In-Building Fusion of Deep Data from Digital-Twins? and their Concomitant Control Schemes

PassiveLogic proposes a novel methodology for automated building science data fusion that piggybacks on the fundamental market need: automation. The proposed project is automatic deep data fusion enabled and managed by an autonomous building platform, where the combined purposes of data fusion and automation enable a platform that provides autonomous testing and validation of the data fusion in real-time. The DOE’s goals of energy-efficiency, energy demand reduction and flexibility, critical water issues, and resiliency are limited by the non-standardization of data sets coming from buildings. Data sets, even those emerging from the same building or the same system within a building, cannot be fused together to provide deeper understanding of the components the data is meant to describe, thus limiting deep analytics solutions such as demand-side management. The proposed solution is an in-building platform using high-speed digital twins that co-simulate and regress the target physical building in a controlled or observed system. By utilizing a continuously regressed physics-based representation of the target building, the system can extract data to a deep physics level of description, even for data points for which specific sensors or instrumentation don’t exist a priori, while also maintaining quality of data measures using statistical reinforcement with the digital twins. An example of deep data extraction is, in comparison to a conventional analytics system which can only track whether a pump is signaled “on”, the proposed system would be able to extract rotor speed, flow rates, pressure, fluid type, and errors, in addition to the corresponding quality of data measures. This additional data extraction is made possible due to the inter-validation of physical properties in the digital twin models. Currently neither automation systems, sensor networks, nor analytics platforms have any inherent understanding about how buildings and building systems actually work. This leaves the analysis of buildings as a complex post-processing exercise, that lacks sufficient data or data quality indicators. Because the current market approach to building data is off- line, validation is difficult without the test subject in the loop. Thus, the opportunity for deep data is lost because physical information can't be extracted through black box operations, and the expertise and labor requirements for accurate analytics are exceedingly high. The proposed technology architecture is designed with the market in mind. Not only does the architecture enable singular building science data fusion, but it also simplifies and optimizes building automation. We have demonstrated our early alpha technology to automation customers and distributors, validating our approach and demand in the marketplace for a future automation platform product.