摘要
Since many industries may not be able to migrate for Industry 4.0 directly, Industry 3.5 is proposed as a hybrid strategy. While facility is crucial for product quality for hightech manufacturing, manufacturers have focused mainly on facility stability and paid less attention for power saving. In practice, the operation of chillers relies on domain experts leading to inconsistent decisions and operational inefficiency. Although a number of studies have proposed intelligent facilities with similar visions of Industry 4.0, they may not be ready to be adopted in the operating facilities. Furthermore, limitations of existing approaches can be traced in part to the lack of an integrated framework within which interrelated decisions can be modeled and related to the optimal configuration of chillers. Following Industry 3.5 stragey, this study aims to develop an approach that integrates a cooling load forecasting model based on SARIMAX model, a prediction model for the operation efficiency of the chillers, and chiller health evaluation model to optimize the combination of operating chillers to fulfill the cooling load demand and minimize the electricity consumption given the existing infrastructure. To estimate the validity, an empirical study was conducted in a leading semiconductor company for carbon reduction and green production. The results have shown its practical viability and the developed solution has been implemented in this company with significant power saving.