Automated Corrosion Analysis Using Neural Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The manual analysis of corrosion coupons for pipeline health is prone to operator bias and errors due to subjective interpretation and calculation transmission, leading to inconsistent and inaccurate determinations of pipeline condition.
Innovation Solution
A system utilizing a corrosion coupon, a database, and a neural network to collect and analyze corrosion data, which is then processed to output a corrosion level based on existing data from multiple pipelines, reducing human bias and automating the analysis process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual analysis by scientist is used, then personal experience and interpretation can be applied, but operator bias and errors lead to inconsistent and inaccurate determinations
Solution Approach 1:
The patent replaces the manual mechanical analysis process performed by scientists with an automated image processing and machine learning system. The neural network automatically analyzes corrosion coupon images, eliminating human operator bias and interpretation errors while maintaining high accuracy in pipeline health determination.
Solution Approach 2:
The system enables self-service analysis where the corrosion coupon images are automatically processed by the neural network without requiring scientist intervention. The automated system independently performs measurement, analysis, and determination of pipeline health status, freeing experts from routine analysis tasks.
2Reliability
If centralized database with neural network is implemented, then operator bias is minimized and accuracy improves, but system complexity increases
Solution Approach 1:
The neural network system serves multiple functions: it analyzes corrosion coupon images, measures corrosion levels, compares results against historical data, and generates pipeline health assessments. This multi-functional approach consolidates what would otherwise require multiple separate systems into a single universal platform.
Solution Approach 2:
The patent introduces a centralized database as an intermediary layer between data collection and analysis. This database stores and manages corrosion coupon images, historical data, and analysis results, serving as a mediator that organizes information flow and enables the neural network to access and process data efficiently.
3Productivity
If automated neural network analysis is used, then processing time is reduced and efficiency increases, but initial setup and training requirements increase complexity
Solution Approach 1:
The system performs preliminary training of the neural network using historical corrosion data before deployment. This preliminary action prepares the model in advance, allowing it to quickly and accurately analyze new corrosion coupons without requiring complex real-time processing or extensive setup during actual operation.
Solution Approach 2:
The patent utilizes parameter changes in the neural network during training and deployment phases. By adjusting model parameters, learning rates, and architectural configurations during development, the system optimizes processing speed and accuracy while managing the complexity of implementation.
Data Source
AI summary
A system for determining a corrosion level of a pipeline includes a corrosion coupon received from the pipeline, a database, a first computer, a second computer, a neural network, and a third computer. The database includes existing data indicative of corrosion of a plurality of previously analyzed pipelines. The first computer receives corrosion data of the corrosion coupon and uploads the corrosion data to the database. The second computer uploads calculations performed on the corrosion data to the database. The neural network receives the corrosion data and the calculations from the database and outputs a corrosion level of the pipeline based on the corrosion data and the calculations. Further, the neural network is trained on the existing data from the plurality of previously analyzed pipelines so that the corrosion level is based on a combination of the corrosion data, the calculations, and the existing data. Finally, a third computer receives the corrosion level and generates a report of the corrosion level of the pipeline.