This paper focuses on the control system performance in industrial control processes and evaluates the controller performance using the assessment criterion based on linear quadratic Gaussian (LQG). The LQG performance benchmark curve is determined by the numerical calculation method, which avoids the calculation of the complex interaction matrix. This method depends on the model-based steady-state optimization technique and combines the LQG benchmark and the performance assessment of the control system. The control performance of the process system under different control strategies is described by establishing a series of the steady-state optimization problems. Compared with existing assessment algorithms, our method provides a simpler and more effective method to evaluate the performance of IMC-PID control system for both model match and model mismatch cases. Finally, the effectiveness of the method is verified by the experiments on a heating furnace.