In this paper, an in situ piezoelectric-fiber hybrid sensor network was developed to monitor the life-cycle of carbon fiber-reinforced plastics (CFRPs), from the manufacturing phase to the life in service. The piezoelectric lead-zirconate titanate (PZT) sensors were inserted inside the composite structures during the manufacturing process to monitor important curing parameters, including the storage modulus of resin and the progress of the reaction (POR). The strain that is related to the storage modulus and the state of resin was measured by embedded fiber Bragg grating (FBG) sensors, and the gelation moment identified by the FBG sensors was very close to those determined by dynamic mechanical analysis (DMA) and POR. After curing, experiments were conducted on the fabricated CFRP specimen to investigate the damage identification capability of the embedded piezoelectric sensor network. Furthermore, a modified probability diagnostic imaging (PDI) algorithm with a dynamically adaptive shape factor and fusion frequency was proposed to indicate the damage location in the tested sample and to greatly improve the position precision. The experimental results demonstrated that the average relative distance error (RDE) of the modified PDI method was 68.48% and 46.97% lower than those of the conventional PDI method and the PDI method, respectively, with an averaged shape factor and fusion frequency, indicating the effectiveness and applicability of the proposed damage imaging method. It is obvious that the whole life-cycle of CFRPs can be effectively monitored by the piezoelectric-fiber hybrid sensor network.

The vacuum-assisted resin infusion (VARI) technique provides considerable advantages in manufacturing large-scale composite structures. An accurate and consecutive structural health monitoring system is urgently required to determine the initial quality and assess the structural integrity of a composite structure. In this paper, a real-time active smart diagnostic system (SDS) based on piezoelectric sensor network is proposed to monitor the whole life-cycle of composite structures. Experiments were conducted on carbon fiber reinforced plastic specimens with different thicknesses to investigate the monitoring capability of piezoelectric lead-zirconate-titanate (PZT) sensors used in the SDS approach. The PZT sensor networks inserted inside the composite structures during the VARI process are used to monitor not only the curing parameters, but also the health status of composite structures when they are in service after curing. To monitor the curing process only, the sensor network can also be installed on the bottom of the mould. Experimental results demonstrate that both 3D resin flow and degree of cure in the VARI process can be effectively monitored by the PZT sensor network. Meanwhile, the embedded PZT sensor network has the potential to identify the different stages in the curing process. It is obvious that the piezoelectric sensor network will provide important technical support for composite materials with the structure and function integrated.

Impact brings great threat to the composite structures that are extensively used in an aircraft. Therefore, it is necessary to develop an accurate and reliable impact monitoring method. In this paper, fiber Bragg grating (FBG) sensors are embedded in unidirectional carbon fiber reinforced plastics (CFRPs) during the manufacturing process to monitor the strain that is related to the elastic modulus and the state of resin. After that, an advanced impact identification model is proposed. Support vector regression (SVR) and a back propagation (BP) neural network are combined appropriately in this stacking-based ensemble learning model. Then, the model is trained and tested through hundreds of impacts, and the corresponding strain responses are recorded by the embedded FBG sensors. Finally, the performances of different models are compared, and the influence of the time of arrival (ToA) on the neural network is also explored. The results show that compared with a single neural network, ensemble learning has a better capability in impact identification.

Establishing an Exhaust Gas Temperature (EGT) baseline model is an important way to realize the overall performance real time monitoring and health management of an aeroengine. An EGT baseline model of aeroengine was constructed by combining Kernel Principal Component Analysis (KPCA) with Deep Belief Network (DBN). The Quick Access Recorder (QAR) data of the airplane with CFMS6-7B aeroengine was taken as the original data sample. The data dimensions were first reduced using KCPA, where Gaussian function was selected as Kernel function. The dimensionality-reduced data was put into DBN to establish the EGT baseline model. The validity and superiority of the model were verified by a large number of QAR data. Compared with traditional neural network modeling techniques, the proposed method not only reduces the complexity of the network structure, but also improves the accuracy of the model.