The hydrothermal scheduling problem aims to determine an operation strategy that produces generation targets for each power plant at each stage of the planning horizon. This strategy aims to minimize the expected value of the operation cost over the planning horizon, composed of fuel costs to operate thermal plants plus penalties for failure in load supply. The system state at each stage is highly dependent on the water inflow at each hydropower generator reservoir. This work focuses on developing a probabilistic model for the inflows that is suitable for a multistage stochastic algorithm that solves the hydrothermal scheduling problem. The probabilistic model that governs the inflows is based on a dynamic linear model. Due to the cyclical behavior of the inflows, the model incorporates seasonal and regression components. We also incorporate climate variables such as precipitation, El Ni\~no, and other ocean indexes, as predictive variables when relevant. The model is tested for the power generation system in Brazil with about 140 hydro plants, which are responsible for more than 80\% of the electricity generation in the country. At first, these plants are gathered by basin and classified into 15 groups. Each group has a different probabilistic model that describes its seasonality and specific characteristics. The inflow forecast derived with the probabilistic model at each stage of the planning horizon is a continuous distribution, instead of a single point forecast. We describe an algorithm to form a finite scenario tree by sampling from the inflow forecasting distribution with interstage dependency, that is, the inflow realization at a specific stage depends on the inflow realization of previous stages. / text

Inflow forecasting is important when planning the use of water in a hydropower plant. The process of making forecasts is characterized by using knowledge from previous events and occurrences to make predictions about the future. Traditionally, inflow is predicted using hydrological models. The model developed by the Hydrologiska Byråns Vattenbalansavdelning (HBV model) is one of the most widely used hydrological models around the world. Machine learning is emerging as a potential alternative to the current HBV models but needs to be evaluated. This thesis investigates machine learning for inflow forecasting as a mixed qualitative and quantitative case study. Interviews with experts in various backgrounds within hydropower illustrated the key issues and opportunities for inflow forecasting accuracy and laid the foundation for the machine learning model created. The thesis found that the noise in the realised inflow data was one of the main factors which affected the quality of the machine learning inflow forecasts. Other notable factors were the precipitation data from the three closest weather stations. The interviews suggested that the noise in the realised inflow data could be due to faulty measurements. The interviews also provided examples of additional data such as snow quantity measurements and ground moisture levels which could be included in a machine learning model to improve inflow forecast performance. One proposed application for the machine learning model was as a complementary tool to the current HBV model to assist in making manual adjustments to the forecasts when considered necessary. The machine learning model achieved an average Mean Absolute Error (MAE) of 1.39 compared to 1.73 for a baseline forecast for inflow to the Lake Kymmen river system 1-7 days ahead over the period 2015-2019. For inflow to the Lake Kymmen river system 8-14 days ahead the machine learning model achieved an average MAE of 1.68 compared to 2.45 for a baseline forecast. The current HBV model in place had a lower average MAE than the machine learning model over the available comparison period of January 2018. / Tillrinningsprognostisering är viktig vid planeringen av vattenanvändningen i ett vattenkraftverk. Prognostiseringsprocessen går ut på att använda tidigare kunskap för att kunna göra prediktioner om framtiden. Traditionellt sett har tillrinningsprognostisering gjorts med hjälp av hydrologiska modeller. Hydrologiska Byråns Vattenbalansavdelning-modellen (HBV-modellen) är en av de mest använda hydrologiska modellerna och används världen över. Maskininlärning växer för tillfället fram som ett potentiellt alternativ till de nuvarande HBV-modellerna men behöver utvärderas. Det här examensarbetet använder en blandad kvalitativ och kvantitativ metod för att utforska maskininlärning för tillrinningsprognostisering i en fallstudie. Intervjuer med experter med olika bakgrund inom vattenkraft påtalade nyckelfrågor och möjligheter för precisering av tillrinningsprognostisering och lade grunden för den maskininlärningsmodell som skapades. Den här studien fann att brus i realiserade tillrinningsdata var en av huvudfaktorerna som påverkade kvaliteten i tillrinningsprognoserna av maskininlärningsmodellen. Andra nämnvärda faktorer var nederbördsdata från de tre närmaste väderstationerna. Intervjuerna antydde att bruset i realiserade tillrinningsdatana kan bero på felaktiga mätvärden. Intervjuerna bidrog också med exempel på ytterligare data som kan inkluderas i en maskininlärningsmodell för att förbättra tillrinningsprognoserna, såsom mätningar av snömängd och markvattennivåer. En föreslagen användning för maskininlärningsmodellen var som ett kompletterande verktyg till den nuvarande HBV-modellen för att underlätta manuella justeringar av prognoserna när det bedöms nödvändigt. Maskininlärningsmodellen åstadkom ett genomsnittligt Mean Absolute Error (MAE) på 1,39 jämfört med 1,73 för en referensprognos för tillrinningen till Kymmens sjösystem 1–7 dagar fram i tiden under perioden 2015–2019. För tillrinningen till Kymmens sjösystem 8–14 dagar fram i tiden åstadkom maskininlärningsmodellen ett genomsnittligt MAE på 1,68 jämfört med 2,45 för en referensprognos. Den nuvarande HBV-modellen hade ett lägre genomsnittligt MAE jämfört med maskininlärningsmodellen under den tillgängliga jämförelseperioden januari 2018.

Orientador: Ieda Geriberto Hidalgo / Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Engenharia Mecânica / Made available in DSpace on 2018-08-25T19:15:52Z (GMT). No. of bitstreams: 1 Signoriello_GiuseppeAlessandro_M.pdf: 31629174 bytes, checksum: 1674c1adcccf93d9b3ee9711be3f709e (MD5) Previous issue date: 2014 / Resumo: Este trabalho apresenta a comparação de dois modelos matemáticos desenvolvidos para prever vazões afluentes à usinas hidrelétricas. O objetivo é abordar os aspectos que determinam a qualidade do insumo fundamental para a programação da operação do sistema hidrotérmico brasileiro: a previsão de vazões. A ferramenta de suporte à avaliação dos modelos matemáticos é o SISPREV, gerenciador/executor de estudos de previsão de vazões desenvolvido na UNICAMP. Esta ferramenta permite ao usuário prever vazões diárias e mensais com base em modelos de Regressão Linear (RL) e Sistema de Inferência Neuro-Fuzzy (SINF). Avaliou-se a qualidade das previsões diárias e mensais dos modelos RL e SINF através da metodologia de mineração de dados Cross Industry Standard Process for Data Mining (CRISP-DM). A CRISP-DM é baseada em um modelo hierárquico de processos comumente usados na descoberta de conhecimento. Os resultados mostram que o modelo RL apresenta um desempenho melhor para previsões diárias e o modelo SINF para as previsões mensais. Além disso, o modelo RL tem a tendência a ter bom desempenho de previsão nas situações típicas de chuva-vazão, enquanto os melhores índices de desempenho do modelo SINF caem nas condições atípicas, em particular com a contemporaneidade de altas vazões e baixas precipitações / Abstract: This work presents a comparison between two different mathematical models developed to predict inflows to hydropower plants. The purpose is to explore the aspects that determine the quality of an important input variable for operation planning of the Brazilian hydrothermal system: the inflows forecasting. The tool that supports the evaluation of the mathematical models is called SISPREV. It is a manager/runner of inflows forecasting studies developed at UNICAMP. This tool allows the user to predict daily and monthly inflows based on Linear Regression (RL) models and Neuro-Fuzzy Inference System (SINF). In this thesis, was evaluated the quality of daily and monthly forecasts of RL and SINF models using the methodology Cross Industry Standard Process for Data Mining. CRISP-DM is used in the discovery of knowledge and based on a hierarchical process model. The results show that the RL model performs better for daily predictions and the SINF model for monthly predictions. Furthermore, the RL model tends to have better performance in typical situations of rainfall-inflow, while the best performance indices of the SINF model fall in atypical conditions, in particular with the simultaneous high inflow rates and low precipitation / Mestrado / Planejamento de Sistemas Energeticos / Mestre em Planejamento de Sistemas Energéticos

Orientador: Secundino Soares Filho / Tese (doutorado) - Universidade Estadual de Campinas, Faculdade de Engenharia Eletrica e de Computação / Made available in DSpace on 2018-08-15T02:32:47Z (GMT). No. of bitstreams: 1 Zambelli_MonicadeSouza_D.pdf: 4122152 bytes, checksum: 3256b0546520645ef065b43111b44374 (MD5) Previous issue date: 2009 / Resumo: O planejamento da operação energética do Sistema Interligado Nacional (SIN) é uma tarefa complexa realizada por meio de uma cadeia de modelos de médio, curto e curtíssimo prazo acoplados entre si, cada um com considerações pertinentes à etapa que aborda. A proposta deste trabalho é apresentar uma alternativa para o planejamento da operação energética de médio prazo. Foi desenvolvida uma metodologia baseada em modelo de controle preditivo, abordando os aspectos estocásticos do problema de forma implícita pela utilização de valores esperados das vazões, e fazendo uso de um modelo determinístico de otimização a usinas individualizadas, que possibilita uma representação mais precisa do sistema hidrotérmico. A análise de desempenho é feita através de simulações da operação, considerando os parques hidrelétrico e termelétrico que compõem o SIN, com restrições operativas reais, em configuração dinâmica, com plano de expansão e a possibilidade de intercâmbio e importação de mercados vizinhos. Os resultados são comparados aos fornecidos pela metodologia em vigor no setor elétrico brasileiro, notadamente o modelo NEWAVE, que determina as decisões de geração por subsistema, e o modelo Suishi-O, que as desagrega por usinas individualizadas / Abstract: The long term hydrothermal scheduling of the Brazilian Integrated System (SIN) is a complex task solved by a chain of long, medium and short term coupled models, each one with considerations pertinent to the stage of operation that it deals with. The proposal of this work is to present an alternative for the long term hydrothermal scheduling. A methodology based on model predictive control was developed, implicitly handling stochastic aspects of the problem by the use of inflows expected values, and making use of a deterministic optimization model to obtain the optimal dispatch for individualized plants, what makes possible a more accurate representation of the hydrothermal system. The performance analysis is made through simulations of the operation, taking into consideration all the hydro and thermal plants that compose the SIN, with real operative constraints, in dynamic configuration, with its expansion plan and the possibility of interchange and importation from neighboring markets. The results are compared with those provided by the approach actually in use by the Brazilian electric sector, specifically the NEWAVE model, which defines the generation decisions for the subsystems, and the Suishi-O model, that disaggregates them for the individualized plants / Doutorado / Energia Eletrica / Doutor em Engenharia Elétrica