The Time Series Prediction Algorithm Based on Improved GSA-ELM

Aiming at the random selection of the input weight and threshold of the extreme learning machine (ELM), the stability is poor. This paper proposes an extreme learning machine model optimized based on the gravitational search algorithm (GSA). The chaotic sequence is used to initialize the population position, and Levi's flight updates the object position to avoid falling into a local optimum, and improves the gravitational search algorithm. Then it optimizes the parameters of the extreme learning machine. Results show that the model has good prediction effect on time series.