The output signal of pressure sensor installed in the dynamic weighing system for loader contains strong vibration, noise, nonlinear signal. The accuracy of the dynamic weighing system is closely related to the pressure signal. An empirical mode decomposition (EMD) algorithm is proposed to preprocessing the signal contaminated. The real weighing signal is filtered out. a new method based on neural network is used to predicate the nonlinear output. in order to solve the problem that it was easily to sink into the partial minimum , the genetic algorithm was put forward .The emulation analysis and the results show that by using the above method, measure precision within 1% can be obtained.