ニューラル・ネットワークと遺伝的アルゴリズムとを用いた高精密マシニングセンタの開発

概要

論文の詳細を見る
This paper describes a system in which the thermal distortion of a machining center is measured through an internal monitoring method using newly investigated deformation sensors and is actively compensated by thermal actuators, consisting of heaters and coolers, which are located in the lower part of the column. The translational error is compensated by shifting the origin of the coordinate system of the numerical controller and the rotational error is compensated by controlling the thermal actuators. The amount of shifting and controlling is determined by a neural network system integrated with a genetic algorithm and a neural network system with the feature of 3-state sigmoid function. Using this method, although the room temperature was varied during the experiment over a range of 14℃, the position and the attitude of the spindle were maintained within 12 micrometers and 21. 4 microradins, respectively. The achieved result is approximately 1/3 of the variation of the uncontrolled machine.
一般社団法人日本機械学会の論文
1995-11-25