Server-Side ML Flow Measurement for Acoustic Meters
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Acoustic flow meters require on-board computing power for calculations, limiting their ability to perform sophisticated algorithms, and lack remote update capabilities and data sharing, leading to inaccurate and cumbersome measurement processes.
Innovation Solution
A machine learning model is used to process data from acoustic devices, transmitting signals into pipelines, where the data is processed on a server to select a reference model, extract features, and apply a multivariate model to predict flow rates, enabling more accurate measurements and remote firmware updates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If on-board computing power is used for flow rate calculations, then the device can perform measurements independently, but the processing power is limited to basic physics-based calculations and cannot handle sophisticated algorithms
Solution Approach 1:
The patent introduces a server as an intermediary computing resource that receives acoustic data from the flow meter device. The server performs sophisticated machine learning analysis on the transmitted data, enabling advanced algorithms without requiring enhanced on-board processing power. This mediator approach allows the device to leverage remote computational resources for improved measurement accuracy.
2Adaptability or versatility
If the processor is local and the code is fixed, then the device operates autonomously, but it cannot easily be updated remotely when new algorithms are available
Solution Approach 1:
The patent extracts the complex algorithm execution and model training functions from the local device to a remote server. The device retains only data collection and transmission capabilities, while sophisticated processing occurs externally. This extraction allows the device to benefit from remote updates and new algorithms without requiring complex local firmware update mechanisms.
3Measurement precision
If standard physics-based calculations are used, then the computation is simple and fast, but the accuracy is dependent on multiple factors that cannot always be accounted for
Solution Approach 1:
The patent substitutes traditional physics-based mechanical calculations with machine learning models that operate on acoustic data. Instead of relying on fixed physical equations that require numerous assumptions about fluid properties and pipe characteristics, the system uses trained neural networks that automatically adapt to various conditions, improving accuracy without increasing device complexity.
4Measurement precision
If manual calibration is performed periodically, then the device maintains measurement accuracy, but it requires physical inspection and is cumbersome to monitor from a central location
Solution Approach 1:
The patent implements self-service through automated machine learning models that continuously learn from incoming data and adapt to changing conditions. The system performs automatic calibration by training models on-site using real-world data, eliminating the need for manual physical calibration. This self-calibrating capability allows the device to maintain accuracy autonomously without requiring periodic human intervention or physical inspection.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enhances the accuracy of flow rate measurements by leveraging machine learning on a server, allowing for remote updates and improved data processing, overcoming the limitations of on-board processing power and manual calibration.
Implementation Method 1
receiving, by the acoustic device, received signals reflected from the contents of the pipeline
Data Source
AI summary
Systems and methods enable predicting flow rate in a pipeline by using machine learning. Data is collected from a pipeline with an acoustic device and transmitted from the acoustic device to a server. The data is processed and, based on the processing, a reference model is selected that is most appropriate for the processed data. Features extracted from the data are input to the reference model. The reference model outputs a predicted flow rate.