PV Inverter Clustering for Fast Accurate Power Station Simulation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing photovoltaic power station simulation methods face challenges in achieving high accuracy while maintaining simulation speed due to the use of detailed models with high complexity and single-machine models with introduced errors.
Innovation Solution
A simulation method that involves collecting output data from inverters in steady-state and transient-state processes, clustering inverters based on this data, and constructing equivalent models for clusters to increase accuracy and speed.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If detailed models are established for each power generation unit, then simulation accuracy is improved, but simulation time increases excessively
Solution Approach 1:
The patent merges multiple detailed inverter models into a single equivalent inverter model by collecting output data from all inverters, performing clustering analysis to identify representative characteristics, and constructing an equivalent model that replicates the aggregate behavior. This combining approach maintains simulation accuracy while dramatically reducing computational time by eliminating the need to simulate each inverter individually.
Solution Approach 2:
The patent creates a simplified copy (equivalent model) that replicates the essential characteristics of the complex system. By using clustering to identify representative patterns from detailed inverter data, the equivalent model serves as a computationally efficient copy that preserves accuracy for stability and fault analysis without requiring the full complexity of individual inverter models.
2Productivity
If a single-machine model is used to represent all power generation units, then simulation speed is improved, but simulation accuracy deteriorates due to introduced errors
Solution Approach 1:
The patent applies local quality by making the equivalent inverter model adaptive rather than uniform. Through clustering analysis of inverter output data, the model captures local variations and characteristics of different inverter groups. The equivalent model is constructed to reflect the specific operational patterns and electrical characteristics of the actual inverter fleet, providing locally accurate representation rather than a generic single-machine approximation.
Solution Approach 2:
The patent utilizes parameter changes by adjusting the equivalent inverter's electrical parameters (such as output impedance, power output, and response characteristics) based on clustering results from actual inverter data. This allows the equivalent model to dynamically reflect the aggregate behavior of multiple inverters under different operating conditions, maintaining accuracy while preserving simulation speed.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Embodiments of this application provide a simulation method for a photovoltaic power station. In a data collection phase, all data of all inverters in steady-state and transient-state processes is collected; then clustering is performed on all the inverters based on the data and line impedances to obtain a plurality of clusters; equivalent models are respectively constructed for the plurality of clusters; and an internal algorithm parameter of the clustering is adjusted, so that an error of equivalent models obtained by using the adjusted clustering is less than a specified value. The clustering of all the inverters and the construction of the equivalent models can increase the simulation speed compared with a detailed model, and can increase the simulation accuracy compared with a single-machine model. In the clustering of all the inverters, the use of all the data in the steady-state and transient-state processes for the clustering leads to an increase in the overall accuracy of the finally simulated equivalent models for steady-state and transient-state fitting.