CONSTRUCTION OF NOVEL NEURAL NETWORK MODEL BASED ON NEUROEVOLUTIONARY ALGORITHM FOR IMPROVING THE PERFORMANCE OF CROP YIELD PREDICTION

CONSTRUCTION OF NOVEL NEURAL NETWORK MODEL BASED ON NEUROEVOLUTIONARY ALGORITHM FOR IMPROVING THE PERFORMANCE OF CROP YIELD PREDICTION

E.Kanimozhi

Ph.D. Research Scholar,

Department of Computer Science, School of Computing Sciences,

Vels Institute of Science, Technology & Advanced Studies (VISTAS), Chennai, India.

Kanimozhi135@gmail.com

Dr.D.Akila,

M.C.A.,M.E.,M.Phil.,Ph.D.,Post

Doctoral Fellow(LUC-Malaysia)

Former Associate Professor,

Department of computer Application, School of Computing Sciences,

Vels Institute of Science, Technology & Advanced Studies (VISTAS), Chennai, India

akiindia@yahoo.com

Dr.R.Priya

Professor

Department of computer Application, School of Computing Sciences,

Vels Institute of Science, Technology & Advanced Studies (VISTAS), Chennai, India

Abstract

The development of machine learning combined with high speed computing power has opened up several multidisciplinary opportunities. Here, we provide a cutting-edge machine learning technique for forecasting agricultural production. The performance enhancement in agricultural yield prediction was proven by the classification approach utilising machine learning algorithm. It depends on the local soil, water irrigations, and meteorological variables that are related to data on climate change. Here, we've provided an example of how to use an ANN-based neuroevolution model to forecast wheat crop production. From June through September, crop yield projections are taken into account. The yields are generated based on climate and fertilizer usage statistics. The capacity to forecast how a season will evolve based on weather information has significantly improved. As a consequence, the model's results help with decision-making before planting a wheat crop. The findings are more helpful for making decisions, planning wheat planting, and performing other agricultural duties at various stages of the wheat crop's development. The same model may be utilised to predict a variety of agricultural data, such as disease and weather predictions.

In this work, a neural evolution system for forecasting different plant illnesses was created. The performance increase in plant disease diagnosis may be proven using the algorithms of machine learning that support several categorization strategies. Disease prediction is based on local soil conditions and meteorological variables that are related to data on climate change. Here, we've shown how to use an ANN-based neuroevolution model to forecast a variety of plant illnesses. Different causes and the sort of illness that may harm various plants at various times of the year are anticipated. As a result, the results of the suggested model help with decision-making and preventative measures for plant diseases. The results may be used to make decisions early on about plant disease prevention as well as numerous farm operations at different stages. The technique used for forecasting different agricultural data, including crop yield and weather prediction, is also utilised here.

The paper's primary goal is to use a cutting-edge neuroevolutionary algorithm to analyse agricultural yields more accurately. One of the biggest issues of the twenty-first century for sustainable and nutrition security for a population increase is climate, which is putting the world's food security in jeopardy now. Water scarcities, exorbitant expenses brought on by supply and demand, and unpredictability of the weather are among the present problems. Farmers were exhorted to enhance their farming practices. The main causes in agriculture are the shortage of crops that can be grown using conventional agricultural methods, the uncertainty surrounding climate change, the absence of irrigation infrastructure, the paucity of land, and the scarcity of crops. To anticipate crops, many machine learning methods like perception, divisions, regress, and aggregation are applied. Some of the mathematical techniques employed to carry out the prediction include artificial network, vectors support networks, linear and linguistic matrices, decisions trees, and Bayesian intelligence. It is clear that choosing the appropriate methods for sample design is a difficult task for academics. The ability of machine learning systems to forecast agricultural yields has been investigated. The techniques created to provide predictions using machine learning. The outcome demonstrates that, when compared to other approaches, the neuroevolutionary algorithm is more successful. We modified this method to learn graphical images of plant diseases and demonstrate that going to discriminate infected areas of a plant can be satisfactorily located and emphasised for disease identification. This work was inspired by latest projects on multi-organ recognition and classification that has demonstrated the capacity of an interest Rnn Model (RNN) to locate applicable regions of various plants without even any prior human annotation.