Multi criteria decision making with machine-learning based load forecasting methods for techno-economic and environmentally sustainable distributed hybrid energy solution

Abstract

Global energy transition demands increasing renewable share in the power mix. Intermittency of resources and the variability of load are the critical challenges of distributed renewable energy systems. Uncertainty of load affects the accurate capacity determination of hybrid energy systems. It creates difficulties to decide the expected cost of electricity. An accurate forecasting may minimize the future load uncertainty. The study integrates accurate load forecasting with subsequent techno-economic and environmental impact assessment (by full life cycle assessment). The proposed methodology helps to determine accurate capacity of the components of energy combination which improves economic and environmental performance. It will enhance the overall sustainability of such energy systems. Predictions of three different forecasting methods: one classical, one hybrid and another fuzzy-based, are compared for determining the most accurate forecasting by minimizing the mean absolute percentage error. This accurate forecasted data is used for better estimation of capacities of component systems for uninterrupted power. The performance indicators for overall sustainability may not converge to the same optimum solution. A multi criteria decision making approach is then adopted. It decides the finally acceptable optimum solution on the basis of the equal weights of different performance criteria. The robustness of this obtained solution is evaluated. Results show that the fuzzy-based forecasting method is the most accurate one for this integrated methodology. The mean absolute percentage error for this forecasting is 6.3% and 3.2% lesser than those of the classical and hybrid forecasting methods respectively. The integrated methodology shows that the combination of photovoltaic-diesel generator-lead acid battery is the acceptable optimum solution with maximum overall sustainability. This combination has a cost of electricity-0.100$/kWh, net present cost-5374$, excess electricity-21.7%, human health- 1.14 DALY, ecosystems- 0.002 species.yr and resources- 386120 USD2013 for sustainability.