Optimizing adsorption efficiency: A novel application of SVM_Boosting_IGWO for methylene blue dye removal using low-cost fruit peels adsorbents

In this work, the potential for employing orange peels (OP) and potato peels (PP) as biosorbents to remove the methylene blue dye (MB) from aqueous solutions is studied. Several physicochemical methods, such as FTIR, SEM-EDX, X-ray diffraction, and pH point of zero charge measurement, were used to analyze the adsorbents. FTIR analysis revealed changes in peak intensities after dye adsorption. SEM analysis confirmed the presence of starch in the PP adsorbent, while no apparent pore structure was observed in the OP adsorbent. EDX analysis showed that carbon and oxygen were the main components on the surfaces of OP and PP. X-ray diffraction patterns indicated that both adsorbents were amorphous materials. The impact of different factors, including adsorbent dosage, contact time, temperature, initial dye concentration, pH and particle size, on the biosorption process was studied. Kinetic studies revealed that equilibrium was reached within a few minutes of contact, and the MB removal followed the pseudo-second-order model.Furthermore, a novel predictive model combining Support Vector Machine (SVM) with Boosting and the Improved Grey Wolf Optimizer (IGWO) algorithm was developed. The SVM-IGWO-Boosting model exhibited excellent performance in predicting methylene blue adsorption, demonstrating perfect correlation and low prediction error. The IGWO optimization approach effectively optimized the input parameters for the adsorbents, resulting in excellent agreement between experimental and predicted values. Moreover, OP showed higher efficiency in removing methylene blue compared to PP, with a maximum capacity of 111.75 mg/g for OP and 96.67 mg/g for PP using IGWO. The use of orange peels and potato peels as agricultural waste for methylene blue removal offers an efficient and sustainable solution. The SVM-IGWO-Boosting predictive model, in conjunction with the IGWO optimization approach, provides a promising tool for predicting and optimizing the adsorption efficiency of MB adsorbents. These findings present valuable prospects for real-world applications requiring accurate and reliable predictions.