Plant electrical signal prediction based on LSTM Neural Network

Abstract

As an important part of facility agriculture, plant factories can directly reflect the level of agricultural development in China, and plant electrical signals are an important physical signal that plants reflect on external stimuli. In view of the complexity of plant growth environment, this paper can not consider the temporal correlation of growth state. It proposes the treatment of plant electrical signals based on long-short-term memory neural network, so that the plants can change under the external environment. The signal is predicted. In this paper, after introducing BP neural network, RNN neural network and LSTM neural network respectively, taking the electric signal prediction of the tiger Piran plant as an example, the parameters such as temperature, humidity and light intensity of the plant are screened, and the constructed LSTM neural network model is used. Training is carried out to compare the measured actual values with the predicted values. The results show that the prediction of plant electrical signals based on long-term and short-term memory neural networks is feasible. This method can be used as a new artificial intelligence method in greenhouses. Or an important indicator of the establishment of a plant factory.