To quickly obtain a mix proportion suitable for a concrete without requiring skill by a method wherein weather data, characteristic values of respective structural materials of the concrete and a necessary consistency is imputted in a processing means having a neural network system equipped with an input and an output layers and one and more intermediate layer and executed an action.

Kinds of structural materials to be mixed with one another for a concrete, a necessary consistency and, when necessary, mechanical characteristic values are selected and specified in advance and then inputted in a computer 1. The computer 1 reads the characteristic values of the corresponding structural materials from a memory 1a and the data sent from an atmospheric temperature and humidity sensor 2 are read at predetermined time intervals so as to input them in the corresponding neural units. The inputted value is calculated to the direction of easy flow from an input layer to an output layer and finally outputs the mix proportion of the respective structural materials through the respective neural units. When the change of the consistency is requested, by inputting its from a transmitter 4, its is inputted through a receiver 3 and a computer 4, resulting in allowing to obtain a new mix proportion responding to the changed condition.