Introduction

Prognostics and health management (PHM) is an enabling discipline that uses sensors to assess the health of systems, diagnoses anomalous behavior, and predicts the remaining useful performance over the life of the asset. PHM originates the idea that the “health” (or degradation) of assets can be determined and the reliability (and remaining useful performance over the life of the asset) predicted with the aid of in situ sensing. PHM methodologies are based on several key elements in which sensors provide the capability of monitoring failure precursors and environmental loading conditions.

Deep Neural Network for Fault Diagnosis of Power Transformers using Dissolved Gas Analysis

-    Partners

-    Abstract
Dissolved gas analysis (DGA) is the most popular diagnostic tool to detect incipient stages of various faults in power transformers. To enhance DGA-based diagnosis accuracy, a novel method using deep neural network (DNN) has been developed to determine high-level features without relying on the handcrafted features. The results indicate that the proposed DNN approach achieves higher accuracy than the existing methods based on shallow learning with the handcrafted features.

Reduction of Li-ion Battery Qualification Time Based on Prognostics and Health Management

-    Partners

–    Abstract
During the lifetime, Li-ion batteries experiences capacity fade over repeated usage. State-of-Health (SOH) is defined the relative capacity compared to its initial value, and battery End-of-Life (EOL) is often considered when SOH reaches 80%. Li-ion battery qualification involves repeated charge and discharge cycles, which usually requires weeks and months of testing. This paper presents a PHM(Prognostics and Health Management)-based qualification for early detection of battery anomalies, as well as prediction of remaining useful life. A model-based prognostic algorithm was used to detect the earliest anomalous capacity fade, and to predict the remaining useful life prior to reaching the EOL.

PHM Architecture for Fault Diagnosis and Prognosis in Smart Factories
–    Partners
-    Abstract
This research introduced a standardized process of PHM implementation for fault diagnosis and prognosis in smart factories. Six common core modules in smart factories were identified to develop optimized PHM implementation process for health reasoning, fault diagnosis and prognosis. Standard architecture included common and practical PHM algorithms surveyed from >300 references with their respective technical maturity. The research outcomes published and distributed to small and medium enterprises, government agencies including MOTIE, and many research institutions.

PHM implementation process with standard architecture