遺伝的アルゴリズムを用いた小口径トンネルロボットの最適ゲインオートチューニング

概要

論文の詳細を見る
We describe an autotuning method of feedback gain for a small-diameter tunnelling robot. We have already proposed a directional control method in which the head angle of the control input is the sum of the deviation multiplied by the feedback gain K_p and the angular deviation multiplied by the feedback gain K_a. We used a neural network to automatically optimize the feedback gains K_p and K_a. However, this conventional method required too much learning time in the neural network. Therefore, in this paper, we use a genetic algorithm to automatically optimize the feedback gains. This genetic algorithm is a search algorithm based on the mechanics of natural selection and natural genetics. The problem of optimizing feedback gains in this paper can be regarded as the problem of searching for a combination of feedback gains which is a individual making the best fitness in individuals. We designed individuals which have an artificial gene consisting of a string expressed by the binary digit vectors of K_p and K_a. We simulated evolving individuals using the genetic algorithm to search for optimum feedback gains. This method finds a solution more quickly than the conventional method. We show computer simulation results and verify the applicability and usefulness of this method.
一般社団法人日本機械学会の論文
1996-09-25