Computational aspects of sensitivity analysis

Summary form only given. This presentation deals with sensitivity analysis of power system transients considering two approaches. The aim of the analysis is to obtain the relationship between a response and one or several parameters, F=f(p). The aim is to obtain the ratio of a response to a cause, that is the derivative of a response with respect to system parameters, ≠F≠p (or the normalized sensitivity [p/F][≠F/≠p]). In the most general case, the response can be any combination of system variables. This presentation attempts to introduce the sensitivity analysis in power systems by providing a summary of the principles of both the adjoint network method and the transpose system method, showing how to extend the application of the latter method to obtain large change sensitivity, and including an introduction to time-domain sensitivities of linear and nonlinear systems. Some illustrative examples are used to justify the importance of this analysis