Inferring Ingrained Remote Information in AC Power Flows Using Neuromorphic Modality Regime

TL;DR

This work addresses cyber-security and reliability challenges in increasingly digitized, DER-dominated power grids by proposing grid-edge neuromorphic intelligence that infers remote voltages and currents from local measurements. It develops an event-driven SNN framework deployed at each DER edge, using binary spike events to encode measurements, rate encoding for inputs, and PWM-derived decoding to drive converters, all trained with backpropagation to track target modulation. The approach demonstrates effective reactive power sharing, voltage regulation, and fault ride-through adaptation in simulations and a two-bus experimental testbed, while exhibiting superior energy efficiency due to sparse, asynchronous spike processing. By bypassing reliance on the cyber layer and enabling local, online adaptation, the method enhances resilience and scalability for future AC power systems with distributed energy resources.

Abstract

In this paper, we infer remote measurements such as remote voltages and currents online with change in AC power flows using spiking neural network (SNN) as grid-edge technology for efficient coordination of power electronic converters. This work unifies power and information as a means of data normalization using a multi-modal regime in the form of spikes using energy-efficient neuromorphic learning and event-driven asynchronous data collection. Firstly, we organize the synchronous real-valued measurements at each edge and translate them into asynchronous spike-based events to collect sparse data for training of SNN at each edge. Instead of relying on error-dependent supervised data-driven learning theory, we exploit the latency-driven unsupervised Hebbian learning rule to obtain modulation pulses for switching of power electronic converters that can now comprehend grid disturbances locally and adapt their operation without requiring explicit infrastructure for global coordination. Not only does this philosophy block exogenous path arrival for cyber attackers by dismissing the cyber layer, it also entails converter adaptation to system reconfiguration and parameter mismatch issues. We conclude this work by validating its energy-efficient and effective online learning performance under various scenarios in different system sizes, including modified IEEE 14-bus system and under experimental conditions.

Figures (9)

• Figure 1: Structural elements of the proposed modality of power and ingrained remote information: (a) conventional cyber-physical control framework, (b) simplified architecture due to the proposed co-transfer methodology, (c) design stages from sampling to PWM generation using Hebbian learning based weight update policy in SNN at each edge.
• Figure 2: Leaky integrate and fire (LIF) model of SNN: (a) Input and output of the SNN at each DER, where only some LIF neurons are active subject to the frequency of spikes, (b) the neuron model that synthesizes input spikes into selective output spikes based on a voltage threshold $\mathrm{V_{th}}$.
• Figure 3: Test cases for various simulation scenarios in Section III carried out on DER at bus #6 in modified IEEE 14-bus system.
• Figure 4: Coordinated control results using the proposed approach in Fig. \ref{['fig_1']}(c): (a) regression accuracy of the modulation signal, (b) output spikes of SNN, (c) reactive power sharing results, (d) average voltage regulation results.
• Figure 5: Fault ride through (FRT) compliance: (a) voltage instability without FRT control, (b) overcurrent issues without FRT control, (c) voltage and (d) current adaption.