DeviceBoard – Natural Language Query for Insights
DeviceBoard – Documentation
DeviceBoard – Natural Language Query for Insights
DeviceBoard includes an intelligent, AI-driven Natural Language Insights Engine that allows users to ask questions in simple English (or other supported languages) and automatically generate:
- Instant insights
- Dynamic dashboards
- Analytical charts
- Data summaries
- AI-enhanced interpretations
- Auto-generated reports
This feature eliminates the need for users to manually explore data, build charts, or configure dashboards—even if they have no technical or analytical background.
1. Overview of Natural Language Insights
DeviceBoard enables users to ask questions like:
- “Show me the last 24 hours of temperature trends for Plant 1.”
- “Which devices have the highest energy consumption this week?”
- “Generate a dashboard for pump efficiency comparison.”
- “Give me a report of all alarms from Zone A in the last month.”
- “Show anomaly patterns for Machine Group 3.”
Based on the query, DeviceBoard:
- Understands the user’s intent
- Identifies relevant devices, groups, or assets
- Selects required telemetry, attributes, alarms, or AI outputs
- Automatically generates suitable visualizations or reports
- Creates dashboards or documents without manual configuration
This reduces analysis time from hours to seconds.
2. Key Capabilities
2.1 Natural Language Understanding (NLU)
DeviceBoard’s insights engine supports:
✔ Device-aware queries
“Show today’s pressure readings for Compressor-02.”
✔ Group-aware queries
“Compare energy usage across Zone B devices.”
✔ Asset-aware queries
“Give occupancy analytics for Building 7.”
✔ Time-window queries
“What happened in the last 6 hours?”
✔ AI-related queries
“Show anomaly spikes and failure probability for the packaging line.”
✔ Alarm/event queries
“Summarize critical alarms for last week.”
✔ Performance KPI queries
“What is the uptime for all pumps this month?”
The engine parses:
- Devices
- Assets
- Telemetry keys
- Time periods
- Metrics
- Trends
- Aggregations
2.2 Automatic Dashboard Generation
When a user asks for insights requiring visualization, DeviceBoard:
- Creates a new dashboard automatically
- Selects appropriate widget types
- Builds charts (line/bar/gauge/histogram/scatter)
- Organizes layout
- Applies filters
- Adds legend + labels
- Saves dashboard for reuse
Examples:
Query:
“Create a dashboard showing temperature, humidity, and vibration trends for Machine A.”
DeviceBoard automatically generates a visually organized dashboard containing:
- Temperature line chart
- Humidity line chart
- Vibration anomaly scatter plot
- AI health score widget
- Alarm timeline
2.3 Automatic Report Generation
Queries that require summary documents generate full reports:
Types of auto-generated reports:
- Performance reports
- Alarm analysis reports
- Anomaly & AI insights reports
- KPI summaries
- Fleet-level comparison reports
- Energy consumption reports
Examples:
Query:
“Generate a weekly health report for all devices in Plant A.”
Output includes:
- Summary table
- Health scores
- Anomaly detection charts
- Trend analysis
- Top alarms
- Recommendations
Reports can be exported in:
- Excel
- CSV (for raw data sections)
2.4 Insight Panel with AI Explanation
After executing a query, DeviceBoard shows an AI-generated explanation, such as:
- “Device A shows abnormal vibration between 2 PM–3 PM.”
- “Energy consumption increased 12% compared to last week.”
- “Machine X has the highest probability of failure next week.”
- “Zone C alarms are 40% higher than other zones.”
This gives context—not just charts.
2.5 Data Aggregation & Transformation
DeviceBoard can perform:
- Min/max/avg
- Percentiles
- Sum/count
- Trend comparison
- Correlation analysis
- Forecasting
Example Queries:
“Compare average flow rate between Pump Line 1 and Line 2.”
“Show correlation between temperature and vibration for Machine 7.”
DeviceBoard automatically computes the required metrics.
2.6 Multi-Device, Multi-Group, Multi-Asset Insights
Users can ask high-level questions:
- “What are the top 10 devices with highest downtime?”
- “Show me device groups with most alarms.”
- “Which assets consume the most energy?”
DeviceBoard analyzes metadata and telemetry to answer.
2.7 AI + Natural Language Synergy
DeviceBoard leverages:
- Anomaly detection models
- Predictive models
- Health scoring models
Users can ask:
- “Which devices are showing early signs of failure?”
- “Explain why Machine 4 is unhealthy.”
- “Show predicted energy usage for next 7 days.”
2.8 Context Awareness
Natural Language queries are tailored to:
- The user’s permissions
- Assigned device groups
- Assigned asset groups
✔ Data privacy
✔ Correct filtering
✔ No accidental exposure to unauthorized data
3. Supported Query Types
3.1 Telemetry Analytics Queries
Examples:
- “Show temperature last 6 hours.”
- “Graph humidity vs time for Device 33.”
- “What is the peak vibration today?”
3.2 Alarm Queries
Examples:
- “List all major alarms this week.”
- “Show a chart of alarms by severity.”
- “Create an alarm summary dashboard.”
3.3 AI & Anomaly Queries
Examples:
- “Show anomaly score trend for Compressor-01.”
- “List devices with anomaly score > 0.8.”
3.4 Performance & KPI Queries
Examples:
- “What is the uptime for all machines in Line-C?”
- “Compare energy consumption monthly.”
3.5 Device Group / Asset Group Queries
Examples:
- “Summarize performance for Cooling Tower Group.”
- “Show CO2 levels in all buildings.”
3.6 Geolocation & Map Queries
Examples:
- “Show live map of all moving assets.”
- “Show route playback for Vehicle-12.”
3.7 Predictive Queries
Examples:
- “Which devices may fail in the next 30 days?”
- “Predict tomorrow’s energy usage.”
4. Internal Workflow of Natural Language Insights Engine
When a query is submitted:
Step 1 — Language Parsing
Intent detection, entity recognition, device/group identification.
Step 2 — Data Mapping
Match telemetry keys, assets, device groups.
Step 3 — Query Translation
Convert natural language → internal analytics request.
Step 4 — Execution
Retrieve telemetry / alarm / AI data from time-series DB.
Step 5 — Visualization or Report Generation
Create widgets, dashboards, or documents automatically.
Step 6 — AI Explanation (optional)
Provide context-based interpretation.
5. Permissions & Access Control
Natural Language results respect:
✔ RBAC — Module access
✔ ABAC — Operational rights
✔ Device Group assignment
✔ Asset Group assignment
If a user asks a query outside their visibility, DeviceBoard adjusts the insight automatically.
6. Benefits of Natural Language Insights
| Benefit | Description |
|---|---|
| Instant insights | No manual dashboard building required |
| Zero training needed | Anyone can ask questions in natural language |
| Automated dashboards | Saves hours of analysis & configuration |
| Automated reports | Generate professional reports instantly |
| AI-enhanced intelligence | Interprets patterns, anomalies & predictions |
| Accelerated decision-making | Managers and operators get answers immediately |
| Data democratization | Insights accessible to non-technical users |
7. Example End-to-End Scenarios
Scenario 1 – Operator asks:
“Show real-time efficiency of all pumps.”
DeviceBoard creates:
- A dashboard with pump KPIs, efficiency gauges, and trend charts.
Scenario 2 – Manager asks:
“Generate monthly alarm report for Plant B.”
DeviceBoard:
- Aggregates alarms
- Creates a PDF report
- Emails or downloads it
Scenario 3 – Maintenance engineer asks:
“Which machines are likely to fail this week?”
DeviceBoard:
- Runs AI models
- Shows ranked list of devices
- Generates anomaly charts
8. Summary of Natural Language Query Features
✔ Ask questions in natural language
✔ Automatically generate dashboards
✔ Automatically create analytical reports
✔ AI-powered data interpretation
✔ Supports telemetry, alarms, AI, KPIs, geolocation
✔ Role-aware, permission-controlled
✔ Works across cloud and edge deployments
✔ Eliminates manual dashboard configuration
DeviceBoard transforms raw IoT data into actionable intelligence instantly—all through human language.