Page 227 - Modern Control of DC-Based Power Systems
P. 227

Control Approaches for Parallel Source Converter Systems     191


                 T should be high for good command tracking at low frequencies, but
              be low for high frequencies since we are not interested in tracking noise
              from the reference command. Ideally, we would prefer low values for S
              to eliminate the impact of disturbance. There is a conflict at higher fre-
              quencies that both S and T are required to be low. This is not possible
              due to the complementarity nature of S and T as evident from
              Eq. (5.224). To resolve this conflict, it is ideal to have S with high pass
              characteristics and T with low pass characteristics. The transfer function
              from noise source to the control input is known as the noise sensitivity
              transfer function RðjωÞ. R is required to have low pass characteristics,
              since the noise generally belongs to the higher side of the spectrum.
              Hence a low pass R would imply that the control input would not be
              sensitive to measurement noise. The expressions of the three key transfer
              functions are shown in (5.225).
                                                    21
                                     SjωðÞ 5 ½I1GKŠ
                                                  21
                                TjωðÞ 5 GK½I1GKŠ     5 I 2 S            (5.225)
                                                   21
                                  RjωðÞ 5 K½I1GKŠ    5 KS


              5.7.4.1 Generalized Plant Modeling
              In the weighted sensitivity approach, some crucial control signals of inter-
              est are augmented with dynamic weighting functions. In the previous sec-
              tion, the three sensitivity transfer functions were introduced. These
              sensitivity functions can be modified by choosing appropriate weighting
              functions. The explanation of how modifying each weighting function
              would modify its respective characteristic function is explained in the
              next section. In this section, the modeling of generalized plant or in other
              words the augmented plant is briefed. The weighting functions W e , W T ,
              and W u are frequency-dependent weightings of error signal v, plant out-
              put y, and the control input u. The augmented or generalized plant P
              consists of the actual plant G and the three weighting functions. It can be
              seen from Fig. 5.54, that the input for P are the reference command w
              and u and the outputs are z 1 , z 2 , z 3 , and v. By using the MIMO rule,
              one can write the transfer functions from each output to input. The
              transfer function of the augmented plant P(s) represents the mapping from
              the exogenous inputs (w,u) to the exogenous outputs (z 1 , z 2 , z 3 , v) and it
              is given by Eq. (5.226).
   222   223   224   225   226   227   228   229   230   231   232