Very frequently the limiting factor in efficiently troubleshooting issues with wind turbines is the quality of the data available for analisis. This is unfortunate because access to good data along with appropriate tools for analysis can be invaluable for quickly fixing turbine issues as they occur or preventing them before they result in unnecessary downtime.
Also, most SCADA systems do not allow operators to access enough data to get a fully detailed overview of their site and the performance of each individual asset. The PLC in each turbine can output far more data than tends to be available from SCADA. Having access to this data in a well organized and easy to use structure would go a long way to helping with management of renewable assets.
SCADAScope is a platform which seeks to solve these problems by collecting high resolution signal data from various sources, extracting the most useful insights from that data and making all data available for further analysis. SCADAScope provides both tools for navigating the data in a visual form and for doing further more sophisticated data analytics on stored information. This post describes the components of SCADAScope and outlines how they work together as a powerful tool for managing renewable assets.
SCADAScope Architecture
SCADAScope is composed of several components. A set of data acquisition systems gather signal data at the highest resolution available. The feature extractor scans through this signal data and extracts the most useful information into files each representing 10 minutes of data. SCADAScope API makes both raw and processed data available to the web viewer or for external data analysis. SignalScope is a web viewer for visual analysis and overviews of asset performance. Asset Insights Engine uses the 10 minute data from the feature extractor to find higher level insights, such as finding high priority problems with specific turbine components or performance.
Data Acquisition Systems
Because SCADAScope can gather data from multiple asset types (e.g., Turbine PLC or Converter Control Unit), different sets of data acquisition systems can be configured to run depending on the needs of the site. For example, if a site has Gamesa wind turbines, the Gamesa DA system would be used.
This system connects to the PLC and gathers high speed data which is stored into parquet files every 10 minutes. Once this data is stored it is easily accessible both from the file system, API, or in the web viewer where any set of data points can be charted.
Feature Extractor
The large amount of data produced by the data acquisition systems is used further on in the system in various ways. Very frequently, the results of calculations on this data show up over and over again for different requirements. By distilling out the most important properties of the data, the Feature Extractor reduces the overall amount of processing required, puts the data into a more useful form, and makes transferring the data to other locations more convenient. The Feature Extractor creates JSON objects containing this distilled information. The feature data is stored to the file system where it is easy to work with, both by data scientists and other components of SCADAScope.
The modular design of the feature extractor makes extending the list of the calculations it does straightforward. All that is required to add additional processing is to add a new “feature calc” as a simple python function. For example, if new knowledge becomes available on how to detect a certina issue from the raw data, a new feature calc can be created to capture this observation.
SignalsScope
Though the SCADAScope system contains features for automated detection of issues, very frequently issues are better understood or solved when all relevant signals are available for visual analysis. SignalsScope provides the ability to conveniently graph any signals related to the problem they are trying to solve.
As SignalsScope is used to troubleshoot some types of problems, the common causes of such problems can become clearer, allowing the manual process of using SignalsScope to be abstracted into a feature calc and added to the system.
Asset Insights Engine
Using the data generated by the Feature Extractor, the Asset Insights Engine produces things like Quality and Health Reports or Troubleshooting Memos. It runs at regular intervals as the feature data becomes available from the Feature Extractor. The Insights Engine can be configured and extended through the use of plugins to automate the process of finding problems with assets; it can also be used to generate reports on the full functioning of the site.
The difference between the Asset Insights Engine and the feature extractor is that the feature extractor extracts properties of the signal data, while the Asset Insights Engine uses those properties to create higher level, actionable insights or bigger picture reports.
SCADAScope API
The SCADAScope API exists to make the signal data, the feature data, and the results from the Asset Insight Engine available. The primary use for the API is to provide data to the web app, but it is also useful for loading data into other systems for analysis.
Data Management
Every 10 minutes, the data acquisition system creates a parquet file for each connected asset which is stored directly to the file system in a directory structure which makes it easy to find later based on time or asset number.
The feature files are also stored to the filesystem for use on site, but can be optionally uploaded to the cloud for off site analysis.
The signal data and feature data can optionally be published to the cloud to make off-sight analisis more convenient. For setups which do not require data to be written to the cloud, the feature and raw data can be accessed by either FTP or the SCADAScope API.
Deployment Process
Because everything required can run on a single server, the deployment process for SCADAScope is straightforward. We create configuration files which list the assets and any extra information required to access them on the site, along with details about which feature calcs to run. With this file in place, the whole system can be started on a field server, making both the data and the web app available to anyone on the site.
Once SCADAScope is running on a new site, we monitor the site to assure uptime of the system, by making sure uploads to the cloud are happening and that all assets are connected. If you are curious about SCADAScope or wonder if it could improve your operations, please let us know (service@aprenewables.com).