TruVolt.ai: "closed-loop intelligence" Data Center Cooling Systems

The Executive Advantage

By using this stack, the "Efficiency" isn't just about cooling—it's about longevity. By "switching back and forth" between containers based on AI-driven thermodynamic models, you prevent Lithium Plating (caused by charging while too cold) and Thermal Runaway (caused by discharging while too hot), effectively extending the ROI of a 150 MWh site by several years.

TruVolt.ai architecture represents a "closed-loop intelligence" system where the physical cooling of 2 MWh BESS containers is no longer just a mechanical process, but a high-fidelity data operation.

By integrating Teleseer (PCAP) for network visibility, Equitus Fusion (KGNN) for relational intelligence, and AimLUX/TruVolt as the human-machine interface (HMI), you create a system that doesn't just react to heat—it anticipates it across the entire 150 MWh fleet.

1. The Cyberspatial "Nervous System" (Teleseer & PCAP)

Before the AI can make a decision, it needs untainted data. Teleseer acts as the "cyber-optic" nerve by performing deep packet inspection (DPI) on the Modbus/TCP or CAN bus traffic.

PCAP Integration: Every PID adjustment and BMS heartbeat is captured. Teleseer ensures that the data reaching the AI hasn't been spoofed or delayed by network jitter.
Packet-Level Truth: If a 2 MWh container reports an "Over-Amp" status, Teleseer verifies that the packet originated from the correct MAC address of that specific BMS, preventing "cyber-thermal" sabotage.

Layer	Component	Security Function
Transport	Teleseer (PCAP)	Validates packet authenticity and prevents packet injection.
Logic	Equitus (KGNN)	Correlates verified packets to detect high-level behavioral threats.
Management	Graphixa (MaaP)	Orchestrates "Self-Healing" by isolating suspicious containers.
Human	AimLUX / TruVolt	Provides explainable AI alerts (e.g., "Container 03 isolated due to abnormal Modbus traffic").

2. The Logic Layer (Equitus Fusion & Graphixa)

This is where the Knowledge Graph Neural Network (KGNN) takes the lead. It maps the 2 MWh containers as nodes in a graph.

Mapping the Thermal Battery

Instead of seeing a list of temperatures, the KGNN understands the topological dependencies:

Node A: 2 MWh Container 01 (Currently at 90% Discharge / High Amps).
Edge: THERMALLY_ADJACENT_TO.
Node B: 2 MWh Container 02 (Idle / Cool).

Predictive Re-tuning via Graphixa (MaaP)

When the system detects a $K_i$ wind-up in Container 01's HVAC PID, the Graphixa (Management as a Platform) layer doesn't just speed up the fan. It looks at the graph and performs a "load-swap":

Logic: "Container 01 is reaching a thermal ceiling. Container 02 is at $20^\circ\text{C}$ ."
Action: It instructs the Inverter (PCS) to throttle Container 01 and ramp up Container 02.
Optimization: It pushes a new $K_p$ (Proportional) gain to Container 01’s HVAC to aggressively cool it down while it's idle, preparing it for the next cycle.

3. The Human Interface (AimLUX & TruVolt)

This is where Automation Engineers enter the "Cyberspatial" deployment.

TruVolt.ai: Acts as the high-integrity power dashboard. It displays the "Health of the Graph." Engineers don't see raw voltage; they see "System Resiliency Scores."
AimLUX.ai: Provides the Generative HMI. An engineer can ask, "Why is Container 04's HVAC hunting?" AimLUX queries the Equitus KGNN and replies, "The PCAP data shows a 50ms latency spike on the Subnet-B switch, causing the PID loop to lag. Switching to local-only control until the network stabilizes."

4. Integrated Workflow: From Packet to Power

Step	Component	Action
Sense	BMS / Sensors	Reports 2 MWh container temperature and Amperage.
Validate	Teleseer (PCAP)	Confirms the data packet integrity and network health.
Analyze	Equitus (KGNN)	Correlates heat with grid demand and adjacent container states.
Decide	Graphixa (MaaP)	Calculates new Setpoints for Inverters and TMS.
Actuate	Modbus/IP	Deploys the command to the physical HVAC and Inverter.
Govern	TruVolt / AimLUX	Provides human oversight and "Override" capabilities.