Fault Diagnosis And Protecting System For Large Steam Turbine

Turbine generator set is regarded as key equipments for the coal fire power station. It will sometime lead to enormous economically and financial losses and creating negative effects once the turbine generator set is out of control.With regard to the turbine generator set to be well developed toward creating a large scale and high parameters orientation,the scale and influencing impacts created by utilizing uni-generator set have been accordingly enhanced.This has increased greatly demands on the operational reliability for the exploration of turbine generator.It is believed to launch large scale of monitoring on large scale of turbine generator set,failure diagnosis method and new technology research,which exercise significant impacts on maintaining the operational quality and the safety for the sophistical turbine generator equipment.The structure of a large turbine generator tends to be complicated with the combination of electro-mechanical and electro-hydraulic.The failure signal is with background noise interfaces,nonlinear and nonstationary character.The rote of its transformation process is in combination with its natured decreasing sophistication often mixed with failure signal source together.This has created great difficulties for the correct diagnosis failure signal,feature study and capturing its technology.Obtaining information from correct failure signal is the critical technology for the failure signal diagnosis.The dissertation is undertaken by consideration the above explanations, together with the national 863 program and it aims at 'supercritical,ultra supercritical turbine generator monitoring,failure diagnosis technology and system research (2008AA04Z410)".Launching large turbine generator set research under the complicated background for the vibration signal monitoring,the primary research of the dissertation include three parts:The first part is to diversify new method from an independent component analysis;The second part is to generate new method based upon high-order statistical analysis;and the third part is to research on the monitoring and protection to the turbine generator.The main achievements and precise contents of the above three primary parts are summarized as the follows.1.With regard to the sophisticated operational environment,research on the failure source separation system and failure obtaining technology undertaken by using independent component analysis-ICA.The research can take advantage of turbine rotation system and vibration signal to testify and it can realize correct and sufficient information from the monitoring signal and the failure feature.The dissertation analysis different measurement on channel frequency,different vibration signals sets and different location to form each ICA result.With regard to the failure diagnosis of the turbine generator operation,the study has explored the possibilities to detect the failure diagnosis source from the measurement signals.It has also provided sufficient analysis case for the application of separation to detecting failure source on the turbine generator.2.Aiming at the engineering application of overcomplete bases ICA,the dissertation has raised virtual channel on the new method of ICA.It take advantage of some testified knowledge to the constitution of ICA known as(addition virtual channel), together with other measurement signal to proceed ICA separation,which can increase frequency on the input signal.Research firstly should be undertaken additional virtual channel model,and it can further testify the separation effects by using simulation signal. The additional virtual channel to the application of ICA method can be applied to turbine rotation system for system vibration.Especially it exercises great and actual impacts to the failure source separation and it can further testify the research on 'delay virtual channel ICA'.It means to take advantage of a specific moment to data on diagnosing independent component as virtual channel and the separation of ICA at future timing.In conducting such way,it can use few sensors measurement signal to realize the separation of the faulty source and it can also provide many good solutions for the application of ICA.3.In the consideration of convolution mixing ICA model for the best fitting the turbine rotation system and separation issue,the dissertation raised a kind of analysis on the fourier transform model,it can transfer from rotation mix to linear mix relation. Hence it can take advantage of ICA calculation method to separate vastly from the ICA analysis.The research can make use of actual turbine measurement signal to testify the above method.The result of conducting such research demonstrates,frequency range of ICA can detect possible some tiny and weaken failure information.The result of the separation effects is clear and it has some obvious advantages in comparison with ICA method.4.Following the principle of the theory in combination with practices,with a series of objective rectification and research on the high-order spectrum natured analysis,this research has explored advance level statistical analysis to obtain the faulty feature variables and accumulated bispectrum and bicoherence spectrum analysis,1(1/2) dimension spectrum analysis including un normal vibration signal of turbine generator, gear cracking fault vibration signal point.5.With the consideration of the foundation on detecting normal vibration signal monitoring technology,the study researches mostly on turbine generator set monitoring system and its reliability design technology.The research is mainly embedded system in operation protection function for on line monitoring system.The research includes building up typical faulty database for the turbine generator,researching intelligence faulty diagnosis system,developing multi physics amount and long range faulty diagnosis system.The research aims at upgrading method,technology and system with further rectification and this is believed to enhance the diagnosis method,technology and reliabilities.