Intelligent production unit layout and energy consumption optimization method and system

Abstract

The invention discloses an intelligent production unit layout and energy consumption optimization method, and relates to the technical field of intelligent production unit scheduling optimization, and the method comprises the steps: taking the maximization of productivity and the minimization of AGV energy consumption as targets, and constructing a mathematical model and a target function; establishing an intelligent production unit simulation model according to the mathematical model; and randomly generating an initial population by adopting an improved NSGA-II algorithm, carrying out iteration, calling a simulation model in the process to obtain an individual fitness value, outputting a non-dominated sorting result after iteration is completed, obtaining equipment layout information and AGV speed information of an optimal solution enabling a target function to meet constraint conditions, and completing optimization of intelligent production unit layout and energy consumption. According to the method, the randomness of a real production scene is effectively reserved by establishing a multi-target mathematical model and a simulation model of equipment layout and AGV energy consumption, equipment layout information and AGV speed information are solved by improving an NSGA-II algorithm, and a productivity and energy-saving balance point is found to obtain an overall optimal solution.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent production unit scheduling optimization, and more specifically, to an intelligent production unit layout and energy consumption optimization method and system. Background Art

[0002] FLP involves the scientific configuration of all production factors (such as equipment, buffers, personnel, etc.) that constitute the production system within a given space (such as a workshop or warehouse). By adjusting their relative positions and arrangements, the optimal performance of the overall system is achieved, which is crucial to achieving the agility, energy saving and economy of the manufacturing system. The layout planning problem mainly includes three parts: layout logic, layout modeling and model solving. The equipment layout of an intelligent manufacturing factory should be highly flexible, modular and easy to reconfigure. Therefore, more attention should be paid to layout logic such as unit layout and robust layout, and higher requirements should be placed on layout modeling and solving algorithms. The layout model mainly includes graph theory models (such as queuing theory), mathematical analysis models (such as nonlinear programming models, mixed integer programming models) and simulation models. The solution method can be optimized by using an exact algorithm according to the scale of the problem, or suboptimal by using a heuristic algorithm or a meta-heuristic algorithm.

[0003] The use of optimization theory and methods to conduct in-depth and systematic research on the two core issues of production cell construction and cell auxiliary resource allocation can provide key theoretical and methodological guidance for the implementation of intelligent cell production methods in discrete manufacturing enterprises. Intelligent Manufacturing Cell Layout Planning (IMCLP) relies on the efficient execution of material handling tasks by handling robots within the cell to achieve compact equipment layout within the cell. It is necessary to solve the two key issues of intelligent production cell layout and intelligent logistics storage and transportation resource allocation. Therefore, the intelligent production cell layout problem considering the intelligent material storage and transportation system not only includes the FLP problem, but also involves path network design and vehicle path planning problems.

[0004] Relevant research in recent years still uses the logistics distance and logistics intensity distribution of classic equipment layout to think about the problem. Intelligent manufacturing needs to form flexible and green production, and should also consider the coupling relationship between the logistics speed of handling equipment (AGV) and the layout of intelligent production units. At the same time, in recent years, the price of production factors, mainly land, has increased rapidly. Considering the compact and coordinated layout of intelligent production units is of great research value in reducing layout costs and reducing operating energy consumption. Therefore, the present invention provides a method and system for optimizing the layout and energy consumption of intelligent production units. Summary of the invention

[0005] The purpose of the present invention is to solve the problem of coordinated optimization of the ring layout and energy consumption of an intelligent production unit with AGV transportation in the prior art, to provide a method and system for optimizing the layout and energy consumption of an intelligent production unit, and to find the overall optimal solution for the production capacity and energy consumption of the intelligent production unit.

[0006] To achieve the above purpose, the technical method of the present invention is as follows:

[0007] Taking maximizing the capacity of the intelligent unit and minimizing the energy consumption of the AGV as the joint optimization goals, a capacity-energy consumption mathematical model and its objective function are constructed;

[0008] According to the capacity-energy consumption mathematical model, a simulation model of an intelligent production unit with AGV circular transportation is established using modeling software;

[0009] The preset improved NSGA-Ⅱ genetic algorithm is used to randomly generate the initial population and perform population iteration. During the iteration process, the intelligent production unit simulation model is called to simulate to obtain the fitness values ​​of individuals in the population. After the iteration is completed, the final non-dominated sorting result, i.e., the approximate solution set of the Pareto optimal frontier, is output;

[0010] Based on the approximate solution set of the Pareto optimal frontier, the equipment layout information and AGV speed information of the optimal solution that makes the objective function meet the constraint conditions are obtained, and the optimization of the layout and energy consumption of the intelligent production unit is completed according to the equipment layout information and AGV speed information.

[0011] Preferably, the objective function of the joint optimization objective is:

[0012] maxΘ(X,v)

[0013] minE(X,v)

[0014] in, It means that the total number of workpieces num produced by the intelligent generation unit under the influence of variables X and v is divided by the simulation time t to obtain the production capacity Θ, E is the energy consumption of AGV, X is the equipment layout information, and v is the cruising speed of AGV; 0-1 decision variable X, that is, the equipment layout information X is:

[0015]

[0016] Among them, X is used to determine the relative position of the equipment and the loading and unloading ports. If the problem scale is n, X is an n×n dimensional 0-1 matrix, where x i,j If it is 1, it means that facility node i is in position j L j On, otherwise 0;

[0017] The calculation process of energy consumption E is as follows:

[0018]

[0019] E s =E am +E um +E a

[0020]

[0021] E a =P a t ij

[0022] F t =m total a ac +F f

[0023] F f =fm total g

[0024] m total =m 1 +c 1 m 2

[0025] l z = l ij -l ac -l de

[0026]

[0027] Where, E: energy consumption of AGV; n: problem scale, the number of locations where equipment can be arranged in the intelligent unit, including the loading port buffer and the unloading port buffer; E s : Total energy consumption of AGV for a single task; E am : Energy consumption during AGV acceleration phase; E um : Energy consumption of AGV during cruising; E a : Additional energy consumed by other energy-consuming components; P a : Power of energy-consuming components; h m : The total number of steps for workpiece m, p = 1, ..., m, q = 1, ..., h m ;t idle : AGV idle time; t ac1 : The acceleration time when the AGV cannot reach the cruising speed v when it moves from position i to position j; t ac2 : The time required to accelerate from 0 to cruising speed; t de1 : The braking time when the AGV cannot reach the cruising speed v when it moves from position i to position j; t de2 : The time required to decelerate from the cruising speed v to 0; t ij: The total time required for the AGV to travel from position i to position j; c 1 : When the AGV moves with a workpiece load, c 1 is 1, otherwise c 1 is 0; c 2 : When the AGV cannot reach the cruising speed v during the process of traveling from position i to position j, c 2 = 1, otherwise c 2 = 0; l z : The moving distance during cruising; l ij : The moving distance of the AGV from position i to position j; l ac : The moving distance when the AGV accelerates; l de : The moving distance when the AGV decelerates; η: Energy efficiency factor; F t : Driving force; F f : Frictional force; f: Rolling friction coefficient; g: Acceleration due to gravity; a ac : Acceleration; a de : Braking acceleration, v ac : v ac < v, the maximum speed when the robot cannot reach the cruising speed v during the process of traveling from position i to position j.

[0028] Preferably, the constraint conditions of the objective function include:

[0029]

[0030] Among them, what the above two expressions jointly express is that only one device or facility node can be arranged at each position, and each device or facility node can only be arranged at one position;

[0031] v min ≤ v ≤ v max

[0032] Among them, v min : The minimum AGV moving speed that can be set; v max : The maximum AGV moving speed that can be set.

[0033] Preferably, before performing population iteration, the improved NSGA-II genetic algorithm needs to first design the coding rules, specifically as follows:

[0034] The device layout information is encoded on the left side of the gene, and the encoding length depends on the problem scale n, that is, if there are n positions on the outer periphery of the circular track of the simulation model, the device layout information encoding is represented as an n-dimensional array, which can accommodate the loading port (P), the unloading port (D), and n - 2 workstations, and the set of workstations is {S 1 , S 2 ,..., S n-2};

[0035] The AGV speed information is encoded on the right side of the gene. At present, the AGV speed in the workshop is generally set at 0-3m / s. To ensure the accuracy of the experiment, three decimal places are retained. Therefore, 14-bit binary encoding is used. After conversion to decimal, the value range is [0,8192]. Dividing by 1000, the moving speed value of the AGV is obtained.

[0036] The equipment layout information encoding and AGV speed information encoding are merged to form an individual encoding. The encoding length depends on the problem scale n. In addition, the merged encoding cautiously adopts crossover and mutation operations to ensure the correctness of the information expressed by the encoding.

[0037] Preferably, the iterative process of the improved NSGA-II genetic algorithm includes:

[0038] Population initialization, randomly generate the initial population;

[0039] Performing a crossover operation on the gene codes of the individuals of the initialized population in sections;

[0040] After the individuals after crossover are subjected to single-point mutation operation and correctness test,

[0041] After the correctness test is passed, the child and parent populations are merged;

[0042] Traverse the merged population, identify isomorphic individuals, mark one of each isomorphic individual, and obtain the fitness value of one of each isomorphic individual and the fitness values ​​of other individuals without isomorphic individuals through the simulation model, share the fitness value of one of each isomorphic individual with other isomorphic individuals, complete the task of removing isomorphic individuals, reduce the number of calls to the simulation model, and compress the solution time;

[0043] The fast non-dominated sorting rule is used to construct a non-dominated set, and the other individuals outside the set are sorted to obtain the non-dominated sorting result;

[0044] Select elite individuals from the non-dominated sorting results to enter the next generation population;

[0045] Determine whether the iteration termination condition is met. If the iteration termination condition is met, output the final non-dominated sorting result, otherwise continue to iterate and execute the steps starting from the crossover operation.

[0046] Preferably, the crossover operation is performed using a partial matching crossover operator, and on the basis of the traditional two-point crossover partial matching operator, an intersection is added to the AGV speed information encoding part, that is, the first and second intersections are randomly generated in the equipment layout information encoding part, and the third intersection is randomly generated in the AGV speed information encoding part, and the codes between the first and second intersections of different individuals and the codes after the third intersection are exchanged. After the exchange is completed, a uniqueness check is performed to ensure that the encoded information is correctly expressed.

[0047] Preferably, for the single-point mutation operation, when the mutation point is encoded in the device layout information, a uniqueness check needs to be performed after the mutation to ensure that the encoded information is correctly expressed.

[0048] Preferably, the specific process of constructing a non-dominated set by using a fast non-dominated sorting rule and sorting other individuals outside the set includes two parts;

[0049] The first part consists of finding the non-dominated set, i.e., individuals of level 1, as follows:

[0050] For each individual in the population from i 1 to the population size and each other individual in the population from j i+1 to the population size, set the i-th individual in the two individuals currently compared to p and the j-th individual to q; if p dominates q, add q to the domination set of p, the domination set of p represents the individuals dominated by p, and the number of times q is dominated increases by 1; if q dominates p, add p to the domination set of q, the domination set of q represents the individuals dominated by q, and the number of times p is dominated increases by 1; if the number of times the i-th individual is dominated is 0, that is, no other individual dominates it, then the individual is classified as the first level and added to the first set of the non-dominated set F, and the level of the individual is set to 1;

[0051] The second part consists of classifying individuals other than those in class 1 as follows:

[0052] Initialize an empty set Q to temporarily store individuals of new levels; for each individual in the non-dominated set F whose current level is r=1 and for all other individuals dominated by this individual, reduce the number of times these individuals are dominated. If the number of times an individual is dominated becomes 0, that is, it is not dominated by any other individual, then set the level of this individual to r+1 and add it to Q; if Q is empty, it means that all individuals have been assigned levels, and exit the loop; otherwise, add the individuals in Q as individuals of the new level r+1 to the next set of the non-dominated set F, increase the value of r, and continue to process individuals of the next level.

[0053] Preferably, in order to speed up the process of non-dominated sorting and ensure the distribution and diversity of the population, the clustering distance of individuals is used as a supplementary rule for sorting between individuals, and individuals with large clustering distances are preferentially retained. The calculation formula of the clustering distance is as follows:

[0054]

[0055] The aggregation distance formula indicates that the aggregation distance of the i-th individual is equal to the sum of the differences between the i-1-th individual and the i+1-th individual in all objective function values. After adding the aggregation distance rule, the process of non-dominated sorting includes: first construct a non-dominated set and calculate the aggregation distance of each individual, then sort them in ascending order according to the level divided by the non-dominated set, and sort the individuals in the same level in descending order according to the size of the aggregation distance.

[0056] To achieve the purpose of the invention, the present invention also provides an intelligent production unit layout and energy consumption optimization system, applying an intelligent production unit layout and energy consumption optimization method described in the above technical solution, the system includes:

[0057] A digital model building unit is used to build a capacity-energy consumption mathematical model and its objective function with maximizing the capacity of the intelligent unit and minimizing the energy consumption of the AGV as the joint optimization goal;

[0058] A simulation unit, used to establish a simulation model of an intelligent production unit with AGV circular transportation using modeling software according to the capacity-energy consumption mathematical model;

[0059] The optimal solution solving unit is used to randomly generate an initial population using a preset improved NSGA-Ⅱ genetic algorithm and perform population iteration. During the iteration process, the intelligent production unit simulation model is called to perform simulation to obtain the fitness value of individuals in the population. After the iteration is completed, the Pareto optimal solution set for constructing the production capacity-energy consumption mathematical model is output;

[0060] The layout and energy consumption optimization unit is used to obtain the equipment layout information and AGV speed information of the optimal solution that makes the objective function meet the constraint conditions based on the Pareto optimal solution set, and complete the optimization of the layout and energy consumption of the intelligent production unit according to the equipment layout information and AGV speed information.

[0061] Compared with the prior art, the present invention has the following beneficial effects:

[0062] The present invention effectively retains the randomness of the real production scene by establishing a multi-objective mathematical model and simulation model of equipment layout and AGV energy consumption in the intelligent production unit, and solves the equipment layout information and AGV speed information through the improved NSGA-II genetic algorithm to find the balance point between the two conflicting goals of production capacity and energy saving to obtain the overall optimal solution. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 This is a flow chart of a method for intelligent production unit layout and energy consumption optimization according to Example 1 of the present application;

[0064] Figure 2 This is a schematic diagram of the equipment layout of an intelligent production unit according to Example 1 of the present application;

[0065] Figure 3 This is a schematic diagram of individual coding of Example 1 of the present application;

[0066] Figure 4 This is a schematic diagram of a simulation model of an intelligent production unit with AGV circular transportation according to Example 1 of the present application;

[0067] Figure 5 This is a schematic diagram of a crossover operation including a partial matching crossover operator according to Embodiment 1 of the present application;

[0068] Figure 6 This is a schematic diagram of the selection operation of the NSGA-Ⅱ genetic algorithm of Example 2 of the present application;

[0069] Figure 7 This is a flow chart of the improved NSGA-Ⅱ genetic algorithm based on the simulation model of Example 2 of the present application. DETAILED DESCRIPTION

[0070] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0071] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.

[0072] Example 1

[0073] See also Figure 1 , Embodiment 1 of the present application provides a method for intelligent production unit layout and energy consumption optimization, comprising the following steps:

[0074] S1: Taking maximizing the capacity of the intelligent unit and minimizing the energy consumption of the AGV as the joint optimization goals, a capacity-energy consumption mathematical model and its objective function are constructed.

[0075] In this embodiment 1, the intelligent production unit is an organizational form of the production unit of the intelligent factory, which more compactly includes the two key production functions of material processing and material storage and transportation. The equipment in the unit is generally arranged around the outer side of the circular track for the robot to move, such as Figure 2 The figure shows a schematic diagram of the equipment layout of an intelligent production unit. Specifically, the material processing subsystem is composed of multiple workstations such as stamping machines, machining centers, CNC machine tools, and quality inspection machines, and the material storage and transportation subsystem is composed of automatic loading and unloading robots and storage buffer areas. The intelligent production unit composed of the two subsystems can realize functions such as processing automation, material transportation automation, fixture and tool replacement automation, and monitoring and detection automation.

[0076] The operation process in the intelligent production unit is carried out by the coordination of three types of resources: workstations, handling robots, and buffer areas. There are resource coordination constraints, that is, the insufficient capacity of one type of resource will cause the system to be blocked and wait, which will reduce efficiency. The handling robot responds to service requests according to the "first come, first served" rule and presents three working states: idle, empty movement, and loaded movement. If the load of the handling robot arrives at a workstation with a full buffer area, it will wait for unloading, and the intelligent production unit will be deadlocked.

[0077] The intelligent production unit meets the following conditions:

[0078] (1) The process of transporting the workpiece from the external area to the unit loading port is not considered, that is, the workpiece has been prepared at the beginning and can be directly selected from the unit loading port;

[0079] (2) After the materials processed outside the unit arrive, they are temporarily stored in the loading port buffer area, waiting for the AGV to transfer them to the workstation for the first process;

[0080] (3) Each workstation has only one machine and can only process one workpiece at a time. When the material arrives at a busy workstation, it can only wait in the front buffer area;

[0081] (4) After the material is processed at the processing station, it is temporarily stored in the back buffer area of ​​the workstation, waiting for the AGV to transfer it to the workstation of the next process. After all the processes of the material are completed, the finished product will be transported by the AGV to the unloading port buffer area of ​​the unit and then leave the unit;

[0082] (5) The capacity of the AGV is 1, and it can only carry one workpiece at a time. The AGV is considered to be moving straight when cruising on the track;

[0083] (6) All buffers can store multiple artifacts;

[0084] (7) Sudden production disturbances such as machine failures and workpiece repairs are not considered.

[0085] AGVs are battery-driven and often start and stop frequently when performing transfer tasks in the workshop. The actual energy consumption is bound to be higher than the energy consumption of the uniform speed. It is inappropriate to estimate the overall energy consumption of AGV tasks only by the energy consumption when driving at a uniform speed. Therefore, the AGV transportation energy consumption includes the energy consumption of loaded transportation and the energy consumption of empty transportation.

[0086] In this embodiment 1, the optimal combination of equipment layout scheme and AGV speed configuration is solved with production capacity and energy consumption as optimization targets. The equipment layout directly determines the upper limit of production capacity. At the same time, the scheme can be selected on the basis of maintaining the production capacity level, so that the AGV moving distance and time are shorter and the production efficiency is improved; when the distance is the same, a faster AGV speed may improve work efficiency, but it will also increase energy consumption. The coupling relationship between production capacity and energy consumption brings challenges to problem modeling and solution. The equipment layout optimization problem is a typical combinatorial optimization problem. The search space grows exponentially with the increase in the number of equipment. Combined with the AGV speed optimization problem, its multi-objective function, namely production capacity and energy consumption, is highly nonlinear and non-convex. Although a single-objective model can be constructed and another objective can be used as a constraint to reduce the difficulty of solution and shorten the solution time, multi-objective optimization can consider multiple objectives at the same time to reduce excessive dependence on a single objective, and by seeking the "Pareto-Optimal Set", the robustness and stability of the optimization results are enhanced. Therefore, in this embodiment 1, maximizing the production capacity of the intelligent unit and minimizing the energy consumption of the AGV are used as joint optimization targets.

[0087] Establish a mathematical model and its objective function:

[0088] The Intelligent Manufacturing Cell Loop Layout Problem (IMCLLP) with AGV transportation is to arrange several intelligent devices and buffers in a limited space outside the circular track, determine the relative positions of the devices and the loading and unloading buffers, so as to minimize the total energy consumption cost, material flow cost, layout area, maximize production capacity and other evaluation indicators. For the convenience of describing the IMCLLP studied in Example 1 of the present invention, the relevant parameters and variables are defined as follows:

[0089] X: equipment layout information; v: AGV cruising speed; v min : The minimum AGV moving speed that can be set; v max : The maximum robot movement speed that can be set; v ac :v ac<v, the maximum speed when the robot cannot reach the cruising speed during the process of moving from position i to position j; Θ: the production capacity of the intelligent unit, p = 1,..., m; E: the energy consumption of the AGV; n: the problem scale, the number of positions of the deployable devices in the intelligent unit, including the loading buffer and the unloading buffer; E s : the total energy consumption of the robot for a single task; E am : the energy consumption during the acceleration phase of the robot; E um : the energy consumption during the cruising of the robot; E a : the additional energy consumed by other energy-consuming components; P a : the power of the energy-consuming component; h m : the total number of processes of workpiece m, q = 1,..., h m ; t idle : the idle time of the AGV; t ac1 : the acceleration time when the robot cannot reach the cruising speed during the process of moving from position i to position j; t ac2 : the time required to accelerate from 0 to the cruising speed; t de1 : the braking time when the robot cannot reach the cruising speed during the process of moving from position i to position j; t de2 : the time required to decelerate from the cruising speed to 0; t ij : the total time required for the robot to move from position i to position j; c 1 : when the AGV moves with a loaded workpiece, c 1 is 1, otherwise c 1 is 0,; c 2 : when the robot cannot reach v during the process of moving from position i to position j, c 2 = 1, otherwise c 2 = 0; l z : the moving distance during cruising; l ij : the moving distance of the robot from position i to position j; l ac : the moving distance during the acceleration of the robot; l de : the moving distance during the deceleration of the robot; η: the energy efficiency factor; F t : the driving force; F f : the frictional force; f: the rolling friction coefficient; g: the acceleration due to gravity; a ac : the acceleration; a de : the braking acceleration.

[0090] The objective function is as follows:

[0091] max Θ(X, v)(1)

[0092] min E(X, v)(2)

[0093]

[0094] Among them, Θ represents the production capacity obtained by dividing the total number of workpieces num produced by the intelligent generation unit under the influence of variables X and v by the simulation time t.

[0095] Calculation process of energy consumption E:

[0096]

[0097] E s =E am +E um +E a (5)

[0098]

[0099] E a =P a t ij (8)

[0100] F t =m total a ac +F f (9)

[0101] F f =fm total g (10)

[0102] m total =m 1 +c 1 m 2 (11)

[0103] l z = l ij -l ac -l de (12)

[0104]

[0105] The 0-1 decision variables are as follows:

[0106]

[0107] The constraints are as follows:

[0108]

[0109] v min ≤v≤v max (18)

[0110] Among them, v min : The minimum AGV moving speed that can be set; v max: The maximum AGV moving speed that can be set. Formula (1) represents the maximum capacity Θ, and formula (2) represents the minimum energy consumption E; Formula (3) represents the calculation of capacity. The total number of workpieces produced by the system under the influence of variables X and v is obtained through simulation, and then divided by the simulation time; Formula (4) represents the total energy consumption of AGV; Formula (5) represents the energy consumption of a single movement of AGV, which is composed of the energy consumption of the acceleration stage, the energy consumption of the cruising stage, and the energy consumption of other energy-consuming components; Formulas (6) to (8) respectively represent the calculation methods of the energy consumption of the acceleration stage, the energy consumption of the cruising stage, and the energy consumption of other energy-consuming components; Formulas (9) and (10) respectively represent the calculation methods of the driving force and the friction force; Formula (11) represents the calculation method of the total weight; Formulas (12) to (14) represent the calculation method of the moving distance of AGV in each stage; In addition to the AGV moving speed v, the decision variables in the model also introduce a 0-1 decision variable X, as shown in Formula (15), to determine the relative position of the equipment or facility node and the loading and unloading port. If the problem scale is n, X is an n×n dimensional 0-1 matrix, where x i,j If it is 1, it means that the equipment or facility node i is in the jth position L j Otherwise, it is 0; Equation (16) and Equation (17) together express that only one device or facility node can be arranged at each location, and each device or facility node can only be arranged at one location; Equation (18) represents the settable range of AGV moving speed.

[0111] S2: Based on the capacity-energy consumption mathematical model, a simulation model of an intelligent production unit with AGV circular transportation is established using modeling software.

[0112] The energy consumption and production capacity of the smart unit with AGV transportation are conflicting goals. It is necessary to find out the key factors that affect the two in order to find the balance point for the optimal solution. The factors that affect the performance indicators of the smart unit mainly include: equipment layout, AGV movement speed, work intensity in the unit, buffer area capacity, etc. Since the smart unit is full of randomness, such as uncertainty in order arrival time, uncertainty in processing procedures and processing requirements for arriving workpieces, uncertainty in AGV transportation cycle, etc., its performance indicators are difficult to calculate using closed-form mathematical formulas, that is, the calculation results are difficult to restore the actual production scene well. Therefore, it is necessary to use simulation technology with strong simulation detail capabilities to model, combined with scientific experimental design, to provide more reliable system performance data and verify the key factors affecting system performance.

[0113] In Example 1 of the present invention, Tecnomatix Plant Simulation 15.0 software is used to build a simulation model of an intelligent production unit with AGV circular transportation. The operating system environment is PC Windows 10, and the hardware environment is Intel (R) CPU 2.50 GHz, 12.0 GB RAM. Figure 4The simulation model of the intelligent production unit with AGV circular transportation is shown. The simulation model is used to calculate the fitness value, and the fitness function is the system capacity and the AGV energy consumption, that is, formula (1) and formula (2), and the function value is calculated by the simulation model.

[0114]

[0115]

[0116] Table 1 Figure 4 Entity and object comparison table

[0117] (1) Source object Figure 4 In the figure, Buf_0 is on the left side. Since the simulation model in this paper only simulates the production in the intelligent unit system, the process of the external workpiece arriving in the system is realized by the Source object generating a corresponding number of MUs (to simulate the workpieces in the generation process) according to the time interval set by the user.

[0118] (2) Drain object Figure 4 In the figure below the Source object, similarly, the process of the workpiece leaving the intelligent unit after completing all the processing steps is realized by the Drain object recycling MU;

[0119] (3)Method object Figure 4 The object with the letter M in the middle) is used to write code programs so that the actual production logic can be better simulated in the simulation model. At the same time, the statistics and output of certain performance indicators also rely on the implementation of the program written in the Method object;

[0120] (4)EventController object Figure 4 The simulation time in the simulation model is controlled and the simulation speed is adjusted. Depending on this object, the operation of triggering a certain behavior at a specific time node can also be realized;

[0121] (5) Table object Figure 4 It is used to record the data that the user needs to record during the simulation process or to store intermediate values.

[0122] S3: The preset improved NSGA-Ⅱ genetic algorithm is used to randomly generate the initial population and perform population iteration. During the iteration process, the intelligent production unit simulation model is called to simulate to obtain the fitness values ​​of individuals in the population. After the iteration is completed, the final non-dominated sorting result, i.e., the approximate solution set of the Pareto optimal frontier, is output.

[0123] Improved NSGA-Ⅱ genetic algorithm design:

[0124] The improved NSGA-Ⅱ genetic algorithm first needs to design a coding rule. The individual coding of Example 1 of the present invention is composed of the equipment layout information coding and the AGV speed information coding. Figure 3 The schematic diagram of individual coding of this embodiment 1 is shown. The equipment layout information is encoded on the left side of the gene. The problem scale is the coding length. That is, if there are n positions on the periphery of the circular track, the equipment layout information coding is represented as an n-dimensional array, which can accommodate the loading port (P), the unloading port (D) and n-2 workstations. The set of workstations is {S 1 ,S 2 ,...,S n-2 For example, for a problem size of n = 6, 6 positions need to be allocated to 4 workstations and a pair of PD ports, and the relational array can be expressed as [S 3 ,S 1 ,D,S 4 ,P,S 2 ], the array indicates that in this layout scheme, position 1 is placed at workstation S 3 , place S in position 2 1 , position 3 puts down the material port, and so on. For the convenience of calculation, the loading port is represented as 0, the unloading port is represented as n-1, and the workstations are coded from 1 to n-2; the equipment layout code of this layout scheme is represented as [3,1,5,4,0,2]. The AGV speed information is encoded on the right side of the gene. At present, the workshop transfer AGV speed is generally set at 0~3m / s. To ensure accuracy, three decimal places are retained, such as 2.965m / s. Therefore, a 14-bit binary code is used, converted to decimal, and its value range is [0, 8192]. For example, the binary 00101110010101, converted to decimal, is 2965, and divided by 1000 to get the AGV moving speed value. In summary, the equipment layout information code and the AGV speed information code are merged to form an individual code. The code length depends on the scale of the problem; in addition, the merged code requires the cautious use of crossover and mutation operations to ensure the correctness of the information expressed by the code.

[0125] See also Figure 7 The iterative process of improving the NSGA-Ⅱ genetic algorithm in this embodiment 1 includes:

[0126] S11: Population initialization, randomly generating an initial population, that is, randomly generating equipment layout information codes and AGV speed codes, and then splicing them into individual codes;

[0127] S12: Perform crossover operation on the genetic codes of individuals in the initialized population in segments;

[0128] S13: Perform single-point mutation operation on the individuals after crossover and pass the correctness test;

[0129] S14: After the correctness test is passed, the child and parent populations are merged;

[0130] S15: traverse the merged population, identify isomorphic individuals, mark one of each isomorphic individual, and obtain the fitness value of one of each isomorphic individual and the fitness values ​​of other individuals without isomorphic individuals through the simulation model, share the fitness value of one of each isomorphic individual with other isomorphic individuals, complete the task of removing isomorphic individuals, reduce the number of calls to the simulation model, and compress the solution time;

[0131] S16: Use the fast non-dominated sorting rule to construct a non-dominated set, and sort the other individuals outside the set to obtain a non-dominated sorting result;

[0132] S17: Select elite individuals from the non-dominated sorting results to enter the next generation population;

[0133] S18: Determine whether the iteration termination condition is met. If the iteration termination condition is met, output the final non-dominated sorting result, otherwise continue to iterate and execute steps S12-S18.

[0134] In this embodiment 1, the population size is set to 70, the crossover operation adopts three-point crossover, the ratio is set to 0.8, and the mutation operation adopts single-point mutation, the ratio is 0.2.

[0135] The crossover and mutation operations are as follows:

[0136] See also Figure 5 , Figure 5 The crossover operation diagram of this embodiment 1 is shown in FIG. 1 . Considering that the equipment layout information encoding must ensure the uniqueness of the gene in the chromosome, that is, the gene only appears once in the chromosome, the partially matched crossover (PMX) operator is more suitable, but the chromosome of the AGV speed information encoding part does not need to comply with the uniqueness principle. The operation process of PMX is as follows: Figure 5 As shown, Figure 5The first step is to record the mapping relationship according to the positions of the two intersections, and exchange the codes between the two intersections; the second step is the situation of the two individuals after the crossover is completed; the third step shows the process of deduplication of duplicate points in the individuals. According to the mapping relationship, the duplicate points are corrected. This step is called "uniqueness check". Therefore, in Example 1 of the present invention, based on the traditional two-point crossover PMX operator, an intersection is added to the AGV speed information encoding part. That is, two intersections (the first and second intersections) are randomly generated in the equipment layout information encoding part, and one intersection (the third intersection) is randomly generated in the AGV speed information encoding part. The codes between the first and second intersections of the chromosomes of different individuals and the codes after the third intersection are exchanged, and a uniqueness check is performed after the exchange is completed.

[0137] Similarly, for mutation operations, when the mutation point is encoded in the device layout information, a uniqueness check is also required after the mutation to ensure that the encoded information is correctly expressed.

[0138] S4: Based on the approximate solution set of the Pareto optimal frontier, the equipment layout information and AGV speed information of the optimal solution that makes the objective function meet the constraints are obtained. According to the equipment layout information and AGV speed information, the optimization of the layout and energy consumption of the intelligent production unit is completed.

[0139] Example 2

[0140] This embodiment 2 is based on embodiment 1 and further illustrates the fast non-dominated sorting and selection operations in embodiment 1.

[0141] Fast non-dominated sort:

[0142] In the IMCLLP problem, production system performance indicators such as production capacity and energy consumption are difficult to express with closed-form mathematical formulas. Therefore, a simulation model will be used to obtain them as the fitness value of the individual. Due to the existence of multiple conflicting fitness values, it is impossible to directly compare and select them in the genetic algorithm. NSGA-Ⅱ introduces the non-dominated sorting rule to solve this problem. The non-dominated sorting rule means that a non-dominated solution set is constructed based on the dominance relationship between solutions, and other solutions are ranked, and individuals with higher ranks are selected to enter the next generation. However, when there are too many individuals of a certain level, it is not easy to obtain elite individuals by random selection, so the search time is long. Therefore, in order to speed up the selection process of non-dominated sorting, this embodiment 1 introduces the clustering distance as the elite individual screening rule, which is embedded in the non-dominated sorting rule to shorten the algorithm search time.

[0143] The dominance relationship is used to compare the advantages and disadvantages of two solutions in multi-objective optimization. When a solution A is said to dominate another solution B, denoted as A < B, it means that A is not inferior to B in all objectives and is better than B in at least one objective, that is, it satisfies two conditions: 1) For all sub-objectives, A is not worse than B, that is, f k (A)≤f k (B)(k=1,2,…,r); 2) There is at least one sub-goal that makes A better than B, that is , so that f l (A) <f l (B); where r is the number of sub-goals.

[0144] At this time, A is called non-dominated, or non-inferior or dominant, and B is dominated. If there are multiple solutions that do not dominate each other, the comprehensive performance of these solutions is relatively high-quality, forming a non-dominated set. Before the selection operation, it is necessary to construct a non-dominated set and sort the other individuals outside the set. The specific process includes two parts:

[0145] The first part consists of finding the non-dominated set, i.e., individuals of level 1, as follows:

[0146] For each individual in the population from i 1 to the population size and each other individual in the population from j i+1 to the population size, set the i-th individual in the two individuals currently compared to p and the j-th individual to q; if p dominates q, add q to the domination set of p, the domination set of p represents the individuals dominated by p, and the number of times q is dominated increases by 1; if q dominates p, add p to the domination set of q, the domination set of q represents the individuals dominated by q, and the number of times p is dominated increases by 1; if the number of times the i-th individual is dominated is 0, that is, no other individual dominates it, then the individual is classified as the first level and added to the first set of the non-dominated set F, and the level of the individual is set to 1;

[0147] The second part consists of classifying individuals other than those in class 1 as follows:

[0148] Initialize an empty set Q to temporarily store individuals of new levels; for each individual in the non-dominated set F whose current level is r=1 and for all other individuals dominated by this individual, reduce the number of times these individuals are dominated. If the number of times an individual is dominated becomes 0, that is, it is not dominated by any other individual, then set the level of this individual to r+1 and add it to Q; if Q is empty, it means that all individuals have been assigned levels, and exit the loop; otherwise, add the individuals in Q as individuals of the new level r+1 to the next set of the non-dominated set F, increase the value of r, and continue to process individuals of the next level.

[0149] The pseudo code of the above process is as follows:

[0150]

[0151]

[0152]

[0153] In order to speed up the process of non-dominated sorting and ensure the distribution and diversity of the population, the individual aggregation distance is used as a supplementary rule for sorting between individuals, and individuals with large aggregation distances are retained first. The calculation formula of the aggregation distance is as follows:

[0154]

[0155] The aggregation distance formula indicates that the aggregation distance of the i-th individual is equal to the sum of the differences between the i-1-th individual and the i+1-th individual in all objective function values. After adding the aggregation distance rule, the process of non-dominated sorting includes: first construct a non-dominated set and calculate the aggregation distance of each individual, then sort them in ascending order according to the level divided by the non-dominated set, and sort the individuals in the same level in descending order according to the size of the aggregation distance.

[0156] Select an action:

[0157] Fast non-dominated sorting assigns all individuals in the parent and child populations to corresponding levels of dominance and non-domination, and then selects a certain proportion of individuals to enter the next generation population. First, select individuals with a high non-dominated level. When all individuals of a certain level cannot enter the next generation at the same time, select individuals with a large aggregation distance to enter the next generation. The selection operation is as follows: Figure 6 As shown in the figure, P t represents the parent generation, Q t F represents the unselected offspring after crossover mutation of the parent generation. 1 , F 2 , F 3 , ... represent various levels, P t+1 Indicates a new population.

[0158] The other steps or technical details of this embodiment 2 are the same as those of embodiment 1 and will not be repeated here.

[0159] Example 3

[0160] This embodiment 3 is based on embodiment 1 and embodiment 2, and applies a smart production unit layout and energy consumption optimization method in embodiment 1 and embodiment 2 to provide a smart production unit layout and energy consumption optimization system, which includes:

[0161] A digital model building unit is used to build a capacity-energy consumption mathematical model and its objective function with maximizing the capacity of the intelligent unit and minimizing the energy consumption of the AGV as the joint optimization goal;

[0162] A simulation unit is used to establish a simulation model of an intelligent production unit with AGV circular transportation using modeling software based on a capacity-energy consumption mathematical model;

[0163] The optimal solution solving unit is used to randomly generate the initial population using the preset improved NSGA-Ⅱ genetic algorithm and perform population iteration. During the iteration process, the intelligent production unit simulation model is called to simulate to obtain the fitness value of the individuals in the population. After the iteration is completed, the Pareto optimal solution set for constructing the capacity-energy consumption mathematical model is output;

[0164] The layout and energy consumption optimization unit is used to obtain the equipment layout information and AGV speed information of the optimal solution that makes the objective function meet the constraints based on the Pareto optimal solution set, and complete the optimization of the layout and energy consumption of the intelligent production unit according to the equipment layout information and AGV speed information.

[0165] The other steps of this embodiment 3 are the same as those of embodiments 1 and 2 and will not be repeated here.

[0166] In summary, the present invention establishes a multi-objective model by considering the equipment layout information and the energy consumption of AGV. Through Plant Simulation, the randomness of the real production scene is effectively retained, and the equipment layout information and AGV speed information are solved by proposing an improved NSGA-II genetic algorithm to find a balance point between the two conflicting goals of production capacity and energy saving, obtain the overall optimal solution, and reasonably set the speed of the AGV robot, the mutual influence of the equipment and the layout of the loading / unloading port of the production unit, further ensure the timely delivery of materials to stabilize the production efficiency of the intelligent unit, and reduce the energy consumption of the intelligent unit.

[0167] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the embodiments of the present invention. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the embodiments here. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the claims of the present invention.

Claims

1. A method for intelligent production unit layout and energy consumption optimization, characterized in that: The following steps are involved: Taking maximizing the capacity of the intelligent unit and minimizing the energy consumption of the AGV as the joint optimization goals, a capacity-energy consumption mathematical model and its objective function are constructed; According to the capacity-energy consumption mathematical model, a simulation model of an intelligent production unit with AGV circular transportation is established using modeling software; The preset improved NSGA-Ⅱ genetic algorithm is used to randomly generate the initial population and perform population iteration. During the iteration process, the intelligent production unit simulation model is called to simulate to obtain the fitness values ​​of individuals in the population. After the iteration is completed, the final non-dominated sorting result, i.e., the approximate solution set of the Pareto optimal frontier, is output; Based on the approximate solution set of the Pareto optimal frontier, the equipment layout information and AGV speed information of the optimal solution that makes the objective function meet the constraint conditions are obtained, and the optimization of the layout and energy consumption of the intelligent production unit is completed according to the equipment layout information and AGV speed information.

2. The method for intelligent production unit layout and energy consumption optimization according to claim 1, characterized in that: The objective function of the joint optimization objective is: maxΘ(X,v) minE(X,v) in, It means that the total number of workpieces num produced by the intelligent generation unit under the influence of variables X and v is divided by the simulation time t to obtain the production capacity Θ, E is the energy consumption of AGV, X is the equipment layout information, and v is the cruising speed of AGV; 0-1 decision variable X, that is, the equipment layout information X is: Among them, X is used to determine the relative position of the equipment and the loading and unloading ports. If the problem scale is n, X is an n×n dimensional 0-1 matrix, where x i,j If it is 1, it means that facility node i is in position j L j On, otherwise 0; The calculation process of energy consumption E is as follows: AND s =And am +E um +E a E a =P a t ij F t =m total a ac +F f F f =fm total g m total =m1+c1m2 l z ＝l ij -l ac -l de Where, E: energy consumption of AGV; n: problem scale, the number of locations where equipment can be arranged in the intelligent unit, including the loading port buffer and the unloading port buffer; E s : Total energy consumption of AGV for a single task; E am : Energy consumption during AGV acceleration phase; E um : Energy consumption of AGV during cruising; E a : Additional energy consumed by other energy-consuming components; P a : Power of energy-consuming components; h m : The total number of steps for workpiece m, p = 1, ..., m, q = 1, ..., h m ;t idle : AGV idle time; t ac1 : The acceleration time when the AGV cannot reach the cruising speed v when it moves from position i to position j; t ac2 : The time required to accelerate from 0 to cruising speed; t de1 : The braking time when the AGV cannot reach the cruising speed v when it moves from position i to position j; t de2 : The time required to decelerate from the cruising speed v to 0; t ij : The total time required for the AGV to move from position i to position j; c1: When the AGV loads the workpiece, c1 is 1, otherwise c1 is 0; c2: When the AGV cannot reach the cruising speed v during the process of moving from position i to position j, c2 = 1, otherwise c2 = 0; l z : moving distance during cruising; l ij : The moving distance of AGV from position i to position j; l ac : Moving distance of AGV when accelerating; l de : Moving distance when AGV decelerates; η: Energy efficiency factor; F t : driving force; F f : friction force; f: rolling friction coefficient; g: gravitational acceleration; a ac : acceleration; a de : Braking acceleration, v ac :v ac <v, the robot cannot reach the maximum speed when the cruising speed v is reached when it moves from position i to position j.

3. The method for optimizing the layout and energy consumption of intelligent production units according to claim 2, characterized in that: The constraints of the objective function include: The above two expressions together express that only one device or facility node can be arranged at each location, and each device or facility node can only be arranged at one location; in min ≤v≤v max Among them, v min : The minimum AGV moving speed that can be set; v max : The maximum AGV moving speed that can be set.

4. The method for intelligent production unit layout and energy consumption optimization according to claim 1, characterized in that: Before the improved NSGA-II genetic algorithm performs population iteration, it is necessary to design coding rules, which are as follows: The equipment layout information is encoded on the left side of the gene, and the length of the encoding depends on the problem scale n. That is, if there are n positions on the periphery of the circular track of the simulation model, the equipment layout information encoding is represented as an n-dimensional array, which can accommodate the loading port (P), the unloading port (D) and n-2 workstations. The set of workstations is {S1, S2, ..., S n-2 ); The AGV speed information is encoded on the right side of the gene. At present, the AGV speed in the workshop is generally set at 0-3m / s. To ensure the accuracy of the experiment, three decimal places are retained. Therefore, 14-bit binary encoding is used. After conversion to decimal, the value range is [0, 8192]. Dividing by 1000, the moving speed value of the AGV is obtained. The equipment layout information encoding and AGV speed information encoding are merged to form an individual encoding. The encoding length depends on the problem scale n. In addition, the merged encoding cautiously adopts crossover and mutation operations to ensure the correctness of the information expressed by the encoding.

5. The method for intelligent production unit layout and energy consumption optimization according to claim 1, characterized in that: The iterative process of the improved NSGA-II genetic algorithm includes: Population initialization, randomly generate the initial population; Performing a crossover operation on the gene codes of the individuals of the initialized population in sections; After the individuals after crossover are subjected to single-point mutation operation and correctness test, After the correctness test is passed, the child and parent populations are merged; Traverse the merged population, identify isomorphic individuals, mark one of each isomorphic individual, and obtain the fitness value of one of each isomorphic individual and the fitness values ​​of other individuals without isomorphic individuals through the simulation model, share the fitness value of one of each isomorphic individual with other isomorphic individuals, complete the task of removing isomorphic individuals, reduce the number of calls to the simulation model, and compress the solution time; The fast non-dominated sorting rule is used to construct a non-dominated set, and the other individuals outside the set are sorted to obtain the non-dominated sorting result; Select elite individuals from the non-dominated sorting results to enter the next generation population; Determine whether the iteration termination condition is met. If the iteration termination condition is met, output the final non-dominated sorting result, otherwise continue to iterate and execute the steps starting from the crossover operation.

6. A method for optimizing the layout and energy consumption of intelligent production units according to any one of claims 4 or 5, characterized in that: The crossover operation is performed using a partial matching crossover operator, and on the basis of the traditional two-point crossover partial matching operator, an intersection is added to the AGV speed information encoding part, that is, the first and second intersections are randomly generated in the equipment layout information encoding part, and the third intersection is randomly generated in the AGV speed information encoding part, and the codes between the first and second intersections of different individuals and the codes after the third intersection are exchanged. After the exchange is completed, a uniqueness check is performed to ensure that the encoded information is correctly expressed.

7. A method for intelligent production unit layout and energy consumption optimization according to claim 4 or 5, characterized in that: For the single-point mutation operation, when the mutation point is encoded in the device layout information, a uniqueness check needs to be performed after the mutation to ensure that the encoded information is correctly expressed.

8. The method for intelligent production unit layout and energy consumption optimization according to claim 5, characterized in that: The specific process of constructing a non-dominated set by using a fast non-dominated sorting rule and sorting other individuals outside the set includes two parts: The first part consists of finding the non-dominated set, i.e., individuals of level 1, as follows: For each individual in the population from i 1 to the population size and each other individual in the population from j i+1 to the population size, set the i-th individual in the two individuals currently compared to p and the j-th individual to q; If p dominates q, add q to p's dominance set, which represents the individuals dominated by p, and the number of times q is dominated increases by 1; If q dominates p, add p to q's domination set, where q's domination set represents the individuals dominated by q, and the number of times p is dominated increases by 1; if the number of times the i-th individual is dominated is 0, that is, no other individual dominates it, then the individual is classified as level 1 and added to the first set of the non-dominated set F, and the level of the individual is set to 1; The second part consists of classifying individuals other than those in class 1 as follows: Initialize an empty set Q to temporarily store individuals of new ranks; for each individual in the non-dominated set F whose current rank is r=1 and for all other individuals dominated by this individual, reduce the number of times these individuals are dominated; if the number of times an individual is dominated becomes 0, that is, it is now not dominated by any other individual, set the rank of this individual to r+1 and add it to Q; If Q is empty, it means that all individuals have been assigned levels and the loop exits. Otherwise, the individuals in Q are added to the next set of the non-dominated set F as individuals of the new level r+1, and the value of r is increased to continue processing the individuals of the next level.

9. The method for intelligent production unit layout and energy consumption optimization according to claim 8, characterized in that: In order to speed up the process of non-dominated sorting and ensure the distribution and diversity of the population, the individual aggregation distance is used as a supplementary rule for sorting between individuals, and individuals with large aggregation distances are retained first. The calculation formula of the aggregation distance is as follows: The aggregation distance formula indicates that the aggregation distance of the i-th individual is equal to the sum of the differences between the i-1-th individual and the i+1-th individual in all objective function values. After adding the aggregation distance rule, the process of non-dominated sorting includes: first construct a non-dominated set and calculate the aggregation distance of each individual, then sort them in ascending order according to the level divided by the non-dominated set, and sort the individuals in the same level in descending order according to the size of the aggregation distance.

10. An intelligent production unit layout and energy consumption optimization system, using an intelligent production unit layout and energy consumption optimization method such as those described in claims 1-9, characterized in that: The system comprises: A digital model building unit is used to build a capacity-energy consumption mathematical model and its objective function by taking maximizing the capacity of the intelligent unit and minimizing the energy consumption of the AGV as the joint optimization goal; A simulation unit, used to establish a simulation model of an intelligent production unit with AGV circular transportation using modeling software according to the capacity-energy consumption mathematical model; The optimal solution solving unit is used to randomly generate an initial population using a preset improved NSGA-Ⅱ genetic algorithm and perform population iteration. During the iteration process, the intelligent production unit simulation model is called to perform simulation to obtain the fitness value of individuals in the population. After the iteration is completed, the Pareto optimal solution set for constructing the production capacity-energy consumption mathematical model is output; The layout and energy consumption optimization unit is used to obtain the equipment layout information and AGV speed information of the optimal solution that makes the objective function meet the constraint conditions based on the Pareto optimal solution set, and complete the optimization of the layout and energy consumption of the intelligent production unit according to the equipment layout information and AGV speed information.

Images