Personalized Route Recommendation Of Hazardous Materials Road Transportation Based On Neural Collaborative Filtering

In recent years,the demand for chemical products is increasing,and the scale of hazardous materials transportation enterprises is expanding,so it is urgent for standardized route selection system to provide preference-based route selection service for enterprises.For general hazardous materials(excluding radioactive,civil explosives,fireworks and precursor hazardous materials),enterprises can choose their own transportation routes,and their route decision-making has enterprise personalized characteristics.The road transportation route recommendation of hazardous materials based on preference and context can better meet the needs of enterprises,improve the quality of transportation service,and standardize the selection of hazardous materials transportation route.Through the processing,analysis and in-depth mining of the GPS trajectory data of hazardous materials transportation vehicles,this paper learns the context scene and preference classification of historical routes,uses the recommendation algorithm to obtain the preference of hazardous materials road transportation enterprises for choosing transportation routes under different context backgrounds,and provides enterprises with transportation routes that fit their preferences according to the preference and context.The main contents of this paper are as follows:(1)Research on route clustering based on preference.This paper proposes a multi angle route preference learning method,which learns route preference from historical trajectory data and constructs preference vector.In addition,a preference clustering algorithm K-PDM combining propensity and distance measurement is proposed,and the same propensity weight is introduced to improve the accuracy of route preference classification judgment.Using this algorithm to realize preference based route clustering,we can obtain the preference classification of routes.(2)Research on route context recognition.This paper proposes a technology to identify route context through historical trajectory data,introduces time context and weather context,and creatively considers other contexts to separate short-distance transportation from medium-distance and long-distance transportation,accurately learns the different preferences of enterprises under different contexts.Learn the effectiveness of the context to each enterprise through the context information and route preference clustering results,and use clustering algorithm Rock to cluster the context weighted by the effectiveness to obtain the route context classification.(3)Research on road transportation route recommendation of hazardous materials based on preference and context.In this paper,the recommendation algorithm is applied to route decision-making,and a hazardous materials road transportation route recommendation algorithm N-HRS is proposed based on the neural collaborative filtering model.The hazardous materials road transportation enterprise is regarded as users,and the preference classification of the transportation route under the context classification is regarded as projects to learn deeper interactions between users and projects,so as to effectively alleviate the problem of data sparsity.The algorithm takes the score of the enterprise on the route preference classification as the basis of route recommendation,learns the preference of the enterprise in choosing the transportation route,and recommends the transportation route in line with its preference for the enterprise under the context.Finally,the GPS trajectory data of hazardous materials road transportation vehicles on a platform are used to verify the effectiveness of preference clustering algorithm and hazardous materials road transportation route recommendation algorithm.The experimental results show that the K-PDM preference clustering algorithm proposed in this paper can effectively identify the route preference and improve the accuracy of transportation route recommendation.Compared with other algorithms,the proposed N-HRS hazardous materials road transportation route recommendation algorithm can learn more potential information,and the recommended route is more in line with the preferences of hazardous materials road transportation enterprises.