Research Of Process Control System Fault Detection And Diagnosis Based On Correlation Coefficient

Automatic control is an important composing part of modern science andtechnology. It is widely used in fields related to national defense, petrochemical, aerospace,power, materials, etc. It is one of the important basic technology which is indispensable inmodern science and technology. The safety and quality of process control system, which isbased on the traditional control theory and technology, can be guaranteed only when theobjects, sensors, actuators and signal transmission channels of the system are in normalworking condition. It will affect normal operation and production of the system once onecertain part of the system is in wrong order. In serious situation, it can threat to propertyand personal safety, even bring immeasurable loss.The security of process systems which are especially in the presence of hightemperature, high pressure parameters, toxic and hazardous materials, has been givenparticular attention. The material transformation and energy conversion of the processproduction system, must comply with the law of conservation of matter and energy, as wellas physical, chemical and related forms/phenomenological constraints, which canguarantee correlation among some process variables. This correlation canbe characterized by correlation coefficient. On this basis, the study proposes a process faultdetection and diagnosis method based on correlation coefficients. The correlationcoefficient between the relevant process variables should be in the normal range (area) or tomeet certain constraints when the system is in normal operation. When the correlationcoefficient does not satisfy the constraints, it indicates that factors related to the productionprocess or the relevent implementing agency, process equipment and sensors occurabnormal or failure. By calculating the correlation coefficient and in-depth analysis of itsnature and physical significance, we give a recursive algorithm based on dynamic process variable correlation coefficient.Finally, a practical engineering case of distillation column control system is used toverify the effectiveness of the system fault detection and diagnosis method based onprocess control correlation coefficients, shows that the method has a good effect of processdiagnostics, industrial applications and promotional values.