| 1. |
- Borga, Magnus, et al.
(author)
-
A Unified Approach to PCA, PLS, MLR and CCA
- 1997
-
Reports (other academic/artistic)abstract
- This paper presents a novel algorithm for analysis of stochastic processes. The algorithm can be used to find the required solutions in the cases of principal component analysis (PCA), partial least squares (PLS), canonical correlation analysis (CCA) or multiple linear regression (MLR). The algorithm is iterative and sequential in its structure and uses on-line stochastic approximation to reach an equilibrium point. A quotient between two quadratic forms is used as an energy function and it is shown that the equilibrium points constitute solutions to the generalized eigenproblem.
|
|
| 2. |
|
|
| 3. |
- Bui, Hai Thanh, 1972-, et al.
(author)
-
Group theoretical investigations of daylight spectra
- 2004
-
In: 2nd European Conference on Color in Graphics, Imaging and Vision,2004. - Springfield, VA, USA : IST: The Society for Imaging Science and Technology. ; , s. 437-
-
Conference paper (peer-reviewed)
|
|
| 4. |
- Campana, Pietro Elia, 1984-, et al.
(author)
-
A gridded optimization model for photovoltaic applications
- 2020
-
In: Solar Energy. - : PERGAMON-ELSEVIER SCIENCE LTD. - 0038-092X .- 1471-1257. ; 202, s. 465-484
-
Journal article (peer-reviewed)abstract
- This study aims to develop a gridded optimization model for studying photovoltaic applications in Nordic countries. The model uses the spatial and temporal data generated by the mesoscale models STRANG and MESAN developed by the Swedish Meteorological and Hydrological Institute. The model is developed based on the comparison between five irradiance databases, three decomposition models, two transposition models, and two photovoltaic models. Several techno-economic and environmental aspects of photovoltaic systems and photovoltaic systems integrated with batteries are investigated from a spatial perspective. CM SAF SARAH-2, Engerer2, and Perez1990 have shown the best performances among the irradiance databases, and decomposition and transposition models, respectively. STRANG resulted in the second-best irradiance database to be used in Sweden for photovoltaic applications when comparing hourly global horizontal irradiance with weather station data. The developed model can be employed for carrying out further detailed gridded techno-economic assessments of photovoltaic applications and energy systems in general in Nordic countries. The model structure is generic and can be applied to every gridded climatological database worldwide.
|
|
| 5. |
|
|
| 6. |
|
|
| 7. |
- Hellgren, Laila, et al.
(author)
-
Severe mitral regurgitation : relations between magnetic resonance imaging, echocardiography and natriuretic peptides
- 2008
-
In: Scandinavian Cardiovascular Journal. - : Informa UK Limited. - 1401-7431 .- 1651-2006. ; 42:1, s. 48-55
-
Journal article (peer-reviewed)abstract
- BACKGROUND:Assessment of the severity of mitral regurgitation by echocardiography can be technically demanding in certain patients and supplementary methods are therefore desirable. This study addressed the agreement between magnetic resonance imaging (MRI) and echocardiography, and their relations to natriuretic peptides (NT-proANP and NT-proBNP), in quantifying severe mitral regurgitation.METHODS:Eighteen patients with severe mitral regurgitation scheduled for surgery underwent MRI, echocardiography and assay of natriuretic peptides preoperatively for clinical assessment.RESULTS:MRI and echocardiography were comparable in measuring severity of regurgitation qualitatively but not quantitatively, mitral regurgitant fraction (mean difference 27.5 (11) ml). There was a correlation between increasing regurgitant fraction on MRI and increased levels of plasma NT-proANP and NT-proBNP. In echocardiography, increasing vena contracta width and increasing PISA correlated to increased levels of plasma NT-proANP and NT-proBNP. No other correlation was found between measures on MRI and echocardiography and natriuretic peptides.CONCLUSIONS:MRI and echocardiography were comparable grading the severity of mitral regurgitation with qualitative measures but not with quantitative measures. MRI might be a complement to echocardiography when a more distinct measure of the regurgitant volume is needed, as in paravalvular leakage.
|
|
| 8. |
- Knutsson, Hans, et al.
(author)
-
Generalized Eigenproblem for Stochastic Process Covariances
- 1996
-
Reports (other academic/artistic)abstract
- This paper presents a novel algorithm for finding the solution of the generalized eigenproblem where the matrices involved contain expectation values from stochastic processes. The algorithm is iterative and sequential to its structure and uses on-line stochastic approximation to reach an equilibrium point. A quotient between two quadratic forms is suggested as an energy function for this problem and is shown to have zero gradient only at the points solving the eigenproblem. Furthermore it is shown that the algorithm for the generalized eigenproblem can be used to solve three important problems as special cases. For a stochastic process the algorithm can be used to find the directions for maximal variance, covariance, and canonical correlation as well as their magnitudes.
|
|
| 9. |
- Knutsson, Hans, et al.
(author)
-
Learning Canonical Correlations
- 1995
-
Reports (other academic/artistic)abstract
- This paper presents a novel learning algorithm that finds the linear combination of one set of multi-dimensional variates that is the best predictor, and at the same time finds the linear combination of another set which is the most predictable. This relation is known as the canonical correlation and has the property of being invariant with respect to affine transformations of the two sets of variates. The algorithm successively finds all the canonical correlations beginning with the largest one. It is shown that canonical correlations can be used in computer vision to find feature detectors by giving examples of the desired features. When used on the pixel level, the method finds quadrature filters and when used on a higher level, the method finds combinations of filter output that are less sensitive to noise compared to vector averaging.
|
|
| 10. |
- Knutsson, Hans, 1950-, et al.
(author)
-
Learning Multidimensional Signal Processing
- 1998
-
In: Proceedings of the 14th International Conference on Pattern Recognition, vol 2. - Linköping, Sweden : Linköping University, Department of Electrical Engineering. ; , s. 1416-1420
-
Reports (other academic/artistic)abstract
- This paper presents our general strategy for designing learning machines as well as a number of particular designs. The search for methods allowing a sufficient level of adaptivity are based on two main principles: 1. Simple adaptive local models and 2. Adaptive model distribution. Particularly important concepts in our work is mutual information and canonical correlation. Examples are given on learning feature descriptors, modeling disparity, synthesis of a global 3-mode model and a setup for reinforcement learning of online video coder parameter control.
|
|
| 11. |
- Koehler, Birgit, 1980-, et al.
(author)
-
Simulation of photoreactive transients and of photochemical transformation of organic pollutants in sunlit boreal lakes across 14 degrees of latitude : A photochemical mapping of Sweden
- 2018
-
In: Water Research. - : Elsevier. - 0043-1354 .- 1879-2448. ; 129, s. 94-104
-
Journal article (peer-reviewed)abstract
- Lake water constituents, such as chromophoric dissolved organic matter (CDOM) and nitrate, absorb sunlight which induces an array of photochemical reactions. Although these reactions are a substantial driver of pollutant degradation in lakes they are insufficiently understood, in particular on large scales. Here, we provide for the first time comprehensive photochemical maps covering a large geographic region. Using photochemical kinetics modeling for 1048 lakes across Sweden we simulated the steady-state concentrations of four photoreactive transient species, which are continuously produced and consumed in sunlit lake waters. We then simulated the transient-induced photochemical transformation of organic pollutants, to gain insight into the relevance of the different photoreaction pathways. We found that boreal lakes were often unfavorable environments for photoreactions mediated by hydroxyl radicals ([rad]OH) and carbonate radical anions (CO3−[rad]), while photoreactions mediated by CDOM triplet states (3CDOM*) and, to a lesser extent, singlet oxygen (1O2) were the most prevalent. These conditions promote the photodegradation of phenols, which are used as plastic, medical drug and herbicide precursors. When CDOM concentrations increase, as is currently commonly the case in boreal areas such as Sweden,3CDOM* will also increase, promoting its importance in photochemical pathways even more.
|
|
| 12. |
- Koehler, Birgit, 1980-, et al.
(author)
-
Sunlight-induced carbon dioxide emissions from inland waters
- 2014
-
In: Global Biogeochemical Cycles. - 0886-6236 .- 1944-9224. ; 28:7, s. 696-711
-
Journal article (peer-reviewed)abstract
- The emissions of carbon dioxide (CO2) from inland waters are substantial on a global scale. Yet, the fundamental question remains open which proportion of these CO2 emissions is induced by sunlight via photochemical mineralization of dissolved organic carbon (DOC), rather than by microbial respiration during DOC decomposition. Also, it is unknown on larger spatial and temporal scales how photochemical mineralization compares to other C fluxes in the inland water C cycle. We combined field and laboratory data with atmospheric radiative transfer modeling to parameterize a photochemical rate model for each day of the year 2009, for 1086 lakes situated between latitudes from 55 to 69°N in Sweden. The sunlight-induced production of dissolved inorganic carbon (DIC) averaged 3.8 ± 0.04 g C m-2 yr-1, which is a flux comparable in size to the organic carbon burial in the lake sediments. Countrywide, 151 ± 1 kt C yr-1 was produced by photochemical mineralization, corresponding to about 12% of total annual mean CO2 emissions from Swedish lakes. With a median depth of 3.2 m, the lakes were generally deep enough that incoming, photochemically active photons were absorbed in the water column. This resulted in a linear positive relationship between DIC photoproduction and the incoming photon flux, which correspond to the absorbed photons. Therefore, the slope of the regression line represents the wavelength- and depth-integrated apparent quantum yield of DIC photoproduction. We used this relationship to obtain a first estimate of DIC photoproduction in lakes and reservoirs worldwide. Global DIC photoproduction amounted to 13 and 35 Mt C yr-1 under overcast and clear sky, respectively. Consequently, these directly sunlight-induced CO2 emissions contribute up to about one tenth to the global CO2 emissions from lakes and reservoirs, corroborating that microbial respiration contributes a substantially larger share than formerly thought, and generate annual C fluxes similar in magnitude to the C burial in natural lake sediments worldwide.
|
|
| 13. |
- Landelius, Tomas, et al.
(author)
-
A Dynamic Tree Structure for Incremental Reinforcement Learning of Good Behavior
- 1994
-
Reports (other academic/artistic)abstract
- This paper addresses the idea of learning by reinforcement, within the theory of behaviorism. The reason for this choice is its generality and especially that the reinforcement learning paradigm allows systems to be designed, which can improve their behavior beyond that of their teacher. The role of the teacher is to define the reinforcement function, which acts as a description of the problem the machine is to solve. Gained knowledge is represented by a behavior probability density function which is approximated with a number of normal distributions, stored in the nodes of a binary tree. It is argued that a meaningful partitioning into local models can only be accomplished in a fused space consisting of both stimuli and responses. Given a stimulus, the system searches for responses likely to result in highly reinforced decisions by treating the sum of the two normal distributions on each level in the tree as a distribution describing the system's behavior at that resolution. The resolution of the response, as well as the tree growing and pruning processes, are controlled by a random variable based on the difference in performance between two consecutive levels in the tree. This results in a system that will never be content but will indefinitely continue to search for better solutions.
|
|
| 14. |
- Landelius, Tomas, et al.
(author)
-
A system for modelling solar radiation parameters with mesoscale spatial resolution
- 2001
-
Reports (other academic/artistic)abstract
- Today, modern analysis systems synthesise meteorological data from a number of sources, e.g.\ground based SYNOP, satellites, radar, etc., into field information which enable us to model radiation at the Earth’s surface on the mesoscale. At the Swedish Meteorological and Hydrological Institute (SMHI) we have set up a model system that produce hourly information in terms of field data with a resolution of about 22x22 km2 for a geographic area covering Scandinavia and the run off region of the Baltic sea. Presently, the model calculates fields of global-, photosynthetically active- (PAR), UV- and direct radiation based on output from a mesoscale analysis system, a high resolution limited area numerical weather prediction model (NWP), an ice model for the Baltic sea together with satellite measurements of total ozone. A spectral clear sky model lies at the heart of the model system. Its output is multiplied by a function which captures the influence of clouds and precipitation. Different cloud effect functions are applied to the different radiation components, with the exception of global- and PAR for which the same relation is assumed. Measurements from the radiation network of SMHI were used for estimation and validation purposes. A first evaluation of the model system suggests that the RMSE for hourly global radiation data is on the order of 28% and about 16% for daily values. These errors are comparable to those obtained for models purely based on synoptic observations (SYNOP) (29% and 13%) . For UV radiation the figures are similar but for the direct radiation component they are worse; 53% and 31% respectively compared to 25% and 15% for the SYNOP models. To some extent the larger errors for the direct component could be explained by its sensitivity to scale differences when model grid squares are validated against point measurements.
|
|
| 15. |
- Landelius, Tomas
(author)
-
Behavior Representation by Growing a Learning Tree
- 1993
-
Licentiate thesis (other academic/artistic)abstract
- The work presented in this thesis is based on the basic idea of learning by reinforcement, within the theory of behaviorism. The reason for this choice is the generality of such an approach, especially that the reinforcement learning paradigm allows systems to be designed which can improve their behavior beyond that of their teacher. The role of the teacher is to define the reinforcement function, which acts as a description of the problem the machine is to solve.Learning is considered to be a bootstrapping procedure. Fragmented past experience, of what to do when performing well, is used for response generation. The new response, in its turn, adds more information to the system about the environment. Gained knowledge is represented by a behavior probability density function. This density function is approximated with a number of normal distributions which are stored in the nodes of a binary tree. The tree structure is grown by applying a recursive algorithm to the stored stimuli-response combinations, called decisions. By considering both the response and the stimulus, the system is able to bring meaning to structures in the input signal. The recursive algorithm is first applied to the whole set of stored decisions. A mean decision vector and a covariance matrix are calculated and stored in the root node. The decision space is then partitioned into two halves across the direction of maximal data variation. This procedure is now repeated recursively for each of the two halves of the decision space, forming a binary tree with mean vectors and covariance matrices in its nodes.The tree is the system's guide to response generation. Given a stimulus, the system searches for responses likely to result in highly reinforced decisions. This is accomplished by treating the sum of the normal distributions in the leaves as distribution describing the behavior of the system. The sum of normal distributions, with the current stimulus held fixed, is finally used for random generation of the response.This procedure makes it possible for the system to have several equally plausible responses to one stimulus. Not applying maximum likelihood principles will make the system more explorative and reduce its risk of being trapped in local minima.The performance and complexity of the learning tree is investigated and compared to some well known alternative methods. Presented are also some simple, yet principally important, experiments verifying the behavior of the proposed algorithm.
|
|
| 16. |
|
|
| 17. |
- Landelius, Tomas, et al.
(author)
-
Depth and Velocity from Orientation Tensor Fields
- 1993
-
In: SCIA8.
-
Conference paper (peer-reviewed)abstract
- This paper presents an algorithm for retrieving depth and velocity by estimating the 3D-orientation in an image sequence under the assumption of pure translation of the camera in a static scene. Quantitative error measurements are presented comparing the proposed algorithm to a gradient based optical flow algorithm.
|
|
| 18. |
|
|
| 19. |
- Landelius, Tomas, et al.
(author)
-
Greedy adaptive critics for LPQ [dvs LQR] problems : Convergence Proofs
- 1996
-
Reports (other academic/artistic)abstract
- A number of success stories have been told where reinforcement learning has been applied to problems in continuous state spaces using neural nets or other sorts of function approximators in the adaptive critics. However, the theoretical understanding of why and when these algorithms work is inadequate. This is clearly exemplified by the lack of convergence results for a number of important situations. To our knowledge only two such results been presented for systems in the continuous state space domain. The first is due to Werbos and is concerned with linear function approximation and heuristic dynamic programming. Here no optimal strategy can be found why the result is of limited importance. The second result is due to Bradtke and deals with linear quadratic systems and quadratic function approximators. Bradtke's proof is limited to ADHDP and policy iteration techniques where the optimal solution is found by a number of successive approximations. This paper deals with greedy techniques, where the optimal solution is directly aimed for. Convergence proofs for a number of adaptive critics, HDP, DHP, ADHDP and ADDHP, are presented. Optimal controllers for linear quadratic regulation (LQR) systems can be found by standard techniques from control theory but the assumptions made in control theory can be weakened if adaptive critic techniques are employed. The main point of this paper is, however, not to emphasize the differences but to highlight the similarities and by so doing contribute to a theoretical understanding of adaptive critics.
|
|
| 20. |
- Landelius, Tomas, et al.
(author)
-
On-Line Singular Value Decomposition of Stochastic Process Covariances
- 1995
-
Reports (other academic/artistic)abstract
- This paper presents novel algorithms for finding the singular value decomposition (SVD) of a general covariance matrix by stochastic approximation. General in the sense that also non-square, between sets, covariance matrices are dealt with. For one of the algorithms, convergence is shown using results from stochastic approximation theory. Proofs of this sort, establishing both the point of equilibrium and its domain of attraction, have been reported very rarely for stochastic, iterative feature extraction algorithms.
|
|
| 21. |
- Landelius, Tomas, et al.
(author)
-
Reinforcement Learning Adaptive Control and Explicit Criterion Maximization
- 1996
-
Reports (other academic/artistic)abstract
- This paper reviews an existing algorithm for adaptive control based on explicit criterion maximization (ECM) and presents an extended version suited for reinforcement learning tasks. Furthermore, assumptions under which the algorithm convergences to a local maxima of a long term utility function are given. Such convergence theorems are very rare for reinforcement learning algorithms working with continuous state and action spaces. A number of similar algorithms, previously suggested to the reinforcement learning community, are briefly surveyed in order to give the presented algorithm a place in the field. The relations between the different algorithms is exemplified by checking their consistency on a simple problem of linear quadratic regulation (LQR).
|
|
| 22. |
- Landelius, Tomas
(author)
-
Reinforcement Learning and Distributed Local Model Synthesis
- 1997
-
Doctoral thesis (other academic/artistic)abstract
- Reinforcement learning is a general and powerful way to formulate complex learning problems and acquire good system behaviour. The goal of a reinforcement learning system is to maximize a long term sum of instantaneous rewards provided by a teacher. In its extremum form, reinforcement learning only requires that the teacher can provide a measure of success. This formulation does not require a training set with correct responses, and allows the system to become better than its teacher.In reinforcement learning much of the burden is moved from the teacher to the training algorithm. The exact and general algorithms that exist for these problems are based on dynamic programming (DP), and have a computational complexity that grows exponentially with the dimensionality of the state space. These algorithms can only be applied to real world problems if an efficient encoding of the state space can be found.To cope with these problems, heuristic algorithms and function approximation need to be incorporated. In this thesis it is argued that local models have the potential to help solving problems in high-dimensional spaces and that global models have not. This is motivated with the biasvariance dilemma, which is resolved with the assumption that the system is constrained to live on a low-dimensional manifold in the space of inputs and outputs. This observation leads to the introduction of bias in terms of continuity and locality.A linear approximation of the system dynamics and a quadratic function describing the long term reward are suggested to constitute a suitable local model. For problems involving one such model, i.e. linear quadratic regulation problems, novel convergence proofs for heuristic DP algorithms are presented. This is one of few available convergence proofs for reinforcement learning in continuous state spaces.Reinforcement learning is closely related to optimal control, where local models are commonly used. Relations to present methods are investigated, e.g. adaptive control, gain scheduling, fuzzy control, and jump linear systems. Ideas from these areas are compiled in a synergistic way to produce a new algorithm for heuristic dynamic programming where function parameters and locality, expressed as model applicability, are learned on-line. Both top-down and bottom-up versions are presented.The emerging local models and their applicability need to be memorized by the learning system. The binary tree is put forward as a suitable data structure for on-line storage and retrieval of these functions.
|
|
| 23. |
- Landelius, Tomas, et al.
(author)
-
Reinforcement Learning Trees
- 1996
-
Reports (other academic/artistic)abstract
- Two new reinforcement learning algorithms are presented. Both use a binary tree to store simple local models in the leaf nodes and coarser global models towards the root. It is demonstrated that a meaningful partitioning into local models can only be accomplished in a fused space consisting of both input and output. The first algorithm uses a batch like statistic procedure to estimate the reward functions in the fused space. The second one uses channel coding to represent the output- and input vectors allowing a simple iterative algorithm based on competing subsystems. The behaviors of both algorithms are illustrated in a preliminary experiment.
|
|
| 24. |
- Landelius, Tomas, et al.
(author)
-
The Learning Tree, A New Concept in Learning
- 1993
-
In: Proceedings of the 2nd International Conference on Adaptive and Learning Systems.
-
Conference paper (peer-reviewed)abstract
- In this paper learning is considered to be the bootstrapping procedure where fragmented past experience of what to do when performing well is used for generation of new responses adding more information to the system about the environment. The gained knowledge is represented by a behavior probability density function which is decomposed into a number of normal distributions using a binary tree. This tree structure is built by storing highly reinforced stimuli-response combinations, decisions, and calculating their mean decision vector and covariance matrix. Thereafter the decision space is divided, through the mean vector, into two halves along the direction of maximal data variation. The mean vector and the covariance matrix are stored in the tree node and the procedure is repeated recursively for each of the two halves of the decision space forming a binary tree with mean vectors and covariance matrices in its nodes. The tree is the systems guide to response generation. Given a stimuli the system searches for decisions likely to give a high reinforcement. This is accomplished by treating the sum of the normal distributions in the leaves, using their mean vectors and covariance matrices as the distribution parameters, as a distribution describing the systems behavior. A response is generated by fixating the stimuli in this sum of normal distribution and use the resulting distribution, which turns out to be a new sum of normal distributions, for random generation of the response. This procedure will also make it possible for the system to have several equally plausible response to one stimuli when this is appropriate. Not applying maximum likelihood principles will lead to a more explorative system behavior avoiding local minima traps.
|
|
| 25. |
- Stridh, Bengt, Universitetslektor, 1957-, et al.
(author)
-
Förbättrad beräkning av solelproduktion i Sverige
- 2020
-
Reports (other academic/artistic)abstract
- Nordligt läge med lägre solstrålning än exempelvis södra Europa och förhållandevis lågt elpris gör att noggranna förutsägelser av energiutbyte från solcellsanläggningar är av stor vikt när man gör investeringskalkyler i Sverige. Noggrannare beräkningar av förväntad solelproduktion ger mindre ekonomisk osäkerhet, vilket resulterar i en mer resurseffektiv utveckling. Val av meteorologiska data och beräkningsmetod för kalkyler av solelproduktion är därför av stor vikt.En fråga är därför vilket simuleringsprogram för solelproduktion som är bäst att använda i Sverige. OptiCE, Polysun, PVsyst och PV*SOL med programmens meteorologiska databaser visade sig här vara relativt likvärdiga för Stockholm, Norrköping och Visby. Överensstämmelsen är relativt god med de uppmätta värdena för solelproduktion under 2019, med skillnader på mindre än ±5%. Men de ger alla 13%-15% för höga värden för Kiruna. PVGIS med databas ERA5 ger lite större avvikelser för Stockholm, Norrköping och Visby än ovan nämnda program men ger ett värde nära det uppmätta under 2019 i Kiruna. SAM och PVGIS med databaserna SARAH eller COSMO ger större avvikelser än ovan nämnda program. Då SARAH i en jämförande studie hade bäst noggrannhet är det tänkbart att beräkningarna i PVGIS skulle kunna förbättras genom att välja SARAH i kombination med ett lägre värde än grundinställningen 14% för systemförluster.Den största osäkerheten vid uppskattning av solcellssystems elproduktion kommer från solstrålningsdata. Genom att förbättra solstrålningsdata och göra dem allmänt tillgängliga hjälps investerare att fatta beslut med minskad osäkerhet. Det finns behov av en branschstandard för solstrålningsdata i Sverige. En vidareutveckling av STRÅNG-modellen för solstrålningsdata är önskvärd. Ett standardförfarande hur man beräknar inverkan av skuggning skulle vara värdefullt, då skuggning vid sidan om val av solstrålningsdatabaser kan ha en stor inverkan på utbytet av solel.Solstrålningsklimatet kan förändras över tid, vilket man kan se i uppmätt solstrålning för Sverige. I framtiden kan även pågående klimatförändring ha betydelse för solinstrålning och därmed solenergiproduktion. Data för solstrålning, vind, temperatur och albedo från klimatscenarion för två tidsperioder (2030-2065 och 2066-2095) användes för att uppskatta hur solelproduktionen kan komma att påverkas. Resultatet pekar på att solelproduktionen minskar något men att förändringen endast är statistiskt signifikant i det scenario som representerar fortsatt höga koldioxidutsläpp och då endast för norra Sverige under den senare tidsperioden. Sett över hela landet beräknas förändringen för denna period hamna mellan -9% (10:e percentilen) och -2% (90:e percentilen) med medelvärde på ca -6%.De kartor för Sverige för optimerade lutningar, solstrålning och solelproduktion som tagits fram med den utvecklade modellen OptiCE är ett verktyg för att bättre förstå, utforma och förbättra installationer av solcellssystem i Sverige.Bland de undersökta modellerna för uppdelning av global horisontell solstrålning i diffus och direkt strålning för att ta fram egna solstrålningsdata för användning i simuleringsprogram är slutsatsen att för timvärden är Engerer2 eller Paulescu och Blaga lämpliga val. För 1-minutvärden visar Yang2 bäst prestanda.
|
|
| 26. |
- van Noord, Michiel, et al.
(author)
-
Snow-induced PV loss modeling using production-data inferred PV system models
- 2021
-
In: Energies. - : MDPI AG. - 1996-1073. ; 14:6
-
Journal article (peer-reviewed)abstract
- Snow-induced photovoltaic (PV)-energy losses (snow losses) in snowy and cold locations vary up to 100% monthly and 34% annually, according to literature. Levels that illustrate the need for snow loss estimation using validated models. However, to our knowledge, all these models build on limited numbers of sites and winter seasons, and with limited climate diversity. To overcome this limitation in underlying statistics, we investigate the estimation of snow losses using a PV system’s yield data together with freely available gridded weather datasets. To develop and illustrate this approach, 263 sites in northern Sweden are studied over multiple winters. Firstly, snow-free production is approximated by identifying snow-free days and using corresponding data to infer tilt and azimuth angles and a snow-free performance model incorporating shading effects, etc. This performance model approximates snow-free monthly yields with an average hourly standard deviation of 6.9%, indicating decent agreement. Secondly, snow losses are calculated as the difference between measured and modeled yield, showing annual snow losses up to 20% and means of 1.5-6.2% for winters with data for at least 89 sites. Thirdly, two existing snow loss estimation models are compared to our calculated snow losses, with the best match showing a correlation of 0.73 and less than 1% bias for annual snow losses. Based on these results, we argue that our approach enables studying snow losses for high numbers of PV systems and winter seasons using existing datasets. © 2021 by the authors.
|
|
| 27. |
- van Noord, Michiel, et al.
(author)
-
Utveckling av prognosmodeller och –verktyg för snöpåverkan på solelproduktion via fjärrmätning
- 2021
-
Reports (other academic/artistic)abstract
- Solcellsanläggningar installeras överallt i Sverige, från Kurland i söder till Kiruna i norr. Förhållandena mellan våra två landsändar är dock rätt stora. Detta projekt har undersökt hur snöfall påverkar elproduktionen från solceller, med fokus på Mellersta och Norra Sverige. Resultaten från drygt 260 anläggningar och upp till sex vintersäsonger visar att snöförluster är något att räkna med. Årliga förluster upp till 20% har konstaterats. I snitt förväntas de flesta anläggningar dock komma undan med årliga förluster under 10% och i många fall under 5%. Tydligt är att förlusterna blir större ju längre norrut och ju närmare fjällen solcellerna befinner sig. För att få en uppskattning på hur stora snöförlusterna kan vara där du befinner dig har ett gratis verktyg publicerats på snosolel.ri.se. Den stora utmaningen i projektet har varit att kunna studera så många anläggningar som möjligt, för att säkerställa att resultaten är relevanta. Därför har historisk produktionsdata för 263 anläggningar analyserats kombinerad med data för solinstrålning, snödjup och temperatur från vädermodeller och satelliter. Med all denna data har projektet lyckats modellera anläggningarnas prestanda över tid, inklusive sådant som skuggning, med relativt bra precision. Genom att jämföra de modellerna med uppmätt produktion under vintersäsongerna har snöförlusterna beräknats. I ett nästa steg jämfördes snöförlusterna för de studerade anläggningarna med två befintliga modeller för att uppskatta snöförluster. Problemet med dessa och liknande modeller har varit att de är svåra att verifiera mot många anläggningar och över stora geografiska områden. Metoden som utvecklades i detta projekt gör det möjligt att utföra verifieringar med befintliga data utan att behöva komplettera med extra mätningar på plats. Det visade sig att ingen av uppskattnings-modellerna var särskilt bra på att uppskatta snöförlusterna per månad, men att den ena gav rätt bra uppskattningar för årliga förluster. Denna modell, utvecklad av Marion m.fl. (2013), har implementerats i ett gratis online verktyg som uppskattar ungefärliga snöförluster för en solcellsanläggning på valfri plats i Sverige (och delar av Finland och Norge). Inledande försök pekar på att det finns potential att utveckla bättre uppskattningsmodeller för snöförluster. Det finns också goda förhoppningar att kunna förbättra precisionen i metoden för att modellera solcellsanläggningarna utifrån deras produktionsdata.
|
|
| 28. |
- Zainali, Sebastian, 1995-, et al.
(author)
-
Site adaptation with machine learning for a Northern Europe gridded global solar irradiance product
- 2023
-
In: Energy and AI. - 2666-5468. ; 15
-
Journal article (peer-reviewed)abstract
- Gridded global horizontal irradiance (GHI) databases are fundamental for analysing solar energy applications' technical and economic aspects, particularly photovoltaic applications. Today, there exist numerous gridded GHI databases whose quality has been thoroughly validated against ground-based irradiance measurements. Nonetheless, databases that generate data at latitudes above 65˚ are few, and those available gridded irradiance products, which are either reanalysis or based on polar orbiters, such as ERA5, COSMO-REA6, or CM SAF CLARA-A2, generally have lower quality or a coarser time resolution than those gridded irradiance products based on geostationary satellites. Amongst the high-latitude gridded GHI databases, the STRÅNG model developed by the Swedish Meteorological and Hydrological Institute (SMHI) is likely the most accurate one, providing data across Sweden. To further enhance the product quality, the calibration technique called "site adaptation" is herein used to improve the STRÅNG dataset, which seeks to adjust a long period of low-quality gridded irradiance estimates based on a short period of high-quality irradiance measurements. This study introduces a novel approach for site adaptation of solar irradiance based on machine learning techniques, which differs from the conventional statistical methods used in previous studies. Seven machine-learning algorithms have been analysed and compared with conventional statistical approaches to identify Sweden's most accurate algorithms for site adaptation. Solar irradiance data gathered from three weather stations of SMHI is used for training and validation. The results show that machine learning can substantially improve the STRÅNG model's accuracy. However, due to the spatiotemporal heterogeneity in model performance, no universal machine learning model can be identified, which suggests that site adaptation is a location-dependant procedure.
|
|
