Preventing Equipment Failures with Automated Battery Anomaly Detection

Impact

A US utility operating a multi-station network relied on battery monitors to protect critical equipment, but spotting abnormal signals in the monitoring data was a struggle. That left it exposed to equipment failures and service disruptions. Blackstraw built an automated system that scans battery monitoring data, detects major and minor abnormalities, and alerts the right people immediately. The solution fully automated battery monitoring, helped the utility avoid major failures, and cut regulatory compliance costs by 12%.

Background

Station batteries are typically a backup power source for critical equipment, so a failing battery can go unnoticed until it is needed. The utility collected readings from its battery monitors, but reviewing that data by hand across many stations was slow and inconsistent. Abnormal signals were easy to miss, and monitors with missing data were hard to spot. The client needed a reliable, repeatable way to check every battery monitor, surface problems early and keep a clear record for compliance.

Solution Highlights

• Automated Data Pipeline: Pulls 90 days of battery monitor data from the utility’s PI Server and cleans it before analysis.

• Rule-Based Anomaly Detection: Thresholds flag major and minor abnormalities in each monitor’s signals in real time.

• Data Gap Detection: Flags monitors with too little data for analysis, so blind spots are visible as well as faults.

• Instant Alerts and Reports: Notifies stakeholders as soon as an anomaly is found, and produces summary reports and charts comparing normal and abnormal signals.

• Daily Monitoring in Power BI: Runs every day and shows results in a Power BI dashboard.

Key Benefits

• 12% Lower Compliance Costs: Regulatory compliance costs fell 12% with automated checks and a clear audit record.

• Major Failures Avoided: Automatic alerts let teams act before battery problems became equipment failures.

• 100% Automated Monitoring: Every battery monitor was checked daily, with no manual review needed.

• Foundation for Predictive Maintenance: The system laid the groundwork for machine learning models that predict and prevent battery failures.