Composite materials have a myriad of applications in complex engineering systems, and multiple structural health monitoring (SHM) strategies have been developed. However, these methods are challenging due to signal attenuation and excessive noise interference in composite materials. Signal processing can capture a small difference between the input–output signals associated with the severity of the damage in composites. Thus, the research question is “can signal processing techniques reduce the required number of features and assess the randomness of fatigue damage classification in composite materials using machine learning (ML) algorithms?” To answer this question, piezo-electric signals for carbon fiber reinforced polymer (CFRP) test specimens were taken from NASA Ames prognostics data repository. A framework based on a comparative analysis of signals was developed. For the first specific aim, the effectiveness of features based on statistical condition indicators of the sensor signals were evaluated. For the second specific aim, actuator-sensor signal pair were analyzed using cross-correlation to extract two features. These features were used to train and test four supervised ML algorithms for damage classification and their performance was discussed. For the third specific aim, randomness in the dataset of fatigue damage of the specimens was assessed. Results showed that by signal processing, the requirement of features for training ML was reduced with the improvement in the performance of ML. The randomness was captured by the utilization of two specimens from the same material. This work contributes to the improvement of intelligent damage classification of composite materials, operating under complex working conditions.

Early diagnosis in rotating machinery has been a challenge when looking toward the concept of intelligent machines. A crucial and critical component in these systems is the lubricated journal bearing, subjected to wear fault by abrasive removing of material in its inner wall, mainly during run-ups and run-downs. In extreme conditions, wear faults can cause unexpected shutdowns in rotating systems. Consequently, advanced condition monitoring is an essential procedure in the wear diagnosis of journal bearings. Although an increasing number of data-driven condition monitoring approaches for rotating machines have been proposed in the past decade, they mostly rely on substantial amounts of experimental data for training, which is expensive and time-consuming to obtain. The objective of this work is to develop a framework using a deep learning algorithm to classify wear faults in hydrodynamic journal bearings using simulated vibrations signals. Numerically simulated data sets under different wear severity levels and operating conditions were used to train and test the diagnostics framework. The results show that the proposed framework can be a promising tool to diagnose wear faults in journal bearings.

Composite materials can be modified according to the requirements of applications, and hence, their applications are increasing significantly with time. Due to the complex nature of the aging of composites, it is equally challenging to establish structural health monitoring techniques. One of the most applied non-destructive techniques for this class of materials is using Lamb waves to quantify the damage. Another important advancement in damage detection is the application of deep neural networks. The data-driven methods have proven to be most efficient for damage detection in composites. For both of these advanced methods, the burning question always has been the requirement of data and quality of data. In this paper, these measurements were used to create a framework based on a deep neural network for efficient fault diagnostics. The research question developed for this paper was as follows: Can data fusion techniques used along with data augmentation improve the damage diagnostics using the convolutional neural network? The specific aims developed to answer this research question were: (1) highlighting the importance of data fusion methods, (2) underlining the importance of data augmentation techniques, (3) generalization abilities of the proposed framework, and (4) sensitivity of the size of the dataset. The results obtained through the analysis concluded that the artificial intelligence techniques along with the Lamb wave measurements can efficiently improve the fault diagnostics of complex materials such as composites.

Rotating machinery has extensive usage in industrial applications, either as leading equipment (power plants) or as auxiliary equipment (oil and gas exploitation). These highly complex systems demand expensive maintenance programs due to the high costs of eventual shutdown. Consequently, critical fault diagnosis and prognosis are essential during the operation of those systems. Fault identification and classification demand robust verification of codes and calculations and a discerning validation of numerical models used for rotating machinery. Hence, verification and validation (V&V) are an essential initial service for a digital twin (DT), offering some advantages in this application. In this context, the following research question is proposed: Does V&V using DT improve data access and reduce the effort of data exchange? The following objectives are created to address the research question: perform a code verification, conduct the calculation verification, validate the models using two different validation approaches, and demonstrate easy access to asset data. For this study, two identical hydrodynamic bearings and a noncentral disk were considered, representing a laboratory experimental setup. The validation metric requirement is promisingly satisfied for the disk and bearings according to validation Approaches 1 and 2. Furthermore, validation Approach 2 generates even more successful results than approach 1. Accurate estimation and reliable interpretation of the numerical model outcomes encourage the application of DT for future fault diagnosis and prognosis.

This study aims to address the question: can the structural reliability of an offshore wind turbine (OWT) under fatigue loading conditions be predicted more consistently? To respond to that question this study addresses the following specific aims: (1) to obtain a systematic approach that takes into consideration the amount of information available for the uncertainty modeling of the model input parameters and (2) to determine the impact of the most sensitive input parameters on the structural reliability of the OWT through a surrogate model. First, a coupled model to determine the fatigue life of the support structure considering the soil-structure interaction under 15 different loading conditions was developed. Second, a sensitivity scheme using two global analyses was developed to consistently establish the most and least important input parameters of the model. Third, systematic uncertainty quantification (UQ) scheme was employed to model the uncertainties of model input parameters based on their available—data-driven and physics-informed—information. Finally, the impact of the proposed UQ framework on the OWT structural reliability was evaluated through the estimation of the probability of failure of the structure based on the fatigue limit state design criterion. The results show high sensitivity for the wind speed and moderate sensitivity for parameters usually considered as deterministic values in design standards. Additionally, it is shown that applying systematic UQ not only produces a more efficient and better approximation of the fatigue life under uncertainty, but also a more accurate estimation of the structural reliability of offshore wind turbine's structure during conceptual design. Consequently, more reliable, and robust estimations of the structural designs for large offshore wind turbines with limited information may be achieved during the early stages of design.

Wind turbines suffer from mass imbalance due to manufacturing, installation, and severe climatic conditions. Condition monitoring systems are essential to reduce costs in the wind energy sector. Many attempts were made to improve the detection of faults at an early stage to plan predictive maintenance strategies, but effective methods have not yet been developed. Artificial intelligence has a huge potential in the wind turbine industry. However, several shortcomings related to the datasets still need to be overcome. Thus, the research question developed for this paper was “Can data augmentation and fusion techniques enhance the mass imbalance diagnostics methods applied to wind turbines using deep learning algorithms?” The specific aims developed were: (i) to perform sensitivity analysis on classification based on how many samples/sample frequencies are required for stabilized results; (ii) to classify the imbalance levels using Gramian angular summation field and Gramian angular difference field and compare against data fusion; and (iii) to classify the imbalance levels using data fusion for augmented data. Convolutional neural network (CNN) techniques were employed to detect rotor mass imbalance for a multiclass problem using the estimated rotor speed as an input variable. A 1.5-MW turbine model was considered and a database was built using the software turbsim, fast, and simulink. The model was tested under different wind speeds and turbulence intensities. The data augmentation and fusion techniques used along with CNN techniques showed improvement in the classification and hence the diagnostics of wind turbines.