Knowing the full state of an industrial plant may be important for both safety and efficient control. However, only a limited amount of measurement data may be available. This may be due to the expense or difficulty of incorporating sensors into a plant. With limited measurement data, various techniques may be used that will produce an estimate of the entire state of the system. State estimators provide a cheap alternative to adding new sensors or upgrading existing ones.
Control engineering concerns the efficient automated control of diverse kinds of industrial and other processes. An example would be regulating the flow of gas within a gas network, while minimising the amount of energy expended on driving the gas through the pipes. This is achieved by monitoring a limited number of flow and pressure measurements from the network, and using this information to guide the control of the compressors and valves. This is known as feedback control.
Fault detection and diagnosis
A related problem is fault detection and diagnosis (FDD). Whether certain types of faults can be detected and diagnosed depends on a detailed analysis of the underlying process model and the type of measurement data available. When a fault, such as a gas leak, is detected this information may be passed to the control system in order for it to respond appropriately.
Combining state estimation, control and fault detection
In many applications there is a need for state estimation and control systems to work in tandem with monitoring systems for fault detection and diagnosis. In some control systems, potential fault states are included with the normal process variables in the state estimation algorithms used to guide the controllers. State estimators, control systems and fault detection methods are therefore often linked.
To contact us: please e-mail Dr Simon Stringer (CEO) at: