A method for cross-scheduling automobile orders

Abstract

The present invention relates to the technical field of automobile manufacturing, and specifically relates to an automobile order cross-scheduling method, which includes establishing a mathematical model, calculating the caking length relaxation interval based on the order data characteristics, and determining the upper limit of the switching times; determining the objective function weights according to the priority levels and calculating the initial objective function value; updating the order sequence through the domain search strategy of the caking phenomenon; and performing iterative solution through an improved genetic algorithm. By deeply understanding the actual automobile production scenario, this method establishes a unique mathematical model, which better reflects the constraint requirements for the order sequence in the actual automobile production scenario. Through the overall crossover of attribute caking, that is, randomly inserting a continuous plurality of orders with the same attribute into the sequence as a whole, the solution speed and solution quality are greatly improved. The improvement of the crossover strategy for specific attribute caking in this method can reduce the solution speed and achieve a higher constraint satisfaction rate.

Description

Technical Field

[0001] The present invention relates to the technical field of automobile manufacturing, and particularly to a method for cross-scheduling automobile orders. Background Art

[0002] With the continuous development of the manufacturing industry and the increasing diversification of user needs, the production mode of manufacturing enterprises has changed from stock production to order-based production centered on the market. The uncertainty of demand poses high requirements on the planning and control of enterprises. Therefore, manufacturing enterprises are committed to obtaining a shorter product manufacturing cycle and lower production costs under various constraints through more reasonable and effective order scheduling strategies. In the scheduling problem, there are usually two types of constraints: hard constraints and soft constraints. Among them, hard constraints must be fully satisfied to make the generated solution a feasible solution, while soft constraints only need to be satisfied as much as possible. The order scheduling problem is an NP-complete problem, and it is impossible to obtain an optimal solution within a limited polynomial time.

[0003] For the automobile order scheduling problem, the commonly used methods at present are mainly divided into exact methods, heuristic methods, and intelligent optimization methods. Among them, exact methods include mixed integer programming methods and dynamic programming methods. For small-scale problems, exact methods can obtain theoretical optimal solutions, but the time complexity and space complexity of such methods are relatively high, and it is difficult to use them to solve practical problems; heuristic methods include various priority rules, greedy methods and their variants, which are methods that summarize past experience and infer sub-optimal solutions to problems. Although heuristic methods are superior to exact methods in terms of method complexity, it is difficult to guarantee the convergence of the methods; intelligent optimization methods mainly include Genetic Algorithm (GA), Simulated Annealing Algorithm (SA), and Differential Evolution Algorithm (DEA). Intelligent optimization methods are more suitable for complex problems under large-scale and multi-constraint conditions, and find the optimal scheduling plan by simulating natural selection and mechanisms. At the same time, intelligent optimization methods do not depend on the gradient of the fitness function, making the solution plan more applicable.

[0004] However, the above methods have problems such as low constraint satisfaction rate and long scheduling time. Therefore, there is an urgent need for a method for cross-scheduling automobile orders to solve this problem. Summary of the Invention

[0005] The purpose of the present invention is to provide a method for cross-scheduling automobile orders, aiming to solve the problems of low constraint satisfaction rate and long scheduling time existing in the existing automobile order production scheduling methods.

[0006] To achieve the above purpose, the present invention provides a method for cross-scheduling automobile orders, including the following steps:

[0007] Establish a mathematical model, calculate the caking length relaxation interval based on the characteristics of order data, and determine the upper limit of the number of switches;

[0008] Determine the objective function weight according to the priority level and calculate the initial objective function value;

[0009] Update the order sequence through the domain search strategy of the caking phenomenon;

[0010] Perform iterative solution through an improved genetic algorithm.

[0011] Among them, the objective function includes the number of configuration classification switches, the number of color classification switches, and the square difference of the caking length of the roof overprint.

[0012] Among them, the number of configuration classification switches is the number of changes from one configuration classification to another in the final production scheduling result.

[0013] Among them, the number of color classification switches is the number of changes in the color attributes of two adjacent orders in actual production.

[0014] Among them, for the square difference of the caking length of the roof overprint, for each roof overprint, it is necessary to divide according to the specified caking size, and then the cakings corresponding to different roof overprints are arranged in a cross pattern.

[0015] A cross-production scheduling method for automobile orders of the present invention establishes a mathematical model, calculates the caking length relaxation interval based on the characteristics of order data, and determines the upper limit of the number of switches; determines the objective function weight according to the priority level and calculates the initial objective function value; updates the order sequence through the domain search strategy of the caking phenomenon; performs iterative solution through an improved genetic algorithm. This method, through a deep understanding of the actual automobile production scenario, establishes a unique mathematical model, which better reflects the constraint requirements for the order sequence in the actual automobile production scenario. Through the overall cross of attribute cakings, that is, randomly inserting a continuous number of orders with the same attribute into the sequence as a whole, the solution speed and solution quality are greatly improved. When applied, it can achieve the purpose of cost reduction and efficiency increase for automobile manufacturing enterprises. The improvement of the cross strategy for specific attribute cakings can reduce the solution speed and achieve a higher constraint satisfaction rate, solving the problems of low constraint satisfaction rate and long production scheduling time existing in the existing automobile order production scheduling methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art.

[0017] Figure 1It is a result comparison chart for scheduling 250 orders. Among them, Curve 1 is the result achieved by this method, and Curves 2 and 3 are the results of two existing methods respectively.

[0018] Figure 2 It is a result comparison chart for scheduling 800 orders. Among them, Curve 1 is the result achieved by this method, and Curves 2 and 3 are the results of two existing methods respectively.

[0019] Figure 3 It is a result comparison chart for scheduling 1600 orders. Among them, Curve 1 is the result achieved by this method, and Curves 2 and 3 are the results of two existing methods respectively.

[0020] Figure 4 It is a result comparison chart for scheduling 2400 orders. Among them, Curve 1 is the result achieved by this method, and Curves 2 and 3 are the results of two existing methods respectively.

[0021] Figure 5 It is a schematic flow chart of a method for cross-scheduling automobile orders according to the present invention.

[0022] Figure 6 It is a schematic diagram of the steps of a method for cross-scheduling automobile orders according to the present invention. Detailed implementation manners

[0023] The embodiments of the present invention will be described in detail below. Examples of the embodiments are shown in the drawings, where the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the drawings are exemplary and are intended to explain the present invention and should not be construed as limiting the present invention.

[0024] Please refer to Figures 1 to 6 , the present invention provides a method for cross-scheduling automobile orders, including the following steps:

[0025] S1 Establish a mathematical model, calculate the caking length relaxation interval based on the order data characteristics, and determine the upper limit of the switching times;

[0026] Specifically, the mathematical model is established as follows:

[0027]

[0028] Among them,

[0029]

[0030] D k represents the caking length of the kth caking car roof overprinting; S represents the specified caking size of each caking car roof overprinting, and the variable x in the 0-1 relaxation interval i,i+1Indicates whether the configuration classification of two adjacent orders has switched, and the slack interval variable y from 0 to 1 i,i+1 Indicates whether the color classification of two adjacent orders has switched.

[0031] Constraint conditions:

[0032]

[0033] D k ≤S max (6)

[0034] D k ≥S min (7)

[0035] T 1 =0 (8)

[0036] Among them,

[0037]

[0038] N max represents the maximum number of switches in the configuration classification; M max represents the maximum number of switches in the color classification; S max represents the upper limit of the number of overcoats for each caked car roof; S min represents the lower limit of the number of overcoats for each caked car roof.

[0039] Equation (1) represents minimizing the number of configuration classification switches; Equation (2) represents minimizing the number of color classification switches; Equation (3) represents minimizing the sum of the squared differences in the lengths of the caked car roof overcoats; Equations (4) and (5) respectively represent that the number of switches in the configuration classification and the color classification cannot exceed the specified values; Equations (6) and (7) represent that the number of overcoats for each caked car roof can only be within the specified interval; Equation (8) represents that the overcoat type of the first caked car roof is two-color.

[0040] S2 determines the weights of the objective function according to the priority level and calculates the initial value of the objective function;

[0041] Specifically, the priority level is caked length constraint > configuration switch number constraint > color switch number constraint, and three indicators of minimizing the number of configuration classification switches, minimizing the number of color classification switches, and minimizing the sum of the squared differences in the lengths of the caked car roof overcoats are used as the objective function.

[0042] The number of configuration classification switches refers to the number of changes from one configuration classification to another in the final production scheduling result. The number of color classification switches refers to the number of changes in the color attributes of adjacent two orders in actual production. Too many configuration switches will significantly reduce the production speed, and each color switch will involve a series of tasks such as pigment cleaning, equipment replacement, and pigment treatment, which will generate relatively high production costs and also affect the production speed. The cross-scheduling of roof color matching means that for each roof color matching, it needs to be divided according to the specified agglomerate size, and then the agglomerates corresponding to different roof color matchings are arranged crosswise. Too large a number of agglomerates will lead to line blockage during the production process. Once line blockage occurs, it will have a huge impact on the production smoothness.

[0043] S3 updates the order sequence through the domain search strategy of the agglomeration phenomenon;

[0044] S4 performs iterative solution through the improved genetic algorithm.

[0045] In this method, in the actual production process, the factors that have a greater impact on productivity are quantified into three objectives, namely the number of color switches, the number of configuration switches, and the sum of the differences between the lengths of each agglomerate and the target length. And a mathematical model is established in this way through linear weighting; the method is improved through a specific domain search strategy, that is, for the agglomerates with too large a difference between the agglomerate length and the target length, they are merged and reorganized or split and reorganized, so as to ensure that in the final result, the length of each agglomerate is relatively close to the target length. This method overcomes the problems of poor convergence effect and low constraint satisfaction rate of the traditional method, and better meets the actual production needs;

[0046] By quantifying the factors that have a greater impact on production speed and production cost during the upgraded production process, that is, the two switching times and the agglomerate length, the selected objectives are more in line with the enterprise's goal of reducing costs and increasing efficiency; effectively avoiding the occurrence of line blockage affecting production speed caused by multiple equipment replacements and too large a number of agglomerates during the production process, improving production efficiency, while reducing color switches, avoiding waste of pigments, saving costs, achieving the enterprise's goal of reducing costs and increasing efficiency, and at the same time avoiding environmental health problems caused by improper treatment of pigments.

[0047] To better understand this technical solution, the following embodiments are provided for further illustration:

[0048] Taking the actual order data provided by a certain automobile manufacturing enterprise as an example, using 250 orders, 800 orders, 1600 orders, and 2400 orders as test data respectively, the method was tested on a workstation equipped with an Intel(R) Core(TM) i7-14700k processor, 32G of memory, and Intel(R) UHD Graphics 770 in the PTYHON 3.8 environment.

[0049] First, the hard constraints of the order (such as a certain order must be arranged in the second position) are preferentially satisfied;

[0050] Secondly, a model is constructed with some constraints of higher priority (such as caking quantity constraint, caking length constraint, switching times constraint);

[0051] Then, it is solved by the method proposed in the present invention.

[0052] Figures 1 - 4 The figure below is a comparison chart of the method results of the present invention with the traditional method and the differential evolution method under four different orders (the three curves are the method (GA), the improved method (IGA), and the differential evolution method (EA)):

[0053] As can be seen from the above Figures 1 - 4 It can be seen that compared with some existing methods, the present method has a faster convergence speed, and the final fitness function value is also smaller, that is, the present invention realizes the optimization in both time and solution quality. In addition to the speed and quality of the solution, the present invention also has good stability and can obtain a better solution more quickly for order data of different scales.

[0054] Specifically, for the scenario of automobile production and manufacturing, due to the caking length constraint, most existing methods cannot guarantee a high constraint satisfaction rate. For example, for the traditional method, when solving the same mathematical model, the constraint satisfaction rate can only reach 73%-85%, while the present invention can reach 83% to 93%. In terms of the convergence speed, through Figures 1 - 4 It can be seen that under different scales, the convergence speed of this method is nearly 50% faster than that of the existing methods.

[0055] The above-disclosed is only a preferred embodiment of a method for cross-scheduling automobile orders of the present application, and it cannot be used to limit the scope of rights of the present application. Those of ordinary skill in the art can understand all or part of the processes of implementing the above embodiments, and the equivalent changes made according to the claims of the present application still fall within the scope covered by the present application.

Claims

1. A method for cross-scheduling of automobile orders, characterized in that: The following steps are involved: Establish a mathematical model and calculate the relaxation interval of agglomeration length based on the characteristics of order data to determine the upper limit of the number of switching times; Determine the objective function weight according to the priority level and calculate the initial objective function value; Update order sequence through the domain search strategy of agglomeration phenomenon; Iterative solution is performed through improved genetic algorithm; A mathematical model is established, and the relaxation interval of the agglomeration length is calculated based on the order data characteristics to determine the upper limit of the switching times. Specifically, the mathematical model is established as follows: in, D k represents the length of the kth block roof color; S represents the specified block size of each block roof color, and the variable x in the 0-1 relaxation interval i,i+1 Indicates whether the configuration classification of two adjacent orders is switched, a 0-1 relaxation interval variable y i,i+1 Indicates whether the color classification of two adjacent orders is switched; Constraints: D k ≤S max (6) D k ≥S min (7) T1＝0 (8) in, N max Indicates the maximum number of switching times for the configuration category; M max Indicates the maximum number of switching times of color classification; S max Indicates the upper limit of the number of colors for each agglomerated roof; S min Indicates the lower limit of the number of colors for each agglomerated roof; Formula (1) represents the minimization of the number of configuration classification switching; Formula (2) represents the minimization of the number of color classification switching; Formula (3) represents the minimization of the sum of the square differences of the lengths of the roof color blocks; Formulas (4) and (5) respectively represent that the number of configuration classification and color classification switching times cannot exceed the specified value; Formulas (6) and (7) represent that the number of roof colors of each block can only be within the specified interval; Formula (8) represents that the roof color type of the first block is two-color; Determine the objective function weight according to the priority and calculate the initial objective function value. Specifically, the priority is block length constraint > configuration switching number constraint > color switching number constraint, and minimize the number of configuration classification switching, minimize the number of color classification switching, and minimize the roof color block length square difference and three indicators as the objective function; The square difference of the length of the roof color block is: For each roof color, it is necessary to divide it according to the specified block size, and then the blocks corresponding to different roof colors are arranged crosswise; The iterative solution is performed through an improved genetic algorithm, including: The factors that have a greater impact on productivity are scalared into three targets, namely the number of color switching times, the number of configuration switching times, and the sum of the differences between the lengths of each block and the target length. A mathematical model is established based on this by means of linear weighting. The method is improved through a specific field search strategy, that is, for blocks with a large difference between their lengths and the target length, they are merged or split and reorganized to ensure that in the final result, the length of each block is relatively close to the target length.

2. The method for cross-scheduling of automobile orders according to claim 1, characterized in that: The number of configuration classification switching times is the number of times a configuration classification is changed from one configuration classification to another in the final production scheduling result.

3. The method for cross-scheduling of automobile orders according to claim 1, characterized in that: The number of color classification switching times is the number of times the color attributes of two adjacent orders change in actual production.

Images