The Application Of GA-BP Nneural Network In The Navigation Path Planning

As is known to all, the intelligent transportation path planning is highly nonlinear, according to which, this paper proposed to conduct the path planning with the hybrid GA-BP neural online algorithm. To verify the advantages of the algorithm in the global optimal solution, we also did some simulation work. This paper studied the algorithm principle and the algorithm implementation carefully, the main work are as follows:(1) To optimize the BP (Back Propagation) neural network with the GA (Genetic Algorithms), we have three different ways. In this paper, we analysis the advantages and disadvantage, as well as the difficulties of the three different ways, and proposed to optimize the BP neural network by modify its initial weights with GA.(2) Established road network models needed by the shortest path planning with the graph theory, also some analysis and configuration about the feasibility of the algorithm were done in this paper.(3) Made a detail comparison with the A*algorithm in the aspect of search time and optimize ability, as a result of which proved the advantages of the hybrid GA-BP neutral network algorithm in optimizing the resolution. And the advantages will be more pronounced while the number of the nods becomes larger.(4) On the basis of the foregoing content, we made a detailed analysis and description of road network models with the limits of traffic turning. At last, we completed the processing with the electronic maps and the geography data on the PC platform, Fedora12operating system,with the QTOPIA run environment. Also we realized the ability to conduct path planning in the electronic maps.