiple pipelines directly affects the total flow resistance performance of the system. This invention decomposes the solution process into two stages: task timing scheduling and dynamic topology optimization. The specific steps are as follows: 1. Extract features from the existing layout, calculate the span of each path using Euclidean metric, and simultaneously identify the physical attributes of the interface entrances and exits, mapping them to energy level factors that characterize the constraint strength.

[0042] 2. Construct a priority evaluation function based on functional level, enforce the "core exit first, core entry second" and "shortest path first" strategies at the same level, and realize the serialization and reorganization of global routing tasks.

[0043] 3. Discretize the planning space using a non-uniform quadtree, perform forced subdivision in the interface neighborhood to ensure precise docking, and perform sparse sampling in open areas to compress node size.

[0044] 4. Optimize according to the task sequence, and improve A. The algorithm defines the existing pipeline as a dynamic target set, combines Steiner tree topology logic to detect the existing network in real time, automatically identifies and establishes the optimal branch point at the point where the cost function is minimized, guides new paths to perform dynamic adsorption to the existing network, and outputs the solution that minimizes the total flow resistance.

[0045] (2) Cabling task priority scheduling

[0046] To address the spatial conflict problem caused by multiple interfaces, a task scheduling strategy based on multiple constraint features is proposed. This strategy aims to transform wiring tasks into an energy-ordered task flow by establishing a priority evaluation function, thereby logically avoiding topology deadlock.

[0047] First, define the original set of routing tasks as follows: In the feature perception stage, the span of each path is calculated using Euclidean metrics. Simultaneously identify the physical attributes of the entrances and exits of the path and construct a priority evaluation function. Reconstruct the task execution sequence:

[0048] Among them, the weighting coefficient The value of is directly related to the constraint severity level of the task. This strategy introduces a dynamic partitioning mechanism based on functional energy levels: for the outlet tasks of core components with strict layer constraints, such as water pumps and check valves (high energy level), is assigned ... Extremely high weights are assigned to ensure that constraint stringency dominates scheduling; inlet tasks of core components (medium-level) have the next highest weight; while general connection tasks such as heat exchanger interfaces (low-level) have the lowest weight. The minimum value is chosen to ensure that the path length factor plays a major guiding role in tasks of the same level.

[0049] This mechanism establishes a topological sequence of "core exit first, core entrance second," preventing critical channels from being prematurely occupied. Within tasks of the same level, a "shortest path first" strategy is triggered, using short paths to quickly fill compact spaces and reduce search branching factors. Furthermore, this strategy simultaneously identifies task characteristics of paths originating from or converging on the same source, reserving logical entry points for subsequent topology merging, thus achieving coordinated optimization of constraint stringency and space utilization.

[0050] (3) Spatial representation of non-uniform quadtrees

[0051] To address the physical characteristics of the thermal management integrated module, which features highly uneven obstacle distribution and intricate local geometric features, a hierarchical non-uniform quadtree spatial representation model is constructed. This model reduces the dimensionality of the complex three-dimensional layout solution space into a series of two-dimensional topological planes with physical hierarchical constraints, and achieves environmental discretization and feature mapping through multi-scale recursive subdivision.

[0052] 1. Subdivision Criteria for Two-Way Constraints

[0053] In traditional spatial partitioning, a single resolution often leads to a trade-off between computational efficiency and geometric accuracy. To address this issue, this invention proposes an adaptive subdivision algorithm driven by both geometric boundary sensitivity and physical vector constraints.

[0054] First, complex boundaries are detected in real time using an envelope detection function. When a sampling unit covers the component outline or a no-access zone, recursive subdivision is immediately triggered until the smallest granularity is reached, thus ensuring that the pipeline has accurate obstacle avoidance margin. For physical interfaces such as water pumps that have strict outlet pipe angle requirements, a neighborhood forced subdivision mechanism is introduced: by monitoring the Euclidean distance between the grid and the interface center in real time, it is ensured that the interface area can be locked at the highest resolution level even in open space, thus avoiding the "jagged" distortion that occurs in the early stages of path planning from the source.

[0055] At the logical representation level, the physical environment is transformed into a non-uniform quadtree structure through binarized sampling. The root node recursively branches downwards based on obstacle density, allowing open areas to be retained as large-scale leaf nodes to compress the search space, while geometric boundaries and interface neighborhoods evolve into control points with the smallest granularity. Furthermore, the algorithm sets a maximum side length limit for open areas, effectively preventing global displacement vector distortion caused by local refinement. This "locally precise, globally sparse" design not only ensures millimeter-level accuracy for interface docking but also implicitly achieves smooth global paths by controlling the mesh step size, realizing a deep integration of representation efficiency and physical constraints.

[0056] 2. Multi-scale node topology extraction and inter-layer association

[0057] After performing non-uniform quadtree partitioning, a discretized graph model is constructed by extracting the geometric features of leaf nodes at various scales. This transforms static geometric blocks into a connected topological search network. Leaf nodes from all non-obstacle regions in each plane are extracted as vertices of the search graph. The position of each vertex is defined as the geometric center of a quadtree node, where each node... It was given a multidimensional attribute vector :

[0058] in, The coordinates of the center within the layer, The scale factor corresponds to the grid subdivision level. This represents the region attribute (such as component interface, free space, or hard restricted area).

[0059] To address the scale abrupt changes between non-uniform grids, this invention proposes a dynamic neighborhood indexing algorithm based on edge-stepping detection. Discrete sampling is performed along the current node boundary with a minimum step size. An addressing function quickly captures leaf nodes of varying scales within adjacent regions, ensuring good connectivity when traversing regions of different resolutions and effectively avoiding vector distortion. For 3D spatial connectivity, vias are used as "bridging vertices" to achieve cross-layer dual indexing, stitching discrete 2D path segments into continuous 3D topological lines. This model dynamically weights the search step size using a scale factor, compressing the global search nodes on a large scale while maintaining high-density integrated spatial precision, effectively overcoming the computational bottleneck across scales.

[0060] Figure 2 This is a schematic diagram of a non-uniform quadtree spatial partitioning. Figure 3 This is a schematic diagram of node topology extraction based on a mesh. (4) Improved A based on Steiner tree Optimization Algorithm In the design of complex 3D integrated flow channel plates, traditional point-to-point optimization algorithms often fall into the "first-come, first-served" local optimum dilemma. Because the initially planned piping occupies valuable space, subsequent piping often generates numerous redundant bends to avoid obstacles, leading to a significant increase in the overall system pressure drop. To address this, a point-set-driven Steiner tree topology merging mechanism is proposed, along with a strategic circuit breaker mechanism, aiming to achieve incremental adaptive evolution of the flow channel system under high-density layouts.

[0061] 1. Strategic Circuit Breaker

[0062] Unlike traditional unrestricted heuristic search, the model will expand the maximum number of nodes. Considered a hard constraint for evaluating the feasibility of current physical layer routing, its mathematical definition is:

[0063] in, The characteristic distance between the starting point and the target. This represents the size of the operator action space (i.e., neighborhood connectivity) of the current node in a non-uniform grid. and These represent the path complexity coefficient and local detour tolerance, respectively, determined through prior experiments on typical industrial layout cases. In practical applications, the coefficients... The selection of is usually positively correlated with the average compression rate of the non-uniform mesh, used to compensate for the additional node overhead generated by long-span paths in the subdivided regions, while the bias term This provides basic search redundancy for navigating complex obstacles. Essentially, the model establishes a relationship between geometric span, grid resolution, and spatial topological complexity, enabling dynamic allocation of search resource quotas.

[0064] During the optimization process, the algorithm dynamically monitors the expanded node scale in real time. If A The algorithm has achieved a breakthrough in the number of nodes expanded at the current physical layer. If connectivity is still not achieved by the threshold, the spatial congestion of the current physical layer is determined to exceed the flow resistance tolerance threshold, meaning that a high-quality path meeting flow resistance requirements cannot be generated within a limited cost. At this point, the algorithm triggers the circuit breaker operator, terminating the invalid lateral expansion and initiating the vertical layer switching logic to guide the search process to jump to the adjacent physical layer.

[0065] 2. Incremental Topological Adsorption Model

[0066] The construction process of Steiner trees is abstracted into the construction of non-uniform quadtree graphs. This addresses the subgraph optimization problem. It overcomes the limitations of traditional isolated "end-to-end" planning by redefining the starting and ending conditions of the search to achieve topological evolution of the path.

[0067] First, a multi-source concurrent initialization strategy is adopted. During the initialization phase of the optimization operator, not only are the component interfaces of the current task used as the starting source, but also key topology nodes in the planned path are extracted by traversing the global topology information database to construct a multi-source starting set. By utilizing priority-based scheduling of Open Set containers, All candidate nodes are pushed into the search heap simultaneously.

[0068] To guide paths toward existing networks, the cost distribution of the open set follows a multi-source mapping rule. For the initial set... The initial path cost of all candidate nodes within it. The initial cost is uniformly and explicitly defined as 0, while for the remaining nodes in the space, the initial cost is set to infinity. ):

[0069] This model simulates the "multicentric growth" phenomenon in physics, mathematically constructing a zero-potential field centered on multiple initial topological points. This allows the optimization process to expand outwards in parallel from multiple potential branch points simultaneously, greatly increasing the probability of finding locally optimal access points in high-density spaces.

[0070] Secondly, a dynamic target point set-driven logic is introduced. All discrete sampling nodes of the planned path are defined as a dynamic target set. During heuristic optimization iteration, the heuristic term in the evaluation function... Instead of pointing to a single fixed endpoint, it solves the current node in real time. to set Minimum estimated cost of the nearest access point:

[0071] This modeling approach simplifies complex topology growth into a "dynamic adsorption" of new branches onto the existing network point set. At the logical execution level, once a search branch touches any node in the target set, the current task is immediately determined to be connected, and the touch point is identified as a Steiner branch point in real time. This incremental merging mechanism can adaptively generate ternary or multi-branch topologies, significantly reducing the total physical length of the flow system while ensuring connectivity.

[0072] Comparative case studies show that the topology merging algorithm, by dynamically identifying logical branch points, not only avoids local congestion deadlocks at the interface that occur with traditional algorithms, but also significantly reduces the physical space occupied by the flow channel board while ensuring 100% connectivity, providing more layout margin for the modular integration of subsequent thermal management components.

[0073] Figure 4 This is a schematic diagram of the end-to-end isolated pipeline planning results; Figure 5 This is a schematic diagram of the pipeline adsorption and merging results based on Steiner trees; To verify the performance of the non-uniform quadtree model in representing complex and constrained spaces, a basic task (Case A) and a high-density ensemble task (Case B) were selected for algorithm performance comparison. The basic task (Case A) is as follows: Figure 6 As shown, high-density integration tasks (Case B) are as follows: Figure 7 As shown in the figure. Experimental results show that the non-uniform quadtree significantly reduces the search space size by merging the scales of open areas. In the basic case A, the number of nodes to be searched decreased from 73,260 to 3,270, with a compression rate of 95.5%; in the high-difficulty case B, the number of nodes was sharply reduced from 316,282 to 12,526, further improving to 96.0%. This "locally rigorous, globally sparse" design deeply couples the representation granularity with obstacle features, eliminating search redundancy from the underlying architecture and greatly reducing memory load.

[0074] In terms of planning efficiency, the comparison of computation time reflects the model's superior performance in handling the "computational explosion" problem. For Case A, which has a relatively simple structure, the optimization time is reduced from 2 minutes and 52 seconds to 1 minute and 10 seconds, improving efficiency by about 1.5 times. For Case B, which has a huge search space and complex constraint logic, the traditional algorithm takes 8 minutes and 53 seconds as the complexity increases, while the algorithm in this paper can complete the global planning in only 1 minute and 3 seconds, with a computational efficiency leap of 8.4 times.

[0075] It is worth noting that as the scene complexity increases from A to B, the time consumption of traditional algorithms skyrockets by nearly four times, while the algorithm of this invention exhibits extremely strong stability in time consumption. This proves that complex physical constraints (such as angle and layer number limitations) play a "search pruning" role in non-uniform grids, giving the algorithm a significant "complexity immunity" characteristic. This makes the algorithm highly practical in engineering applications when facing the increasingly stringent integration requirements of thermal management systems for new energy vehicles.

[0076] Case 1

[0077] enter: Panel size: [380, 260] Component information: { Component symbol: "5WV", Number of interfaces: 5 Component area: 8100 "Interface radius": 8, "interface": [ {"Interface ID": 1,"Interface relative position coordinates":[0,-29.5], "Interface connection angle range": [0,360]}, {"Interface ID": 2,"Interface Relative Position Coordinates":[0,0], "Interface Connection Angle Range": [0,360]}, {"Interface ID": 3,"Interface relative position coordinates":[0,29.5], "Interface connection angle range": [0,360]}, {"Interface ID": 4,"Interface Relative Position Coordinates":[-29.5,0], "Interface Connection Angle Range": [0,360]}, {"Interface ID": 5,"Interface Relative Position Coordinates":[29.5,0], "Interface Connection Angle Range": [0,360]} ], "Polygonal shape": [ {"Outline Number": 1, "Relative Coordinates of Outline Position": "(-45, -45), (45, -45), (45, 45),(-45, 45)"} ] } { "Component Symbol": "ECP", Number of interfaces: 2, Component area: 9503 "Interface radius": 8, "interface": [ {"Interface ID": 1,"Interface Relative Position Coordinates":[0,0], "Interface Connection Angle Range": [0,360],"Interface Attributes":"Entry Point"}, {"Interface ID": 2,"Interface Relative Position Coordinates":[0,-34.5], "Interface Connection Angle Range":[120,122],"Interface Attributes":"Exit", "inner_ring_radius": 34.5} ], "Circular shape": [ {"Outline Number": 1, "Outline Radius": 34.5}, {"Outline Number": 2, "Outline Radius": 55} ] } { "Component symbol": "EC", Number of interfaces: 2, Component area: 17 "Interface radius": 8, "interface": [ {"Interface ID": 1,"Interface relative position coordinates":[0,0], "Interface connection angle range": [0,360]}, {"Interface ID": 2,"Interface Relative Position Coordinates":[44,0], "Interface Connection Angle Range": [0,360]} ] } { "Component Symbol": "EW", Number of interfaces: 2, Component area: 17 "Interface radius": 8, "interface": [ {"Interface ID": 1,"Interface relative position coordinates":[0,0], "Interface connection angle range": [0,360]}, {"Interface ID": 2,"Interface Relative Position Coordinates":[0,137], "Interface Connection Angle Range": [0,360]} ] } { "Component symbol": "P", Number of interfaces: 1, Component area: 200 "Interface radius": 8, "interface": [ {"Interface ID": 1,"Interface Relative Position Coordinates":[0,0], "Interface Connection Angle Range": [0,360]} ] } { "Component symbol": "1WV", Number of interfaces: 2, Component area: 2551 "Interface radius": 8, "interface": [ {"Interface ID": 1,"Interface Relative Position Coordinates":[0,0], "Interface Connection Angle Range": [0,360]} ], "Circular shape": [ {"Outline Number": 1, "Outline Radius": 28.5} ] } Location information:

[0078] Connection information:

[0079] Figure 8 This is a top-down view of the 3D pipeline planning results in a high-density integrated environment, as shown in Case B.

[0080] Figure 9 This is a 3D pipeline planning result diagram (isoaxial view) in the high-density integrated environment of Case B.

[0081] The preferred embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make numerous modifications and variations based on the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of existing technology should be within the scope of protection defined by the claims.

Claims

1. An automatic pipeline planning method for 3D pipelines based on Steiner trees with non-uniform networks, characterized in that, This includes using functional levels to establish routing priorities for core components and improving coupling A. The algorithm and Steiner tree model are used to dynamically absorb pipelines into the existing network and optimize branch points.

2. The Steiner tree 3D pipeline automatic planning method based on non-uniform networks as described in claim 1 includes task priority scheduling, constructing a priority evaluation function, establishing a time sequence of core exit first, then general connections and shortest paths of the same level first, performing task serialization, and avoiding deadlock and interference.

3. The Steiner tree-based 3D pipeline automatic planning method according to claim 1, including an improved A method based on Steiner trees. The optimization process defines existing pipelines as a dynamic target set, guides new paths to the network based on Steiner tree logic to perform dynamic adsorption, and automatically establishes branch points at the lowest cost to optimize the topology.

4. The Steiner tree-based 3D pipeline automatic planning method based on non-uniform networks as described in claim 1 adopts a non-uniform quadtree representation framework and adaptively switches between a forced subdivision strategy for the interface neighborhood or a sparse sampling strategy for open areas based on geometric features.

5. The Steiner tree 3D pipeline automatic planning method based on non-uniform networks as described in claim 4, wherein the non-uniform quadtree performs recursive subdivision in the interface neighborhood and sparse sampling in the open area.

6. The Steiner tree 3D pipeline automatic planning method based on non-uniform networks as described in claim 1, including task timing scheduling and dynamic topology optimization.

7. The Steiner tree-based 3D pipeline automatic planning method according to claim 1, comprising the following steps: Step 1: Extract features from the layout, calculate the span of each path using Euclidean metric, and simultaneously identify the physical attributes of the interface entrances and exits, mapping them to energy level factors that characterize the constraint strength. Step 2: Construct a priority evaluation function based on functional level, enforce the core exit first, core entry second, and shortest path first strategy at the same level, and realize the serialization and reorganization of global routing tasks; Step 3: Discretize the planning space using a non-uniform quadtree, perform forced subdivision in the interface neighborhood to ensure precise docking, and perform sparse sampling in the open area to compress the node size. Step 4: Optimize according to the task sequence, and improve A. The algorithm defines the existing pipeline as a dynamic target set, combines Steiner tree topology logic to detect the existing network in real time, automatically identifies and establishes the optimal branch point at the point where the cost function is minimized, guides new paths to perform dynamic adsorption to the existing network, and outputs the solution that minimizes the total flow resistance.

8. The Steiner tree 3D pipeline automatic planning method based on non-uniform networks as described in claim 7, wherein step 2, by establishing a priority evaluation function, transforms the wiring task into an energy-ordered task flow.

9. The Steiner tree 3D pipeline automatic planning method based on non-uniform networks as described in claim 7, wherein step 3, the spatial representation of the non-uniform quadtree, reduces the dimensionality of the three-dimensional layout solution space into a series of two-dimensional topological planes with physical hierarchical constraints, and performs environmental discretization and feature mapping through multi-scale recursive subdivision.

10. The Steiner tree 3D pipeline automatic planning method based on non-uniform network as described in claim 7, wherein step 4 is based on point set-driven Steiner tree topology merging and introduces a strategic circuit breaker mechanism to perform incremental adaptive evolution of the flow channel system.

Images