【摘要】 最优 PWM脉宽调制是逆变器控制的一种常用方案 ,但是由于最优 PWM开关角计算的复杂性 ,一般都是离线计算最优开关角 ,用开关点预置的方法进行控制 ,十分不利于在线的电压调节。文中介绍了一种最优 PWM开关角的在线计算方法 ,即利用神经网络的非线性逼近能力 ,对离线求解的最优 PWM开关角样本点进行辨识拟合 ,从而得到最优 PWM开关角对基波电压传输比的函数 ,用于实时计算 ,为利用最优 PWM控制逆变器进行在线调压做好了准备。另外 ,还提出了一种简便的三相最优 PWM开关角排序方法 ,解决了最优 PWM开关角的在线排序问题。本算法在 DSP实验平台上进行了实验验证 ,证明了算法的可行性。更多还原

【Abstract】 Optimal PWM is one of the most popular methods for inv er ter control, but the online calculation of optimal PWM angles is always a proble m. A strategy based on neutral network is used to solve the problem. Because the neutral network has the ability to match nonlinear functions, eight neutral net works are trained to get eight functions between optimal PWM angles and voltage transfer ratio. Once the voltage transfer ratio is determined, eight optimal PWM angles can be got by calculating these eight functions. All calculations of the se eight angles need 0 287 ms, which basically satisfy the requirement of the o nline calculation. Besides a simple method to put the optimal PWM angles in orde r is also presented. Experiments based on TMS320C32 DSP chip prove the correctio n of the strategy.更多还原