An object is to accurately estimate loss-on-ignition in a short time. A loss-on-ignition estimation apparatus includes at least one processor configured to carry out an estimation step, the estimation step including estimating the loss-on-ignition of foundry sand with use of a learned model constructed by means of machine learning. The learned model is configured to receive, as input, (1) sand weight data relating to a weight of the foundry sand detected in a calcination period and (2) at least one of (i) sand property data relating to one or more properties of the foundry sand, (ii) additive data relating to one or more additives added to the foundry sand, and (iii) calcination environment data relating to a calcination environment detected in the calcination period. The learned model is configured to generate, as output, an estimated loss-on-ignition of the foundry sand.

An object is to accurately estimate loss-on-ignition in a short time. A loss-on-ignition estimation apparatus includes at least one processor configured to carry out an estimation step, the estimation step including estimating the loss-on-ignition of foundry sand with use of a learned model constructed by means of machine learning. The learned model is configured to receive, as input, (1) sand weight data relating to a weight of the foundry sand detected in a calcination period and (2) at least one of (i) sand property data relating to one or more properties of the foundry sand, (ii) additive data relating to one or more additives added to the foundry sand, and (iii) calcination environment data relating to a calcination environment detected in the calcination period. The learned model is configured to generate, as output, an estimated loss-on-ignition of the foundry sand.