This paper presents a novel parallel neural network clustering approach based particle swarm optimisation (PSO) for the large spatial database suitable in data mining. A large number of pieces of evidence are clustered into subsets, a nonlinear connection function is adopted, the centre of connection function is regarded as a particle, a PSO approach that can describe the group intelligence behavior use to solve the optimisation problem. A numerical example has been used to illustrate the effect of the algorithm on the load characteristics clustering of power system. Many sets of load data measured from a power system have also been dealt with using the method. The results of the study clearly indicate that the proposed method is very useful to load characteristics clustering for power system