Example based Anomaly Sign Detection
| Project/Area Number |
24300072
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (B)
|
| Allocation Type | Partial Multi-year Fund |
|---|
| Section | 一般 |
|---|
| Research Field |
Perception information processing/Intelligent robotics
|
|---|
| Research Institution | Wakayama University |
|---|
Principal Investigator |
WADA Toshikazu 和歌山大学, システム工学部, 教授 (00231035)
|
|---|
| Co-Investigator(Renkei-kenkyūsha) |
MAEDA Shunji 株式会社日立製作所, 研究開発本部 (00626799)
SHIBUYA Hisae 株式会社日立製作所, 研究開発本部 (50626801)
|
|---|
| Project Period (FY) |
2012-04-01 – 2016-03-31
|
| Project Status |
Completed (Fiscal Year 2015)
|
| Budget Amount *help |
¥13,910,000 (Direct Cost: ¥10,700,000、Indirect Cost: ¥3,210,000)
Fiscal Year 2015: ¥2,990,000 (Direct Cost: ¥2,300,000、Indirect Cost: ¥690,000)
Fiscal Year 2014: ¥2,340,000 (Direct Cost: ¥1,800,000、Indirect Cost: ¥540,000)
Fiscal Year 2013: ¥2,860,000 (Direct Cost: ¥2,200,000、Indirect Cost: ¥660,000)
Fiscal Year 2012: ¥5,720,000 (Direct Cost: ¥4,400,000、Indirect Cost: ¥1,320,000)
|
|---|
| Keywords | パターン認識 / 異常予兆検出 / 非線形回帰 / ガウス過程回帰 / Dynamic Active Set / 時間多重解像度解析 / Spectro Anomaly Gram / Gaussian Process Regression / Similarity Based Modeling |
|---|
| Outline of Final Research Achievements |
Anomaly sign detection is a problem detecting subtle deviation of sensor data from normal data for monitoring patients, industrial plant, and so on. Anomaly sign detection can be realized as a problem measuring the discrepancy between observed data and estimated sensor data by non-linear regression. In this research, we developed anomaly sign detector based on Gaussian Process Regression (GPR). As the results of this research, we proposed 1) “Anomaly Measure” representing the ratio of the discrepancy and GPR estimated standard deviation, 2) multi-scale derivation and visualization of anomaly measure “Spectro Anomaly Gram(SAG)”, 3) an acceleration of GPR computation “Dynamic Active Set(DAS)”, 4) “MultiVariate GPR (MVGPR)” estimating vector output and covariant matrix, 5) “Reweighted MVGPR” to improve improper covariant matrix estimated by MVGPR, and so on. We applied our methods to industrial plant data and ECG data, and confirmed the effectiveness.
|
Report
(5 results)
Research Products
(18 results)
