Data Multiple Interpolation Technique Based on Convolutional Neural Networks

The current electricity consumption data from the electricity information collection system is caused by the external environment such as transmission and interference, resulting in low data quality. If such data is used for high-precision power system evaluation and related decisions, multiple data interpolation operations are required. This paper constructs multiple data interpolation models based on convolutional neural networks. Firstly, mathematical statistics method is used to identify abnormal data, and a convolutional neural network model is constructed to interpolate missing and outliers. Experiments on the real power database show that the data interpolation model proposed in this paper is simple and efficient, and is widely applicable to power data interpolation processing, which can improve the power data quality of the electricity information collection system.