Knowledge and Data Integration with Linguistic Equation

Linguistic equations are designed for integrating knowledge and data in development of nonlinear multivariable systems for intelligent process analysis, process control, fault diagnosis and forecasting. Insight to the process operation is maintained since all the modules can be assessed by expert knowledge and membership definitions relate measurements to appropriate linguistic terms on different operating areas. The linguistic equation (LE) approach increases the performance by combining various specialised models in a case-based approach: models can be generated automatically from data; various fuzzy models can be represented by LE models; neural networks and evolutionary computing can be used in tuning. Individual LE modules can be transformed to fuzzy rule-based systems. The LE approach is also successfully extended to dynamic simulation and used in intelligent controller design.