Electric vehicle anti-collision path planning method based on neural network and double-variation genetic algorithm

The invention discloses an electric vehicle anti-collision path planning method based on a neural network and a double-variation genetic algorithm. The method specifically comprises the following steps: firstly, establishing a two-dimensional coordinate system to describe an obstacle area and encode a path; secondly, establishing a neural network model to limit the initialized population; thirdly, an initial population is randomly generated, the fitness of the initial population is calculated, feasible solutions in the population are counted, elite retention operation is carried out, and for a transition population formed after non-feasible solutions are subjected to crossover, mutation and statistical operation, the fitness is calculated to improve the convergence capacity of the population; and then, preferentially reserving and replacing the transition population. And finally, setting a termination condition, and judging whether to evolve to a preset algebra or population to achieve convergence. According