Swinging door deadband: rate only

The following diagram depicts an ideal case, where the incoming signal is noise-free and with a proper rate deadband specification only (no value deadband or deadband override period).

Assume point 0 has been stored on disk. The system waits for point 2 to arrive before making its next storage decision. When point 2 is received, the storage engine calculates the change in slope as follows:

Slope0_1 is considered the base slope, and Slope1_2 is considered the current slope.

Slope0_1 = (Value1 - Value0) / (Time1 - Time0)

Slope1_2 = (Value2 - Value1) / (Time2 - Time1)

Slope_Change_Percent = 100* | (Slope1_2 - Slope0_1) / Slope0_1 |

If

Slope_Change_Percent > Rate_Deadband_Specified

In other words, if the percentage change in slope is greater than the specified rate deadband, the storage engine goes ahead and stores point 1 on disk. Next, it receives point 3. The base slope for point 2 will be the slope between points 1 and 2. The current slope will be the slope between points 2 and 3 only if point 1 was stored. If point 1 was not stored, then the base slope for point 2 will be the slope between points 0 and 1, and the current slope will be the slope between points 2 and 3.

The base slope for an evaluation point is not changed unless the previous point is stored; otherwise, the base slope will be the last known current slope that caused a point to be stored on disk.

Assuming point 1 is stored, because the slope between points 2 and 3 is about the same as the slope between points 1 and 2, the rate deadband criterion is not satisfied, and point 2 is discarded. When point 4 is received, the slope change calculation results in point 3 being discarded, and so on until point 6 arrives. Now the rate deadband criterion is satisfied (slope change between points 5 and 6 and points 1 and 2 is greater than the rate deadband specified), and point 5 is stored on disk.

The arrival of point 7, likewise, discards point 6 even though the actual slope between point 6 and point 7 may be quite high, and may even be higher than the rate deadband specified, it is not sufficiently different from the slope between points 5 and 6 to qualify point 6 to be stored. Following this logic through until point 12 is received results in the storage on disk of points 10 and 11, discarding all the other points in between.

Point 13 illustrates the effect of the real-time window setting. Under normal circumstances, point 12 would not qualify to be stored. If, however, the elapsed time between receiving point 12 and point 13 exceeds the time window in which the storage engine is able to store point 12 as a real-time point, point 12 is stored anyway, and the value of the SysRateDeadbandForcedValues system tag is incremented. In other words, if, while the system waits for point 13 to arrive, the timestamp of point 12 becomes so old that it reaches the limit for the real-time window, point 12 is stored regardless of whether it is outside the deadband.

The SysRateDeadbandForcedValues system tag counts the number of "extra" points stored as a result of an insufficient real-time window for swinging door storage.

When point 14 arrives, the base slope for evaluating point 13 is between points 11 and 12, and not between points 12 and 13, because point 12 was stored due to the real-time window expiration. A point stored due to the real-time window does not re-establish the base slope; only points stored due to exceeding the rate change causes the base slope to be re-established. Then "normal" rate change evaluation resumes, resulting in point 13 being stored, and so on.