Dsp. High-speed Test System In The Power Spectrum Estimation Algorithm

²⁵²C_f-source-driven neutron noise analysis method is a way of measuring and analyzing the reactivity of reactor or subcritical reactor, which detects subcritical neutron multiplication factor, prompt neutron decay constant, and so on. In order to obtain these data, we must at first calculate APSDs (Auto-Power Spectral Density) and CPSDs (Cross-Power Spectral Density) of the three signals in the system. Studies have been made on arithmetic realization of APSD estimate in a high-speed detect system, since the way of calculating CPSD is only a little different from that of calculating APSD.On the basis of the hardware of the detect system which uses ²⁵²C_f-source-driven neutron noise analysis method, I designed the emulation platform of the APSD estimate, as an important part of the whole detect system. The platform uses a floating point DSP (Digital Signal Processor) TMS320C6713, and a SDRAM (Synchronous Dynamic Random Access Memory). The whole emulation platform also includes a user's interface, which was developed by using VC++ 6.0, and a serial communication module that connects DSP and PC.The software for emulation is the most important part in this thesis. Various APSD estimate arithmetics including periodogram method, correlogram method, Blackman-Tukey method, Barlett method, Yule-Walker method, LDA, and DGA were implemented on CCS (Code Composer Studio). Some code optimization techniques especial using DSP Library highly accelerate the execution speed of the programs. At the end of the thesis there are the emulation results, followed by result analysis, with which one can select the right arithmetic more easily when using the detect system.