An autoencoder wavelet based deep neural network with attention mechanism for multi-step prediction of plant growth

Multi-step-ahead prediction is considered of major significance for time series analysis in many real life problems. Existing methods mainly focus on one-step-ahead forecasting, since multiple step forecasting generally fails due to accumulation errors. This paper presents a novel approach predicting plant growth agriculture, focusing Stem Diameter Variations (SDV). The proposed consists three main steps. At first, wavelet decomposition applied the original data, so as facilitate model fitting and reduce noise. Then an encoder-decoder framework developed using Long Short Term Memory (LSTM) used appropriate feature extraction from data. Finally, recurrent neural network including LSTM attention mechanism modelling long-term dependencies Experimental results are presented which illustrate good performance that it significantly outperforms existing models, terms error criteria such RMSE, MAE MAPE.