Control parameter computation method and control parameter computation device

A non-transitory recording medium storing a control parameter computation program that is executable by a computer to perform a process, the process comprising:

• reading divided, paired data representing a pairing of time series data of an manipulated variable subject to control and time series data of a control variable with respect to the manipulated variable at each time for every predetermined time segment from a storage section storing the paired data for each time segment, and identifying, for each time segment, a plurality of mathematical models of different structures based on the paired data corresponding to the time segment, using a system identification method;
• for each time segment and the plurality of mathematical models identified for the time segment, using the mathematical models to compute predicted values of the control variable with respect to the manipulated variable in the paired data for the respective paired data corresponding to time segments different from the time segment employed in identification, and computing an adaptiveness representing a level of adaptation between a time series of the computed control variable predicted values and time series data of the control variable in the paired data corresponding to the time segments different from the time segment employed in the identification;
• for each time segment, selecting, as a mathematical model for the paired data of the time segment, a mathematical model for which the adaptiveness is greater than a predetermined threshold value;
• for each time segment, configuring a PID controller based on an internal model control method and on the respective selected mathematical model, generating a transfer function represented by a product of the PID controller and the selected mathematical model, and generating a function representing a gain margin of the transfer function and a function representing a phase margin of the transfer function;
• and
• computing a parameter Lambda in the internal model control method such that the function representing the gain margin is equal to or greater than a gain margin setting threshold value and the function representing the phase margin is equal to or greater than a phase margin setting threshold value, and so as to minimize a function represented by a sum of a difference between the function representing the gain margin and the gain margin setting threshold value added to a difference between the function representing the phase margin and the phase margin setting threshold value, and then computing a control parameter in the PID controller from the parameter Lambda.

A control parameter computation device comprising:

• a memory;
• and
• a processor coupled to the memory, the processor being configured to:
• read divided, paired data representing a pairing of time series data of an manipulated variable subject to control and time series data of a control variable with respect to the manipulated variable at each time for every predetermined time segment from a storage section storing the paired data for each time segment, and identify, for each time segment, a plurality of mathematical models of different structures based on the paired data corresponding to the time segment, using a system identification method;
• for each time segment and the plurality of mathematical models identified for the time segments, use the mathematical models to compute predicted values of the control variable with respect to the manipulated variable in the paired data for the respective paired data corresponding to time segments different from the time segment employed in identification, and compute an adaptiveness between a time series of the computed control variable predicted values and time series data of the control variable in the paired data corresponding to the time segments different from the time segment employed in the identification;
• for each time segment, select, as a mathematical model for the paired data of the time segment, a mathematical model for which the adaptiveness is greater than a predetermined threshold value;
• for each time segment, configure a PID controller based on an internal model control method and on the respective selected mathematical model, generate a transfer function represented by a product of the PID controller and the selected mathematical model, and generate a function representing a gain margin of the transfer function and a function representing a phase margin of the transfer function;
• compute a parameter Lambda in the internal model control method such that the function representing the gain margin is equal to or greater than a gain margin setting threshold value and the function representing the phase margin is equal to or greater than a phase margin setting threshold value, and so as to minimize a function represented by a sum of a difference between the function representing the gain margin and the gain margin setting threshold value added to a difference between the function representing the phase margin and the phase margin setting threshold value;
• and
• compute a control parameter in the PID controller based on the computed parameter Lambda.