Evaluating and Improving Model-Based Assessment of Contextual Data Quality in Smart Grids

Data science has great potential in smart grids, but applying it must not compromise dependability—thus requiring high-quality data. Ideally, quality is ensured in early development stages. To enable this, we propose using architecture models for data quality assessment. The study focuses on a domain-specific language based on the Smart Grid Architecture Model and models created with it. The first goal is to evaluate how suitable such models are for assessing contextual data quality—the data’s fitness for use. We discovered the assessability is mainly limited by the modeling language not facilitating standards-independent data definition. Consequently, we set out to improve the language. We arrived at three proposed modifications using a case study–based design-science approach: including an element for generic data definition, separating data and information, and adding attributes to data flows. Our research demonstrates the feasibility of model-based data quality assessment and takes a step towards integrating data science into smart-grid architecture.