Data Mining and Machine Learning in Building Energy Analysis

4Artificial Intelligence for Building Energy Analysis

4.1. Introduction

In this chapter, we apply artificial intelligence models in building energy analysis. There are two main applications, one is predicting energy consumption and the other is to perform faulty consumption detection and diagnosis. Both of them are crucial for energy conservation in building design, retrofit and operation as described in Chapter 1.

In the first application, we will investigate the performance of the SVR model in the prediction of energy consumption in the unknown future based on the historical behavior. Then, we try to extract models from multiple buildings’ performance and carry out the prediction of the consumption for a new building. Two types of energy, electricity consumption and heating demand, will be used as the targets in the experiments. Furthermore, we will profoundly test the robustness of this model by considering various situations and try to find out in which circumstance the better performance is achieved. For this purpose, we design three sets of experiments that differ in the dataset selection, then analyze the trend of the model performance.

In the second application, we introduce an effective ANN model, RDP neural network, to implement fault detection and diagnosis (FDD) of building energy consumption. Based on the knowledge from previous faulty consumption, this model is able to report faults automatically and with a high accuracy. It also shows high performance in a newly ...