改进无差拍控制在单相并网逆变器中的应用  

Applicaton of Improved Deadbeat Control to Single-Phase Grid-Connect Inverter

作　　者：吴海涛[1] 陈晓[1] 孙以泽[1] 孟婥[1] 徐洋[1] 

机构地区：[1]东华大学机械工程学院,上海201620

出　　处：《控制工程》2012年第3期531-534,共4页Control Engineering of China

基　　金：东华大学博士论文创新基金资助项目(9D10312)

摘　　要：为解决在单相并网逆变器中应用无差拍控制时,由于采样及处理器计算的延时而导致系统不稳定的问题,提出了一种基于CMAC神经网络的无差拍控制策略。控制策略通过MPPT算法跟踪光伏电池的最大输出功率,然后将最大功率减去电路中功率损失后除以电网电压的有效值平方得到逆变器参考电流的系数k,再根据CMAC神经网络预测出下一时刻的电网电压和参考电流,最后计算出下一时刻的开关周期的占空比对逆变器进行控制。此方法实现了对采样及处理器计算的延时的补偿,可提高了系统的稳定性,使并网逆变器的输出电流快速、准确地跟踪电网电压,降低逆变器输出电流中的总谐波歧变。通过对该策略进行了仿真和试验,结果表明该控制策略有效性和可行性。To solve the instability problems caused by the sampling and calculation of the delay when single-phase grid-connect inverter is controlled by deadbeat control strategy, a deadbeat control strategy based on a cerebellar-model-articulation-eontroller （CMAC） neu- ral network is proposed. The control strategy tracks the maximum output power of solar cells, then subtracts the power losses in tire circuit and divided by the square of the grid voltage of the RMS and get the coefficients of the inverter reference current. According to CMAC neural network to predict the grid voltage and reference current of next time, and finally get the switch duty cycle of next time. the delay caused by the sampling and calculation is compensated and it improve the stable of system, moreover, il make the output current of inverter rapidly and accurately traek the grid voltage and reduce the total Harmonic distortion of output current. The effectiveness and feasibility of the proposed control strategy are proved by the simulation and experiment.

关 键 词：光伏并网逆变器 无差拍控制 CMAC神经网络 延时补偿 

分 类 号：TP27[自动化与计算机技术—检测技术与自动化装置]