A multi-control parameter optimization method for a merging end of an automatic logistics sorting system

Abstract

The application particularly relates to a multi-control parameter optimization method for a merging end of an automatic logistics sorting system, and belongs to the technical field of automatic logistics sorting systems.The application constructs a multi-control parameter optimization problem model for the merging end of the automatic logistics sorting system according to actual operation conditions of the automatic logistics sorting system, comprehensively considers the influence of control parameters such as a virtual window control mode, a collection belt conveyor operation speed, a virtual window length and the number of simultaneously opened injection belt conveyors on the merging efficiency of the automatic logistics sorting system, proposes a simulation optimization framework for solving the model, designs a multi-objective adaptive parallel wolf swarm algorithm to solve the model, and obtains an optimal solution set, thereby providing multiple optimization schemes for decision makers.The simulation optimization model is established according to actual operation conditions of the merging end of a general automatic logistics sorting system, the method is in line with the actual production, can effectively shorten the average waiting time of goods merging, reduce the system energy consumption, and improve the operation efficiency of the automatic logistics sorting system.

Description

Technical Field

[0001] This invention relates to a method for optimizing multiple control parameters at the merging end of an automated logistics sorting system, belonging to the technical field of automated logistics sorting systems. Background Technology

[0002] Automated sorting systems, as a fast and accurate picking tool, are increasingly widely used, gradually replacing manual picking operations. Currently, my country has extensively adopted automated sorting technologies in picking-intensive distribution centers such as those in the pharmaceutical, tobacco, military, and express delivery industries. These include automated storage and retrieval systems, rotating shelf automated picking systems, modular automated sorting systems, and parallel automated sorting systems with pre-sorting capabilities. Compared to general manual picking systems, automated sorting systems can continuously sort goods and offer significant advantages such as more sorting points, lower error rates, and unmanned operation. With the rapid development of the commodity economy, enterprises are constantly increasing their performance requirements for automated sorting systems. Therefore, exploring how to better control automated sorting systems, optimize their control parameters, and ensure optimal performance indicators such as operational efficiency and system energy consumption during sorting operations has become a research hotspot in the field of automated sorting systems.

[0003] The composition of automated logistics sorting systems is quite complex, consisting of various types of transmission equipment, information acquisition systems, automated sorting management systems, and control systems. Sorting operations can be broadly divided into four processes: merging, sorting information input, sorting, and loading. The efficiency of each process affects the overall operational efficiency of the entire automated logistics sorting system. Currently, many scholars have studied information system management in the sorting information input stage, control strategies in the sorting stage, and management strategies in the loading stage, but research on the impact of the merging stage on the efficiency of goods merging is lacking. Therefore, addressing the problems of long waiting times and high energy consumption during the merging process of automated logistics sorting systems, this invention proposes a multi-control parameter optimization method for the merging end of the automated logistics sorting system. This method comprehensively considers the impact of control parameters such as virtual window control mode, collection belt conveyor speed, virtual window length, and the number of simultaneously open injection belt conveyors on the merging efficiency of the automated logistics sorting system, thereby comprehensively optimizing the operational efficiency of the merging end of the automated logistics sorting system from multiple perspectives. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing automated logistics sorting systems by proposing a multi-control parameter optimization method for the merging end of an automated logistics sorting system, in order to meet the needs of setting control parameters and multi-objective optimization at the merging end of the automated logistics sorting system under different cargo flow conditions.

[0005] The technical solution of this invention is as follows: First, a multi-control parameter optimization problem model for the merging end of an automated logistics sorting system is established; then, a simulation optimization framework for solving this problem model is constructed; finally, a multi-objective adaptive parallel wolf pack algorithm is designed to solve the problem.

[0006] The model for optimizing multiple control parameters at the merging end of an automated logistics sorting system is established based on the impact of control parameters such as virtual window technology and control method, collection belt conveyor speed, virtual window length, and the number of simultaneously open injection belt conveyors on the merging efficiency of the automated logistics sorting system. The problem model is described as follows:

[0007] Based on the principle of minimizing the average waiting time of goods injected into the collection belt conveyor, the first optimization objective is established as follows:

[0008]

[0009] In the formula, F1 represents the average waiting time for each item injected into the collection conveyor; i∈{1,2,…,l} represents the number of the injection conveyor, and l represents the maximum number of the injection conveyor; k i ∈{1,2,…,x i} represents the item number injected onto belt conveyor i, x i Indicates the maximum item number injected onto belt conveyor i; Indicates the kth injection point on the belt conveyor i. i Waiting time for each item to be injected into the collection belt conveyor;

[0010] Based on the principle of minimizing the total energy consumption of sorting and conveying equipment in an automated logistics sorting system, the second optimization objective is established as follows:

[0011]

[0012] In the formula, F2 represents the total energy consumption of the sorting and conveying equipment in the automated logistics sorting system; a∈{1,2,…,m} represents the sorting equipment number in the automated logistics sorting system, and m represents the maximum sorting equipment number in the automated logistics sorting system; E a Let E represent the energy consumption of sorting equipment a in the automated logistics sorting system; b∈{1,2,…,n} represents the number of conveyor equipment in the automated logistics sorting system, and n represents the maximum number of conveyor equipment in the automated logistics sorting system; E b This indicates the energy consumption of conveyor equipment b in the automated logistics sorting system;

[0013] The following constraints also need to be met:

[0014] In the formula, Indicates the kth injection point on the belt conveyor i.i The length of each item; Indicates the kth injection point on the belt conveyor i. i The width of each item; Indicates the kth injection point on the belt conveyor i. i The height of each item; Indicates the kth injection point on the belt conveyor i. i The maximum permissible volume of each item;

[0015] In the formula, Indicates the kth injection point on the belt conveyor i. i The weight of each item; G max Indicates the maximum permissible weight of the goods;

[0016] In the formula, This indicates the number of goods that can be served simultaneously by belt conveyor i.

[0017] In the formula, T represents the arrival time of the first item that needs to be sorted in the system. start Indicates the start time set for the simulation model;

[0018] In the formula, Indicates the arrival time of the last item to be sorted in the system; T end Indicates the end time set for the simulation model;

[0019] In the formula, Indicates cargo k i The actual injected virtual window; Indicates cargo k i The actual virtual window used for the application;

[0020] N W-e ≠0; where N W-e Indicates the number of empty virtual windows in the system;

[0021] In the formula, This indicates the number of goods that each virtual window can serve simultaneously.

[0022] L Wmin ≤L W ≤L Wmax In the formula, L W L represents the length of the virtual window W; Wmin Indicates the minimum allowed length of the virtual window W; L Wmax Indicates the maximum allowed length of the virtual window W;

[0023] M W ∈{1,2,…,h}; where M W ∈{1,2,…,h} represents the set of control methods for virtual window W, and h represents the last element in the set of control methods for virtual window W;

[0024] N min ≤N i ≤N max In the formula, N i N represents the number of injection belt conveyors that are open simultaneously. min Indicates the minimum number of injection belt conveyors allowed to be open simultaneously; N max Indicates the maximum number of injection belt conveyors that can be opened simultaneously;

[0025] v bmin ≤v b ≤v bmax In the formula, v b This indicates the operating speed of the conveyor equipment in an automated logistics sorting system; v bmin This indicates the minimum permissible operating speed of conveyor equipment in an automated logistics sorting system; v bmax This indicates the maximum permissible operating speed of the conveyor equipment in an automated logistics sorting system;

[0026] The simulation optimization framework for solving this problem model is specifically described as follows:

[0027] The framework consists of three modules: initialization, simulation optimization, and data processing. The initialization module initializes the parameters of the conveying equipment, control parameters, cargo information, and algorithm parameters of the model and generates an initial population. The initial population is then fed back to the algorithm model in the simulation optimization module, which controls the simulation model to perform optimization iterations. Finally, the data processing module statistically analyzes and evaluates the data obtained from the simulation optimization module. If the algorithm termination condition is met, an optimized solution set is output, and the optimized solution set is decoded to generate an optimized scheme for decision-makers to choose from. If the algorithm termination condition is not met, the population encoding is updated and fed back to the simulation optimization module to continue optimization iterations.

[0028] The multi-objective adaptive parallel wolf pack solution algorithm is specifically described as follows:

[0029] Step 1: Initialize the population;

[0030] Step 2: Combine random methods and reverse learning strategies to generate the initial population and the reverse population, calculate the target fitness value of each initial individual and the reverse individual, and evaluate the target fitness value to determine the initial population;

[0031] Step 3: Sort the initial population according to its target fitness value using Pareto non-dominated sorting, calculate the crowding degree of individuals of the same fitness level, eliminate redundant individuals, and temporarily store non-dominated individuals in the external space Ψ1.

[0032] Step 4: Implement the wolf pack classification mechanism for non-dominant individuals;

[0033] Step 5: Based on the wolf roaming mechanism and the alpha wolf summoning mechanism, the local neighborhood search mechanism and heuristic optimization strategy are executed in parallel on the classified wolf roaming individuals and wolf alpha individuals, and the target fitness values ​​of the roaming population and the summoned population are calculated.

[0034] Step 6: Based on the wolf siege mechanism, the global neighborhood search mechanism and heuristic optimization strategy are jointly executed on the individual scout wolves after the scout wolves have moved and the individual wolf wolves after the alpha wolf has summoned them. The target fitness value of the population after the siege is calculated and the population after the siege is temporarily stored in the external space Ψ2.

[0035] Step 7: A new population is formed by combining the populations in outer space Ψ1 and outer space Ψ2. The new population is sorted by Pareto non-dominated ordering based on the corresponding target fitness value. Crowding is calculated for individuals of the same level, and redundant individuals are eliminated.

[0036] Step 8: Determine if the termination condition is met. If it is, proceed to Step 9; otherwise, return to Step 4 to continue iterating.

[0037] Step 9: Output the Pareto front, and the algorithm terminates.

[0038] The beneficial effects of this invention are:

[0039] 1. When establishing a multi-control parameter optimization problem model for the merging end of an automated logistics sorting system, this invention comprehensively considers the impact of multiple control parameters on the merging efficiency of the automated logistics sorting system. Compared with optimization methods that consider a single influencing factor, it has better applicability.

[0040] 2. The simulation optimization framework for solving the problem model proposed in this invention can meet the merging requirements of various automated logistics sorting systems according to different input parameters, and has good robustness;

[0041] 3. The multi-control parameter optimization problem model of the merging end of the automated logistics sorting system established in this invention is based on the actual operating conditions of the merging end of a general automated logistics sorting system. It has good practicality and versatility and can be widely applied to related systems in various industries.

[0042] 4. The multi-objective adaptive parallel wolf pack algorithm proposed in this invention is adaptively adjusted and improved according to the characteristics of the model, and can obtain a better optimal solution set. Attached Figure Description

[0043] Figure 1 This is an overall flowchart of the present invention;

[0044] Figure 2 This is a simulation optimization framework diagram of the problem-solving model proposed in this invention;

[0045] Figure 3 The flowchart of the multi-objective adaptive parallel wolf pack algorithm proposed in this invention is shown below.

[0046] Figure 4 This is a schematic diagram illustrating the optimization problem of multiple control parameters at the merging end of the automated logistics sorting system proposed in this invention.

[0047] Figure 5 This is a schematic diagram of the initialization encoding of the multi-objective adaptive parallel wolf pack algorithm proposed in this invention. Detailed Implementation

[0048] Example 1: As Figure 1-5 As shown, a method for optimizing multiple control parameters at the merging end of an automated logistics sorting system is proposed. First, a problem model for optimizing multiple control parameters at the merging end of the automated logistics sorting system is established; then, a simulation optimization framework for solving this problem model is constructed; finally, a multi-objective adaptive parallel wolf pack algorithm is designed to solve the problem.

[0049] The model for optimizing multiple control parameters at the merging end of an automated logistics sorting system is established based on the impact of control parameters such as virtual window technology and control method, collection belt conveyor speed, virtual window length, and the number of simultaneously open injection belt conveyors on the merging efficiency of the automated logistics sorting system. The problem model is described as follows:

[0050] Based on the principle of minimizing the average waiting time of goods injected into the collection belt conveyor, the first optimization objective is established as follows:

[0051]

[0052] In the formula, F1 represents the average waiting time for each item injected into the collection conveyor; i∈{1,2,…,l} represents the number of the injection conveyor, and l represents the maximum number of the injection conveyor; k i ∈{1,2,…,x i} represents the item number injected onto belt conveyor i, x i Indicates the maximum item number injected onto belt conveyor i; Indicates the kth injection point on the belt conveyor i. i Waiting time for each item to be injected into the collection belt conveyor;

[0053] Based on the principle of minimizing the total energy consumption of sorting and conveying equipment in an automated logistics sorting system, the second optimization objective is established as follows:

[0054]

[0055] In the formula, F2 represents the total energy consumption of the sorting and conveying equipment in the automated logistics sorting system; a∈{1,2,…,m} represents the sorting equipment number in the automated logistics sorting system, and m represents the maximum sorting equipment number in the automated logistics sorting system; E a Let E represent the energy consumption of sorting equipment a in the automated logistics sorting system; b∈{1,2,…,n} represents the number of conveyor equipment in the automated logistics sorting system, and n represents the maximum number of conveyor equipment in the automated logistics sorting system; E b This indicates the energy consumption of conveyor equipment b in the automated logistics sorting system;

[0056] The following constraints also need to be met:

[0057] In the formula, Indicates the kth injection point on the belt conveyor i. i The length of each item; Indicates the kth injection point on the belt conveyor i. i The width of each item; Indicates the kth injection point on the belt conveyor i. i The height of each item; Indicates the kth injection point on the belt conveyor i. i The maximum permissible volume of goods; this formula ensures that the volume of goods meets the sorting and conveying requirements of automated logistics sorting systems;

[0058] In the formula, Indicates the kth injection point on the belt conveyor i. i The weight of each item; G max This indicates the maximum permissible weight of the goods; this formula ensures that the weight of the goods meets the sorting and conveying requirements of the automated logistics sorting system.

[0059] In the formula, This indicates the number of goods that injection belt conveyor i can serve simultaneously; this formula ensures that each injection belt conveyor can only serve one item at a time.

[0060] In the formula, T represents the arrival time of the first item that needs to be sorted in the system. start This indicates the start time set by the simulation model; this formula ensures that the start time set by the simulation model is less than the arrival time of the first item to be sorted in the system.

[0061] In the formula, Indicates the arrival time of the last item to be sorted in the system; T end This indicates the end time set by the simulation model; this formula ensures that the end time set by the simulation model is greater than the arrival time of the last item to be sorted in the system.

[0062] In the formula, Indicates cargo k i The actual injected virtual window; Indicates cargo k i The actual virtual window applied for; this method ensures that the virtual window into which the goods are actually injected matches the applied virtual window;

[0063] N W-e ≠0; where N W-e This indicates the number of empty virtual windows in the system; this formula ensures that goods can always obtain a window.

[0064] In the formula, This expression represents the number of goods that each virtual window can serve simultaneously; it ensures that each virtual window can be filled with at most one goods.

[0065] L Wmin ≤L W ≤L Wmax In the formula, L W L represents the length of the virtual window W; Wmin Indicates the minimum allowed length of the virtual window W; L Wmax This indicates the maximum allowable length of the virtual window W; this formula ensures that the length of the virtual window meets the actual design requirements of the automated logistics sorting system.

[0066] M W ∈{1,2,…,h}; where M W ∈{1,2,…,h} represents the set of control methods for virtual window W, and h represents the last element in the set of control methods for virtual window W; this expression ensures that the control method for virtual window is selected from the set of virtual window control methods set by the model;

[0067] N min ≤N i ≤N max In the formula, N i N represents the number of injection belt conveyors that are open simultaneously. min Indicates the minimum number of injection belt conveyors allowed to be open simultaneously; N max This indicates the maximum number of injection belt conveyors that can be open simultaneously; this formula ensures that the number of injection belt conveyors open simultaneously meets the actual design requirements of the automated logistics sorting system.

[0068] vbmin ≤v b ≤v bmax In the formula, v b This indicates the operating speed of the conveyor equipment in an automated logistics sorting system; v bmin This indicates the minimum permissible operating speed of conveyor equipment in an automated logistics sorting system; v bmax This represents the maximum permissible operating speed of the conveyor equipment in an automated logistics sorting system; this formula ensures that the operating speed of the conveyor equipment in the automated logistics sorting system meets the actual design requirements of the automated logistics sorting system;

[0069] The key control parameters considered in this invention are specifically described as follows:

[0070] (1) Virtual window control method: M W

[0071] The virtual window control method constructed by this invention based on the actual design constraints of the automated logistics sorting system is as follows:

[0072] ①. Specify the virtual window control method;

[0073] ②. Mixed specified and random virtual window control method;

[0074] ③. Random virtual window control method.

[0075] (2) Collect the operating speed of the belt conveyor: v b

[0076] (3) Virtual window length: L W

[0077] (4) Number of simultaneously open injection belt conveyors: N i

[0078] The simulation optimization framework for solving this problem model is specifically described as follows:

[0079] The framework consists of three modules: initialization, simulation optimization, and data processing. The initialization module initializes the parameters of the conveying equipment, control parameters, cargo information, and algorithm parameters of the model and generates an initial population. The initial population is then fed back to the algorithm model in the simulation optimization module, which controls the simulation model to perform optimization iterations. Finally, the data processing module statistically analyzes and evaluates the data obtained from the simulation optimization module. If the algorithm termination condition is met, an optimized solution set is output, and the optimized solution set is decoded to generate an optimized scheme for decision-makers to choose from. If the algorithm termination condition is not met, the population encoding is updated and fed back to the simulation optimization module to continue optimization iterations.

[0080] The multi-objective adaptive parallel wolf pack solution algorithm is specifically described as follows:

[0081] Based on the multi-control parameter optimization problem model of the merging end of the automated logistics sorting system established according to the present invention, the multi-objective adaptive parallel wolf pack algorithm proposed in this invention designs the encoding of individual wolves, the initial population generation method, and the intelligent behavior mechanism of the wolf pack, and obtains the optimal solution set through Pareto non-dominated sorting iterative optimization; the multi-objective adaptive parallel wolf pack algorithm includes the following steps:

[0082] Step 1: Initialize the population;

[0083] Regarding the encoding method, this invention proposes a hybrid integer single-chain encoding. The first segment represents the virtual window control method, selected from a set of virtual window control methods; the second segment represents the operating speed of the belt conveyor (unit: m·s). -1 ), from real number [v jmin ,v jmax The value is taken from [L]; the third segment represents the length of the virtual window (unit: m), from the real number [L]. Wmin ,L Wmax The value is taken from [N]; the fourth segment represents the number of simultaneously open injection belt conveyors (unit: number), starting from the integer [N]. min N max The values ​​are taken from []. Since the fitness value of each individual in the population is obtained through simulation rather than a function, to facilitate determining whether individuals in each population need to be updated during the optimization process of the multi-objective adaptive parallel wolf pack algorithm, two coding bits are added to the encoding, namely segment V and segment VI, which represent the fitness values ​​of the two optimization objectives F1 and F2, respectively. At the same time, considering that during the iteration process of the multi-objective adaptive parallel wolf pack algorithm, some individuals in the current generation and the previous generation may be the same, especially when the algorithm converges in the later stages of iteration, the differences between individuals in the current generation and the previous generation are smaller; therefore, in the encoding Add a seventh segment of encoding, defining it as a flag bit, taking a value from binary 0 or 1; if there is the same individual in the current generation and the previous generation, there is no need to start the simulation of that individual again, shortening this part of the time will effectively improve the simulation optimization efficiency. The target fitness value can be taken from the corresponding encoding bit of the corresponding individual in the previous generation, and the flag bit is set to 1; if there is no same individual, the flag bit is set to 0; according to the Cartesian product theory, due to the independence between the four control parameters, each individual must have the above four control parameters, and each control parameter can only exist in one of the corresponding control parameter sets;

[0084] Step 2: Combine random methods and reverse learning strategies to generate the initial population and the reverse population, calculate the target fitness value of each initial individual and the reverse individual, and evaluate the target fitness value to determine the initial population;

[0085] Step 3: Sort the initial population according to its target fitness value using Pareto non-dominated sorting, calculate the crowding degree of individuals of the same fitness level, eliminate redundant individuals, and temporarily store non-dominated individuals in the external space Ψ1.

[0086] Step 4: Implement the wolf pack classification mechanism for non-dominant individuals;

[0087] The individual with the smallest F1 value on the Pareto front is defined as the first alpha wolf; the individual with the smallest F2 value (excluding the first alpha wolf) on the Pareto front is defined as the second alpha wolf. If multiple individuals have the smallest F1 or F2 values, one of them is randomly selected. Next, an integer S between [y / (β+1), y / β] is randomly selected according to the wolf detection ratio factor β. num The number of individual scouting wolves is defined as y, where y represents the total number of individuals in the current population; the remaining individuals are defined as individual wolves.

[0088] Step 5: Based on the wolf roaming mechanism and the alpha wolf summoning mechanism, the local neighborhood search mechanism and heuristic optimization strategy are executed in parallel on the classified wolf roaming individuals and wolf alpha individuals, and the target fitness values ​​of the roaming population and the summoned population are calculated.

[0089] An adaptive walk probability mechanism is introduced into the walk mechanism, and its walk probability is determined by the following formula;

[0090]

[0091] In the formula, P m Indicates the adaptive walk probability; These represent the walk probabilities in the early and late stages of the walk, respectively, and d represents the current iteration number. max Indicates the maximum number of iterations;

[0092] The roaming mechanism is executed by individual wolf scouts, introducing an adaptive local neighborhood search mechanism and a heuristic optimization strategy. A coding bit of each individual wolf scout within the roaming probability is randomly selected and a coding different from the original coding is randomly generated according to the corresponding coding method. The target fitness value of each individual wolf scout after the roaming is calculated. In this mechanism, if the leader wolf individual is updated, the original leader wolf individual is regarded as an individual wolf scout.

[0093] The summoning mechanism is executed by individual wolves. Since each individual wolf needs to quickly move towards the alpha wolf during the hunt, the summoning mechanism employs an adaptive local neighborhood search mechanism and a heuristic optimization strategy. All the coding bits of each individual wolf are compared with those of a random alpha wolf. Two random coding bits of the individual wolf are replaced with the corresponding coding bits of the alpha wolf, and the target fitness value of each individual wolf after summoning is calculated. If the alpha wolf is updated, the original alpha wolf is considered as an individual wolf.

[0094] Step 6: Based on the wolf siege mechanism, the global neighborhood search mechanism and heuristic optimization strategy are jointly executed on the individual scout wolves after the scout wolves have moved and the individual wolf wolves after the alpha wolf has summoned them. The target fitness value of the population after the siege is calculated and the population after the siege is temporarily stored in the external space Ψ2.

[0095] The siege mechanism is jointly executed by all individual scout wolves after the scout wolves have roamed and all individual predatory wolves after the alpha wolf has summoned them. An adaptive global neighborhood search mechanism and a heuristic optimization strategy are introduced. The coding bits of each individual scout wolf are compared with those of a random alpha wolf, and the coding bits of each individual predatory wolf are compared with those of another alpha wolf. The two random codes are swapped with the different codes of the corresponding coding bits of the corresponding alpha wolf, and the target fitness value of the population after the siege is calculated.

[0096] Step 7: A new population is formed by combining the populations in outer space Ψ1 and outer space Ψ2. The new population is sorted by Pareto non-dominated ordering based on the corresponding target fitness value. Crowding is calculated for individuals of the same level, and redundant individuals are eliminated.

[0097] Step 8: Determine if the termination condition is met. If it is, proceed to Step 9; otherwise, return to Step 4 to continue iterating.

[0098] Step 9: Output the Pareto front, and the algorithm terminates;

[0099] Problem Setting: This study focuses on an automated logistics sorting system with two collection belt conveyors and 10 injection belt conveyors on each side. The system will perform a 5-day sorting task with a total of 43,265 items. The system will be controlled via a virtual window system. W It consists of specified virtual window control mode, random virtual window control mode, and mixed specified and random virtual window control mode; collecting the running speed v of the belt conveyor. b Values ​​range from [0.6, 1.5], unit: m / s; virtual window length L W Values ​​range from [1.5, 3.0], unit: m; Number of simultaneously open injection belt conveyors N i Values ​​are taken between [10, 20], in units of: items;

[0100] Algorithm parameter settings: The algorithm parameters were determined by orthogonal experiments, including population size y = 70, wolf probing factor β = 3, and the probabilities of wolf movement in the early and late stages. Maximum number of iterations d max =200;

[0101] Table 1 shows the Pareto solution set obtained by this method on a 5-day scale;

[0102] Table 1. Pareto solution set obtained under a 5-day timeframe.

[0103] <![CDATA[F1]]> 1.386 1.427 1.635 2.006 2.097 2.163 2.241 <![CDATA[F2]]> 3488.16 3206.15 3202.06 3189.28 3153.93 3144.97 3142.36

[0104] The specific embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the above embodiments. Within the scope of knowledge possessed by those skilled in the art, various changes can be made without departing from the spirit of the present invention.

Claims

1. A method for optimizing multiple control parameters at the merging end of an automated logistics sorting system, characterized in that: First, a multi-control parameter optimization problem model is established at the merging end of an automated logistics sorting system; then, a simulation optimization framework is constructed to solve this problem model; finally, a multi-objective adaptive parallel wolf pack algorithm is designed to solve the problem. The model for optimizing multiple control parameters at the merging end of an automated logistics sorting system is established based on the impact of virtual window technology and control methods, the operating speed of the collection belt conveyor, the length of the virtual window, and the number of simultaneously open injection belt conveyors as control parameters on the merging efficiency of the automated logistics sorting system. The problem model is described as follows: Based on the principle of minimizing the average waiting time of goods injected into the collection belt conveyor, the first optimization objective is established as follows: In the formula, This indicates the average waiting time for each item injected into the collection belt conveyor; This indicates the injection belt conveyor number. Indicates the maximum number of the injection belt conveyor; This indicates an injection belt conveyor. The item number on the package Indicates injection belt conveyor The highest cargo number on the list; Indicates injection belt conveyor Upper Waiting time for each item to be injected into the collection belt conveyor; Based on the principle of minimizing the total energy consumption of sorting and conveying equipment in an automated logistics sorting system, the second optimization objective is established as follows: In the formula, This represents the total energy consumption of sorting and conveying equipment in an automated logistics sorting system. This indicates the sorting equipment number in the automated logistics sorting system. Indicates the maximum sorting device number in the automated logistics sorting system; This refers to the sorting equipment in an automated logistics sorting system. Energy consumption; This indicates the serial number of the conveyor equipment in the automated logistics sorting system. This indicates the number of the largest conveyor in the automated sorting system. This refers to the conveyor equipment in an automated logistics sorting system. Energy consumption; The following constraints also need to be met: In the formula, Indicates injection belt conveyor Upper The length of each item; Indicates injection belt conveyor Upper The width of each item; Indicates injection belt conveyor Upper The height of each item; Indicates injection belt conveyor Upper The maximum permissible volume of each item; In the formula, Indicates injection belt conveyor Upper The weight of each item; Indicates the maximum permissible weight of the goods; In the formula, Indicates injection belt conveyor The number of goods that can be served simultaneously; In the formula, This indicates the arrival time of the first item that needs to be sorted in the system; Indicates the start time set for the simulation model; In the formula, This indicates the arrival time of the last item that needs to be sorted in the system; Indicates the end time set for the simulation model; In the formula, Indicates goods The actual injected virtual window; Indicates goods The actual virtual window used for the application; In the formula, Indicates the number of empty virtual windows in the system; In the formula, This indicates the number of goods that each virtual window can serve simultaneously. In the formula, Indicates virtual window Length; Indicates virtual window Minimum allowed length; Indicates virtual window Maximum allowed length; In the formula, , indicating a virtual window A set of control methods, Indicates virtual window The last element in the set of control methods; In the formula, Indicates the number of injection belt conveyors that are open simultaneously; Indicates the minimum number of injection belt conveyors that can be open simultaneously; Indicates the maximum number of injection belt conveyors that can be opened simultaneously; In the formula, This indicates the operating speed of the conveyor equipment in an automated logistics sorting system; This indicates the minimum permissible operating speed of conveyor equipment in an automated logistics sorting system; This indicates the maximum permissible operating speed of the conveyor equipment in an automated logistics sorting system; The simulation optimization framework for solving this problem model is specifically described as follows: The framework consists of three modules: initialization, simulation optimization, and data processing. The initialization module initializes the parameters of the conveying equipment, control parameters, cargo information, and algorithm parameters of the model and generates an initial population. The initial population is then fed back to the algorithm model in the simulation optimization module, which controls the simulation model to perform optimization iterations. Finally, the data processing module statistically analyzes and evaluates the data obtained from the simulation optimization module. If the algorithm termination condition is met, the optimized solution set is output, and the optimized solution set is decoded to generate an optimized scheme for decision-makers to choose from. If the algorithm termination condition is not met, the population encoding is updated and fed back to the simulation optimization module to continue the optimization iteration; The multi-objective adaptive parallel wolf pack solution algorithm is specifically described as follows: Step 1: Initialize the population; Step 2: Combine random methods and reverse learning strategies to generate the initial population and the reverse population, calculate the target fitness value of each initial individual and the reverse individual, and evaluate the target fitness value to determine the initial population; Step 3: Sort the initial population according to its target fitness value using Pareto non-dominant sorting. Calculate the crowding degree of individuals of the same fitness level, eliminate redundant individuals, and temporarily store non-dominant individuals in the external space. middle; Step 4: Implement the wolf pack classification mechanism for non-dominant individuals; Step 5: Based on the wolf roaming mechanism and the alpha wolf summoning mechanism, the local neighborhood search mechanism and heuristic optimization strategy are executed in parallel on the classified wolf roaming individuals and wolf alpha individuals, and the target fitness values ​​of the roaming population and the summoned population are calculated. Step 6: Based on the wolf siege mechanism, perform a global neighborhood search mechanism and a heuristic fitness preservation strategy on both the individual scout wolves after the scout wolves have moved and the individual wolf wolves after the alpha wolf has summoned them. Calculate the target fitness value of the population after the siege and temporarily store the population in the external space. middle; Step 7: From external space and external space The population in the middle is used to form a new population. The new population is sorted by Pareto non-dominated ordering according to the corresponding target fitness value. The crowding degree of individuals of the same level is calculated and redundant individuals are eliminated. Step 8: Determine if the termination condition is met. If it is, proceed to Step 9; otherwise, return to Step 4 to continue iterating. Step 9: Output the Pareto front, and the algorithm terminates.

Images