Limited Edge Compute
Industrial PCs struggle with ultra-high-frequency sampling, downsampling, feature computation, and anomaly detection — leading to alert delays or missed alerts that compromise operational safety.
High Transmission and Storage Costs
Uploading full-resolution data strains bandwidth and storage at significant cost. Uploading only aggregated metrics risks losing critical waveform detail, limiting fault analysis, historical lookback, and model training. There is no easy middle ground.
Fragmented Edge-Cloud Logic
Edge alerting logic and cloud analytics models operate independently with inconsistent definitions, requiring separate rework and validation on each side — making it difficult to close the loop across alerting, analysis, and optimization.
No Unified Data View
Event data, downsampled data, and raw waveforms across transmission, substation, and distribution segments lack unified organization, making cross-regional, cross-device, and cross-voltage analysis nearly impossible.
DolphinDB Solution
Real-Time Edge Processing
A lightweight DolphinDB instance deployed on edge industrial PCs serves as the local real-time compute engine, connecting directly to data streams from acquisition systems. Protocol parsing, real-time downsampling, feature extraction, and anomaly detection are all handled at the edge — shifting latency-sensitive computation closer to the source and ensuring timely, reliable alerting.
Efficient Data Transmission
For each alert, DolphinDB automatically captures the surrounding waveform segments, downsampled curves, and feature metrics into a compact event data package, which is then uploaded to the cloud via secure channels (TCP / MQTT / HTTP / Kafka). This preserves full analytical and traceability capability while significantly reducing bandwidth consumption and cloud storage overhead.
Unified Cloud Analytics
A distributed DolphinDB cluster in the cloud consolidates event data, downsampled data, and key waveform segments from transmission lines, substations, and distribution networks — alongside external data such as dispatch, weather, and distributed resource information — supporting high-concurrency queries and long-term historical analysis.
Scene-Specific Analytics
Built on a unified time-series data foundation, DolphinDB supports targeted analysis across segments:
- Transmission: Ice accretion, thermal, and wind deflection risk assessment, and corridor safety boundary evaluation
- Substation: Equipment condition assessment, defect identification, and degradation trend analysis
- Distribution: Overvoltage analysis, reverse power flow impact assessment, line loss analysis, and fault location and recovery evaluation
Closed-Loop Edge-Cloud Intelligence
Edge and cloud share the same technology stack and scripting language. Alerting rules, feature computation logic, and analytics models can be validated and refined in the cloud using historical and event data, then pushed down to edge nodes for execution — enabling fast, consistent iteration across the entire system.
Key Benefits
Faster Alerting
Low-latency edge computation and vectorized processing enable real-time analysis of high-frequency data locally, ensuring timely alerts and greater control over operational risk across transmission, substation, and distribution networks.
Lower and More Predictable Costs
An event-driven upload strategy with selective waveform retention preserves the data needed for fault analysis and model training, while avoiding full-resolution uploads — striking the optimal balance between bandwidth, storage cost, and analytical depth.
Faster Iteration
A shared technology stack and scripting language across edge and cloud means algorithms are reusable and portable, significantly reducing iteration overhead and accelerating the evolution of intelligent monitoring and O&M capabilities.