Filtering by
- All Subjects: Jazz
DOE prototype medium office building compliant with ASHRAE 90.1-2010 was selected for the analysis and four different HVAC systems in three US climates were simulated.
The interaction between building design parameters related to envelope characteristics and geometry (total of seven variables) has been studied using two different statistical methods, namely the ‘Morris method’ and ‘Predictive Learning via Rule Ensembles’.
Subsequently, a simple graphical tool based on sensitivity analysis has been developed and demonstrated to present the results from parametric simulations. This tool would be useful to better inform design decisions since it allows imposition of constraints on various parameters and visualize their interaction with other parameters.
It was observed that the Radiant system performed best in all three climates, followed by displacement ventilation system. However, it should be noted that this study did not deal with performance optimization of HVAC systems while there have been several studies which concluded that a VAV system with better controls can perform better than some of the newer HVAC technologies. In terms of building design parameters, it was observed that ‘Ceiling Height’, ‘Window-Wall Ratio’ and ‘Window Properties’ showed highest importance as well as interaction as compared to other parameters considered in this study, for all HVAC systems and climates.
Based on the results of this study, it is suggested to extend such analysis using statistical methods such as the ‘Morris method’, which require much fewer simulations to categorize parameters based on their importance and interaction strength. Usage of statistical methods like ‘Rule Ensembles’ or other simple visual tools to analyze simulation results for all combinations of parameters that show interaction would allow designers to make informed and superior design decisions while benefiting from large reduction in computational time.
This research aims at developing such a methodology for generating cost effective net zero energy solutions for school buildings. The Department of Energy (DOE) prototype primary school, meant to serve as the starting baseline, was modeled in the building energy simulation software eQUEST and made compliant with the requirement of ASHRAE 90.1-2007. Commonly used efficiency measures, for which credible initial cost and maintenance data were available, were selected as the parametric design set. An initial sensitivity analysis was conducted by using the Morris Method to rank the efficiency measures in terms of their importance and interaction strengths. A sequential search technique was adopted to search the solution space and identify combinations that lie near the Pareto-optimal front; this allowed various minimum cost design solutions to be identified corresponding to different energy savings levels.
Based on the results of this study, it was found that the cost optimal combination of measures over the 30 year analysis span resulted in an annual energy cost reduction of 47%, while net zero site energy conditions were achieved by the addition of a 435 kW photovoltaic generation system that covered 73% of the roof area. The simple payback period for the additional technology required to achieve NZE conditions was calculated to be 26.3 years and carried a 37.4% premium over the initial building construction cost. The study identifies future work in how to automate this computationally conservative search technique so that it can provide practical feedback to the building designer during all stages of the design process.
The purpose of this project is to create a useful tool for musicians that utilizes the harmonic content of their playing to recommend new, relevant chords to play. This is done by training various Long Short-Term Memory (LSTM) Recurrent Neural Networks (RNNs) on the lead sheets of 100 different jazz standards. A total of 200 unique datasets were produced and tested, resulting in the prediction of nearly 51 million chords. A note-prediction accuracy of 82.1% and a chord-prediction accuracy of 34.5% were achieved across all datasets. Methods of data representation that were rooted in valid music theory frameworks were found to increase the efficacy of harmonic prediction by up to 6%. Optimal LSTM input sizes were also determined for each method of data representation.
My proposed project is an educational application that will seek to simplify the<br/>process of internalizing the chord symbols most commonly seen by those learning<br/>musical improvisation. The application will operate like a game, encouraging the<br/>user to identify chord tones within time limits and award points for successfully<br/>doing so.