Revealing Microclimate around Buildings with Long-Term Monitoring through the Neural Network Algorithms

The profile of urban microclimates is important in many engineering fields, such as occupant’s thermal comfort and health, other building engineering. To predict the microclimate, this study applies artificial neural network long short-term memory predictive models, an microclimate dataset was obtained with a long-term monitoring from year 2017 to 2019 5-min resolution including temperature, relative humidity, solar radiation. Two models were applied, first (Model 1) apply techniques real-time sequence, then extract characteristics while second 2) directly microclimate. Backpropagation (BP-ANN) long-short term (LSTM) applied both models. results show Model 1 time-series prediction can reach best (99.92%) correlation coefficient 98% mean average percentage error (MAPE), for 99.66% 98.18% respectively, accuracies 2 decreased 79% 88.6% MAPE temperature respectively. radiation using ANN LSTM are 51.1% 57.8% coefficient,