Automatic Identification of Fault Types When a Vacuum Circuit Breaker Switches Off a Shunt Reactor - Blog

info-1280-853

Fault type identification should be based on?

The switching-off of a shunt reactor by a vacuum circuit breaker is a typical case of inductive small-current switching, in which current chopping, restrike/reignition, and overvoltage are typical phenomena. Automatic fault-type identification cannot rely on a single electrical quantity alone; it should be based on multi-source signal fusion.

Acquire measurable signals

First, measurable signals should be collected: the circuit breaker opening-coil current, contact travel and speed, three-phase reactor current, busbar and reactor-side voltage, and, where conditions permit, post-arc current of the interrupter chamber or high-frequency transient recording. Features are extracted from these signals: current chopping value, number and timing of post-arc restrikes, overvoltage amplitude and steepness, three-phase asynchronism, contact bounce, and inflection points of the coil current. These features correspond respectively to mechanisms such as current chopping, restrike, mechanical abnormality, and three-phase inconsistency, and they have a factual basis.

Determine whether it is "normal breakage/abnormal breakage"

The identification logic can first determine "normal opening/abnormal opening," and then classify abnormalities according to rules: if the chopping value is high and overvoltage is obvious, it is judged as chopping-dominated; if post-arc current appears repeatedly and is accompanied by high-frequency voltage oscillation, it is judged as restrike or reignition; if the coil current and travel waveforms are abnormal and bounce exceeds the limit, it is judged as a mechanical abnormality of the mechanism; if the three-phase opening time difference is abnormal, it is judged as three-phase asynchronism. Control thresholds should be set on the basis of the manufacturer's test results and real-world samples recorded in the field; machine learning may be utilised to assist with classification, but must not deviate from the relevant mechanisms.

Multiple parameters should be cross-validated against one another

It should be emphasized that automatic identification should use multiple parameters for mutual verification and must not draw conclusions based on a single threshold. Specific criteria must be determined in combination with equipment structure and measured data, and must not be fabricated.

info-500-416