Substation PQ and RFI

Risk Reduction Category

Grid Monitoring

Technology Description

Power quality (PQ) events are recorded by PQ meters whenever anomalous events are detected on the power grid. Such events can be indicative of grid conditions that presage wildfire events, or can be indicative of their occurrence. Waveform data is among the most important types of measurement records due to its rich technical content; however, it is also one of the most difficult to use as, often, only detailed visual review by highly trained personnel can identify the pertinent data. The objective is to enable utilities to obtain faster situational awareness using automated waveform identification.

Using neural networks with machine learning can aid in accurately classifying the recorded waveforms and help power system engineers diagnose and rectify the root causes of problems. However, many of the waveforms captured during a disturbance in the power system need to be labeled for supervised learning, leaving a large number of data recordings for engineers to process manually or go unseen. For example, faulted voltage and current waveforms can be analyzed by AI to determine root causes. Common and unique signatures can be identified and used to classify a large data set based on the nature and cause of the fault. These classes include cable faults (cable, joint or splice, and termination failures), animal and tree contact faults, lightning induced faults, and faults cleared by current-limiting fuses.[1]

There are a number of important known power quality phenomena for which automated waveform identification would be valuable. Three such examples are described here:

  • Arcing: An obvious factor in preventing wildfire is detection of electrical arching. This phenomenon is readily detectible in waveforms and automating that ability could be pivotal.
  • Conductor slap: Overhead conductors coming into contact also produces unique signatures. Automated identification of these occurrences can dramatically improve mitigating and preventative measures.
  • Foliage ingress: Proactive detection of the impingement of plant life on electrical circuits is an on- going and expensive task for utilities. Identification of direct contact between plant life and conductors as it is occurring can allow quick intervention as well as optimization of investment in mitigation and prevention.
  • Incipient faults: Early detection of incipient faults can prevent catastrophic failure of equipment. Recognizing these faults and remediating the issue proactively can prevent fires and improve grid performance.

Current Waveform having Incipient Faults and Their Reactance-to-Fault Estimate [1]

Voltage Waveforms of a Cable Splice Failure [1]

Technical Readiness (Commercial Availability)

Next steps that are required in order to advance the usefulness of waveform data and achieve the visionary objectives of an AI sandbox would include these broad tasks:

  • Assembly and deployment of a large and comprehensive waveform and event signature library by incorporating large, existing resources available at EPRI (over 600,000 waveforms currently), and adding additional libraries from key stakeholders such as electric utilities, national labs, universities, etc.
  • Training of the waveform library using modern AI and Machine Learning techniques.

A remaining challenges in achieving the ultimate objective includes integrating the monitoring data from a wide variety of data sources like relays and meters, integrating circuit information from Cyme or GIS, and incorporating a user interface into a control center.

Implementations / Deployments

Innovations as of Mid 2023

Potential Enrichment Work Opportunity

References

[1] Distribution Fault Location and Waveform Characterization. EPRI, Palo Alto, CA: 2009. 1017842.