Identification of motor cylinder pressure mainly refers to the identification of dynamic loading of structure through assessing the response of structures (e.g., movements, speed, acceleration and response) and according to dynamic characteristics as well. This is an approach to inverse problems in dynamic structure and is called identification of structural loads, aiming at achieving the model of its system information as to an unchanging system of single input and output. Its inverse \(H^{ - 1} (f)\) can be obtained once after achieving FRF (frequency response function). Based on the response, loading can be identified after distinguishing the loading spectrum and conducting a Fourier converse transformation. The under problems will be researched: Identification of Cylinder Pressure of Motor, Measures on Improving Accuracy Identification of Cylinder Pressure, Coherent amendment of the frequency response function, Inverse Spectrum Level and Smoothing of FRF. These problems can be solved by inverse spectrum windowing with nonlinear inverse filtering. However, the types and width of windows are difficult to solve and there is a great amount of relative calculation. As a matter of fact, the key reason why the inverse method of frequency function affects the results of loading identification is that there are a number of sharp low points in the amplitude of frequency function and these low points will be amplified, resulting in serious distortion of inversion signals. As long as the amplitude of the transfer function is smoothened, identification results can be improved.