Production Path Fault Detection Using Defect Influence Ranking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In manufacturing processes, such as those for liquid crystal panels, identifying the specific production paths and nodes responsible for product defects is challenging due to various factors involved across the production line, necessitating a method to quickly and accurately determine faulty paths and nodes to improve efficiency and reduce defects.
Innovation Solution
A method and device that analyze production record data to calculate defective rates, influence scores, and weight of evidence to identify faulty production paths and nodes by determining their impact on product quality, using statistical methods like mean, standard deviation, and association rule analysis to rank paths and nodes based on their contribution to defects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional manual inspection methods are used to identify defective production paths, then operators can detect product defects, but the process is time-consuming and cannot quickly determine the specific faulty nodes in the production line
Solution Approach 1:
The patent replaces manual inspection methods with an automated computer-based system that uses machine learning algorithms to analyze production data. The system automatically identifies defective production paths and faulty nodes by processing production record data, replacing the mechanical manual inspection process with an automated information processing system that achieves both high accuracy and rapid fault identification
Solution Approach 2:
The patent introduces an intermediary computational system that acts as a mediator between production data and fault identification. This system uses trained machine learning models to process production record data and generate fault predictions, serving as an intermediary layer that translates raw production data into actionable fault diagnostics without requiring direct manual analysis
2Reliability
If comprehensive data from all production nodes is analyzed to identify faulty paths, then accurate fault identification can be achieved, but the system complexity increases
Solution Approach 1:
The patent segments the complex production line into multiple production paths, each consisting of specific production nodes. The machine learning model analyzes these segmented paths independently, evaluating the probability of each path being defective. This segmentation approach maintains high detection reliability by comprehensively analyzing all nodes while reducing system complexity through modular path-based analysis
Solution Approach 2:
The patent performs preliminary actions by pre-training machine learning models with historical production data before actual fault detection. The models are trained in advance to recognize patterns associated with defective production paths. This preliminary training reduces the complexity of real-time analysis by having the system already equipped with learned knowledge, enabling accurate fault identification without requiring complex real-time computation
Data Source
AI summary
According to the embodiments of the present disclosure, there is provided a method and device of detecting fault in production, and a computer readable storage medium. The method includes: determining whether a plurality of production paths in a production line are faultless in one or more production batches, based on production record data; and determining at least one of the plurality of production paths to be faulty, at least partially based on whether the plurality of production paths are faultless in the one or more production batches.