PURPOSE: To optimize a storage water amount so that stagnating water can be eliminated, by estimating a demand amount learned in a neural net work with a fluctuating factor of a supply water amount serving as an input and with the supply water amount serving as an output.
CONSTITUTION: By a supply water amount calculating memory means 27, from an inflow amount of water to a storage water tank and from a storage water amount, a supply water amount in each unit time is calculated and stored. A fluctuating factor of fluctuating the supply water amount is input to a fluctuating factor input means 28. Next by a demand amount estimating means 29, a demand amount is learned by a neural network with the fluctuating factor serving as an input and with the supply water amount serving as an output, to estimate the demand amount of supply water after the present point of time. By a storage water tank control means 30, based on the present storage water amount and the demand amount estimated value, a water amount in the storage water tank is controlled so as to eliminate stagnating water by optimizing to an amount matched with the demand amount.
OZAKI MINORU
MATSUSHITA MASAHITO
FUKAZAWA YUTAKA
Next Patent: FREEZING WATER CUT-OFF CONSTRUCTION METHOD