Detecting nonlinear modules in a dynamic network: a step-by-step procedure^⁎

Adopting a dynamic network viewpoint allows one to analyze and identify subsystems of a complex interconnected system. When studying a network of dynamic systems, it is important to know if significant nonlinear behavior is present in a dynamic network under study and where the nonlinearity is located in the network. This work extends the Best Linear Approximation framework from the closed-loop to the networked setting. The framework is illustrated using a practical step-by-step estimation and analysis procedure. It is shown how nonlinear behavior can be quantified and located in a dynamic network using this framework.