AI with Model-Based Design: Virtual Sensor Modeling

2026-05-26 · Source: IEEE Spectrum · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Robotics & Autonomous Systems, Software Development & Engineering · Depth: Advanced, quick

Summary

This webinar explores the application of AI-based virtual sensors for estimating signals that are challenging or expensive to measure, such as battery state of charge (SOC) in Battery Management Systems. It demonstrates how these AI models can be integrated into system-level designs and validated against performance, resource, and deployment constraints. The session outlines a comprehensive workflow covering the design, verification, compression, and deployment of AI-based virtual sensors to embedded processors, all within a unified environment. Key topics include integrating AI models into Simulink® for simulation, applying formal verification to neural networks, optimizing models for memory and speed, and generating library-free C code for deployment.

Key takeaway

For embedded systems engineers developing Battery Management Systems, understanding AI-based virtual sensor integration is crucial. You should explore methods for formally verifying neural network behavior and optimizing models for memory footprint and execution speed on embedded processors. This approach enables efficient estimation of complex signals like battery SOC, reducing reliance on costly physical sensors and streamlining deployment processes.

Key insights

AI-based virtual sensors efficiently estimate hard-to-measure signals, integrating into system-level designs for embedded deployment.

Principles

• Integrate AI models into system-level design.
• Formally verify neural network behavior.
• Optimize models for embedded constraints.

Method

The workflow involves designing, verifying, compressing, and deploying AI-based virtual sensors to embedded processors within a single environment.

In practice

• Integrate AI models into Simulink®.
• Generate library-free C code.
• Optimize models for memory and speed.

Topics

• AI-based Virtual Sensors
• Battery Management Systems
• Simulink
• Embedded Systems
• Model Optimization
• Formal Verification

Best for: Machine Learning Engineer, AI Engineer, AI Hardware Engineer

Related on AIssential

• Sift raises $42 million in Series B round — Sift structures raw hardware sensor data for AI-controlled systems, enabling fluent interaction between AI and physical devices.
• Explainable Data-driven Deep Reinforcement Learning Methods for Optimal Energy Management in Buildings — Explainable DRL (XRL) optimizes building energy management, reducing costs and providing transparent decision insights, outperforming black-box DRL.
• From Vision-Language Models to Physical AI: Embedded Intelligence Enters a New Phase — Embedded AI is evolving towards practical, multimodal intelligence at the edge, requiring innovation in models and deployment.
• ChatSVA: Bridging SVA Generation for Hardware Verification via Task-Specific LLMs — ChatSVA significantly improves SVA generation by decomposing reasoning and synthesizing high-purity, domain-specific data for LLMs.
• A faster way to estimate AI power consumption — EnergAIzer rapidly and accurately estimates AI workload power consumption on GPUs, improving data center efficiency.
• VBFDD-Agent for Electric Vehicle Battery Fault Detection and Diagnosis: Descriptive Text Modeling of Battery Digital Signals — VBFDD-Agent uses descriptive text modeling and LLM reasoning for interpretable EV battery fault diagnosis.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by IEEE Spectrum.