Abstract: For the problem that the input and output of real systems is a continuous process relative to time,this paper proposed a process neural network model for continuous function approximation.Using the nonlinear mapping ability of neural network,this model performs the continuous-mapping relation between input and output of system.Considering the computation complexity of process neural network,a group of function orthogonal bases are selected in input space.Input function and network weight functions are expressed as the expansion form of the function orthogonal bases.Thus operation of process neuron is simplified by using orthogonality of functions.The learning algorithm is given,and the effectiveness of the model and algorithm is proved by tertiary oil recovery process simulation of oil reservoir development.