Use Plant Electrical Signals to Classify the Growth Status of Plants

In this paper, rape is used as the experimental object, and the plant electrical signals of two kinds of rapes with normal and abnormal growth states are collected by a biological function experimental device, and the collected data are labeled with labels N and A, corresponded to their growth states, and then plant electrical signals are filtered by Butter Worth low-pass filter and denoised by wavelet transform. Then the Instantaneous frequency and the Spectral entropy which are two features of the data are extracted after it is filtered and denoised. The plant electrical signals are divided into test sets (there are 248 abnormal signals and 248 normal signals) and training sets (there are 2236 abnormal signals and 2236 normal signals) after the data processing. The training sets are put into long short-term memory (LSTM) networks to train the classification model. And the classification accuracy of the trained model is 90.9274%, through the test sets test the LSTM model. It is concluded that the state of plant growth can be judged through the plant electrical signal. This provides a reliable basis that exploring the relationship between the state and the environment factor of plant growth, in order to provide the optimal growth environment for plants in the future work.