PhD FindDoctoral opportunities · PhDFind
Menu
中文

PhD in smart-building control | Safe and preference-aware model predictive control

Project 2753 · Tallinn University of Technology (TalTech) · FinEst Centre for Smart Cities and Nearly Zero Energy Buildings Research Group

Application workspace

Save this opportunity and prepare around its actual requirements. Confirm deadlines and eligibility at the original source.

Read the research report View the recruitment source
Country or region
Estonia
Application deadline
2026-12-05 (confirm the exact time with the source)
Funding
Four-year fully funded doctoral position as advertised; salary, contract category and benefits are not stated. SMARTWIN runs from 2027 to 2029, so confirm written funding and contract coverage for the full four years.

The research opportunity

Project overview

TalTech advertises a four-year fully funded PhD combining building-energy modelling, machine learning and control within SMARTWIN. The research develops transferable, physically consistent and uncertainty-aware building models using campus data and experimental facilities, to support energy prediction and safe model predictive control. The deadline is 5 December 2026 at 23:59 Europe/Tallinn.

Doctoral work and research environment

The researcher will develop hybrid physical/data-driven thermal and energy models; investigate grey-box models, physics-informed neural networks, structured Gaussian processes and neural state-space models; integrate energy, indoor-climate, weather, occupancy and building data; quantify uncertainty and reliability; and validate using campus datasets and facilities. The notice also lists reusable interfaces for MPC and explainable AI, open software, benchmark datasets, reproducible workflows, publications and conference presentations. It does not require implementing every model family or identify a fixed controller architecture.

The title refers to preference-aware control, but the vacancy does not explain whose preferences are involved, how they are represented, or how they enter the control method. The wider SMARTWIN programme mentions preference-based Bayesian optimization, but this does not establish an individual thesis task. Clarify the scope, available datasets and facilities and the work split from the related transfer-learning position.

The position involves TalTech's FinEst Centre for Smart Cities and Nearly Zero Energy Buildings Research Group. The notice names Avleen Kaur Malhi as principal supervisor and Tuule Mall Parts as co-supervisor and describes joint supervision with TU Darmstadt, possible visits there and EPFL collaboration. A saved TalTech directory entry notes that Parts's employment contract was suspended; confirm her current role. International collaboration does not by itself establish a joint degree, and the doctoral specialty and registration route remain unconfirmed.

Application and funding

The vacancy promises four years of full funding but does not specify salary, contract category, start date or benefits. The SMARTWIN grant runs from January 2027 to December 2029, shorter than the advertised doctoral period; this does not prove that year four is unfunded. Request written funding and contract coverage for all 48 months.

Applicants need a relevant master's degree, grounding in machine learning, modelling, data analysis or control, familiarity with time series, dynamical systems and/or building energy, programming/data-analysis skills and English at CEFR C1. Experience with physics-informed or transfer learning, relevant software frameworks, sensor/BMS data, HVAC and research publications is beneficial. A short essay describing motivation and research questions is required; confirm the complete application checklist.

Research and application brief

Report generated: . This date does not confirm that recruitment remains open.

Information updated: . Recheck the original notice before applying.

Show report contents

PhD in smart-building control | Safe and preference-aware model predictive control

Quick Decision

This is a four-year, fully funded PhD advertised at Tallinn University of Technology (TalTech), combining building-energy modelling, machine learning and control. The strongest match is a candidate who can connect physical/system models with real sensor data and evaluate whether forecasts are useful inside a controller. The vacancy names transferable, physically consistent, uncertainty-aware building models and safe energy control as the core, with several candidate model families and a substantial validation/software remit. It does not specify the individual salary, contract type, doctoral speciality or how preferences enter the control method. TalTech’s SMARTWIN grant record runs from 2027 to 2029, whereas the advertised PhD lasts four years; ask how the fourth year is funded. Applications close 5 December 2026 at 23:59 Tallinn time, and the vacancy requests a short motivation/research-interest essay. SMARTWIN project record

The position-specific terms below follow the full vacancy notice. The other sources provide project, university, policy, research and labour-market context.

Research Problem and Doctoral Work

What Is Unsolved and Why It Matters

The project targets a gap between building-energy prediction and safe control. A data-driven model may need extensive measurements from each building and can produce predictions that do not respect thermal or HVAC constraints. A model used by a controller also has to generalize to buildings with limited calibration data, make uncertainty visible, and remain dependable in operating conditions that differ from its training data. The vacancy says the work will use TalTech campus-building data and experimental facilities to build models for energy forecasting and learning-supported model predictive control (MPC).

This problem sits within a broader push to make buildings more energy efficient and responsive to changing grid and user needs. The recast EU Energy Performance of Buildings Directive makes building decarbonisation and smart-readiness part of the policy setting, but it does not specify this PhD’s algorithms or promise deployment. Energy Performance of Buildings Directive (2024/1275)

The TalTech campus dataset offers a concrete starting point: its public version contains heat and electricity time series for many campus buildings over 2022–2025. The published dataset describes meters and measured variables, but does not establish that this doctoral researcher will have access to every indoor-climate, occupancy, building-management-system or experimental-facility stream listed in the vacancy. TalTech campus energy dataset

Work, Methods, and Potential Outputs

Advertised work: develop hybrid physics-informed and data-driven building thermal and energy models; investigate grey-box models, physics-informed neural networks, structured Gaussian processes and neural state-space models; combine energy, indoor-climate, weather, occupancy and building metadata; assess prediction uncertainty and model reliability; validate with TalTech campus data and experimental facilities; create reusable interfaces suitable for MPC and explainable-AI methods; and produce open-source software, benchmark datasets and reproducible workflows. The vacancy also anticipates journal publications and conference presentations. These are the listed task families; it does not say that every model family will be implemented or that a particular controller architecture is already selected.

The wider Horizon Europe project SMARTWIN is more specific about its programme-level method transfer: physics-informed meta-learning, safe learning-supported MPC and preference-based Bayesian optimization, along with co-supervision and mobility involving TU Darmstadt and EPFL. CORDIS also says a smart-building-control laboratory is planned at TalTech. SMARTWIN is scheduled for 2027–2029, so these are project-wide plans rather than proof that every facility or partner activity is operational when the PhD begins. SMARTWIN project record

Preference-aware control needs clarification. The vacancy title uses this phrase, but its research questions and task list do not say who expresses a preference, what variables represent it, or how it would be combined with comfort, energy, grid and safety constraints. The CORDIS reference to preference-based Bayesian optimization shows that preference methods exist in the wider grant, but does not define the candidate’s role. Ask whether preference elicitation or preference-based optimization is an assigned doctoral task, and whether the focus is occupants, building operators, owners or energy-system stakeholders.

Recent work makes a useful methodological point: a forecast that looks accurate by a standalone metric may not be the best forecast for control. An experimental-building MPC study involving TalTech researcher Tuule Mall Kull/Parts found that multi-step-ahead prediction error was more informative for assessing models used inside MPC than common model indicators. This supports testing predictive models through their effect on controller performance, while leaving the new project’s safety criteria and benchmarks open. Predictive-model impact on experimental-building MPC

School, Programme and Field Context

TalTech is the advertised PhD host. The announcement places the work with the FinEst Centre for Smart Cities and the Nearly Zero Energy Buildings Research Group, and names a co-supervisor at TU Darmstadt. It also mentions possible research stays at TU Darmstadt, collaboration with EPFL and conference participation. These are supervision, mobility and collaboration arrangements; the announcement does not state that the partners jointly register or award the degree.

TalTech’s doctoral education page describes a four-academic-year route leading to the degree “Doctor of Philosophy” followed by the curriculum speciality. The programme includes research, coursework, teaching activities, external research mobility, a thesis and public defence. The position announcement does not identify the exact programme or speciality. Its host department and near-zero-energy-building focus make Engineering Sciences a relevant programme family to ask about. TalTech lists Building and Civil Engineering and Architecture and Chemical, Materials and Energy Technology among its specialities; the first looks contextually plausible from the named department but is not confirmed for this post. TalTech doctoral education and programmes, School of Engineering doctoral studies, Civil Engineering and Architecture study programmes

TalTech regulations make admission and matriculation formal steps, and its current admissions page distinguishes Early Stage Researcher, Industrial Doctoral Student, Doctoral Student and External Student routes. The captured vacancy does not identify which route applies. In particular, do not treat “fully funded PhD” as enough to infer the employment form, exact programme or a joint TalTech–Darmstadt degree. Confirm the registration sequence, programme director and degree wording before accepting. TalTech doctoral admission, TalTech Regulations for Doctoral Studies

The relevant institutional context is technically coherent. The Near-Zero Energy Buildings group works on high-performance buildings and control approaches; the FinEst Centre is an applied research and innovation centre within TalTech. QS ranks TalTech #=600 overall in 2027. This is institution-wide context, not a ranking of the School of Engineering, building-energy research, the group or the supervisors. No TalTech Engineering & Technology subject placement is claimed here. Nearly Zero Energy Buildings Research Group, FinEst Centre for Smart Cities, QS World University Rankings 2027: TalTech, QS Engineering & Technology subject page

Supervisor and Lab Assessment

Supervisor Research Directions

The vacancy names Adjunct Professor Avleen Kaur Malhi as main supervisor and Senior Researcher Tuule Mall Parts as co-supervisor. TalTech’s profiles place Malhi at the FinEst Centre and Parts in the Nearly Zero Energy Buildings group and Department of Civil Engineering and Architecture. Parts’ current profile notes a suspended employment contract; ask about her availability and supervisory role for this project. Avleen Kaur Malhi at TalTech, Tuule Mall Parts at TalTech

Recent Important Publications

  • Machine-learning applications for smart-building energy use (2024): Huotari, Malhi and Främling survey ML methods across energy management and control, grid interaction, maintenance and personalization. It provides a broad map for the project’s prediction/control scope, not a prescription for this thesis. Survey
  • Heating-control model fidelity (2023): Parts and colleagues empirically model a wax actuator and test increasing levels of controller-model detail for underfloor heating. Their results illustrate how physical details and controller settings can affect simulated energy use and temperature. The case concerns a particular heating system and conventional On/Off/PI control, not safe learning MPC. Paper
  • Battery storage and near-zero-energy requirements (2025): Parts and colleagues assess PV/battery contributions across 24 projects and nine building types under Estonian requirements. This adds local evidence about building heterogeneity and flexibility; it is not a predictive-control study. Paper
  • Portfolio-based energy-savings estimates (2025): Matson and colleagues, including Parts, use portfolio-based modelling to estimate building energy-saving potential. It is relevant to variation between buildings, but it does not address safe MPC or preference representation. Paper

Lab, Resources, and Collaboration Environment

TalTech publishes a multi-building heat/electricity dataset, and the position describes additional experimental facilities. The CORDIS grant record says a smart-building-control laboratory is planned as part of SMARTWIN. Because the grant starts in 2027, clarify what equipment and datasets will be available at the start, what will be built later, who manages access, and whether the planned lab is the same facility named in the vacancy. TalTech campus dataset, SMARTWIN project record

Funding, Application, and Living Cost

The captured vacancy describes a fully funded four-year PhD within SMARTWIN, but does not state a salary amount, employment contract type, weekly hours, start date or benefit package. The EU project record lists SMARTWIN’s period as January 2027 through December 2029, shorter than the advertised doctoral term. This does not show that year four is unfunded; it does mean the funding mechanism beyond the grant period should be documented in the offer.

TalTech’s current rules state that doctoral studies have no tuition. Applicants may, however, face a €100 one-time fee for evaluation of application documents and qualifications from 1 June 2026; the University lists exemptions for specified applicant groups. If the position is formally an Early Stage Researcher appointment, TalTech policy sets a full-time minimum gross salary of €2,300 per month and guarantees salary during the employment contract, up to 48 months of full-time work. The vacancy does not confirm this appointment form or quote its actual salary. For scale only, Statistics Estonia reports the Q2 2026 national average gross wage at €2,243/month; that is not a TalTech pay promise or a Tallinn engineering salary. TalTech doctoral admission and fee rules, Early-Stage Researcher funding procedure, Statistics Estonia wages

For housing, a July 2026 ERR report cites a TalTech double-room dorm bed at about €160–€170/month plus utilities and small nearby apartments around €400–€650, with newer flats higher. The University’s Tallinn page separately gives older private-rent and winter-utility estimates, while the national Study in Estonia portal gives broad student living-cost ranges. Use these as planning references, not an employee budget or accommodation guarantee; allow for food, utilities, insurance, deposits and individual circumstances. ERR Tallinn housing report, TalTech cost-of-living page, Study in Estonia living-cost FAQ

The vacancy’s application window runs from 4 November through 5 December 2026, 23:59 Europe/Tallinn. It specifically asks for a short essay explaining motivation and the research questions/directions of interest; applicants may build on the posted questions or suggest related ones. TalTech’s general process additionally requires a CV and qualification/identity documents, with formal verification and possible language-document requirements. Confirm the exact file list and fee exemption on the application page. TalTech doctoral admission

Applicant Fit and Preparation

Required Background and Skills

The vacancy requires a master’s degree in electrical engineering, control engineering, civil or energy engineering, computer science, applied mathematics or a related field; a strong base in machine learning, system modelling, data analytics or control; and familiarity with time series, dynamical systems and/or building-energy systems. It also asks for programming and data-analysis skills (Python, MATLAB or Julia are examples), English at CEFR C1 or above, strong analytical/mathematical problem-solving, independent work and teamwork, and willingness to join research visits, joint experiments, training and organizational work.

Preferred Experience and Suggested Preparation

Beneficial, not required: physics-informed or scientific ML, hybrid modelling, transfer/meta/few-shot learning, PyTorch/TensorFlow/JAX, sensor/weather/BMS data, system identification, building-energy modelling, HVAC/building physics, publications or open-source software.

For the required essay, a concise proposal could identify one or two questions that connect building physics, data transfer, model reliability and controller performance. A useful portfolio item would show how a model handles limited calibration or distribution shift and how the evaluation reflects a downstream engineering decision. If discussing preferences, present possible interpretations as questions rather than assuming occupant-comfort optimization is already assigned. These are preparation suggestions based on the announced tasks, not additional selection criteria.

Next Actions and Questions to Verify

Before Applying

  1. Use the short essay to explain a technically grounded research direction and the strongest relevant modelling/control evidence.
  2. Ask which TalTech doctoral programme and speciality are attached to the post, and which admission form (including Early Stage Researcher status) applies.
  3. Check the full document checklist and whether the €100 evaluation fee exemption applies.

At Interview

  1. What does “preference-aware” mean for this thesis: which stakeholders and preference variables are in scope, and how do preferences interact with safety constraints?
  2. Which campus, BMS, occupancy, indoor-climate and weather data are available, and what restrictions govern access and sharing?
  3. Which experimental facilities will be usable at the start, and when is the SMARTWIN lab expected to become operational?
  4. Which modelling approaches will be compared, how will uncertainty and safety be measured, and will validation be simulation-only, experimental or closed-loop in an occupied building?
  5. What supervision and mobility plan is confirmed with TU Darmstadt and EPFL?

Before Accepting an Offer

  1. Obtain the written salary, employment type, hours, start date, benefits and funding source for all four years, including the year after the current SMARTWIN grant period.
  2. Confirm the exact registration sequence, programme/speciality, degree title and whether any partner university has a formal degree-awarding role.
  3. Confirm housing availability, utilities and fallback costs before setting a monthly budget.

Evidence and Sources

The vacancy notice sets out the position-specific terms. The sources below provide project and broader research context.

Project, university, doctoral programme, supervisors and laboratory

Research publications and policy

Salary, Tallinn costs, and career context

Sources

The listing may come from a university or recruitment platform. Check the notice and official sources in the report for current application requirements. View the recruitment source

Explore opportunities with related research or the same country, prioritising similar funding where relevant. Recheck each notice before applying.

Tallinn University of Technology (TalTech) · Estonia

Power Systems Security PhD | Interpretable Cyberattack Detection

Full-time employment, minimum €2,300 gross per month and tuition-free doctoral study; contract duration, start date and other benefits not specified.

2026-12-04 (confirm the exact time with the source)

Tallinn University of Technology (TalTech) · Estonia

PhD in smart-building energy modelling | Physics-informed transfer learning

The notice promises four years of full funding but does not specify salary or employment category. SMARTWIN runs from January 2027 to December 2029; request written confirmation of contract and funding for all 48 months.

2026-12-05 (confirm the exact time with the source)