Self-Regulation of Spectrum and Photon Densities of LED Lighting System Using Control Algorithms

Tan, Marcus Yen Wai (2020) Self-Regulation of Spectrum and Photon Densities of LED Lighting System Using Control Algorithms. Final Year Project (Bachelor), Tunku Abdul Rahman University College.

Text Marcus Tan Yen Wai.pdf Restricted to Registered users only Download (10MB)

Abstract

With the issue of shortage of land for agriculture and increasing demand of food supply from the society, controlled environment agriculture, CEA has to be realised in order to address and resolve these issues. The advancement of technology has allowed the agriculture industry to flourish as CEA is able to aid in achieving maximum crops yield by utilising shorter time for plants to grow and mature. The research project proposes the implementation of artificial intelligence into a LED lighting system which allows the optimization of environmental parameters for crops growing, specifically the intensity of light. As light source is an essential for the development and growth of plants, being able to optimize the amount of light source exposed would not only promote plants growth, but prevents leaf scorching phenomenon to occur. The research project has the objectives of providing control and data acquisition for users remotely, regulation of the intensity of light source, and to establish a conducive growing environment for plants. All these would be achieved by creating a lighting system which implements neural network for optimization and a mobile application for the ease of control and data acquisition for users. The back-propagation neural network was implemented as the control algorithm for the optimization of light intensity where environmental parameters such as temperature, humidity, and soil moisture are made inputs for the prediction model. By a series of training and testing process conducted, the learning rate, epoch and number of neurons in the hidden layer were identified where the trained neural network achieved an accuracy of 93.8095%. With the respective trained weights, the prediction model then obtains environmental parameters from ThingSpeak platform where the prediction process is conducted and the output will be communicated to the lighting system through ThingSpeak where necessary changes in the light intensity would be made. The research project has realised the functionality of a LED lighting system in contributing to modern agriculture and with the implementation of back-propagation neural network, the prediction of light intensity would greatly contribute to the reduction in leaf scorching phenomenon in crops.