Time Series Snippets: A Novel Distance Measure for Data Mining in Challenging Scenarios

As we continue to shift towards renewable energy sources, optimizing energy storage becomes crucial for minimizing emissions and maximizing efficiency. In this article, we explore a novel approach called "time series snippets" that helps in selecting the optimal duration and capacity of energy storage systems based on representative periods. By leveraging these snippets, we can effectively capture the fullspace value of interday energy sharing while avoiding emissions violations.
The article begins by providing an overview of the original time series algorithm and its limitations. It then introduces the proposed method, which is built on the matrix profile distance (MPdist) measure that compares two-time series and considers them similar if they have similar subsequences. The goal is to select the best sequences from a time series of length t to generalize the full-time series for energy storage optimization.
The article then delves into the discussion of representative period length, highlighting its impact on duration and capacity of installed storage. As expected, longer periods result in higher emissions due to less predictable effects with lengths over 3 days. However, by representing intermediate periods, time series snippets are better able to capture the fullspace value of interday energy sharing, leading to lower operating costs without significantly increasing total costs.
The article concludes by presenting the results of numerical studies that demonstrate the effectiveness of time series snippets in optimizing energy storage systems. The study shows that the d = 3 scenario (representative period of 4 days) has the lowest overall cost despite larger investment, with the total costs for 2030 being 1.1% higher but emissions 7.1% lower.
In simple terms, time series snippets are like puzzle pieces that help us optimize energy storage systems. By selecting the right sequence of these pieces based on representative periods, we can minimize emissions while maximizing efficiency. The longer the period, the more predictable the effects become, but at the cost of higher emissions. Finding the sweet spot between duration and capacity is crucial for optimizing energy storage systems, and time series snippets offer a powerful tool to achieve this goal.