Reliable sensor measurements are essential for the undisturbed system operation. If a measurement of a crucial system parameter fails to deliver an accurate value, the system’s controls can lead the system out of an optimal operation. In worst-case scenarios, an unwanted system shut down and/or component faults can occur. By applying sensor validation techniques, critical sensors are monitored in a manner where the sensor validation software alerts an operator about a faulty sensor. When a faulty sensor is identified, it is isolated and a model-based calculated value is used. In this way, undisturbed operation of a system and time for sensor maintenance are ensured.
The sensor validation procedure that we offer consists of three steps:
Faulty sensor identification
Faulty sensor isolation
Faulty sensor accommodation
In the faulty-sensor-identification phase a problem with a sensor readings is identified. Once a faulty sensor is detected, the faulty-sensor-isolation phase is initiated disabling the values of failing sensor to be propagated to a control system. Faulty-sensor-accommodation phase follows by providing model-based estimated values. The fault accommodation provides “healthy” sensor values which enables proceeding of undisturbed system operation and provides time for execution of sensor maintenance.
The practice of sensor validation approach is to use multiple sensors for the same parameter measurements. In such a case, the deviation between the readings is used to detect a failing sensor. Such solution can be associated with high cost due to multiple sensors’ installations and their maintenance. Furthermore, redundant sensor applications are not always possible due to practical constraints of a system. In this regard, we offer the sensor validation solution with several advantages and relevant features:
- Additional costs related to redundant sensors and their maintenance can be avoided
- Suitable for applications where redundant sensors cannot be installed due to practical constraints of a system
- Further increase in safety is ensured when applied in parallel with classical methods
- Multiple sensors can be monitored simultaneously, thus further decreasing the costs of redundancy
- A failing sensor can be detected, isolated and replaced by a calculated values, ensuring undisturbed system operation and providing time for sensor maintenance
- Applicable on complex processes where analytical connections among parameters are not available (alternative to Kalman filter and Gas Path Analysis methods)
- Detection of simultaneous occurrence of multiple faulty sensors is in some cases possible as well
Application areas of the sensor validation solution are numerous. Applications extend practically to all types of measurements, i.e. pressures, temperatures, mass flows etc., and to all types of systems and/or components with certain amount of correlations among measured parameters.