摘要 |
PROBLEM TO BE SOLVED: To provide a prediction learning method capable of quickly acquiring teacher data and correcting a model in feedforward control using learnable control logic. SOLUTION: The prediction learning method in feedforwd control using learnale control logic judges whether a controlled system is in a steady state or not from a driven state and a controlled variable, and when the controlled system is in the steady state, feeds back the controlled variable of the controlled system, predicts a deviation between the model and the controlled system based on the feedback information, prepares predictive teacher data for correcting the deviation, and learns the control logic of the model by using the predictive teacher data in a control method for modeling the controlled system by control logic and executing the feedforward control of the controlled system so as to operate the system along a prescribed objective value by using the model.
|