Plant control supporting apparatus, plant control supporting method, and recording medium

A plant control supporting method using a computer, the plant control supporting method comprising:

• selecting, by the computer from among a plurality of segments defined in units of control loops in a plant, a segment for which first learning for acquiring an optimal value of at least one parameter representing an operation state is executed;
• defining, by the computer, a reward function used for the first learning;
• extracting, by the computer, at least one parameter that is a target for the first learning in the selected segment based on input and output information of a device used in the plant and based on segment information representing a configuration of a device included in the selected segment;
• performing, by the computer, the first learning for acquiring the optimal value of the at least one parameter for each segment based on the reward function and the extracted at least one parameter;
• performing second learning for acquiring a whole optimization value of the at least one parameter for a whole of the plurality of segments on the basis of the optimal value of the at least one parameter acquired in the first learning performed for each segment;
• and
• outputting an operation instruction used for giving an instruction for operating the plurality of segments on the basis of the whole optimization value acquired in the second learning,
• wherein
• the selecting of the segment comprises sequentially selecting the plurality of segments defined in the plant from a segment in a further downstream process to a segment in a further upstream process in accordance with a flow of a series of processes in the plant, or from a segment in a further upstream process to a segment in a further downstream process in accordance with the flow of the series of processes in the plant.

A non-transitory computer-readable storage medium storing a plant control supporting program, which when executed by a computer, causes the computer to:

• select, from among a plurality of segments defined in units of control loops in a plant, a segment for which first learning for acquiring an optimal value of at least one parameter representing an operation state is executed;
• define a reward function used for the first learning;
• extract at least one parameter that is a target for the first learning in the selected segment based on input and output information of a device used in the plant and based on segment information representing a configuration of a device included in the selected segment;
• and
• perform the first learning for acquiring the optimal value of the at least one parameter for each segment based on the reward function and the extracted at least one parameter;
• perform second learning for acquiring a whole optimization value of the at least one parameter for a whole of the plurality of segments on the basis of the optimal value of the at least one parameter acquired in the first learning performed for each segment and
• output an operation instruction used for giving an instruction for operating the plurality of segments on the basis of the whole optimization value acquired in the second learning,
• wherein
• the selecting of the segment comprises sequentially selecting the plurality of segments defined in the plant from a segment in a further downstream process to a segment in a further upstream process in accordance with a flow of a series of processes in the plant, or from a segment in a further upstream process to a segment in a further downstream process in accordance with the flow of the series of processes in the plant.