In consideration of the shortage of water resources in Yantai, in order to maintain more reasonable exploitation and utilization of water resources, based on building RVA model to assess the influence of hydrological regime that Yan tai Dagujia river basin build conservancy project in recent years, combinating with BP neural network, reduce the runoff to the natural state without the influnces of the water conservancy project. Firstly, the results show that the construction of water conservancy projects like reservoir cause significant changes in the natural runoff, which month flow change degree is between 0.3 and 0.7 and limit maximum and minimum average flow changes degree is between 0.3 and 0.6, belonging to medium altitude change. Secondly, the results were analyzed from the RVA model, combined with BP neural network method to establish Dagujia river basin natural runoff reduction calculation model, through the establishment of the model reduction of Dagujia river basin natural runoff calculation, combined with the measured values to error comparison，found that the BP neural network is feasible，reliable and convenient. It is good for the water resources and hydrology engineering construction and provides a scientific basis for reasonable development and utilization of water resources in this region.