Case Studies


Back Case Studies

Client: EDI challenge proposed by Virtual Power Solutions SA.
Domain: Energy & Environment
Length: 5+ months (ongoing)
Goal: HVAC usage management and control optimization
Tech: Python, Spark (Scala), Docker, Kafka​


More than 30,000 measurement points are monitored at the moment. Typical building installations include the monitoring of income and other circuits, such as HVAC (heating, ventilation, air conditioning) and lighting.

Approximately 15% of VPS measurement points are related to HVAC circuits with remote on/off control. These circuits represent a major part of total energy consumption and, for that reason, there is an opportunity to use data mining and advance control techniques to reduce waste without compromising comfort and thus realize direct savings. Additionally, the volume of electrical power under management offers load balancing opportunities in the VPP context.

As a challenge, VPS proposes the development of a model, using available historical consumption data and other attributes like location, to identify and predict consumption patterns with the following objectives: (1) HVAC consumption forecasts; (2) Anomalous consumption detection; (3) Manage shedding events.


SmartCat is developing Optimus Power, a solution for “HVAC usage management and optimal control”, using the dataset from VPS data provider as well as additional open datasets and state-of-the-art reinforcement learning approaches.


SmartCat has developed the MVP of Optimus Power, a solution to “HVAC usage management and optimal control” EDI challenge. We have used dataset from VPS data provider as well as historical and forecasted weather data from OpenWeatherMap API. Solution architecture is shown in figure below. Time series of measurements from 100+ physical objects is collected from VPS API, for the time period of 3 years (~30M data points). For some branches we are continuously collecting real-time data. Ingestion component is robust to missing data and deployed and run in scalable manner. Parsed data is passed to Apache Kafka for downstream consumers - processing components. Load prediction and anomaly detection are implemented in distributed manner by Apache Spark, while optimization is implemented as dedicated python component.

Optimus Power solves all three problem areas specified by data provider: a) it infers base load profiles and does load prediction, b) detects multiple types of anomalies and c) does optimal control, ie. prepares policies for savings by shedding. Optimus Power gives reliable predictions, is adaptive to changes in users behaviour or simple change of seasons, is robust and scalable (thanks to chosen tech-stack and algorithms) and saves money and energy (up to 35% compared to baselines)!. This way, Optimus Power plays the descriptive, predictive and, with optimal control, also a very important prescriptive analytics role in energy management.


Working solution is best presented with the promotional video below. Stay tuned for further news about development of Optimus Power.

Next post