Anti-noise Performance Of Discrete Spectrum Correction Theories And Their Application On Engineering

The discrete spectrum correction theories can greatly improve the parameters estimation accuracy of a sinusoidal signal when the signal is free of noise. But when the signal is corrupted with noise, the estimation accuracy of the parameter estimator methods is discounted and even unavailable. It becomes an urgent issue that how to reduce the influence of noise on the estimation accuracy and consequently improve the anti-noise performance of the parameter estimators. Condition monitoring and diagnostics on vehicles in used is an effective method to reduce accidents and improve the vehicles operational performances. The accuracy and effectiveness of the vehicles performance diagnosis will be improved by the application of the parameter estimation methods on the engine speed measurement and rolling bearing failure diagnosis of automobile transmission gearbox.According to the characteristics of the energy centrobaric correction method and the FT continuous zoom method, the sources of error of the two parameter estimators are studied theoretically, and the improving algorithms or evaluation indicators are proposed accordingly. By the study of the influences of the window functions and the number of spectrum lines used in the estimation, the energy centrobaric correction method using four spectrum lines is proposed. Analytical expressions of the estimator variances for frequency, amplitude and phase correction in the presence of white Gaussian noise are derived in detail. As for the FT continuous zoom method, the influencing factors of estimation error with and without noise are compared. The conclusion is given that the maximum frequency estimation error of this method will not inversely proportional to the zoom multiple anymore when the signal is corrupted with noise. The maximum frequency estimation error is deduced under the white Gaussian noise background. The best zoom multiples under different signal-to-noise ratios are given accordingly. Based on the researches above, the performances of four commonly used parameter estimation methods are compared: the interpolation method, the phase difference method, the energy centrobaric correction method and the FT continuous zoom method. The conclusions provide the theory basis for the proper estimator and parameters selection in practical situations.The discrete spectrum correction theories studied are applied to the engine speed measurement and rolling bearing failure diagnosis of automobile transmission gearbox. In view of the disadvantages of the existing engine speed measuring system, a new engine speed measuring method, which based on the vehicle vibration and the discrete spectrum correction methods, is proposed. The new speed measuring method can recognizes the engine speed quickly and precisely and doesn't need the priori knowledge about the engine cylinder number. Both the engine test-bed experiments and tests on the vehicle in used all confirm the validity of the new method. By the foundation experiments of rolling element bearing, a set of effective rolling bearing failure diagnosis techniques based on the discrete spectrum correction theories are summarized, which improve the accuracy of rolling bearing failure monitor and diagnosis. The conclusions are applied in the automobile gear box failure diagnosis effectively.