Data Fusion Algorithm in Agricultural Internet of Things System

Agricultural production relies on a large number of sensors to obtain environmental data, and how to effectively integrate this data has become a major challenge in current agricultural IoT applications. Therefore, this article proposes a novel data fusion algorithm. Firstly, the weighted average method is used to preliminarily fuse data from different types of sensors; Then, deep neural networks (DNNs) are used to further extract potential features from the fused data. Finally, the weighting strategy and network model were optimized through experiments. The experimental results indicate that the research method performs well in terms of accuracy, spatiotemporal consistency, and real-time performance. Compared with the weighted average method, this research method reduced the mean square error (MSE) by about 29% and increased the dynamic time warping (DTW) score by about 40.7%. The results verified that the method is practical in improving data quality and suitable for large-scale data processing in agricultural IoT.