Transformer Failure Diagnosis Using Integrated Deep Belief Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
A single deep belief network is inadequate for effectively diagnosing various types of transformer failures, as it tends to prefer certain types of transformers over others during failure diagnosis.
Innovation Solution
An integrated deep belief network is created by training multiple deep belief networks with different learning rates, using vibration signal features obtained through Fourier transform and normalization, and selecting target networks based on high failure diagnosis correct rates to build an integrated model for comprehensive transformer failure diagnosis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a single deep belief network is used for transformer failure diagnosis, then the device complexity is reduced, but the diagnosis accuracy and reliability deteriorate because the network cannot effectively diagnose all types of transformer failures
Solution Approach 1:
The patent divides the single deep belief network into multiple specialized deep belief networks, where each network is trained to diagnose specific types of transformer failures. This segmentation allows each network to specialize in particular failure modes, improving overall diagnosis accuracy while maintaining manageable complexity through modular architecture.
Solution Approach 2:
The patent combines multiple deep belief networks into an integrated diagnosis system where each network's output is aggregated to form the final diagnosis result. This merging approach leverages the strengths of individual networks for different failure types, achieving high comprehensive accuracy without requiring any single network to handle all failure modes alone.
2Measurement precision
If multiple deep belief networks with different learning rates are trained, then the diagnosis accuracy improves, but the training time and computational resources increase
Solution Approach 1:
The patent varies the learning rate parameter across different deep belief networks, training each network with a specific learning rate optimized for its particular failure type. This parameter diversification allows parallel training of multiple networks without sequential overhead, improving diagnosis accuracy while managing training time through concurrent processing.
3Adaptability or versatility
If an integrated deep belief network is built by combining multiple target networks, then the versatility and adaptability improve for diagnosing various transformer types, but the system complexity increases
Solution Approach 1:
The patent creates an integrated deep belief network system that functions universally across multiple transformer types and failure modes. Each component network maintains its specialized function while the integration layer provides universal diagnostic capability, allowing the system to adapt to various transformer configurations without requiring complete redesign for each application.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The integrated deep belief network achieves a higher correct rate in diagnosing transformer failures compared to single deep belief networks, effectively addressing the limitations of single network diagnostics and improving overall diagnosis accuracy.
Implementation Method 1
A Fourier transform is performed on each of the vibration signals to obtain a Fourier coefficient of each of the vibration signals
Data Source
AI summary
A transformer failure diagnosis method and system based on an integrated deep belief network are provided. The disclosure relates to the fields of electronic circuit engineering and computer vision. The method includes the following: obtaining a plurality of vibration signals of transformers of various types exhibiting different failure types, retrieving a feature of each of the vibration signals, and establishing training data through the retrieved features; training a plurality of deep belief networks exhibiting different learning rates through the training data and obtaining a failure diagnosis correct rate of each of the deep belief networks; and keeping target deep belief networks corresponding to the failure diagnosis correct rates that satisfy requirements, building an integrated deep belief network through each of the target deep belief networks, and performing a failure diagnosis on the transformers through the integrated deep belief network.