Recognition of gas-liquid flow regimes in helically coiled tube using wire-mesh sensor and KNN algorithm

Liu, Li; Hu, Bing; Liu, Shuai; Wang, Ke; Gu, Hanyang

doi:10.1016/j.ijmultiphaseflow.2022.104144

Cited by 12 publications

(4 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This search is based on some distance metric, where the shorter the distance, the higher the degree of similarity. The prediction samples are predicted using these k neighboring points, and the average of these k neighboring points is used as the output in the regression prediction [29][30]. The steps are as follows:…”

Section: Fundamentals Of the Knn Algorithmmentioning

confidence: 99%

Research on the Application of KNN Algorithm Incorporating Gaussian Functions in Precision Marketing Classification of E-commerce Platforms

Wang

2023

Applied Mathematics and Nonlinear Sciences

View full text Add to dashboard Cite

The technology can fully explore the user’s consumption behavior habits and help the e-commerce platform formulate more precise marketing strategies in a targeted manner. This paper firstly analyzes the optimization of marketing strategy based on the 3R marketing theory, gives the design process of the precise marketing strategy of an e-commerce platform, and analyzes the personalized service based on consumer classification. Secondly, for the shortcomings of the KNN algorithm in the process of accurate classification, the Gaussian function is introduced to weight the optimization of the algorithm, which further realizes the construction of the G-KNN algorithm. Finally, the testing and application analysis of the algorithm model was carried out using the actual user consumption data of the e-commerce platform. The results show that the classification accuracy of the G-KNN algorithm has been maintained at about 95% when the K value exceeds 800, and the F1 composite value of this paper’s algorithm fluctuates around 56% when the K value exceeds 1000. On the e-commerce platform, except for the electrical appliances category classification test, the fit and accuracy of other categories basically match. Using the KNN algorithm incorporating the Gaussian function can effectively realize the accurate classification of user characteristics on the e-commerce platform and provide data support for the e-commerce platform to formulate accurate marketing strategies based on consumer preferences.

show abstract

Section: Fundamentals Of the Knn Algorithmmentioning

confidence: 99%

Research on the Application of KNN Algorithm Incorporating Gaussian Functions in Precision Marketing Classification of E-commerce Platforms

Wang

2023

Applied Mathematics and Nonlinear Sciences

View full text Add to dashboard Cite

show abstract

“…Helical-coiled One-Through Steam Steam generator (H-OTSG) has been widely used in small modular reactors because of its compact structure and high thermal efficiency (Xu et al, 2021;Liu et al, 2022). The reactor primary coolant flows through the outside of the H-OTSG helical tube and exchanges heat with the secondary loop, the coolant inside the tube is heated and generates superheated steam.…”

Section: Introductionmentioning

confidence: 99%

Experimental study on helical tubes heat transfer characteristics considering corrosion products

Yulong,

Shuqi,

Xianghui

et al. 2023

Front. Energy Res.

View full text Add to dashboard Cite

During the power operation of nuclear reactors, the corrosion products in the secondary circuit will deposit on the inside of Helical-coiled One-Through Steam Generator (H-OTSG) heat transfer tubes and form fouling. The main component of fouling is iron-based oxide, which generates from the heat exchangers made of cast iron. Fouling may change the local heat transfer characteristics of heat transfer tubes, and then affect the safety and economy of reactors. In order to obtain the distribution of fouling thickness and its action mechanism on helical tubes heat transfer characteristics, the heat transfer experiment under small pressurized reactors service environment was conducted. The experimental results indicate: corrosion products mainly deposit on onset of saturated boiling point with the maximum fouling thickness of 50 μm, and induce local wall temperature elevating about 773 K; fouling will increase the thermal equilibrium void fraction and hinder the heat transfer of helical tube, but has little impact on the onset position of saturated boiling point; corrosion products will release and re-deposit near dry-out point because of the flush of high speed vapor, causing the periodic fluctuation of thermal resistance of helical tube.

show abstract

“…Ten parameters were analyzed by these ML algorithms, which indicated that the plastic viscosity and true vertical depth highly influenced the drilling mud density. As ML algorithms are very powerful in data and image classification tasks, Liu et al 23 trained the k-nearest neighbors model with experimental void fraction data and managed to recognize and predict the flow pattern of gas-liquid flows in inclined tubes. Their results were comparable to previous experimental results reported by Dennis et al 24 and Zhu et al 25 Another common application of ML is to be coupled with numerical methods (e.g., CFD) to improve modeling performance.…”

mentioning

confidence: 99%

“…32 In this work, the hybrid ML algorithm approach is used to analyze turbulent particle transport in a horizontal pipe, due to its simplicity, straightforward implementation, and robustness for classification as well as regression. 23,33,34 High-quality training data are obtained from an experimental Lagrangian technique of 3D positron imaging particle tracking (PEPT), and the minimum amount required to accurately predict phase velocity and concentration distributions in the pipe by the hybrid ML algorithm is determined. In addition, by combining datasets corresponding to different flow conditions of particle concentration, size, and density, the hybrid ML model is trained to predict the multiphase flow behavior under new conditions within and without the range of experimental data available.…”

mentioning

confidence: 99%

Experimentally trained hybrid machine learning algorithm for predicting turbulent particle-laden flows in pipes

Yang,

Li,

Barigou

2023

Physics of Fluids

View full text Add to dashboard Cite

A hybrid learning algorithm consisting of a preprocessor, a k-nearest neighbors regressor, a noise generator, and a particle–wall collision model is introduced for predicting features of turbulent single-phase and particle–liquid flows in a pipe. The hybrid learning algorithm has the ability to learn and predict the behavior of such complex fluid dynamic systems using experimental dynamic databases. Given a small amount of typical training data, the algorithm is able to reliably predict the local liquid and particle velocities as well as the spatial distribution of particle concentration within and without the limits of the range of training data. The algorithm requires an order of magnitude less training data than a typical full set of experimental measurements to give predictions on the same level of accuracy (typically, 20 cf. 100 trajectories for phase velocity distribution and 40 cf. 500 trajectories for phase concentration distribution), thus leading to huge reductions in experimentation and simulation. A feature importance analysis revealed the effects of the different experimental variables on the particle velocity field in a two-phase particulate flow, with particle–liquid density ratio and particle vertical radial position being the most influential and particle concentration the least. The algorithm is amenable to extension by using more complex databanks to address a much more comprehensive range of flow situations.

show abstract

Recognition of gas-liquid flow regimes in helically coiled tube using wire-mesh sensor and KNN algorithm

Cited by 12 publications

References 37 publications

Research on the Application of KNN Algorithm Incorporating Gaussian Functions in Precision Marketing Classification of E-commerce Platforms

Research on the Application of KNN Algorithm Incorporating Gaussian Functions in Precision Marketing Classification of E-commerce Platforms

Experimental study on helical tubes heat transfer characteristics considering corrosion products

Experimentally trained hybrid machine learning algorithm for predicting turbulent particle-laden flows in pipes

Contact Info

Product

Resources

About