Harmonic Modeling of Voltage Source Converters using Basic Numerical Methods

14Parameter Estimation for Multiple VSCs with Domain Adaptation

“Seek success, but prepare for vegetables.”

– An AI‐driven bot

Chapter discusses how we can use simulation to generate a large number of training data for constructing an ANN model. We have shown that such a model still yields accurate parameter estimation (PE) even when the testing data fed to the model are the actual experimental measurement data. However, if the measurement data are contaminated with severe noises, we need to use domain adaptation (DA) techniques from deep learning (DL) to narrow the gap between simulation and measurement data. We believe that DA, which bridges the simulation and real‐world domains, will find its useful roles in many power system applications. Parameter estimation PE is just one of them!

This chapter is to extend the work of Chapter such that the following objectives can be achieved:

1. a wide range of parameters of a VSC can be estimated
2. the parameters of parallel‐connected multiple VSCs can be estimated
3. the estimation can still be accurate even if severe noise is present in the measurement data.

It is to be noted that an ANN model would have difficulty in achieving these three objectives. Thus, this chapter uses DL methods for achieving them. The organization of this chapter is as follows: Section 14.1 provides a brief introduction on DL. Section 14.2 gives a brief introduction on DA. Section 14.3 describes the tasks of parameter estimations of multiple VSCs. Section 14.4 ...