US-12624979-B2 - Method of measuring physical properties

US12624979B2US 12624979 B2US12624979 B2US 12624979B2US-12624979-B2

Abstract

A method of measuring physical properties includes a number of operations. Simulated mass flow rate data of training physical property parameter sets are generated by a computing unit based on the training physical property parameter sets and vessel shape information. Deep learning model is trained by a processor based on a training data set with an input feature vector of the simulated mass flow rate data and an output feature vector of the training physical property parameter sets. Measured fluid is received by a loader through an opening of a vessel of the vessel shape information. Weight accumulation data of weight of the loader is measured by a scale during a time period to obtain measured mass flow rate data. The measured mass flow rate data is inputted to the deep learning model to obtain measured physical property parameter set of the measured fluid.

Inventors

Chih-Ang CHUNG
Chun-Hung WENG
Wei-Xu CAI
Han-Xiang LI

Assignees

NATIONAL CENTRAL UNIVERSITY

Dates

Publication Date: 20260512
Application Date: 20230502
Priority Date: 20230203

Claims (14)

1 . A method of measuring physical properties, comprising: generating a plurality of simulated mass flow rate data corresponding to a plurality of training physical property parameter sets by a computing unit based on the training physical property parameter sets and a vessel shape information; training a deep learning model by a processor based on a training data set, wherein an input feature vector of the training data set comprises the simulated mass flow rate data and an output feature vector of the training data set comprises the training physical property parameter sets; containing a measured metal in a vessel corresponding to the vessel shape information, wherein the vessel is opaque and accommodated in a furnace; heating the measured metal to a determined temperature by the furnace, so that the measured metal is melted into a molten metal fluid as a measured fluid; receiving the measured fluid by a loader through an opening of the vessel; measuring a weight accumulation data of a weight of the loader by a scale during a time period; obtaining a measured mass flow rate data of the measured fluid based on the weight accumulation data; and inputting the measured mass flow rate data to the deep learning model to obtain a measured physical property parameter set of the measured fluid.
2 . The method of claim 1 , wherein the training physical property parameter sets comprises a plurality of densities, a plurality of viscosities and a plurality of surface tensions, the measured physical property parameter set comprises a measured density, a measured viscosity and a measured surface tension.
3 . The method of claim 1 , wherein the vessel shape information comprises a cross-section area of the vessel, a cross-section area of the opening of the vessel and a length of the opening of the vessel.
4 . The method of claim 1 , wherein the computing unit is a computational fluid dynamics unit, generating the simulated mass flow rate data corresponding to the training physical property parameter sets comprises: performing a simulation of a computational fluid dynamics algorithm by the computing unit to obtain the simulated mass flow rate data.
5 . The method of claim 1 , wherein the simulated mass flow rate data comprises a plurality of simulated mass flow rates with respect to a plurality of time points and/or with respect to a plurality of height heads, the height heads are a plurality of fluid height of the measured fluid in the vessel from a simulation performed by the computing unit.
6 . The method of claim 1 , wherein a guide angle is provided at a top of the opening of the vessel.
7 . The method of claim 1 , wherein the vessel is spatially separated from the loader.
8 . A method of measuring physical properties, comprising: providing a vessel and recording a vessel shape information of the vessel; training a deep learning model by a processor, comprising: measuring a weight accumulation data of a weight of a fluid flowing from the vessel to a loader during a time period by a scale; generating a mass flow rate data of the fluid based on the weight accumulation data; converting the weight accumulation data into a first length criterion occupied by the fluid in the vessel through a predicted property parameter set and the vessel shape information; simulating a flow of the fluid based on the predicted property parameter set and the vessel shape information to obtain a second length criterion occupied by the fluid in the vessel; calculating a difference function based on a difference between the first length criterion and the second length criterion; adjusting the predicted property parameter set to minimize the difference function, wherein the predicted property parameter set is selected as a physical property parameter set of the fluid when the difference function is minimized; and training the deep learning model through a training data set, wherein an input feature vector of the training data set comprises the mass flow rate data and an output feature vector of the training data set comprises the physical property parameter set; containing a measured metal in the vessel, wherein the vessel is opaque and accommodated in a furnace; heating the measured metal to a determined temperature by the furnace, so that the measured metal is melted into a molten metal fluid as a measured fluid; receiving the measured fluid flowing out from an opening of the vessel by the loader; measuring a measured weight accumulation data of a weight of the loader receiving the measured fluid relative to a plurality of time points by the scale; obtaining a measured mass flow rate data of the measured fluid based on the measured weight accumulation data; and inputting the measured mass flow rate data to the deep learning model to obtain a measured physical property parameter set of the measured fluid.
9 . The method of claim 8 , wherein simulating the flow of the fluid based on the predicted property parameter set and the vessel shape information to obtain the second length criterion occupied by the fluid in the vessel comprises: performing a simulation of a computational fluid dynamics algorithm to obtain the second length criterion.
10 . The method of claim 8 , wherein the first length criterion is a first height head obtained by converting the weight accumulation data into a first fluid height of the fluid in the vessel, and the second length criterion is a second height head obtained by simulating a second fluid height of the fluid in the vessel.
11 . The method of claim 10 , wherein the first height head and the second height head are functions relative to the time points.
12 . The method of claim 8 , wherein the predicted property parameter set comprises a density, a viscosity and a surface tension, and the measured physical property parameter set comprises a measured density, a measured viscosity and a measured surface tension of the measured fluid.
13 . The method of claim 8 , wherein a guide angle is provided at a top of the opening of the vessel, and the vessel shape information comprises a cross-section area of the vessel, a cross-section area of the opening of the vessel, a length of the opening of the vessel and the guide angle.
14 . The method of claim 8 , wherein the vessel is spatially separated from the loader.

Description

CROSS-REFERENCE TO RELATED APPLICATION This application claims priority to Taiwan application Serial Number 11/210,3935, filed Feb. 3, 2023, which is herein incorporated by reference in its entirety. BACKGROUND Field of Invention The present disclosure relates to method of measuring physic properties. Description of Related Art Measurement of fluid properties is an important issue in engineering. In early developments, fluid properties such as density could be obtained experimentally through Bernoulli's principle and corresponding flow channel design. For example, in some designed measurement methods, fluid properties can be obtained through the draining vessel method, especially for high-temperature molten fluids. In general, in the measurement of the draining vessel method, it takes a considerable amount of time to perform the corresponding fluid simulation calculations and costs a lot of time. Therefore, providing a solution that can improve the time required for measurement in the draining vessel method is one of the problems to be solved for the industry. SUMMARY An aspect of the present disclosure is related to a method of measuring physical properties. According to one or more embodiments, a method of measuring physical properties includes a number of operations. A plurality of simulated mass flow rate data corresponding to a plurality of training physical property parameter sets is generated by a computing unit based on the training physical property parameter sets and vessel shape information. A deep learning model is trained by a processor based on a training data set, wherein an input feature vector of the training data set includes the simulated mass flow rate data and an output feature vector of the training data set includes the training physical property parameter sets. A measured fluid is received by a loader through an opening of a vessel corresponding to the vessel shape information. A weight accumulation data of a weight of the loader is measured by a scale during a time period. A measured mass flow rate data of the measured fluid is obtained based on the weight accumulation data. The measured mass flow rate data is inputted to the deep learning model to obtain a measured physical property parameter set of the measured fluid. In one or more embodiments of the present disclosure, the training physical property parameter sets including a plurality of densities, a plurality of viscosities and a plurality of surface tensions. The measured physical property parameter set includes a measured density, a measured viscosity and a measured surface tension. In one or more embodiments of the present disclosure, the vessel shape information includes a cross-section area of the vessel, a cross-section area of the opening of the vessel and a length of the opening of the vessel. In one or more embodiments of the present disclosure, the computing unit is a computational fluid dynamics unit. Generating the simulated mass flow rate data corresponding to the training physical property parameter sets includes performing a simulation of a computational fluid dynamics algorithm by the computing unit to obtain the simulated mass flow rate data. In one or more embodiments of the present disclosure, the vessel is accommodated in a furnace. Receiving the measured fluid by the loader through the opening of the vessel corresponding to the vessel shape information includes a number of operations. A measured metal is contained in the vessel. The measured metal is heated to a determined temperature by the furnace, so that the measured metal is melted into a molten metal fluid as the measured fluid. In one or more embodiments of the present disclosure, the simulated mass flow rate data includes a plurality of simulated mass flow rates with respect to a plurality of time points and/or with respect to a plurality of height heads. The height heads are a plurality of fluid height of the measured fluid in the vessel from a simulation performed by the computing unit. In one or more embodiments of the present disclosure, a guide angle is provided at a top of the opening of the vessel. In one or more embodiments of the present disclosure, the vessel is spatially separated from the loader. An aspect of the present disclosure is related to a method of measuring physical properties. According to one or more embodiments, a method of measuring physical properties includes a number of operations. A vessel and recording a vessel shape information of the vessel is provided. A deep learning model is trained by a processor. Training the deep learning model by the processor includes a number of operations. A weight accumulation data of a weight of a fluid flowing from the vessel to a loader during a time period is measured by a scale. A mass flow rate data of the fluid is generated based on the weight accumulation data. The weight accumulation data is converted into a first length criterion occupied by the fluid in the vessel through a pre