Multi-objective optimization method for improving double-branch delimitation

Abstract

The invention discloses a multi-objective optimization method for improving double-branch delimitation of service strategy generation and task intelligent analysis scheduling, and the method mainly comprises the following steps: extracting customer information, carrying out the multi-dimensional classification of tobacco marketing service specialists and terminal customers, and integrating service task features, conventional visit plans, time, regions and other elements; service strategy information is obtained, and a logistics undirected connected graph is constructed for delivery points and customer distribution points according to information such as customer service orders; according to the characteristics of the double-branch delimitation model, in combination with constraint condition rules and resources, an objective function and constraint conditions are determined, and a double-branch delimitation-multi-objective optimization algorithm is designed for solving. With the adoption of the method, a large number of non-optimal solutions can be quickly eliminated in the double-branch and bound algorithm while the customer satisfaction is considered, so that the search space is reduced, the scheduling of service strategy tasks is intelligently analyzed, and the solving efficiency is improved. And optimization is carried out aiming at different targets, and a balanced solution is obtained among the multiple targets.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to an improved double-branch-and-bound multi-objective optimization method applied to tobacco logistics service strategy generation and task intelligent analysis and scheduling. Background Art

[0002] With the rapid development of the domestic economy, especially in the field of tobacco logistics, how to efficiently generate service strategies and reasonably schedule tasks to accurately meet customer service needs, improve customer satisfaction and work efficiency, and reduce service costs is an important research topic. With the continuous changes in market demand and the intensification of competition, traditional service strategy generation and task scheduling methods, such as empirical scheduling and heuristic algorithms, can provide enterprises with feasible solutions in most cases. However, when faced with complex multi-objective optimization problems, it is difficult to meet the increasingly complex transportation needs. When dealing with large-scale data and complex constraints, traditional methods are prone to fall into local optimal solutions and it is difficult to find global optimal or approximate optimal solutions, resulting in low service efficiency, high costs and waste of resources.

[0003] In order to realize the automatic generation of service strategies, accurately meet the service needs of customers and provide customer satisfaction. In order to solve the problems that the information transmission of service needs is not timely and accurate, the tasks of customer specialists are not clear, and the service strategies are not clear, the existing methods usually only focus on a single goal, such as customer satisfaction or logistics costs, and fail to consider the optimization of multiple goals at the same time. In addition, the information transmission of service needs is not timely and accurate, the task allocation of customer specialists is not clear, and the service strategies are not clear. These problems also seriously affect the service quality. Therefore, an optimization method that can comprehensively consider multiple goals and effectively solve them is needed. Summary of the invention

[0004] The technical problem to be solved by the embodiments of the present invention is to provide an improved double-branch-and-bounded multi-objective optimization method for service strategy generation and task intelligent analysis and scheduling, which can effectively narrow the search space and improve the convergence speed of the algorithm while focusing on improving customer satisfaction and reducing the economic cost of the service process, and obtain the optimal solution set of the intelligent scheduling solution.

[0005] In order to solve the above technical problems, an embodiment of the present invention provides a multi-objective optimization method of improved double branch and bound for service strategy generation and task intelligent analysis and scheduling, comprising the following steps: S1: Obtain multi-dimensional information from the system database, construct a distribution logistics network, and establish a bi-objective branch-and-bound algorithm that comprehensively considers customer satisfaction and service strategy costs; S2: Determine the objective function of the dual-objective branch-and-bound algorithm, which includes two sub-algorithms: one is a customer satisfaction branch-and-bound sub-algorithm with maximizing the customer satisfaction of the service strategy as the objective function; the other is a service cost branch-and-bound sub-algorithm with minimizing the cost of the order delivery link as the objective function; S3: Using the NSGA-II algorithm and combining with resource constraints, an improved dual branch and bound-multi-objective genetic optimization algorithm model is constructed, and the dual-objective branch and bound algorithm is combined with the multi-objective optimization algorithm for optimization and solution.

[0006] The multi-dimensional information obtained in S1 includes obtaining at least one of the dimensional information of customer characteristics, level, service list, service subscription, and service specialist capabilities.

[0007] Among them, the construction of the distribution logistics network described in S1 includes the steps of: extracting distribution information based on customer information and customer entrusted orders, and obtaining the customer's specific delivery address information from the database; combining map data to obtain relevant information between each node in the distribution network, including distance and transportation time, and then constructing a logistics undirected connected graph, and calculating the transportation distance, total time and total transportation cost of a single logistics vehicle in logistics transportation.

[0008] The comprehensive objective function of the customer satisfaction branch and bound sub-algorithm in S2 is:

[0009] The above formula represents the objective function of the customer satisfaction branch and bound sub-algorithm to maximize customer satisfaction. Indicates customer satisfaction, Indicates the work efficiency of service specialists. Indicates resource utilization, represents the fairness and matching degree of task allocation, E is the penalty function of the customer service process, , , , , Respectively represent the weights of their respective sub-goals;

[0010] The above formula represents the customer satisfaction objective function, O represents the set of customer service strategy activities, , Represents customer satisfaction feedback, and calculates customer ratings based on historical data and simulated service results. Represents the service response time, which is the time difference from the time the customer requests the service to the time the service starts. Indicates the quality of service completion. Based on historical data and the completion quality of different simulated service strategies, the customer satisfaction score after the service is completed is calculated. Indicates customer loyalty and calculates the customer repurchase rate based on historical data and simulated service results. , , , Respectively represent the specific weight of each indicator;

[0011] The above formula represents the service specialist efficiency objective function, O represents the set of customer service strategy activities, , represents the task completion time of service specialist z in service activity i, represents the number of tasks completed by service specialist z in service activity i, represents the task complexity of service specialist z in performing service activity i, , , Respectively represent the specific weight of each indicator;

[0012] The above formula represents the resource utilization objective function, O represents the set of customer service strategy activities, , It represents the total time that resource j occupies in task i within a certain period of time. represents the idle time of resource j in task i, Indicates specific weight;

[0013] The above formula represents the objective function of task allocation, D represents the set of service specialists, , represents the maximum value of task load among all service specialists, represents the minimum value of task load among all service specialists, Indicates that the task of service specialist g meets the requirements. Indicates specific weight;

[0014] The above formula represents the penalty function of the customer service process. P(s) is a large penalty coefficient used to satisfy the service constraints. is the service start time of task i, represents the task completion time of service specialist z in service activity i, is the latest allowed time for service completion.

[0015] The comprehensive objective function of the service cost branch and bound sub-algorithm in S2 is:

[0016] The above formula represents the objective function of the branch-and-bound algorithm for logistics cost, which minimizes the time cost and expense cost of serving customer order delivery. Indicates the time cost, Represents the delivery logistics cost. , Respectively represent the weights of their respective sub-goals;

[0017] The above formula represents the time cost objective function of transportation and distribution scheduling, A represents the transportation and distribution task information, , E represents the node set of the distribution logistics network, ,, represents the required distance between the service delivery customer nodes i and j for the transportation delivery task a, P(t) represents a large penalty coefficient to ensure that the time window constraint is met, represents the starting time of the delivery task a from the service customer node r, represents the transportation time from service delivery customer node i to transportation service node j for transportation delivery task a, Indicates the latest delivery time allowed for task a;

[0018] The above formula represents the objective function of transportation and distribution cost, where Y represents the set of dispatchable vehicle information. , represents the yth logistics vehicle, k is a weight coefficient used to adjust the impact of loading waste in the total cost, represents the loading rate of delivery vehicle y, is the fixed cost of starting vehicle y, represents the transportation cost between customer nodes i and j served by delivery task a, P(m) is a large penalty coefficient used to penalize solutions that exceed the load capacity, P(v) is a large penalty coefficient used to penalize solutions that exceed the volume limit, U represents the delivery task set of the current delivery logistics vehicle, , represents the cargo weight of the delivery task u that vehicle y is responsible for, represents the cargo volume of the delivery task u that vehicle y is responsible for, Indicates that the total weight of the delivery task of vehicle y does not exceed the maximum load of the vehicle; It means that the total volume of the delivery task of vehicle y does not exceed the maximum volume of the vehicle.

[0019] Among them, the constraints of the service cost branch and bound sub-algorithm are: Considering the vehicle cargo capacity limit and the specific requirements of the order delivery task, the encoding scheme needs to meet the following constraints; The total weight of each vehicle’s delivery mission shall not exceed the vehicle’s maximum load capacity Mmax; The total volume of each vehicle’s delivery mission shall not exceed the vehicle’s maximum volume, Vomax; The execution time of each delivery task must be within the time window required by the customer; The weight of each newly added delivery task shall not exceed the vehicle's current loadable weight Mcurrent; The volume of each newly added delivery task shall not exceed the current loading volume of the vehicle Vocurrent; Constraints for pruning rules, define the following variables; ; For the load constraint, the delivery logistics vehicle y and its delivery task set U ensure that the load of the delivery task u will not exceed the maximum load of vehicle y if , then prune; ; For the volume constraint, the distribution logistics vehicle y and its distribution task set U ensure that the volume of the distribution task u does not exceed the maximum volume of vehicle y if , then prune; ; For time window constraints, ensure that they are completed within the time window [Ei, Li], where Ei is the earliest start time and Li is the latest completion time. If , then prune; ; For each logistics vehicle z and newly added task order i, ensure that the weight of the current task order i does not exceed the current loadable weight of the assigned vehicle if , then prune; ; For each logistics vehicle z and newly added task order i, ensure that the volume of the current task order i does not exceed the current loading capacity of the assigned vehicle if , then prune.

[0020] The dual-objective branch and bound algorithm model in S2 is solved using a dual-branch and bound-multi-objective genetic optimization algorithm.

[0021] The dual branch and bound multi-objective genetic optimization algorithm comprises the following steps: S21, encoding, in the dual-objective branch-and-bound algorithm, the customer satisfaction branch-and-bound sub-algorithm and the service cost branch-and-bound sub-algorithm both use real number encoding, where some chromosome gene fragments of the genetic algorithm represent different dimensional information, including service strategy selection, service specialist allocation, service task sequence, logistics distribution sequence and vehicle scheduling allocation; S22, initialize the population, first generate a service task scheduling priority list based on the service task time window, task dependency and priority and other logical relationships, and generate a service specialist allocation list based on the service specialist's resource allocation and service specialist's work ability; at the same time, generate a logistics distribution task scheduling priority list based on the order delivery time window and vehicle cargo capacity limit, and formulate a vehicle allocation list based on the vehicle scheduling resource allocation, vehicle cargo capacity limit and order delivery task; then generate an initial solution population based on the actual constraints of the customer information and service resource information provided by the system; each individual in the solution population represents a complete service and logistics solution, and through coding, fully represents the comprehensive solution of service tasks, service specialist allocation, logistics tasks and vehicle scheduling; S23, a dual branch-and-bound method, respectively executing a customer satisfaction branch-and-bound sub-algorithm and a service cost branch-and-bound sub-algorithm, wherein the customer satisfaction branch-and-bound sub-algorithm comprises the steps of maximizing customer satisfaction by optimizing service strategies and allocation of service specialists, removing solutions that do not meet customer satisfaction constraints through pruning operations, and ensuring that service specialists are maximized in efficiency; the service cost branch-and-bound sub-algorithm comprises the steps of improving vehicle utilization and minimizing transportation costs by optimizing vehicle scheduling and cargo load ratio, removing solutions that exceed vehicle cargo load limits or cannot meet time windows through pruning operations, ensuring that the cargo load on each vehicle does not exceed its maximum cargo load, and ensuring that the cost is optimal and the task is completed on time; S24. Multi-objective optimization. The solution processed by the dual branch and bound algorithm is input into the multi-objective genetic algorithm for optimization. In the iterative process of the multi-objective optimization algorithm, the population is merged, and a comprehensive evaluation is performed on each individual in terms of the two objectives of customer satisfaction and service cost. The population diversity is maintained by non-dominated sorting and crowding distance calculation. The optimal individuals and individuals that meet the diversity requirements are selected to enter the next generation population. In the genetic operator part, the tournament selection strategy is used to select excellent individuals from the current population, and the two-point crossover method is used to exchange some gene fragments of the chromosome to explore new solutions. The mutation operation uses the perturbation mutation method to mutate the chromosome genes to increase the population diversity. The above evaluation, selection, crossover and mutation processes are repeated, and the results are traced back to the dual-objective branch and bound algorithm. The solution is iterated until the preset number of iterations is reached. Finally, the Pareto frontier is formed, and the solution set that cannot be further optimized between different optimization objectives is obtained. By analyzing the Pareto frontier, the service and scheduling plan that best meets the actual needs is selected to achieve the dual optimization of customer satisfaction and cost control.

[0022] The implementation of the embodiment of the present invention has the following beneficial effects: the present invention decomposes the problem into two sub-problems through double branch and bound, and optimizes them respectively for different objectives. This method can effectively narrow the search space and improve the convergence speed of the algorithm while ensuring the diversity of solutions. On the one hand, it focuses on improving the satisfaction of service customers and the service efficiency of service specialists, and ensures that service specialists can execute service strategies in a timely and accurate manner by optimizing the scheduling of service tasks; on the other hand, it focuses on reducing service costs, and reduces time cost waste and resource waste through reasonable allocation of resources, thereby reducing overall service costs, helping to improve the accurate transmission of service strategy information and the efficiency of service personnel, and optimizing resource allocation. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0024] In order to make the objectives, technical solutions and advantages of the present invention more clear, the present invention will be further described in detail below with reference to the accompanying drawings.

[0025] In this embodiment, an improved double branch and bound multi-objective optimization method for service strategy generation and task intelligent analysis and scheduling is provided. Figure 1 As shown, and implemented through the following steps.

[0026] S1. Aiming at realizing the automatic generation of service strategies, aiming to accurately meet customer service needs and maximize customer satisfaction, while striving to minimize service delivery costs, obtain multi-dimensional information from the system database, build a distribution logistics network, and establish a dual-objective branch-and-bound algorithm that comprehensively considers customer satisfaction and service strategy costs.

[0027] Among them, multi-dimensional information is obtained to provide necessary data support for encoding and initializing populations and to realize automatic generation of service strategies. Multi-dimensional information related to customers is obtained from the system database, including customer characteristics, levels, service lists, service orders, service specialist capabilities and other dimensional information. By integrating these dimensional information and combining the preset strategies in the company's service strategy library, the system accurately matches service strategy activities suitable for customers, including matching between customer levels and service activities, matching between service specialist capabilities and task adaptability, and selecting and matching basic service strategies. This process is achieved by integrating tobacco marketing service specialists, terminal customers, service task characteristics, regular visit plans, and information on time, region and other factors. The obtained service strategy activity set is used as input data for the customer satisfaction branch and bound sub-algorithm in the dual-objective branch and bound algorithm.

[0028] When constructing a distribution logistics network, the system extracts distribution information based on customer information and customer orders, and obtains the customer's specific delivery address information from the database. Combined with map data, the system obtains relevant information between nodes in the distribution network (including shipping points and customer points), including distance and transportation time, and then constructs a logistics undirected connected graph, and calculates the transportation distance, total time and total transportation cost of a single logistics vehicle in logistics transportation. These calculation results will be used as input data for the service cost branch and bound sub-algorithm in the dual-objective branch and bound algorithm to optimize vehicle scheduling and logistics distribution solutions.

[0029] S2. Determine the objective function and constraints of the dual-objective branch-and-bound algorithm. The dual-objective branch-and-bound algorithm is divided into two sub-algorithms: one is a customer satisfaction branch-and-bound sub-algorithm with maximizing customer satisfaction of the service strategy as the objective function; the other is a service cost branch-and-bound sub-algorithm with minimizing the cost of the order delivery link as the objective function; Among them, the comprehensive objective function of the customer satisfaction branch and bound sub-algorithm is:

[0030] The above formula represents the objective function of the customer satisfaction branch and bound sub-algorithm to maximize customer satisfaction. Indicates customer satisfaction, Indicates the work efficiency of service specialists. Indicates resource utilization, represents the fairness and matching degree of task allocation, E is the penalty function of the customer service process, , , , , They represent the weights of their respective sub-goals.

[0031]

[0032] The above formula represents the customer satisfaction objective function, O represents the set of customer service strategy activities, , Represents customer satisfaction feedback, and calculates customer ratings based on historical data and simulated service results. Represents the service response time, which is the time difference from the time the customer requests the service to the time the service starts. Indicates the quality of service completion. Based on historical data and the completion quality of different simulated service strategies, the customer satisfaction score after the service is completed is calculated. Indicates customer loyalty and calculates the customer repurchase rate based on historical data and simulated service results. , , , Represent the specific weight of each indicator respectively.

[0033]

[0034] The above formula represents the service specialist efficiency objective function, O represents the set of customer service strategy activities, , represents the task completion time of service specialist z in service activity i, represents the number of tasks completed by service specialist z in service activity i, represents the task complexity of service specialist z in performing service activity i, , , Represent the specific weight of each indicator respectively.

[0035]

[0036] The above formula represents the resource utilization objective function, O represents the set of customer service strategy activities, , It represents the total time that resource j occupies in task i within a certain period of time. represents the idle time of resource j in task i, Indicates the specific weight.

[0037]

[0038] The above formula represents the objective function of task allocation, D represents the set of service specialists, , represents the maximum value of task load among all service specialists, represents the minimum value of task load among all service specialists, Indicates that the task of service specialist g meets the requirements. Indicates the specific weight.

[0039] The penalty function is:

[0040] The above formula represents the penalty function of the customer service process. P(s) is a large penalty coefficient used to satisfy the service constraints. is the service start time of task i, represents the task completion time of service specialist z in service activity i, is the latest allowed time for service completion. It also includes the service cost branch and bound sub-algorithm, and the comprehensive objective function is:

[0041] The above formula represents the objective function of the branch-and-bound algorithm for logistics cost, which minimizes the time cost and expense cost of serving customer order delivery. Indicates the time cost, Represents the delivery logistics cost. , They represent the weights of their respective sub-goals.

[0042]

[0043] The above formula represents the time cost objective function of transportation and distribution scheduling, A represents the transportation and distribution task information, , E represents the node set of the distribution logistics network, ,, represents the required distance between the service delivery customer nodes i and j for the transportation delivery task a, P(t) represents a large penalty coefficient to ensure that the time window constraint is met, Represents a delivery task a The starting time from serving client node r, represents the transportation time from service delivery customer node i to transportation service node j for transportation delivery task a, Indicates the latest delivery time allowed for task a.

[0044]

[0045] The above formula represents the objective function of transportation and distribution cost, where Y represents the set of dispatchable vehicle information. , represents the yth logistics vehicle, k is a weight coefficient used to adjust the impact of loading waste in the total cost, represents the loading rate of delivery vehicle y, is the fixed cost of starting vehicle y, represents the transportation cost between customer nodes i and j served by delivery task a, P(m) is a large penalty coefficient used to penalize solutions that exceed the load capacity, P(v) is a large penalty coefficient used to penalize solutions that exceed the volume limit, U represents the delivery task set of the current delivery logistics vehicle, , represents the cargo weight of the delivery task u that vehicle y is responsible for, represents the cargo volume of the delivery task u that vehicle y is responsible for, Indicates that the total weight of the delivery task of vehicle y does not exceed the maximum load of the vehicle; It means that the total volume of the delivery task of vehicle y does not exceed the maximum volume of the vehicle.

[0046] Constraints of the service cost branch and bound subalgorithm: Considering the vehicle cargo capacity limit and the specific requirements of the order delivery task, the encoding scheme needs to meet the following constraints; The total weight of each vehicle's delivery task does not exceed the vehicle's maximum load M. max ; The total volume of each vehicle's delivery mission does not exceed the vehicle's maximum volume Vo max ; The execution time of each delivery task must be within the time window required by the customer; The weight of each newly added delivery task shall not exceed the current loadable weight M of the vehicle. current ; The volume of each newly added delivery task shall not exceed the current loading volume Vo of the vehicle. current ; Constraints for pruning rules, define the following variables; ; For the load constraint, the delivery logistics vehicle y and its delivery task set U ensure that the load of the delivery task u will not exceed the maximum load of vehicle y if , then prune.

[0047] ; For the volume constraint, the distribution logistics vehicle y and its distribution task set U ensure that the volume of the distribution task u does not exceed the maximum volume of vehicle y if , then prune.

[0048] ; For the time window constraint, ensure that it is within the time window [E i , L i ], where E i is the earliest start time, L i is the latest completion time, if , then prune.

[0049] ; For each logistics vehicle z and newly added task order i, ensure that the weight of the current task order i does not exceed the current loadable weight of the assigned vehicle if , then prune.

[0050] ; For each logistics vehicle z and newly added task order i, ensure that the volume of the current task order i does not exceed the current loading capacity of the assigned vehicle if , then prune.

[0051] S3. Model construction. To further improve the performance and solution efficiency of the algorithm, a comprehensive model combining the double branch and bound method and the multi-objective genetic optimization algorithm (NSGA-II) was designed. The model uses the double branch and bound method to perform preliminary search and constraint pruning on the solution space, and quickly screens out candidate solution sets that meet basic constraints. On this basis, the genetic operation is further optimized through NSGA-II to achieve efficient search, and finally output the optimal solution set that meets customer satisfaction and cost control.

[0052] In step S2 of the above method, the dual branch and bound-multi-objective genetic optimization algorithm model mainly includes the following steps: S21, encoding, in the dual-objective branch-and-bound algorithm model, the customer satisfaction branch-and-bound sub-algorithm and the service cost branch-and-bound sub-algorithm both use real number encoding. Some chromosome gene fragments of the genetic algorithm represent different dimensional information, including service strategy selection, service specialist allocation, service task sequence, logistics distribution sequence, and vehicle scheduling allocation; S22. Initialize the population. First, generate a service task scheduling priority list based on the service task time window, task dependency, priority and other logical relationships. Generate a service specialist allocation list based on the service specialist’s resource allocation and service specialist’s work ability. At the same time, generate a logistics distribution task scheduling priority list based on the order’s delivery time window and vehicle cargo capacity limit. Formulate a vehicle allocation list based on the vehicle’s scheduling resource allocation, vehicle cargo capacity limit and order delivery tasks. On this basis, generate an initial solution population based on the actual constraints of the customer information and service resource information provided by the system. Each individual in the solution population represents a complete service and logistics solution. Through coding, it fully represents the comprehensive solution of service tasks, service specialist allocation, logistics tasks and vehicle scheduling. S23, dual branch and bound method, respectively executes customer satisfaction branch and bound sub-algorithm and service cost branch and bound sub-algorithm. Customer satisfaction branch and bound sub-algorithm: maximizes customer satisfaction by optimizing service strategies and allocation of service specialists. Through pruning operations, remove solutions that do not meet customer satisfaction constraints to ensure that service specialists are maximized in efficiency. Service cost branch and bound sub-algorithm: optimize vehicle scheduling and cargo load ratio to improve vehicle utilization and minimize transportation costs. Through pruning operations, remove solutions that exceed vehicle cargo load limits or cannot meet time windows to ensure that the cargo loaded on each vehicle does not exceed its maximum cargo load, ensuring optimal costs and on-time completion of tasks.

[0053] S24, multi-objective optimization, the solution processed by the dual branch and bound algorithm is input into the multi-objective genetic algorithm (NSGA-II) for optimization. In the iterative process of the multi-objective optimization algorithm, the population is merged, and each individual is comprehensively evaluated on the two objectives of customer satisfaction and service cost. The population diversity is maintained by non-dominated sorting and crowding distance calculation, and the optimal individuals and individuals that meet the diversity requirements are selected to enter the next generation population. In the genetic operator part, the tournament selection strategy is used to select excellent individuals from the current population, and the two-point crossover method is used to exchange some gene fragments of the chromosome to explore new solutions. The mutation operation uses the perturbation mutation method to mutate the chromosome genes to increase the population diversity. Repeat the above evaluation, selection, crossover and mutation process, and trace the results back to the dual-objective branch and bound algorithm, iteratively solve until the preset number of iterations is reached. Finally, the Pareto frontier is formed, and the solution set that cannot be further optimized between different optimization objectives is obtained. By analyzing the Pareto frontier, the service and scheduling scheme that best meets the actual needs can be selected to achieve the dual optimization of customer satisfaction and cost control.

[0054] The above disclosure is only a preferred embodiment of the present invention, which certainly cannot be used to limit the scope of the present invention. Therefore, equivalent changes made according to the claims of the present invention are still within the scope of the present invention.

Claims

1. An improved double-branch-and-bound multi-objective optimization method for service strategy generation and task intelligent analysis and scheduling, characterized in that: The following steps are involved: S1: Obtain multi-dimensional information from the system database, construct a distribution logistics network, and establish a bi-objective branch-and-bound algorithm that comprehensively considers customer satisfaction and service strategy costs; S2: Determine the objective function of the dual-objective branch-and-bound algorithm, which includes two sub-algorithms: one is a customer satisfaction branch-and-bound sub-algorithm with maximizing the customer satisfaction of the service strategy as the objective function; the other is a service cost branch-and-bound sub-algorithm with minimizing the cost of the order delivery link as the objective function; S3: Using the NSGA-II algorithm and combining with resource constraints, an improved dual branch and bound-multi-objective genetic optimization algorithm model is constructed, and the dual-objective branch and bound algorithm is combined with the multi-objective optimization algorithm for optimization and solution.

2. The method according to claim 1, characterized in that The multi-dimensional information is obtained in S1, including obtaining at least one of the dimensional information of customer characteristics, level, service list, service subscription, and service specialist capabilities.

3. The method according to claim 1, characterized in that The construction of the distribution logistics network described in S1 includes the following steps: extracting distribution information based on customer information and customer entrusted orders, and obtaining the customer's specific delivery address information from the database; combining map data to obtain relevant information between nodes in the distribution network, including distance and transportation time, and then constructing a logistics undirected connected graph, and calculating the transportation distance, total time spent and total transportation cost of a single logistics vehicle in logistics transportation.

4. The method according to claim 3, characterized in that: The comprehensive objective function of the customer satisfaction branch and bound sub-algorithm in S2 is: The above formula represents the objective function of the customer satisfaction branch and bound sub-algorithm to maximize customer satisfaction. Indicates customer satisfaction, Indicates the work efficiency of service specialists. Indicates resource utilization, represents the fairness and matching degree of task allocation, E is the penalty function of the customer service process, , , , , Respectively represent the weights of their respective sub-goals; The above formula represents the customer satisfaction objective function, O represents the set of customer service strategy activities, , Represents customer satisfaction feedback, and calculates customer ratings based on historical data and simulated service results. Represents the service response time, which is the time difference from the time the customer requests the service to the time the service starts. Indicates the quality of service completion. Based on historical data and the completion quality of different simulated service strategies, the customer satisfaction score after the service is completed is calculated. Indicates customer loyalty and calculates the customer repurchase rate based on historical data and simulated service results. , , , Respectively represent the specific weight of each indicator; The above formula represents the service specialist efficiency objective function, O represents the set of customer service strategy activities, , represents the task completion time of service specialist z in service activity i, represents the number of tasks completed by service specialist z in service activity i, represents the task complexity of service specialist z in performing service activity i, , , Respectively represent the specific weight of each indicator; The above formula represents the resource utilization objective function, O represents the set of customer service strategy activities, , It represents the total time that resource j occupies in task i within a certain period of time. represents the idle time of resource j in task i, Indicates specific weight; The above formula represents the objective function of task allocation, D represents the set of service specialists, , represents the maximum value of task load among all service specialists, represents the minimum value of task load among all service specialists, Indicates that the task of service specialist g meets the requirements. Indicates specific weight; The above formula represents the penalty function of the customer service process. P(s) is a large penalty coefficient used to satisfy the service constraints. is the service start time of task i, represents the task completion time of service specialist z in service activity i, is the latest allowed time for service completion.

5. The method according to claim 4, characterized in that The comprehensive objective function of the service cost branch and bound sub-algorithm in S2 is: The above formula represents the objective function of the branch-and-bound algorithm for logistics cost, which minimizes the time cost and expense cost of serving customer order delivery. Indicates the time cost, Represents the delivery logistics cost. , Respectively represent the weights of their respective sub-goals; The above formula represents the time cost objective function of transportation and distribution scheduling, A represents the transportation and distribution task information, , E represents the node set of the distribution logistics network, ,, represents the required distance between the service delivery customer nodes i and j for the transportation delivery task a, P(t) represents a large penalty coefficient to ensure that the time window constraint is met, represents the starting time of the delivery task a from the service customer node r, represents the transportation time from service delivery customer node i to transportation service node j for transportation delivery task a, Indicates the latest delivery time allowed for task a; The above formula represents the objective function of transportation and distribution cost, where Y represents the set of dispatchable vehicle information. , represents the yth logistics vehicle, k is a weight coefficient used to adjust the impact of loading waste in the total cost, represents the loading rate of delivery vehicle y, is the fixed cost of starting vehicle y, represents the transportation cost between customer nodes i and j served by delivery task a, P(m) is a large penalty coefficient used to penalize solutions that exceed the load capacity, P(v) is a large penalty coefficient used to penalize solutions that exceed the volume limit, U represents the delivery task set of the current delivery logistics vehicle, , represents the cargo weight of the delivery task u that vehicle y is responsible for, represents the cargo volume of the delivery task u that vehicle y is responsible for, Indicates that the total weight of the delivery task of vehicle y does not exceed the maximum load of the vehicle; It means that the total volume of the delivery task of vehicle y does not exceed the maximum volume of the vehicle.

6. The method according to claim 5, characterized in that Constraints of the service cost branch and bound sub-algorithm: Considering the vehicle cargo capacity limit and the specific requirements of the order delivery task, the encoding scheme needs to meet the following constraints; The total weight of each vehicle’s delivery mission shall not exceed the vehicle’s maximum load capacity Mmax; The total volume of each vehicle’s delivery mission shall not exceed the vehicle’s maximum volume, Vomax; The execution time of each delivery task must be within the time window required by the customer; The weight of each newly added delivery task shall not exceed the vehicle's current loadable weight Mcurrent; The volume of each newly added delivery task shall not exceed the vehicle's current loadable volume Vocurrent; Constraints for pruning rules, define the following variables; ； For the load constraint, the delivery logistics vehicle y and its delivery task set U ensure that the load of the delivery task u will not exceed the maximum load of vehicle y if , then prune; ； For the volume constraint, the distribution logistics vehicle y and its distribution task set U ensure that the volume of the distribution task u does not exceed the maximum volume of vehicle y if , then prune; ； For time window constraints, ensure that they are completed within the time window [Ei, Li], where Ei is the earliest start time and Li is the latest completion time. If , then prune; ； For each logistics vehicle z and newly added task order i, ensure that the weight of the current task order i does not exceed the current loadable weight of the assigned vehicle if , then prune; ； For each logistics vehicle z and newly added task order i, ensure that the volume of the current task order i does not exceed the current loading capacity of the assigned vehicle if , then prune.

7. The method according to claim 1, characterized in that The dual-objective branch and bound algorithm model in S2 is solved using a dual-branch and bound-multi-objective genetic optimization algorithm.

8. The method according to claim 7, characterized in that The dual branch and bound-multi-objective genetic optimization algorithm comprises the following steps: S21, encoding, in the dual-objective branch-and-bound algorithm, the customer satisfaction branch-and-bound sub-algorithm and the service cost branch-and-bound sub-algorithm both use real number encoding, where some chromosome gene fragments of the genetic algorithm represent different dimensional information, including service strategy selection, service specialist allocation, service task sequence, logistics distribution sequence and vehicle scheduling allocation; S22, initialize the population, first generate a service task scheduling priority list based on the service task time window, task dependency and priority and other logical relationships, and generate a service specialist allocation list based on the service specialist's resource allocation and service specialist's work ability; at the same time, generate a logistics distribution task scheduling priority list based on the order delivery time window and vehicle cargo capacity limit, and formulate a vehicle allocation list based on the vehicle scheduling resource allocation, vehicle cargo capacity limit and order delivery task; then generate an initial solution population based on the actual constraints of the customer information and service resource information provided by the system; each individual in the solution population represents a complete service and logistics solution, and through coding, fully represents the comprehensive solution of service tasks, service specialist allocation, logistics tasks and vehicle scheduling; S23, a dual branch-and-bound method, executing a customer satisfaction branch-and-bound sub-algorithm and a service cost branch-and-bound sub-algorithm respectively; S24. Multi-objective optimization. The solution processed by the dual branch and bound algorithm is input into the multi-objective genetic algorithm for optimization. In the iterative process of the multi-objective optimization algorithm, the population is merged, and a comprehensive evaluation is performed on each individual in terms of the two objectives of customer satisfaction and service cost. The population diversity is maintained by non-dominated sorting and crowding distance calculation. The optimal individuals and individuals that meet the diversity requirements are selected to enter the next generation population. In the genetic operator part, the tournament selection strategy is used to select excellent individuals from the current population, and the two-point crossover method is used to exchange some gene fragments of the chromosome to explore new solutions. The mutation operation uses the perturbation mutation method to mutate the chromosome genes to increase the population diversity. The above evaluation, selection, crossover and mutation processes are repeated, and the results are traced back to the dual-objective branch and bound algorithm. The solution is iterated until the preset number of iterations is reached. Finally, the Pareto frontier is formed, and the solution set that cannot be further optimized between different optimization objectives is obtained. By analyzing the Pareto frontier, the service and scheduling plan that best meets the actual needs is selected to achieve the dual optimization of customer satisfaction and cost control.

9. The method according to claim 8, characterized in that The customer satisfaction branch and bound sub-algorithm in S23 includes the steps of maximizing customer satisfaction by optimizing service strategies and allocation of service specialists, and removing solutions that do not meet customer satisfaction constraints through pruning operations to ensure maximum efficiency of service specialists.

10. The method according to claim 9, characterized in that The service cost branch and bound sub-algorithm described in S23 includes the steps of: optimizing vehicle scheduling and cargo load ratio to improve vehicle utilization and minimize transportation costs, removing solutions that exceed the vehicle cargo load limit or cannot meet the time window through pruning operations, ensuring that the cargo loaded on each vehicle does not exceed its maximum cargo load, ensuring optimal costs and on-time completion of tasks.

Images