Abstract: To solve the problems of hot cracks, pores and softening of welded joints in the traditional welding process of 7 series super hard aluminum, the mechanism of nonlinear relationship between welding process parameters and the strength of welded joints during vibration welding process is studied in the paper. The nonlinear time series prediction model of 7075 super hard aluminum vibration welded joints with process parameter measurement data. Used the parameter measurement data of 7075 super hard aluminum vibration welding process, the time series of welding process parameters is established, and the phase space reconstruction parameters and deterministic test methods of the welding process system are established based on time series. The artificial neural network model of the phase space phase point evolution trajectory is established in the basis of the nonlinear relationship between the phase evolution trajectory of the reconstructed phase space and the strength parameters of the welded joint. The model is used to calculate the physical parameters such as the elongation after fracture, tensile strength, hardness, weld reinforcement, maximum coarseness of the crystal, and the number of crystal grains. A series of welded joint strength tests for the established model show that the prediction results of this model can meet the engineering practical value.