An evaluation value prediction unit predicts an evaluation value at a time point after a given time passes from a predetermined evaluation time based on input data related to an operation of a factory at the evaluation time using a learned prediction model. The prediction model is a learned model that is learned so that the evaluation value related to the operation of the factory at a time point after the given time passes from one time point is output by inputting a plurality of kinds of data related to the operation of the factory at one time point. An evaluation value output unit outputs information related to the evaluation value.

A self-optimizing controller for equipment of a plant provides a manipulated variable as an input to the plant and receives an output variable as feedback. The controller generates a performance variable model defining the performance variable as a function of the manipulated variable and an output variable model defining the output variable as a function of the manipulated variable. The controller uses the performance variable model to determine a gradient of the performance variable, uses the output variable model to determine a gradient of the output variable, and generates a self-optimizing variable based on the gradient of the performance variable model and the gradient of the output variable model. The controller operates the equipment of the plant to affect a variable state or condition of the building based on the value of the self-optimizing variable from the self-optimizing variable model.

A self-optimizing controller for equipment of a plant provides a manipulated variable as an input to the plant and receives an output variable as feedback. The controller generates a performance variable model defining the performance variable as a function of the manipulated variable and an output variable model defining the output variable as a function of the manipulated variable. The controller uses the performance variable model to determine a gradient of the performance variable, uses the output variable model to determine a gradient of the output variable, and generates a self-optimizing variable based on the gradient of the performance variable model and the gradient of the output variable model. The controller operates the equipment of the plant to affect a variable state or condition of the building based on the value of the self-optimizing variable from the self-optimizing variable model.

A control system is configured to operate a plant to achieve an optimal value for a performance variable of the plant. The system comprise a feedforward controller configured to receive a measurable disturbance to the plant and generate a feedforward contribution to a control input to the plant using the measurable disturbance. The system also comprises an extremum-seeking controller configured to receive the performance variable from the plant and generate an extremum-seeking contribution to the control input to drive the performance variable to the optimal value. The system further comprises a control input element configured to generate the control input by combining the extremum-seeking contribution and the feedforward contribution and provide the control input to the plant.

An object of this invention is to provide a plant operation support-system server and a plant operation support system which can manage the current situation of overall resources that are required for performing jobs corresponding to a procedure manual, and can further achieve resource management with enough margin based on future prediction. Using a current procedure-manual location determination unit, a control unit of a plant operation support-system server determines a current location in procedure-manual data D.sub.M, based on a condition of the plant acquired through sensors by a plant data acquisition unit. When there is a change in the current location, the control unit causes a current-situation management determination unit to assign currently-required resources according to the current position and causes a predictive management determination unit to predict presence/absence of resource deficiency.

An evaluation value prediction unit predicts an evaluation value at a time point after a given time passes from a predetermined evaluation time based on input data related to an operation of a factory at the evaluation time using a learned prediction model. The prediction model is a learned model that is learned so that the evaluation value related to the operation of the factory at a time point after the given time passes from one time point is output by inputting a plurality of kinds of data related to the operation of the factory at one time point. An evaluation value output unit outputs information related to the evaluation value.

A control system is configured to operate a plant to achieve an optimal value for a performance variable of the plant. The system comprise a feedforward controller configured to receive a measurable disturbance to the plant and generate a feedforward contribution to a control input to the plant using the measurable disturbance. The system also comprises an extremum-seeking controller configured to receive the performance variable from the plant and generate an extremum-seeking contribution to the control input to drive the performance variable to the optimal value. The system further comprises a control input element configured to generate the control input by combining the extremum-seeking contribution and the feedforward contribution and provide the control input to the plant.

An object of this invention is to provide a plant operation support-system server and a plant operation support system which can manage the current situation of overall resources that are required for performing jobs corresponding to a procedure manual, and can further achieve resource management with enough margin based on future prediction. Using a current procedure-manual location determination unit, a control unit of a plant operation support-system server determines a current location in procedure-manual data D.sub.M, based on a condition of the plant acquired through sensors by a plant data acquisition unit. When there is a change in the current location, the control unit causes a current-situation management determination unit to assign currently-required resources according to the current position and causes a predictive management determination unit to predict presence/absence of resource deficiency.

An actuator control method and an actuator control device that incorporate an element of feedback control in time optimal control, including: a calculation step of calculating a switching time at which an acceleration output is switched to a deceleration output and an end time of the deceleration output expressed by time elapsed from a calculation time at which calculation for control is performed using a maximum acceleration and a maximum deceleration, which are measured in advance, at the time of the maximum output of control force of an actuator; a control output step of setting the control force of the actuator to a maximum acceleration output from the calculation time to the switching time, setting the control force of the actuator to a maximum deceleration output from the switching time to the end time, and ending the output of the control force at the end time, and an update step of calculating and updating the switching time and the end time by repeating the calculation step at each preset time.