Deep learning for assessment of environmental satisfaction using BIM big data in energy efficient building digital twins

Energy efficient Building Digital Twins (BDTs) are researched using Building Information Model (BIM) to explore the key techniques of Digital Twins (DTs). DTs in buildings can be regarded as an expression of “BIM+,” born to digital descriptions. Comprehensive perception of physical systems is the preconditions for DTs implementation. BIM’s energy-saving design includes the selection of building orientation and building shape. BIM energy consumption analysis can compare different materials, examine the performance of various materials, and select the most suitable and most energy-efficient materials for building structure maintenance. Data Fusion Algorithm (DFA) in Wireless Sensor Networks (WSNs) is improved. A novel DFA is constructed by combining Backpropagation Neural Network (BPNN) with Dynamic Host Configuration Protocol (DCHP), recorded as BP-DCHP. Simulation experiment proves that BP-DCHP can prolong sensor nodes’ survival time and provide the highest data fusion quality. BP-DCHP runs for about 310 s, 500 s, and 705 s in WSNs consisting of 20, 50, and 100 WSNs, respectively. Moreover, BP-DCHP can provide higher quality given insufficient data fusion degree. Once the WSNs consume 50% of the total initial energy, BP-DCHP presents a shorter network delay, only 0.6 s on average in the 100-sensor-node-WSN. To validate BDTs’ effectiveness, the environmental satisfaction of residents from two Beijing intelligent communities is assessed using Deep Learning (DL) approach. Taking the data as the clue, the study establishes DTs serving the application of urban scene, which plays a certain role in promoting the technological innovation of BDTs, better optimizing the city and managing the city.