NONLINEAR OPTIMIZATION PROGRAM FOR CONTINUOUS VALUE OPTIMIZATION PROBLEM, AND ROUTE SEARCH PROGRAM AND ROUTE SEARCH DEVICE

PROBLEM TO BE SOLVED: To provide a nonlinear optimization program for a continuous value optimization problem that is enabled to perform high-speed processing on optimization control, and a route search program and a route search device.SOLUTION: A search program is adapted to: prepare a variable of an evaluation function independently separated from a variable of equality constraint in applying a Lagrange undetermined multiplier method with the equality constraint using a Lagrange constant for the evaluation function; and perform processing using additional equality constraint for equalizing the variable of the evaluation function to the variable of the equality constraint. With this search program, an optimization device is made to alternately and repeatedly execute: a procedure for updating the variable of the evaluation function so as to gradually approximate the equality constraint while searching for a condition under which the evaluation function becomes closer to an extremal value; and a procedure for updating the variable of the equality constraint to gradually approximate the condition under which the evaluation function comes closer to the extremal value while the equality constraint is satisfied.SELECTED DRAWING: Figure 6 【課題】最適制御を高速処理できるようにした連続値最適化問題の非線形最適化プログラム、経路探索プログラム、及び経路探索装置を提供する。【解決手段】探索プログラムは、評価関数にラグランジュ定数を用いて等式制約付きのラグランジュ未定乗数法を適用するときに等式制約の変数と独立的に分離した評価関数の変数を用意し、評価関数の変数と等式制約の変数とを一致させる追加等式制約を用いて処理するプログラムである。この探索プログラムによれば、最適化装置に、評価関数が極値に近くなる条件を探索しながら等式制約に徐々に近づくように評価関数の変数を更新する手順と、等式制約を満足したまま前記評価関数が極値に近くなる条件に徐々に近づくように等式制約の変数を更新する手順と、を交互に繰り返し実行させるようにしている。【選択図】図６