三相变流器神经网络滞环控制研究

神经网络技术在人工智能，自动控制以及模式识别等领域的研究与应用正方兴未艾，而滞环电流控制是一种传统常规的电流控制方式，在功率因数和无功补偿等领域有着广泛的应用。该文介绍了三相变流器的BP神经网络滞环电流内环控制，该方案可实现神经网络对快速变量的控制，提高滞环控制的性能，使系统对参数的变化有较强的不灵敏性和鲁棒性。该文分析了三相电源不平衡、某一路电流反馈丢失的工况下，系统的控制特性。为了使系统在轻负载下得到良好的频谱特性，采用实时变误差增益的控制策略，并讨论了容差下限，同时借助于矢量调制的思想，结合神经网络滞环调节器，优化系统性能，减小系统EMI和开关损耗。

A RESEARCH ON NEURAL NETWORK HYSTERESIS CONTROLLER OF THREE-PHASE CONVERTER

The field of neural networks is still developing rapidly in automata control, artificial intelligence, pattern evaluation and so on. The hysteresis current control method is widely used in power factor correction and reactive power compensation. This paper introduces a three-phase converter controlled with a back propagation neural network as a current inner controller. The controller has a high performance in response of current and robustness and insensitivity to the parameter changes. The converter can work well when one phase current feedback signal losing or input supply unbalanced. A hysetresis band changed in terms with load condition. And space vector pulse width modulation (SVPWM) technique is used to decrease the switch losses and EMI.