It is well known tendency that the demands to keep the best safety and availability of a plant and for minimizing of maintenance work and cost.
As one solution to meet these demands, a condition based monitoring and diagnosis system will be put to use.
A steam power plant is consists of several thermo-flow subsystems excluding the rotating part and they have relatively slow dynamic responses. Even though the measurement can be carried out by the conventional sensors applicable to the process control system, the location and the degree of abnormal cannot detect directly with them.
In such kind of subsystem, a particular abnormality is often indicated by changes in several measurable parameters. Then the diagnosis shall be carried out by means of a discrimination software.
It is necessary to maintain the requirements for above mentioned discrimination among the possible cause of abnormalities and for making available instruction of the abnormal condition, the software of diagnosis shall have a quantitative manner. However the quantitative processing of the measured parameters may lead some erroneous results owing to their numerical errors. We have to evaluate the measurement repeatabilities and accuracies and to consider an appropriate countermeasures in the diagnosis system.
This paper reports the following items based upon our research and development.
General concept and effectiveness evaluation of plant diagnosis.
Outline of diagnosis for steam power plant esp. thermo-flow subsystem.
Detailed and practical explanation of diagnosis for thermo-flow subsystem.
Instrumentation and output of diagnosis system.
View full abstract