| 摘要: |
| 提出了一种基于单神经元的直流电压在线辨识方法。该方法根据逆变器的开关平均模型建立了直流电压辨识的参考模型和参数可调的单神经元，以直流电压作为单神经元的加权系数，通过学习算法不断调整加权系数，从而实现辨识值跟踪实际值。基于有功平衡理论直接对电网电流进行控制，因而无需检测负载电流和逆变器输出电流，有效地降低了有源滤波器的成本和硬件的复杂度，使得单相有源滤波器只需2个传感器用于检测电网的电压和电流。研制了一台实验样机，并利用Matlab绘出直流电压的辨识曲线，实验结果证明了所提方法的可行性。 |
| 关键词: 有源滤波器 谐波 单神经元 开关平均模型 |
| DOI： |
| 分类号: |
| 基金项目:湖南省自然科学基金项目（06JJ50101） |
|
| APF without DC voltage sensor |
|
TANG Xin, HU Xinxin
|
|
College of Electrical Engineering，Changsha University of Science and Technology，Changsha 410076，China
|
| Abstract: |
| A method of online DC voltage identification based on single neuron is proposed，which，according to the DC voltage identification reference model based on switch average model，takes the DC voltage as the weights of single neural and adjusts them by learning algorithm to realize the identification values well tracking the actual values. As the supply current is directly controlled based on the energy balance theory，the detection of load current and inverter output current is not necessary，which effectively reduces the hardware complexity and cost of APF（Active Power Filter）. Only two sensors are needed for supply voltage and current detection in a single-phase APF. A prototype is developed and DC voltage identification curve is plotted with Matlab. The experimental results verify the feasibility of the proposed method. |
| Key words: APF harmonics single neuron switch average model |
