Tire physical information estimation system and arithmetic operation model generation system

An arithmetic operation model generation system comprising:

• a physical information estimator that includes a learning type arithmetic operation model based on a neural network including an input layer through an output layer and estimates tire physical information generated in accordance with movement of a tire;
• a data acquisitor configured to acquire input data input to the input layer;
• a learning processor that trains the arithmetic operation model based on training data derived from normalizing the tire physical information measured in the tire;
• and
• a vehicle control device that uses the tire physical information estimated by the physical information estimator for at least one of estimation of braking distance, application to vehicle control, and notification of the driver of information related to the safe driving of the vehicle,
• wherein the arithmetic operation model includes a feature extractor configured to perform a convolution operation to the data input to the input layer and the output layer, the arithmetic operation model outputting normalized tire physical information from the output layer, and the tire physical information is a tire force in each of three axial directions among longitudinal, lateral and vertical directions,
• wherein the arithmetic operation model includes a plurality of fully-connected units corresponding to respective axial directions and inputs an output of the feature extractor to the plurality of fully-connected units, and
• wherein the output layer outputs the normalized tire forces in longitudinal and lateral directions defined as values of divided by a constant value, and the normalized tire force in vertical direction defined as a value of divided by a constant value after subtracted by a median value.