# AI based optimization of injection pressure for hydrogen and spirogyra biodiesel dual fuel engine to enhance combustion performance and emission characteristics

## TL;DR

This study uses AI to optimize injection pressure in a dual-fuel engine running on hydrogen and biodiesel, improving performance and reducing emissions.

## Contribution

The novel contribution is the application of machine learning to optimize injection pressure for a hydrogen and biodiesel dual-fuel engine.

## Key findings

- An injection pressure of 220 bar improved combustion efficiency and reduced CO and HC emissions compared to higher pressures.
- The decision tree machine learning model achieved high accuracy in predicting engine performance metrics.
- Smoke emissions were significantly reduced at 220 bar compared to diesel and other fuel combinations.

## Abstract

The principal objective of this research is to employ modern machine learning techniques to optimize high-pressure biofuel injection strategies for sustainable energy applications. An engine powered with biofuel and hydrogen (H₂) under dual-fuel (DF) mode was tested under a varied fuel injection pressure range from 180 to 240 bar for optimization and modeling. The results demonstrate that an injection pressure of 220 bar produces enhanced engine performance. At this pressure, enhancements were noted in combustion characteristics, efficiency, and emission levels. The ignition delay (ID) at 220 bar injection pressure was 9.4% longer than at 240 bar injection pressure. The 220 bar IP mix demonstrated reduced peak cylinder pressure (PCP) and heat release rate (HRR) compared to the 240 bar. A 12.4% rise in brake-specific fuel consumption (BSFC) was observed at 220 bar inlet pressure. Nevertheless, although brake thermal efficiency (BTE) increased with increasing injection pressure (IP), the increase at 220 bar was somewhat less than that at 240 bar. Despite elevated nitrogen oxide (NOx) emissions with the 220 bars compared to pure diesel, carbon monoxide (CO) and hydrocarbon (HC) emissions were markedly decreased. Smoke emissions were reduced with the 220 bars in comparison to diesel and other fuel combinations. Three machine learning models were employed to establish a predictive control framework. The decision tree (DT) model had the greatest accuracy, with R² values of 0.9792 for PCP and 0.9710 for HC, alongside near-zero MAPE for BTE and HC This study underscores the potential of AI-driven biofuel optimization for fostering sustainable transportation and renewable fuel strategies, paving the way for large-scale adoption of low-carbon, high-efficiency energy solutions.