Abstract: Nonlinear forecasting of sea clutter is important field of radar signal processing. Neural network has the advantage of approximating the nonlinear function, applies to time series forecasts of sea clutter . With that in mind, this paper proposes weak targets detection based on forecasting error of radial basis function with regression weight neural network under sea clutter background, to achieve the result that effectively detect weak target signal under complex clutter background. And, simulates target echo signal detection progress of high resolution radar in this method under background of sea clutter. The simulation results show that this way can effectively detect target signal in relatively low signal to clutter ratio, and the detection performance of this method is superior to the detection performance of the method with radial basis function neural network. This method is valuable to the research of weak targets detection under complex clutter background.