This paper reviews non-intrusive load monitoring (NILM) approaches that employ deep neural networks to disaggregate appliances from low frequency data, i.e., data with sampling rates lower than the AC base frequency. The overall purpose of this review is, firstly, to gain an overview on the state of the research up to November 2020, and secondly, to identify worthwhile open research topics. Accordingly, we first review the many degrees of freedom of these approaches, what has already been done in the literature, and compile the main characteristics of the reviewed publications in an extensive overview table. The second part of the paper discusses selected aspects of the literature and corresponding research gaps. In particular, we do a performance comparison with respect to reported mean absolute error (MAE) and F1-scores and observe different recurring elements in the best performing approaches, namely data sampling intervals below 10 s, a large field of view, the usage of generative adversarial network (GAN) losses, multi-task learning, and post-processing. Subsequently, multiple input features, multi-task learning, and related research gaps are discussed, the need for comparative studies is highlighted, and finally, missing elements for a successful deployment of NILM approaches based on deep neural networks are pointed out. We conclude the review with an outlook on possible future scenarios.

No abstract

In this paper right coprime factorization results are derived for i. general class of nonlinear plants and stabilizing feedback controllers Both input-output descriptions and state space realizations of th(plant and controller are used. It is first shown that if there exist stable right coprime factorizt ions for the plant and controller, and if a certain matrix of nonlinear operators has a stable inverse then the feedback system is well-posed and internally stable. The links between the right and left coprime factorization for a stable plant controller pair will be explored for this purpose. A generalization of the notion of linear fractional maps is explored as a means of characterizing the class of plants stabilized by this controller, and dually classes of controllers which stabilize the plant. It is then shown how to apply this theory to nonlinear plants which have a state space realization of a given form. It is also shown that if there exists a stabilizing state feedback for a plant in the class of int crest, then there exists a right cop rime factorization for the plant. Additionidly if there exists a stabilizing output injection, then there will exist a stabilizing controller with a right coprime factorization. An important sasumption in this work is to assume that the plant and controller have the same initial conditions, an approach is developed to allow for the stabfization of the plant by a controller with a different initial condition. A similar approach may also be used to stabilize a plant which has unmodeled dynamics. These results, of course, specialize to familiar linear system ones, and just as such linear systems results have had a wide application in robust and adaptive control system design, it is believed that the results developed here will facilitate the development of corresponding nonlinear robust and adaptive control system design.

No abstract

Abstract-In this paper a general approach is taken to yield a characterization of the class of stable plant controller pairs which is a generalization of the Youla parameterization for linear systems. This is based on the idea of representing the input-output pairs of the plant and controller as elements of the kernel of some related operator which is denoted the kernel representation of the system. It is demonstrated that in some sense the kernel representation is a generalization of the left coprime factorization of a general nonlinear system. Results giving one method of deriving a kernel representation for a nonlinear plant with a general state-space description are presented.