A novel seasonal grey prediction model with external impacts and its application

Purpose Seasonal data are characterized by multiple complex features such as seasonal fluctuations and stage differences, which presents challenges for the scientifically sound construction of its predictive models. This article proposes a novel seasonal grey prediction model with a weighted fractional order accumulation operator for accurate prediction of China’s natural gas production. The grey models proposed in this article optimize the accumulation method of the traditional grey model, and flexibly adjust the generation sequence through the two parameters introduced, so as to explore the internal law of the data information at a deeper level, and achieve the purpose of improving the model accuracy. Design/methodology/approach Firstly, based on the seasonal fluctuations of the raw data, the raw data is divided into four seasonal groups. Secondly, when an external disturbance affects the system, the classic average weakening buffer operator is used to weaken its effects. Then, the new information accumulation parameters ? and the fractional-order cumulative generating operator parameter r, are optimized using the particle swarm optimization (PSO) technique. Findings The experimental results show that the new grey prediction model (DGGM(1,1,λ,r)and DGGM(1,N,λ,r)) perform better than other models. This model is more suitable for solving the prediction problem of seasonal fluctuation systems with external disturbances. Practical implications The seasonal grey model proposed in this article is suitable for predicting data sequences with seasonal changes while considering the impact of external shock. Two models proposed in this article are used in two cases, i.e. estimating China’s natural gas production and predicting total electricity consumption in Nanjing city. Accurately predicting to provide a decision-making basis for the Chinese government to formulate energy policies, plan energy layouts, and adjust industrial structures. Originality/value Simultaneously considering seasonal changes and external shock disturbances, the data grouping approach is used to identify seasonal changes. This article proposes a new weighted fractional order accumulation operator that combines the new information accumulation operator and the fractional order accumulation operator. A novel seasonal grey prediction model with external impacts is proposed to improve the predictive performance of the system. The PSO algorithm is used to solve the parameter optimization problem of the model.