DESIGN AND IMPLEMENTATION OF PID NEURAL NETWORK FOR AIR-FUEL RATIO CONTROL SYSTEM ON 2-STROKE INTERNAL COMBUSTION ENGINE

• Main Author: MADE ASWIN NAHRENDRA - NIM : 13214023, I
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/27991
• Institution: Institut Teknologi Bandung

<p align="justify">In this research was developed an Engine Control Unit (ECU) based on an Electronic Fuel Injection (EFI) to control the Air-Fuel Ratio (AFR) of a 2-stroke engine with the capacity of 157 cc. The EFI is controlled by implementing a PID Neural Network (PIDNN) algorithm on the microcontroller. PIDNN algorithm is adequate in this research because a black-box identification is applied for the engine’s plant, and theoritically the system’s parameter will change non-linearly based on engine’s temperature and angular speed. The ECU will monitor the engine’s condition periodically for every engine’s cycle . One of the measured quantity is AFR from the lambda sensor. Then, the system will process all of the data to calculate the new weight of the PIDNN and the needed amount of fuel injected in the next cycle so that the desired AFR will be achieved. In this reasearch, the updating computation for the PIDNN is done by using the online training method, which is every received data will become a new input dataset for the PIDNN to learn and updating the network’s weight.<p align="justify">