Selection of secondary variables is an effective way to reduce redundant information and to improve efficiency in nonlinear system modeling. The combination of Kernel Principal Component Analysis (KPCA) and K-Nearest Neighbor (KNN) is applied to fault diagnosis of bearing. In this approach, the integral operator kernel functions is used to realize the nonlinear map from the raw feature space of vibration signals to high dimensional feature space, and structure and statistics in the feature space to extract the feature vector from the fault signal with the principal component analytic method. Assessment method using the feature vector of the Kernel Principal Component Analysis, and then enter the sensitive features to K-Nearest Neighbor classification. The experimental results indicated that this method has good accuracy.

The post-evaluation of electric power construction project is an important link, which is necessary in electric grid projects cycle. An advanced method of post-evaluation is proposed by combining Rough Set (RS) theory and Grey System (GS) theory, defined as Grey Rough Set (GRS). According to post-evaluation project experiences, this paper provides the method of index screening by attribute reduction and weight decision.

<p>The characteristics of quasi-zero stiffness(QZS) system with nonlinear positive and negative stiffness is researched. A modified QZS model with nonlinear spring element is established and the stiffness curves are obtained based on the analysis of relationship between spring force and displacement. A non-dimensional form of QZS is deduced to discover its essential laws, and simulation is presented with different nonlinear springs. Then the force transmissibility of QZS is verified with the experiment, which shows that the QZS isolation performance is better than the linear one in the low frequency band, and there exists no resonant peak in this system.</p>

Rotor rub-impact fault may be diagnosed through several kind signal such as rotor vibration, stator vibration and rotor transient speed, and every signal include fault features of different side, so it is possible to improve diagnosis successful probability by mul-information fusing method. The fault identifying frame and combination diagnosis rules are determined using stator vibration and rotor transient speed signals. It is adopted to determine mass function by the S-function, and the deducing method is put up. After peak value of stator resonance demodulation and rotor transient speed fluctuation amplitude information are fused, the method is applied to diagnose rotor rub-impact fault successfully.

For the purpose of improving adaptive performance of chaotic signals de-noising with wavelet transform, a method of Memetic-algorithm-based adaptive wavelet de-noising (MAWD) is presented. The MAWD based on generalized cross validation (GCV) is competent to obtain the global optimum thresholds and to raise the efficiency of adaptive searching computation. The de-noising results of simulative Lorenz time series are presented. The results show that the chaotic signals de-noised by MAWD can remove the white noise more effectively than the signals de-noised by using standard soft threshoding method (STM) and genetic-algorithm-based adaptive wavelet de-noising (GAWD), and the advantages are more apparent under the condition of lower SNR. The Lorenz time series with lower SNR de-noised by MAWD and GAWD respectively are predicted by Volterra adaptive filters, and the results show that the prediction absolute error of Lorenz time series de-noised by MAWD is nearly nine times smaller than that by GAWD. This method has a promising prospect in practical Chaotic signals de-noising.