Combined forecasting tool for renewable energy management in sustainable supply chains

Abstract

Effective managing sustainable supply chain is on the most cutting-edge position of various organizations in the process of delivering products. There are many optimization technologies that have been applied in sustainable supply chain management, aiming to decrease the carbon emission level of production process. Among these methods, designing reliable renewable energy forecasting approaches is conductive to power management and optimization in sustainable and circular supply chain. However, irregular and non-stationary fluctuations of wind speed is a major obstacle to optimize the applications of renewable energy in sustainable supply chain management. There are various approaches for wind speed prediction, yet most of them ignore the significance of hyperparametric selection and combined forecasting strategy, resulting in unsatisfactory prediction results. To remedy the drawbacks, a combined prediction framework is proposed, which uses grid search to select suited hyperparameters of sub-models and employs an improved intelligent optimization algorithm (ranking-based adaptive cuckoo search algorithm) to calculate the optimal weighting coefficient of sub-models. Nine datasets are collected to validate the proposed model and 15 benchmark models. The simulations revel that the proposed model yields the satisfactory prediction level, which precedes comparative models based on statistical test results. Hence, it is a valuable tool for decision makers to provide key reference information in sustainable supply chain management and optimization.