Application of modified particle swarm optimisation on forecasting diffusion of mobile internet

In order to make accurate forecasts of mobile internet diffusion trend, this paper proposes a method which is based on modified bass innovation diffusion model in which the values of three parameters change over time. A novel particle swarm optimisation (PSO) algorithm is introduced to find the most precise parameters. This algorithm employs opposition-based learning strategy during the stage of initialisation and execution. The application of the index of population density helps determine the convergence status of population, and inertia weight is adjusted dynamically according to the value of population density. When the algorithm is trapped into local optima, the combination of Cauchy mutation and Gaussian mutation is applied on the best particle. The results demonstrate good performance of the novel algorithm on convergence accuracy and convergence velocity and the modified Bass model has the capability to forecast the diffusion of mobile internet more accurately.