With the help of this model and Liquid Rocket Engine(LRE) historical test data,the linear correlation hypothesis testing of 7 turbopump vibration parameters and that of 13 traditional statistical features were performed here. Then,the turbopump fault detection sensitivity and stability of statistical features and the normality of turbopump vibration data were also analyzed.

With a liquid rocket engine(LRE) historical test data,this algorithm is validated. These results show that there is only one class when the algorithm is used to healthy turbopump vibration data,and the distance between the neighboring neuron is less than 0.1;

(2) The framework and approaches of turbopump post test data analysis were presented. Using the traditional time domain features of vibration signals, the turbopump history test data were analyzed, and based on the results some conclusions were drawn.

One wishes for a new Max Weber to put their broad statements to some kind of historical test.

While previous studies were based on re-enactments from historical test data, the present study is the first to examine how well these adaptive methods function in a real-time testing situation.

All historical test reports were produced by Lakefield, now SGS Lakefield Research.

For distributors who filed on an historical test year, this is 2004 information.

Historical test results are included in appendices for traceability and completeness.

In order to enhance the safety of liquid rocket engine (LRE) turbopump and minimize the damage of its faults, a health monitoring system of turbopump (TP-HMS) is designed for a large LRE. After the realization of the subsystems for signal acquisition, real-time fault detection, post test data analysis and real-time database support, the function and the flow of execution are investigated and analyzed. Then, the multi-feature adaptive threshold compositive decision-making algorithm (MATA) is validated...

In order to enhance the safety of liquid rocket engine (LRE) turbopump and minimize the damage of its faults, a health monitoring system of turbopump (TP-HMS) is designed for a large LRE. After the realization of the subsystems for signal acquisition, real-time fault detection, post test data analysis and real-time database support, the function and the flow of execution are investigated and analyzed. Then, the multi-feature adaptive threshold compositive decision-making algorithm (MATA) is validated with the historical data from LRE test as well as real-time data from rotor test platform. To obtain the information of fault signals, the adaptive spectrogram is applied in the data analysis. It is shown that MATA can detect the faults in real time and give no false alarm in this case. Through the application of adaptive spectrogram, the cross-term interference is suppressed. The time frequency information is obtained accurately. Therefore, the conclusion is drawn that the TP-HMS design is suitable for the health monitoring of LRE turbopump.

In order to abscind the redundant vibration parameters and statistical features, and enhance the real time capability of turbopump fault detection,the linear correlation hypothesis test model was built.With the help of this model and Liquid Rocket Engine(LRE) historical test data,the linear correlation hypothesis testing of 7 turbopump vibration parameters and that of 13 traditional statistical features were performed here.Then,the turbopump fault detection sensitivity and stability of statistical features and...

In order to abscind the redundant vibration parameters and statistical features, and enhance the real time capability of turbopump fault detection,the linear correlation hypothesis test model was built.With the help of this model and Liquid Rocket Engine(LRE) historical test data,the linear correlation hypothesis testing of 7 turbopump vibration parameters and that of 13 traditional statistical features were performed here.Then,the turbopump fault detection sensitivity and stability of statistical features and the normality of turbopump vibration data were also analyzed.The data statistic analysis results show that a majority of turbopump vibration and statistical features are evidently linear correlation;and the amplitude statis-tical features fault sensitivity is weak,but their fault stability is strong;while the dimensionless ones are just the contrary.What's more,the normal turbopump vibration data obey normal distribution;however the abnormal ones do not obey normal distribution anymore.So,3 vibration parameters and 3 statistical features,which are weak linear correlation and well reflect the faults characteristic,are selected for the LRE turbopump fault detection.

In order to detect the turbopump fault short of fault samples,the spectrums of turbopump vibration signals were analyzed,and the frequency band energy ratio was selected as the fault feature of those signals.After SOM competitive learning theory and U matrix description of clustering results were discussed,the frequency-band-energy-ratio-based SOM algorithm for turbopump fault detection is presented,and the selection of the best matching unit(BMU) and the adaptive upgrade of their weight vectors are also realized...

In order to detect the turbopump fault short of fault samples,the spectrums of turbopump vibration signals were analyzed,and the frequency band energy ratio was selected as the fault feature of those signals.After SOM competitive learning theory and U matrix description of clustering results were discussed,the frequency-band-energy-ratio-based SOM algorithm for turbopump fault detection is presented,and the selection of the best matching unit(BMU) and the adaptive upgrade of their weight vectors are also realized in this algorithm.With a liquid rocket engine(LRE) historical test data,this algorithm is validated.These results show that there is only one class when the algorithm is used to healthy turbopump vibration data,and the distance between the neighboring neuron is less than 0.1;while there are two or more classes when the algorithm is used for faulty turbopump vibration data,and the distance between the neighboring neuron is greater than 0.1.Therefore the algorithm can effectively detect the turbopump fault.