Utility Asset RUL Analysis Using Irrelevance-Filtered SPRT Alarms
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for monitoring power transformer health, such as dissolved gas analysis (DGA), are time-consuming and expensive, and prognostic-surveillance methods generate high false alarm rates due to insufficient historical failure data, leading to unnecessary maintenance and premature asset replacement.
Innovation Solution
A system using an inferential model, sequential probability ratio test (SPRT), and an irrelevance filter to analyze time-series sensor signals, filtering out irrelevant alarms and calculating a remaining useful life (RUL) based on logistic-regression, reducing false alarms and optimizing maintenance schedules.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If dissolved gas analysis (DGA) is used to monitor transformer health, then detection capability is improved, but time consumption and cost increase
Solution Approach 1:
The patent replaces the mechanical/chemical DGA process with an electrical signal-based monitoring system. Instead of periodically extracting oil samples for chemical analysis, the system continuously monitors transformer health through electrical sensors that detect patterns in voltage, current, and other electrical parameters, eliminating the time-consuming physical sampling and laboratory analysis processes.
Solution Approach 2:
The patent creates a virtual model of transformer behavior by continuously collecting and analyzing electrical signal patterns. This digital copy of the transformer's operational state allows for ongoing health assessment without physically interrupting the transformer's operation or requiring periodic oil sampling, enabling continuous monitoring at minimal cost and time.
2Reliability
If prognostic-surveillance techniques are used to estimate RUL, then proactive maintenance is improved, but false alarm rate increases
Solution Approach 1:
The patent performs preliminary training of the neural network model using historical transformer data before deployment. By pre-training the system with extensive historical patterns of both normal and degraded transformer states, the model learns to distinguish between benign variations and true degradation signals, reducing false alarms while maintaining proactive detection capability.
Solution Approach 2:
The system incorporates feedback mechanisms where alarm predictions are continuously validated against actual transformer outcomes. When false alarms occur, the system learns from these errors by adjusting its detection thresholds and patterns, and when actual failures occur, the system refines its prediction algorithms, creating a self-improving feedback loop that reduces false alarms over time.
3Measurement precision
If frequent monitoring is performed to improve detection accuracy, then measurement precision is improved, but cost and complexity increase
Solution Approach 1:
The patent designs a monitoring system that uses existing electrical sensors and communication infrastructure already present in modern transformers. By making the system universal and compatible with standard transformer components rather than requiring specialized sensors or complex hardware additions, the patent achieves continuous high-precision monitoring without significantly increasing device complexity or cost.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
During operation, the system receives time-series signals gathered from sensors in a utility system asset. Next, the system uses an inferential model to generate estimated values for the time-series signals, and performs a pairwise differencing operation between actual values and the estimated values for the time-series signals to produce residuals. The system then performs a sequential probability ratio test (SPRT) on the residuals to produce SPRT alarms. Next, the system applies an irrelevance filter to the SPRT alarms to produce filtered SPRT alarms, wherein the irrelevance filter removes SPRT alarms for signals that are uncorrelated with previous failures of similar utility system assets. The system then uses a logistic-regression model to compute an RUL-based risk index for the utility system asset based on the filtered SPRT alarms. When the risk index exceeds a threshold, the system generates a notification indicating that the utility system asset needs to be replaced.