Energy consumption forecasting for buildings has immense value in energy efficiency and sustainability research. Accurate energy forecasting models have numerous implications in planning and energy optimization of buildings and campuses. For new buildings, where past recorded data is unavailable, computer simulation methods are used for energy analysis and forecasting future scenarios. However, for existing buildings with historically recorded time series energy data, statistical and machine learning techniques have proved to be more accurate and quick. This study presents a comprehensive review of the existing machine learning techniques for forecasting time series energy consumption. Although the emphasis is given to a single time series data analysis, the review is not just limited to it since energy data is often co-analyzed with other time series variables like outdoor weather and indoor environmental conditions. The nine most popular forecasting techniques that are based on the machine learning platform are analyzed. An in-depth review and analysis of the ‘hybrid model’, that combines two or more forecasting techniques is also presented. The various combinations of the hybrid model are found to be the most effective in time series energy forecasting for building.

A review on time series forecasting techniques for building energy consumption

Practical applications inherently exhibit high levels of linguistic and numerical uncertainty. Conventional type-1 fuzzy sets have been employed in many real-world applications, but type-1 fuzzy sets use precise and crisp membership functions, which are unable to accommodate high levels of uncertainty. For example, the failure rate and the repair rate for any component of a system are uncertain; therefore, a membership representation using precise and crisp functions seems unreasonable from a pragmatic perspective. Type-2 fuzzy set theory represents a paradigm shift in which uncertainties related to the membership representation are incorporated through secondary membership functions. Whereas attempts have been made by researchers to accommodate the additional uncertainties related to linguistic and numerical data using type-1 fuzzy set theory, this paper investigates type-2 fuzzy set theory to address these uncertainties in an efficient way. The results of this study demonstrate the superiority of the proposed approach for analyzing the reliability of phasor measurement units, which are a technology so recent that the lack of sufficient field data creates a milieu of uncertainty.

Reliability Evaluation of Phasor Measurement Unit Using Type-2 Fuzzy Set Theory

Multilayer perceptron neural network is one of the widely used method for load forecasting. There are hyperparameters which can be used to determine the network structure and used to train the multilayer perceptron neural network model. This paper aims to propose a framework for grid search model based on the walk-forward validation methodology. The training process will specify the optimal models which satisfy requirement for minimum of accuracy scores of root mean square error, mean absolute percentage error and mean absolute error. The testing process will evaluate the optimal models along with the other ones. The results indicated that the optimal models have accuracy scores near the minimum values. The US airline passenger and Ho Chi Minh city load demand data were used to verify the accuracy and reliability of the grid search framework.

/pdf/grid-search-of-multilayer-perceptron-based-on-the-walk-1wjn87z1bw.pdf

Grid search of multilayer perceptron based on the walk-forward validation methodology

Type-2 (T2) fuzzy set was introduced to model vagueness associated with primary membership function of type-1 (T1) fuzzy set. While it was invented to handle more fuzzy information, there are only a few algorithms (models) to deal with data in the form of T2 fuzzy variables given their three-dimensional features. To solve the problem, we define the expected value of a T2 fuzzy variable using credibility theory in this paper. And by substituting the expected value for the original T2 fuzzy set, the vertical uncertainties of data are transferred to horizontal ones without much distortion of information. Calculations between three-dimensional T2 fuzzy sets are thus transferred to two-dimensional range calculations between T1 fuzzy sets. Based on that principle, we also build a T2 fuzzy expected regression model and apply it to the arbitrage pricing theory. © 2016 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.

Building a type‐2 fuzzy regression model based on credibility theory and its application on arbitrage pricing theory

These days, fuzzy logic system has widely been applied in many applications, including wave forecasting. In current fuzzy logic development, researchers have started to use interval type 2 fuzzy as a solution to solve the uncertainties problem. The aim of this research is to propose a method to forecast significant wave height by using interval type 2 fuzzy logic. As part of the research, this paper focuses on two objectives; to extend type-1 fuzzy to type-2 fuzzy by assigning uncertainty to its type 1 counterpart and to predict significant wave height for 1 hour, 3 hours, 6 hours, 12 hours and 24 hours lead time. The root mean square error (RMSE) has been used to score the performance of both type 1 and type 2 fuzzy methods. The best result has shown at 6 hours lead time with RMSE of 0.1529.

Interval type 2 fuzzy logic for significant wave height forecasting

This paper describes about electrical load time series data forecasting using interval type-2 fuzzy logic system. This interval type-2 fuzzy logic is used as the method to forecast electrical load in East Java-Bali area from January until March 2007. The training data used in this research are electric load data from September 2005 until December 2006. The structure of Interval Type-2 Fuzzy Logic used in this research, has three fuzzy sets per input with uncertain mean Gaussian membership function. The number of input varies from two until five inputs in order to predict single output value. Steepest descent training algorithm is used to train interval type-2 fuzzy logic system. The training process was done by adjusting the used parameter so that this system can produce an output with minimum error. Experimental result showed that the interval type-2 fuzzy logic system could forecast East Java-Bali's electrical load data well with the best root mean square error value of 0.082691. This was resulted from the experiment using 2 inputs, variance value of 0.2, and learning rate of 0.4.

http://fportfolio.petra.ac.id/user_files/97-031/Paper%20ICCSIT%202010.pdf

Yongky Kurniawan Thiang

Papers

Electrical load time series data forecasting using interval type-2 fuzzy logic system