How NormNet™ Works
NormNet™ is a real-time data derived anomaly detection system based on a five-stage approach that eliminates the need for physical models or detailed knowledge of a system.
NormNet™ uses transfer function models that incorporates data from all sensors and data sources for both the current and immediate past time periods. NormNet™ can easily adapt to the use of many sensors and other data sources to detect the probability and severity of future failures.
Input: NormNet™ accepts data from a variety of inputs, including on-line data bases, sensor systems and off-line data storage media. Pre-processors included in the input function ensure that the existing data can be used without modification at its source.
Learn: The NormNet™ PHM algorithms go through a learning process where they are "trained" on the characteristics of the system being monitored using data representing normal or expected environmental and operational conditions and derived from system data sources.
Detect: The PHM algorithms apply statistical methods and pattern recognition techniques using the relationship of data from all sources to detect abnormalities that, while small are statistically significant departures from normal and foreshadow potential failures. The algorithms detect extremely small abnormalities (<2% variance) while maintaining low false alarm rates (less than 0.5%).
Identify: The PHM algorithms evaluate the values and relationships of the information used to compute the probability of abnormality to perform a "most likely cause" analysis.
Presentation: The presentation element distributes the results from the PHM algorithms to the designated databases, management systems and personnel. Similar to the input element, post-processors ensure the easy distribution of PHM data to existing logistics and maintenance data repositories.
NormNet™ technology has been successfully tested using data from a variety of sources including Pratt & Whitney/USAF; Rolls Royce/Royal Air Force; Allison/Plains Exploration and Production Company. It is currently being adapted for use in trucking, shipping, helicopter, and water quality management applications. In every test, NormNet™ has demonstrated its ability to identify failures with sufficient warning that corrective action can take place before a catastrophic failure.