Multilayer Principal Component Research Articles

Monitoring of low voltage grids with multilayer principal component analysis

This article presents a monitoring strategy based on multilayer principal component analysis (PCA) to detect and diagnose power system disturbances in large amounts of data collected by intelligent electronic devices in low voltage smart grids. The PCA models are built on multiple sliding windows, sized (in terms of length and sampling time) according to the type of phenomena to detect. Abnormalities are detected with use of two complementary statistical indexes, then diagnosed by computing the individual contributions of each monitored variable to the constraint violation of those statistics. As a result, its implementation enables an automatic analysis of multiple phenomena of interest in parallel over time using distinct electrical quantities. Furthermore, the method is demonstrated within the RESOLVD project with data from the OpenLV project containing measurements of active and reactive power gathered at different low voltage distribution substations.

Lumbar posture assessment with fabric strain sensors

Astronauts are at risk for low back pain and injury during extravehicular activity because of the deconditioning of the lumbar region and biomechanical demands associated with wearing a spacesuit. To understand and mitigate injury risks, it is necessary to study the lumbar kinematics of astronauts inside their spacesuit. To expand on previous efforts, the purpose of this study was to develop and test a generalizable method to assess complex lumbar motion using 10 fabric strain sensors placed on the torso. Anatomical landmark positions and corresponding sensor measurements were collected from 12 male study participants performing 16 static lumbar postures. A multilayer principal component and regression-based model was constructed to estimate lumbar joint angles from the sensor measurements. Good lumbar joint angle estimation was observed (<9° mean error) from flexion and lateral bending joint angles, and lower accuracy (13.7° mean error) was observed from axial rotation joint angles. With continued development, this method can become a useful technique for measuring suited lumbar motion and could potentially be extrapolated to civilian work applications.

Fault Diagnosis System for NPC Inverter based on Multi-Layer Principal Component Neural Network

This paper presents a fault diagnosis method for a neutral point clamped (NPC) inverter using a multi-layer artificial neural network (MANN). The considered possible faults of NPC inverter include the open-circuit fault occurring in one single device or more devices. The upper, middle and down bridge voltages are adopted the test signals because of the difficulties in isolating some fault modes. A novel multi-layer neural network is proposed to diagnose all possible open-circuit faults. Furthermore, the principal component analysis (PCA) is utilized to reduce the input size of neural network. The comparison between neural network with and without PCA is performed. The simulation and experimental results prove the feasibility of the diagnostic method and show that the proposed method has the advantages of good classification performance and high reliability.