Power System Transient Stability Assessment Across Similar Failures
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data-driven methods for transient stability assessment in electric power systems face challenges in accurately determining stability under different failure scenarios due to limited data utilization and model performance, as similar data sets are not adequately used for training distinct models.
Innovation Solution
A Multi-Task Learning and Siamese Network approach is employed to cluster data sets based on similarity evaluation indices, such as Jaccard and Hausdorff distances, allowing for parameter-sharing across similar failures, thereby increasing training data and improving model generalization and accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If separate machine learning models are constructed for each predetermined failure using steady-state variables, then the model can capture failure-specific characteristics, but the training data for each model is limited and model performance deteriorates
Solution Approach 1:
The patent merges multiple separate machine learning models into a single multi-task learning framework where models for different failure types share common parameters. This allows training data from multiple failure scenarios to be combined, increasing the effective training data amount while maintaining failure-specific assessment capabilities through task-specific output layers.
Solution Approach 2:
The patent creates a universal multi-task learning model that can perform transient stability assessment for multiple different failure types simultaneously. The shared parameter structure enables the model to learn common patterns across different failures, making the system multi-functional rather than requiring separate specialized models for each failure type.
2Reliability
If data sets under different predetermined failures are used to construct multiple machine learning models, then failure-specific assessment is achieved, but similar data sets cannot be used adequately which reduces productivity
Solution Approach 1:
The patent combines datasets from multiple failure scenarios into a unified training framework. By merging the training processes of multiple separate models into a single multi-task learning process, the system can adequately utilize similar data sets across different failures, improving data utilization efficiency while maintaining assessment accuracy.
Solution Approach 2:
The patent changes the parameter structure from separate independent models to a shared parameter structure with task-specific outputs. This parameter sharing enables the model to learn common features from similar datasets across different failures, improving both data utilization efficiency and assessment reliability.
3Reliability
If multiple separate transient stability assessment models are trained for different failures, then each model can be specialized, but the device complexity increases
Solution Approach 1:
The patent merges multiple separate assessment models into a single multi-task learning model. This consolidation reduces the number of independent models from multiple separate models to one unified model, thereby reducing device complexity while maintaining the capability to assess different failure types through shared parameters and task-specific output layers.
Solution Approach 2:
The patent creates a universal model structure that can handle multiple failure types simultaneously. This multi-functional design eliminates the need for multiple separate specialized models, reducing system complexity while preserving failure-specific assessment capabilities through a single versatile framework.
Data Source
AI summary
A transient stability assessment method for an electric power system is disclosed. Transient stability tags and steady-state data of the electric power system before a failure occurs are collected from transient stability simulation data. Data sets under different predetermined failures are obtained based on a statistical result of the transient stability tags and a maximum-minimum method. A similarity evaluation index between different predetermined failures is constructed based on a Jaccard distance and a Hausdorff distance. Different predetermined failures are clustered based on a clustering algorithm. A parameters-shared siamese neural network is trained for different predetermined failures in each cluster to obtain a multi-task siamese neural network for the transient stability assessment. Transient stability assessment results of the electric power system under all the predetermined failures are obtained based on the statistical result of the transient stability tags and the multi-task siamese neural network for the transient stability assessment.