Automated MICs Extraction from Aircraft Engine Manuals
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Manual extraction of Master Inspection Characteristics (MICs) information from large technical manuals for aircraft engines is time-consuming and prone to human errors, making it inefficient and inaccurate.
Innovation Solution
A scalable end-to-end Machine Learning (ML) architecture using a Linear Support Vector Machine (SVM) Classifier coupled with Chi Square correlation to automate the extraction process, converting technical manuals into XML, and organizing extracted information in an engineer-friendly format.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual extraction of MICs information is performed from large technical manuals, then the extracted information can be organized and used by engineers, but the process consumes a lot of time and energy and is prone to human errors
Solution Approach 1:
The patent replaces the manual mechanical extraction process with an automated computer-based system that uses Natural Language Processing (NLP) and Machine Learning models to extract MICs information from technical manuals, thereby eliminating human error and significantly reducing extraction time
Solution Approach 2:
The patent introduces an intermediary automated extraction system that acts as a mediator between the technical manual and the engineer, using NLP techniques and ML models to process and extract relevant MICs information, thus resolving the contradiction between accuracy and time consumption
2Productivity
If manual extraction of MICs information is performed from large technical manuals, then the extracted information can be obtained, but the process is prone to human errors that may slow down or pause the work
Solution Approach 1:
The patent replaces the manual extraction process with an automated computer-based system using NLP and Machine Learning, which eliminates human error and ensures high reliability of extracted MICs information, thereby maintaining high engineering productivity
Solution Approach 2:
The patent implements a self-service automated extraction system that independently processes technical manuals and extracts MICs information without human intervention, ensuring both high productivity and reliability through consistent automated performance
3Loss of information
If the technical manual has a relatively large content size, then it contains comprehensive technical information, but finding and extracting desired technical information may not be practical as this may take time
Solution Approach 1:
The patent applies extraction by selectively identifying and extracting only the relevant MICs information from the large technical manual using NLP keywords and Machine Learning models, thus maintaining information completeness while reducing the time required to find and extract desired information
Solution Approach 2:
The patent segments the large technical manual into manageable units by identifying specific MICs information points using NLP techniques and Machine Learning, allowing comprehensive information to be processed and extracted efficiently without requiring time-consuming manual search
Data Source
Figure 1
Figure 2A~2B
Figure 3A
AI summary
A method of extracting information from a technical manual (100, 400), the method being implemented on a computer system (500). The method includes extracting, by the computer system (500), selected keywords from the technical manual (100, 400), appending the selected keywords to a table (104) to provide master data (105), receiving a portion of the master data (105) and a first table of labels (109) corresponding to portion of the master data (105), performing a text pre-processing (110) of the master data (105) to format the master data (105) to provide pre-processed data (111, 412), applying a Machine Learning model (118) to the pre-processed data (111, 412) to generate a second table of labels (120) corresponding to an entire master data (105), and processing labels in the second table of labels (120) to correspond to the master data (105) to enable a user to review or to use, or both, information contained in the technical manual (100, 400) without reading or searching in the entire technical manual (100, 400).