Search

The ability to forecast the concentration of air pollutants in an urban region is crucial for decision-makers wishing to reduce the impact of pollution on public health through active measures (e.g. temporary traffic closures). In this study, we present a machine learning approach applied to forecasts of the day-ahead maximum value of ozone concentration for several geographical locations in southern Switzerland. Due to the low density of measurement stations and to the complex orography of the use-case terrain, we adopted feature selection methods instead of explicitly restricting relevant features to a neighborhood of the prediction sites, as common in spatio-temporal forecasting methods. We then used Shapley values to assess the explainability of the learned models in terms of feature importance and feature interactions in relation to ozone predictions. Our analysis suggests that the trained models effectively learned explanatory cross-dependencies among atmospheric variables. Finally, we show how weighting observations helps to increase the accuracy of the forecasts for specific ranges of ozone’s daily peak values.

The precise calculation of solar irradiance is pivotal for forecasting the electric power generated by PV plants. However, on-ground measurements are expensive and are generally not performed for small and medium-sized PV plants. Satellite-based services represent a valid alternative to on-site measurements, but their space-time resolution is limited. In this paper we present a method for estimating the global horizontal irradiance (GHI) from the power measurements of one or more photovoltaic (PV) systems located in the same neighborhood. The method is completely unsupervised and is based on a physical model of a PV plant. It can estimate the nominal power and orientation of multiple PV fields, using only the aggregated power signal from their PV power plant. Moreover, if more than one PV power plant is available, the different signals are reconciled using outliers detection and assessing shading patterns for each PV plant. Results from two case studies located in Switzerland are presented here. The performance of the proposed method at estimating GHI is compared with that of free and commercial satellite services. Our results show that the method presented here is generally better than satellite-based services, especially at high temporal resolutions.

In this paper, we report the results of simulations of a demand side management strategy, developed within the context of the GridSense project. GridSense is a multi-objective energy management system that aims at decreasing both the end user energy costs and the congestions on the local feeder. The algorithm controls single appliances, among which boilers, heat pumps, EV chargers and stationary batteries. The control strategy is completely decentralized and does not require communication outside the building. A multiphysics simulation framework, which allows simulating the electrical grid, buildings and appliances, has been implemented using the Modelica language, while the control algorithms are coded in C. The performance of the proposed solution against the two main optimization objectives (cost and grid friendliness) has been tested in simulation under different scenarios. We systematically varied the capacity and E-rate of the energy storage. Our analysis shows that savings are mostly dependent on the battery capacity, while being only slightly affected by the battery inverter-charger size. The break even system price of the different solutions was calculated for an example house in a typical Swiss tariff scheme.

We present a comparative study of different probabilistic forecasting techniques on the task of predicting the electrical load of secondary substations and cabinets located in a low voltage distribution grid, as well as their aggregated power profile. The methods are evaluated using standard KPIs for deterministic and probabilistic forecasts. We also compare the ability of different hierarchical techniques in improving the bottom level forecasters’ performances. Both the raw and cleaned datasets, including meteorological data, are made publicly available to provide a standard benchmark for evaluating forecasting algorithms for demand-side management applications.

Demand side management is a promising approach towards the integration of renewable energy sources in the electric grid, which does not require massive infrastructural investments. In this paper, we report the analysis of the performance of a demand side management algorithm for the control of electric boilers, developed within the context of the GridSense project. GridSense is a multi-objective energy management system that aims at decreasing both the end user energy costs and the congestions on the local feeder. The latter objective is minimized exploiting the existent correlation between the voltage measured at the connection point to the grid and the power flow measured at the low voltage transformer. The algorithm behavior has been firstly investigated by means of simulation, using typical water consumption profiles and a simplified thermodynamic model of the electric boiler. The simulation results show that the algorithm can effectively shift the boiler's electric consumption based on voltage and price profiles. In the second phase, we analyzed the results from a pilot project, in which the GridSense units were controlling the boilers of four households, located in the same low voltage grid.

BIPV technologies applied to façades are strongly affected by complex and dynamic shadings especially when located in dense urban environments. In this case, the shading effects need to be evaluated in detail in order to properly estimate the energy yield and optimise the energy harvest of such PV systems.The S.M.O (String Matching Optimisation) solution represents a new approach, entirely based on open-source software, which allows a very accurate study of the shading effects on façades and the investigation of electrical behaviour of BIPV strings under complex irradiation patterns. This solution has been validated with a real pilot BIPV façade located in southern Switzerland and the model was simulated on other theoretical PV façades with different levels of complexity. The presented method allows an accurate evaluation and optimisation of the energy yield of BIPV facades in complex urban environments.

In this paper, we propose a distributed control strategy for the design of an energy market. The method relies on a hierarchical structure of aggregators for the coordination of prosumers (agents which can produce and consume energy). The hierarchy reflects the voltage level separations of the electrical grid and allows aggregating prosumers in pools, while taking into account the grid operational constraints. To reach optimal coordination, the prosumers communicate their forecasted power profile to the upper level of the hierarchy. Each time the information crosses upwards a level of the hierarchy, it is first aggregated, both to strongly reduce the data flow and to preserve the privacy. In the first part of the paper, the decomposition algorithm, which is based on the alternating direction method of multipliers (ADMM), is presented. In the second part, we explore how the proposed algorithm scales with increasing number of prosumers and hierarchical levels, through extensive simulations based on randomly generated scenarios.

We propose a method to design a decentralized energy market which guarantees individual rationality (IR) in expectation, in the presence of system-level grid constraints. We formulate the market as a welfare maximization problem subject to IR constraints, and we make use of Lagrangian duality to model the problem as a n-person non-cooperative game with a unique generalized Nash equilibrium (GNE). We provide a distributed algorithm which converges to the GNE. The convergence and properties of the algorithm are investigated by means of numerical simulations.

We consider the problem of estimating the (unobserved) amounts of PV generation and demand in a power distribution network starting from measurements of the aggregated power flow at the point of common coupling (PCC) by leveraging local global horizontal irradiance (GHI) measurements. The estimation principle consists in modeling the PV generation as a function of the measured GHI, enabling to identify PV production patterns in the aggregated power flow measurements. Four estimation algorithms are proposed: the first assumes that variability in the aggregated PV generation is mostly given by variations of PV generation, two use a model of the demand to improve estimation performance, and the fourth assumes that, in a certain frequency range, the aggregated power flow is dominated by PV generation dynamics. Algorithms take advantage of irradiance transposition models to explore several azimuth/tilt configurations and explain PV generation patterns from the aggregated power flow profile accounting for non-uniform configurations of the plants. Algorithms estimation performance is compared and validated by using experimental measurements from a real-life setup with 4 dwellings with rooftop PV installations and battery systems for self-consumption.

The incremental deployment of small scale stochastic generators has a significant impact low voltage grids. We investigated the applicability of local voltage measurements at household sockets as predictors of the power at the low voltage branch of the transformer. The general goal is to evaluate the feasibility of a decentralized demand-side control algorithm using local voltage as the regulation input. In this paper we introduce the approach adopted by our study and describe the experimental results, which demonstrate the possibility of using a local voltage measurement as an input signal for decentralized control.

The environmental impacts of buildings can significantly vary with the dynamics of their energy demand and production. Significant variations have been modelled for buildings in the U.S., France, Denmark and Switzerland but the levels of variation are different between these countries. This difference can be explained by factors like the existing energy sources, the availability of renewable energy and the importation of electricity from nearby countries. With its high share of renewable energy and significant electricity exchanges with neighbouring countries, Switzerland presents a specific case where benchmark values from dynamic life cycle assessment should be well understood. The project’s goal is to provide results from a dynamic life cycle assessment with a detailed study of the influence from temporal fluctuations in the national electricity production, electricity imports, decentralised generation and electricity demand from buildings. Additionally, consequences of changing the temporal precision (i.e. hourly, daily, monthly and annual) of energy dynamics are analysed. This assessment is conducted with demand and production estimations for the design of a residential building in Switzerland. Disparities of results are assessed for all temporal precision levels with a comparison to the values that are obtained with the current national methodology which operates with values based on average annual electricity production. Results thus suggest some methodological recommendations to develop the temporal aspects of the environmental impact assessment methodology for the Swiss building sector.

The growth of energy demand and decentralised renewable energy generation (e.g. photovoltaic, eolic) can lead to electric grid imbalances requiring extra investments in the electric grid infrastructure. One of the goals of Smart-Home and Smart-Grid solutions is to solve this issue. The majority of the solutions are focused on centralised load management. Furthermore most of the Smart-Home and Smart-Grid publications analyse the topics from the application point of view so that there is a lack of specific descriptions of the infrastructure and of the used technology. SUPSI has proposed and verified a decentralised and innovative Smart-Grid approach. To demonstrate the feasibility of this approach a custom HAC (Household Appliance Controller) and a communication infrastructure has been developed. This paper describes this design, and its use in several pilot projects performed in Switzerland (Swiss2Grid, RiParTi 2.0, HCD 2.0) which in their turn have demonstrate the feasibility and the benefits of a decentralised Smart-Grid management.

Experimental evolution is a powerful tool for investigating complex traits. Artificial selection can be applied for a specific trait and the resulting phenotypically divergent populations pool-sequenced to identify alleles that occur at substantially different frequencies in the extreme populations. To maximize the proportion of loci that are causal to the phenotype among all enriched loci, population size and number of replicates need to be high. These requirements have, in fact, limited evolution studies in higher organisms, where the time investment required for phenotyping is often prohibitive for large-scale studies. Animal size is a highly multigenic trait that remains poorly understood, and an experimental evolution approach may thus aid in gaining new insights into the genetic basis of this trait. To this end, we developed the FlyCatwalk, a fully automated, high-throughput system to sort live fruit flies (Drosophila melanogaster) based on morphometric traits. With the FlyCatwalk, we can detect gender and quantify body and wing morphology parameters at a four-old higher throughput compared with manual processing. The phenotyping results acquired using the FlyCatwalk correlate well with those obtained using the standard manual procedure. We demonstrate that an automated, high-throughput, feature-based sorting system is able to avoid previous limitations in population size and replicate numbers. Our approach can likewise be applied for a variety of traits and experimental settings that require high-throughput phenotyping., G3: Genes, Genomes, Genetics, 5 (3), ISSN:2160-1836

The robust and efficient flight control of insects provide a powerful model system for autonomous microrobots. Conversely, robots offer a robust experimental platform on which to test biological hypotheses. This interaction of biology and robotics is an exciting but challenging task, because the vast disparities between both can lead to inaccurate or even misleading conclusions. In this paper, we present a biorobotic platform that can arbitrarily define the dynamic couplings between a fruit fly and a robot. The platform is used to explore the stability and emergent properties of the biorobotic couple. The fruit flypsilas wing kinematics are measured in real time and used to drive an autonomous robot. In turn, the robotpsilas sensory information is transformed back into visual feedback to the fly. Using different case studies, we explore how the choice of feedback influences the success of the biorobotic device. We discuss the meaning of such feedback in view of biomimetic implementations.

In the context of a pilot project, the Lugaggia Innovation Community (LIC), we address the problem of non-intrusive load monitoring for the purpose of demand side management on low voltage grids in presence of distributed power generation (photovoltaic). From the power readings of smart meters, we estimate the photovoltaic production and detect the activation of major loads (heatpumps and domestic water heaters). Experiments, conducted with real data and in silico, show that exploiting meter readings only, we can estimate PV production with MAPE ranging from 4.6% (best case) to 41.9% (worst case). Even with non negligible photovoltaic production estimation errors, the proposed method is capable of detecting the activation of heatpumps and domestic water heaters.