Multi-View Feature Regression for Industrial Fault Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing fault detection technologies in industrial systems are inadequate for comprehensively diagnosing complex production processes due to limitations in handling multi-source heterogeneous data and inconsistencies in data categories.
Innovation Solution
A method for detecting abnormal working conditions based on feature regression is developed, which involves acquiring and preprocessing multi-view data from industrial processes, establishing a collaborative modeling algorithm, and using an objective function model to identify abnormal conditions through projection vectors and regression centers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If multi-source heterogeneous data is used for comprehensive fault detection, then detection comprehensiveness is improved, but data inconsistency and modeling difficulty increase
Solution Approach 1:
The patent transforms multi-source heterogeneous data into a unified representation by changing the parameter space. It projects data from different views (image, signal, text) into a common feature space using autoencoders and attention mechanisms, converting inconsistent data categories into comparable parameter forms that can be jointly analyzed for comprehensive fault detection
Solution Approach 2:
The patent introduces an intermediary attention mechanism that mediates between multi-source heterogeneous data and the final fault detection output. This attention mechanism weights and integrates features from different data views, acting as a bridge that reconciles data inconsistencies while preserving the comprehensive diagnostic information from all sources
2Stability of the object's composition
If single data source is used for fault detection, then data consistency is maintained, but detection comprehensiveness deteriorates
Solution Approach 1:
The patent creates a universal multi-view detection framework that can process multiple types of data (image, signal, text) through a single integrated system. Each data view is processed by dedicated sub-networks that output to a shared attention mechanism, allowing the system to maintain data-specific characteristics while achieving comprehensive multi-functional detection capabilities
3Ease of operation
If traditional data processing methods are used, then processing simplicity is maintained, but fault detection accuracy deteriorates
Solution Approach 1:
The patent segments the fault detection process into distinct functional modules: data acquisition from multiple views, separate preprocessing pipelines for each data type, view-specific autoencoders for feature extraction, an attention mechanism for integration, and a final classification output. This segmentation maintains operational clarity while achieving high detection accuracy through specialized processing at each stage
Data Source
AI summary
Provided is a method for detecting abnormal working conditions of multi-view data based on feature regression. By the method, data capable of being acquired in a production process is collected together, a big data pool is established, and historical data information is fully utilized; by analyzing the data in the data pool, the method for detecting abnormal working conditions based on the multi-view data is established by a feature regression method, and a general mathematical model is established for preprocessed data acquired by different sensors; left and right projection vectors solved through the model can make similar sample points have better clustering effects in a low dimensional space; and by comparing a correlation between vectors after dimensionality reduction and various category vectors, production working conditions at a current time can be recognized.