The high-fidelity dynamic model is the basis for seismic margin analysis,seismic performance monitoring and continuous improvement of seismic design of in-service nuclear power valve piping system.However,its establishment process is faced with the problems of great difficulty and low efficiency.Based on the high-fidelity dynamic model of a practical valve piping system,the anti-resonance frequency characteristics of various piping systems are studied.The results show that the origin anti-resonance frequency and the resonance frequency appear alternately.In the low frequency region,the theoretical values of the origin anti-resonance frequency of various piping systems match the measured values well.At the connection position between the valve and the pipeline,different structural parameters have close sensitivity to the origin anti-resonance frequency and resonance frequency,so the measuring points should be preferentially arranged at these positions in practical engineering.Based on this theory,a finite element model updating method combining Gaussian Radial Basis Function-Response Surface Method and Improved Adaptive Genetic Algorithm is proposed with resonance frequency and anti-resonance frequency as the objective function.The example shows that the proposed method overcomes the difficulties of insufficient test data and slow convergence speed,and can identify the unknown structural parameters in nuclear valve piping with high precision and high efficiency,as well as establishes its high-fidelity dynamic model.The research results provide the possibility for further improvement of the economy and safety of seismic design of nuclear power valve piping system.