Application of Genetic Algorithm to Optimal Flight Path Problem. On the Maximum Downrange Problem of Space-Plane.
Yasuhiko Takae, Yoshio ISHIKAWA, Katsuhito Yamaguchi · The Journal of the Japan Society of Aeronautical Engineering · 1996
A genetic algorithm (GA) was applied to a maximum downrange problem of a space-plane to study its effectiveness as a solving tool of optimum path problems. It is generally impossible to obtain analytical solutions of optimal control problems in engineering because of the existence of many variables and constraints. Numerical methods, e. g., steepest descent method etc., however, can hardly find global minimums of such problems with strong nonlinearities. In such cases, to choose proper initial value is the essential matter, and it requires much experience and takes long time. Therefore, new methods searching for global optimum are widely developing, especially GA that can simultaneously search the solution space using many test solutions. Conventional methods of GA using bit notation, however, could not be applied to engineering problems, since which could only treat discrete numbers. Therefore we propose a new GA method using coefficients of a polynomial expansion to code genes of a chromosome, and then we can apply it to time continuous control problems. At first we used the method to solve a linear regulator problem to check accuracy and convergence of GA solutions, and next maximum downrange problem of space-plane.