In a controlling device that uses a model for simulating characteristics of a control target to learn a method of generating a model input that satisfies a target model output value and generates an operation signal according to a learning result, a target model output value for satisfying an operation target value is determined from measured signals from the control target.The controlling device comprising: a model for predicting the value of a measured signal obtained when an operation signal is given to the control target, a function for learning a method of generating a model input so that a model output satisfies the target value, a function for determining an operation signal to be given to the control target according to the learning result, a database for storing measured signal limit values set in advance, an external input interface for fetching measured signals from the control target, a measured signal database for storing the values of the fetched measured signals, and a function for determining an initial value of the target model output value by using limit values of the measured signals and at least one of an average, a maximum value, or a minimum value that are calculated from the measured signals stored in the measured signal database.