Intelligent power grid material distribution planning method based on three-dimensional boxing

Abstract

The invention provides an intelligent power grid material distribution planning method based on three-dimensional encasement, and designs a space dimension reduction method and a heuristic evolutionary algorithm in combination with the problems of wide material distribution of an electric power system and the vehicle space utilization rate in the transportation process. The problem of power infrastructure material distribution optimization under the constraint of three-dimensional boxing is solved. According to the method, through adaptive expansion and a virtual boxing technology, the method can better adapt to power equipment of various different shapes, sizes and weights. Meanwhile, according to the method, the distribution time is shortened and the distribution efficiency is improved by optimizing the distribution route, so that the distribution cost is reduced and the operation efficiency of the power system is improved. In addition, the cargo stability is considered in detail in the process of designing the boxing operator and heuristic evolution, and the cargo safety in the distribution process is guaranteed. In general, the invention provides an effective, efficient and intelligent power system material distribution planning scheme.

Description

Technical Field

[0001] The present invention belongs to the field of power grid logistics and distribution, and more specifically relates to an intelligent power grid material distribution planning method based on three-dimensional packaging. Background Art

[0002] In the operation of power systems, material distribution has become a major problem due to the wide distribution of material resources. To solve this problem, a smart grid material distribution planning method based on three-dimensional packing is proposed.

[0003] The three-dimensional bin packing problem is a typical combinatorial optimization problem, primarily concerned with how to place multiple objects into one or more given bins to achieve an optimal target value. This is a very complex optimization problem, and to date, no algorithm has been found that can solve it in polynomial time. Currently, solutions rely primarily on heuristic algorithms.

[0004] Logistics and delivery are complex tasks involving numerous issues, including vehicle planning, route selection, and time scheduling. Numerous constraints must be considered, such as vehicle weight, load volume, transport distance, and service time. In power systems, many of these issues are exacerbated by the unique characteristics of power equipment. For example, power equipment is typically large, and the size, weight, and shape of different equipment vary significantly, requiring appropriate loading planning and selection of appropriate delivery routes based on these characteristics.

[0005] Based on the solution process of the three-dimensional packing problem, this paper designs an intelligent power grid material distribution planning method. This method fully considers the space utilization of vehicles and the optimization of paths, thereby effectively reducing the time and cost of material distribution and improving the efficiency and stability of power system operation. Summary of the Invention

[0006] The main technical problem solved by the present invention is how to optimize the distribution space by combining the three-dimensional packing problem in the material distribution process of the power system, intelligently plan the distribution route, improve the loading efficiency and distribution efficiency of the vehicle, reduce the distribution time and cost, and improve the efficiency and stability of the power system operation.

[0007] In order to achieve the above object, the present invention is implemented by adopting the following technical solutions: the method comprises:

[0008] Design a spatial dimensionality reduction method for the three-dimensional bin packing problem;

[0009] Design of bin packing operators and heuristic evolution;

[0010] Describe the optimization problem of power infrastructure material distribution under three-dimensional packing constraints;

[0011] Provide the vehicle routing optimization model assumptions;

[0012] Establish vehicle routing optimization model and constraints under three-dimensional packing constraints;

[0013] A vehicle routing optimization genetic algorithm is designed to solve the vehicle routing optimization model under three-dimensional packing constraints.

[0014] In one embodiment, the spatial dimensionality reduction method for the three-dimensional packing problem includes: performing preliminary virtual packing through an adaptively expanded virtual container body, and then loading the virtual container into the loaded container to construct a new loading space, so that the loading process is more adaptable to changes in the loaded cargo and the robustness of the space allocation is enhanced;

[0015] The setting rules are as follows: 1) Grouping is based on an anthropomorphic packing algorithm with strong constraints; 2) Each group expands the virtual container body based on the larger cargo; 3) Each group contains at least 3 cargoes and the total volume of the virtual container body is between 900dm3 and 5000dm3; 4) After each grouping, the cargo occupies more than 90% of the total volume of its corresponding virtual container.

[0016] In one solution, the design of the packing operator and the heuristic evolution includes: designing a data-driven crossover operator, introducing similar posture replacement and similar goods replacement into the crossover operator, thereby improving the convergence speed of the algorithm;

[0017] During the algorithm iteration, first, two goods with the same loading position in the two similar solutions to be optimized are found, and the cargo numbers of the two goods are judged to be the same. If the cargo numbers are the same, the posture approximation rule of the goods is introduced. If the cargo posture codes of the two similar solutions are similar, the coding position is classified as the first type of similar coding position.

[0018] If the cargo numbers are different, the shape information (X1, Y1, Z1) and (X2, Y2, Z2) of the two cargoes are introduced. If |X1-X2| < 10, |Y1-Y2| < 10, |Z1-Z2| < 10, it means that the shapes of the two cargoes are similar. The cargo number coding and cargo posture coding positions of the two similar solutions are the second type of similar coding positions.

[0019] Then, the first type of similar coding positions are cross-operated according to the cross-probability R1, and the second type of similar coding positions are cross-operated according to the cross-probability R2.

[0020] In one scenario, the power infrastructure material distribution optimization problem is described as follows: The vehicle routing optimization problem under three-dimensional packing constraints can be described as follows: Given an undirected graph G = (D, E) representing the distribution network, where D = {D0, D1, D2, ..., D n} is the vertex set, E={(D i ,Dj ) / D i ,D j ∈D,0≤i,j≤N,i≠j} is a path set; the vertex D0 represents the distribution center, (D1,D2,...,D n ) represents the customer point set; (D i ,D j ) indicates customer D i and D j The lines between them are connected with d ij Indicates customer D i and Customer D j The distance between ij =d ji The distribution center has k trucks of different types, and the rated load mass and maximum volume of each truck are represented by G k and V k Indicates: The length, width and height of the truck are expressed in L. k ,W k ,H k Indicated by; the length, width, height and mass of each piece of cargo are expressed by l i ,w i ,h i ,g i express.

[0021] In one embodiment, the vehicle routing optimization model assumes that:

[0022] (1) There is only one distribution center, and the goods in the distribution center can meet the needs of all customers;

[0023] (2) The number of vehicles owned by the distribution center meets all distribution requirements;

[0024] (3) The vehicle used departs from the distribution center and returns to the distribution center after delivery;

[0025] (4) Each customer's needs are met and served only once;

[0026] (5) All goods required by customers are known;

[0027] (6) The locations and distances between customers and between customers and distribution centers are known;

[0028] (7) The mass and volume of cargo loaded on each vehicle cannot exceed the vehicle's permitted rated load mass and maximum volume constraints;

[0029] (8) Assume that all cargoes are regular and uniform rectangular parallelepipeds and the length, width and height of each cargo do not exceed the dimensions of the vehicle box;

[0030] (9) The delivery vehicles are box trucks of the same model;

[0031] (10) The cargo must be placed parallel to the carriage and only horizontal rotation is allowed. The schematic diagram of the cargo rotation direction in three-dimensional space is shown in the figure below;

[0032] (11) To ensure the stability of the vehicle, the center of gravity of the cargo must be within the allowable range of the vehicle compartment;

[0033] (12) Goods delivery does not consider time windows;

[0034] (13) The placement of goods must comply with the first-in-last-out principle, that is, if customers are on the same service route, the goods required by the first-served customer cannot be blocked or pressed by the goods required by the later-served customer.

[0035] In one embodiment, the constraints include: (1) vehicle load constraints; (2) vehicle volume constraints; (3) rotation direction constraints; (4) cargo stability constraints; (5) vehicle size constraints; (6) delivery constraints; and (7) loading sequence constraints.

[0036] In one embodiment, a vehicle routing optimization genetic algorithm includes:

[0037] 1) Calculate the fitness of each individual in the population f(r=1,2,3...,M), where M is the size of the population;

[0038] 2) Calculate the probability of each individual being inherited into the next generation population, that is, the probability of individual r being selected;

[0039] 3) Calculate the cumulative probability of each individual;

[0040] 4) Generate a uniformly distributed pseudo-random number e in the interval [0,1];

[0041] 5) If e < q(1), select individual 1; otherwise, select individual v;

[0042] 6) Repeat steps 4) and 5) M times.

[0043] Beneficial effects of the present invention:

[0044] The present invention is based on the three-dimensional packing problem and performs intelligent planning for the distribution of materials in the power system, which can improve the loading efficiency of vehicles, reduce space waste, and thus reduce the cost of material distribution.

[0045] By optimizing the delivery route, the present invention can significantly reduce the delivery time, improve the delivery efficiency, and help improve the operating efficiency and stability of the power system.

[0046] Through adaptive expansion and virtual packaging technology, the present invention can better adapt to power equipment of various shapes, sizes and weights, which is conducive to improving the flexibility and robustness of packaging.

[0047] The method of the present invention is highly intelligent, and the heuristic evolutionary algorithm it adopts is powerful and efficient, and can maintain efficient solution performance even when faced with large-scale complex problems.

[0048] The design of the packing operator and the heuristic evolution process of the present invention also takes into account the stability of the goods, so that while ensuring the distribution efficiency, it can also ensure that the goods are not prone to accidents such as dumping during transportation, thereby ensuring the safety of the goods. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 Flow chart of the method of the present invention;

[0050] Figure 2 For grouping operations;

[0051] Figure 3 is the crossover operator;

[0052] Figure 4 is a random mutation strategy;

[0053] Figure 5 It is a single-point mutation strategy;

[0054] Figure 6 It is a sequential reversal mutation strategy;

[0055] Figure 7 This is a schematic diagram of material distribution for power infrastructure;

[0056] Figure 8 Schematic diagram of the direction of cargo rotation;

[0057] Figure 9 is the coding scheme;

[0058] Figure 10 space for cargo;

[0059] Figure 11 This is a schematic diagram of a single-point crossover operation;

[0060] Figure 12 Schematic diagram of basic bit variation. DETAILED DESCRIPTION

[0061] To facilitate understanding of the present invention, the present invention will be described more fully below with reference to the accompanying drawings. The drawings illustrate exemplary embodiments of the present invention. However, the present invention may be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to provide a more thorough and comprehensive understanding of the present invention.

[0062] Unless otherwise defined, all technical and scientific terms used herein have the same meanings as those understood by those skilled in the art to which the present invention pertains. The terms used in the present specification are for the purpose of describing specific embodiments only and are not intended to limit the present invention. To facilitate understanding of the present invention, a more comprehensive description of the present invention will be provided below with reference to the accompanying drawings. Typical embodiments of the present invention are shown in the drawings. However, the present invention may be embodied in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to provide a more thorough and comprehensive understanding of the present invention.

[0063] like Figure 1 As shown, a smart grid material distribution planning method based on three-dimensional packing specifically includes the following steps:

[0064] Step 1: Design a spatial dimensionality reduction method suitable for the three-dimensional bin packing problem

[0065] The decision space for the three-dimensional packing problem is extremely large. Therefore, designing a spatial dimensionality reduction method suitable for this problem can significantly aid in the search for feasible solutions. Grouping strategies are a classic dimensionality reduction strategy for large-scale multi-objective optimization problems. While their advantage is that they can significantly reduce the search space, their disadvantage is that the effectiveness of grouping can affect algorithm performance. Since the encoding of the three-dimensional packing problem allows for evaluation of grouping effectiveness, a rule-based grouping strategy can accelerate the search for feasible solutions. The grouping process for the three-dimensional packing problem involves performing preliminary virtual packing using an adaptively expanded virtual container body, then loading the virtual container into the loaded container to construct a new loading space. This makes the loading process more adaptable to changes in the loaded cargo and enhances the robustness of the space allocation. Based on the container used in the experiment, the following rules were set: 1) Grouping was performed based on a strongly constrained anthropomorphic packing algorithm. 2) Each group expanded the virtual container body based on the larger cargo. 3) Each group contains at least three cargoes and the total volume of the virtual container is between 900dm3 and 5000dm3; 4) After each grouping, the cargo occupies more than 90% of the total volume of the corresponding virtual container.

[0066] The diagram of group operation is as follows Figure 2 As shown in the figure, 6 goods are divided into one group, that is, S = (1,1,1,2,1,3,3,1,5,4,1,7,5,1,9,6,1,11) is simplified to S = (1,1,1) in the solution space, achieving a large degree of dimensionality reduction.

[0067] Step 2: Binning Operator and Heuristic Evolution Design

[0068] Considering that the core of evolutionary algorithms is the evolutionary operator, three common crossover methods for bin packing problems are partial matching crossover, cyclic crossover, and sequential crossover. Experimental verification shows that partial matching crossover yields better results, but often leads to slower evolution. The optimization process of combinatorial optimization differs from that of simple function optimization, primarily in that continuous optimization can select the best solution for mean optimization. However, in three-dimensional bin packing, the decision variables and the objective function are weakly correlated. Conventional crossover operators or binomial crossover-mutation operators cannot meet the basic requirements of optimization methods such as mean optimization. Therefore, it is necessary to design a new crossover-mutation operator suitable for three-dimensional bin packing optimization.

[0069] Based on this, the present invention designs a data-driven crossover operator, which introduces similar posture replacement and similar cargo replacement into the crossover operator, thereby improving the convergence speed of the algorithm. During the algorithm iteration, first, find two cargoes with the same loading position in the two similar solutions to be optimized, and determine whether the cargo numbers of the two cargoes are the same. If the cargo numbers are the same, the posture approximation rule of the cargo is introduced. If the cargo posture codes of the two similar solutions are similar, then the coding position is the first type of similar coding position; if the cargo numbers are different, the shape information (X1, Y1, Z1) and (X2, Y2, Z2) of the two cargoes are introduced. If |X1-X2|<10, |Y1-Y2|<10, |Z1-Z2|<10, it means that the shapes of the two cargoes are similar, and the cargo number codes and cargo posture coding positions of the two similar solutions are the second type of similar coding positions. Then, the first type of similar coding positions are cross-operated according to the cross-probability R1, and the second type of similar coding positions are cross-operated according to the cross-probability R2. The crossover operation is as follows Figure 3 shown.

[0070] The mutation operation will enhance the diversity of the population. A variety of mutation strategies are used to avoid the occurrence of immature convergence and local convergence caused by crossover operations. Random mutation strategy randomly replaces a few identical variables of similar solutions, as shown in Figure 4 below. Single point mutation selects a random position in the coding segment and replaces it with a random real number, as shown in Figure 5 below. Sequence reversal mutation randomly selects two points in the parent generation as the mutation points of the mutation, and then selects two points in the offspring in the opposite order of the parent generation for rearrangement. It is a bionic mutation method, as shown below Figure 6 As shown in the figure. During the algorithm iteration, if the algorithm is stuck in a local optimum, that is, if there are many optimal solutions near a certain region of the solution set, a random mutation strategy is used to explore the unknown space. The suboptimal solution is the second-order solution in the non-dominated sort. If there is an optimal solution near a certain region of the solution set, but the optimal solutions are relatively few, a single-point mutation strategy is used to accelerate convergence and improve the degree of convergence. Sequence reversal mutation occurs in any solution set with a mutation probability β.

[0071] S1,S2,...,S mare multiple feasible solutions sorted by non-dominated solution sets, f1,f2,...,f m For each solution, the normalized solution vector is obtained by Euclidean distance d k.i The similarity of the solutions is determined as shown in formula (1). The similarity threshold is changed by the deformed negative exponential function, as shown in formula (2), so that the algorithm expands the search space in the early stage of iteration and accelerates convergence in the later stage.

[0072]

[0073] Where: c k (f k ,f i ) is the projection distance from the i-th solution vector to the k-th solution vector; d k,i (f k ,f i ) is the vertical distance from the i-th solution vector to the k-th solution vector; m is the number of feasible solutions; θ is the similarity threshold parameter; G is the maximum number of iterations; G T is the current iteration number. k,i (f k ,f i )<θ, it means f i Solution vector and f k Too similar, that is, solution S i With S k Too similar. If the number of similar solutions is greater than 8, it means there are too many similar solutions; if the number of similar solutions is less than 3, it means there are too few similar solutions; otherwise, the solution set characteristics are not obvious.

[0074] Step 3: Describe the optimization problem of power infrastructure material distribution under three-dimensional packing constraints

[0075] After completing the loading of power infrastructure materials, its distribution link is still worthy of in-depth invention. Cargo packing and vehicle path optimization are the basic elements for the distribution center to complete the logistics distribution tasks. The solution of these two problems directly affects the level of distribution efficiency. These two problems are not isolated problems, they are interrelated, mutually influential and mutually restricted. The present invention integrates the three-dimensional packing problem and the vehicle path optimization problem to find the entry point for the invention of this combined optimization problem. Taking into account the packing constraints and vehicle path constraints, a multi-objective combined optimization model with the goals of shortest transportation path, maximum vehicle loading space utilization, and maximum vehicle load utilization is proposed. In view of the complexity of multi-objective algorithm solution, the model is established by adding weight coefficients to the multi-objective and converting it into a single-objective optimization.

[0076] The vehicle routing optimization problem under three-dimensional packing constraints can be described as follows: Given an undirected graph G = (D, E) representing the distribution network of the present invention, where D = {D0, D1, D2, ..., D n} is the vertex set, E={(D i ,D j ) / D i ,D j ∈D,0≤i,j≤N,i≠j} is a path set. Vertex D0 represents the distribution center, (D1,D2,...,D n ) represents the customer point set. (D i ,D j ) indicates customer D i and D j The lines between them are connected with d ij Indicates customer D i and Customer D j The distance between ij =d ji The distribution center has k trucks of different types, and the rated load mass and maximum volume of each truck are represented by G k and V k Indicates: The length, width and height of the truck are expressed in L. k ,W k ,H k The length, width, height and mass of each piece of cargo are represented by l i ,w i ,h i ,g i The distribution process of this problem is as follows Figure 7 As shown:

[0077] Step 4: Establish the assumptions of the vehicle routing optimization model under the three-dimensional packing constraint;

[0078] Before modeling, in order to facilitate the establishment of the mathematical model, the following assumptions are made regarding the definition of the combinatorial optimization problem:

[0079] (1) There is only one distribution center, and the goods in the distribution center can meet the needs of all customers;

[0080] (2) The number of vehicles owned by the distribution center meets all distribution requirements;

[0081] (3) The vehicle used departs from the distribution center and returns to the distribution center after delivery;

[0082] (4) Each customer's needs are met and served only once;

[0083] (5) All goods required by customers are known;

[0084] (6) The locations and distances between customers and between customers and distribution centers are known;

[0085] (7) The mass and volume of cargo loaded on each vehicle cannot exceed the vehicle's permitted rated load mass and maximum volume constraints;

[0086] (8) Assume that all cargoes are regular and uniform rectangular parallelepipeds and the length, width and height of each cargo do not exceed the dimensions of the vehicle box;

[0087] (9) The delivery vehicles are box trucks of the same model;

[0088] (10) The cargo must be placed parallel to the carriage and only horizontal rotation is allowed. The schematic diagram of the cargo rotation direction in three-dimensional space is as follows Figure 8 As shown;

[0089] (11) To ensure the stability of the vehicle, the center of gravity of the cargo must be within the allowable range of the vehicle compartment;

[0090] (12) Goods delivery does not consider time windows;

[0091] (13) The placement of goods must comply with the first-in-last-out principle (First In Last Out). That is, if customers are on the same service route, the goods required by the first-served customer cannot be blocked or pressed by the goods required by the later-served customer.

[0092] According to the model to be established in this article, the relevant specific symbols are explained as follows:

[0093] k represents the vehicle number, k = 1, 2, 3..., k;

[0094] G k ,V k , represents the rated loading mass and maximum volume of vehicle k;

[0095] L k ,W k ,H k represents the length, width and height of the vehicle k box;

[0096] i represents the node number, i=0 represents the distribution center, i=1,2,3...j...,n represents the customer node;

[0097] d ij Represents the distance between two client nodes;

[0098] T i Indicates the total number of goods required by customer i;

[0099] l it ,w it ,h it represents the length, width, and height of the t-th item required by customer i;

[0100] git represents the quality of the t-th item required by customer i;

[0101] x it ,y it ,z it represents the center of gravity of the t-th item of customer i in the vehicle;

[0102] (a′,a″),(b′,b″),(c′,c″) represent the center of gravity range of the vehicle, and the values ​​of a′,a″,b′,b″,c″ are determined by the vehicle cabin specifications.

[0103] Indicates the coordinate of the lower right corner of the t-th piece of cargo of customer i in the k-th car.

[0104] Indicates the coordinate of the upper left corner of the t-th item of customer i in the k-th car.

[0105] The decision variables are as follows:

[0106]

[0107]

[0108] Step 5: Establish the vehicle routing optimization model and constraints under three-dimensional packing constraints

[0109] Before establishing a distribution model for power infrastructure materials, we should consider the actual needs of distribution centers and customers and consider practical constraints, including vehicle constraints, packing constraints, and delivery priorities. The specific constraints before establishing the model are described below:

[0110] (1) Vehicle load constraints: The sum of the load mass of the goods on a single vehicle must not exceed the rated load mass of the vehicle.

[0111] (2) Constraints on vehicle volume: The sum of the loading volume of the cargo on a single vehicle must not exceed the maximum volume of the vehicle.

[0112] (3) Rotational direction constraint: This article stipulates that the cargo cannot rotate arbitrarily in the carriage. It must be parallel to the carriage and can only rotate in the horizontal direction. That is, on the one hand, the length direction of the cargo is parallel to the length direction of the carriage or the width direction of the carriage (ignoring the gaps between the cargo).

[0113]

[0114] or

[0115]

[0116] On the other hand, the cargo cannot be inverted or turned sideways in the carriage, and must be facing upwards, that is, the height direction of the cargo must be parallel to the height direction of the carriage.

[0117] (4) Constraints on cargo stability: The purpose of constraining the center of gravity of the cargo loaded on the vehicle is to ensure the safety of the vehicle during driving. That is, the center of gravity of all cargo in the vehicle compartment must be within the required range of the center of gravity of the vehicle compartment, that is, the center of gravity in the longitudinal direction of the compartment must meet the requirements. Its center of gravity in the width direction of the carriage satisfies The center of gravity in the height direction of the car meets the requirements

[0118] 5) Vehicle size constraint: For any vehicle k, the cargo must be loaded inside the vehicle and cannot exceed the length, width and height of the vehicle. That is, the upper left coordinate of any cargo loaded on vehicle k must satisfy

[0119] (6) Distribution constraints: Distribution constraints are mainly considered from three aspects. First, all vehicles involved in distribution activities must start from the distribution center and return to the distribution center after delivering the goods. That is, any vehicle must meet Secondly, the same customer can only be served by the same car, that is, for any customer, there should be Finally, every customer's needs must be met, that is, for any customer

[0120] (7) Loading order constraint: During the distribution process, the placement of goods on the same route must meet the first-in-last-out principle (FILO), that is, the goods of the customer served first should be loaded last, and the goods loaded first cannot block or press the goods loaded later. If the customer i is served first, that is, r ijk = 1, it must satisfy that for any customer i's t-th item, the difference between the coordinate value of its upper left front corner and the corresponding coordinate value of customer j's t-th item is greater than or equal to the length, width and height corresponding to customer i's t-th item. and At least one holds true; when r ijk =0 is the opposite. In summary, the multi-objective optimization mathematical model of three-dimensional packing and vehicle path optimization established by the present invention is as follows:

[0121]

[0122]

[0123] Formula (8) indicates that the transportation distance of the delivery vehicle is the shortest; Formula (9) indicates that the space utilization rate of the vehicle is the highest during the delivery process; Formula (10) indicates that the load utilization rate of the vehicle is the highest during the delivery process; Formula (11) indicates that the load of each vehicle cannot exceed the rated load mass required by the vehicle; Formula (12) indicates that the loading volume of each vehicle cannot exceed the maximum volume required by the vehicle; Formulas (13) and (14) indicate that the goods loaded on the vehicle must be loaded parallel to the vehicle compartment; Formula (15) indicates that the goods cannot be inverted in the vehicle compartment; Formulas (16) to (18) indicate that the center of gravity range of the goods loaded in the vehicle compartment must be within the range required by the center of gravity of the vehicle; Formulas (19) to (21) indicate that the goods loaded on the vehicle cannot exceed the size range of the vehicle compartment; Formula (22) indicates that the delivery vehicle departs from the distribution center and returns to the distribution center after the delivery service is completed; Formula (23) indicates that the same customer must be served by the same vehicle; Formula (24) indicates that all goods of all customers must be delivered.

[0124] Step 6: Design of genetic algorithm for vehicle routing optimization under three-dimensional packing constraints

[0125] Chromosome encoding operations

[0126] Individual encoding is the entry-level operation for genetic algorithms. It converts data into a computer's recognition space, continuously searching for individuals with higher fitness within the computer's search space, and ultimately finding the optimal or near-optimal solution to the problem. Encoding is the primary issue to be addressed when applying genetic algorithms and is a key step in their design. In addition to determining the arrangement of individual chromosomes, the encoding method also determines the decoding method used when transforming the individual's genotype in the search space to the phenotype in the solution space. This largely determines how the genetic algorithm evolutionary operation of the population is performed and the efficiency of the genetic algorithm operation. In the present invention, encoding is the process of converting the actual feasible solutions in the distribution process from reality to a search solution space that the genetic algorithm can handle. During the genetic algorithm operation, each individual in the population is composed of genes. Therefore, to match the chromosomes to the solution to be solved, the genes must be correctly encoded. Throughout the distribution process, information such as customer demand information, vehicle information, cargo information, cargo placement location, direction, and cargo placement level in the order form all need to be encoded. The encoding methods of genetic algorithms mainly include binary encoding, real number encoding, Gray code encoding, floating point encoding, symbol encoding, multi-parameter cascade encoding and multi-parameter cross encoding. Commonly used encoding methods are binary encoding and real number encoding. Binary encoding is simple and easy to operate, but it takes up a lot of space when solving practical problems, while real number encoding is to parse the problem space into real number space, which takes up less space. Therefore, according to the characteristics of the problem of this invention and the advantages of real number encoding, this article adopts the real number encoding method to encode the chromosomes. In view of the actual loading requirements in distribution, the encoding scheme of the present invention consists of 6 parts as follows Figure 9 shown.

[0127] In terms of the placement of the cargo, the present invention defines the loading position of the cargo as follows: Figure 10 shown.

[0128] Points A, B, and C represent the three loading reference points generated when a cargo is loaded into a carriage. At the same time, this paper assumes that the initial loading reference point coordinates of the first cargo loaded are (0, 0, 0).

[0129] Regarding the direction of placing the goods, two directions of placing the goods are allowed according to the above description. Therefore, the coding table corresponding to the direction of placing the goods defined in the present invention is shown in the following table.

[0130] Cargo placement direction coding table

[0131]

[0132] Regarding the placement level of cargo, the present invention defines the maximum allowable loading layer number of the vehicle box as the ratio of the height H of the vehicle box when the cargo to be loaded is placed parallel to the vehicle box to the smallest height h of the cargo to be loaded, that is:

[0133] The initial population is the starting point for genetic algorithm evolution, and this starting point has a certain impact on the evolutionary speed and quality of the population within the genetic algorithm. A good initial population (initial solution) can improve the convergence speed of the genetic algorithm and accelerate the population's evolution. The initial solution is a randomly generated sequence of natural numbers. Based on a designed encoding scheme, a feasible sequence is randomly generated, and this process is repeated until an initial population that meets the population size is generated. Genetic algorithms use the concept of fitness to measure the degree to which each individual in a population finds the optimal solution during an optimization calculation. Individuals with higher fitness have a greater probability of being passed on to the next generation, while individuals with lower fitness have a lower probability of being passed on to the next generation. In genetic algorithms, the population evolution process is based on the fitness of each individual in the population. Through repeated iterations, individuals with higher fitness are continuously sought, ultimately achieving the optimal solution or a near-optimal solution to the problem. To directly link the fitness function with the quality of individuals in the population, this paper uses the fitness function as the normalized objective function of the optimization model. As can be seen, the larger the function value, the greater the fitness value, i.e., the better the chromosomes, the better the population adaptability.

[0134] Selection operators are used in genetic algorithms to select individuals from a population for survival of the fittest. Individuals with high fitness have a greater probability of being selected for the next generation, while individuals with low fitness have a lower probability of being selected for the next generation. The selection operation in genetic operations determines how to select superior individuals for inheritance to the next generation. The selection operation is based on fitness evaluation, and its main purpose is to prevent the elimination of superior genes and improve the global convergence and computational efficiency of the genetic algorithm. Therefore, the choice of selection operator operation method is crucial for genetic algorithms. Common selection operator operation methods include proportional selection, deterministic sampling selection, random selection without replay, random selection with remainder without replay, sorted selection, and random league selection. Proportional selection is a type of random selection with a return, also known as roulette wheel selection. It calculates the probability of each individual appearing in the offspring based on its fitness value and randomly selects individuals to form the offspring population based on this probability. The starting point of the roulette wheel selection strategy is that individuals with better fitness values ​​are more likely to be selected. The specific operation steps of the roulette wheel selection operator are:

[0135] 1) Calculate the fitness of each individual in the population f(r=1,2,3...,M), where M is the size of the population;

[0136] 2) Calculate the probability of each individual being inherited into the next generation, that is, the probability of individual r being selected. The formula is as follows:

[0137]

[0138] 3) Calculate the cumulative probability of each individual. The formula is as follows:

[0139]

[0140] 4) Generate a uniformly distributed pseudo-random number e in the interval [0,1];

[0141] 5) If e<q(1), select individual 1; otherwise, select individual v such that: q(ν-1)<e≤q(ν);

[0142] 6) Repeat steps 4) and 5) M times.

[0143] Crossover operator, the crossover operation in the genetic algorithm refers to the process in which two paired chromosomes exchange some genes with each other through a certain crossover method to produce new individuals. Crossover operation is an important feature that distinguishes the genetic algorithm from other algorithms. Through crossover, the search ability of the genetic algorithm can be greatly improved. The crossover operation is the main method for generating new individuals in the genetic algorithm, so the crossover probability should generally take a larger value. The generally recommended value range is 0.4 to 0.99. At this stage, the main crossover methods are: single-point crossover, multi-point crossover, uniform crossover and arithmetic crossover. Generally, as the number of crossover points increases, the probability of individual structure being destroyed also gradually increases, and the difficulty of effectively preserving better individual patterns will also become greater. Because single-point crossover is the simplest and most commonly used crossover operator, the present invention selects the single-point crossover method as the crossover operator in this article. The specific implementation process of the single-point crossover operator is: pairing individuals in a population of size M, randomly setting a crossover point for each pair of paired individuals, and finally, for each pair of paired individuals, setting the crossover probability p according to the set crossover probability p. c At the crossover point, part of the chromosomes of two individuals are exchanged to produce a new individual. The schematic diagram of the single-point crossover operation is as follows Figure 11 shown.

[0144] Mutation operator, the mutation operation in the genetic algorithm refers to the process of replacing the gene value in the individual chromosome coding string with other alleles to produce new individuals. The mutation operation is performed after the selection and crossover operations, and is an auxiliary method for producing new individuals. In the mutation operation, if the mutation probability is large, although more new individuals can be produced, many better patterns will be destroyed. If the mutation probability is small, the function of the mutation operation to produce new individuals and inhibit premature maturation will be reduced. The generally recommended value range is 0.0001 to 0.1. Commonly used mutation operator operations mainly include: basic bit mutation, uniform mutation, boundary mutation, non-uniform mutation and Gaussian mutation. The basic bit mutation operator operation changes the gene value at the individual gene locus in the individual coding string. It is the simplest, most basic and most commonly used mutation operation operator. Compared with the operational complexity of other mutation operations, the basic bit mutation operator reduces the running time of the genetic algorithm. The specific execution process of the basic bit mutation operator is: for each gene locus of the individual, according to the mutation probability p m Designate it as a mutation point and then perform an inverse operation on the gene for each designated mutation point, and finally generate a new individual. According to the characteristics of the optimization problem and the advantages of the basic bit mutation operator, the mutation method used in this invention is the basic bit mutation method. The schematic diagram of the basic bit mutation operation is as follows Figure 12 shown.

[0145] Those skilled in the art will appreciate that all or part of the processes in the above-described method embodiments can be implemented by instructing related hardware through a computer program. The program can be stored in a computer-readable storage medium, and when executed, the program can include the processes in the above-described method embodiments. The storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0146] It should be understood that the detailed description of the technical solutions of the present invention using the preferred embodiments above is illustrative and not restrictive. A person skilled in the art, after reading the present specification, may modify the technical solutions described in the embodiments or replace some of the technical features therein with equivalents; such modifications or replacements do not deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A smart grid material distribution planning method based on three-dimensional packaging, characterized by: The method includes: Design a spatial dimensionality reduction method for the three-dimensional bin packing problem; Design of bin packing operators and heuristic evolution; Describe the optimization problem of power infrastructure material distribution under three-dimensional packing constraints; Provide the vehicle routing optimization model assumptions; Establish vehicle routing optimization model and constraints under three-dimensional packing constraints; A vehicle routing optimization genetic algorithm is designed to solve the vehicle routing optimization model under three-dimensional packing constraints.

2. The intelligent power grid material distribution planning method based on three-dimensional packaging according to claim 1 is characterized by: The spatial dimensionality reduction method for the three-dimensional packing problem includes: performing preliminary virtual packing through an adaptively expanded virtual container body, and then loading the virtual container into the loaded container to construct a new loading space, so that the loading process can better adapt to changes in the loaded cargo and enhance the robustness of space allocation; The setting rules are as follows: 1) Grouping is based on an anthropomorphic packing algorithm with strong constraints; 2) Each group expands the virtual container body based on the larger cargo; 3) Each group contains at least 3 cargoes and the total volume of the virtual container body is between 900dm3 and 5000dm3; 4) After each grouping, the cargo occupies more than 90% of the total volume of its corresponding virtual container.

3. The intelligent power grid material distribution planning method based on three-dimensional packaging according to claim 1 is characterized by: The design of the packing operator and the heuristic evolution include: designing a data-driven crossover operator, introducing similar posture replacement and similar goods replacement into the crossover operator, thereby improving the convergence speed of the algorithm; During the algorithm iteration, first, two goods with the same loading position in the two similar solutions to be optimized are found, and the cargo numbers of the two goods are judged to be the same. If the cargo numbers are the same, the posture approximation rule of the goods is introduced. If the cargo posture codes of the two similar solutions are similar, the coding position is classified as the first type of similar coding position. If the cargo numbers are different, the shape information (X1, Y1, Z1) and (X2, Y2, Z2) of the two cargoes are introduced. If |X1-X2| < 10, |Y1-Y2| < 10, |Z1-Z2| < 10, it means that the shapes of the two cargoes are similar. The cargo number coding and cargo posture coding positions of the two similar solutions are the second type of similar coding positions. Then, the first type of similar coding positions are cross-operated according to the cross-probability R1, and the second type of similar coding positions are cross-operated according to the cross-probability R2.

4. The intelligent power grid material distribution planning method based on three-dimensional packaging according to claim 1 is characterized by: The optimization problem of power infrastructure material distribution is described as follows: The vehicle routing optimization problem under three-dimensional packing constraints can be described as follows: Given an undirected graph G = (D, E) representing the distribution network, where D = {D0, D1, D2, ..., D n } is the vertex set, E={(D i ,D j ) / D i ,D j ∈D,0≤i,j≤N,i≠j} is a path set; the vertex D0 represents the distribution center, (D1,D2,...,D n ) represents the customer point set; (D i ,D j ) indicates customer D i and D j The lines between them are connected with d ij Indicates customer D i and Customer D j The distance between ij =d ji The distribution center has k trucks of different types, and the rated load mass and maximum volume of each truck are represented by G k and V k Indicates: The length, width and height of the truck are expressed in L. k ,W k ,H k Indicated by; the length, width, height and mass of each piece of cargo are expressed by l i ,w i ,h i ,g i express.

5. The intelligent power grid material distribution planning method based on three-dimensional packaging according to claim 1 is characterized by: The vehicle routing optimization model assumptions include: (1) There is only one distribution center, and the goods in the distribution center can meet the needs of all customers; (2) The number of vehicles owned by the distribution center meets all distribution requirements; (3) The vehicle used departs from the distribution center and returns to the distribution center after delivery; (4) Each customer's needs are met and served only once; (5) All goods required by customers are known; (6) The locations and distances between customers and between customers and distribution centers are known; (7) The mass and volume of cargo loaded on each vehicle cannot exceed the vehicle's permitted rated load mass and maximum volume constraints; (8) Assume that all cargoes are regular and uniform rectangular parallelepipeds and the length, width and height of each cargo do not exceed the dimensions of the vehicle box; (9) The delivery vehicles are box trucks of the same model; (10) The cargo must be placed parallel to the carriage and only horizontal rotation is allowed. The schematic diagram of the cargo rotation direction in three-dimensional space is shown in the figure below; (11) To ensure the stability of the vehicle, the center of gravity of the cargo must be within the allowable range of the vehicle compartment; (12) Goods delivery does not consider time windows; (13) The placement of goods must comply with the first-in-last-out principle, that is, if customers are on the same service route, the goods required by the first-served customer cannot be blocked or pressed by the goods required by the later-served customer.

6. The intelligent power grid material distribution planning method based on three-dimensional packaging according to claim 1 is characterized by: The constraints include: (1) vehicle load constraints; (2) vehicle volume constraints; (3) rotation direction constraints; (4) cargo stability constraints; (5) vehicle size constraints; (6) delivery constraints; and (7) loading sequence constraints.

7. The intelligent power grid material distribution planning method based on three-dimensional packaging according to claim 1 is characterized by: The vehicle routing optimization genetic algorithm includes: 1) Calculate the fitness of each individual in the population f(r=1,2,3...,M), where M is the size of the population; 2) Calculate the probability of each individual being inherited into the next generation population, that is, the probability of individual r being selected; 3) Calculate the cumulative probability of each individual; 4) Generate a uniformly distributed pseudo-random number e in the interval [0,1]; 5) If e < q(1), select individual 1; otherwise, select individual v; 6) Repeat steps 4) and 5) M times.

Images