Fill-level prediction in online valley-filling algorithms for electric vehicle charging

Abstract

Due to the large increase in electric vehicles (EVs), smart charging strategies are required in order for the distribution grid to accommodate all these EVs. Many charging strategies either assume that future loads are known in advance, or use predictions of these loads as input. However, accurate prediction of uncontrollable load is very difficult. Online valleyfilling algorithms circumvent this problem by determining the charging profile based on a prediction of the fill-level: a single parameter that characterizes the optimal EV schedule. This paper presents a simple, but accurate, method to predict this fill-level. We show that near-optimal charging profiles with an optimality gap of less than 1% can be realized when our method is used to predict the input level for the online valley-filling approach. Furthermore, our method is very fast and thus suitable for use in decentralized energy management systems that employ the online valley-filling approach.

Offline and online scheduling of electric vehicle charging with a minimum charging threshold

“…We generate predictions of Z act and Z fill by computing the optimal EV schedule for 70 previous days under the same circumstances as for the current charging session, i.e., the same values of C, X min and X max , using the house power profile of each of the days respectively. In line with the findings in [21], we expect that the resulting predictions properly represent the behavior of Z act and Z fill . For each previous day, we compute an online solution using Algorithm 2 and the predictions corresponding to that day as input.…”

ODDO: Online Duality-Driven Optimization

“…However, this historical weather has little to no influence on the current household consumption and thus also little to no influence on the optimal Lagrange multipliers corresponding to the current day. In earlier work [49], we observed a similar relation between the optimal Lagrange multiplier and the choice of number of training instances.…”

“…Predicting values such that they are preferably beneath a (given) threshold value corresponds to the concept of quantile functions in statistics [53]. In earlier work on scheduling the charging of EVs [49], we used this concept to successfully predict optimal Lagrange multipliers (also called "fill-levels" in this application). Thus, it is worthwhile to investigate the possibility of generalizing this approach to the general Problem P.…”

Duality-driven optimization in energy management: offline and online algorithms for resource allocation problems

“…We note that all analyses, algorithms, and online optimization approaches in this section can be generalized to the case where the maximum charging rates X max are stage-dependent, i.e., we have X min y t ≤ x t ≤ X t max y t for all t ∈ T , provided that X t max ≥ 2X min (see [SU:8]). Moreover, they can also be generalized to the case where the objective function is (0, p, f )-separable for some convex function f , i.e., where we replace each term 1 2 (x t + p t ) 2 of the objective function by f (x t + p t ) for a given convex function f (see also Section 2.2.2).…”