Photovoltaic Array Fault Diagnosis Using Composite Data Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current fault diagnosis methods for photovoltaic arrays are limited by insufficient comprehensiveness and low accuracy, as they rely on single-type information such as infrared images or electrical characteristics, failing to effectively identify fault types and locations.
Innovation Solution
A method that combines image data and text data using deep convolutional neural networks and support vector machines, respectively, and fuses them with a logistic regression algorithm to create a comprehensive fault diagnosis model, enhancing the robustness and accuracy of fault classification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If single-type information (infrared images or electrical characteristics) is used for fault diagnosis, then the diagnosis method is simple, but the diagnosis accuracy and comprehensiveness are insufficient
Solution Approach 1:
The patent combines multiple types of information (infrared images, electrical characteristics including voltage and current, and environmental data) into a unified fault diagnosis system. The fusion model integrates features from different data types to achieve comprehensive fault identification, resolving the contradiction between method simplicity and diagnosis accuracy by merging multiple information sources into a single integrated system.
Solution Approach 2:
The patent uses composite information fusion, analogous to composite materials, where multiple types of data (thermal images, electrical parameters, environmental conditions) are combined to create a more robust and accurate diagnosis system. This composite approach leverages the strengths of each data type to overcome the limitations of single-type information.
2Ease of operation
If infrared image detection method is used, then fault detection is easy and temperature difference characteristics can be reflected, but fault types cannot be identified
Solution Approach 1:
The patent merges infrared image data with electrical characteristic data (voltage, current) and environmental data to create a multi-dimensional diagnosis system. This combination preserves the ease of infrared detection while adding the capability to identify specific fault types through complementary information from other data sources.
Solution Approach 2:
The patent adds additional dimensions of information (electrical characteristics, environmental data) to the infrared image data. This dimensional expansion transforms the diagnosis system from single-type detection to multi-type identification, enabling both ease of operation and comprehensive fault type recognition.
3Measurement precision
If electrical characteristic detection method is used, then fault diagnosis can be realized through voltage and current analysis, but a large number of sensors are required which limits application
Solution Approach 1:
The patent creates a universal diagnosis system that processes multiple types of information (infrared, electrical, environmental) through a single fusion model. This multi-functional approach reduces the need for separate dedicated systems for each data type, thereby reducing overall device complexity while maintaining comprehensive fault diagnosis capability.
Solution Approach 2:
The patent combines multiple data sources and processing functions into a unified fusion model, reducing the need for separate sensor systems and processing units. This integration lowers device complexity by consolidating functionality while preserving the diagnostic precision of electrical characteristic analysis.
4Reliability
If time-domain reflection method is used, then fault detection can be achieved by comparing input and reflected signals, but fault location precision is limited
Solution Approach 1:
The patent merges time-domain reflection data with infrared image data and electrical characteristic data to improve fault location precision. The fusion model cross-validates fault locations across multiple data types, achieving more precise localization than any single method alone while maintaining the reliability of fault detection.
Data Source
AI summary
A photovoltaic array fault diagnosis method based on composite information is provided. The method includes: collecting and preprocessing composite information data of photovoltaic array working state, including image data and text data; using the image data of photovoltaic array working state to train a pre-established fault classification model of deep convolutional neural network, to thereby obtain an image fault classification model; using the text data of photovoltaic array working state to train a pre-established fault classification model based on a support vector machine, to thereby obtain a text fault classification model; fusing the image fault classification model and the text fault classification model by logistic regression algorithm to obtain a fusion model, and training the fusion model using the composite information data of photovoltaic array working state to thereby obtain the photovoltaic array fault diagnosis model.