Intelligent routing optimization method for electric vehicles in multiple distribution centers based on dynamic fuzzy prediction
Through dynamic fuzzy prediction and improved genetic algorithm, the path scheme of electric vehicles in multi-distribution centers is optimized, and the problems of demand uncertainty and performance limitations of electric vehicles in the multi-distribution centers are solved, thereby improving the distribution efficiency and customer satisfaction , reduces costs.
Patent Information
- Application Number
- CN202411225914.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-03
- Publication Date
- 2024-11-08
- Estimated Expiration
- 2044-09-03
AI Technical Summary
In the multi-distribution center electric vehicle distribution network, due to the uncertainty of customer demand, real-time changes in traffic conditions, and limitations of the electric vehicle's own performance, traditional static path optimization methods are difficult to meet actual needs, resulting in inefficient distribution efficiency and increased costs.
Using the intelligent path optimization method of multi-distribution center electric vehicles based on dynamic fuzzy prediction, by building a distribution system model, combining fuzzy number calculation and improved genetic algorithms, we can capture demand changes in real time, optimize the path plan of electric vehicles, and consider the status and geography of electric vehicles. Position and charging status, and develop the optimal path for collaborative operations.
The distribution efficiency of electric vehicles in multiple distribution centers has been improved, the air driving and waiting time during the distribution process has been reduced, the delivery costs have been reduced, and the customer needs have been better met, and customer satisfaction has been improved.
Smart Images

Figure CN118761534B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of electric vehicle path optimization, and in particular to a multi-distribution center electric vehicle intelligent path optimization method based on dynamic fuzzy prediction. Background Art
[0002] With the increasing global environmental awareness and the rapid development of science and technology, electric vehicles are increasingly used in the field of logistics and distribution. With its low energy consumption and zero emissions, electric vehicles have become an important choice for urban logistics and distribution. However, in actual applications, electric vehicle distribution faces many challenges, such as limited driving range, insufficient charging infrastructure, and dynamic changes in distribution demand. These problems make the scheduling and path optimization of electric vehicles in electric vehicle distribution networks particularly complex.
[0003] At present, the multi-distribution center joint distribution model has gradually become the mainstream trend in the logistics industry. Through the collaborative operation of multiple distribution centers, the operating cost of the logistics distribution network can be effectively reduced and the transportation efficiency can be improved. However, in the multi-distribution center electric vehicle distribution network, due to the uncertainty of customer demand, the real-time changes in traffic conditions and the limitations of the electric vehicle's own performance, the traditional static path optimization method has been difficult to meet the actual needs. Summary of the invention
[0004] The present invention provides a multi-distribution center electric vehicle intelligent path optimization method based on dynamic fuzzy prediction to overcome the above technical problems.
[0005] The intelligent path optimization method for electric vehicles in multiple distribution centers based on dynamic fuzzy prediction includes:
[0006] S1. Construct a distribution system, which includes a set of customer points, a set of distribution centers, a set of electric vehicles, and a set of public charging piles, and design assumptions for the distribution system;
[0007] S2. Constructing the fuzzy demand of the customer points, calculating the remaining load of the electric vehicle when delivering to any customer point according to the fuzzy demand of the customer points, and performing credibility evaluation on the remaining load based on fuzzy number calculation and calculating the credibility;
[0008] S3. Set a decision threshold, construct a first constraint condition according to the credibility and the decision threshold, design an objective function and its corresponding constraint condition set based on the assumptions of the distribution system, wherein the objective function includes the dispatching cost of electric vehicles, the transportation cost, the time window penalty cost, and the charging cost of electric vehicles, and merge the first constraint condition and the constraint condition set corresponding to the objective function as the system constraint condition;
[0009] S4. Improving the genetic algorithm, wherein the improving the genetic algorithm includes generating an initial population based on a double-layer encoding technique, generating a parent population based on a binary tournament selection operator and the initial population, obtaining a child population from the parent population based on a two-point crossover operator and a BreederGA mutation operator, and obtaining a new initial population from the parent population and the child population based on an elite retention operator;
[0010] S5. Set a maximum number of iterations, and solve the objective function within the system constraints based on the improved genetic algorithm. The method of solving the objective function within the system constraints based on the improved genetic algorithm includes iteratively training the improved genetic algorithm and calculating population information, wherein the population information includes the value and fitness of the objective function. When the number of iterations reaches the maximum number of iterations, the optimal population information is obtained, and a distribution plan is generated based on the optimal population information.
[0011] Preferably, S2 includes setting the maximum load capacity of the electric vehicle to Q , for a given customer point i , its fuzzy demand is expressed in the form of fuzzy numbers as ,in , and represent the lower limit, most likely value and upper limit of the fuzzy demand respectively, and , when an electric car has completed the P After the delivery task of customer points, the current total delivery volume It can be obtained from formula (1):
[0012] (1)
[0013] The remaining load capacity of the electric vehicle Maximum load capacity Q Determine by subtracting the current total delivery volume, which is:
[0014] (2)
[0015] The credibility of the remaining load is evaluated according to formula (3):
[0016] (3)
[0017] in, Indicates ambiguous events credibility, .
[0018] Preferably, the objective function is as shown in formula (4):
[0019] (4)
[0020] (5)
[0021] (6)
[0022] (7)
[0023] (8)
[0024] in, Z represents the total cost, S Represents the collection of all nodes, including customer points, distribution centers and public charging piles, represents the fixed dispatch cost of electric vehicles; represents the unit transportation cost of electric vehicles; represents the penalty cost per unit time when the electric vehicle arrives earlier than the lower bound of the service time window of the customer point; represents the penalty cost per unit time when the electric vehicle arrives later than the upper bound of the customer service time window; Represents the unit charging cost of electric vehicles at public charging stations; M Represents a set of customer points; N represents a collection of distribution centers; K Represents a collection of delivery electric vehicles; A Represents a collection of public charging piles; is a 0-1 variable. k Slave Node i Travel to the node j The value is 1 when it is, otherwise it is 0; Representation Node i With Node j The distance between Indicates customer point i The upper limit of the allowed service time window; Indicates customer point i The lower bound of the allowed service time window; Indicates that the electric vehicle arrives at the node i time; Indicates electric vehicle k At public charging stations i of charge.
[0025] Preferably, the system constraints are as shown in equations (9)-(25):
[0026] (9)
[0027] (10)
[0028] (11)
[0029] (12)
[0030] (13)
[0031] (14)
[0032] (15)
[0033] (16)
[0034] (17)
[0035] (18)
[0036] (19)
[0037] (20)
[0038] (twenty one)
[0039] (twenty two)
[0040] (twenty three)
[0041] (twenty four)
[0042] (25)
[0043] Constraint (9) ensures that the reliability of each vehicle used to load customer goods does not exceed the maximum load of electric vehicles is not less than the preset preference value of the decision maker. Constraint (9) is the first constraint. As the decision threshold, represents the credibility; constraint (10) limits the load of the electric vehicle on each path to not exceed its maximum load; constraint (11) ensures that each customer is served by only one vehicle; constraint (12) stipulates that if the delivery path contains a customer point, the customer must be served; constraint (13) represents the time it takes for the electric vehicle to travel from a charging station to any customer point; constraint (14) represents the time it takes for the electric vehicle to travel from the current customer point to the next customer point; constraint (15) indicates that the electric vehicle starts from a distribution center and returns to any distribution center after replenishing goods or completing the task; constraint (16) indicates that the electric vehicle cannot visit the same node repeatedly, that is, to avoid invalid paths; constraint (17) indicates that the electric vehicle The electric vehicle reaches a full charge after each charge; constraint (18) indicates that the power consumption of the electric vehicle is not considered when it is waiting for work at the customer point; constraint (19) indicates the charging time required for the electric vehicle to charge at the charging point; constraint (20) stipulates the charging time required for the electric vehicle to charge at the charging point, and the charging time is positive; constraint (21) stipulates the battery power constraint of the electric vehicle during the distribution process to ensure that the electric vehicle has enough power to complete the tasks of the remaining nodes at any time; constraint (22) indicates the power added by the electric vehicle when charging at the charging pile; constraint (23) ensures that the power of the electric vehicle is positive when it arrives at each customer point; constraints (24)-(25) represent the value range of the decision variables; among them, y ik is a 0-1 variable. k Serving Nodes i The value is 1 when it is, otherwise it is 0; Indicates that the electric vehicle arrives at the node i time; Indicates that the electric vehicle arrives at the node j time, Indicates electric vehicle k At public charging stations i Charging time; Representation Node i With Node j The distance between M Represents a set of customer points; N represents a collection of distribution centers; K Represents a collection of delivery electric vehicles; A Represents a collection of public charging piles; S represents the set of all nodes, including customer points, distribution centers and public charging piles, that is, ; Indicates that electric vehicles are at customer points i service time; Indicates electric vehicle k Leave the node iThe remaining power when Indicates electric vehicle k At public charging stations i The charge capacity; Indicates electric vehicle k Arrival Node i The remaining power; is a 0-1 variable. k Slave Node i Travel to the node j The value is 1 when it is, otherwise it is 0; Q Indicates the maximum load capacity of the electric vehicle; V represents the average speed of electric vehicles; Indicates customer point i The fuzzy demand quantity is expressed as the fuzzy demand number, that is, ; Indicates customer point The upper limit of the allowed service time window; Indicates customer point i The lower bound of the allowed service time window; Indicates the maximum battery capacity of an electric vehicle; B Indicates the battery power consumption rate of electric vehicles, Indicates that when electric vehicles k Slave Node i Travel to the node i When the value is 0, Indicates the charging power.
[0044] Preferably, generating an initial population based on double-layer coding includes:
[0045] S41, set the population size to G, F to the total number of customer points, and H to the maximum number of electric vehicles that can be used;
[0046] S42, designing a two-layer coding strategy, the two-layer coding strategy includes setting the first layer as a customer point delivery priority sequence, which is composed of real numbers in the range of 0 to 1, each real number represents the delivery priority of a customer point, and setting the second layer as an electric vehicle code, which is composed of integers in the range of 0 to (H-1), each integer represents a customer point delivered by an electric vehicle corresponding to the electric vehicle code;
[0047] S43, generating a chromosome according to the double-layer coding strategy, wherein the chromosome is composed of 2F real numbers, the first part includes F real numbers in the range of 0 to 1, and the second part includes F integers in the range of 0 to (H-1), and a set of values is randomly generated according to the value ranges of the two parts to obtain a chromosome;
[0048] S44. Repeat S43 until G chromosomes are generated, that is, a G×2F array, which is the initial population.
[0049] The present invention provides a multi-distribution center electric vehicle intelligent path optimization method based on dynamic fuzzy prediction. By introducing dynamic fuzzy prediction technology, it can capture and analyze the changing trend of these uncertain demands in real time, and provide a basis for dynamic adjustment for path optimization; dynamic fuzzy prediction technology combined with improved genetic algorithm can comprehensively consider the electric vehicle status, remaining load, geographical location and other factors of each distribution center, and formulate the optimal path plan for collaborative operation for multiple distribution centers; improve the distribution efficiency of electric vehicles in multiple distribution centers, reduce the empty driving and waiting time in the distribution process, and thus reduce the distribution cost. At the same time, the optimized path plan can better meet customer needs and improve customer satisfaction. BRIEF DESCRIPTION OF THE DRAWINGS
[0050] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.
[0051] Figure 1 is a flow chart of the method of the present invention;
[0052] Figure 2 It is a flow chart of the improved genetic algorithm of the present invention;
[0053] Figure 3 The encoding and decoding method of the present invention;
[0054] Figure 4 It is the two-point crossover operator of the present invention. DETAILED DESCRIPTION
[0055] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0056] Figure 1 is a flow chart of the method of the present invention, such as Figure 1 As shown, the method of this embodiment may include:
[0057] S1. Construct a distribution system, which includes a set of customer points, a set of distribution centers, a set of electric vehicles, and a set of public charging piles, and design assumptions for the distribution system;
[0058] S2. Constructing the fuzzy demand of the customer points, calculating the remaining load of the electric vehicle when delivering to any customer point according to the fuzzy demand of the customer points, and performing credibility evaluation on the remaining load based on fuzzy number calculation and calculating the credibility;
[0059] S3. Set a decision threshold, construct a first constraint condition according to the credibility and the decision threshold, design an objective function and its corresponding constraint condition set based on the assumptions of the distribution system, wherein the objective function includes the dispatching cost of electric vehicles, the transportation cost, the time window penalty cost, and the charging cost of electric vehicles, and merge the first constraint condition and the constraint condition set corresponding to the objective function as the system constraint condition;
[0060] S4. Improving the genetic algorithm, wherein the improving the genetic algorithm includes generating an initial population based on a double-layer encoding technique, generating a parent population based on a binary tournament selection operator and the initial population, obtaining a child population from the parent population based on a two-point crossover operator and a BreederGA mutation operator, and obtaining a new initial population from the parent population and the child population based on an elite retention operator;
[0061] S5. Set a maximum number of iterations, and solve the objective function within the system constraints based on the improved genetic algorithm. The method of solving the objective function within the system constraints based on the improved genetic algorithm includes iteratively training the improved genetic algorithm and calculating population information, wherein the population information includes the value and fitness of the objective function. When the number of iterations reaches the maximum number of iterations, the optimal population information is obtained, and a distribution plan is generated based on the optimal population information.
[0062] Based on the above scheme, the present invention can capture and analyze the changing trends of these uncertain demands in real time by introducing dynamic fuzzy prediction technology, and provide a basis for dynamic adjustment for route optimization; dynamic fuzzy prediction technology combined with improved genetic algorithm can comprehensively consider the status of electric vehicles, remaining load, geographical location and other factors of each distribution center, and formulate the optimal route plan for collaborative operation for multiple distribution centers; improve the distribution efficiency of electric vehicles in multiple distribution centers, reduce the empty driving and waiting time during the distribution process, and thus reduce the distribution cost. At the same time, the optimized route plan can better meet customer needs and improve customer satisfaction.
[0063] Specifically, this embodiment provides the details of the intelligent path optimization method for electric vehicles, including:
[0064] 1. Establish a multi-distribution center electric vehicle intelligent path optimization model based on dynamic fuzzy prediction, which mainly includes determining symbols and variables, constructing a dynamic fuzzy demand prediction function, and building a model:
[0065] (1) Symbols and variables:
[0066] 1) Collection:
[0067] M Represents a set of customer points;
[0068] N represents a collection of distribution centers;
[0069] K Represents a collection of delivery electric vehicles;
[0070] A Represents a collection of public charging piles;
[0071] S represents the set of all nodes, including customer points, distribution centers and public charging piles, that is, ;
[0072] 2) Parameters:
[0073] Indicates the maximum load capacity of the electric vehicle;
[0074] represents the fixed dispatch cost of electric vehicles;
[0075] represents the unit transportation cost of electric vehicles;
[0076] represents the penalty cost per unit time when the electric vehicle arrives earlier than the lower bound of the service time window of the customer point;
[0077] represents the penalty cost per unit time when the electric vehicle arrives later than the upper bound of the customer service time window;
[0078] Represents the unit charging cost of electric vehicles at public charging stations;
[0079] Representation Node i With Node j The distance between
[0080] Indicates customer point i The fuzzy demand quantity is expressed as the fuzzy demand number, that is, ;
[0081] represents the average speed of electric vehicles;
[0082] Indicates customer point i The upper limit of the allowed service time window;
[0083] Indicates customer point i The lower bound of the allowed service time window;
[0084] Indicates that electric vehicles are at customer points i service time;
[0085] Indicates the maximum battery capacity of an electric vehicle;
[0086] Indicates the battery power consumption rate of electric vehicles;
[0087] 3) Decision variables:
[0088] is a 0-1 variable. k Slave Node i Travel to the node j The value is 1 when it is, otherwise it is 0;
[0089] is a 0-1 variable. k Serving Nodes i The value is 1 when it is, otherwise it is 0;
[0090] Indicates that the electric vehicle arrives at the node i time;
[0091] Indicates electric vehicle k At public charging stations i Charging time;
[0092] Indicates electric vehicle k At public charging stations i The charge capacity;
[0093] Indicates electric vehicle k Arrival Node i The remaining power;
[0094] Indicates electric vehicle k Leave the node i The remaining power when
[0095] Indicates that when electric vehicles k Slave Node i Travel to the node i When the value is 0;
[0096] Indicates the charging power.
[0097] (2) Constructing dynamic fuzzy demand prediction function:
[0098] In the logistics distribution system, it is assumed that the maximum load of the distribution electric vehicle is Q For a given customer point, its fuzzy demand is expressed in the form of fuzzy numbers as ,in , and represent the lower limit, most likely value and upper limit of demand respectively, and When a delivery electric vehicle has completed the P After the delivery task of customer points, the current total delivery volume It can be obtained by the following formula:
[0099] (1)
[0100] In function (1) It is also a triangular fuzzy number, reflecting the uncertainty of the cumulative delivery volume. At this time, the remaining load capacity of the delivery electric vehicle is Maximum load capacity Q Determine by subtracting the current total delivery volume, which is:
[0101] (2)
[0102] Obviously, the remaining load is also a triangular fuzzy number, which represents the fuzziness of the remaining carrying capacity. According to the fuzzy credibility theory, when considering the next customer point to be served, P +1, it is crucial to assess the credibility of its demand not exceeding the remaining load of the current electric vehicle. This credibility can be obtained through the operation and comparison of fuzzy numbers:
[0103] (3)
[0104] in, Indicates ambiguous events The credibility measure, Obviously, The larger the value, the more successful the delivery electric vehicle is in serving the customer. P In order to deal with the actual situation more flexibly, the present invention introduces the decision maker preference value As the decision threshold. Specifically, if the credibility , it is determined that the delivery electric vehicle can continue to serve customers P +1 service; on the contrary, if the credibility , it is considered a service failure. At this time, the electric vehicle needs to choose the nearest distribution center to reload the goods, and then go to the "failed" customer point to complete the delivery task;
[0105] (3) Model building:
[0106] (4)
[0107] (5)
[0108] (6)
[0109] (7)
[0110] (8)
[0111] Constraints:
[0112] (9)
[0113] (10)
[0114] (11)
[0115] (12)
[0116] (13)
[0117] (14)
[0118] (15)
[0119] (16)
[0120] (17)
[0121] (18)
[0122] (19)
[0123] (20)
[0124] (twenty one)
[0125] (twenty two)
[0126] (twenty three)
[0127] (twenty four)
[0128] (25)
[0129] The objective function (4) aims to minimize the total cost of the entire distribution system Z , which consists of four parts: electric vehicle dispatch cost (5), transportation cost (6), time window penalty cost (7), and electric vehicle charging cost (8). Constraint function (9) ensures that the credibility of each vehicle used to load customer goods does not exceed the maximum load of the electric vehicle is not less than the preset preference value of the decision maker. Constraint (9) is the first constraint. As the decision threshold, represents credibility; constraint function (10) limits the load of electric vehicles on each path to not exceed its maximum load; constraint function (11) ensures that each customer is served by only one vehicle; constraint function (12) stipulates that if the delivery path contains a customer point, the customer must be served; constraint function (13) represents the time it takes for an electric vehicle to travel from a charging station to any customer point; constraint function (14) represents the time it takes for an electric vehicle to travel from the current customer point to the next customer point; constraint function (15) represents that an electric vehicle starts from a distribution center and returns to any distribution center after replenishing goods or completing the task; constraint function (16) represents that an electric vehicle cannot visit the same node repeatedly, that is, to avoid invalid paths; constraint function (17) represents that the electric vehicle The electric vehicle reaches a full charge after each charge; constraint function (18) indicates that the power consumption of the electric vehicle is not considered when it is waiting for work at the customer point; constraint function (19) indicates the charging time required for the electric vehicle to charge at the charging point; constraint function (20) stipulates the charging time required for the electric vehicle to charge at the charging point, and adjusts the charging time to be positive; constraint function (21) stipulates the battery power constraint of the electric vehicle during the distribution process to ensure that the electric vehicle has enough power to complete the tasks of the remaining nodes at any time; constraint function (22) indicates the power added by the electric vehicle when charging at the charging pile; constraint function (23) ensures that the power of the electric vehicle is positive when it arrives at each customer point; constraint functions (24)-(25) represent the value range of the decision variables.
[0130] 2. Design of intelligent path optimization algorithm for electric vehicles:
[0131] The present invention is aimed at the optimization of electric vehicle routes in multiple distribution centers in NP-hard problems, especially when the model integrates fuzzy predictions to cope with dynamic demand changes. Traditional solvers such as CPLEX, Lingo, etc. can provide high-quality solutions in small-scale examples, but when faced with large-scale data, their computational complexity and time consumption increase significantly, limiting their practicality in practical applications. In order to break through this bottleneck, the present invention proposes an improved genetic algorithm, which not only inherits the powerful ability of genetic algorithms in global optimization, but also improves the performance of genetic algorithms through double-layer coding technology, binary tournament selection operators, two-point crossover operators, BreederGA mutation operators, and elite retention operators, effectively coping with the complexity and diversity of the problem. This improvement aims to improve the solution efficiency, ensure rapid response and decision optimization in a dynamic logistics environment, while maintaining the solution quality, and achieve multiple iterations in a limited time through heuristic search strategies to more efficiently approach the optimal path solution. The improved genetic algorithm flow chart is as follows Figure 2 shown.
[0132] Step 1: Chromosome encoding and decoding:
[0133] The present invention adopts a two-layer coding strategy. The first layer is the customer point delivery priority sequence, which is composed of real numbers ranging from 0 to 1. Each real number represents the delivery priority of a customer point. The larger the value, the higher the priority of the customer point. When decoding, the delivery order of the customer point can be obtained by sorting these priority numbers from large to small. The second layer is the electric vehicle code, which is composed of a series of integers ranging from 0 to , is the number of electric vehicles, which is used to specify the set of customer points served by each vehicle.
[0134] Combining these two layers of coding, we can clearly know which customer points each vehicle is responsible for and the order of delivery to these customer points. In addition, considering the existence of multiple distribution centers and the need for electric vehicles to be charged during the distribution process, the present invention adopts a series of intelligent decision-making mechanisms. First, based on the principle of shortest distance, each vehicle starts from the distribution center closest to its first customer point, and tries to return to the distribution center closest to the last customer point after completing the delivery. Then, the triangular fuzzy theory is used to determine whether the electric vehicle needs to return to the nearest distribution center to pick up the goods on the way. Finally, when the electric vehicle is low on power, the nearest charging pile will be selected for charging based on the principle of shortest distance.
[0135] Let's take a specific example: Suppose there are 5 customer points , 2 distribution centers , 2 charging stations And 2 delivery electric vehicles. Figure 3Represents 1 chromosome, and the delivery priority of the customer point has been encoded as a real number, that is, the delivery priorities of customer points c0, c1, c2, c3, and c4 are 0.12, 0.51, 0.43, 0.67, and 0.75 respectively.
[0136] First, the delivery priorities of the customer points are arranged in descending order, and the order of delivery is determined as: c4, c3, c1, c2, c0. According to the allocation of the second-level code, electric vehicle 0 is assigned to deliver c4 and c0 in turn, while electric vehicle 1 is responsible for the delivery tasks of c3, c1 and c2.
[0137] Then, based on the principle of shortest distance, the nearest distribution center is assigned to electric vehicle 0 and electric vehicle 1. Assuming that the distribution centers closest to customer points c4, c3, c2, c1, and c0 are d0, d1, d1, d0, and d1 respectively, the distribution path of electric vehicle 0 is: d0→c4→c0→d1; the distribution path of electric vehicle 1 is: d1→c3→c1→c2→d1.
[0138] Secondly, considering the ambiguity of customer demand, when an electric vehicle arrives at a certain customer point, there may be a situation where the remaining cargo of the electric vehicle is insufficient. At this time, the electric vehicle needs to return to the distribution center closest to the current customer point to replenish the goods, and then return to the customer point to continue the delivery task. Assuming that electric vehicle 1 finds that the remaining cargo is insufficient when it arrives at customer point c1, and c1 is closest to d0, then the delivery path of electric vehicle 1 needs to be corrected to: d1→c3→c1→d0→c1→c2→d1.
[0139] Finally, by calculating the remaining power of each vehicle and the expected power consumption, the charging requirements of electric vehicles on each delivery route are determined. Based on the principle of shortest distance, the nearest charging pile is selected for the electric vehicle that needs to be charged. Assuming that vehicle 1 needs to be charged before serving customer point c2, and the charging pile closest to customer point c2 is f1, then the final delivery path of vehicle 1 is: d1→c3→c1→d0 (replenishment)→c1→f1 (charging)→c2→d1.
[0140] Step 2: Initial population generation:
[0141] According to the double-layer coding rule, the initial population is randomly generated. Assume that the population size is G, F is the total number of customer points, and H is the maximum number of electric vehicles that can be used. According to the coding rule, a chromosome is composed of 2F real numbers. The first part contains F real numbers in the range of 0~1, and the second part contains F integers in the range of 0~(H-1). A set of values is randomly generated according to the value range of these two parts to obtain a complete chromosome. Repeat this process G times to construct a G×2F array, which is the initial population.
[0142] Step 3, fitness function:
[0143] In genetic algorithms, the fitness function is used to evaluate the quality of each individual in the population. It assigns a fitness value to each individual based on the specific requirements of the problem, which directly reflects the performance of the individual in solving a specific problem. The higher the fitness value, the closer the individual is to the optimal solution of the problem; otherwise, the farther away.
[0144] For the algorithm of the present invention, the fitness function adjusts the original fitness value through linear scaling to improve the search efficiency and convergence of the algorithm. Specifically, the fitness function is as follows:
[0145] (26)
[0146] In formula (26), is the fitness after linear scaling, fit is the original fitness, , Z is the objective function, a and b is the transformation coefficient, and are the average and maximum values of the fitness of the contemporary population, c is the expected number of copies of the individual with the maximum fitness.
[0147] Step 4. Binary Tournament Selection Operator:
[0148] In each selection operation, the binary tournament operator randomly selects two individuals from the population to compare their fitness values, and retains the individuals with higher fitness to enter the next generation. This competition mechanism not only promotes the diversity of the population, effectively prevents the algorithm from falling into the local optimum, but also promotes the continuous evolution of the population towards a better solution. In addition, the operator also cleverly designs a certain probability to retain individuals with slightly lower fitness, thereby expanding the search range and enhancing the algorithm's ability to find the global optimal solution in a larger solution space.
[0149] Step 5: Crossover operator:
[0150] In the evolutionary framework of the genetic algorithm, the present invention selects two-point crossover as the key operation mode of the genetic algorithm, aiming to fuse specific gene fragments of two parent chromosomes with a certain probability in order to generate offspring individuals with higher fitness. An embodiment of the two-point crossover operator is as follows:
[0151] First, we still assume that there are 5 customer points and 2 cars, and thus determine that the length of the chromosome is 10. For the initial population containing 4 parent individuals, Figure 4As shown. In order to perform the crossover operation in an orderly manner and maintain population diversity, the population is evenly divided into two sub-populations: the first half of the population and the second half of the population. This grouping strategy not only simplifies the management of the crossover operation, but also promotes the maintenance of the genetic diversity of the population, laying the foundation for the algorithm to explore a better solution space.
[0152] Subsequently, for each pair of parent chromosomes in the two subgroups, a two-point crossover operation is performed according to the preset crossover probability. This strategy ensures that the high-quality genetic information in the parent chromosomes can be effectively passed to the offspring, thereby enhancing the fitness potential of the offspring individuals.
[0153] Step 6: Mutation operator:
[0154] The Breeder GA mutation operator, as an efficient real number mutation strategy in genetic algorithms, has the core function of achieving local search and optimization by finely regulating the gene values on chromosomes, thereby enhancing the diversity of the population and effectively preventing the algorithm from falling into the trap of local optimality during the solution process.
[0155] In the mutation operation, first, define and set key parameters such as mutation rate, gene value range (including upper and lower limits), and fixed constants. For each gene value on the chromosome, randomly generate a value between 0 and 1 to compare with the mutation rate to determine whether to mutate the gene. Compare the generated random number with the mutation rate. If the random number is less than the mutation rate, perform a mutation operation on the gene value to generate a new gene value and generate a new chromosome based on it; otherwise, keep the gene value unchanged.
[0156] Then, for the gene value that is decided to be mutated, the new gene value is calculated according to formula (27) based on the specific mutation formula:
[0157] (27)
[0158] In formula (27) and are the gene values before and after mutation, respectively. L is the gene value The difference between the upper and lower limits, and is a fixed constant, by The probability of taking is 1, The probability of is 0, Generate a random value between 0 and 1.
[0159] Finally, the mutated gene value replaces the original gene value to generate a new chromosome. This process is repeated on each gene value on the chromosome until the mutation operation on the entire chromosome is completed.
[0160] Step 7: Elite selection operator:
[0161] In order to ensure the continuous inheritance of excellent individuals, it is not enough to rely solely on the binary tournament selection operator. Therefore, the present invention also introduces an elite retention operator as a supplement. The core of this operator is to copy the best performing individuals in each generation directly to the next generation to ensure that these elite individuals are not eliminated during the evolution process. This strategy not only makes full use of known excellent solutions and reduces repeated searches, but also significantly accelerates the convergence process of the algorithm and improves the optimization effect of the algorithm within a limited number of iterations. The present invention integrates two key selection operators, binary tournament and elite retention. The synergistic effect of the two ensures the efficiency and stability of the search process and enhances the optimization performance of the genetic algorithm.
[0162] Step 8. Algorithm termination criteria:
[0163] If the number of iterations reaches the maximum number of iterations maxGen , then stop the iteration and output the operation result.
[0164] Through the above content, the technical problems solved by this embodiment include: Dynamic processing of demand uncertainty: In the process of logistics distribution, customer demand often presents a high degree of dynamics and uncertainty, such as real-time changes in order volume, insertion of urgent orders, customer cancellation or change of orders, etc. Traditional static path planning methods are difficult to cope with such real-time changing demands, resulting in low distribution efficiency and increased costs. By introducing dynamic fuzzy prediction technology, it is possible to capture and analyze the changing trends of these uncertain demands in real time, providing a basis for dynamic adjustment for path optimization. Path optimization for collaborative operation of multiple distribution centers: In a logistics network served by multiple distribution centers, how to coordinate resource allocation and path planning between centers to minimize distribution costs and meet customer needs is a complex optimization problem. Dynamic fuzzy prediction technology combined with intelligent optimization algorithms can comprehensively consider factors such as the status of electric vehicles, remaining load, geographical location, and real-time traffic conditions of each distribution center, and formulate the optimal path plan for collaborative operation for multiple distribution centers. Real-time path adjustment and dynamic optimization: In the face of demand uncertainty, traditional path planning solutions are difficult to flexibly adjust once they are determined, and it is difficult to meet real-time changing distribution needs. The intelligent path optimization method proposed in this application can adjust the delivery path of electric vehicles in real time based on the results of dynamic fuzzy prediction to cope with emergencies and changes in demand, and ensure the smooth completion of delivery tasks. Improve delivery efficiency and reduce costs: Through the combination of dynamic fuzzy prediction and intelligent optimization algorithms, this application aims to improve the delivery efficiency of electric vehicles in multiple distribution centers, reduce empty driving and waiting time during the delivery process, and thus reduce delivery costs. At the same time, the optimized path plan can better meet customer needs and improve customer satisfaction.
[0165] Overall beneficial effects:
[0166] The present invention can significantly improve the response speed and flexibility of the logistics distribution system, and effectively respond to challenges such as demand uncertainty and electric vehicle range limitations. Through the combination of dynamic fuzzy prediction and fuzzy opportunity constraint programming model, real-time optimization and intelligent adjustment of distribution paths are achieved, which significantly reduces the total distribution cost and improves distribution efficiency. At the same time, the method promotes collaborative operations among multiple distribution centers and enhances the overall efficiency of the logistics network. Its application not only brings significant economic benefits to logistics companies, but also promotes the development of green logistics, and provides strong support for the intelligent and sustainable urban logistics distribution.
[0167] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.
Claims
1. A multi-distribution center electric vehicle intelligent path optimization method based on dynamic fuzzy prediction, characterized in that: include: S1. Construct a distribution system, which includes a set of customer points, a set of distribution centers, a set of electric vehicles, and a set of public charging piles, and design assumptions for the distribution system; S2. Constructing the fuzzy demand of the customer points, calculating the remaining load of the electric vehicle when delivering to any customer point according to the fuzzy demand of the customer points, and performing credibility evaluation on the remaining load based on fuzzy number calculation and calculating the credibility; S3. Set a decision threshold, construct a first constraint condition according to the credibility and the decision threshold, design an objective function and its corresponding constraint condition set based on the assumptions of the distribution system, wherein the objective function includes the dispatching cost of electric vehicles, the transportation cost, the time window penalty cost, and the charging cost of electric vehicles, and merge the first constraint condition and the constraint condition set corresponding to the objective function as the system constraint condition; S4. Improving the genetic algorithm, wherein the improving the genetic algorithm includes generating an initial population based on a double-layer encoding technique, generating a parent population based on a binary tournament selection operator and the initial population, obtaining a child population from the parent population based on a two-point crossover operator and a BreederGA mutation operator, and obtaining a new initial population from the parent population and the child population based on an elite retention operator; S5. Set a maximum number of iterations, and solve the objective function within the system constraints based on the improved genetic algorithm. The method of solving the objective function within the system constraints based on the improved genetic algorithm includes iteratively training the improved genetic algorithm and calculating population information, wherein the population information includes the value and fitness of the objective function. When the number of iterations reaches the maximum number of iterations, the optimal population information is obtained, and a distribution plan is generated based on the optimal population information.
2. The method for intelligent path optimization of electric vehicles for multiple distribution centers based on dynamic fuzzy prediction according to claim 1 is characterized in that: The S2 includes setting the maximum load capacity of the electric vehicle to Q , for a given customer point i , its fuzzy demand is expressed in the form of fuzzy numbers as ,in , and represent the lower limit, most likely value and upper limit of the fuzzy demand respectively, and , when an electric car has completed the P After the delivery task of customer points, the current total delivery volume It can be obtained from formula (1): (1) The remaining load capacity of the electric vehicle Maximum load capacity Q Determine by subtracting the current total delivery volume, which is: (2) The credibility of the remaining load is evaluated according to formula (3): (3) in, Indicates ambiguous events credibility, .
3. The method for intelligent path optimization of electric vehicles for multiple distribution centers based on dynamic fuzzy prediction according to claim 1 is characterized in that: The objective function is shown in formula (4): (4) (5) (6) (7) (8) in, Z represents the total cost, S Represents the collection of all nodes, including customer points, distribution centers and public charging piles, represents the fixed dispatch cost of electric vehicles; represents the unit transportation cost of electric vehicles; represents the penalty cost per unit time when the electric vehicle arrives earlier than the lower bound of the service time window of the customer point; represents the penalty cost per unit time when the electric vehicle arrives later than the upper bound of the customer service time window; Represents the unit charging cost of electric vehicles at public charging stations; M Represents a set of customer points; N represents a collection of distribution centers; K Represents a collection of delivery electric vehicles; A Represents a collection of public charging piles; is a 0-1 variable. k Slave Node i Travel to the node j The value is 1 when it is, otherwise it is 0; Representation Node i With Node j The distance between Indicates customer point i The upper limit of the allowed service time window; Indicates customer point i The lower bound of the allowed service time window; Indicates that the electric vehicle arrives at the node i time; Indicates electric vehicle k At public charging stations i of charge.
4. The method for intelligent path optimization of electric vehicles for multiple distribution centers based on dynamic fuzzy prediction according to claim 1 is characterized in that: The system constraints are shown in equations (9)-(25): (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (21) (22) (23) (24) (25) Constraint (9) ensures that the reliability of each vehicle used to load customer goods does not exceed the maximum load of electric vehicles is not less than the preset preference value of the decision maker. Constraint (9) is the first constraint. As the decision threshold, represents the credibility; constraint (10) limits the load of the electric vehicle on each path to not exceed its maximum load; Constraint (11) ensures that each customer is served by only one vehicle; Constraint (12) stipulates that if the delivery route includes a customer point, the customer must be served; Constraint (13) represents the time it takes for an electric vehicle to travel from a charging station to any customer point; Constraint (14) represents the time it takes for an electric vehicle to travel from the current customer point to the next customer point; Constraint (15) represents that an electric vehicle starts from a distribution center and returns to any distribution center after replenishing goods or completing a task; Constraint (16) represents that an electric vehicle cannot visit the same node repeatedly, thus avoiding invalid paths; Constraint (17) represents that the electric vehicle is fully charged after each charge; Constraint (18) represents that when an electric vehicle is waiting for work at a customer point, power consumption is not considered; Constraint (19) represents the charging time required for an electric vehicle to charge at a charging point; Constraint (20) specifies the charging time required for an electric vehicle to charge at a charging point, and all charging times are positive; Constraint (21) specifies the battery power constraints of the electric vehicle during the delivery process, ensuring that the electric vehicle has enough power to complete the tasks of the remaining nodes at any time; Constraint (22) represents the amount of electricity that the electric vehicle replenishes when charging at the charging station; constraint (23) ensures that the amount of electricity that the electric vehicle has when it arrives at each customer point is positive; constraints (24)-(25) represent the value range of the decision variables; where, is a 0-1 variable. k Serving Nodes i The value is 1 when it is, otherwise it is 0; Indicates that the electric vehicle arrives at the node i time; Indicates that the electric vehicle arrives at the node j time, Indicates electric vehicle k At public charging stations i Charging time; Representation Node i With Node j The distance between M Represents a set of customer points; N represents a distribution center set; K Represents a collection of delivery electric vehicles; A Represents a collection of public charging piles; S represents the set of all nodes, including customer points, distribution centers and public charging piles, that is, ; Indicates that electric vehicles are at customer points i service time; Indicates electric vehicle k Leave the node i The remaining power when Indicates electric vehicle k At public charging stations i The charge capacity; Indicates electric vehicle k Arrival Node i The remaining power; is a 0-1 variable. k Slave Node i Travel to the node j The value is 1 when it is, otherwise it is 0; Q Indicates the maximum load capacity of the electric vehicle; V represents the average speed of electric vehicles; Indicates customer point i The fuzzy demand quantity is expressed as the fuzzy demand number, that is, ; Indicates customer point i The upper limit of the allowed service time window; Indicates customer point i The lower bound of the allowed service time window; Indicates the maximum battery capacity of an electric vehicle; B Indicates the battery power consumption rate of electric vehicles, Indicates that when electric vehicles k Slave Node i Travel to the node i When the value is 0, Indicates the charging power.
5. The method for intelligent path optimization of electric vehicles for multiple distribution centers based on dynamic fuzzy prediction according to claim 1 is characterized in that: The generating of the initial population based on the double-layer coding comprises: S41, set the population size to G, F to the total number of customer points, and H to the maximum number of electric vehicles that can be used; S42, designing a two-layer coding strategy, the two-layer coding strategy includes setting the first layer as a customer point delivery priority sequence, which is composed of real numbers in the range of 0 to 1, each real number represents the delivery priority of a customer point, and setting the second layer as an electric vehicle code, which is composed of integers in the range of 0 to (H-1), each integer represents a customer point delivered by an electric vehicle corresponding to the electric vehicle code; S43, generating a chromosome according to the double-layer coding strategy, wherein the chromosome is composed of 2F real numbers, the first part includes F real numbers in the range of 0 to 1, and the second part includes F integers in the range of 0 to (H-1), and a set of values is randomly generated according to the value ranges of the two parts to obtain a chromosome; S44. Repeat S43 until G chromosomes are generated, that is, a G×2F array, which is the initial population.
