by Sunil Kotagiri, Lead Consultant, Utilities and Geospatial BU, Cyient
As the Internet of Things (IoT) proliferates across the utility network, the increased number of smart devices, Distributed Energy Resources (DER), and Information and Communication Technology (ICT) infrastructure are creating operational complexities, new business paradigms and benefit scenarios.
An increase in DER across the distribution network mandates real-time generation and supply modelling, bringing fresh challenges to the utilities in managing and interpreting large volumes of information.
To streamline this transformation, utilities throughout the world are turning to Advanced Distribution Management Systems (ADMS) and related operational analytics. Data, however, remains a significant stumbling block to realising the benefits.
For instance, ADMS and operational analytics can enable power flow modelling to help dispatch smaller energy resources only when the system accurately understands actual load characteristics based on real-time phase and system load.
While ADMS play a vital role in enhancing grid resiliency, it heavily relies on the data coming from various IT systems and grid sensing devices to develop advanced applications and operational analytics.
Utilities are exploring ways to validate if the input data is operationally viable – complete, accurate, consistent and current.
Moreover, utilities are looking for best approaches to sustain the data quality for meeting their stretched goals.
The traditional way of data validation using static tools and processes may no longer be productive, and indeed a transformation leveraging advanced techniques like machine learning is the need of the hour. The quality of network connectivity and data governance can be significantly enhanced when machine learning based algorithms are supplemented with real-time grid intelligence.
Smart meters are increasingly gaining acceptance, leading to a proliferation of data along with voltage and load profiles at regular intervals. However, the usage of such data in operations is limited, while this data is immensely useful in meter-to-cash processes.
If voltage signature data from the meters and connected grid content is incorporated into the data validation routines, they can bring in real intelligence to undertake the relational check of key attributes such as phasing and meter-to-transformer ties.
As machine learning algorithms continue to ‘learn’ from the voltage and load data, they become increasingly valuable in data validation and augmentation processes, which ensures the availability of right data in the GIS at all times, and accurate data for ADMS operations.
When an ADMS gets access to validated ‘true-state’ representation of the network model, it makes energy dispatch more accurate even after the integration of distributed energy resources. It improves fault localisation, isolation and restoration processes, and enhances the accuracy of switching operations for the safety of workforce and assets.
Grid resiliency can only get better if static data and real-time data from sensing devices are used in conjunction. Therefore, data governance plays a key role along with machine learning based data validation algorithms in measuring, protecting and assuring data quality in a manner sufficient to meet the targeted business objectives for which the data is sought.