Trend analysis methods and system for incipient fault prediction
View Patent ↗There is described herein the extension of the application of the Laplace Test Statistic for trend analysis and prediction of incipient faults for power systems. The extensions consider the situation where two parameters believed to contribute explicitly to the eventual failure are monitored. The developed extensions applied to actual incipient failure events provide promising results for prediction of the impending failure. It is demonstrated that by incorporating two parameters in the trend analysis, the robustness to outliers is increased and the flexibility is augmented by increasing the degrees of freedom in the generation of the alarm signal.
1. A method of performing trend analysis on incipient events in a power network, the method comprising:
receiving data including a first parameter and a second parameter, said first parameter representing the time of an incipient event and said second parameter representing a characteristic of said incipient event;
determining a laplace test statistic using said first parameter and said second parameter;
comparing said laplace test statistic against a threshold;
outputting an alarm signal if said laplace test statistic is greater than said threshold; setting an event window size, wherein said event window operates according to first-in-first-out principle; and
wherein the laplace test statistic is determined using only data from said event window.
2. The method of claim 1 wherein said second parameter is incorporated as a weighting factor for said first parameter.
3. The method of claim 2 wherein said laplace test statistic is a weighted laplace test statistic (WM LTS) and is calculated according to:
WMLTS
=
(
∑
i
=
1
N
ω
i
t
i
∑
i
=
1
N
ω
i
-
t
N
+
1
2
)
t
N
+
1
·
1
12
N
.
4. The method of claim 1 wherein said laplace test statistic is determined using said first and said second parameters in two dimensional Cartesian space.
5. The method of claim 4 wherein said laplace test statistic is calculated according to:
2
DLTS
=
{
(
1
N
∑
i
=
1
N
par
1
i
-
par
1
N
+
1
2
)
par
1
N
+
1
·
1
12
N
}
+
j
{
(
1
N
∑
i
=
1
N
par
2
i
-
par
2
N
+
1
2
)
par
2
N
+
1
·
1
12
N
}
.
6. The method of claim 1 wherein said second parameter is the energy of the incipient event.
7. The method of claim 1 wherein the second parameter is a value indicative of environmental conditions at the time of the incipient event.
8. The method of claim 1 wherein the second parameter is a value indicative of baseline load at the time of the incipient event.