Applying a multi-layer perceptron model for predicting gasification process outcomes

Recently, gasification processes have seen the integration of artificial neural networks and artificial intelligence-driven regression techniques. In the pursuit of efficient and environmentally responsible conversion of carbonaceous feedstock into valuable gases, this research presents a novel approach for predictive modeling. The primary driver behind the adoption of predictive models is the quest for cost-effective and highly accurate solutions. These models serve to streamline estimation procedures, adapt to diverse operational scenarios, improve safety measures, and support data-driven optimization, thereby addressing the complex challenges associated with gasification processes. This research is focused on utilizing the Multi-layer Perceptron (MLP) model, known for its proficiency in capturing intricate connections between input variables and gas production outcomes by leveraging historical data. Additionally, the incorporation of the Reptile Search Algorithm Optimization (RSA) and the African Vultures Optimization (AVOA) further enhances the model's predictive capacity, elevating its precision. The study marks a substantial advancement in optimizing gasification processes, providing a roadmap for more effective and sustainable conversion of carbonaceous feedstock. It serves as a testament to the potential of data-driven sustainability in this domain. The analysis conducted reveals the outstanding predictive capabilities of the MLRS model, situated in the third layer, particularly in forecasting Hydrogen (H₂). This model achieved an impressive R² value of 0.994 during the validation phase. Conversely, regarding Nitrogen (N₂) prediction, the MLRS model in the second layer consistently outperformed all other models, boasting an impressive R² of 0.997 during both the testing and validation phases. The superiority of the MLRS model over the conventional MLP model is further emphasized through the evaluation of various accuracy metrics, such as RMSE, MSE, and MARE. These results underscore the reliability and effectiveness of the MLRS model in the context of gasification process prediction.