Structural health monitoring for mechanical structures using multi-sensor data

Structural health monitoring (SHM) is now one of the key and crucial issues for structures and machinery due to the increase in accident ratio amount. At present, more and more attention has been attracted toward the reliability and safety evaluations for the inspected structures.^1–3 On reviewing the history of SHM, which is proposed in 1970s, we can see that the development of SHM always accompanies the development of sensor technologies and measurement technologies. Taking the measurement of strain and stress, for example, these two terms, which are important for damage detection in SHM, are usually registered by strain gauge or estimated by the derivative of displacement. ⁴ However, the development of sensor allows the directed measure of the derivative modal shape, ⁵ and the laser technology^6,7 also allows the measure of the full wave field or vibration field. These are all fully involved and considered in the decision and the assessment of damage for SHM.

The development of sensor technologies provides more alternative choices for SHM. Different from the classical approaches used in SHM, namely the acceleration-based methods,^8–10 the novel methodologies contain the wave propagation method,^11,12 fiber Bragg grating method,^13–15 and the advanced modal curvature methods.^16,17 These technologies detect damage via the totally different physical phenomenon. Thereafter, the determination of faults could be disparate. From the view of practice, more types and a larger number of sensors or array can be implemented on structures, thanks to the cost reduction of sensors and the higher requirement of reliability. This makes it possible to consider an SHM system based on multi-sensor data. On the other hand, the monitoring based on single-type sensor could not provide enough information to detect the operational condition of complex mechanical structures. ¹⁸ The multi-sensor data SHM process includes data measurements taken using an array of sensors, the extraction of damage-sensitive features from these measurements, and the statistical analysis of these features to predict the current state of the structures. “There are a thousand Hamlets in a thousand people’s eyes,” this also works for SHM, especially for the structures could be monitored by various types of sensors. The results of multi-sensor system may induce the conflict on conclusion due to the different sensitives of methodologies or phenomenon. To address the conflicts among methodologies, some novel methods for data fusion have been proposed, like the hybrid multivariate analysis method,^19,20 the multiple step methods,^21–23 and the hybrid methods. ²⁴ Although the present methodologies and techniques are feasible to resolve some problems, the emergence of non-destructive tests, SHM, and equipment evaluation using vibration information demands for the further development to achieve the essential data fusion.

The appreciate processing of multi-sensor data should be based on the clear understanding of damage mechanism and modeling, and then the fusion of signal processing are all the crucial issue for this subject. On the aspect of damage mechanism and modeling, Xu et al. ²⁵ proposed fatigue mechanism–based dependent modeling with stochastic degradation and random shocks. They considered both fatigue degradation and applied random shock damage to have a coupled effect on the crack propagation process, which addresses the retardation phenomenon problem in multi-sensor SHM. Yang and colleagues^26–29 proposed the wavelet finite element method to illustrate the relationships between crack propagation and the dynamic properties in frequency/wavenumber domain. The scaling functions and the corresponding wavelets functions are used to replace the polynomials utilized in traditional finite element modeling for higher accurate on the descriptions of crack and the other singularities. On the aspect of signal processing for multi-sensor data, Park et al. ³⁰ introduced a wireless displacement sensing system for bridges using multi-sensor fusion. In their work, they proposed an indirect displacement estimation using two different types of measurements fusion such as strain and acceleration. Soman et al. ³¹ proposed a genetic algorithm–based method to fuse different measurement and features. Their further investigation focused on the noise suppression of multi-sensor fusion using Kalman-filtering. ³² In the present special collection, 10 research articles from America, Australia, British, China, and Mexico after a strict peer review process. The research areas were within the topic of the special collection, for example, damage detection strategy based on multi-sensor data, multi-sensor data fusion method, non-destructive testing and evaluation dependent on multi-sensor data, machine learning and streaming data–based SHM, multi-sensor data–based condition monitoring, multi-sensor data SHM hardware system development, case studies and industrial applications, and so on.

For damage-detection strategy based on multi-sensor data, the paper “Enhanced frame expansion via configuring filter-bank topology for rapid processing of multi-sensor vibration data with applications to turbo-machinery fault diagnosis” by He et al. proposed the enhanced frame expansion via integration of multiple translation-invariant frames. It is verified by the authors that the enhanced frame expansion satisfies the necessary constraints of complex-valued wavelet frame and other beneficial merits. The paper “A multi-sensor fault detection strategy for axial piston pump using the Walsh transform method” by Gao et al. developed a novel multi-sensor fault-detection strategy to realize more effective diagnosis of axial piston pump. The discharge pressure, vibration, and acoustic signal are all considered in their fusion work.

For non-destructive testing and evaluation dependent on multi-sensor data, the paper “Connection looseness detection of steel grid structures using piezoceramic transducers” by Yan et al. proposed an evaluation method for bolt-sphere joint looseness of steel grid structures using piezoceramic guided wave–based method through experiments and numerical simulations.

For imagery-based SHM, the paper “Surface reconstruction based on the camera relative irradiance” by Yao et al. proposed a novel surface reconstruction method that uses camera relative irradiance via the image gray-scale value information under fixed ring light. After calibrations of the measurement condition, just one image of the object is necessary to reconstruct the surface.

For multi-sensor data fusion method, the paper “Multi-sensor image fusion based on regional characteristics” by Meng et al. presented a fusion algorithm based on regional characteristics for combining infrared and visible light images in order to achieve an image with clear objects and high-resolution scene. The paper “Control of steering wheel idle jitter based on optimization of engine suspension system with verifications using multi-sensor measurement” by He et al. proposed a multi-sensor-based measurement strategy that was utilized to evaluate the idle jitter severity of the steering wheel.

For machine learning and streaming data–based SHM, the paper “Optimization of acousto-ultrasonic sensor networks using genetic algorithms based on experimental and numerical data sets” by Marks et al. presented a methodology for optimizing acousto-ultrasonic transducer placement for adhesive dis-bond detection on metallic aerospace structures. The paper “Dynamic cluster heads selection and data aggregation for efficient target monitoring and tracking in wireless sensor networks” by Feng et al. proposed an efficient target-tracking approach, in which the nodes on the edge of a cluster instead of the centered nodes are chosen as cluster heads so that cluster heads can serve as manager and monitoring node.

For multi-sensor data-based condition monitoring, the paper “Quadcopter localization and health monitoring method based on multiple virtual silhouette sensor integration” by Hou et al. presented a quadcopter flight regime extraction algorithm for quadcopter localization and health monitoring using imageries captured by general purpose monocular cameras.

For case studies and industrial applications, the paper “Performance analysis of a wireless sensor network with cognitive radio capabilities in SHM applications: A discrete model” by Garrido et al. investigated the performance of a wireless sensor network with cognitive radio capabilities to gather information about SHM of buildings in case of seismic activity.

In summary, much effort should be done to connect the novelty methods and the SHM for the real-world mechanical and civil structures.