US20250335223
2025-10-30
Physics
G06F9/45504
The patent application discusses a hardware emulator designed to mimic the dynamics of a biological system using advanced technology. It integrates an artificial neural network (ANN) reconstruction model with a memristor-based circuit to emulate state space representations of dynamic systems. This combination allows for the emulation of complex system behaviors based on input data, reflecting the intricate operations of natural neuronal networks.
The hardware emulator consists of two main components: the ANN-based reconstruction model and the memristor-based circuit. The reconstruction model processes input data to recreate the dynamics of the system, while the memristor circuit emulates these dynamics in a physical form. The emulator can reflect features of dynamic systems by approximating differential equations or hidden states within the ANN, allowing for real-time adjustments and fine-tuning to maintain fidelity in emulation.
The emulation system includes a control circuit with programmable electronic components to adjust and replicate neural data operations. This system fine-tunes the memristor-based emulator by modifying control elements until desired emulation fidelity is achieved. The system can be integrated into various devices, such as wafer monitoring tools, video and audio synthesis devices, robots, home appliances, and communication devices.
The operating method involves using the ANN-based reconstruction model to process input data and reconstruct dynamic systems. The memristor-based circuit then emulates these systems' state space representation. This process includes iterative fine-tuning using normalized differential equations and chaotic attractors for precise emulation, ensuring the hardware emulator accurately reflects dynamic behaviors.
The described hardware emulator can be adapted for various technological applications, including personal computers, smart devices, intelligent vehicles, and wearable technology. Its design allows for modifications and enhancements to fit different use cases while maintaining core functionalities. The flexibility in its design ensures it can evolve with advancements in neuromorphic engineering and related fields.