Data-Driven Load Modeling and Forecasting of Residential Appliances

Abstract

The expansion of residential demand response programs and increased deployment of controllable loads will require accurate appliance-level load modeling and forecasting. This paper proposes a conditional hidden semi-Markov model to describe the probabilistic nature of residential appliance demand, and an algorithm for short-term load forecasting. Model parameters are estimated directly from power consumption data using scalable statistical learning methods. Case studies performed using submetered 1-minute power consumption data from several types of appliances demonstrate the effectiveness of the model for load forecasting and anomaly detection.Index Terms-Hidden semi-Markov model, residential appliances, load model, short-term load forecast.

Scenario Forecasting of Residential Load Profiles

“…We can see from Fig. 5 that there is a change in the pattern of the realized 2 Since the residential load from single household is notoriously difficult to be forecasted accurately because of the large randomness, and the forecasting for an aggregation of 100 households is typically a easier task [4], we only use the forecasting results for 10 households as examples in the remaining part of Section IV to illustrate the performance of our approach. For the complete data and code, please refer to https://github.com/zhhhling/June2019.git.…”

“…Since these resources are stochastic and intermittent, accurate forecasting of residential load for a single or a small number of households becomes important for operators to decide on whether to integrate distributed energy resources and where to deploy energy storage so as to match customers' demand and make better use of energy [1]. In addition, accurate load forecasting on a small scale also allows customers to manage costs such as peak demand charges [2], [3].…”

Extraction of Dynamic Frequency Response Characteristics and Modelling of Modern Air Conditioners

“…The modelling of home appliances (e.g., air conditioner-AC) can be classified into two categories: steady-state modelling and dynamic-state modelling. The steady-state AC modelling has been investigated extensively in the literature [3][4][5][6][7][8], which mainly focus on modelling with normal operating conditions. For the AC modelling in the dynamic state, it is usually modelled as an induction motor [9][10][11].…”

Using Machine Learning and Deep Learning for Load disaggregation and Recognition of Activities in Household

“…In order to demonstrate the effectiveness of the proposed load forecast, the performance of the proposed ensemble learning(LSTM-RF) based load forecast was compared with the obtained results with other reference papers [6,12,53,54,55].…”

“…We first compared the performance of the proposed LSTM-RF with other forecasting methods [12,53] that were tested on the same AMPDs dataset. As shown in table 7.6 and figure 7.12.…”