在Donoho D L和Johnston IM提出的多分辨分析小波阈值去噪方法的基础上,提出了一种新的双变量阈值函数。采用新的阈值函数的去噪效果无论在视觉效果,还是在信噪比增益和最小均方意义上均优于传统的硬阈值和软阈值,克服了采用硬阈值法去...在Donoho D L和Johnston IM提出的多分辨分析小波阈值去噪方法的基础上,提出了一种新的双变量阈值函数。采用新的阈值函数的去噪效果无论在视觉效果,还是在信噪比增益和最小均方意义上均优于传统的硬阈值和软阈值,克服了采用硬阈值法去噪效果不佳和软阈值法过度光滑使信号失真的缺点。通过仿真实验结果,表明该方法的有效性和优越性。展开更多
In order to reduce the feedback load of multi-user orthogonal frequency division multiplexing （ OFDM ） -based wireless systems, a practiral limited bits feedback precoding algorithm is proposed with direct source-de...In order to reduce the feedback load of multi-user orthogonal frequency division multiplexing （ OFDM ） -based wireless systems, a practiral limited bits feedback precoding algorithm is proposed with direct source-destination link based on amplify-and- forward cooperative relay network under frequency selective fading channels. Using joint minimum mean square error（MMSE） filter, the receiving decoding matrix is designed for each user in the paper. Source precoding （beamforming） matrix is optimized with convex function of weight mean square error （MSE）. Relay precoding matrix is obtained under MSE decomposition and convex optimization. The precoding matrix index is fed back for clustered subcarrier of OFDM with limited feedback. Then using interpolation algorithm, all precoding matrices are achieved at base station （BS） and relay nodes. Simulations indicate the effectiveness of the proposed limited feedback joint precoding and beam_formlng design. The proposed method can improve bit error rate （BER） performance and obtain better sum-rate performance in contrast to existing algorithms. It displays the BER performance is close to that of the unquantified precoding feedback method.展开更多
It is an important step in the online monitoring of partial discharge (PD) to extract PD pulses from various background noises. An adaptive de-noising method is introduced for adaptive noise reduction during detection...It is an important step in the online monitoring of partial discharge (PD) to extract PD pulses from various background noises. An adaptive de-noising method is introduced for adaptive noise reduction during detection of PD pulses. This method is based on Wavelet Transform (WT), and in the wavelet domain the noises decomposed at the levels are reduced by independent thresholds. Instead of the standard hard thresholding function, a new type of hard thresholding function with continuous derivative is employed by this method. For the selection of thresholds, an unsupervised learning algorithm based on gradient in a mean square error (MSE) is present to search for the optimal threshold for noise reduction, and the optimal threshold is selected when the minimum MSE is obtained. With the simulating signals and on-site experimental data processed by this method, it is shown that the background noises such as narrowband noises can be reduced efficiently. Furthermore, it is proved that in comparison with the conventional wavelet de-noising method the adaptive de-noising method has a better performance in keeping the pulses and is more adaptive when suppressing the background noises of PD signals.展开更多
文摘在Donoho D L和Johnston IM提出的多分辨分析小波阈值去噪方法的基础上,提出了一种新的双变量阈值函数。采用新的阈值函数的去噪效果无论在视觉效果,还是在信噪比增益和最小均方意义上均优于传统的硬阈值和软阈值,克服了采用硬阈值法去噪效果不佳和软阈值法过度光滑使信号失真的缺点。通过仿真实验结果,表明该方法的有效性和优越性。
基金National Natural Science Foundation of China-Guangdong，Guangdong-Hong Kong Key Projects of Science and Technology,China，University-Industry Key Project of Department of Education of Guangdong Province,China，National Natural Science Foundation of China
文摘In order to reduce the feedback load of multi-user orthogonal frequency division multiplexing （ OFDM ） -based wireless systems, a practiral limited bits feedback precoding algorithm is proposed with direct source-destination link based on amplify-and- forward cooperative relay network under frequency selective fading channels. Using joint minimum mean square error（MMSE） filter, the receiving decoding matrix is designed for each user in the paper. Source precoding （beamforming） matrix is optimized with convex function of weight mean square error （MSE）. Relay precoding matrix is obtained under MSE decomposition and convex optimization. The precoding matrix index is fed back for clustered subcarrier of OFDM with limited feedback. Then using interpolation algorithm, all precoding matrices are achieved at base station （BS） and relay nodes. Simulations indicate the effectiveness of the proposed limited feedback joint precoding and beam_formlng design. The proposed method can improve bit error rate （BER） performance and obtain better sum-rate performance in contrast to existing algorithms. It displays the BER performance is close to that of the unquantified precoding feedback method.
文摘It is an important step in the online monitoring of partial discharge (PD) to extract PD pulses from various background noises. An adaptive de-noising method is introduced for adaptive noise reduction during detection of PD pulses. This method is based on Wavelet Transform (WT), and in the wavelet domain the noises decomposed at the levels are reduced by independent thresholds. Instead of the standard hard thresholding function, a new type of hard thresholding function with continuous derivative is employed by this method. For the selection of thresholds, an unsupervised learning algorithm based on gradient in a mean square error (MSE) is present to search for the optimal threshold for noise reduction, and the optimal threshold is selected when the minimum MSE is obtained. With the simulating signals and on-site experimental data processed by this method, it is shown that the background noises such as narrowband noises can be reduced efficiently. Furthermore, it is proved that in comparison with the conventional wavelet de-noising method the adaptive de-noising method has a better performance in keeping the pulses and is more adaptive when suppressing the background noises of PD signals.
