Indicator State Classification for Power Grid Voltage and Reactive Power Scheduling

5.5 KB of experimental results from a multi-agent reinforcement learning method applied to the IEEE 33-node power grid system. The data, authored by Zhida Lin and last updated in April 2026, likely contains performance metrics for a proposed hierarchical scheduling framework. Results include a voltage qualification rate of 98.7% and a 30.5% reduction in system power loss.

Use Cases

• Benchmarking new reactive power scheduling algorithms based on the reported MAE prediction error of < 0.01 p.u.
• Analyzing the impact of sensitivity-driven feedback regulation on voltage deviation, based on the reduction from 1.679 p.u. to 1.589 p.u.
• Evaluating decision-time improvements in high-dimensional action spaces based on the reported 34.2% reduction in average scheduling decision time.

Strengths

• Results are validated on the standard IEEE 33-node test system, a common benchmark in power engineering.
• Performance metrics are quantitatively reported, including a voltage qualification rate of 98.7% and a 30.5% reduction in system power loss.
• The underlying method achieved a high-precision prediction error (MAE < 0.01 p.u.) for reactive power and voltage states.

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• The 5.5 KB file size suggests a very limited scope, likely containing summary results rather than raw operational data.

Provenance

Source

figshare, author Zhida Lin

Collection Method

Experimental results from a proposed multi-agent reinforcement learning framework applied to a simulated power grid.

Time Range

Simulation timeframe is unspecified.

Freshness

Last updated 2026-04-24 17:29:07; freshness should be verified.

Geography

Simulation based on the IEEE 33-node system, a standard test case not tied to a specific geographic region.