Open AccessData Preprocessing for Agricultural IoT Based on RBF Neural Network
Author(s) -
Qihui Huang,
Yuexin Ma,
Jin Zhang
Publication year - 2019
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1288/1/012040
Subject(s) - preprocessor , artificial neural network , data pre processing , computer science , raw data , data mining , nonlinear system , time series , test data , artificial intelligence , pattern recognition (psychology) , machine learning , physics , quantum mechanics , programming language
The collected data obtained in the agricultural environment is not simply linear and stable, complex nonlinear functional relationships can be found. While RBF neural networks have the ability to approximate arbitrary nonlinear mappings and can implement the functions of nonlinear predictors with superior performance for data forecasting. Considering the strong time correlation of agricultural data, this paper proposes a time series model based on RBF neural network for preprocessing raw data. A four layer of agricultural IoT system is designed, a data processing layer is inserted into the traditional three-layer IoT system. Then the abnormal value is identified and eliminated by the “ t testing criteria” test on the data preprocessing. The experimental results indicate that, comparing with auto regressive moving average model, the proposed model can achieve more competitive prediction ability.