Pipeline Sensor Fusion for Accurate Hydrocarbon Product Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Pipeline transportation networks for hydrocarbons face challenges in real-time mapping and predicting the behavior and location of different hydrocarbon products due to complex operations, shared pipeline usage, and integrity issues like leaks, which affect flow rates and product mixing.
Innovation Solution
An automated method and system for integrating pipeline sensors using machine learning to acquire, integrate, and analyze sensor responses, enabling accurate identification of hydrocarbon products and detection of leaks, while predicting arrival times and controlling valves for efficient product management.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple sensors are integrated to improve measurement accuracy, then product mapping precision improves, but system complexity increases
Solution Approach 1:
The patent combines multiple sensor responses (flow meters, pressure transducers, temperature sensors) into a unified machine learning model that processes all sensor data together to identify hydrocarbon products and their locations, thereby improving measurement precision while managing system complexity through integration
Solution Approach 2:
The patent introduces a machine learning model as an intermediary that processes and integrates sensor responses, transforming raw data from multiple sensors into accurate product identification and location mapping, reducing the complexity of direct multi-sensor integration
2Measurement precision
If real-time sensor data integration is performed to improve product identification accuracy, then mapping accuracy improves, but processing time increases
Solution Approach 1:
The patent pre-trains the machine learning model using historical sensor data and product information before real-time operation, so that during actual product mapping, the pre-trained model can quickly identify products and locations without extensive real-time processing
Solution Approach 2:
The patent replaces traditional mechanical signal processing methods with a machine learning-based system that can integrate and analyze multiple sensor responses more efficiently, reducing processing time while maintaining or improving mapping accuracy
3Measurement precision
If machine learning models are used to integrate sensor responses, then product identification accuracy improves, but computational requirements increase
Solution Approach 1:
The patent performs computationally intensive model training in advance using historical data, allowing the deployed model to operate with lower computational requirements during real-time product identification, thus reducing ongoing energy consumption while maintaining high accuracy
Data Source
AI summary
An automated method of pipeline sensor integration for product mapping of a pipeline network is provided. The method includes acquiring, by a plurality of sensors of the pipeline network, first sensor responses of a pipeline in the pipeline network when a first hydrocarbon product is flowing through the pipeline. The method further includes using a prediction circuit to receive the acquired first sensor responses, integrate the received first sensor responses into one or more integrated first sensor responses in order to improve accuracy of the received first sensor responses, and identify the first hydrocarbon product in the pipeline based on the integrated first sensor responses. The prediction circuit is built from training data using a machine learning process. The training data includes first training sensor responses of the pipeline by the plurality of sensors acquired at a previous time when the first hydrocarbon product was flowing through the pipeline.