Classification of Organic and Inorganic Waste Types Based on Neural Networks

Abstract

Garbage is the   residue of unused industrial production and household consumption. In Indonesia, waste is divided into 2 types, namely organic and inorganic waste. The two types of waste can be recycled in diverse ways, so they must be separated. So far, it is often difficult for the community to sort waste. This paper presents the process of recognizing and sorting waste automatically by utilizing Artificial Intelligence technology, especially Artificial Neural Networks (ANN). The ANN architecture used in this study consists of 4 layers. The number of neurons in each layer consists of 3 neurons in the input layer, 4 neurons in the hidden layer-1, 4 neurons in the hidden layer-2 and 1 neuron in the output layer. The ANN model that has been designed is trained, so that the best weight and bias model will be obtained, which in turn gives the ANN the ability to be able to sort waste properly. The best weights and biases will then be implanted into the Arduino UNO Microcontroller hardware. In this developed system, the microcontroller is given input obtained from 3 kinds of sensors, namely capacitive proximity, inductive proximity, and photodiode. While the input consists of 2 pieces of organic or in organic waste conditions. From the test results, it was found that the system has 100% training accuracy and 100% test accuracy.