电力变压器有源噪声控制系统抗干扰技术研究

发布时间：2018-05-26 21:41

本文选题：变压器 + 有源噪声控制　；参考：《武汉大学》2017年硕士论文

【摘要】：随着我国经济的高速发展,城镇化的迅速推进,使得原本位于城郊的高电压等级、大容量的变电站逐渐进入居民区和城市中心,而且随着人们对生活质量要求的提高,变压器噪声污染的问题受到广泛的社会关切。所以,治理变压器的噪声污染问题有很重要的意义。通过优化变压器的制作工艺来降低其辐射的噪声,不仅需要巨额投入,而且许多关键技术问题尚未解决。而被动噪声控制技术,如隔音、吸声和消声等,对于以低频为主的变压器噪声控制效果不明显。所以,研究变压器的有源噪声控制(Active Noise Control,ANC)技术有实际应用价值。论文研究了冲击性干扰和次级声反馈对变压器ANC系统的影响,并从算法和系统结构上做了相应的改进,进一步提高系统的抗干扰能力,为其实用化做铺垫。声传感器拾取的冲击性干扰信号对ANC系统的稳定性影响很大。论文详细的分析参考信号和误差信号中的冲击性干扰对ANC系统控制器的影响,即当采用经典的滤波最小均方算法对控制器自适应更新时,控制器易受冲击性干扰的影响而剧烈震荡甚至是发散。在总结分析已有改进算法的基础上,论文提出归一化语音压缩μ函数变换的最小均方(Normalized Filtered-xμ Law Mean square,NFxμLMS)算法。该算法采用语音压缩μ函数变换后的误差信号的均方作为目标函数,并对参考信号做归一化处理以提高算法的鲁棒性。NFxμLMS算法可以根据不同变电站的噪声水平选择不同的压缩系数以提高算法的实用性。最后,四组对比仿真实验说明NFxμLMS算法的收敛性和鲁棒性。参考声传感器拾取到次级声源辐射的次级噪声,影响了参考信号的质量,进而影响了 ANC系统的稳定性。为了从根本上避免次级声反馈问题,论文研究了内部参考信号方法应用到变压器有源噪声控制中的可行性。首先,对比研究了基于内部参考信号方法的并联型和直接型多频率ANC系统的系统特性,采用并联型结构构建变压器有源噪声控制系统。然后,分析了内部合成参考信号的变压器有源噪声控制系统的降噪量与迭代步长和频率偏差的关系。分析得出,系统降噪量随着频率偏差的增大而降低,而当变压器噪声与内部给定参考信号的频率偏差较小时,可以通过选择较大的迭代步长来保持系统降噪量的结论。最后,几组仿真算例验证了内部给定参考信号的方法应用到变压器噪声控制的可行性。在并联型多频率ANC系统中,各子系统均采用误差声传感器采集的系统误差信号更新,导致各子系统的收敛特性和降噪量较差。论文首先分析了其他子系统残余误差信号和背景噪声对子系统收敛特性的影响。然后,论文构造了完全并联型多频率变压器ANC系统。新系统采用限制零极点方法设计出一组窄带滤波器,对误差信号做分频处理,以便将误差信号中各频率分量输入到相应地子系统中,以实现完整意义上各子系统的并联运行,使得各子系统控制器的自适应更新过程互不影响。最后,仿真算例对比了改进前后系统的收敛速度和收敛精度,验证了新系统的实用性。
[Abstract]:With the rapid development of China's economy and the rapid development of urbanization, the large capacity substation has gradually entered the residential area and the city center, which is originally located in the suburb of the city. And with the improvement of people's quality of life, the problem of noise pollution of transformer is widely concerned. Therefore, the noise of transformer is harnessing. The problem of pollution is of great significance. By optimizing the production process of the transformer to reduce the noise of its radiation, it not only needs a huge amount of input, but many key technical problems have not been solved. But the passive noise control technology, such as sound insulation, sound absorption and noise elimination, is not obvious to the noise control effect of the low-frequency transformer. The Active Noise Control (ANC) technology has practical application value. The impact of impact interference and secondary acoustic feedback on the transformer ANC system is studied in this paper, and the corresponding improvement is made from the algorithm and system structure to further improve the anti interference ability of the system and to pave the way for its utility. The impact interference signal has great influence on the stability of the ANC system. In this paper, the impact of the impact interference on the ANC system controller is analyzed in detail, that is, when the classical least mean square algorithm is used to update the controller adaptively, the controller is easily affected by the impact of the impact disturbance and even shocks the controller. On the basis of summarizing and analyzing the existing improved algorithms, the paper proposes the least mean square (Normalized Filtered-x Law Mean square, NFx LMS) algorithm for the normalized speech compression muon transform. The algorithm uses the mean square of the error signal of the speech compression muon function as the target function, and the normalization of the reference signal is done. In order to improve the robustness of the algorithm,.NFx mu LMS algorithm can select different compression coefficients according to the noise level of different substations. Finally, four groups of comparison simulation experiments show the convergence and robustness of the NFx mu LMS algorithm. The reference sound sensor picked up secondary noise from the secondary sound source and affected the reference signal. In order to avoid the secondary acoustic feedback, the feasibility of applying the internal reference signal to the active noise control of the transformer is studied in this paper. First, the system characteristics of the parallel and direct multi frequency ANC system based on the internal reference signal method are compared and studied. The active noise control system of the transformer is constructed with parallel structure. Then, the relationship between the noise reduction of the active noise control system of the transformer with internal synthesis reference signal and the iteration step and the frequency deviation is analyzed. The analysis shows that the noise reduction of the system decreases with the increase of frequency deviation, while the transformer noise and the internal reference signal are given. When the frequency deviation is small, the noise reduction of the system can be maintained by selecting the larger iteration step. Finally, several simulation examples verify the feasibility of applying the internal given reference signal to the transformer noise control. In the parallel multi frequency ANC system, the system errors collected by the error acoustic sensors are all mistaken. This paper first analyzes the influence of the residual error signals and the background noise on the convergence of the subsystems. Then, the paper constructs a fully parallel multi frequency transformer ANC system. The new system designs a group of narrow band filters using the restricted zero pole method. In order to realize the parallel operation of each subsystem in a complete sense, the adaptive updating process of each subsystem controller does not affect each other. Finally, the simulation example is compared to the convergence speed and convergence accuracy of the system before and after the improvement. The practicability of the new system is verified.
【学位授予单位】：武汉大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM41

【参考文献】