An optimization method for flexible intelligent manufacturing unit system

Abstract

The present invention discloses a flexible intelligent manufacturing unit system and an optimization method thereof, and relates to the technical field of flexible intelligent manufacturing unit design. The system includes: a loading port buffer, a unloading port buffer, a robot, and a plurality of workstations; the method includes the following steps: based on the process path of the flexible intelligent manufacturing unit system, a target optimization problem is constructed with the goal of maximizing the average output rate of the system workpieces; the target optimization problem is solved to obtain a system facility layout optimization strategy; and the facility layout optimization of the flexible intelligent manufacturing unit system is completed according to the system facility layout optimization strategy. The optimization method provided by this application can effectively solve the deadlock phenomenon that facilities in the flexible intelligent manufacturing unit system are prone to, and effectively reduce the production cost of workpieces.

Description

Technical Field

[0001] The present invention relates to the technical field of flexible intelligent manufacturing unit design, and in particular to an optimization method for a flexible intelligent manufacturing unit system. Background Art

[0002] Intelligence and service-oriented development are key trends in industrial modernization. Smart factories are the vehicles for intelligent manufacturing. Factory planning is the first step in building a smart factory, and the layout of production facilities is a crucial component of smart factory planning. The layout of production facilities significantly impacts the manufacturing system's production efficiency, material handling efficiency, and production cycle time.

[0003] A vast body of research has been published on the Facility Layout Problem (FLP). Most of this literature examines deterministic static layout optimization problems and evaluates the effectiveness of layout solutions based on minimizing logistics costs, without considering key performance indicators of the system, such as average throughput. In static FLP research, logistics costs are not factored into the empty travel of material handling, and the stochastic factors of dynamic systems, such as the uncertainty of task arrival, processing, and handling, are ignored. This results in layout solutions lacking practical application value.

[0004] A key challenge in considering stochastic layout optimization lies in the performance evaluation methods for stochastic systems. Currently, the main approaches used are simulation and queuing theory. While simulation methods can yield relatively accurate solutions, they require multiple independent experiments, resulting in lengthy simulation times and high time complexity for simulation-based layout optimization algorithms. Analytical methods based on queuing theory are effective tools for evaluating the performance of stochastic systems, and their fast runtime significantly improves the efficiency of solving layout optimization problems. Queuing networks, based on the traditional single-queue model of queuing theory, describe the entire stochastic system as a network of coupled queues. This requires considering the coupling between queues, making analysis and solution difficult. Furthermore, in queuing network models with finite buffers, blocking often prevents product-form solutions, making analysis and solution even more challenging.

[0005] Production systems consist of two key subsystems: the material processing system and the material handling system. Existing queuing network-based production system performance evaluation methods mostly focus on each of these systems separately, rarely considering their integrated integration. Most queuing network models for production systems that integrate material processing and handling are based on Markovian or infinite-capacity buffer assumptions (i.e., ignoring blocking). In summary, current state-of-the-art technologies fail to effectively establish queuing network models with blocking effects and generally distributed service times for stochastic production systems coupled with material processing and handling. This results in facility deadlocks within flexible intelligent manufacturing cell systems, leading to higher production costs for workpieces. Summary of the Invention

[0006] In order to overcome the problem that the existing technology fails to effectively establish a queuing network model with blocking effects and generally distributed service times, resulting in the deadlock phenomenon easily occurring in the facilities of the flexible intelligent manufacturing unit system, the present invention proposes an optimization method for the flexible intelligent manufacturing unit system, which can quickly and effectively establish a queuing network model with blocking effects and generally distributed service times, thereby solving the deadlock phenomenon easily occurring in the facilities of the flexible intelligent manufacturing unit system and effectively reducing the production cost of the workpiece.

[0007] To achieve the purpose of the present invention, the present invention adopts the following technical solutions:

[0008] One of the purposes of the present invention is:

[0009] A flexible intelligent manufacturing unit system, comprising: a loading port buffer, a unloading port buffer, a robot, and several workstations;

[0010] The loading port buffer sends a workpiece handling request to the robot;

[0011] The robot transports the workpiece to the workstation according to the workpiece transport request issued by the loading port buffer;

[0012] The workstation processes the workpiece and sends a workpiece handling request to the robot after the processing is completed;

[0013] The robot transports the workpiece to the unloading port buffer according to the workpiece transport request issued by the workstation.

[0014] Preferably, the workstation includes a workpiece processing buffer and processing equipment;

[0015] The processing equipment processes the workpiece, and after the processing is completed, sends a workpiece transportation request to the robot. The robot transports the workpiece to the workpiece processing buffer area or the unloading port buffer area according to the workpiece transportation request sent by the processing equipment.

[0016] In the above technical solution, each workstation contains a buffer with limited capacity and a processing device. The task processing time follows the general distribution (General Distribution), and the service rule is first-come, first-served (FCFS). The speed of the robot is constant, that is, the transportation time of the workpiece between workstations is a constant distribution. The blocking mechanism of the processing process is post-service blocking; the blocking mechanism of the transportation process is pre-service blocking, so as to avoid deadlock and effectively reduce the production cost of the workpiece.

[0017] The second purpose of the present invention is:

[0018] A method for optimizing a flexible intelligent manufacturing unit system is provided, wherein the method is applied to a flexible intelligent manufacturing unit system and comprises the following steps:

[0019] Based on the process path of the flexible intelligent manufacturing cell system, a target optimization problem is constructed with the goal of maximizing the average output rate of the system workpieces;

[0020] Solving the target optimization problem to obtain a system facility layout optimization strategy;

[0021] The facility layout optimization of the flexible intelligent manufacturing unit system is completed according to the system facility layout optimization strategy.

[0022] In the above technical scheme, based on the process path of the flexible intelligent manufacturing unit system, the target optimization problem can be constructed according to the general layout problem of the flexible intelligent manufacturing unit system, so as to fully reflect the problems existing in the facility layout of the current flexible intelligent manufacturing unit system from the data, and then consider the blocking impact and the general distribution service time, construct the target optimization problem, and solve the target optimization problem to obtain the facility layout optimization strategy of the flexible intelligent manufacturing unit system, and complete the facility layout optimization of the flexible intelligent manufacturing unit system according to the system facility layout optimization strategy, which can effectively solve the deadlock phenomenon of the facilities in the flexible intelligent manufacturing unit system and effectively reduce the production cost of the workpiece.

[0023] Furthermore, the process of constructing the target optimization problem includes:

[0024] Assign m+2 workstations, loading port buffers, and unloading port buffers to m+2 known layout positions, and set node position constraints and position capacity constraints. The expressions are:

[0025] ;

[0026] ;

[0027] Based on the node location constraints and location capacity constraints, the target optimization problem is constructed and expressed as:

[0028] max Θ( X );

[0029] Where m represents the number of nodes in the system, including workstations, loading port buffers, and unloading port buffers; X ={ x i,k} represents the decision variable; Θ represents the average output rate of the system; i represents the sequence number of the node in the system, i =0, 1, …, m +1, where i =0 and m +1 represents the unit loading / unloading port buffer zone; k represents the node i at the kth position.

[0030] In the above technical solution, based on the process path of the flexible intelligent manufacturing unit system, the target optimization problem can be constructed according to the general layout problem of the flexible intelligent manufacturing unit system, so as to fully reflect the problems existing in the current flexible intelligent manufacturing unit system facility layout from the data, and then consider the blocking impact and general distribution service time to construct the target optimization problem.

[0031] Furthermore, the process of solving the target optimization problem includes:

[0032] Generate an initial solution based on the system's process path;

[0033] Based on the initial solution, a 2-Opt exchange strategy is used to design the neighborhood structure, and a homogeneous layout is performed based on the robot transport path network.

[0034] Setting up a queuing network model to approximately solve the output rate of the flexible intelligent manufacturing unit system and obtain the objective function value of the optimization problem;

[0035] Based on the objective function value of the optimization problem, an improved variable neighborhood search algorithm is set to iteratively optimize and solve the target optimization problem to obtain the system facility layout optimization strategy.

[0036] In the above technical scheme, an initial solution is generated based on the process path of the system, which can reflect the workpiece output efficiency of the current system and provide initial parameters for the target optimization problem to be solved, so as to optimize based on the initial solution to improve the workpiece output rate of the system; a 2-Opt exchange strategy is used for neighborhood structure design, and isomorphic layout processing is performed based on the robot transportation path network, which can preliminarily optimize the layout of the system facilities, thereby facilitating the use of the set queuing network model to approximate the output rate of the flexible intelligent manufacturing unit system to obtain the objective function value of the optimization problem. The set queuing network model can quickly and effectively establish connectivity with blocking effects and generally distributed service times between facilities in the system; then, based on the objective function value of the optimization problem, an improved variable neighborhood search algorithm is set to iteratively optimize and solve the target optimization problem, and obtain a system facility layout optimization strategy, which effectively solves the deadlock phenomenon of facilities in the flexible intelligent manufacturing unit system and effectively reduces the production cost of workpieces.

[0037] Furthermore, the process of setting a queuing network model to approximate the output rate of the flexible intelligent manufacturing unit system includes:

[0038] Add virtual nodes to reconstruct the queuing network model, and obtain a queuing network model in which each node is independent of each other;

[0039] The reconstructed queuing network model is used to solve the workpiece processing input or output rate of the workstation node;

[0040] The reconstructed queuing network model is used to solve the input or output rate of the workpiece transportation of the robot node;

[0041] Based on the workpiece processing input or output rate of the workstation node and the workpiece transportation input or output rate of the robot node, the average output rate of the workpiece processing in the system is calculated, and the average output rate is iteratively optimized to obtain the system facility layout optimization strategy;

[0042] Wherein, the queuing network model is a neural network model.

[0043] Furthermore, the process of solving the workpiece processing input or output rate of the workstation node using the reconstructed queuing network model includes:

[0044] Set the arrival rate constraint and blocking probability constraint of the workpiece arriving at the workstation node and return blocking probability , the expressions are:

[0045] ;

[0046] ;

[0047] ;

[0048] According to the arrival rate constraint and blocking probability constraint and return blocking probability , calculate the workpiece input or output rate of the workstation node, the expression is:

[0049] ;

[0050] in, Indicates the rate at which the material outside the system reaches the unit loading port. α i,j Representation node i arrive j The path probability, j =1, 2, …, m +1; ρ j Indicates a workstation j The workpiece flow intensity, represents the rate of distribution of the remaining processing time of the workpiece in workstation node j, The coefficient representing the difference in workpiece flow, , Representation node j The effective input rate, For virtual nodes h The arrival rate; represents the blocking probability of the robot node, For virtual nodes h The workpiece flow intensity, Indicates a workstation j The exponential function value of the workpiece flow intensity with respect to the coefficient of variation, .

[0051] Furthermore, the process of solving the input or output rate of the workpiece transportation of the robot node using the reconstructed queuing network model includes:

[0052] Set the input rate constraint of the robot, the expression is:

[0053] ;

[0054] Set the robot's saturation probability constraint, the expression is:

[0055] ;

[0056] Set the probability constraint of the robot being in a non-idle state, the expression is:

[0057] ;

[0058] According to the input rate constraint, saturation probability constraint and probability constraint of the robot in the non-idle state, the input or output rate of the robot is calculated. The expression is:

[0059] ;

[0060] in, ρ r Represents a robot node r The workpiece flow intensity, Indicates the no-load workpiece flow intensity, Indicates the workpiece flow intensity of the load.

[0061] Furthermore, based on the workpiece processing output rate of the workstation node and the input or output rate of the workpiece transportation of the robot node, the average output rate of the workpiece processing in the system is calculated, and the expression is:

[0062] ;

[0063] Determine whether the average output rate of workpiece processing in the system meets the node position constraints and position capacity constraints. If so, the maximum system output rate obtained in the iterative solution process will be used as the final system facility layout optimization strategy; if not, continue to iterate until the average output rate obtained by the solution meets the node position constraints and position capacity constraints, and use the maximum system output rate obtained as the final system facility layout optimization strategy.

[0064] In the above technical solution, virtual nodes are added to reconstruct the grid of the queuing network model, which can better reflect the status of each facility in the system from the data, thereby obtaining a queuing network model in which each node is independent of each other, and using the queuing network model in which each node is independent of each other to solve the workpiece processing input or output rate of the workstation node and the input or output rate of the workpiece transportation of the robot node; then based on the workpiece processing input or output rate of the workstation node and the input or output rate of the workpiece transportation of the robot node, the average output rate of the workpiece processing in the system is calculated, and the average output rate is iteratively optimized to obtain the system facility layout optimization strategy, and finally, the facility layout optimization of the flexible intelligent manufacturing unit system is completed according to the system facility layout optimization strategy, effectively solving the deadlock phenomenon of the facilities in the flexible intelligent manufacturing unit system and effectively reducing the production cost of the workpiece.

[0065] The third purpose of the present invention is:

[0066] An electronic device comprises a memory, a processor and a computer program stored in the memory and running on the processor, wherein when the processor executes the computer program, the steps of an optimization method of a flexible intelligent manufacturing unit system are implemented.

[0067] Compared with the prior art, the present invention has the following beneficial effects:

[0068] The present invention provides an optimization method for a flexible intelligent manufacturing unit system. First, based on the process path of the flexible intelligent manufacturing unit system, a target optimization problem can be constructed according to the general layout problem of the flexible intelligent manufacturing unit system to fully reflect the problems existing in the facility layout of the current flexible intelligent manufacturing unit system from the data, and then considering the blocking effect and the general distribution service time, the target optimization problem is constructed, and a queuing network model with blocking effect and general distribution service time can be effectively established, and the queuing network model is used to solve the target optimization problem to obtain the facility layout optimization strategy of the flexible intelligent manufacturing unit system, and the facility layout optimization of the flexible intelligent manufacturing unit system is completed according to the system facility layout optimization strategy, which can effectively solve the deadlock phenomenon of the facilities in the flexible intelligent manufacturing unit system and effectively reduce the production cost of the workpiece. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 A schematic structural diagram of a flexible intelligent manufacturing unit system provided in an embodiment of the present application;

[0070] Figure 2 A flowchart of the steps of an optimization method for a flexible intelligent manufacturing unit system provided in an embodiment of the present application;

[0071] Figure 3 A schematic diagram of encoding in an arranged encoding manner provided in an embodiment of the present application;

[0072] Figure 4 A schematic diagram of the central rotation isomorphism of the coding isomorphic layout provided in an embodiment of the present application;

[0073] Figure 5 A schematic diagram of a vertex symmetry axis flip isomorphism of a coding isomorphic layout provided in an embodiment of the present application;

[0074] Figure 6 A schematic diagram of arc symmetry axis flip isomorphism of the coding isomorphic layout provided in an embodiment of the present application;

[0075] Figure 7 A schematic diagram of the vertex-arc symmetry axis flip isomorphism of the coding isomorphic layout provided in an embodiment of the present application;

[0076] Figure 8 A schematic diagram of a queuing network model provided in an embodiment of the present application;

[0077] Figure 9 A schematic diagram of a queuing network model after network reconstruction provided in an embodiment of the present application;

[0078] Figure 10A pseudocode diagram of the improved variable neighborhood search algorithm provided in an embodiment of the present application. DETAILED DESCRIPTION

[0079] To facilitate understanding of the present invention, the present invention will be described more fully below with reference to the accompanying drawings. Preferred embodiments of the present invention are shown in the accompanying drawings. However, the present invention may be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to provide a more thorough and comprehensive understanding of the present disclosure.

[0080] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0081] The terms used in the specification of the present invention herein are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the associated listed items.

[0082] Example 1:

[0083] A flexible intelligent manufacturing unit system, see Figure 1 , the system includes: a loading port buffer, a unloading port buffer, a robot and several workstations;

[0084] The loading port buffer sends a workpiece handling request to the robot;

[0085] The robot transports the workpiece to the workstation according to the workpiece transport request issued by the loading port buffer;

[0086] The workstation processes the workpiece and sends a workpiece handling request to the robot after the processing is completed;

[0087] The robot transports the workpiece to the unloading port buffer according to the workpiece transport request issued by the workstation.

[0088] As a preferred embodiment, see Figure 1 , the workstation includes a workpiece processing buffer and processing equipment;

[0089] The processing equipment processes the workpiece, and after the processing is completed, sends a workpiece transportation request to the robot. The robot transports the workpiece to the workpiece processing buffer area or the unloading port buffer area according to the workpiece transportation request sent by the processing equipment.

[0090] For example, the arrival of tasks outside the system follows a Poisson process; each workstation contains a buffer with finite capacity and a processing device, the task processing time follows a general distribution, and the service rule is first-come, first-served (FCFS); the speed of the robot is a constant, that is, the transportation time of the workpiece between workstations is a constant distribution; the blocking mechanism of the processing process is post-service blocking; the blocking mechanism of the transportation process is pre-service blocking to avoid deadlock; the layout positions of the workstations along the robot guide rail are known, and the distances between each layout position are also known.

[0091] In this embodiment, each workstation contains a buffer with limited capacity and a processing device. The task processing time follows a general distribution, and the service rule is first-come, first-served (FCFS). The speed of the robot is constant, that is, the transportation time of the workpiece between workstations follows a constant distribution. The blocking mechanism of the processing process is post-service blocking; the blocking mechanism of the transportation process is pre-service blocking, so as to avoid deadlock and effectively reduce the production cost of the workpiece.

[0092] Example 2:

[0093] This embodiment provides an optimization method for a flexible intelligent manufacturing unit system for the system of embodiment 1. The method is applied to a flexible intelligent manufacturing unit system. Figure 2 , the method comprises the following steps:

[0094] Step S1: Based on the process path of the flexible intelligent manufacturing unit system, a target optimization problem is constructed with the goal of maximizing the average output rate of the system workpieces;

[0095] Step S2: Solve the target optimization problem to obtain a system facility layout optimization strategy;

[0096] Step S3: completing the facility layout optimization of the flexible intelligent manufacturing unit system according to the system facility layout optimization strategy.

[0097] Specifically, in step S1, the process of constructing the target optimization problem includes:

[0098] Assign m+2 workstations, loading port buffers, and unloading port buffers to m+2 known layout positions, and set node position constraints and position capacity constraints. The expressions are:

[0099] ;

[0100] ;

[0101] Based on the node location constraints and location capacity constraints, the target optimization problem is constructed and expressed as:

[0102] max Θ( X );

[0103] Decision variables: X ={ x i,k}, , i, k =0, 1, …, m +2;

[0104] Where m represents the number of nodes in the system, including workstations, loading port buffers, and unloading port buffers; X ={ x i,k} represents the decision variable; Θ represents the average output rate of the system; i represents the sequence number of the node in the system, i =0, 1, …, m +1, where i =0 and m +1 represents the unit loading / unloading port buffer zone; k represents the node i at the kth position.

[0105] It can be understood that based on the process path of the flexible intelligent manufacturing unit system, the target optimization problem can be constructed according to the general layout problem of the flexible intelligent manufacturing unit system, so as to fully reflect the problems existing in the current flexible intelligent manufacturing unit system facility layout from the data, and then consider the blocking impact and general distribution service time to construct the target optimization problem.

[0106] As a preferred embodiment, in step S2, the process of solving the target optimization problem includes:

[0107] S21: Generate an initial solution based on the system’s process path;

[0108] S22: Based on the initial solution, the 2-Opt exchange strategy is used to design the neighborhood structure, and homogeneous layout processing is performed based on the robot transportation path network;

[0109] S23: Setting a queuing network model to approximately solve the output rate of the flexible intelligent manufacturing unit system to obtain the objective function value of the optimization problem;

[0110] S24: Based on the objective function value of the optimization problem, an improved variable neighborhood search algorithm is set to iteratively optimize and solve the target optimization problem to obtain the system facility layout optimization strategy.

[0111] Specifically, in step S21, see Figure 3, using natural numbers to identify nodes (workstations, unit loading / unloading buffers) and unit layout positions, encoding them in a permutation encoding manner, and implicitly storing the layout position information through the permutation index. π Index i Starting from 0, sort by i The element (i.e. node) at position is represented as π ( i ).

[0112] Based on the principle of minimizing logistics costs, work units with strong logistics correlation (i.e., process path sequence) are usually arranged adjacent to each other to shorten the logistics transportation distance and improve the system output rate.

[0113] Based on this, the method of producing the initial solution is improved. Define a directed graph D =( V , A ) Description contains n ( n ≤ m ) process path of the first process, where the vertex set V ={ W i , 1≤ i ≤ n}∪{ B 0, B m+1}; for ∀ W i , W j ∊ V , two-element group ( W i , W j )express W i The corresponding process is W j The immediate preceding process of the corresponding process, a two-element group ( W i , W j ) set composition D Arc Set A .

[0114] From the vertex W i Start along the arc to reach the vertex W j for W i arrive W j The path is denoted as { W i , …,W j}.because D A path can be represented as a set of processes with a close relationship between the preceding and the following, so it can be based on D The path to generate the initial solution is as follows:

[0115] Step A: Find all B 0 to B m+1 If there is n ( n If the vertices of the paths do not intersect (there are no common vertices between the paths), execute step B; otherwise, execute step C.

[0116] Step B: From n Randomly select a road from the roads and record it as π ={ B 0, …, B m+1};implement n -1 repeated insertion operation, each time randomly selecting a path, deleting the elements before and after the path, and inserting it π At the second to last position of D Delete the arcs contained in the road; if the arrangement π Include V If all vertices are found, execute step 5; otherwise, execute step 4.

[0117] Step C: Select the longest path, recorded as π ={ B 0, …, B m+1};exist D Delete all arcs on the longest path; if the arrangement π Include V If all vertices are found, execute step E; otherwise, execute step D.

[0118] Step D: D Select an arc ( u , v ),satisfy u In the arrangement u In, and in D There is u The arc leaving is u Go along the arc direction as the starting point, and when you pass the vertex y Also there π Stop in the middle and get a path u , v , …, x , y};turn up u In the arrangement uPosition in and insert the arrangement { u , v , …, x}; If the arrangement π Include V If all vertices are found, execute step E; otherwise, execute step D.

[0119] Step E: Arrange π Encode and convert into an initial solution.

[0120] Specifically, in step S22, the process of using the 2-Opt exchange strategy to design the neighborhood structure includes: π ={ a 1, a 2, …, a n}, using the local search "2-Opt" exchange strategy, take i ∊[1, n ], swap two adjacent elements π ( i )and π ( i +1) position (if i = n Then exchange a 1 and a n ), which can generate n Neighborhood solutions.

[0121] If you take j ∊[2, n -1], the exchanged elements can be π ( i )and π (( i + j ) mod n ). For any two permutations π 1 and π 2. Through a limited number of exchanges π The two elements of 2 can be converted into π 1.

[0122] Based on this, the neighborhood structure is designed by exchanging the positions of two elements in this application. π ={ a 1, a 2, …, a n},make k ∊[1, n ], take ∀ i ∊[1, n ], neighborhood structure 𝒩 k (π ) for exchange π ( i )and π (( i + j ) mod n ) is the solution set formed by i ∊[1, n ], we have the equation 𝒩 i ( π ) = 𝒩 n-i ( π ) holds, so the number of neighborhood structures is ⌊ n / 2⌋.

[0123] Specifically, in step S22, see Figure 4-Figure 7 The process of isomorphic layout processing based on the robot transport path network includes: defining a directed graph D′ =( V′ , A′ ) describes the robot's transport path, where the vertex set V′ = π ; For ∀ π ( i ), π ( j )∊ V , two-element group ( π ( i ), π ( j )) represents the reachable path of the robot, and the two-element group ( π ( i ), π ( j ))Set composition D′ Arc Set A′ .

[0124] According to the isomorphism definition of directed graphs, two isomorphic directed graphs D′ 1 and D′ 2 have the same topological properties, and performing specific geometric transformations on them (such as rotation, flipping, etc.) can maintain the topological properties of the graphics.

[0125] Based on this, different encodings of mutually isomorphic graphs are defined as "isomorphic encodings". For the transportation path networks with different robot guide rail structures, the scale of the solution space of the neighborhood search is reduced by eliminating isomorphic encodings.

[0126] See also Figure 4 , Definition 1 (Center Rotation Isomorphism): Encoding π 1={1, 2, …, x -1, x ,x +1, …, n}, randomly pick an element x , then the encoding π 2={ x , x +1, …, n , 1, 2, …, x -1} with π 1 are mutually isomorphic codes.

[0127] See also Figure 5 , Definition 2 (Vertex symmetry axis flip isomorphism): The lengths are n Encoding π 1 and π 2. If n is an even number, u = n / 2, then for ∀ i ∊[1, u ],set up π 2( u - i )= π 2( u + i ), that is, exchange i 1st and 2nd i + u The position of the bit element after the exchange π 2 and π 1 are mutually isomorphic codes.

[0128] See also Figure 6 , Definition 3 (arc symmetry axis flip isomorphism): the lengths are n Encoding π 1 and π 2. If n is an even number, u = n / 2, then for ∀ i ∊[0, u -1], let π 2(1+ i )= π 2( n - i ), that is, exchange 1+ i 1st and 2nd n - i The position of the bit element after the exchange π 2 and π 1 are mutually isomorphic codes.

[0129] See also Figure 7 , Definition 4 (Vertex-arc symmetry axis flip isomorphism): The lengths are n Encoding π 1 andπ 2. If n is an odd number, let k= ⌈ n / 2⌉, then for ∀ i ∊[0, ⌊ n / 2⌋], let π 2( ki )= π 2( k+i ), that is, exchange ki 1st and 2nd k+i The position of the bit element after the exchange π 2 and π 1 are mutually isomorphic codes.

[0130] For the circular bidirectional guide rail unit, the robot moves along the circular bidirectional guide rail, and the directed graph corresponding to the encoding is a bidirectional graph. At this time, Definition 1, Definition 2, Definition 3 and Definition 4 are all valid; for the circular unidirectional guide rail unit, the robot moves along the circular unidirectional guide rail, and the directed graph corresponding to the encoding is a unidirectional graph. At this time, only Definition 1 is valid; for the linear bidirectional guide rail unit, the robot moves along the linear bidirectional guide rail, and there is no adjacent relationship between the first and last positions in the directed graph corresponding to the encoding. At this time, only Definition 3 is valid.

[0131] Taking the circular track as an example, the coding isomorphism diagram is as follows Figure 4-Figure 7 shown.

[0132] If Definition 1 holds, the length is n The encoding has n -1 center rotation isomorphic code; if Definition 2 holds, the length is n The encoding has n / 2 vertex symmetry axis flip isomorphism encoding; if Definition 3 holds, the length is n The encoding has n / 2 vertex symmetry axis flip isomorphism encoding; if Definition 4 holds, the length is n The encoding of has only one vertex-arc symmetry axis flip isomorphism encoding.

[0133] For units with different guide rail structures, the solution space after eliminating isomorphic coding is shown in Table 1.

[0134] Table 1:

[0135] The size of the solution space after eliminating homomorphic coding

[0136]

[0137] Specifically, in step S23, see Figure 8 ,The process of setting up the queuing network model includes : ,establishing an open queuing network model with a finite capacity buffer for ,unit system performance analysis, such as Figure 2The figure shows a schematic diagram of a queuing network model for a unit system consisting of three workstations and an arbitrary process path structure. Based on the model assumptions and Kendall notation, the queuing model for each workstation node is the GI / G / 1 / K model, and the queuing model for the robot node is the GI / D / 1 / 1 model. The purpose of queuing network modeling is to describe the operation of a stochastic system based on stochastic process theory and calculate system performance indicators (such as the average system output rate). This allows us to determine whether the objective function or constraints are met during the iterative optimization process of the optimization algorithm for the layout optimization problem.

[0138] It can be understood that generating an initial solution based on the system's process path can reflect the workpiece output efficiency of the current system and provide initial parameters for the target optimization problem to be solved, so as to optimize based on the initial solution to improve the system's workpiece output rate; using the 2-Opt exchange strategy for neighborhood structure design and performing isomorphic layout processing based on the robot transportation path network can preliminarily optimize the layout of the system facilities, thereby facilitating the use of the set queuing network model to approximate the output rate of the flexible intelligent manufacturing unit system to obtain the objective function value of the optimization problem. The set queuing network model can quickly and effectively establish connectivity with blocking effects and generally distributed service times between facilities in the system; then, based on the objective function value of the optimization problem, an improved variable neighborhood search algorithm is set to iteratively optimize and solve the target optimization problem to obtain a system facility layout optimization strategy, which effectively solves the deadlock phenomenon of facilities in the flexible intelligent manufacturing unit system and effectively reduces the production cost of workpieces.

[0139] Specifically, in step S23, the process of setting a queuing network model to approximately solve the output rate of the flexible intelligent manufacturing unit system includes:

[0140] S231: adding virtual nodes to reconstruct the queuing network model, thereby obtaining a queuing network model in which each node is independent of each other;

[0141] S232: solving the workpiece processing input or output rate of the workstation node using the reconstructed queuing network model;

[0142] S233: using the reconstructed queuing network model to solve the input or output rate of the workpiece transportation of the robot node;

[0143] S234: Based on the workpiece processing input or output rate of the workstation node and the workpiece transportation input or output rate of the robot node, calculate the average output rate of the workpiece processing in the system, and iteratively optimize the average output rate to obtain a system facility layout optimization strategy;

[0144] Wherein, the queuing network model is a neural network model.

[0145] Specifically, in step S231, the process of adding virtual nodes to reconstruct the queuing network model into a Jackson network in which each node is independent of each other by adding virtual nodes based on the generalized expansion method. Figure 9 As shown, at each workstation node j Add a virtual node with queuing model GI / G / ∞ h Used to accommodate blocked artifacts. When a node j When saturated, the node i The completed workpiece is blocked with probability Enter the virtual node h After a blocking delay, the artifact attempts to enter the node j , when the node j While still in saturation, the artifact will be Enter the virtual node again until the node j The workpiece can enter the node only when it turns to a non-saturated state and the robot is in an available state j .

[0146] Specifically, in step S232, in the material processing system, the process of using the reconstructed queuing network model to solve the workpiece processing input or output rate of the workstation node includes:

[0147] Set the arrival rate constraint and blocking probability constraint of the workpiece arriving at the workstation node and return blocking probability The process includes:

[0148] The workpieces arriving at each workstation include the workpieces output from the unit loading port and the workpieces completed and output from other workstations, so the workstation node j The arrival rate constraint is:

[0149] , (4)

[0150] in, It is the rate at which the material outside the system reaches the unit loading port.

[0151] The input process of each workstation node is a branch of the output process of the robot node. Therefore, without considering the blocking, according to the property that the renewal interval distribution of the branch process is the random sum of the geometric distribution of the original renewal interval, the workstation node j The input process squared coefficient of variation is:

[0152] ;

[0153] Based on the Diffusion Approximation Method, the blocking probability constraint of the GI / G / 1 / K model is approximated. and return blocking probability constraints as follows:

[0154] ,

[0155] Among them, the workpiece flow intensity ,coefficient ;

[0156] ,

[0157] in, r 1 and r 2 are equations The roots of , For nodes j The effective input rate, For virtual nodes h arrival rate.

[0158] If you reach the node j The artifact is blocked, the artifact needs to be in the virtual node h Waiting time for node i Therefore, the delay time of the blocked workpiece is the same as the remaining processing time of the node. j The remaining processing time of the workpieces processed in the same way has the same distribution. Based on the renewal theory, the rate of the remaining processing time distribution is:

[0159] ,

[0160] in, σ j 2 is the variance of processing time.

[0161] Assume that the arrival node j In the workpiece flow, the blocked workpiece flow and the unblocked workpiece flow are separated, then the squared coefficient of variation of the input process distribution in its decomposition form is:

[0162] ;(9)

[0163] Since the nodes of the reconstructed network are independent of each other, according to Marshall's formula, the calculation nodes j The squared coefficient of variation of the output process distribution is:

[0164] ,

[0165] in, For artifacts in node j The average queue time is , For nodes j The average captain.

[0166] Based on the principle of balanced input and output rates of nodes under steady-state conditions, and considering the resource coordination constraints between workstations and robots, nodes j The output rate is:

[0167] ,

[0168] in, For virtual nodes h The workpiece flow intensity, Indicates the rate at which the material outside the system reaches the unit loading port. α i,j Representation node i arrive j The path probability, j =1, 2, …, m +1; ρ j Indicates a workstation j The workpiece flow intensity, represents the rate of distribution of the remaining processing time of the workpiece in workstation node j, The coefficient representing the difference in workpiece flow, , Representation node j The effective input rate, For virtual nodes h The arrival rate; represents the blocking probability of the robot node, For virtual nodes h The workpiece flow intensity, Indicates a workstation j The exponential function value of the workpiece flow intensity with respect to the coefficient of variation, .

[0169] Specifically, in step S233, in the material storage and transportation system, the process of using the reconstructed queuing network model to solve the input or output rate of the workpiece transportation of the robot node includes:

[0170] In addition to the node m +2, the workpieces output from all nodes need to reach the robot node first, so the input rate constraint of the robot is set, and the expression is:

[0171] ;

[0172] Robot responds from the node i To Node j When requesting a transport task, you need to first start from the location where the last transport task was completed. k , no-load running to the node i , and then execute the node i To Node j The load stroke of the robot is composed of the no-load stroke and the load stroke.

[0173] ,

[0174] in, is the average service time of the robot, and are the workpiece flow intensities of the robot's no-load stroke and loaded stroke, respectively.

[0175] The input process of the robot consists of the unit loading port and the output process of all workstation nodes. Therefore, without considering the blocking, the square coefficient of variation of the robot's arrival process is:

[0176] ;

[0177] The queuing model of the robot node is a GI / D / 1 / 1 model without a buffer. The saturation probability constraint of the robot is set as follows:

[0178] ;

[0179] In each handling process, the robot's empty travel and loaded travel are two interdependent continuous processes. Therefore, the probability constraint of the robot being in a non-idle state is set, and the expression is:

[0180] ;

[0181] Similar to the blocking and non-blocking diversion process of the workstation node, the squared coefficient of variation of the input process of the robot node is:

[0182] .

[0183] According to the input rate constraint, saturation probability constraint and probability constraint of the robot in the non-idle state, the input or output rate of the robot is calculated. The expression is:

[0184] ;

[0185] The square coefficient of variation of the robot's output process is:

[0186] ,

[0187] in, is the square coefficient of variation of the robot's running time, is the average captain of the robots, and , ρ r Represents a robot node r The workpiece flow intensity, Indicates the no-load workpiece flow intensity, Indicates the workpiece flow intensity of the load.

[0188] Specifically, in step S234, based on the workpiece processing output rate of the workstation node and the input or output rate of the workpiece transportation of the robot node, the average output rate of the workpiece processing in the system is calculated, and the expression is:

[0189] ;

[0190] Determine whether the average output rate of workpiece processing in the system meets the node position constraints and position capacity constraints. If so, the maximum system output rate obtained in the iterative solution process will be used as the final system facility layout optimization strategy; if not, continue to iterate until the average output rate obtained by the solution meets the node position constraints and position capacity constraints, and use the maximum system output rate obtained as the final system facility layout optimization strategy.

[0191] Specifically, in step S24, see Figure 10 , set up an improved variable neighborhood search algorithm to iteratively optimize and solve the target optimization problem, and obtain the system facility layout optimization strategy.

[0192] Definitions of variables and symbols in this embodiment:

[0193] in, m Indicates the total number of nodes in the unit (including workstations and unit loading / unloading buffers); i represents The node's sequence number, i =0, 1, …, m +1, where i =0 and m +1 indicates the buffer zone of the unit loading / unloading port; λ 0 represents the external arrival rate of the system; μ i Indicates a workstation i The processing rate, i =1, 2, …, m ; Indicates a workstation iSquared coefficient of variation (SCV) of the processing distribution; ρ i Indicates a workstation i Workpiece flow intensity; N i Indicates a workstation i The upper limit of the buffer capacity; K i Indicates a workstation i The upper capacity limit of Indicates a workstation i The blocking probability of α i,j Representation node i arrive j The path probability, j =1, 2, …, m +1; d i,j Representation node i arrive j Distance; Λ i Representation node i Input rate; Θ i Representation node i Output rate; Representation node i The squared coefficient of variation of the input process distribution; Representation node i The squared coefficient of variation of the output process distribution; v Indicates the running speed of the robot; represents the saturation probability of the robot node; represents the blocking probability of the robot node; Λ r represents the input rate of the robot node; Θ r represents the output rate of the robot node; represents the squared coefficient of variation of the input process distribution of the robot node; represents the squared coefficient of variation of the output process distribution of the robot node; Θ represents the average output rate of the system, that is, the number of finished products output by the system per unit time. B i Representation node i buffer zone; M i Indicates a workstation i processing equipment; W i Indicates a workstation i ; R Represents a robot node; L k Indicates the klayout positions.

[0194] It can be understood that adding virtual nodes to reconstruct the grid of the queuing network model can better reflect the status of each facility in the system from the data, thereby obtaining a queuing network model in which each node is independent of each other, and using the queuing network model in which each node is independent of each other to solve the workpiece processing input or output rate of the workstation node and the input or output rate of the workpiece transportation of the robot node; then based on the workpiece processing input or output rate of the workstation node and the input or output rate of the workpiece transportation of the robot node, the average output rate of the workpiece processing in the system is calculated, and the average output rate is iteratively optimized to obtain the system facility layout optimization strategy, and finally, according to the system facility layout optimization strategy, the facility layout optimization of the flexible intelligent manufacturing unit system is completed, effectively solving the deadlock phenomenon of facilities in the flexible intelligent manufacturing unit system and effectively reducing the production cost of the workpiece.

[0195] In this embodiment, based on the process path of the flexible intelligent manufacturing unit system, a target optimization problem can be constructed according to the general layout problem of the flexible intelligent manufacturing unit system, so as to fully reflect the problems existing in the facility layout of the current flexible intelligent manufacturing unit system from the data, and then consider the blocking impact and general distribution service time, construct a target optimization problem, and solve the target optimization problem to obtain the facility layout optimization strategy of the flexible intelligent manufacturing unit system, and complete the facility layout optimization of the flexible intelligent manufacturing unit system according to the system facility layout optimization strategy, which can effectively solve the deadlock phenomenon of facilities in the flexible intelligent manufacturing unit system and effectively reduce the production cost of workpieces.

[0196] Example 3:

[0197] This embodiment provides an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein when the processor executes the computer program, steps of a method for optimizing a flexible intelligent manufacturing unit system are implemented.

[0198] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention's description and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for optimizing a flexible intelligent manufacturing unit system, characterized in that: The system includes: a loading port buffer, a unloading port buffer, a robot and several workstations; The loading port buffer sends a workpiece handling request to the robot; The robot transports the workpiece to the workstation according to the workpiece transport request issued by the loading port buffer; The workstation processes the workpiece and sends a workpiece handling request to the robot after the processing is completed; The robot transports the workpiece to the discharge port buffer according to the workpiece transport request issued by the workstation; The workstation includes a workpiece processing buffer and processing equipment; The processing equipment processes the workpiece, and after the processing is completed, sends a workpiece transport request to the robot, and the robot transports the workpiece to the workpiece processing buffer or the unloading port buffer according to the workpiece transport request sent by the processing equipment; The method comprises the following steps: Based on the process path of the flexible intelligent manufacturing cell system, a target optimization problem is constructed with the goal of maximizing the average output rate of the system workpieces; Solving the target optimization problem to obtain a system facility layout optimization strategy; Complete the facility layout optimization of the flexible intelligent manufacturing unit system according to the system facility layout optimization strategy; The process of constructing the target optimization problem includes: Assign m+2 workstations, loading port buffers, and unloading port buffers to m+2 known layout positions, and set node position constraints and position capacity constraints. The expressions are: ； ； Based on the node location constraints and location capacity constraints, the target optimization problem is constructed and expressed as: max Θ( X ); Where m represents the number of nodes in the system, including workstations, loading port buffers, and unloading port buffers; X ={ x i,k } represents the decision variable; Θ represents the average output rate of the system; i represents the sequence number of the node in the system, i =0, 1, …, m +1, where i =0 and m +1 represents the unit loading / unloading port buffer zone; k represents the node i at the kth position; The process of solving the target optimization problem includes: Generate an initial solution based on the system's process path; Based on the initial solution, a 2-Opt exchange strategy is used to design the neighborhood structure, and a homogeneous layout is performed based on the robot transport path network. Setting up a queuing network model to approximately solve the output rate of the flexible intelligent manufacturing unit system and obtain the objective function value of the optimization problem; Based on the objective function value of the optimization problem, an improved variable neighborhood search algorithm is set to iteratively optimize and solve the target optimization problem to obtain the system facility layout optimization strategy.

2. The optimization method of the flexible intelligent manufacturing unit system according to claim 1, characterized in that: The process of setting up a queuing network model to approximate the output rate of the flexible intelligent manufacturing unit system includes: Add virtual nodes to reconstruct the queuing network model, and obtain a queuing network model in which each node is independent of each other; The reconstructed queuing network model is used to solve the workpiece processing input or output rate of the workstation node; The reconstructed queuing network model is used to solve the input or output rate of the workpiece transportation of the robot node; Based on the workpiece processing input or output rate of the workstation node and the workpiece transportation input or output rate of the robot node, the average output rate of the workpiece processing in the system is calculated, and the average output rate is iteratively optimized to obtain the system facility layout optimization strategy; Wherein, the queuing network model is a neural network model.

3. The optimization method of the flexible intelligent manufacturing unit system according to claim 2, characterized in that: The process of solving the workpiece processing input or output rate of the workstation node using the reconstructed queuing network model includes: Set the arrival rate constraint and blocking probability constraint of the workpiece arriving at the workstation node and return blocking probability , the expressions are: ； ； ； According to the arrival rate constraint and blocking probability constraint and return blocking probability , calculate the workpiece input or output rate of the workstation node, the expression is: ； in, Indicates the rate at which the material outside the system reaches the unit loading port. α i,j Representation node i arrive j The path probability, j =1, 2,…, m +1; ρ j Indicates a workstation j The workpiece flow intensity, represents the rate of distribution of the remaining processing time of the workpiece in workstation node j, The coefficient representing the difference in workpiece flow, , Representation node j The effective input rate, For virtual nodes h The arrival rate; represents the blocking probability of the robot node, For virtual nodes h The workpiece flow intensity, Indicates a workstation j The exponential function value of the workpiece flow intensity with respect to the coefficient of variation, .

4. The optimization method of the flexible intelligent manufacturing unit system according to claim 3, characterized in that: The process of solving the input or output rate of the workpiece transportation of the robot node using the reconstructed queuing network model includes: Set the input rate constraint of the robot, the expression is: ； Set the robot's saturation probability constraint, the expression is: ； Set the probability constraint of the robot being in a non-idle state, the expression is: ； According to the input rate constraint, saturation probability constraint and probability constraint of the robot in the non-idle state, the input or output rate of the robot is calculated. The expression is: ； in, ρ r Represents a robot node r The workpiece flow intensity, Indicates the workpiece flow intensity without load, Indicates the workpiece flow intensity of the load.

5. The optimization method of the flexible intelligent manufacturing unit system according to claim 4, characterized in that: Based on the workpiece processing output rate of the workstation node and the input or output rate of the workpiece transportation of the robot node, the average output rate of the workpiece processing in the system is calculated, and the expression is: ； Determine whether the average output rate of workpiece processing in the system meets the node location constraints and location capacity constraints. If so, the maximum system output rate obtained during the iterative solution process is used as the final system facility layout optimization strategy; If not, continue to iterate until the average output rate obtained satisfies the node location constraint and location capacity constraint, and use the maximum system output rate obtained as the final system facility layout optimization strategy.

6. An electronic device, characterized in that: The method comprises a memory, a processor, and a computer program stored in the memory and running on the processor, wherein when the processor executes the computer program, the steps of the method according to any one of claims 1 to 5 are implemented.

Images