Optimizing Spark Ignition Engine Performance and Emission Control with Spike Neural Networks Using Gasoline–Ethanol Blends

This study introduces a novel approach utilizing Spike Neural Networks (SNNs) to analyze the performance and emissions of pollutants of a 4-stroke Spark Ignition (SI) engine running on ethanol–gasoline blends. By dynamically adjusting synaptic weights based on various application data, the SNNs technique aims to enhance the solution model for improved exploration and exploitation. Isoamyl alcohol is utilized at high Compression Ratios (CR) in SI engines to boost performance and reduce emissions. The main objectives of the SNNs approach are to evaluate waste products as potential alternative fuels, improve the performance of the engine and lower emissions. Using a variety of gasoline–ethanol combinations and speed conditions as input data, the SNNs model predicts relationships between braking power, Brake-Specific Fuel Consumption (BSFC), thermal and volumetric efficiency, torque and components of emissions. Performance evaluation is conducted using MATLAB, comparing against existing methods like Heap-based Optimizer (HBO), Color Harmony Algorithm (CHA) and Random Forest Algorithm (RFA). Results show low Mean Relative Errors (MRE) ranging from 0.46% to 5.57% and relatively low Root Mean Square Errors (RMSE) compared to existing approaches.