The study first proposes a heterogeneous fleet, multi-compartment electric vehicle routing problem for perishable products (MCEVRP-PP). We capture a lot of practical demands and constraints of the MCEVRP-PP, such as multiple temperature zones, the hard time window, charging more than once during delivery, various power consumption per unit of refrigeration, etc. We model the MCEVRP-PP as a mixed integer program and aim to optimize the total cost including vehicle fixed cost, power cost, and cooling cost. A hybrid ant colony optimization (HACO) is developed to solve the problem. In the transfer rule, the time window is introduced to improve flexibility in route construction. According to the features of multi-compartment electric vehicles, the capacity constraint judgment algorithm is developed in route construction. Six local search strategies are designed with time windows, recharging stations, etc. Experiments based on various instances validate that HACO solves MCEVRP-PP more effectively than the ant colony optimization (ACO). Compared with fuel vehicles and single-compartment vehicles, electric vehicles and multi-compartment electric vehicles can save the total cost and mileage, and increase utilization of vehicles.

This paper presents the Smart Garment Ecosystem (SG-ECO) and enriches the garment customization production scheduling model theory. On the basis of SG-ECO, the author designs a regional collaborative production alliance (RCPA) based on the idea of collaborative production management and then establishes a flexible production scheduling model (FPSM) that is oriented to the RCPA model under multiple constraints and aims to maximize weighted cost savings. The RCPA model and research on FPSM can enrich the theoretical system of production scheduling research to a certain extent and provide new ideas for the latter’s research on customized production scheduling. Although the calculation example proves that the genetic algorithm based on double-layer integer coding (DIC-GA) can effectively solve the FPSM problem, the feasible solution space of the algorithm increases when the order size increases, and the number of iterations and search time required gradually increase. An improved genetic algorithm with double-layer integer coding for processes and workshops is designed, which can not only ensure that each chromosome is a legal individual but also reduce the complexity of the algorithm. The crossover operation and compilation operation are designed based on the coding method to ensure that the genetic operations in the algorithm can produce feasible solutions.

Crowd intelligence based transaction network (CIbTN) is a new generation of e-commerce. In a CIbTN, buyers, sellers, and other institutions are all independent and intelligent agents. Each agent stores the commodity information in a local node. The agents interconnect through a circle of friends and construct an unstructured network. To conduct the commodity search task in a network more efficiently and in an energy-saving manner when a buyer presents a commodity demand, a hybrid breadth-depth search algorithm (HBDA) is proposed, which combines the search logic of the breadth-first search algorithm and the depth-first search algorithm. We defined the correlation degree of nodes in a network, optimized the rules of search and forwarding paths using the correlation degree between a node and its neighboring nodes in the circle of friends, and realized the HBDA based on the PeerSim simulation tool and Java. Experimental results show that, in general, the proposed HBDA has a better search success rate, search time, commodity matching degree, and search network consumption over the two blind search algorithms. The HBDA also has good expansibility, thus allowing it to be used for commodity search efficiently with a high success rate in large-scale networks.

The crowd intelligence-based e-commerce transaction network (CIeTN) is a distributed and unstructured network structure. Smart individuals, such as buyers, sellers, and third-party organizations, can store information in local nodes and connect and share information via moments. The purpose of this study is to design a product search algorithm on the basis of ant colony optimization (ACO) to achieve an efficient and accurate search for the product demand of a node in the network. We introduce the improved ideas of maximum and minimum ants to design a set of heuristic search algorithms on the basis of ACO. To reduce search blindness, additional relevant heuristic factors are selected to define the heuristic calculation equation. The pheromone update mechanism integrating into the product matching factor and forwarding probability is used to design the network search rules among nodes in the search algorithm. Finally, the search algorithm is facilitated by Java language programming and PeerSim software. Experimental results show that the algorithm has significant advantages over the flooding method and the random walk method in terms of search success rate, search time, product matching, search network consumption, and scalability. The search algorithm introduces the idea of improving the maximum and minimum ant colony system and proposes new ideas in the design of heuristic factors in the heuristic equation and the pheromone update strategy. The search algorithm can search for product information effectively.

Based on the present situation of usage based insurance (UBI) research and application, this paper puts forward the UBI rating model based on driving behavior classification, and applies the technology of data mining to the evaluation of driving behavior. The actual driving behavior data and the risk data of 400 drivers are used as experimental data. Finally, an example shows that the driving behavior classification model is superior to the driving behavior score model for the identification of accident risk, which can make UBI rate more scientific and reasonable.