Optimal filtering of piecewise deterministic processes for source detection and separation in electric load monitoring

Abstract

We wish to compute in discrete real time the best estimate for the composition of the total active power recorded at a customer's house, without any intrusion. A new solution for online characterization of the electric load is proposed. The solution is based on issues from the applied stochastic analysis of Markov processes. We specifically suggest that optimal filtering of a hybrid stochastic differential system covers entirely the detection and separation subproblems underlying the main problem.

Keywords:

• electric load monitoring
• hybrid stochastic differential systems
• piecewise deterministic processes
• Fokker–Planck equation
• non-linear filtering