Miao Qian
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Miao Qian has a total of 43 co-authors. Here are the top co-authors by number of shared publications:
- Zhong Xiang
School of Mechanical Engineering, Zhejiang Sci-Tech University, Hangzhou 310018, China
11 shared publications Xudong Hu
Zhejiang Provincial Key Lab of Modern Textile Machinery, Zhejiang Sci-Tech University, China
6 shared publicationsZhong Xiang
5 shared publicationsJianxin Zhang
3 shared publications- Weitao Wu
School of Mechanical Engineering, Zhejiang Sci-Tech University, Hangzhou, China
3 shared publications Binbin Xie
School of Mechanical Engineering, Zhejiang Sci-Tech University, 928/2 Main Street, Xiasha Higher Education Park, Xiasha Street, Jianggan, Hangzhou 310018, China
2 shared publications
Latest Publications
- 27 November 2025
- Published by Wiley in Chemie Ingenieur Technik
- Vol. 98 (4) , 133-143
Machine‐Learning‐Enabled Structural Design and Optimization of Methanol Reforming Hydrogen Reactors
A methanol steam reforming and combustion channel reactor was developed to supply hydrogen for high‐power fuel cells. Numerical simulation combined with a backpropagation (BP) neural network, whose maximum error is below 5 %, quantified how channel dimensions and inlet flows influence peak temperature difference, flow maldistribution, methanol conversion, and hydrogen output. Multi‐objective optimization using non‐dominated sorting genetic algorithm II (NSGA–II) yielded the optimal layout: length 284 mm, width 171 mm, combustor inlet flow 8.3 × 10 3 mL min −1 , and reformer inlet flow 4.7 × 10 4 mL min −1 . Compared with the initial design, the optimized reactor reduces peak temperature and flow differences by 6.6 % and increases methanol conversion and hydrogen production by about 7 %, meeting the hydrogen demand of roughly 23 kW fuel cells.
- 27 November 2025
- Published by Wiley in Chemie Ingenieur Technik
- Vol. 98 (4) , 133-143
Machine‐Learning‐Enabled Structural Design and Optimization of Methanol Reforming Hydrogen Reactors
A methanol steam reforming and combustion channel reactor was developed to supply hydrogen for high‐power fuel cells. Numerical simulation combined with a backpropagation (BP) neural network, whose maximum error is below 5 %, quantified how channel dimensions and inlet flows influence peak temperature difference, flow maldistribution, methanol conversion, and hydrogen output. Multi‐objective optimization using non‐dominated sorting genetic algorithm II (NSGA–II) yielded the optimal layout: length 284 mm, width 171 mm, combustor inlet flow 8.3 × 10 3 mL min −1 , and reformer inlet flow 4.7 × 10 4 mL min −1 . Compared with the initial design, the optimized reactor reduces peak temperature and flow differences by 6.6 % and increases methanol conversion and hydrogen production by about 7 %, meeting the hydrogen demand of roughly 23 kW fuel cells.
- Open Access
- 9 October 2025
- Published by MDPI AG in Algorithms
- Vol. 18 (10) , 638
Texture-Adaptive Fabric Defect Detection via Dynamic Subspace Feature Extraction and Luminance Reconstruction
Defect detection in textile manufacturing is critically hampered by the inefficiency of manual inspection and the dual constraints of deep learning (DL) approaches. Specifically, DL models suffer from poor generalization, as the rapid iteration of fabric types makes acquiring sufficient training data impractical. Furthermore, their high computational costs impede real-time industrial deployment. To address these challenges, this paper proposes a texture-adaptive fabric defect detection method. Our approach begins with a Dynamic Subspace Feature Extraction (DSFE) technique to extract spatial luminance features of the fabric. Subsequently, a Light Field Offset-Aware Reconstruction Model (LFOA) is introduced to reconstruct the luminance distribution, effectively compensating for environmental lighting variations. Finally, we develop a texture-adaptive defect detection system to identify potential defective regions, alongside a probabilistic ‘OutlierIndex’ to quantify their likelihood of being true defects. This system is engineered to rapidly adapt to new fabric types with a small number of labeled samples, demonstrating strong generalization and suitability for dynamic industrial conditions. Experimental validation confirms that our method achieves 70.74% accuracy, decisively outperforming existing models by over 30%.
- 9 October 2025
- Published by MDPI AG in Algorithms
- Vol. 18 (10) , 638
Open AccessTexture-Adaptive Fabric Defect Detection via Dynamic Subspace Feature Extraction and Luminance Reconstruction
Defect detection in textile manufacturing is critically hampered by the inefficiency of manual inspection and the dual constraints of deep learning (DL) approaches. Specifically, DL models suffer from poor generalization, as the rapid iteration of fabric types makes acquiring sufficient training data impractical. Furthermore, their high computational costs impede real-time industrial deployment. To address these challenges, this paper proposes a texture-adaptive fabric defect detection method. Our approach begins with a Dynamic Subspace Feature Extraction (DSFE) technique to extract spatial luminance features of the fabric. Subsequently, a Light Field Offset-Aware Reconstruction Model (LFOA) is introduced to reconstruct the luminance distribution, effectively compensating for environmental lighting variations. Finally, we develop a texture-adaptive defect detection system to identify potential defective regions, alongside a probabilistic ‘OutlierIndex’ to quantify their likelihood of being true defects. This system is engineered to rapidly adapt to new fabric types with a small number of labeled samples, demonstrating strong generalization and suitability for dynamic industrial conditions. Experimental validation confirms that our method achieves 70.74% accuracy, decisively outperforming existing models by over 30%.
- 9 June 2025
- Published by Elsevier in International Journal of Thermal Sciences
- Vol. 217
The effect of porous lattice structures on jet impingement heat dissipation performance
- 9 June 2025
- Published by Elsevier in International Journal of Thermal Sciences
- Vol. 217
The effect of porous lattice structures on jet impingement heat dissipation performance
- 14 May 2025
- Published by Springer Nature in The Visual Computer
- p. 1-19
Block-wise feature fusion for high-precision industrial surface defect detection
- 14 May 2025
- Published by Springer Nature in The Visual Computer
- p. 1-19
Block-wise feature fusion for high-precision industrial surface defect detection
- 12 September 2024
- Published by SAGE Publications in Textile Research Journal
- Vol. 95 (5-6) , 554-568
Reconstructing hyperspectral images of textiles from a single RGB image utilizing the multihead self-attention mechanism
Hyperspectral images possess abundant information and play a pivotal role in enhancing the accuracy of color difference detection in textiles. However, traditional hyperspectral imaging methods necessitate costly equipment and intricate operational procedures. A novel deep learning model based on a multihead attention mechanism was proposed in this article to facilitate the extensive application of hyperspectral imaging technology in textile quality inspection. This model enabled the reconstruction of the hyperspectral information of plain weave textiles from a single RGB image. In this model, encoder-decoder architecture and pyramid pooling convolutional operations were employed to integrate multiscale features of plain weave cotton-linen textiles. This could capture details and contextual information in textile images more precisely, enhancing the accuracy of hyperspectral image reconstruction. Simultaneously, an attention mechanism was introduced to increase the model’s receptive field and improve its focus on key regions in the input image and feature maps. This resulted in a reduced weighting of redundant information during network learning, leading to an improved feature extraction capability of the network. Through these methods, successful reconstructions of plain weave textiles hyperspectral information from a single RGB image was achieved. Quantitative and qualitative tests were conducted on two datasets, namely, the NTIRE 2020 dataset and a self-made textile dataset, to evaluate the performance of the proposed method. The approach proposed in this article exhibited promising results on both datasets. Specifically, the reconstructed textile hyperspectral images achieved a root mean square error of 0.0344, a peak signal-to-noise ratio of 29.945, a spectral angle mapper of 3.753, and a structural similarity index measure of 0.955 on the textile dataset. In the reconstructed hyperspectral colorimetric experiment, the maximum value of average color difference was 2.641. These results demonstrate that the method can meet the requirements for textile color measurement applications.
- 12 September 2024
- Published by SAGE Publications in Textile Research Journal
- Vol. 95 (5-6) , 554-568
Reconstructing hyperspectral images of textiles from a single RGB image utilizing the multihead self-attention mechanism
Hyperspectral images possess abundant information and play a pivotal role in enhancing the accuracy of color difference detection in textiles. However, traditional hyperspectral imaging methods necessitate costly equipment and intricate operational procedures. A novel deep learning model based on a multihead attention mechanism was proposed in this article to facilitate the extensive application of hyperspectral imaging technology in textile quality inspection. This model enabled the reconstruction of the hyperspectral information of plain weave textiles from a single RGB image. In this model, encoder-decoder architecture and pyramid pooling convolutional operations were employed to integrate multiscale features of plain weave cotton-linen textiles. This could capture details and contextual information in textile images more precisely, enhancing the accuracy of hyperspectral image reconstruction. Simultaneously, an attention mechanism was introduced to increase the model’s receptive field and improve its focus on key regions in the input image and feature maps. This resulted in a reduced weighting of redundant information during network learning, leading to an improved feature extraction capability of the network. Through these methods, successful reconstructions of plain weave textiles hyperspectral information from a single RGB image was achieved. Quantitative and qualitative tests were conducted on two datasets, namely, the NTIRE 2020 dataset and a self-made textile dataset, to evaluate the performance of the proposed method. The approach proposed in this article exhibited promising results on both datasets. Specifically, the reconstructed textile hyperspectral images achieved a root mean square error of 0.0344, a peak signal-to-noise ratio of 29.945, a spectral angle mapper of 3.753, and a structural similarity index measure of 0.955 on the textile dataset. In the reconstructed hyperspectral colorimetric experiment, the maximum value of average color difference was 2.641. These results demonstrate that the method can meet the requirements for textile color measurement applications.
- 6 May 2024
- Published by Taylor & Francis in Numerical Heat Transfer, Part B: Fundamentals
- Vol. 86 (8) , 2454-2472
Enhancing thermal transfer in airfoil structures through advanced simulation techniques and MOPSO optimization
- 6 May 2024
- Published by Taylor & Francis in Numerical Heat Transfer, Part B: Fundamentals
- Vol. 86 (8) , 2454-2472
Enhancing thermal transfer in airfoil structures through advanced simulation techniques and MOPSO optimization
- 17 August 2023
- Published by Wiley in Coloration Technology
- Vol. 140 (2) , 287-299
Segmented modelling and analysis of disperse dye concentration based on multidimensional spectrum
Since the water‐insoluble dispersing dye has both absorption and scattering characteristics, a spatial resolution hyperspectral measurement approach and experimental testing was proposed in this article, which can collect spectral and spatial data from samples simultaneously. The concentration of 81 groups of three‐component disperse dye samples were measured. However, the hyperspectral data of dye solutions in the 420–800 nm band is saturated, resulting in the inability for multispectral data processing. A segmented concentration quantitative analysis model was developed. For the unsaturated band (420–510 nm), the partial least squares (PLS), the N‐way partial least squares (NPLS), and support vector machine (SVM) models using the data points on the X‐axis of a two‐dimensional light intensity distribution map were established. The predicted performance of PLS model was worse slightly than that of the other two models, The coefficient of determination (R2) values of concentrations for red, orange and blue disperse dye were 0.888, 0.796 and 0.959, respectively. For saturated band (520–670 nm), the NPLS and SVM models using the data points on the X‐ and Y‐axis were established. Results shows that the prediction accuracy of concentrations of the three‐component disperse dye was increased by adding additional data points on the Y‐axis, with R2 values of 0.944, 0.807, and 0.912, respectively. For the strong scattering band (680–800 nm), a SVM model was established, and R2 of concentration of the three dyes reached 0.974, 0.933 and 0.995, respectively. The results showed that multidimensional spectroscopy method can improve the prediction accuracy of component concentration of disperse dye solution, by using more spectral information from X and Y directions.
- 17 August 2023
- Published by Wiley in Coloration Technology
- Vol. 140 (2) , 287-299
Segmented modelling and analysis of disperse dye concentration based on multidimensional spectrum
Since the water‐insoluble dispersing dye has both absorption and scattering characteristics, a spatial resolution hyperspectral measurement approach and experimental testing was proposed in this article, which can collect spectral and spatial data from samples simultaneously. The concentration of 81 groups of three‐component disperse dye samples were measured. However, the hyperspectral data of dye solutions in the 420–800 nm band is saturated, resulting in the inability for multispectral data processing. A segmented concentration quantitative analysis model was developed. For the unsaturated band (420–510 nm), the partial least squares (PLS), the N‐way partial least squares (NPLS), and support vector machine (SVM) models using the data points on the X‐axis of a two‐dimensional light intensity distribution map were established. The predicted performance of PLS model was worse slightly than that of the other two models, The coefficient of determination (R2) values of concentrations for red, orange and blue disperse dye were 0.888, 0.796 and 0.959, respectively. For saturated band (520–670 nm), the NPLS and SVM models using the data points on the X‐ and Y‐axis were established. Results shows that the prediction accuracy of concentrations of the three‐component disperse dye was increased by adding additional data points on the Y‐axis, with R2 values of 0.944, 0.807, and 0.912, respectively. For the strong scattering band (680–800 nm), a SVM model was established, and R2 of concentration of the three dyes reached 0.974, 0.933 and 0.995, respectively. The results showed that multidimensional spectroscopy method can improve the prediction accuracy of component concentration of disperse dye solution, by using more spectral information from X and Y directions.
- 11 August 2023
- Published by American Chemical Society (ACS) in Industrial & Engineering Chemistry Research
- Vol. 62 (33) , 13249-13259
Development of an Improved One-Hot Encoding Method for Bubbly Flow Image Prediction Generation under Continuous Superficial Gas Velocities
- 11 August 2023
- Published by American Chemical Society (ACS) in Industrial & Engineering Chemistry Research
- Vol. 62 (33) , 13249-13259
Development of an Improved One-Hot Encoding Method for Bubbly Flow Image Prediction Generation under Continuous Superficial Gas Velocities
- 5 June 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 72 (00189456) , 1-11
HookNet: Efficient Multiscale Context Aggregation for High-Accuracy Detection of Fabric Defects
Automatic defect detection is a critical stage of quality control in the textile industry. Because defects vary in texture structure, size, and spatial distribution, existing defect detection methods have difficulty in achieving a perfect trade-off between detection efficiency, accuracy, and generalizability. This paper presents an advanced fabric defect detection model based on a deep learning algorithm. A novel parallel dilated attention module was designed: this is nested in the deep layers of the neural network backbone to establish global channel dependencies and capture multiscale contextual information. Further, a hook-shaped feature pyramid network was developed for multiscale context aggregation: this directs the network to focus on lower-level features and is helpful for further improving detection efficiency without sacrificing detection accuracy. In addition, the Alpha-GIoU loss is used to improve the accuracy of bounding box regression because of its modulated power parameter α. The advantages and effectiveness of the proposed method were statistically analyzed using data from several public datasets. The code is available at https://github.com/aabb605/HookNet.
- 5 June 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 72 (00189456) , 1-11
HookNet: Efficient Multiscale Context Aggregation for High-Accuracy Detection of Fabric Defects
Automatic defect detection is a critical stage of quality control in the textile industry. Because defects vary in texture structure, size, and spatial distribution, existing defect detection methods have difficulty in achieving a perfect trade-off between detection efficiency, accuracy, and generalizability. This paper presents an advanced fabric defect detection model based on a deep learning algorithm. A novel parallel dilated attention module was designed: this is nested in the deep layers of the neural network backbone to establish global channel dependencies and capture multiscale contextual information. Further, a hook-shaped feature pyramid network was developed for multiscale context aggregation: this directs the network to focus on lower-level features and is helpful for further improving detection efficiency without sacrificing detection accuracy. In addition, the Alpha-GIoU loss is used to improve the accuracy of bounding box regression because of its modulated power parameter α. The advantages and effectiveness of the proposed method were statistically analyzed using data from several public datasets. The code is available at https://github.com/aabb605/HookNet.
- 2 May 2023
- Published by Taylor & Francis in The Journal of the Textile Institute
- Vol. 115 (5) , 688-702
Optimum design of stenter machine hot air supply chamber by coupling CFD & DOE
The efficiency of the textile hot air-drying process fundamentally requires the uniformity of the airflow that impinges on the textile fabric. Optimization of a stenter machine’s hot air supply chamber shape was conducted in this paper, based on the actual real-size model and considering both the outlet flow uniformity and the standard velocity deviation as responses. The purpose is to improve the airflow uniformity at the chamber outlet, and modifications applied to the baseline must be applicable to both old and next-generation stenter machines. Coupled Computational Fluid Dynamics and Design of Experiment (CFD-DoE) optimization on the stenter machine chamber geometry was carried out. The CFD model was validated by measuring the chamber outlet velocity using a 2D velocity scanner. After choosing seven factors based on mechanical manufacturing and fluid dynamics theories, a main effect screening design was conducted to highlight relevant factors that influence the outlet airflow distribution. By fixing the best categorical factors, a central composite design (CCD) was applied using three other continuous factors to generate a quadratic prediction model that could be used to simultaneously maximize and minimize the uniformity index (UI) and the standard deviation (SD), respectively. With the embedded JMP SAS desirability function, the optimal value was calculated based on a statistical approach. Those values were used to design an optimal stenter air supply chamber that has been simulated with Fluent CFD code to validate the statistical prediction model. Finally, optimized geometry was obtained and experimentally validated. Results show that this new chamber improved by 16.48% for the uniformity index and 38.07% for the standard deviation.
- 2 May 2023
- Published by Taylor & Francis in The Journal of the Textile Institute
- Vol. 115 (5) , 688-702
Optimum design of stenter machine hot air supply chamber by coupling CFD & DOE
The efficiency of the textile hot air-drying process fundamentally requires the uniformity of the airflow that impinges on the textile fabric. Optimization of a stenter machine’s hot air supply chamber shape was conducted in this paper, based on the actual real-size model and considering both the outlet flow uniformity and the standard velocity deviation as responses. The purpose is to improve the airflow uniformity at the chamber outlet, and modifications applied to the baseline must be applicable to both old and next-generation stenter machines. Coupled Computational Fluid Dynamics and Design of Experiment (CFD-DoE) optimization on the stenter machine chamber geometry was carried out. The CFD model was validated by measuring the chamber outlet velocity using a 2D velocity scanner. After choosing seven factors based on mechanical manufacturing and fluid dynamics theories, a main effect screening design was conducted to highlight relevant factors that influence the outlet airflow distribution. By fixing the best categorical factors, a central composite design (CCD) was applied using three other continuous factors to generate a quadratic prediction model that could be used to simultaneously maximize and minimize the uniformity index (UI) and the standard deviation (SD), respectively. With the embedded JMP SAS desirability function, the optimal value was calculated based on a statistical approach. Those values were used to design an optimal stenter air supply chamber that has been simulated with Fluent CFD code to validate the statistical prediction model. Finally, optimized geometry was obtained and experimentally validated. Results show that this new chamber improved by 16.48% for the uniformity index and 38.07% for the standard deviation.