Control system having learning control function and control method

A control system comprising a driven device driven by a servomotor, a control device for controlling the servomotor to operate the driven device, and a sensor for detecting a value related to the position of a certain portion of the driven device, wherein

• the control device includes:
• an operation control unit for generating a command value for the servomotor in each control cycle based on an operation program generated in advance, to control the operation of the driven device by the command value;
• and
• a learning control processing unit for making the operation control unit repeatedly perform a learning control in which, when an operation control is performed using the operation program, a vibration correction amount is calculated to correct a vibration occurring in the certain portion and the vibration correction amount is applied to the next operation control by the operation program, and
• the learning control processing unit includes:
• a vibration amount estimation unit for estimating time-series vibration data occurring in the certain portion during the single learning control, based on time-series data of a detection value detected by the sensor and time-series data of the command value generated by the operation control unit while the operation control unit performs the single learning control;
• a vibration correction amount calculation unit for calculating the vibration correction amount to be applied to the next operation control by the operation program, based on the time-series vibration data;
• a storage unit, whenever the operation control unit performs the single learning control, the storage unit storing the number of learning controls indicating how many learning controls have been performed and the time-series vibration data in correspondence with each other;
• and
• a determination unit for calculating a convergence determination value to determine whether or not the vibration of the certain portion converges based on the time-series vibration data at each number of learning controls stored in the storage unit, and determines the number of learning controls having a minimum convergence determination value out of the calculated convergence determination values as the optimal number of learning controls.