Towards Detecting Red Palm Weevil Using Machine Learning and Fiber Optic Distributed Acoustic Sensing

Biwei Wang,Changyuan Yu,Islam Ashry,Yousef Al-Fehaid,Abdulmoneim Al-Shawaf,Boon S Ooi,Tien Khee Ng,Yuan Mao

doi:10.3390/s21051592

Abstract

Red palm weevil (RPW) is a detrimental pest, which has wiped out many palm tree farms worldwide. Early detection of RPW is challenging, especially in large-scale farms. Here, we introduce the combination of machine learning and fiber optic distributed acoustic sensing (DAS) techniques as a solution for the early detection of RPW in vast farms. Within the laboratory environment, we reconstructed the conditions of a farm that includes an infested tree with ∼12 day old weevil larvae and another healthy tree. Meanwhile, some noise sources are introduced, including wind and bird sounds around the trees. After training with the experimental time- and frequency-domain data provided by the fiber optic DAS system, a fully-connected artificial neural network (ANN) and a convolutional neural network (CNN) can efficiently recognize the healthy and infested trees with high classification accuracy values (99.9% by ANN with temporal data and 99.7% by CNN with spectral data, in reasonable noise conditions). This work paves the way for deploying the high efficiency and cost-effective fiber optic DAS to monitor RPW in open-air and large-scale farms containing thousands of trees.

Highlights

The date palm is a high-value fruit crop that provides healthy nutrition security to millions of people around the world [1]
We recently reported a solution of using a fiber optic distributed acoustic sensor (DAS) for the early detection of red palm weevil (RPW), such that a single optical fiber is noninvasively wound around the palm trees to possibly scan a large-scale farm within a short time [11,12]
To pave the way for utilizing the fiber optic distributed acoustic sensing (DAS) to monitor real farms, here, we introduce neural network-based machine learning algorithms to classify healthy and infested trees, based on the data collected by a fiber optic DAS

Summary

Introduction

The date palm is a high-value fruit crop that provides healthy nutrition security to millions of people around the world [1]. Existing acoustic detection technologies rely on inserting acoustic probes within the individual tree trunks and building a wireless network to communicate with the sensors [8] It is cost-ineffective to assign a sensor per tree, especially for vast farms containing thousands of trees. We recently reported a solution of using a fiber optic distributed acoustic sensor (DAS) for the early detection of RPW, such that a single optical fiber is noninvasively wound around the palm trees to possibly scan a large-scale farm within a short time [11,12]. Identifying infested trees in open-air farms, where the optical fiber might be subjected to harsh environmental noises, would require a more advanced signal processing technique to classify the larvae sound and other noise sources. This work would be highly beneficial towards the future deployment of the fiber optic DAS for the early detection of RPW in vast real farms

Experimental Design

Investigating the Impact of the Noise Sources on the DAS System

Classifying Infested and Healthy Trees Using Machine Learning Methods

Findings

Discussion

Conclusions