• Refine Query
  • Source
  • Publication year
  • to
  • Language
  • 3
  • Tagged with
  • 3
  • 3
  • 3
  • 3
  • 3
  • 3
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • About
  • The Global ETD Search service is a free service for researchers to find electronic theses and dissertations. This service is provided by the Networked Digital Library of Theses and Dissertations.
    Our metadata is collected from universities around the world. If you manage a university/consortium/country archive and want to be added, details can be found on the NDLTD website.
1

Modeling of the primary sludge thickening process at a wastewater treatment plant with the use of machine learning / Modellering av förtjockningsprocessen av primärslam på ett avloppsreningsverk

Bröndum, Eric January 2022 (has links)
This thesis focuses on modeling the primary sludge in the thickening process at Henrikdals wastewater treatment plant in Stockholm, Sweden. The thickening process is one of the core processes at the wastewater treatment plant, where the goal is to thicken a residual product called primary sludge. Two thickener belts are used to thicken the sludge gravimetrically. Polymer is also added to increase the dewaterability and to thicken the sludge. The thickness of the sludge is measured by the total solids content (TS) in the sludge and is measured with total solid measurement sensors. These sensors have, however, been shown to be inaccurate. A long short-term memory network (LSTM) and a feed-forward neural network were compared by using sensor and instrument data to predict the TS in the thickened primary sludge. To validate the performance of the models, manual laboratory testing samples were compared with the predictions of the models. Simulations in Simulink were also performed with the intent of simulating the thickening process. By using a machine learning model that could predict the TS, hypotheses regarding reductions in the polymer dosage were explored. A feed-forward and feedback control strategy in combination with the LSTM architecture were used and it was shown that the TS of the thickened sludge could be controlled by regulating the polymer dosage. Thus, using a feedback control strategy gives further opportunities for the wastewater treatment plant to choose whether a lower polymer consumption or a higher TS is preferred, as these two variables correlate with each other. / Syftet med detta arbete var att ta fram maskininlärningsmodeller av primärslamsförtjockningen på Henriksdals avloppsreningsverk i Stockholm, Sverige. Förtjockningsprocessen är en av de viktigaste delerna i avloppsreningsverk, där målet är att förtjocka en restprodukt som kallas primärslam. Förtjockningen sker i två separata linjer. Polymer tillsätts och slammet förtjockas genom gravimetrisk avvattning på ett silband. Slammets torrsubstanshalt (TS) är ett mått på slammets tjockhet och beräknas med hjälp av att använda sensorer. Dessa sensorer har dock visats sig vara opålitliga. Genom att använda tillgänglig process-, maskin- och instrumentdata så har en long short-term memory (LSTM) arkitektur och ett framkopplat neuralt nätverk jämförts för att uppskatta torrsubstansen i primärslammet. Manuell provtagning och labbanalys utfördes för att validera prestandan i de två modellerna. Hypoteser kring att kunna optimera TS-halten eller minska polymerförbrukningen utforskades genom att simulera processen i Simulink. Resultaten visade att användandet av en fram och återkopplingsregulator tillsammans med en LSTM arkitektur kan minska polymerförbrukningen och kan ge en jämnare TS-halt i det förtjockade slammet. Däremot måste en avvägning mellan hög TS-halt och låg polymerförbrukning göras, då dessa två variabler korrelerar med varandra
2

A comparative study between algorithms for time series forecasting on customer prediction : An investigation into the performance of ARIMA, RNN, LSTM, TCN and HMM

Almqvist, Olof January 2019 (has links)
Time series prediction is one of the main areas of statistics and machine learning. In 2018 the two new algorithms higher order hidden Markov model and temporal convolutional network were proposed and emerged as challengers to the more traditional recurrent neural network and long-short term memory network as well as the autoregressive integrated moving average (ARIMA). In this study most major algorithms together with recent innovations for time series forecasting is trained and evaluated on two datasets from the theme park industry with the aim of predicting future number of visitors. To develop models, Python libraries Keras and Statsmodels were used. Results from this thesis show that the neural network models are slightly better than ARIMA and the hidden Markov model, and that the temporal convolutional network do not perform significantly better than the recurrent or long-short term memory networks although having the lowest prediction error on one of the datasets. Interestingly, the Markov model performed worse than all neural network models even when using no independent variables.
3

Evaluating deep learning models for electricity spot price forecasting

Zdybek, Mia January 2021 (has links)
Electricity spot prices are difficult to predict since they depend on different unstable and erratic parameters, and also due to the fact that electricity is a commodity that cannot be stored efficiently. This results in a volatile, highly fluctuating behavior of the prices, with many peaks. Machine learning algorithms have outperformed traditional methods in various areas due to their ability to learn complex patterns. In the last decade, deep learning approaches have been introduced in electricity spot price prediction problems, often exceeding their predecessors. In this thesis, several deep learning models were built and evaluated for their ability to predict the spot prices 10-days ahead. Several conclusions were made. Firstly, it was concluded that rather simple neural network architectures can predict prices with high accuracy, except for the most extreme sudden peaks. Secondly, all the deep networks outperformed the benchmark statistical model. Lastly, the proposed LSTM and CNN provided forecasts which were statistically, significantly superior and had the lowest errors, suggesting they are the most suitable for the prediction task. / Elspotspriser är svåra att förutsäga eftersom de beror på olika instabila och oregelbundna faktorer, och också på grund av att elektricitet är en vara som inte kan lagras effektivt. Detta leder till ett volatilt, fluktuerande beteende hos priserna, med många plötsliga toppar. Maskininlärningsalgoritmer har överträffat traditionella metoder inom olika områden på grund av deras förmåga att lära sig komplexa mönster. Under det senaste decenniet har djupinlärningsmetoder introducerats till problem inom elprisprognostisering och ofta visat sig överlägsna sina föregångare. I denna avhandling konstruerades och utvärderades flera djupinlärningsmodeller på deras förmåga att förutsäga spotpriserna 10 dagar framåt. Den första slutsatsen är att relativt simpla nätverksarkitekturer kan förutsäga priser med hög noggrannhet, förutom för fallen med de mest extrema, plötsliga topparna. Vidare, så övertränade alla djupa neurala nätverken den statistiska modellen som användes som riktmärke. Slutligen, så gav de föreslagna LSTM- och CNN-modellerna prognoser som var statistiskt, signifikant överlägsna de andra och hade de lägsta felen, vilket tyder på att de är bäst lämpade för prognostiseringsuppgiften.

Page generated in 0.1086 seconds