Graduate Student Research

we describe a Cyber-Physical system approach to fault detection in Photovoltaic (PV) arrays. More specifically, we explore customized neural network algorithms for fault detection from monitoring devices that sense data and actuate at each individual panel. We develop a framework for the use of feedforward neural networks for fault detection and identification. Our approach promises to improve efficiency by detecting and identifying eight different faults and commonly occurring conditions that affect power output in utility scale PV arrays.

Author:  Sunil Rao, SenSIP Center, School of ECEE, ASU

The positive and unlabeled learning problem is a semi-supervised binary classification problem over a set of known positive samples, no known negative samples, and a large amount of unlabeled data. The probability of an unlabeled sample being positive is unknown. We build on previous work and introduce a new algorithm using a modified logistic regression algorithm to solve the problem.

Index Terms—PU learning, positive unlabeled learning,
machine learning, semi-supervised learning, classification

Authors:  Kristen Jaskie, SenSIP Center, School of ECEE, ASU

For datasets that reside on irregular and complex domains, using graph models to represent them can bring advantages. Graph signal processing (DSPG) offers tools for such data sets and combines graph theory with traditional digital signal processing (DSP). In this project, we focus on improving the accuracy of graph-based
classifier for semi-supervised classification. We propose an approach using multiple graph shift matrices, one
for each feature, which provides better performance when the feature qualities are uneven. We study on multiple optimization solutions for the parameters of the model, which are graph filter coefficients and graph combing coefficients.

Authors:  Jie Fan, SenSIP Center, School of ECEE, ASU

In this project, a cyber-physical system (CPS) approach for optimizing the output power of photovoltaic (PV) energy systems is proposed. In particular, a novel connection topology reconfiguration strategy for PV arrays to maximize power output under a set of partial shading conditions using neural networks is put forth. Depending upon an irradiance/shading profile of the panels, topologies such as series parallel (SP) and total cross tied (TCT) produce different maximum power points (MPP). The connection topology of the PV array that provides the maximum power output is chosen using a multi-layer perceptron model. The method proposed can be implemented in any CPS PV system with switching capabilities and is simple to implement.

Authors:  Vivek Sivaraman Narayanaswamy, SenSIP Center, School of ECEE, ASU