Artificial intelligence techniques applied to fault detection systems
Bibliographic record
Abstract
This thesis presents novel applications of Artificial Intelligence-based algorithms to Failure Detection Systems. It shows the benefits intelligent, adaptive blocks can provide along with potential pitfalls. A new fault detection structure is introduced which has desirable properties when dealing with missing data, or data corrupted by extraneous disturbances. A classical alarm generation procedure is extended by transformation into an optimum, real-time, adaptive block. Two techniques, artificial Neural Networks, and Partial Least Squares, complement each other in one of the failure detection applications exploiting their respective non-linear and de-correlation strengths. Artificial Intelligence techniques are compared side by side with classical approaches and the results are analyzed. Three practical examples are examined: The Static Security Assessment of Electric Power Systems, the Oil Leak Detection in Underground Power Cables, and the Stator Overheating Detector. These case studies are demonstrated since each one represents a class of failure detection problems. The Static Security Assessment of Electric Power Systems is a class of problems with inputs which are somewhat correlated, and which has very little learning data. While the time required for the system to learn is not a concern, the recall time must be short, providing for real-time performance. The Oil Leak Detection in Underground Power Cables represents the class of problems where one has vast amounts of data indicative of a properly functioning system, however data from a failed system are very sparse. Unlike the Static Security Assessment problem, the oil leak detector has to consider the time dynamics of the system. Special provisions must be made to accommodate missing data which would interrupt contiguous data sets required for proper operation. This case study shows ways to exploit the slight sensor redundancy in order to detect sensor breakdown along with the detection of the main system failure. A third class of problems is showcased by the Electric Generator Stator Overheating detector. This application must deal with highly correlated inputs, along with the lack of fault data to be used for learning. Physical system non-linearities as well as time dynamics must also be addressed.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.001 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".