A branch harness path planning and optimization method and system for multiple wiring process constraints

Abstract

The present invention proposes a branch harness path planning and optimization method and system for multiple wiring process constraints, which is suitable for multi-source path planning and layout optimization of cables. The method is based on the A*-ant colony search algorithm. Under the constraints of multiple wiring processes, it weighs the influence of local path turns and global path length on wiring quality, and solves the problems of inaccurate design data, unreasonable wiring, and low change efficiency in traditional wiring modes; at the same time, a closed loop elimination method based on undirected graph expression is proposed, which solves the structural interference problem caused by local closed loops in the multi-source path planning of branch harnesses. There is no need to assume the number and position of branches in advance, and the branch paths and branch structures of the harnesses can be optimized in terms of length, turns, bundling, etc., so as to realize the path planning of branch harnesses with large size, long span, large number, and wide distribution in complex environments. The feasibility and effectiveness of the method of the present invention are verified through relevant application examples.

Description

Technical Field

[0001] The present invention relates to a branch harness path planning and optimization method and system oriented to multiple wiring process constraints. A typical example is applicable to multi-source path planning and layout optimization of aircraft cables. Background Art

[0002] The aviation cable interconnection system EWIS is spread throughout all areas of the aircraft and runs through the entire stage of aircraft development. It organically connects various systems such as fuel, environmental control, flight control, and avionics to achieve effective transmission of various signals, commands, and energy to ensure the normal and safe flight of the aircraft. The quality of cable layout design plays a decisive role in aircraft functional indicators, finished product delivery, and safety of use.

[0003] The EWIS wiring design involves a huge number of cables and a heavy design task. Due to the compact structure of the aircraft's internal environment, numerous functional systems, and complex wiring rules, wiring design and verification currently rely mostly on engineers' design experience and physical trial installation, which can easily lead to unreasonable cable layouts, resulting in constant rework and thus extending the delivery cycle. At the same time, the design of EWIS is highly dependent on the space occupied by other systems and structures. Once the structure undergoes a slight change, it may have a huge impact on the cable wiring path, etc. With the development of aircraft systems, traditional cable laying methods can no longer meet the requirements of precise design and efficient assembly of EWIS cables. It is necessary to design an automated layout method for EWIS cables to optimize the manufacturing and process costs of the entire machine cable, and to improve wiring efficiency while ensuring wiring quality.

[0004] The automatic layout problem of multi-branch wire harnesses is an NP problem, which requires finding the optimal layout plan that optimizes both the wiring path and the branch structure of the wire harness. Both the wiring path and the branch structure are unknown quantities and are mutually optimized indicators. The set of multiple wire paths determines the specific branch structure, and as the branch structure is optimized, the wire paths also change accordingly, which brings challenges to solving the problem. At the same time, the particularity and complexity of the onboard wiring environment also bring difficulties to the automatic arrangement of flexible cables. On the one hand, the wire harness is long, the number of wires is large, and the distribution range is wide, so the model geometry analysis and calculation are time-consuming; on the other hand, the cable path is highly dependent on the rigid structure, and the narrow wiring space has a great influence on the cable path selection, turning conditions, etc.

[0005] At present, there are few related algorithms and achievements at home and abroad, and the main research focuses on the improvement of non-deterministic search algorithms, such as GA algorithm, ACO algorithm, PSO algorithm, etc. The multi-branch harness layout problem is converted into a minimum Steiner tree problem for solution, and the harness branch path results are determined and optimized through population evolution and iteration. The research process gradually approaches engineering practice from theoretical exploration. However, in the specific harness layout process, it is necessary to assume the number and position of branch points in advance, and then use the branch points as the source point for path planning, without considering the subsequent optimization of specific branch positions and branch structures. At the same time, the current method fails to consider the impact of actual engineering constraints such as the occupancy of the laid harness on the path selection of the current harness to be laid, and the wiring verification is carried out in specific parts or concentrated areas, and the adaptability to large-size, long-span, large-number, and widely distributed airborne cable wiring application scenarios is not high.

[0006] Therefore, the present invention aims at the problems of inaccurate design data, unreasonable wiring and low change efficiency existing in the traditional wiring harness design mode, and combines the limitations of the existing automatic layout method that is not suitable for automatic wiring of large quantities of cables in complex environments. A branch wiring harness path planning and optimization method for multiple wiring process constraints is proposed. This method can achieve the overall optimal process layout of multi-branch cables in complex environments, obtain wiring results with excellent wire paths and branch structures, and effectively improve wiring speed and wiring quality. Summary of the invention

[0007] The present invention provides a branch harness path planning and optimization method for multiple wiring process constraints. The method is based on the actual wiring process requirements, performs path planning for multi-branch harnesses based on the A*-ant colony algorithm, and proposes a closed loop elimination method based on undirected graph expression to locally optimize the harness layout, thereby avoiding the interference with the structure caused by the closed loop during the laying process of the long-span harness, and optimizing the wiring path. Therefore, the method does not need to pre-set the number and position of branch points, and can effectively realize the joint optimization of the wire path, branch structure and branch path. At the same time, it solves the problem of low search efficiency of large-size cables in complex environments, and weighs the impact of process constraints on path selection in terms of path length, turning, bundling, electromagnetic compatibility, etc., so as to optimize the wiring length and improve the wiring quality and efficiency.

[0008] The technical solution adopted by the present invention is:

[0009] A branch harness path planning and optimization method for multiple wiring process constraints includes the following specific contents:

[0010] 1) According to the constraints of multiple wiring processes, based on the A*-ant colony path search algorithm, the optimization target and concentration increment model are designed to improve the turning quality of the cable path and optimize the path length cost; the path planning of all cables in the harness is realized;

[0011] 2) According to the path results obtained in 1), a closed loop elimination method based on undirected graph expression is used to locate and divide the loops in the adjacency table for the closed loop structures existing in the multi-branch harness wiring results. By judging the type of the loop, the branches within the loop are split or merged to further optimize the harness branch path and branch structure.

[0012] Among them, the first aspect is the path planning method, which combines various constraints of the wiring process to design the optimization goal and iteration model of the A*-ant colony search algorithm to improve the turning quality of the cable path and optimize the path length cost. Specifically including:

[0013] Take each discrete point in the three-dimensional environment map as a feasible point for path search, assuming that the starting point of the path is P start , the search target node is P end , the current state node is P n Next, search for node P n+1 You can choose to search for adjacent nodes with different moving directions within the step range.

[0014] (1) Combined with actual wiring process constraints such as specific physical constraints, turning constraints, support constraints, electromagnetic compatibility constraints and bundling constraints, the heuristic function H between adjacent nodes in the ant colony algorithm is designed. ij (t), as shown in formula 1.1:

[0015]

[0016] f(n)=g(n)+h(n) (1.2)

[0017] f(n) is introduced as the evaluation function of the current node of the A* algorithm, and its specific definition is shown in formula 1.2. g(n) is the cumulative cost from the starting point to the current node n, and h(n) is the shortest distance from node n to the target point. Introducing the evaluation function of the A* algorithm in the ant colony algorithm provides a global inspiration direction for the algorithm and avoids blind search in the early stage of the algorithm.

[0018] The cable turning evaluation function bend(n) is introduced to evaluate the turning angle θ of the cable when selecting the next search node. n,n+1 With bending radius r n,n+1 Impact on the quality of the wiring path.

[0019] Introducing the clamp capacity evaluation function clip(D n,n+1 ) to reflect the impact of the completed cables on the path search of the cables to be laid.

[0020] Introducing the electromagnetic compatibility evaluation function emc(e n,n+1) to reflect the impact of laying channels with different EMC category codes on the current cable path point selection.

[0021] Introduce the bundle node evaluation function bundle(b n,n+1 ) to reflect the impact of the selected path point on the current cable path point selection.

[0022] (2) Based on the heuristic function between nodes, calculate the probability of an ant transferring from the current node i to the next node j according to Formula 1.3.

[0023]

[0024] Among them, τ ij (t) is the pheromone concentration between node i and node j at time t, H ij (t) is the heuristic function between node i and node j, α and β respectively reflect the influence of the function on the state transition probability. n is any node that the ant can choose in the next step, allowed m is the set to be visited by the current ant. The ant selects the next node according to the roulette method, so the heuristic function H ij The larger the (t) value is, the greater the probability that the ant will choose this path point.

[0025] (3) After each iteration, the pheromone concentration on the path taken by the ants is updated. The better the path, the higher the pheromone concentration increment. The pheromone concentration increment model is expressed as Equation 1.4.

[0026]

[0027] Among them, ρ is the pheromone volatility coefficient, τ ij (t-1) is the pheromone concentration between node i and node j at time t-1, Δτ ij is the total pheromone accumulated between (i, j) after this iteration, and M is the number of ants. The pheromone concentration increment of each ant m between (i, j) is updated using global information, and the specific definition is as shown in Formula 1.5.

[0028]

[0029] In addition to the path length cost L m In addition, the number of path inflection points T is introduced m Accumulation of turn evaluation m bend(n) ratio as the path turning cost, with the number of repeated path points N bm As the path bundling cost evaluation, c L 、c t With c bis the weight of each cost. In the formula, k represents the pheromone concentration coefficient; Q is the pheromone intensity coefficient, which is related to the path search step length and is defined as follows:

[0030] Q=L max -L min (1.6)

[0031] Where, L max is the length of the maximum path generated after this iteration, L min is the length of the shortest path generated after this iteration.

[0032] The average length of the paths experienced by all ants in this iteration is recorded as When L m =L max When k takes the minimum value k min = 0, reducing the increase of pheromone concentration on the worst path; when When , k takes a relatively small value k∈(0,1); when When k takes a larger value k∈(1,k max ); when L m =L min When k takes the maximum value k=k max , thereby increasing the difference in pheromone concentration increments between the better search results and the worse search results, accelerating the convergence speed of subsequent algorithm iterations, and ultimately obtaining path search results with better length process costs and turning process costs. The second aspect is the layout optimization method, which splits or merges the branches and loops in the structure to obtain the wiring harness layout solution with the lowest overall process cost. Specifically including:

[0033] 1) Multi-objective path planning of all wires in the harness is implemented, the positions of branch points are calculated according to the path results, and the preliminary topological structure of the multi-branch harness expressed by an undirected graph is established, and the connection relationship of the branch points is stored in the form of an adjacency list; the depth-first search (DFS) algorithm is used to traverse the current harness adjacency list and locate the loops in the topological structure according to the storage order of the elements.

[0034] The current loop is identified and divided. If there is a nested loop in the current loop, the loop with the smallest total path length inside it is called the minimum loop, and the minimum loop is taken as the current loop.

[0035] For a loop with a connector, the connector node has and can only have one adjacent point. All feasible branch point locations are traversed, and branch path search is performed in a local range. When processing the branches in the loop, the wire length cost L in the loop is used. i,j With path turn cost B i,jThe overall cost evaluation of branch optimization is used as the optimization target to make the target as small as possible, as shown in formula 1.7:

[0036] cost i,j =c L L i,j +c t B i,j (1.7)

[0037] c L 、c t are the weights of length cost and turning cost respectively;

[0038] The branch structure and branch paths of multi-branch wiring harnesses are optimized, and the closed loop problem formed during multi-source path planning is avoided.

[0039] In addition, the method also includes an automatic wiring process, batch extracting assembly data and design information of the original wiring environment, realizing automatic path planning of branch wiring harnesses on the machine, building a wiring harness geometry model and obtaining corresponding manufacturing data. Specifically including:

[0040] Step 1: Input the original wiring environment structure model and wiring harness electrical principle design information, and batch obtain the installation positions and directions of the electrical connectors and fixing clamps in the wiring environment.

[0041] Step 2: Using the edge collapse simplification algorithm (QEM) based on quadratic error metric, calculate the corresponding contraction points of the grid points of the environment model The position and the collapse cost matrix Q are used to perform edge contraction operations, and the environment model is lightweight while ensuring the detailed features such as holes. Obstacles are divided into objects based on the hierarchical bounding box structure (BVH), and the BVH structure of the obstacle space is constructed from top to bottom. The signed distance field (SDF) of the obstacle space is calculated, combined with the minimum wall spacing constraint, the corresponding contour value is set, and the moving cube algorithm (MC) is used to construct the offset surface grid model to form a path search three-dimensional environment map.

[0042] Step three: Use the path planning method to search for the paths of the wires in the harness in turn, and use the loop processing method based on the path results to further optimize the branch positions and branch paths of the harness, so as to achieve large-size, long-span, large-volume, and widely distributed branch harness path planning under multiple wiring process constraints.

[0043] Step 4: According to the branch harness wiring results, a harness geometry model is constructed based on a cubic non-uniform B-spline curve, and according to the wire list contained in the branch, manufacturing information such as wire gauge, wiring relationship, and path length is output.

[0044] The beneficial effects of the present invention are:

[0045] The present invention is a branch wiring harness path planning and optimization method facing multiple wiring process constraints.

[0046] 1) It solves the problems of inaccurate design data, unreasonable wiring and low change efficiency in the traditional wiring mode of physical trial installation dominated by engineer experience, weighs the impact of process constraints on path selection in terms of path length, turns, bundling, electromagnetic compatibility, etc., so that the design data can effectively guide cable manufacturing and laying.

[0047] 2) It avoids the situation where the long-span harness interferes with the structure due to the closed loop during the laying process. The method does not need to pre-set the number and position of branch points, and can effectively achieve the joint optimization of the wire path, branch structure and branch point position, thereby optimizing the wiring length and improving the wiring quality and efficiency.

[0048] 3) Design the overall process of automatic wiring and pre-process the complex wiring environment to solve the problem of low search efficiency of large-size cables in complex environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 It is a schematic diagram of the process of the present invention;

[0050] Figure 2 It is a schematic diagram of the wiring environment preprocessing process included in the method of the present invention;

[0051] Figure 3 Schematic diagram of the process of constructing a feasible search space provided in an embodiment of the present application; (a) lightweight model; (b) constructing an obstacle offset surface; (c) constructing a feasible search space;

[0052] Figure 4 A schematic diagram of a single cable planning process for multiple laying constraints contained in the method of the present invention;

[0053] Figure 5 It is a schematic diagram of the multi-branch harness layout optimization process included in the method of the present invention;

[0054] Figure 6 A schematic diagram of the multi-branch wiring harness layout optimization provided in an embodiment of the present application. DETAILED DESCRIPTION

[0055] Reference will now be made in detail to embodiments of the invention, examples of which are illustrated in the accompanying drawings.

[0056] The present invention proposes a branch wire harness path planning and optimization method for multiple wiring process constraints, referring to Figure 1The flowchart for the implementation of the method mainly includes the following steps: inputting wiring data, preprocessing the wiring environment, cable path planning, multi-branch harness layout optimization and harness geometry modeling.

[0057] Figure 1 The input wiring data specifically includes the original fuselage structure model, connector assembly model and electrical principle design information. Among them, the fuselage structure model is processed into a mesh model in STL or STP format, and the connector assembly model will extract the corresponding installation position and direction in batches through the secondary development function of the original design software. Through the data conversion interface module, the XML format information in the electrical design software is processed into the wiring relationship and selection information of the wires in the harness.

[0058] According to a specific example of the present invention, a branch wire harness path planning and optimization method for multiple wiring process constraints proposed by the present invention mainly includes the following three steps. After completing the input of relevant wiring data, the steps are processed in sequence.

[0059] Step 1: Pre-process the wiring environment. Figure 2 The figure shows the wiring environment preprocessing process. Through lightweight processing, surface reconstruction and other processes, the complex on-board wiring environment model is processed into a three-dimensional environment map that can be used by the path planning algorithm.

[0060] 1) Model lightweight. The edge collapse simplification algorithm (QEM) based on quadratic error metric is used to simplify each mesh vertex v of the original environment space mesh model. i Calculate the collapse cost matrix Q. Let the plane p near the vertex be expressed as ax+by+cz+d=0 and a 2 +b 2 +c 2 =1, then the square of the distance from the vertex to the plane can be expressed as

[0061] dist(v,p) 2 =(v T p) 2 =v T pp T v=v T K p v (1.8)

[0062] The cost matrix Q is the point v i The basic error quadratic form K of all adjacent triangles p The sum of .

[0063] Set the maximum allowable gap value t and find all the mergeable point pairs in the model, and calculate the vertices after each pair of points is contracted. position, and update the collapse cost As a new vertex The error measurement values ​​are uniformly stored in the heap.

[0064] In each iteration, the point pair with the smallest collapse cost in the heap, such as (v1, v2), is popped out, the corresponding contraction point v position and the collapse cost matrix Q are calculated, and the edge contraction operation is performed. At the same time, all point pairs involving v1 and v2 in the heap are updated, such as (v1, v i ) and (v2,v i ) Click Yes, need to update to At the same time, modify the corresponding collapse cost value, update the corresponding arrangement order in the heap, and perform the next iteration until the elements in the heap are empty.

[0065] On the premise of ensuring the detailed features of the original structural model such as holes and cavities, a lightweight environmental space grid model is generated.

[0066] 2) Construct obstacle space. Use a hierarchical bounding box structure (BVH) to implement object division for obstacles, and construct the BVH structure of the obstacle space from top to bottom. First, calculate the overall bounding box AABB of the model and use it as the root node. Then find a suitable division direction based on the object size stored in the current node, divide the triangular faces in the current bounding box into two parts according to the median, and calculate new bounding boxes as the left and right child nodes of the current node. Repeat the above process to create child nodes for the current space until the number of triangular meshes contained in the current bounding box is 1. This node is set as the leaf node of the BVH tree, and the iterative process of dividing the node subspace is terminated. Reduce the time complexity of subsequent interference detection and improve the efficiency of path search and verification.

[0067] 3) Calculate the signed distance field. The obstacle space is gridded to obtain a set of N×N×N sampling grid points. The value of the signed distance function dist(n) of each grid point in the field is calculated and stored in a three-dimensional matrix P. Each element in the matrix represents the closest distance between the grid point and the surface of the spatial model. The grid points with dist(n)>0 are located outside the model surface, the grid points with dist(n)<0 are located inside the model surface, and the grid points with dist(n)=0 are exactly on the model surface.

[0068] 4) Construct the offset surface. According to the SDF values ​​of these sampling grid points, combined with the minimum wall spacing constraint, the corresponding contour values ​​are set, and the moving cube algorithm (MC) is used to construct the offset surface mesh model. The unit cube is constructed with grid points, and there are three cases for the function value corresponding to each vertex: higher than or equal to the equal value indicates that it is inside the surface, and lower than the equal value indicates that it is outside the surface. Calculate the function values ​​of the eight vertices of the cube, determine whether there are equal-value points, and construct triangular patch units based on the equal-value points on each edge. Calculate the unit normal of each voxel vertex and interpolate the normal to each vertex of the triangular patch. By traversing all voxels, find the triangular patches therein, and merge all the obtained triangles, the reconstructed triangular mesh model data is obtained.

[0069] 5) Discretize the feasible search space. Discretize the grid model after the wall constraint processing to obtain three-dimensional ordered discrete points as the feasible search space for routing. The feasible search space U generated in the preprocessing stage is env and obstacle space U obs Together they form a three-dimensional environment map for path planning.

[0070] Step 2: Single cable path planning for multiple wiring process constraints. Figure 4 The figure shows the algorithm flow of single cable path planning. Based on the A*-ant colony algorithm combined with wiring process constraints, the path planning of large-size and long-span single cables in complex environments can be realized.

[0071] (1) Combined with actual wiring process constraints such as specific physical constraints, turning constraints, support constraints, electromagnetic compatibility constraints and bundling constraints, the heuristic function H between adjacent nodes in the ant colony algorithm is designed. ij (t), as shown in formula 1.9:

[0072]

[0073] f(n)=g(n)+h(n) (1.10)

[0074] f(n) is introduced as the estimation function of the A* algorithm for the distance cost of the current node, and its specific definition is shown in formula 1.10. g(n) is the cumulative cost from the starting point to the current node n, and its definition is as follows:

[0075]

[0076] In the formula, l i Indicates the length of the i-th cable segment, i=1,2,3……

[0077] When there is interference with the obstacle space, the distance value of g(n) is infinite. h(n) is the shortest distance from node n to the target point. The evaluation function of the A* algorithm is introduced into the ant colony algorithm to provide a global inspiration direction for the algorithm and avoid blind search in the early stage of the algorithm.

[0078] The cable turning evaluation function bend(n) is introduced to evaluate the turning angle θ of the cable when selecting the next search node. n,n+1 With bending radius r n,n+1 The impact on the quality of the wiring path is specifically defined as shown in Equation 1.12.

[0079]

[0080] Among them, coefficients C1 and C2 respectively represent the influence of turning angle and bending radius on the cable bending evaluation function. According to the wiring harness design principle, different turning angle ranges correspond to different bends (θ n,n+1 ) value. The specific value is shown in formula 1.13.

[0081]

[0082] Select the turning radius coefficient ζ to generate the minimum bending radius wiring process standard value r min for:

[0083] r min =ζD bundle (1.14)

[0084] Where D bundle is the outer diameter of the bundle segment, and ζ is the allowable bending radius ratio.

[0085] At the same time, set the turning safety factor s r . bend(r n,n+1 )When the specific value is as shown in formula 1.15.

[0086]

[0087] Introducing the clamp capacity evaluation function clip(D n,n+1 ) is used to reflect the influence of the completed cables on the path search of the cables to be laid. The specific definition is as shown in Formula 1.16.

[0088]

[0089] Among them, s d is the clamp capacity safety factor, D n,n+1 Indicates searching for the next node P n+1 The outer diameter of all cables bundled at the time. n+1 Indicates passing through node P n+1The cable list is calculated, and the minimum circumscribed circle of all cable cross sections is used as D at this time. n,n+1 The value of .

[0090] Introducing the electromagnetic compatibility evaluation function emc(e n,n+1 ) to reflect the impact of laying channels with different EMC category codes on the current cable path point selection. The specific definition is as shown in formula 1.17.

[0091]

[0092] Among them, e n,n+1 Indicates P n+1 EMC category code for cable channels, e bundle Search the EMC category code of the cable for the current path.

[0093] Introduce the bundle node evaluation function bundle(b n,n+1 ) to reflect the influence of the selected path point on the current cable path point selection. The specific definition is as shown in formula 1.18.

[0094]

[0095] (2) Based on the heuristic function between nodes, calculate the probability of an ant transferring from the current node i to the next node j according to Formula 1.19.

[0096]

[0097] Among them, τ ij (t) is the pheromone concentration between node i and node j, H ij (t) is the heuristic function between node i and node j, α and β respectively reflect the influence of the function on the state transition probability. n is any node that the ant can choose in the next step, allowed m is the set to be visited by the current ant. The ant selects the next node according to the roulette method, so the heuristic function H ij The larger the (t) value is, the greater the probability that the ant will choose this path point.

[0098] (3) After each iteration, the pheromone concentration on the path taken by the ants is updated. The better the path, the higher the pheromone concentration increment. The pheromone concentration increment model is expressed as Equation 1.20.

[0099]

[0100] Among them, ρ is the pheromone volatility coefficient, τ ij (t-1) is the pheromone concentration between node i and node j at time t-1, Δτ ijis the total pheromone accumulated between (i, j) after this iteration. The pheromone concentration increment between (i, j) for each ant is updated using global information, and the specific definition is as shown in formula 1.21.

[0101]

[0102] In addition to the path length cost L m In addition, the number of path inflection points T is introduced m Accumulation of turn evaluation m bend(n) ratio as the path turning cost, with the number of repeated path points N bm As the path bundling cost evaluation, c L 、c t With c b is the weight of each cost. k represents the pheromone concentration coefficient; Q is the pheromone intensity coefficient, which is related to the search step length. It is defined as follows:

[0103] Q=L max -L min (1.22)

[0104] Where, L max is the length of the maximum path generated after this iteration, L min is the length of the shortest path generated after this iteration.

[0105] The average path length experienced by all ants in this iteration is recorded as When L m =L max When k takes the minimum value k min = 0, reducing the increase of pheromone concentration on the worst path; when When , k takes a relatively small value k∈(0,1); when When k takes a larger value k∈(1,k max ); when L m =L min When k takes the maximum value k=k max , thereby increasing the difference in pheromone concentration increment between the better search results and the worse search results, accelerating the convergence speed of subsequent algorithm iterations, and finally obtaining a path search result with better length process cost and turning process cost.

[0106] Step 3: Multi-branch harness layout optimization. Figure 5 As a schematic diagram of the multi-branch harness layout optimization process, a branch merging method of harness loops based on undirected graph expression is proposed to achieve layout optimization of multi-branch harnesses with large numbers and wide distribution.

[0107] According to the preliminary topological structure of the multi-branch harness, the depth-first search (DFS) algorithm is used to traverse the current harness adjacency table and locate the loops in the topological structure according to the storage order of the elements.

[0108] Due to the uncertainty of the order of elements in the adjacency list, there may be nested loops inside the currently located loop. If there are nested loops in the loop, the loop with the smallest total path length is called the smallest loop. Divide the loop and take the smallest loop as the current loop. When processing the branches in the loop, the wire length cost L in the loop is used. i,j With path turn cost B i,j The overall cost evaluation of branch optimization is used as the optimization target to make the target as small as possible, as shown in formula 1.23:

[0109] cost i,j =c L L i,j +c t B i,j (1.23)

[0110] c L 、c t are the weights of length cost and turning cost respectively;

[0111] Specific:

[0112] For a loop with a connector, the connector node has and can only have one adjacent point. All feasible branch point positions are traversed, and branch path search is performed in a local range. The optimal branch point position is determined by optimizing the objective as shown in formula 1.23. At this time, the above formula can also be expressed as:

[0113]

[0114] Among them, L i,j Path length cost T i,j is the number of inflection points contained in the wire path in the loop, and ∑bend(n) is the cumulative evaluation of the turning points of the wire path in the loop. The overall process cost of establishing a new branch in the loop is evaluated with the optimization goal.

[0115] For loops containing interference points, all feasible source points are traversed, and branch path search is performed in a local range avoiding interference points. Similarly, the new branch path is determined by the optimization objective shown in Formula 1.23.

[0116] After processing the above two types of special loops into ordinary loops, traverse all mergeable branches b i , calculate the corresponding wire merging cost according to formula 1.23 , taking the minimum value of the overall target as the final layout result, we can obtain a layout solution with better overall length, turning and bundling cost. At this time, Equation 1.23 can also be expressed as:

[0117]

[0118] Where, T bi The total number of inflection points of the loop branch path after merging

[0119] when When taking branch b i To branch b j The combined length cost is calculated as follows:

[0120]

[0121] In the formula, len bi Contains branch b i cable length.

[0122] Finally, according to the topological structure of the multi-branch harness, cubic non-uniform NURBS curves are used to realize harness geometric modeling.

[0123] In order to verify the feasibility of the present invention, the electrical principle design information of the relevant wiring harness was collected in a wiring area of ​​a certain type of aircraft, and an example verification was carried out based on the 3D wiring environment. The size range of the original fuselage structure model is 21.3m×5.5m×5.5m, the lightweight degree of the obstacle model is set to 95%, the minimum wall spacing is 30mm, the search step is 600mm, and the reference result of the automatic layout design of multi-branch wiring harness is obtained. Figure 6 The automatic wiring results were compared with the verified manual wiring results of the same fuselage section, and good optimization effects were achieved in terms of path quality such as cable length and turns, and the average error of the wire length in the harness was -10.88 mm / m. Compared with the A* algorithm and the classic ant colony algorithm, the method of the present invention has a good effect in improving the wiring speed.

Claims

1. A branch harness path planning and optimization method for multiple wiring process constraints, characterized in that: The method comprises: 1) According to the constraints of multiple wiring processes, based on the A*-ant colony path search algorithm, the optimization target and concentration increment model are designed to improve the turning quality of the cable path and optimize the path length cost L m ;Realize the path planning of all cables in the harness; 2) According to the path results obtained in 1), a closed loop elimination method based on undirected graph expression is used to locate and divide the loops in the adjacency table for the closed loop structure existing in the multi-branch wiring harness wiring results. By judging the type of the loop, the branches in the loop are split or merged to further optimize the branch path and branch structure of the harness; The above 1) specifically includes: the above multiple wiring process constraints include cable turning constraints, support constraints, electromagnetic compatibility constraints and bundling constraints, using A*-ant colony search algorithm, combined with multiple wiring process constraints, to design the heuristic function H between adjacent nodes in the ant colony algorithm ij (t), as shown in formula 1.1: In the above formula, the A* algorithm’s evaluation function f(n) for the current node distance cost is introduced into the heuristic function of the ant colony algorithm, and the path length is used as the heuristic standard to provide a global optimization direction for path planning; the cable turning evaluation function bend(n) and the clamp capacity evaluation function clip(D n,n+1 ), electromagnetic compatibility evaluation function emc(e n,n+1 ) and the bundle evaluation function bundle(b n,n+1 ), provides the local heuristic size of node state transfer, and reflects the size of node state transfer probability through the size of the heuristic value; the corresponding indexes ω1, ω2, ω3 and ω4 respectively reflect the importance of the corresponding evaluation functions on the heuristic function; According to the heuristic function between nodes, the probability of ants transferring from node i to the next node j is calculated. After each iteration, the pheromone concentration on the path passed by the ants is updated, and the pheromone concentration increment model of the ant colony algorithm is designed, as shown in Formula 1.2, introducing the number of path inflection points T m Accumulation of turn evaluation Σ m bend(n) is used as the path turning cost, and the number of common path points N in the path is introduced bm As the path bundling cost, τ ij (t) is the pheromone concentration between node i and node j at time t, ρ is the pheromone volatility coefficient, τ ij (t-1) is the pheromone concentration between node i and node j at time t-1, Δτ ij is the total accumulated pheromone between (i, j) after this iteration, and the pheromone concentration increment between (i, j) for each ant m Use global information to update, c L 、c t With c b are the weights of length cost, turning cost and bundling cost respectively, k represents the pheromone concentration coefficient, and Q represents the pheromone intensity coefficient.

2. A branch wire harness path planning and optimization method for multiple wiring process constraints according to claim 1, characterized in that: Q is the pheromone intensity coefficient, which is related to the path search step length and is defined as follows: Q=L max -L min (1.3) Where, L max is the length of the maximum path generated after this iteration, L min is the length of the shortest path generated after this iteration; The average length of the paths experienced by all ants in this iteration is recorded as When L m =L max When k takes the minimum value k min = 0, reducing the increase of pheromone concentration on the worst path; when When , k takes a relatively small value k∈(0,1); when When k takes a larger value k∈(1,k max ); when L m =L min When k takes the maximum value k=k max , thereby increasing the difference in pheromone concentration increment between better search results and worse search results, and accelerating the convergence speed of subsequent algorithm iterations.

3. The method for planning and optimizing branch wiring harness paths for multiple wiring process constraints according to claim 1, characterized in that: The 2) mentioned above specifically includes: 1) Multi-objective path planning of all wires in the harness is implemented, the location of the branch point is calculated according to the path result, and the branch harness topology structure expressed by an undirected graph is established, and the connection relationship of the branch points is stored in the form of an adjacency table; The depth-first search (DFS) algorithm is used to traverse the current harness adjacency table, locate and identify the closed loop formed locally in the undirected graph, determine whether the current loop has loop nesting, and take the smallest loop with the shortest total length as the current loop; When processing the branches in the loop, the wire length cost L in the loop is used. i,j With path turn cost B i,j The overall cost evaluation of branch optimization is used as the optimization target to make the target as small as possible, as shown in Equation 1.4; cost i,j =c L L i,j +c t B i,j (1.4) c L 、c t are the weights of length cost and turning cost respectively; When the current loop contains an electrical connector, the path points on the adjacent edges of the connector nodes are traversed as new branch points, and the path planning is performed for the newly created branches in a local range. The overall cost of the loop corresponding to each branch point is calculated according to formula 1.4, and the merging scheme with the lowest cost is selected to optimize the branch point position; When the current loop contains an interference point, the path points on the adjacent edges of the interference point are traversed as the source points of the new branch, and the path planning is performed in the local range to bypass the interference point. The overall cost of the loop corresponding to each source point is calculated according to formula 1.4, and the path plan with the lowest overall cost is selected to optimize the branch path; After processing the above two types of special loops into ordinary loops, traverse all mergeable branches, calculate the wire merging cost corresponding to each branch according to formula 1.4, optimize the branch position and branch structure, and obtain a layout plan with an overall better length, turning and bundling cost.

4. The method for planning and optimizing branch wiring harness paths for multiple wiring process constraints according to claim 1, characterized in that: According to the search algorithm and optimization method, the overall process of automatic wiring of branch harnesses is designed, the assembly data and design information of the original wiring environment are extracted in batches, the path planning of branch harnesses in complex environments is realized, the harness geometry model is constructed and the corresponding manufacturing data is obtained, including: Step 1: Preprocessing the wiring environment, through lightweight processing and surface reconstruction, the complex wiring environment model is processed into a three-dimensional environment map that can be used by the path planning algorithm; Step 2: Use the A*-ant colony path search algorithm to achieve cable path planning for multiple wiring process constraints; Step 3: Optimize the layout of the multi-branch harness, and use cubic non-uniform B-spline curves to construct the harness geometry model of the optimized multi-branch harness topology. According to the list of wires contained in the branch, output the wire gauge, wiring relationship, and path length manufacturing information.

5. The method for planning and optimizing branch wiring harness paths for multiple wiring process constraints according to claim 4, characterized in that: The step 1 specifically includes: Input environment model: input the original wiring environment structure model and wiring harness electrical principle design information, and batch obtain the installation positions and directions of the electrical connectors and fixing clamps in the wiring environment; Model lightweighting: The edge collapse simplification algorithm QEM based on quadratic error metric is used to lightweight the environment model while ensuring the detailed features of the holes; Construct obstacle space: Based on the hierarchical bounding box structure BVH, obstacles are divided into objects, and the BVH structure of the obstacle space is constructed from top to bottom; Calculate the signed distance field: grid the obstacle space to obtain a set of N×N×N sampling grid points, calculate the value of the signed distance function dist(n) of each grid point in the field and store it in a three-dimensional matrix P, and calculate the signed distance field SDF of the obstacle space; Construct the offset surface: Combine the minimum wall spacing constraint, set the corresponding contour value, and use the marching cube algorithm MC to construct the offset surface mesh model; Discretize the feasible search space: discretize the grid model after the wall constraint processing to obtain three-dimensional ordered discrete points as the feasible search space for routing; The generated feasible search space and obstacle space together form a three-dimensional environment map for path search.

6. A branch harness path planning and optimization system for multiple wiring process constraints, characterized in that: The system is used to implement the method according to any one of claims 1 to 5.

Images