Wind speed forecasting plays a crucial role in power system operations, power grid security management, and in the electricity market. This is of great significance for society and is still a challenging task. However, most previous studies were based on simple data preprocessing and they only focused on improving either the forecasting accuracy or stability while ignoring the significance of improving these two aspects simultaneously, which will lead to poor forecasting performance. Therefore, a hybrid forecasting system based on a newly developed algorithm proposed herein referred to as the multi-objective sine cosine algorithm (MOSCA)-is developed, which includes four modules, specifically, data preprocessing, optimization, forecasting, and evaluation module. For this system, a modified data decomposition approach is successfully developed to further improve its forecasting performance. In addition, a hybrid wavelet neutral network (WNN) based on MOSCA is developed to obtain high accuracy and strong stability simultaneously. Case studies utilizing eight wind speed datasets collected from two wind farms are performed as examples to analyze the performance of the developed forecasting system. The results clearly reveal that the developed forecasting system is superior to all the considered models herein in terms of both accuracy and stability. As a result, it is concluded that the proposed approach can be an efficient and effective technique for wind speed forecasting.