AI4PublicPolicy / Blog | AI4PublicPolicy Pilots: Lisbon, Portugal

Blog | AI4PublicPolicy Pilots: Lisbon, Portugal

Location: Lisbon, Portugal

Pilot Leader(s)Lisboa E-Nova – Agência Municipal de Energia e Ambiente

Theme – Policies Involved: Energy Management and Optimization Policies

The pilot focuses on gathering relevant and valid data sources, such as weather, buildings’ characteristics and energy consumption, and with the aid of AI and machine learning algorithms detect patterns and problems regarding energy efficiency, so that possible data-driven policies can be defined to ensure a more sustainable and efficient environment in the city. Data regarding energy consumption will be gathered using smart meters and IoT devices. Apart from the current energy consumption time interval of 15 minutes, with the implementation of IoT devices, a deeper analysis will be made, to analyse specific types of energy consumption in each building, assuming a breakdown of consumption considering the equipment existing in each delivery point. Data regarding buildings archetypes is gathered with the support from other projects of LIS. With these main data sources, AI models can be implemented to analyse patterns and support policy making on general energy performance improvement. Deep Learning models such as long/short-term memory network (LSTM Network), Convolutional Neural Networks (CNNs) and Deep Belief Networks (DBNs) can also be used to provide forecast on energy consumption. Clustering algorithms, e.g., k-Means, Hierarchical Cluster Analysis and Expectation Maximization can be used to group clients on similar energy usage. Other deep learning models, such as deep neural networks, can be used to detect patterns on the behaviour of different buildings. With the definition of different patterns in buildings, and with the integration of the different information layers, it is possible to create and analyse common behavioural scenarios on each archetype of buildings to better understand real energy performance. This information can be used for data-driven decisions on policies for energy efficiency.

Two main use cases of policy development will be considered:

  1. Energy performance analysis: This use case will identify problems in the buildings’ energy performance and will perform root cause analysis;
  2. Budget planning for energy usage and buildings renovation: This use cases will focus on policy making decisions on budget planning and effective use of budget and public resources towards buildings renovation and energy usage. For instance, policies for the definition of specific financial supporting mechanisms or retrofit schemes will be recommended.

The following KPIs will be tracked:

  • Number of datasets to be gathered and integrated;
  • Number of policy instruments to be extracted/recommended;
  • Level of Compliance to SECAP.