Buildings and the built environment account for a significant portion of the global energy use and greenhouse gas emissions, and reducing the energy demand in this sector is crucial for a sustainable energy transition. This highlights the need for accurate and large-scale estimations and predictions of the future energy demand in buildings. Urban building energy modeling (UBEM) is an analytical tool for precise and high-quality energy modelling of city-scale building stocks, which is growing in interest as a useful tool for researchers and decision-makers worldwide. This thesis contributes to the understanding and future development in the field of UBEM and multi-variate cluster analysis. Based on a review of contemporary literature, possible improvements and knowledge gaps regarding UBEM are identified. The majority of UBEM studies are developed for similar applications, and some challenges are close to universal. Difficulties in data acquisition and the identification and characterisation of building archetypes are frequently addressed. Drawing on conclusions from the review, a clustering methodology for identifying building archetypes for hybrid UBEM was developed. The methodology utilised the k-means cluster analysis algorithm for multiple diverse parameters, including socio-economic indicators, and is based on open data sets which eliminates data acquisition issues and allows for easy adaptation. Building archetypes were successfully identified for two large data sets, and proved to be representative of the sample building stock. The results of the analysis also show that the error metric values diverge after a certain number of clusters, for multiple runs of the algorithm. This property of the algorithm in combination with the use of both existing and novel error metrics provide a reliable method for determining the optimal number of clusters. The methodology developed in this thesis enables for an improved modelling process, as a part of a complete UBEM.

<p>This thesis discusses novel developments in Bayesian analytics for high-dimensional multivariate time series. The focus is on the class of multiregression dynamic models (MDMs), which can be decomposed into sets of univariate models processed in parallel yet coupled for forecasting and decision making. Parallel processing greatly speeds up the computations and vastly expands the range of time series to which the analysis can be applied. </p><p>I begin by defining a new sparse representation of the dependence between the components of a multivariate time series. Using this representation, innovations involve sparse dynamic dependence networks, idiosyncrasies in time-varying auto-regressive lag structures, and flexibility of discounting methods for stochastic volatilities.</p><p>For exploration of the model space, I define a variant of the Shotgun Stochastic Search (SSS) algorithm. Under the parallelizable framework, this new SSS algorithm allows the stochastic search to move in each dimension simultaneously at each iteration, and thus it moves much faster to high probability regions of model space than does traditional SSS. </p><p>For the assessment of model uncertainty in MDMs, I propose an innovative method that converts model uncertainties from the multivariate context to the univariate context using Bayesian Model Averaging and power discounting techniques. I show that this approach can succeed in effectively capturing time-varying model uncertainties on various model parameters, while also identifying practically superior predictive and lucrative models in financial studies. </p><p>Finally I introduce common state coupled DLMs/MDMs (CSCDLMs/CSCMDMs), a new class of models for multivariate time series. These models are related to the established class of dynamic linear models, but include both common and series-specific state vectors and incorporate multivariate stochastic volatility. Bayesian analytics are developed including sequential updating, using a novel forward-filtering-backward-sampling scheme. Online and analytic learning of observation variances is achieved by an approximation method using variance discounting. This method results in faster computation for sequential step-ahead forecasting than MCMC, satisfying the requirement of speed for real-world applications. </p><p>A motivating example is the problem of short-term prediction of electricity demand in a "Smart Grid" scenario. Previous models do not enable either time-varying, correlated structure or online learning of the covariance structure of the state and observational evolution noise vectors. I address these issues by using a CSCMDM and applying a variance discounting method for learning correlation structure. Experimental results on a real data set, including comparisons with previous models, validate the effectiveness of the new framework.</p> / Dissertation

Energy demand forecasting has been more vital in recent years with countries setting goals to become climate neutral by 2050. Indeed, energy demand forecasting allows the understanding of drivers of the energy demand in all sectors of the economy. It also allows the planning of transformation of the future energy system. This study focuses on forecasting energy demand in Europe using a multi-country bottom-up modelling approach. The work explores ways of collecting large quantity of data to feed an energy model and method of completion for missing data series. It also aims at studying attributes that make a model user friendly and easy to use for the modelling of several countries. A model and a database are developed to answer these questions. A case application is conducted on the specific topic of the phase out of internal combustion engines in the EU to validate model dynamics and practical use. It is found that an energy demand forecasting model is easier and more time efficient to use with an included historical database. The case study shows that multi-country modelling can be relevant for policy assessment. Finally, improvements and future developments are proposed for the present work. / Prognoser för energiefterfrågan har blivit allt viktigare under de senaste åren i och med att länder har satt upp mål om att bli klimatneutrala senast 2050. Prognoser för energiefterfrågan gör det möjligt att förstå drivkrafterna bakom energiefterfrågan inom alla ekonomiska sektorer. Det gör det också möjligt att planera omvandlingen av det framtida energisystemet. Denna studie fokuserar på prognoser för energiefterfrågan i Europa med hjälp av en bottom-up-modelleringsmetod för flera länder. I arbetet undersöks olika sätt att samla in stora mängder data för att mata en energimodell och metoder för att komplettera saknade dataserier. Det syftar också till att studera attribut som gör en modell användar vänlig och lätt att använda för modellering av flera länder. En modell och en databas utvecklas för att besvara dessa frågor. För att validera modellens dynamik och praktiska användning genomförs en fallstudie om utfasningen av förbränningsmotorer i EU. Det visar sig att en modell för prognostisering av energiefterfrågan är enklare och mer tidseffektiv att använda med en inkluderad historisk databas. Fallstudien visar att modeller för flera länder kan vara relevanta för policybedömning. Slutligen föreslås förbättringar och framtida utveckling för det aktuella arbetet.

No description available.