Predictive maintenance is an ideal that will become a must as the solar industry continues to grow. A new report from DNV GL highlights just how cool/valuable predictive maintenance can be when implemented, but also how far we are from it. The good news is the biggest obstacle seems (in a way) fairly simple to overcome – it just takes a more focused, collective approach to O&M.
First, what is predictive maintenance? DNV GL’s ‘Predictive maintenance of solar PV plants: The time is now‘ defines it as a combo of monitoring data and machine learning algorithms to anticipate system failure and thereby better plan any required maintenance intervention. The potential benefits being significantly reduced maintenance costs and reduced operational downtime.
As an example, the report highlights DNV GL’s work with GreenPowerMonitor on a system based on machine learning techniques, that “given the input from a solar inverter, in addition to environmental variables, identifies anomalous behavior within observed data channels based on the operational history for any given inverter.” Their testing and modeling has demonstrated that it can detect up to 83% of failures a week ahead of inverter failure.
The challenge: The short-coming of their system, alluded to earlier, is that it requires detailed and high-quality maintenance logs to train against – “something that the majority of systems lack and has prevented further development and validation of the method.” DNV GL has been focusing on a more generalizable approach that could be used without a history of maintenance logs. That system works to an extent, but it’s tough to hone in a sweet spot – “a balance that allows it to predict the maximum number of failures possible, while maintaining a high true alarm rate.”
Bottom line is the technology is here, but “the absence of detailed maintenance logs and standardization is hindering its development and expansion within the solar industry. Maintenance logs are required for accurate model training and validation, while the lack of standardization in logs, error codes and naming conventions is a severe obstacle for the automation of data analysis and model generation.”
So, OK, that is pretty daunting, but it’s also a totally reasonable and achievable goal that plant owners, operators and manufacturers can start to work toward right now.
“The path towards the future is clear: Owners, operators, and manufacturers must come together and establish procedures and services that enforce the proper and standardized documentation of all relevant events that occur in a PV plant. Moreover, they must exploit the digitalization tools at their disposal in order to facilitate the upkeep and automated access to O&M logs for data analysis routines.”
— Solar Builder magazine