Comprehensive optimization capabilities for modern chiller plants
Probabilistic deep learning models predict chiller COP with uncertainty quantification. Provides confidence intervals for every prediction enabling intelligent fallback decisions.
Gradient boosting decision tree models predict cooling loads using weather forecasts, temporal patterns, and building-specific features with <10% MAPE.
When prediction uncertainty exceeds threshold, system automatically switches to physics-based chiller models ensuring robust operation under all conditions.
Optimal chiller sequencing algorithm that reduces computation time from minutes to seconds by exploiting problem structure and eliminating redundant evaluations.
EWMA-based bias correction adapts to changing conditions in real-time, continuously improving prediction accuracy without model retraining.
Real-time tracking of prediction accuracy, model confidence (sigma), and fallback rates. Automated alerts when model performance degrades below threshold.
Smart Cool Tech leverages Brick Schema, a standardized semantic metadata framework for building systems. This enables:
Beyond automated optimization, our platform features a state-of-the-art LLM Assistant that enables facility managers to interact with building data using natural language.
"Why was the power consumption high yesterday at 3 PM?" - Get instant, data-backed answers.
Generate daily performance summaries or executive energy reports in seconds.
AI identifies anomalies and cross-references them with maintenance history and Brick Schema relations.
Action recommendations front and center. Clear chiller status indicators. One-click access to detailed performance metrics.
4-layer architecture: Global overview, Building details, Chiller specifics, and Deep analysis. Intuitive breadcrumb navigation.
Role-based access control (Viewer/Operator/Admin). Multi-instance deployment for complete data isolation between tenants.
Interactive charts (gauge, radar, sankey, heatmap). Real-time updates. Light/Dark mode support with Design Tokens system.
High-performance time-series database for real-time monitoring. Efficient storage and retrieval of operational data.
Deep dive into chiller performance trends. Compare actual vs predicted. Export data for further analysis or reporting.
Automated alerts for CHWS temperature violations, model accuracy degradation, and system anomalies. Multi-level severity (Info/Warning/Critical).
Track key indicators: energy savings, COP, PLR, prediction accuracy (CVRMSE, MAPE). Compare baseline vs optimized operation.
Log in to the dashboard to explore all capabilities and see real-time optimization in action.
Access Dashboard