Support - Technical Resources
Big Data Approximating Control (BDAC)
Big Data Approximating Control (BDAC) is a new model-free approach, simultaneously addressing model development/system identification, state estimation, and control. It replaces model development/system identification with collection of a set of representative system trajectories. New advances in real time clustering were developed for collecting the representative system trajectories: State estimates and control actions are directly synthesized based on approximate pattern matching.
Online clustering with filtering for real time applications involving time series data
Simple techniques for online clustering for real time use in estimation and control were developed as part of the BDAC framework. However, these techniques are useful in any application where there is a need to collect online time series data and store a representative set for training purposes or direct use as in BDAC. This is especially applicable in cases where it is necessary to recognize new operating modes, and not forget rarely-seen operating conditions (a problem with typical techniques for forgetting old data such as recursive least squares). It is also useful when adaptation to slowly changing systems is needed. For instance, the real time clustering/filtering methods are useful for development and automatic updating of training sets for neural nets.
Fault detection and fault diagnosis
Guide to Fault Detection and Diagnosis
One primary technique for fault diagnosis used in Performity is based on causal directed graph models, described in general terms at Causal Models . However, these can be combined in hybrid systems when additional models are available
Guide to filtering
Guide to data reconciliation
Product documentation and examples
The following demonstration description serves as an example usage of Performity for fault diagnosis:
Pipeline Diagnosis Emphasizing Leak Detection: An Approach and Demonstration
Other technical resources
Other technical resources are also available at https://gregstanleyandassociates.com/whitepapers/whitepapers.htm