Research Of Condenser Fault Diagnosis Based On Pattern Recognition

The condenser is a major auxiliary equipment of steam turbine,that can raise the decreased value of steam enthalpy effectively,that has a great influence to the steam turbine unit safe and economic operation.Due to the condenser running in high vacuum environment,the condenser are prone to failure in the process of running,Condenser failure is one of the main causes of turbine downtime.With the unit capacity of steam turbine increasing,the damage failure will also continue to increase.so in-depth study of the working mechanism of steam systems and operational characteristics,the establishment of the condenser system optimization model to study the fault diagnosis method and implemented to condenser Fault diagnosis for improving crew safety,economic operation has important theoretical significance and practical value.This article introduces the effect of condenser system,principle and system structure,divide the condenser into non-condensing zone of shell side,steam zone of shell side,water zone of shell side and condenser circulating water side.These mathematical models based on STS model supporting are combined with other systems of power plant,into a simulation system.Which static and dynamic characteristics meet the requirements of the scene.After creating a mathematical model of condenser,the model was used to extract the fault data,First,we analyze the common faults of condenser in theory,list the common condenser fault phenomenon,combining with the characteristics of the structure of the condenser,the existing literature and the DCS distribution of measuring points and summed up the typical fault symptoms of fractional condenser from many causes of condenser fault.Based on the fault symptom set and failure data extracted from simulator,using the nearest neighbor method,similarity theory and support vector machine three kinds of pattern recognition methods for fault diagnosis,compared the three methods of diagnosis effect and time.Finally similarity modeling approach is applied to the condenser fault early warning,that increased the amount of advance for fault diagnosis,and achieved good results.