Building Heat Demand Forecasting by Training a Common Machine Learning Model with Physics-Based Simulator

Abstract

Accurate short-term forecasts of building energy consumption are necessary for profitable demand response. Short-term forecasting methods can be roughly classified into physics-based modelling and data-based modelling. Both of these approaches have their advantages and disadvantages and it would be therefore ideal to combine them. This paper proposes a novel approach that allows us to combine the best parts of physics-based modelling and machine learning while avoiding many of their drawbacks. A key idea in the approach is to provide a variety of building parameters as input for an Artificial Neural Network (ANN) and train the model with data from a large group of simulated buildings. The hypothesis is that this forces the ANN model to learn the underlying simulation model-based physics, and thus enables the ANN model to be used in place of the simulator. The advantages of this type of model is the combination of robustness and accuracy from a high-detail physics-based model with the inference speed, ease of deployment, and support for gradient based optimization provided by the ANN model. To evaluate the approach, an ANN model was developed and trained with simulated data from 900–11,700 buildings, including equal distribution of office buildings, apartment buildings, and detached houses. The performance of the ANN model was evaluated with a test set consisting of 60 buildings (20 buildings for each category). The normalized root mean square errors (NRMSE) were on average 0.050, 0.026, 0.052 for apartment buildings, office buildings, and detached houses, respectively. The results show that the model was able to approximate the simulator with good accuracy also outside of the training data distribution and generalize to new buildings in new geographical locations without any building specific heat demand data.

Accelerating Energy-Economic Simulation Models via Machine Learning-Based Emulation and Time Series Aggregation

“…The prediction of heat demands in buildings is simplified in [15] by emulating physical models. The goal was to combine robustness and accuracy in the case of detailed calculations with high speed and simple development.…”

“…Reviewed papers make use of this approach, e.g., in chemistry and medicine [4,18,22,25], the automotive industry [23,24], geoscience [4,21], astrophysics [4], fluid dynamics [26], or various engineering challenges [16,18]. Use cases in the energy sector include fusion simulation [14], the heat demand of buildings [15], an urban energy simulator [17], vehicle energy consumption [28], and smart grids [18][19][20]. A simulation and emulation of large quantities of P2P communities, as performed in Section 6, has not been performed so far.…”

“…The goals of the reviewed papers include the reduction of computational complexity and speed increases by machine learning [14][15][16][17][18][19]21,26] as well as the calculation of inbetween states that are hard or impossible to calculate by the simulation model [14,[23][24][25][26]. The advantages in saving time through the reapplication of the trained ML models are especially emphasized but vary considerably.…”

“…Other papers, e.g., [16,18] still rely on simple random sampling to choose the appropriate training data if only a limited number of simulations are feasible. Most papers have sufficient training data available [4,14,15,17,21,[24][25][26] and hence the effects of sampling methods are diminished [30].…”

Analysis of an Economic Coupling Relationship Model of the Coastal Ecological Fragile Zone Based on a Machine Learning Model

“…Extreme Random Tree. Extra tree does not use the best points when selecting branch points but uses the updated random weights to pick out a feature value to divide the tree classifier [19][20]. Let the base classifier focus on training classification, so the whole algorithm is very random, as shown in Figure 2.…”