Virtual Prototyping and Distributed Control for Solar Array with Distributed Multilevel Inverter

Luan Viet Nguyen, Taylor T. JohnsonarXiv 2014 · OpenReview

TL;DR

This paper introduces a novel virtual prototyping framework for a distributed solar array grid-tie system. The system uses a decentralized control strategy, enabling fault tolerance by allowing individual solar panel inverter modules to fail while the overall grid connection remains operational with acceptable performance.

What Problem Does This Paper Address?

Traditional solar power systems often rely on centralized inverters. When these fail, the entire system goes offline. This paper addresses the challenge of creating a more robust and fault-tolerant solar grid-tie system. The authors propose a distributed architecture where each solar panel is paired with its own inverter module and microcontroller. This modularity, combined with a distributed control strategy, aims to ensure that the failure of individual components or modules does not lead to a complete system shutdown. The goal is to maintain grid connectivity and power output even when some parts of the system fail.

Methods

The core of the paper lies in its virtual prototyping approach using Simulink/Stateflow. The system is modeled as a collection of "agents," where each agent represents a solar panel, its buck-boost converter (for voltage regulation), and an H-bridge inverter (for grid connection), all managed by a dedicated microcontroller.

Key methodological contributions include:

  • Distributed Control Architecture: Each agent operates semi-autonomously but communicates with its neighbors to coordinate its actions. This avoids a single point of failure in the control system.
  • Distributed Identifier Algorithm: A novel algorithm allows each operational agent to dynamically determine the number of functioning agents in the array. This is crucial for regulating the DC voltage output of the buck-boost converters and for calculating the appropriate switching times for the H-bridges to generate the AC waveform.
  • Hybrid Automata Modeling: Both the buck-boost converters and H-bridges are modeled using hybrid automata. This formalism allows for the representation of systems with both continuous dynamics (voltages, currents) and discrete events (switching, failures).
  • Fault Tolerance Abstraction: The paper models failures as "crash faults" for microcontrollers and "stuck-at" errors for switches. The system is designed to tolerate these failures through software-based solutions, rather than relying solely on redundant hardware.

Datasets

Not discussed in the paper. The authors generated their own simulation models and parameters for evaluation.

Key Insights and Ideas

  • Decentralization is Key to Resilience: By distributing control and power electronics across individual solar panel modules, the system inherently gains fault tolerance. The failure of one module doesn't cascade to the entire system.
  • Dynamic Failure Detection and Adaptation: The distributed identifier algorithm is critical. It allows the system to adapt its control strategy (voltage regulation and switching times) in real-time based on the number of currently operational modules, effectively compensating for failures.
  • Software-Based Fault Tolerance: The paper emphasizes that fault tolerance can be achieved through intelligent software and distributed algorithms, complementing traditional hardware redundancy. This opens up possibilities for more flexible and cost-effective resilient systems.
  • Achievable THD with Failures: Even with a significant number of failed modules, the system can still achieve a total harmonic distortion (THD) low enough for grid connection, especially in larger arrays. This suggests that a substantial portion of the array can remain operational and useful.

Key Results

The paper presents simulation results demonstrating the system's performance under various failure scenarios.

  • Static Failures: For an array of 10 panels, the THD was around 5%. With 5 static failures in a 15-panel array, the THD remained around 5%. For larger arrays (e.g., 16+ functioning panels), THD could be as low as 2.5%, indicating grid-tie feasibility even with failures.
  • Dynamic Failures: When a single agent fails dynamically (while the system is operational), the system recovers quickly (within one grid period). The THD returns to the level expected for an array with one fewer functioning agent. The paper suggests that simultaneous dynamic failures are unlikely to occur before recovery.
  • Scalability: The virtual prototyping framework was used to evaluate arrays from 5 to dozens of panels, demonstrating its scalability.

Limitations and Future Work

  • Simulation-Based: All results are based on simulations. The authors acknowledge the need for a physical prototype to validate these findings in real-world scenarios.
  • Simplified Failure Model: The failure model is an abstraction. While it covers crash faults and stuck-at errors, more complex failure modes might require further investigation.
  • Future Work: The authors plan to:
  • Construct an actual prototype to evaluate fault-tolerance in real-world conditions.
  • Implement a switching scheme to vary H-bridge identifiers periodically to reduce wear.
  • Formally verify critical specifications of the H-bridge control algorithm using timed automata, ensuring safety properties like preventing modules with opposite polarities from being connected simultaneously.

Summary updated

Abstract

In this paper, we present the virtual prototyping of a solar array with a grid-tie implemented as a distributed inverter and controlled using distributed algorithms. Due to the distributed control and inherent redundancy in the array composed of many panels and inverter modules, the virtual prototype exhibits fault-tolerance capabilities. The distributed identifier algorithm allows the system to keep track of the number of operating panels to appropriately regulate the DC voltage output of the panels using buck-boost converters, and determine appropriate switching times for H-bridges in the grid-tie. We evaluate the distributed inverter, its control strategy, and fault-tolerance through simulation in Simulink/Stateflow. Our virtual prototyping framework allows for generating arrays and grid-ties consisting of many panels, and we evaluate arrays of five to dozens of panels. Our analysis suggests the achievable total harmonic distortion (THD) of the system may allow for operating the array in spite of failures of the power electronics, control software, and other subcomponents.
BibTeX
@misc{nguyen2014virtual,
  author = {Luan Viet Nguyen and Taylor T. Johnson},
  title = {{Virtual Prototyping and Distributed Control for Solar Array with Distributed Multilevel Inverter}},
  year = {2014},
  eprint = {1404.2259},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1404.2259},
}
Figure