This paper proposes a weighted two-box model to solve the behavior modeling and distortion compensation problem of broadband RF power amplifier.The model

including the memoryless sub-model and memory sub-model

is used to model the static nonlinear and dynamic nonlinear of broadband power amplifiers.First of all

the input signal amplitude is compressed in weighted memoryless sub-model

which can effectively reduce the fitting error of the static distortion model.And the output value of the sub model is decompressed to ensure the saturation driving performance of the amplifier.Second

the weight memory sub-model is constructed

and is used to adjust the low and high power dynamic distortion sub model dynamically based on the weight function related to signal amplitude.The experimental results show that under different signal driving conditions

the proposed method can reduce the computational complexity while maintain better modeling accuracy and distortion compensation capability.