Identification and control of giant African snail in cauliflower farms by low frequency sound measurement technique and back propagation neural

Abstract

The Giant African snail is the world's largest, destructive and most damaging land pest in agricultural plantations. These pests are sometimes called as the natural enemies of farmers. High devouring capacity of these snails causes heavy damages to vegetation. In this work, various snails are collected and placed in the room temperature. The snails were feed with the consumables, such as cauliflower, tomatoes and soil, indeed the sound frequency is measured. In addition, the various activities also monitored and the respective frequencies are observed. For instance, the movement of a snail, retraction of the head in shell, head movements are recorded from the acoustic room with the recorder. However, monitoring the snail low level sound were challenging for recording. Thus the high precision device is required for recording the signals. The recording has been done for a number of days in order to evade uncertainty. The recorded accurate signals are chosen for the analysis. These signals are analyzed in the time and frequency domain. The statistical features are extracted from the signal. Further, the power spectral density, sound pressure level and equal loudness contour of the signals are calculated. The wide range of frequency of the snail of its various movements is determined. The extracted statistical features from the signals are fed into the neural network for training. Back propagation algorithm is used for classification along with the snail's signal, few noise signals also fed into the neural network to identify the effectiveness of network in the identification of the pest. It gives 99% efficiency in identifying the snail's signal. This work is also compared with capacitance based pest identification. This work has been experimentally validated with low-cost acoustic sensor that wirelessly communicates to the base, to monitor and control snail activity, the work suggested in this paper is a non-destructive form of pest identification which does not harm the crops as well as the environment.