Improved Hybrid Bat Algorithm And Its Application In Path Planning

How to have high operation efficiency without consuming too many resources has always been the focus and difficulty of academic research.The traditional optimization strategy is often not ideal for this problem,and the emergence of the swarm intelligence optimization algorithm has changed this situation,so that scholars’ optimal thinking is no longer too limited.This kind of algorithm can obtain the final result through continuous iteration and search,which can not only show high intelligence but also greatly reduce the cost of human resources.Bat Algorithm(BA)mainly imitates the echo characteristics of bat ultrasonic waves in nature.As a representative swarm intelligence optimization algorithm,BA has good searching ability while keeping a simple structure.However,this algorithm is not perfect,and there are still some defects in iterative searches,such as low accuracy of the obtained results,stagnation of the search after obtaining local optimal solutions,and inability to maintain stable optimization results.To improve the optimization performance,this thesis improves the bat algorithm and applies it to the path planning of mobile robots and the path planning of UAV.The main work and research results are as follows:(1)A new hybrid bat algorithm(Golden Sine Bat Algorithm,GSBA)is proposed,which mainly introduces the golden sine operator and updates the individual position in combination with the average position of the population.At the beginning of the iteration,the initialization strategy is improved to make the population evenly distributed in the decision space,to reduce the influence of random initialization on the results.According to fitness,individuals are updated in different ways.The better individuals update with the golden sine operator,which effectively improves the search efficiency,while the worse individuals use the average position of the population to reduce the constraint of local optimization.The two strategies coordinate with each other to make the bat population in a balanced state,and the performance of the algorithm can be guaranteed.In the local search stage,to search the neighborhood of the global optimal position more carefully,a combination of full dimension and single dimension is adopted.In the early iteration,the single dimension search is used to fully obtain the information of each dimension,while in the later stage,the full dimension search is used to achieve rapid convergence,which makes the search in this stage more comprehensive.(2)To solve the problem that the mobile robot can’t reach the target point effectively when performing tasks,a new path planning algorithm based on GSBA is proposed.In the planning process,robots should avoid collision with obstacles,and at the same time,the distance should be shortened as much as possible.It is difficult to get better results when the planning results of traditional algorithms are poor.In this thesis,coordinates are transformed in the stage of environment modeling,which reduces the difficulty of selecting path nodes.Based on GSBA,a random jump strategy is proposed to get rid of the constraint of a local optimal solution,and the search results are re-optimized to delete redundant points in the path.Experimental results show that the planned path quality of the improved algorithm is much better than that of other algorithms in various environments,and the path length is shorter than that of the comparison algorithm.Meanwhile,it shows that the introduction of a random jump strategy can effectively enhance the stability of algorithm optimization.(3)The improved hybrid bat algorithm is applied to the path planning of UAVs,and a solution is proposed for the cooperative operation of multiple UAVs.In the complex mountain environment,the model is built first.To make the algorithm easier to realize and formulate specific coding methods,the influencing factors such as terrain obstacle,weather threat,angle constraint,and flight altitude are considered in the planning process and the corresponding fitness function is set.In the optimization stage,the path nodes are selected to determine the safety of the track,and then the cubic B-spline curve is used to fit the planning results to increase the feasibility of the final track.Simulation results show that GSBA planning results are better than particle swarm optimization and the original bat algorithm.Aiming at the route planning problem of multi-UAV cooperative operation,this thesis expands the single route planning algorithm,and divides the population into several sub-populations to obtain several pre-selected routes during the iterative search,and selects the routes that meet the cooperative constraints.If the strategy fails to complete the planning,virtual unreachable areas are added to the environment to re-plan the shorter routes,so that the related constraints of the multi-UAV problem are satisfied.Experimental results in different environments show that the proposed strategy can better solve the problem of multi-path planning,make multiple UAVs reach the designated point within the specified time,and minimize the cooperative cost of UAV flight.