ARTIFICIAL NEURAL NETWORKS FOR PREDICTING THE NUMBER OF FIELD CROP PESTS

Abstract

Every year, farms face the problem of ensuring the necessary development and growth of field crops due to the high probability of field crops being affected by certain types of pests. Pests can significantly impair the development of crops if their population is not controlled. This will reduce the harvest. To ensure a certain level of field crop production, it is necessary to take a series of measures to reduce the risk of harvest losses and optimize the costs of protecting plant growth. A key element of effective farmland management is the reliable prediction of the number of pests using artificial neural networks and their appropriate configuration. This approach will reduce harvest losses and preserve the ecosystem of a particular region. Reliable forecasting of pest numbers is guaranteed to create conditions for minimizing the cost of growing crops. However, machine learning can only be implemented if there are relevant results of monitoring the number of pests and the factors that influence changes. These factors include solar activity, temperature, and humidity. Such studies were conducted and samples were formed. Neural networks of different structures were used for forecasting, such as the radial basis function and the multilayer perceptron. The results of the forecasting show a sufficiently high accuracy, which will significantly improve production efficiency.