Note1: This article is based on chapter 4 of my white paper titled 7 Key CBTC Functions Transit Operators Must Understand. Downloaded it here.
Note2: This post is part 2 and it's better understood if part 1 is read first.
Predicting a failure before it occurs is the Holy Grail for maintenance personnel and predictive maintenance is the purpose of level 3 diagnostics. Relying on the actual condition of the LLRU to predict when maintenance is required enables maintenance personnel to proactively plan corrective maintenance activities versus the reactive approach of the previous two diagnostic levels.
However, prediction demands data because the prediction algorithm creates a normal operating baseline based on historical data for each LLRU. The algorithm devours data for each LLRU and the architecture must feed this appetite.
The prediction algorithm resides on the DMS but the key element of a level 3 architecture is its ability to store large volumes of data for an extended period of time to feed the prediction algorithm.
Therefore, the CBTC system must have a built in long term storage capacity to allow the DMS and the prediction algorithm to see trends and predict failures; referred to as level 3 alarms.
Unlike level 1 and level 2 alarms, where the information has a direct impact to Service and therefore must be transmitted immediately, level 3 faults are not immediate threats (predicting failures in the future) but if not addressed, they will become Service-affecting faults.
For example, if the pulses of a speed sensor drift outside the normal pattern, the sensor is flagged as an LLRU needing corrective maintenance; if the voltage fluctuation of a power supply falls outside of its normal range, it is flagged as an LLRU needing corrective maintenance.
Predictive maintenance is the gold standard for any maintenance program but no Supplier has this capability; at the same time Operators are demanding prediction without understanding the infrastructure required for such a function.
Once a level 3 diagnostic architecture is implemented, predictive maintenance becomes a possibility; even if the Supplier can’t provide a predictive maintenance algorithm, the Operator can develop one on their own.
Capturing Telemetry
Capturing telemetry is a side benefit of a level 3 diagnostic architecture. It enables key data to be recorded, assisting with incident investigation such as an accident or a near miss. Data such as speed of the train, when brakes were applied, position of the switch or how accurately trains align at the station can be recorded.
Conclusion
Diagnostic capability is a critical function in a CBTC signalled system because the complexity is an order of magnitude higher than a traditional relay-based conventional signaled system. The high demand for diagnostic data requires a diagnostic architecture that supports the easy flow of information from the source of the problem to a central location where the maintenance personnel make decisions.
An effective diagnostic architecture will alert the maintenance personnel about the cause of the problem before they arrive at the equipment room or train; whereas an ineffective diagnostic architecture forces the maintenance personnel to investigate the cause of the problem at the equipment room or train.