Zhousuo Zhang

Research on bearing life prediction based on support vector machine and its application

Life prediction of rolling element bearing is the urgent demand in engineering practice, and the effective life prediction technique is beneficial to predictive maintenance. Support vector machine (SVM) is a novel machine learning method based on statistical learning theory, and is of advantage in prediction. This paper develops SVM-based model for bearing life prediction. The inputs of the model are features of bearing vibration signal and the output is the bearing running time-bearing failure time ratio. The model is built base on a few failed bearing data, and it can fuse information of the predicted bearing. So it is of advantage to bearing life prediction in practice. The model is applied to life prediction of a bearing, and the result shows the proposed model is of high precision.

A Non-Probabilistic Metric Derived From Condition Information for Operational Reliability Assessment of Aero-Engines

The aero-engine is the heart of an airplane. Operational reliability assessment that aims to identify the reliability level of the aero-engine in the service phase is of great significance for improving flight safety. Traditionally, reliability assessment is carried out by statistical analysis on large failure samples. Because the operational reliability of a specific aero-engine is an individual problem lacking statistical sample data, traditional reliability assessment methods may be insufficient to assess the operational reliability of an individual aero-engine. The operational states of the aero-engine can be identified by its condition information. Changes in the condition information reflect the performance degradation of the aero-engine. Aiming at the assessment of the operational reliability of individual aero-engines, a novel similarity index (SI) is proposed by analyzing the condition information from the fault-free state, and the current state. A condition subspace is first obtained by kernel principal component analysis (KPCA). Subspace similarity is then represented by subspace angles, i.e., kernel principal angles (KPAs). The cosine function is finally utilized as a mapping function to transform the subspace angles into a similarity index. The index can be used as a non-probabilistic metric for operational reliability assessment. Only the condition information is needed for computation of the similarity index, thus it can be performed conveniently for online assessment. The effectiveness of the proposed method is validated by three case studies regarding the health assessment of aero-engines subjected to system-level and component-level degradation. The positive results demonstrate that the proposed SI is an effective metric for operational reliability assessment of individual aero-engines.

Novel method for bearing performance degradation assessment – A kernel locality preserving projection-based approach:

Bearing performance degradation assessment is meaningful for keeping mechanical reliability and safety. For this purpose, a novel method based on kernel locality preserving projection is proposed in this article. Kernel locality preserving projection extends the traditional locality preserving projection into the non-linear form by using a kernel function and it is more appropriate to explore the non-linear information hidden in the data sets. Considering this point, the kernel locality preserving projection is used to generate a non-linear subspace from the normal bearing data. The test data are then projected onto the subspace to obtain an index for assessing bearing degradation degrees. The degradation index that is expressed in the form of inner product indicates similarity of the normal data and the test data. Validations by using monitoring data from two experiments show the effectiveness of the proposed method.

Source Contribution Evaluation of Mechanical Vibration Signals via Enhanced Independent Component Analysis

Extraction of effective information from measured vibration signals is a fundamental task for the machinery condition monitoring and fault diagnosis. As a typical blind source separation (BSS) method, independent component analysis (ICA) is known to be able to effectively extract the latent information in complex signals even when the mixing mode and sources are unknown. In this paper, we propose a novel approach to overcome two major drawbacks of the traditional ICA algorithm: lack of robustness and source contribution evaluation. The enhanced ICA algorithm is established to escalate the separation performance and robustness of ICA algorithm. This algorithm repeatedly separates the mixed signals multiple times with different initial parameters and evaluates the optimal separated components by the clustering evaluation method. Furthermore, the source contributions to the mixed signals can also be evaluated. The effectiveness of the proposed method is validated through the numerical simulation and experiment studies. [DOI: 10.1115/1.4005806]

Manifold subspace distance derived from kernel principal angles and its application to machinery structural damage assessment

Damage assessment of machinery structure is beneficial for identifying structural health states and preventing sudden failures. A novel scheme for damage assessment is presented by using manifold subspace distance in this study. Vibration response signals from the machinery structure are collected by accelerometers first, and feature matrices are extracted to characterize the acceleration response comprehensively. Thereafter, a manifold learning method, namely kernel locality preserving projection (KLPP), is performed to obtain manifold subspace from the feature matrix. KLPP is available to mine nonlinear information hidden in the feature matrix, which makes it different from the linear subspace analysis method. This merit enables KLPP to be more effective in exploring the intrinsic model of the feature matrix. Further, kernel principal angles that represent similarity between the manifold subspaces are calculated. Finally, a manifold subspace distance is derived from the kernel principal angles. This distance is an appropriate metric for measuring closeness or similarity between the subspaces embedded on a manifold. The distance between manifold subspaces from normal state and damage state of a machinery structure is defined as a damage index. Effectiveness of the proposed scheme is validated by two case studies with regard to damage assessment for different machinery structures. The results show that the defined damage index is not only sensitive to the occurrence of structural damage but also increases obviously with the increasing damage level. These positive results illustrate that the proposed scheme has promise for future performance and is a valuable method for damage assessment. (Some figures may appear in colour only in the online journal)

State recognition of the viscoelastic sandwich structure based on the adaptive redundant second generation wavelet packet transform, permutation entropy and the wavelet support vector machine

The viscoelastic sandwich structure is widely used in mechanical equipment, yet the structure always suffers from damage during long-term service. Therefore, state recognition of the viscoelastic sandwich structure is very necessary for monitoring structural health states and keeping the equipment running with high reliability. Through the analysis of vibration response signals, this paper presents a novel method for this task based on the adaptive redundant second generation wavelet packet transform (ARSGWPT), permutation entropy (PE) and the wavelet support vector machine (WSVM). In order to tackle the non-linearity existing in the structure vibration response, the PE is introduced to reveal the state changes of the structure. In the case of complex non-stationary vibration response signals, in order to obtain more effective information regarding the structural health states, the ARSGWPT, which can adaptively match the characteristics of a given signal, is proposed to process the vibration response signals, and then multiple PE features are extracted from the resultant wavelet packet coefficients. The WSVM, which can benefit from the conventional SVM as well as wavelet theory, is applied to classify the various structural states automatically. In this study, to achieve accurate and automated state recognition, the ARSGWPT, PE and WSVM are combined for signal processing, feature extraction and state classification, respectively. To demonstrate the effectiveness of the proposed method, a typical viscoelastic sandwich structure is designed, and the different degrees of preload on the structure are used to characterize the various looseness states. The test results show that the proposed method can reliably recognize the different looseness states of the viscoelastic sandwich structure, and the WSVM can achieve a better classification performance than the conventional SVM. Moreover, the superiority of the proposed ARSGWPT in processing the complex vibration response signals and the powerful ability of the PE in revealing the structural state changes are also demonstrated by the test results.

Investigations of denoising source separation technique and its application to source separation and identification of mechanical vibration signals

A novel vibration source separation and identification method using the denoising source separation (DSS) technique is proposed for the mixed mechanical vibration signals from engines in ships. Denoising source separation enables us to extract the source signals from the mixed signals without prior knowledge of sources and their mixing mode, and thus the important source information extracted by DSS can be used to monitor or actively control engine noises. Different denoising functions such as energy, skew, kurtosis, and tangential functions in DSS are applied to both simulation studies and experimental data to evaluate their separating performances. The tangential function provides the best outperformance with both numerical study and engineering application. In addition, the effectiveness of the proposed DSS method is validated by correlation analysis and the frequency marker tracking method.

A novel manifold-manifold distance index applied to looseness state assessment of viscoelastic sandwich structures

Viscoelastic sandwich structures (VSS) are widely used in mechanical equipment; their state assessment is necessary to detect structural states and to keep equipment running with high reliability. This paper proposes a novel manifold–manifold distance-based assessment (M2DBA) method for assessing the looseness state in VSSs. In the M2DBA method, a manifold–manifold distance is viewed as a health index. To design the index, response signals from the structure are firstly acquired by condition monitoring technology and a Hankel matrix is constructed by using the response signals to describe state patterns of the VSS. Thereafter, a subspace analysis method, that is, principal component analysis (PCA), is performed to extract the condition subspace hidden in the Hankel matrix. From the subspace, pattern changes in dynamic structural properties are characterized. Further, a Grassmann manifold (GM) is formed by organizing a set of subspaces. The manifold is mapped to a reproducing kernel Hilbert space (RKHS), where support vector data description (SVDD) is used to model the manifold as a hypersphere. Finally, a health index is defined as the cosine of the angle between the hypersphere centers corresponding to the structural baseline state and the looseness state. The defined health index contains similarity information existing in the two structural states, so structural looseness states can be effectively identified. Moreover, the health index is derived by analysis of the global properties of subspace sets, which is different from traditional subspace analysis methods. The effectiveness of the health index for state assessment is validated by test data collected from a VSS subjected to different degrees of looseness. The results show that the health index is a very effective metric for detecting the occurrence and extension of structural looseness. Comparison results indicate that the defined index outperforms some existing state-of-the-art ones.

A Comparative Study of Information-Based Source Number Estimation Methods and Experimental Validations on Mechanical Systems

This paper investigates one eigenvalue decomposition-based source number estimation method, and three information-based source number estimation methods, namely the Akaike Information Criterion (AIC), Minimum Description Length (MDL) and Bayesian Information Criterion (BIC), and improves BIC as Improved BIC (IBIC) to make it more efficient and easier for calculation. The performances of the abovementioned source number estimation methods are studied comparatively with numerical case studies, which contain a linear superposition case and a both linear superposition and nonlinear modulation mixing case. A test bed with three sound sources is constructed to test the performances of these methods on mechanical systems, and source separation is carried out to validate the effectiveness of the experimental studies. This work can benefit model order selection, complexity analysis of a system, and applications of source separation to mechanical systems for condition monitoring and fault diagnosis purposes.

Detecting the bonding state of explosive welding structures based on EEMD and sensitive IMF time entropy

With the increasing application of explosive welding structures in many engineering fields, interface bonding state detection has become more and more significant to avoid catastrophic accidents. However, the complexity of the interface bonding state makes this task challenging. In this paper, a new method based on ensemble empirical mode decomposition (EEMD) and sensitive intrinsic mode function (IMF) time entropy is proposed for this task. As a self-adaptive non-stationary signal analysis method, EEMD can decompose a complicated signal into a set of IMFs with truly physical meaning, which is beneficial to allocate the structural vibration response signal containing a wealth of bonding state information to certain IMFs. Then, the time entropies of these IMFs are calculated to quantitatively assess the bonding state of the explosive welding structure. However, the IMF time entropies have different sensitivities to the bonding state. Therefore, the most sensitive IMF time entropy is selected based on a distance evaluation technique to detect the bonding state of explosive welding structures. The proposed method is applied to bonding state detection of explosive welding pipes in three cases, and the results demonstrate its effectiveness.