IoT Data Forecasting Based on Deep Learning

Abstract

With the development of IoT technology in recent years, IoT sensors are widely used in various scenarios and provide continuous data support for the corresponding IoT services. However, the sensors may be interrupted in the process of data collection and transmission due to external interference or their own factors, which may lead to the interruption of the corresponding IoT services, and eventually affect the normal production order and even cause economic losses. IoT data is a kind of time series data. Therefore, this paper constructs a time series data prediction model based on Convolutional Neural Network's feature extraction ability and Gate Recurrent Unit's time series relationship extraction ability combined with attention mechanism. Finally, the model proposed in this paper is used to predict Beijing PM2.5 data under certain conditions. The results show that compared with the traditional time series data prediction model, the prediction error of this model is reduced by 11.7% at most and 20.9% at least, which effectively improves the accuracy of time series data prediction.