Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

On Noise Resiliency of Neuromorphic Inferential Communication in Microgrids

Yubo Song, Subham Sahoo, Xiaoguang Diao

TL;DR

The paper addresses the challenge of reliable microgrid coordination under signaling noise by leveraging neuromorphic inferential communication via spiking neural networks (SNNs). It models DC microgrid control with Leaky Integrate-and-Fire (LIF) neurons and Spike Response Model (SRM), enabling event-driven inference where remote states are gleaned from locally sensed data. Through simulations on a two-bus DC microgrid and experimental validation on a down-scaled setup, it demonstrates that the neuromorphic approach can maintain estimation accuracy under noise, while highlighting techniques such as adaptive event thresholds and intrinsic LIF filtering to enhance robustness. The work points to a path toward energy-efficient, noise-tolerant edge computing for secure and resilient microgrid operation, with future extensions in frequency-domain analysis and Hebbian learning robustness.

Abstract

Neuromorphic computing leveraging spiking neural network has emerged as a promising solution to tackle the security and reliability challenges with the conventional cyber-physical infrastructure of microgrids. Its event-driven paradigm facilitates promising prospect in resilient and energy-efficient coordination among power electronic converters. However, different from biological neurons that are focused in the literature, microgrids exhibit distinct architectures and features, implying potentially diverse adaptability in its capabilities to dismiss information transfer, which remains largely unrevealed. One of the biggest drawbacks in the information transfer theory is the impact of noise in the signaling accuracy. Hence, this article hereby explores the noise resiliency of neuromorphic inferential communication in microgrids through case studies and underlines potential challenges and solutions as extensions beyond the results, thus offering insights for its implementation in real-world scenarios.

On Noise Resiliency of Neuromorphic Inferential Communication in Microgrids

TL;DR

The paper addresses the challenge of reliable microgrid coordination under signaling noise by leveraging neuromorphic inferential communication via spiking neural networks (SNNs). It models DC microgrid control with Leaky Integrate-and-Fire (LIF) neurons and Spike Response Model (SRM), enabling event-driven inference where remote states are gleaned from locally sensed data. Through simulations on a two-bus DC microgrid and experimental validation on a down-scaled setup, it demonstrates that the neuromorphic approach can maintain estimation accuracy under noise, while highlighting techniques such as adaptive event thresholds and intrinsic LIF filtering to enhance robustness. The work points to a path toward energy-efficient, noise-tolerant edge computing for secure and resilient microgrid operation, with future extensions in frequency-domain analysis and Hebbian learning robustness.

Abstract

Neuromorphic computing leveraging spiking neural network has emerged as a promising solution to tackle the security and reliability challenges with the conventional cyber-physical infrastructure of microgrids. Its event-driven paradigm facilitates promising prospect in resilient and energy-efficient coordination among power electronic converters. However, different from biological neurons that are focused in the literature, microgrids exhibit distinct architectures and features, implying potentially diverse adaptability in its capabilities to dismiss information transfer, which remains largely unrevealed. One of the biggest drawbacks in the information transfer theory is the impact of noise in the signaling accuracy. Hence, this article hereby explores the noise resiliency of neuromorphic inferential communication in microgrids through case studies and underlines potential challenges and solutions as extensions beyond the results, thus offering insights for its implementation in real-world scenarios.
Paper Structure (16 sections, 18 equations, 11 figures)

This paper contains 16 sections, 18 equations, 11 figures.

Table of Contents

  1. Introduction
  2. Neuromorphic Inferential Communication in Microgrids
  3. Neuromorphic Algorithms and Spiking Neural Network
  4. Neuromorphic Inferential Communication in Microgrids
  5. Performance Validation
  6. Validation of System Performance by Simulation
  7. Validation of Noise Resiliency in Training and Execution Process with Simulation Data
  8. Validation of Spike Synthesis with Experimental Data
  9. Further Perspectives on the Noise Resiliency
  10. Low Pass Filter in the LIF neuron model
  11. Noise modeling for LIF neurons
  12. Bandwidth of controllers
  13. Impacts on Hebbian learning
  14. Conclusions and Future Scope of Work
  15. SNN Parameters -- Simulation Studies
  16. ...and 1 more sections

Figures (11)

  • Figure 1: Concept of the the Leaky Integrate-and-Fire (LIF) neuron model: (a) equivalent RC circuit, and (b) input and output spikes driven by the model, where the $V_\mathrm{th}$ based criteria is employed.
  • Figure 2: Spike response model (SRM) for simulation of the leaky and integrate fire (LIF) neuron.
  • Figure 3: Architecture of a neuromorphic inferential communication in DC microgrid: (a) infrastructure of SNN based control framework for converter $k$, and (b) flowchart of neuromorphic inferential state estimation.
  • Figure 4: A two-bus DC microgrid as the test case.
  • Figure 5: Performance of microgrid governed by neuromorphic controllers: comparison of (a) estimated and measured currents, (b) estimated and measured voltages, and (c) estimated and measured powers.
  • ...and 6 more figures