| Bioelectricity Generation Using Algal Fuel Cells: A Machine Learning Approach |
| Paper ID : 1165-IUGRC6 (R2) |
| Authors |
|
Montaser Bellah Yasser *1, Omar Yasser2, Habiba Hany2, Abdullah Ayman2, Habiba Emad2, Mariam Hossam2, Abdelrahman Yasser3 1Department of Biochemical Engineering, Faculty of Energy and Environmental Engineering, The British University in Egypt (BUE), El-Sherouk City, Cairo, Egypt 2Biochemical Engineering Programme, Faculty of Energy and Environmental Engineering, The British University in Egypt (BUE), El-Sherouk City, Cairo, Egypt 3Biochemical Engineering Programme, Faculty of Energy and Environmental Engineering, The British University in Egypt (BUE), El-Sherouk City, Cairo, Egypt |
| Abstract |
| Microalgae are regarded as one of the most promising and futuristic solutions for green and sustainable generation of bioelectricity. In this study, machine learning (ML) algorithms were applied on a constructed dataset of 648 data points to predict optimal cultivation parameters required for achieving high microalgal biomass concentration, which is a critical factor for algal fuel cell (AFC) efficiency. Upon implementation of decision tree (DT) algorithm, results showed significant classification for microalgal biomass concentration in relation to target microalgal parameters. Microalgal biomass concentrations were recorded at highest values when microalgae were Monoraphidium and Nannochloropsis. N and light levels were also contributing factors to biomass concentration. Bio-oil was at its highest values when pH at was high at ≥7.1 and microalgae were Chlorella, Desmodesmus, and Monoraphidium. Also, light, K, and biomass concentration were contributing factors for enhancing bio-oil content. Microalgal samples of Chlamydomonas, Desmodesmus, Monoraphidium and Scenedesmus required average light intensity levels to achieve their highest bio-oil content. Temperature at 27 °C was efficient for Chlorella, Ettlia, Monoraphidium and Nannochloropsis samples to achieve high biomass concentrations. pH levels were recorded optimal when microalgal bio-oil content was at highest levels ≥21 %w/w. Prediction levels were validated using RMSE and R2 and both showed accurate results using generated decision trees from assigned data partitions. |
| Keywords |
| Machine learning; Microalgae; Algal fuel cells; Bioenergy; Bioelectricity |
| Status: Accepted |