The process of erosion, sediment transport, and sedimentation commonly causes various sediment disasters. Because the formation of a sediment disaster is a complicated system evolution process, it is difficult to model the process according to hydrodynamic mechanisms. Models relating to statistics and based on the dynamic information of sediment disasters, such as experiential models and normal neural network models, mostly are built on the assumption that the process of a sediment disaster is steady state, and the dynamic characteristics cannot be taken into consideration effectively, especially the lag phenomenon, so that the accuracy of model results is not satisfactory. The main problem is that the modeling methods do not match up to the dynamic sediment-disaster process well, so it is important to develop new modeling methods to simulate the dynamic characteristics of the process. Time series models are commonly used to simulate dynamic system processes and the artificial neural networks are widely used in many fields to model nonlinear systems. In this paper, a new model based on the dynamic characteristics of sediment disasters and the integration between neural network and time series models is presented. The new model includes a time series layer, input layer, hidden layer, output layer, and layer of lag effect parameters. The training algorithms of the hybrid networks and the calculation algorithms of the lag effect parameters are put forward according to the characteristics of the dynamic information. Two models, with a single input layer and single output layer and a multiple input layer and single output layer, respectively, were built using the new method. Compared with a regular neural network model, the complexity of the network topological structure is reduced and the results fit well with the lag behavior of the dynamic process. Therefore, the proposed method is valuable to build new models of the dynamic information of the sediment disaster process. The process of erosion, sediment transport, and sedimentation commonly causes various sediment disasters. Because the formation of a sediment disaster is a complicated system evolution process, it is difficult to model the process according to hydrodynamic mechanisms. dynamic information of sediment disasters, such as experiential models and normal neural network models, mostly are built on the assumption that the process of a sediment disaster is steady state, and the dynamic characteristics can not be taken into consideration effectively, especially the lag phenomenon, so that the accuracy of model results is not satisfactory. The main problem is that the modeling methods do not match up to the dynamic sediment-disaster process well, so it is important to develop new modeling methods to simulate the dynamic characteristics of the process. Time series models are commonly used to simulate dynamic system processes and the artificial neural n e this new model includes a time series layer, input layer, hidden layer, output layer, and layer of lag effect parameters. The training algorithms of the hybrid networks and the calculation algorithms of the lag effect parameters are put forward according to the characteristics of the dynamic information. Two models, with a single input Compared with a regular neural network model, the complexity of the network topological structure is reduced and the results fit well with the lag behavior of the dynamic process. Therefore, the proposed method is valuable to build new models of the dynamic information of the sediment disaster procesYes.