Elsevier

Computer Communications

Volume 157, 1 May 2020, Pages 150-161
Computer Communications

Intelligent early structural health prognosis with nonlinear system identification for RFID signal analysis

https://doi.org/10.1016/j.comcom.2020.04.026Get rights and content

Abstract

The mechanical state of non-linear processes includes these characteristics such as multivariable, strong coupling, multiple vibration sources, large signal noise, and various random factors. The intrinsic relationship and overall consistency optimization of the characteristic factor direction paths of multi-source matrix signals are a new research hotspot in multi-source dynamic characteristic signal identification. A intelligent fault diagnosis and prognosis method based on NARMAX-FRF and PCA is proposed in this paper. This method can be widely used in industrial system fault diagnosis. This system solves many basic key problems, including identifying a non-linear model from a detected system, accurately solving the frequency response function, extracting a representative frequency domain from the frequency response function, and applying extracted system frequency domain features for large-scale structural health assessment. In order to verify the performance of the NARMAX_FRF and PCA method for nonlinear defect signal analysis, the experiment of intelligent RFID system of corrosion monitoring and the TOFD experimental system are analyzed for the structural health monitoring in this paper. The set of samples in the experiment of intelligent RFID system of corrosion monitoring consists of coated and uncoated mild steel plates, which have a patch that has been exposed to the environment for different durations to create different levels of corrosion. The results show the effectiveness and robustness of the proposed method.

Introduction

The production equipment evolves under the development of the modern industrial technology with the characteristics of large-scale, automatic and high efficiency. Mechanical multi-fault mode diagnosis technology is used in complex process industrial production lines to achieve intelligent, remote online status monitoring and fault diagnosis of large key equipment, which is an inevitable demand for informatization, intelligence and automation of complex process manufacturing [1], [2], [3], [4], [5], [6]. The research of structural health detection technology for early identification and modeling of non-linear ultrasonic systems with small defects belongs to the category of modern acoustic detection theory and non-linear system science. Due to the complexity of the problem itself, there are many difficulties to be solved. Researchers encountered various difficulties when conducting online early fault diagnosis and structural health testing of major equipment. In order to explore an effective method for the characterization and identification of nonlinear weak defect characteristic signals, a large number of researchers have been constantly exploring [7], [8], [9], [10]. There are the large number of the requirements for the fault diagnosis and Non-destructive evaluation (NDE) techniques of the engineering system, which is applied to assess the working condition of engineering system to detect any incipient damages to prevent any possibility of a catastrophic failure [11], [12].

Pulse eddy current (PEC) testing technology is a rapidly developing technology. Its applications in structural non-destructive testing (NDT) include metal thickness measurement, defect detection and corrosion detection of multilayer materials. PEC based characterization theory and feature extraction techniques have been developed by understanding the influence and responses in the time and frequency domain [13], [14], [15]. Conventional eddy current NDE techniques just used the interpretation techniques to analyze the structural output response to extract the features with meaningful information to assess the structural health and integrity by quantitative characterization such as defect size, location and categories. The interpretation techniques for eddy current NDT technique need a reference with a defect free condition to calculate the differential signal for the structural health detection [16]. However, the reference signal is easy to be affected by the other unstable environmental condition in industrial applications and the differential signal cannot reveal the intrinsic system mechanism, which is generated by the structural defect. It is difficult to find the interpretation techniques to extract the intrinsic information to explain the effects of the defect on the structural system condition [17], [18], [19].

There are many applications of the model-based methods in fault diagnosis of engineering system including system identification [20], [21], [22]. The faults are considered as unknown inputs or disturbances to the model parameters, which is used as features to represent the component failures. However, this method requires the detailed physical model of the engineering system, which is not feasible in practice. In order to overcome the difficulty in physical model, this paper proposed a novel model-based method, which used the discrete input and output time domain data to establish the black-box modeling. The frequency response function (FRF) analysis is used to analyze the time domain model and extract features to diagnose the fault condition of the engineering system. The system fault is reflected in the variation of the time domain model, which is represented in the changes of the system FRFs. Consequently, the changes in the system FRFs can be used to conduct system fault diagnosis structural NDT.

Principal component analysis (PCA) is the most frequently used data analysis method in multivariate analysis (MVA) for fault detection, dimension reduction and isolation. PCA is a simple and effective process monitoring and fault diagnosis method, which has been widely used in practical industrial systems. Based on the kernel principal component analysis (KPCA) and kernel partial minimum analysis (KPLS) models at different scales, reference [23] proposes a new method for nonlinear process monitoring and fault diagnosis, called multi-scale KPCA and multi-scale KPLS, which are applied to these multi-scale data to capture process variable correlations that occur at different scales. This paper on reference [24] proposed a SSAE-based support vector machine (SVM) and principal component analysis (PCA) network to improve the accuracy of power system fault diagnosis. A new method based on principal component analysis (PCA) and sequential probability ratio test (SPRT), the main component with the largest contribution rate is selected as the input signal for SPRT evaluation, is proposed about the multi-fault condition monitoring of slurry pump by Hanxin Chen and Wenjian Huang [24]. In order to improve the recognition rate of traffic detector fault data, the improved MSPCA model combines wavelet packet energy analysis with PCA to realize the fault recognition of traffic detector data [25]. In [26], researchers proposed an independent component analysis-principal component analysis (ICA-PCA) integrated with correlation vector machine (RVM) for multivariable process monitoring.

In order to evaluate the performance of the proposed new NDE method, two sets of experimental procedures are designed. One set consists of specimen of three different defects with eighteen defect sizes. Another set consists of a growing defect with 2 mm step up to the length of 100 mm. The results of the experimental data analyses sufficiently verify the effectiveness of the new technique, and demonstrate that the model-based pulsed eddy current NDE approach has great potential in engineering applications.

Section snippets

Nonlinear system model with NARMAX_FRF

Pulsed eddy current testing has many applications such as defect detection of multi-layered structure [27], corrosion detection [27] etc. The previous researches focused on the feature extraction of the PEC response in time domain such as peak values, rising times and zero-crossing time. However, the specific spectral response due to physical properties of the material has not yet been fully investigated. It is important to address the spectral response of PEC under the different properties of

Theory about RFID for intelligent structural health monitoring

RFID (called the Radio Frequency Identification) system consists of two hardware components: a reader and a transponder. The reader itself usually consists of a microcontroller and RF circuits, such as envelope detectors and filters that need to send and receive RF energy. The reader includes a coil or antenna for transmitting and receiving. The function of the reader is to write data to the tag memory, and at the same time, it can transfer enough power to the tag to power it on and receive the

RFID experimental setup

Atmospheric corrosion of steel is a general term for an arrangement of iron oxides (hematite Fe2o3 and magnetite Fe3o4) and hydroxides (ferrous hydroxide Fe (OH) and ferric hydroxide Fe (OH)3). Magnetite is frequently in rust created in marine climates, which is a ferromagnetic mineral. In order to demonstrate the ability of RFID sensor system to detect and differentiate different levels of corrosion two sets of samples have been applied for validation. A LF (125 kHz) RFID development kit has

Ultrasonic crack detection with PCA for frequency domain feature analysis

A typical TOFD experimental system is set up as shown in Fig. 9. Our experiment used miniature angle beam transducers (MSW-QC style, Benchmark Series) and a model W-211 (45°) wedge, products of GE Inspection Technologies. The carrier frequency and bandwidth of the transducers are 2.25 MHz and 1.5 MHz, respectively. The wedge angle is 45°. A regular steel block is used as the specimen. The thickness of the specimen is 2.25 cm, and the depth of the surface crack is 1.25 cm. The two transducers

Conclusion

A novel nonlinear system identification method based on linear time-domain model and frequency-domain analysis is proposed to establish a relationship model between dynamic characteristics of mechanical systems and frequency-domain parameters in nonlinear multi-fault mode. This model can reveal the dynamic physical mechanism change caused by the mechanical structure change. The mathematical model of dynamic mechanism of mechanical structure system can be determined by system input and output.

CRediT authorship contribution statement

Hanxin Chen: Conceptualization, Data analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing - original draft, Writing - review & editing. Yongting Chen: Data analysis, Investigation, Writing - original draft. Liu Yang: Investigation, Data curation, Writing - review & editing.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

The authors would like to thank EPSRC, UK and the AISP industrial partners for funding this work through grant EP/J012343/1 (NEWTON). The experimental data is provided by the Reliability Research Lab in the Department of Mechanical Engineering at the University of Alberta in Canada. This work was supported by the Special Major Project of the Ministry of Science and Technology of Hubei Province of China (Grant No. 2016AAA056), Major project of Hubei Provincial Department of Education, China (

References (31)

  • JunShi et al.

    Short-term mechanical analysis of polyethylene pipe reinforced by winding steel wires using steel wire spiral structural model

    J. Press. Vessel Technol.-Trans. ASME

    (2018)
  • ChenHanxin et al.

    Fault identification of gearbox degradation with optimized wavelet neural network

    Shock Vib.

    (2013)
  • ChenH. et al.

    Model-based method with nonlinear ultrasonic system identification for mechanical structural health assessment

    Trans. Emerging. Tel. Tech.

    (2020)
  • GaoBin et al.

    Automatic defect identification of eddy current pulsed thermography using single channel blind source separation

    IEEE Trans. Instrum. Meas.

    (2014)
  • G.Y. Tian, A. Sophian, D. Taylor, J. Rudlin, Wavelet-based PCA defect classification and quantification for pulsed eddy...
  • Cited by (0)

    View full text