The present invention provides a D-optimal inner table design based Taguchi experimental design method. The method comprises specific steps of: 1. providing basic experimental information including experimental responses, factors, inter-factor restrictions and the like according to experimental objectives, conditions and engineering experience; 2. selecting a proper uniform design table according to the number of noise factors and the number of proposed levels; 3. determining a relationship model between the responses and the factors, and the number of times of inner table experiments; 4. in a design area, providing an inner table design scheme with given experimental times by using the D-optimal design method, and completing the inner table design; 5. completing external table design by using an empirical distribution function based point equal-division method; 6. performing an experiment according to an experimental scheme, calculating a signal-to-noise ratio of mass properties by using performance indexes obtained from the experiment, and calculating the optimal level of a controllable factor that enables the signal-to-noise ratio to be maximized; 7. determining the optimal level of a stable factor by a sensitivity analysis method; and 8. changing and adjusting the factors to determine an optimal factor level combination.