تخمين زواياي بهينه كليدزني در اينورترهاي چند سطحي توسط شبكه هاي عصبي براي بازه ي مدولاسيون وسيع

OPTIMUM SWITCHING ANGLES ESTIMATION FOR MULTI-LEVEL INVERTERS USING NEURAL NETWORKS IN A WIDE RANGE OF MODULATION INDEX

Usually the multi-level inverters are controlled by using only one switching scheme such as staircase waveform. By using one switching scheme, the modulation index can be controlled only in a limited range. It is possible to solve this problem by changing the switching schemes for different values of modulation index. In this way, the modulation index can be controlled in a wide range from zero to over-modulation values. In this paper, a neural network is trained to estimate the switching angles of each switching scheme. In this case, the number of neural networks will be equal to the number of switching schemes. For example, a seven levels multi-level inverter is used in this paper to prove that by using three different switching schemes and thereby three different neural networks, it is possible to control the modulation index from zero to over-modulation values, continuously. The proposed neural network structure can solve the custom difficulties of practical utilization of look-up tables such as large size of memory, complex digital circuits and controlling the magnitude of output voltage in a discrete manner.