| CPC G07C 5/008 (2013.01) [G06N 3/09 (2023.01); G07C 5/0808 (2013.01)] | 12 Claims |
|
1. A vehicle evaluation system configured to evaluate a target vehicle, which is a vehicle to be evaluated, using sound data obtained by recording sound emitted from the target vehicle, the vehicle evaluation system comprising:
processing circuitry; and
a memory device, wherein
the memory device stores data of a pretrained model trained with training data, and
the training data includes training sound data and training operation data,
the training sound data is data of sound emitted from a reference vehicle in a state serving as a reference for evaluation, the sound being recorded while the reference vehicle is operated in a measurement operation pattern for a specified time period,
the training operation data is data indicating changes in operation information in the measurement operation pattern,
the pretrained model is a model trained by supervised learning using the training data, to generate, from evaluation data, generated data, which is data of the operation information,
the processing circuitry is configured to execute a generation process and an evaluation process,
the generation process is a process of outputting the generated data which is generated by inputting the evaluation data to the pretrained model,
the evaluation data includes evaluation sound data that is data of sound emitted from the target vehicle and recorded while operating the target vehicle in the measurement operation pattern,
the evaluation process is a process of comparing operation data of the target vehicle with the generated data and evaluating the target vehicle according to a magnitude of a deviation between the operation data of the target vehicle and the generated data,
the pretrained model is a neural network in which a feature quantity of data cropped from the data for the specified time period is used as an explanatory variable, and the operation information at a time point corresponding to the cropped data is used as an objective variable,
the evaluation process includes calculating, as an evaluation index value, a sum total for the specified time period of the deviation between the operation data of the target vehicle and the generated data output for each piece of the data that is repeatedly cropped while changing a cropping start time, wherein
the evaluation index is used as a quantitative measure of vehicle health, such that a higher a value of the evaluation index, the more the vehicle's health deviates from a healthy reference vehicle.
|