All-integrator approach to linear SISO control system analysis using block pulse functions (BPF) (Q677511)

This paper makes use of the \(m\)-term block pulse function (BPF) domain operational matrix (with \(h\) as the duration of each component BPF) for the following differentiation \[ \begin{multlined}\text{D1}_{(m)}=\text{P1}^{-1}_{(m)}={1\over h}\left\lceil\begin{matrix} {1\over 2} & 1 &\cdots &\cdots & 1 & 1\\ &{1\over 2} & 1 &&& 1\\ &&\ddots &\ddots &&\vdots\\ &&&\ddots &\ddots &\vdots\\ & 0 &&&\ddots & 1\\ &&&&&{1\over 2}\end{matrix}\right\rceil^{-1}\\ ={1\over h}\left[\begin{matrix} 2 & -4 & 4 &\cdots &\cdots &(-1)^{m-1}4\\ & 2 &-4 & 4 &&\vdots\\ &&\ddots &\ddots &\ddots &\vdots\\ &&&\ddots &\ddots & 4\\ & 0 &&&\ddots &-4\\ &&&&& 2\end{matrix}\right]\end{multlined} \] to find out a block pulse operational transfer function (BPOTF) in the form of \[ |G(s)|_{s=\text{D1}_{(m)}}= G[\text{D1}_{(m)}]=\text{BPOTF}_{(m)}. \] Analysis of simple linear control systems using the BPOTF shows that the results are not accurate when compared with the direct expansion of the exact solution in the BPF domain. To remove this defect, one shot operational matrices for repeated integration introdudced by Rao et al. (1983), \[ \begin{multlined}\text{Pn}_{(m)}={1\over h}\begin{vmatrix} pn1 & pn2 &\cdots &\cdots & pnm\\ & pn1 & pn2 &&\vdots\\ &&\ddots &\ddots &\vdots\\ & 0 &&\ddots & pn2\\ &&&& pn1\end{vmatrix},\\ pnj=\begin{cases} 1,\quad & j=1\\ j- 2(j-1)^{n+1}+ (j-2)^{n+1},\quad &j\geq 2\end{cases}\end{multlined} \] are used to develop a modified BPOTF for linear SISO control system analysis in the BPF domain. A few typical linear SISO control systems are analyzed using the modified BPOTFs as illustrative examples. The results are found to match exactly with the direct BPF expansions of the exact solutions.