Alstom MiCOM P846, Техническое руководство

Страница: 86 / 298

P846/EN OP/D Operation
(OP) 5-8
MiCOM P846
OP
1.3.4 Fuzzy-logic engine stage
Although fast and reliable relaying algorithms can be achieved by applying conventional
Boolean logic, conflicts may arise in doubtful cases. For instance, during line energization at
no-load, CVT (Capacitive Voltage Transformers) induced electro-magnetic transients may
suggest that the active power is not zero (line closed) while the line is actually open.
Experience has demonstrated that such a conflict can be more easily resolved using multi-
criteria fuzzy Logic based approach which results in the following interesting properties:
•
Tripping decision is based on several criteria with weighting factors.
•
Uncertainty with respect to signals and settings is modeled quantitatively
•
Delay of tripping initiation depends on the amount or inflow of information related to
relaying signals, and through them, to the disturbance analyzed by the protection.
A sample general structure of the fuzzy–logic engine is shown in Figure 5. It consists of three
main steps:
1. Fuzzyfication of 10 selected decision features from the pre-processor
2. Fuzzy logic inference on these features, using sixteen rules or criteria
3. Crisp decision sent to the output
1.3.4.1 Fu
tive Power and
zzyfication of 10 selected decision features from the pre-processor
Fuzzification is a necessary step prior to a fuzzy-logic based reasoning system. It consists in
transforming the crisp variables provided by the pre-processor into categories easily
described by linguistic spellings from the common language such as “Normal”, “Small,
“Large”, “Very Large”.
Figure 6 illustrates this fuzzification process for two typical variables, the Ac
the DeltaP. The first feature is defined as "SMALL" when its crisp value is below a given
threshold
θ L while the DeltaP feature is defined as "LARGE" for crisp values higher that θ H.
P4089ENa
Figure 5: Graphical representation of the DLO’s fuzzy decision system