How to determine if the clock deviation of a dual sensor is too large?

A deviation of more than one sample period will impact the comparison results and is deemed excessive; a deviation of less than half a sampling period has no effect and is deemed acceptable. This is the fundamental criterion for assessing whether the clock deviation is too great. In particular, the following two techniques can be used to swiftly ascertain it:

 

I. Rapid Engineering Judgment Method (no complicated computations needed; simply see the outcomes)

With results in ten minutes, this is the most used technique in industrial settings:

Operation Steps: Record the timestamps of the two sensors that detected the step change after applying a fast step signal to both sensors at the same time (e.g., quickly opening and closing a valve to generate a sudden pressure change, or quickly injecting hot water to cause a sudden temperature change).

Criteria for Judgment:

Size of Clock Deviation

Judgment Outcome

Effects

Deviation ≤ ½ × Sampling Time

A little, acceptable divergence

No effect on comparison outcomes, no need for calibrating

Deviation ≤ 1 × Sampling Period ½ × Sampling Period

Crucial, able to be somewhat modified for recompense

Results are barely affected, and only fixed deviation adjustment is needed.

Deviation greater than 1 x Sampling Time

Excessive; calibration is necessary

Data mismatch at the same time point may result from this, which would immediately raise the false alert rate.

For instance, 10Hz sampling, sampling period = 100 ms, variation > 100 ms is too great, and ≤ 50 ms is acceptable.

 

II. Statistical Judgment Method for Data (Accurate Quantification, Fit for Formal Experiments)

The false alarm rate can be used to confirm the impact of variance if exact proof is needed:

After aligning the clock and testing constantly for two days, determine the current false alarm rate while maintaining the same algorithm parameters.

Calculate the false alarm rate after purposefully misaligning the clock to simulate severe deviation and testing constantly for two days.

Judgment Criterion: The initial clock deviation has been exceeded if the false alarm rate rises by more than 2% following clock misalignment. If the rise is less than 1%, the variation has no effect on the results; however, large variances necessitate calibration.

 

III. Visual Observations to Support Decision-Making

It is essentially possible to conclude that the clock deviation is excessively significant if the following things take place:

During regular production, there are no issues, however false alarms are regularly set off, and inspecting the sensors and installation reveals no issues.

The algorithm does not sound an alarm despite the obvious occurrence of sensor drift, and the false alarm rate is unusually high.

The standard deviation of the deviation under steady state is more than twice as large as it was before to calibration, and the deviation data varies much more than usual.

In summary, it can be concluded that the deviation is too big and that calibration will resolve the issue if it surpasses the single sampling time you specified or if there is an inexplicable rise in false alarms.