Data acquisition and analysis software Industrial Sensing Monitoring

calibrationdata archivingdata logger

What if your industrial sensors could stay vendor-independent and automatically build your next AI dataset?
Banalytics acquires raw data from cameras, DAQ hardware, and process instruments, hands it to your own processing or analysis module over a locked interface, and turns the structured results back into monitoring dashboards, alarms, and historian records. Your algorithm never leaves the module boundary; only structured outputs, health status, and timestamps cross it.
Banalytics provides operational visibility across your sensing pipeline - from device connectivity and data freshness to processing readiness, result delivery and edge-node health.

Problem this solves
Research and DAQ teams need to know their sensing pipeline is healthy: devices connected, data flowing, processing module alive, without building device connectivity and health monitoring themselves, and without exposing their proprietary algorithm to a third-party platform.

How it works
Field devices (cameras, DAQ hardware, sensors) → Banalytics acquisition (buffering, timestamp sync, local storage) → your processing module (over ZeroMQ IPC/TCP, shared memory, or in-process) → Banalytics orchestration (events, dashboards, historian, MQTT) → external consumers (SCADA, maintenance systems, downstream services).

Architecture example

• Site A (Test Bench): 4 acoustic-emission sensors + DAQ hardware, ZeroMQ IPC to an in-process analysis module.
• Site B (Production Line): 6 vibration sensors, Modbus RTU, Python analysis module over ZeroMQ TCP.
• Site C (Remote Pump Station): 2 ONVIF cameras + pressure transducers, cloud-hosted module over TCP.
• All sites → Banalytics Dashboard · Event History · MQTT to SCADA.

Who this is for

• Research, DAQ, and measurement teams;
• instrumentation engineers;
• plant reliability teams who need operational visibility into a sensing pipeline without exposing their processing algorithm.

Key features

• Acquisition buffering and timestamp synchronization across devices.
• Transport-tiered architecture (ZeroMQ IPC/TCP, shared memory, in-process) matched to throughput and latency needs.
• Event Manager and Actions for alarms and rule-based responses.
• Cesium/Mapbox/OpenStreetMap/SVG/3D dashboards for monitoring views.
• Event History as a queryable, timestamped historian.
• MQTT publish for downstream integration.
• Locked module-interface contract that keeps your processing algorithm private. Vendor-independent acquisition across ONVIF, RTSP, Modbus, and USB devices.

Technical specifications

• Supported protocols: ONVIF, RTSP, Modbus (RTU/TCP), MQTT (v3/v5), ZeroMQ (IPC/TCP), USB, OPC UA, I2C
• Deployment: runs alongside your existing processing or analysis module via a locked interface contract; no mandatory cloud migration; acquisition and storage remain local.
• Module interface: input/output categories, timestamps, and health tags are defined per engagement; your algorithm never leaves your own module.
• Alerts: dashboard alarms and notifications driven by device, health, and processing status. Historian: Event History provides a queryable, timestamped record of acquisition and processing events.
• Scalability: transport tier (in-process, shared memory, ZeroMQ IPC, ZeroMQ TCP) is selected to match throughput and latency needs, from a single test bench to a multi-site deployment.

Way of work

• Phase 0: Technical demo (free). Architecture walkthrough using your own use case, no data required.
• Phase 1: PoC (always free). Validate the module-interface contract, health tags, and dashboards/alarms/event history using mock or replayed data; no real hardware, no device-specific engineering.
• Phase 2: Funded Pilot. Connect real devices at one site and validate acquisition, processing, and monitoring on live data. Commercial once real hardware is connected; scope and price depend on site/device count and integration complexity.
• Phase 3: Production. Harden and scale to additional lines or sites; ongoing support agreed separately.