Siemens Energy, Inc. has been investigating the potential of a new approach to measuring the process gas temperature leaving the turbine of their heavy industrial gas turbine engines using an acoustic pyrometer system. This system measures the bulk temperature crossing a plane behinds the last row of turbine blades and is a non-intrusive measurement. It has the potential to replace the current intrusive multiple point measurement sensor arrays for both engine control and performance evaluation. The acoustic pyrometer is a device that measures the transit time of an acoustic pulse across the exhaust duct of the engine. An estimate of the temperature of the process fluid can be made from the transit time. Multiple passes may be made at various radial positions to improve the measurement. The gas turbine exhaust is a challenging environment for acoustic temperature measurement where there can be significant temperature stratification and high velocity. Previous applications of acoustic pyrometers to measure process gas temperature in power plants have been confined to applications such as boilers where rapid temperature changes are not expected and fluid velocity patterns are well known. The present study describes the results of acoustic pyrometer testing in an operating gas turbine engine under load using an active acoustic pyrometer system containing eight sets of transmitters and receivers, all external to the turbine exhaust flow path. This active method technology is based on the temperature dependence of the isentropic speed of sound from the simple ideal gas assumptions. Sound transmitters and receivers are mounted around the exhaust duct to measure the speed of sound. Very sophisticated topographical mapping techniques have been developed to extract temperature distribution from using any where from 2 to 8 sensors with up to 24 paths and 400 points. Cross correlation of sensor results to obtain topographical mapping of gas isotherms in a plane in full engine field tests have been conducted to prove the feasible of this technology on a gas turbine engine. The initial installation of the active acoustic pyrometer system in an engine exhaust was accomplished in 2009. All the tests indicate that the steady state measurements of the acoustic pyrometer system fall within 10C of the measured exhaust thermocouple data. An additional installation on a different model engine was subsequently made and data have been gathered and analyzed. Results of these tests are presented and future evaluation options discussed.

Noisy or distorted video/audio training sets represent constant challenges in automated identification and verification tasks. We propose the method of Mutual Interdependence Analysis (MIA) to extract "mutual features" from a high dimensional training set. Mutual features represent a class of objects through a unique direction in the span of the inputs that minimizes the scatter of the projected samples of the class. They capture invariant properties of the object class and can therefore be used for classification. The effectiveness of our approach is tested on real data from face and speaker recognition problems. We show that "mutual faces" extracted from the Yale database are illumination invariant, and obtain identification error rates of 2.2% in leave-one-out tests for differently illuminated images. Also, "mutual speaker signatures" for text independent speaker verification achieve state-of-theart equal error rates of 6.8% on the NTIMIT database.

The mean of a data set is one trivial representation of data from one class. Recently, mutual interdependence analysis (MIA) has been successfully used to extract more involved representations, or "mutual features", accounting for samples in the class. For example a mutual feature is a speaker signature under varying channel conditions or a face signature under varying illumination conditions. A mutual representation is a linear regression that is equally correlated with all samples of the input class. We present the MIA optimization criterion from the perspectives of regression, canonical correlation analysis and Bayesian estimation. This allows us to state and solve the above criterion concisely, to contrast the MIA solution to the sample mean, and to infer other properties of its closed form, unique solution under various statistical assumptions. We define a generalized MIA solution (GMIA) and apply MIA and GMIA in a text-independent speaker verification task using the NTIMIT database. Both methods show competitive performance with equal-error-rates of 7.5 % and 6.5 % respectively over 630 speakers.

Siemens has developed a novel approach for measuring the process gas temperature leaving the power turbine in their heavy industrial gas turbine engines using active acoustic tomography. Siemens has deployed this measurement technique on two test engines of different power ranges and different combustion and exhaust duct configurations. These engine tests have demonstrated that this technology is effective and robust. All working parts are outside the heat effective zone so, unlike the traditional intrusive point temperature measurement method, sensors are easily replaceable during engine operation. Bulk exhaust temperature is used in performance testing of industrial gas turbine engines and is a critical measurement for power production. Temperature distribution information in the exhaust plane is valuable for safe engine operation and can be used to prevent lifetime reduction due to hotspots or to monitor the burner flames. Siemens used broadband sound sources for the previously reported acoustic pyrometer experiments. This paper extends this work utilizing sparse time-frequency encoded sources to improve the robustness of time of flight estimation in the high noise area of the turbine exhaust. The goal is to achieve a higher signal to noise ratio between the emitted and received signals by focusing the acoustic energy into narrow time-frequency bins that are little affected by turbine noise. Different acoustic patterns are tested and compared to the previously used broadband source both in laboratory experiments and a turbine test bed. The patterns are evaluated regarding their noise robustness, sound pressure levels and narrow autocorrelation which are important for accurate time of flight estimation in high noise environments.