دانلود رایگان مقاله مقایسه فیلتر کالمن و نابرابری ماتریس خطی بر اساس فیلتر H بینهایت

Abstract

This paper deals with delay compensation of vanadium Self Powered Neutron Detectors (SPNDs) using Linear Matrix Inequality (LMI) based H-infinity filtering method and compares the results with Kalman filtering method. The entire study is established upon the framework of neutron flux estimation in large core Pressurized Heavy Water Reactor (PHWR) in which delayed SPNDs such as vanadium SPNDs are used as in-core flux monitoring detectors. The use of vanadium SPNDs are limited to 3-D flux mapping despite of providing better Signal to Noise Ratio as compared to other prompt SPNDs, due to their small prompt component in the signal. The use of an appropriate delay compensation technique has been always considered to be an effective strategy to build a prompt and accurate estimate of the neutron flux. We also indicate the noise-response trade-off curve for both the techniques. Since all the delay compensation algorithms always suffer from noise amplification, we propose an efficient adaptive parameter tuning technique for improving performance of the filtering algorithm against noise in the measurement.

5. Conclusion

In this paper, we have presented a study that compares the Kalman filter based approach and H1 filter based approach with respect to delay compensation of SPNDs. The results indicate that although both the method can be modified to build a suitable delay compensation algorithm, the H1 due to its strict upper bound on performance, results into higher filter gain and provides a lower response time with slightly increase noise. We have built the trade-off curves for both the technique that makes the selection process much simpler for choosing the parameters. The recursive implementation and more number of tuning parameters makes the H1 a better choice. We also propose the adaptive technique for H1 filtering based delay compensation, which reduces the steady state noise significantly. The results were simulated for real time sensor data during reactor trip from full power and show that the SPND with adaptive delay compensation can become a good alternative with respect to better SNR as compared to the prompt SPNDs.