Mid-Term and Short-Term Load Forecasting Using Rough Neural Networks and Grasshopper Mutation

 Changes in consumption load in power networks are inevitable with the increasing population and the growth of industrial societies, and it is necessary to forecast the required load of networks. The forecasting of the hourly load in the medium term can be a good measure for evaluating load and energy. This process will also be a good example for short-term forecasting. In this paper, a new method is presented for hourly load forecasting in the short and medium-term load using rough neural networks and the grasshopper mutation algorithm. An improved rough neural network is also presented. Rough neural networks are a type of neural structures designed based on rough neurons. A rough neuron is a pair of neurons that are conventional to upper and lower boundary neurons, similar to multilayer neural perceptron networks. The rough neural network can be trained using a descending-based gradient error post-propagation algorithm. However, this algorithm has certain problems such as being trapped in local minima, a shortcoming which has been overcome in this paper with the help of the grasshopper jump algorithm. The Dobie power global network is proposed to apply the rough neural network and its combination with the grasshopper mutation algorithm to simulate the proposed method in daily load forecasting, and the results indicate the success of the proposed methods.