【摘要】：有源電力濾波器(Active Power Filter, APF)因其能實(shí)時(shí)、動(dòng)態(tài)地跟蹤補(bǔ)償諧波和無(wú)功電流,已成為目前治理諧波污染最有效的手段。然而,提高APF的性能的必要前提是實(shí)現(xiàn)快速并精確的諧波電流或畸變電流檢測(cè)。本文通過(guò)總結(jié)國(guó)內(nèi)外諧波電流檢測(cè)方法,并以傅氏級(jí)數(shù)和自適應(yīng)濾波算法為理論基礎(chǔ),對(duì)諧波電流或畸變電流的自適應(yīng)檢測(cè)算法進(jìn)行了理論研究。 本文首先給出了基于自適應(yīng)噪聲對(duì)消技術(shù)(ANCT)和基于Adaline的兩種自適應(yīng)諧波電流檢測(cè)模型的基本原理,并分析了兩種模型在相同自適應(yīng)算法的條件下的檢測(cè)性能、缺陷和適用范圍,以及不同算法對(duì)相同模型的檢測(cè)性能的影響。其結(jié)果表明：在相同算法的條件下,兩種模型的收斂速度一致,后者的穩(wěn)態(tài)誤差小(理論上可以為0)但計(jì)算量較大,前者只適用于APF的諧波檢測(cè)環(huán)節(jié),而后者還可用于評(píng)估電能質(zhì)量；不同算法對(duì)相同模型檢測(cè)性能的影響由算法的固有性能決定。另外,本文也介紹了基于ANCT的諧波電流檢測(cè)模型應(yīng)用在三相電路中的基本原理。 其次,通過(guò)分析并總結(jié)基于LMS的諧波電流檢測(cè)算法和現(xiàn)有的一些變步長(zhǎng)LMS諧波電流檢測(cè)算法的缺點(diǎn),本文提出了基于改進(jìn)的箕舌線函數(shù)的變步長(zhǎng)LMS自適應(yīng)畸變電流檢測(cè)算法。該算法采用誤差信號(hào)(即當(dāng)前時(shí)刻負(fù)載電流與濾波器輸出的差)與參考輸入信號(hào)的瞬時(shí)互相關(guān)的時(shí)間均值估計(jì)來(lái)消除畸變電流對(duì)步長(zhǎng)更新的干擾,并通過(guò)改進(jìn)的箕舌線函數(shù)來(lái)調(diào)節(jié)該算法的步長(zhǎng)。因此,在保證動(dòng)態(tài)響應(yīng)時(shí)間不增大的前提下,該算法具有更小的穩(wěn)態(tài)誤差,并且當(dāng)負(fù)載電流發(fā)生突變時(shí),該算法也有較強(qiáng)的跟蹤能力。 然后,針對(duì)LMS和RLS算法在諧波檢測(cè)中的缺點(diǎn),本文提出了加權(quán)LMS自適應(yīng)畸變電流檢測(cè)算法。在相同穩(wěn)態(tài)誤差的條件下,該算法的收斂速度比LMS算法更快,并且對(duì)于突變的負(fù)載電流,該算法也比RLS算法有更好的跟蹤能力,這是因?yàn)樵撍惴ǖ臋?quán)值更新不受畸變電流的干擾。此外,該算法的計(jì)算復(fù)雜度居于LMS和RLS算法之間。 通過(guò)執(zhí)行單相電路和三相電路的畸變電流檢測(cè)的計(jì)算機(jī)仿真,驗(yàn)證了本文所做的工作。
[Abstract]:Active power filter (Active Power Filter, APF) has become the most effective method to control harmonic pollution because of its real-time and dynamic tracking and compensating harmonic and reactive current. However, fast and accurate detection of harmonic current or distortion current is necessary to improve the performance of APF. In this paper, based on the theory of Fourier series and adaptive filtering algorithm, the adaptive detection algorithm of harmonic current or distortion current is studied theoretically by summing up the detection methods of harmonic current at home and abroad. In this paper, the basic principles of two adaptive harmonic current detection models based on adaptive noise cancellation technique (ANCT) and adaptive harmonic current detection model based on Adaline are presented, and the detection performance, defects and application range of the two models under the same adaptive algorithm are analyzed. And the effect of different algorithms on the detection performance of the same model. The results show that the convergence rate of the two models is the same under the condition of the same algorithm, the steady-state error of the latter is small (theoretically it can be 0) but the calculation is large, the former is only suitable for the harmonic detection of APF. The latter can also be used to evaluate power quality, and the influence of different algorithms on the performance of the same model is determined by the inherent performance of the algorithm. In addition, this paper also introduces the basic principle of harmonic current detection model based on ANCT in three-phase circuit. Secondly, by analyzing and summarizing the shortcomings of the harmonic current detection algorithm based on LMS and some existing LMS harmonic current detection algorithms with variable step size, this paper proposes a variable step size LMS adaptive distortion current detection algorithm based on the improved tongue line function. In this algorithm, the error signal (i.e. the difference between the current load current and the output of the filter) and the time mean of the instantaneous cross-correlation between the reference input signal and the current load current are used to eliminate the interference of the distortion current to the update of the step size. The step size of the algorithm is adjusted by the improved tongue line function. Therefore, without increasing the dynamic response time, the algorithm has a smaller steady-state error, and it also has a strong tracking ability when the load current changes. Then, aiming at the shortcomings of LMS and RLS algorithms in harmonic detection, a weighted LMS adaptive distortion current detection algorithm is proposed in this paper. Under the same steady state error condition, the convergence speed of the algorithm is faster than that of the LMS algorithm, and the algorithm has better tracking ability than the RLS algorithm for the sudden load current. This is because the weight update of the algorithm is not disturbed by the distorted current. In addition, the computational complexity of the algorithm is between LMS and RLS. The work done in this paper is verified by the computer simulation of detecting the distortion current of single-phase circuit and three-phase circuit.
【學(xué)位授予單位】：西南交通大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2014
【分類號(hào)】：TM933.1

【參考文獻(xiàn)】

相關(guān)期刊論文 前10條

1 石敏,吳正國(guó),尹為民;基于RLS算法的時(shí)變諧波檢測(cè)[J];電工技術(shù)學(xué)報(bào);2005年01期

2 劉國(guó)海;呂漢聞;陳兆嶺;劉慧;;基于修正遺忘因子RLS算法的諧波電流檢測(cè)新方法[J];電工技術(shù)學(xué)報(bào);2010年01期

3 何英杰;鄒云屏;李輝;劉飛;;用于有源濾波器的一種新型諧波檢測(cè)算法[J];電力電子技術(shù);2006年02期

4 周曉霞;劉星橋;;一種新穎的變步長(zhǎng)自適應(yīng)諧波電流檢測(cè)法[J];電力電子技術(shù);2008年11期

5 李輝,李亦斌,鄒云屏,劉飛;一種新的變步長(zhǎng)自適應(yīng)諧波檢測(cè)算法[J];電力系統(tǒng)自動(dòng)化;2005年02期

6 曲軼龍;譚偉璞;馬W，

本文編號(hào)：2194737