.png){kind=icon}
返回
Statistical methods for interpolating missing meteorological data for use in building simulation
.png){kind=icon}
PRE
AI
DOI：10.1007/s12273-014-0174-7.png)
.png){kind=icon}
原文求助.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
Building performance simulation is increasingly used to aid in decision making about the design, construction, retrofit, operation, and maintenance of new and existing buildings. Such simulations require a complete set of meteorological data sampled at regular intervals. A data file with even a single missing measurement value becomes useless for simulation. Unfortunately, it is extremely rare to find such a perfect body of data. Measurement errors and sensor failure are frequent occurrences in meteorological data collection and are among a host of reasons for missing measurement values. To overcome this problem, simulation users may rely on Typical Meteorological Years (TMYs) instead of actual historical data, or they may apply an existing interpolation method to fill the gaps in historical data. Historical data is often preferable, since TMYs fail to account for atypical weather conditions. Clearly, this could lead to poor decision making when the decision outcomes are strongly affected by the occurrence of atypical conditions. This paper presents several methods for statistical interpolation between discrete weather-data points. A normalization procedure is first used to transform meteorological data into a set of Gaussian-distributed sample data. Next, a vector autoregressive model is calibrated using the normalized site-specific meteorological data, and is then used to determine the most likely value for one or more missing data points. Variations of the model are described to address specific combinations of missing data, and the methods are validated for several cities in the USA. Results show that the normalization procedure is the most important contributor towards a significant improvement in accuracy relative to other interpolation methods.
Keyword:
weather data
statistical interpolation
vector autoregression
risk-conscious design
AI总结
对已上传原文的论文进行重点信息的提取，主要内容包括：简要概述、研究摘要、背景介绍、关键亮点、图文解析、展望与总结。