Machine Learning in Cyber Physical Systems for Agriculture: Crop Yield Prediction Using Cyber Physical Systems and Machine Learning

Introduction

Industry 4.0, often known as the fourth industrial revolution, is being realized thanks to the advent of cyber-physical systems and precision agriculture (Shubham Tripathi and Manish Gupta, 2021). One of the current concerns in this field is agricultural cyber-physical systems (ACPS), composed of computer systems and the agricultural environment (Tariq Masoodab & Paul Sonntaga, 2020). The growth of systems has created a significant opportunity to improve the food supply chain system (FSC). It is imperative to humans and has a complex cycle that begins and encompasses farm produce. Logistics and distribution, retail, consumer communication, and trash management are all areas that need to be addressed. Furthermore, those cycles encounter various obstacles, mainly when using CPS technology (Goap et al., 2018). Precision agriculture, a management technique that uses conditions and specific information to precisely regulate production inputs such as soil and crop characteristics unique to each field region, is one of the most current problems in this sector (Amami & Smiti, 2017; Chen et al., 2015; Dadhich, Hiran, et al., 2021).

According to (Diale et al., 2019), precision agriculture strives to optimize production inputs in tiny areas of the field in real-time, such as water, fertilizers, herbicides, insecticides, and farm equipment. The Internet of Underground Things has lately enabled precision agricultural technologies (IOUT). IOUT stands for autonomous devices that collect relevant information about the Earth and are linked through communication. Information, transactions from the fields to the farmers, and decision-making systems are all aided by networking technologies (Dusadeerungsikul et al., 2019). IOUT has the advantage of real-time sensing by using deep learning algorithms with CPS (Mahrishi, M., et al.,2020). Sensing has aided in the adoption of precision agriculture technologies while also increasing the efficiency of agricultural production and operations. Essentially, changes in the intrinsic and extrinsic conditions of underground objects such as communication mediums in soil, seasonal changes, and crop growth cycles are caused by product modifications (Goap et al., 2018). The technology should isolate and adjust the conditions to provide farmers with a precise decision-making mechanism promptly. This paper presents an early machine learning approach to handle the adaptation problem. There are five dimensions to the machine learning framework, including data normalization and handling erroneous data for a decision-making system, time-series prediction, information fusion, and a classifier are used (Cui et al., 2021). The researchers tried with a dataset of fifteen sensors to get a more accurate assessment. Several attributes are included in the sensor data, including soil moisture, air temperature, soil temperature, wind velocity, and air pressure. This chapter is organized into five sections. The first section consists of the introduction of green yield prediction. The second part outlines the applications of cyber-Physical Systems for the agriculture domain. The third section explained the related work in the selected domain. The fourth section delineates the conceptual framework based on theories followed by experimental work. The last section assessed the result, analysis, and conclusion.