[en] The amount of electric power used in the operation of accelerators tends to increase in summer due to rising temperature. In recent years, summer temperatures have tended to increase, so understanding the amount of power usage in summer from weather information has become important in terms of contract power and power saving measures. The relation between the amount of power usage and weather conditions is that the accelerator facility has many cooling facilities, so it can be thought that the amount of power used to cool each device increases as the temperature and humidity increase. It seems that it has not been investigated specifically what kind of dependence there is. Therefore, considering the neural network as a kind of fitting function or a model of calculation, we investigated the influence of weather conditions on the power usage. Specifically, we let the neural network learn the temperature and humidity information as input data and the power usage of the accelerators as teaching data, and investigate the influence of the weather conditions on the power usage using the learned neural network. (author)