| Summary: | 博士 === 國立臺北科技大學 === 設計學院設計博士班 === 107 === Using the trend prediction function of artificial intelligence to assist in processing a large amount of information generated in the physical environment of the building, and maintaining the technical and energy consumption corresponding to the quality of the indoor environment of the building. Benefit management in saving construction costs and reducing operating costs can be positively beneficial, and in terms of situation management, optimization and individualized needs can be achieved. This research method establishes the ANN model in different programming languages, and collects the interactive experimental PMV of the existing building environment using the existing people as the main experimental data, and eliminates the data that cannot be typed into the ANN and then splits the analysis. And design the ANN model through the iterative training of supervised learning to the accurate artificial neural network model training ANN model and can be fast and fast fitting, the research results calculate the weight and weight bias value, the process analysis gets the standard way of ANN model design, as each time The ANN model re-establishes and references the method of inputting big data.
The contribution of the research is:
1. The data of the building physical environment needs to be shunted and then designed with the appropriate ANN model and the indicators obtained after training can be moved to the next level of the ANN model to continue the calculation.
2. The ANN model established by Matlab and the computational function Levenberg-Marquardt is suitable for winter PMV prediction and knows that the data of continuous data stream is adapted to the ANN model type.
3. The designed ANN model is applicable to the establishment of sound, light, heat, air of quality and energy saving in the building physical environment.
|
|---|
