At Qualitrol, we understand that the analysis of fault records during a system disturbance can be a laborious task. Getting to the correct records and extracting the relevant information quickly can help reduce outage times and result in significant cost savings. Quailtrol have been working on a new machine learning approach to fault analysis.
Leveraging data from a worldwide install base to train learning algorithms has allowed faults to be categorized with 99% accuracy. Parameters can be extracted from the waveforms automatically and fault reports can be generated with all the relevant information within minutes of a fault occurring.
Many utilities have a large install base of fault recorders and/or protection relays. Sometimes up to and beyond 200-400 units. During a storm a system this size could accumulate close to 1000 records. GridAssist was born out of the need to reduce the time required to manually process 1000's of records. GridAssist will automatically highlight the critical few records to allow the correct decissions to be made rapidly.
Efficiently categorize all fault records to allow immediate response
The AI built into the software has been trained on Qualitrols global data library of hundreds of thousands of records to achieve analysis accuracy greater than 99%
Automate reporting to support colleagues in Operation and in the field to improve response time and post-mortem investigations
Automatically categorise data from large or small populations of fault recorders or protection relays into record types such as “Circuit trip”, ”Failed to trip”, “Close onto fault”, “Circuit de-energized”, “Circuit energized”, “Through fault” etc. Automatically calculate relevant fault parameters such as protection pick up time, distance to fault (single and double ended), fault duration, faulted phase, maximum current flowed, etc.
Cluster fault records together from the same fault to ensure all necessary data is in one place. For example, circuit trip and circuit reclose (both ends). Present all the data in an easy to navigate user interface.
Fill in the contact us form to register for a demo of the fault classifier tool.