Control of a jacketed continuous stirred tank reactor (CSTR) is challenging due to nonlinear dynamics, complexity, and rapid reactor dynamics under imperfect mixing in the jacket. Current controller designs mainly foc...
详细信息
Control of a jacketed continuous stirred tank reactor (CSTR) is challenging due to nonlinear dynamics, complexity, and rapid reactor dynamics under imperfect mixing in the jacket. Current controller designs mainly focus on the two-state model, neglecting the potential of three-state models in scenarios with nonperfect mixing and fast reactor dynamics. This study proposes a sliding mode controller (SMC) design scheme based on the transfer function model using a newly developed jellyfish optimisation algorithm. Further, a fractional-order sliding mode control (FO-SMC) strategy is proposed, which integrates modifications to the SMC to mitigate chattering, enhance control robustness, and provide better disturbance rejection capability. PID and fractional-order PID (FOPID) controllers were also designed for comparative analysis. The simulation results demonstrated that FO-SMC outperformed other designed controllers, shown by a 37.14% reduction in settling time, 10.69% reduction in integral absolute error (IAE), and 19.06% reduction in time-weighted absolute error (ITAE) compared to SMC and various other improved performance indicators. Parameter variation and noise analysis highlighted the ability of the controller to maintain stability and performance under dynamic conditions.
A facial image analysis system for monitoring customer interest employs cutting-edge facial detection technology for evaluating and analyzing customers' expressions, offering real-time perceptions of their respons...
详细信息
A facial image analysis system for monitoring customer interest employs cutting-edge facial detection technology for evaluating and analyzing customers' expressions, offering real-time perceptions of their responses and preferences. Leveraging advanced neural networks (NNs) can dynamically and correctly analyze facial expressions, allowing retailers to separate and interpret customers' emotions with remarkable accuracy. Deep learning (DL) systems surpass at capturing difficult patterns and nuances in facial features. This study paper presents a novel jellyfish optimizer algorithm with Deep Learning for Consumer Interest Monitoring Advanced Face Analysis (JOADL-CIMAFA) model. The main intention of the JOADL-CIMAFA method is to analyze the facial images of the consumer using the DL model for the detection and classification of customer interest. In the presented JOADL-CIMAFA technique, the EfficientNet model can be applied to the feature extraction process. For the hyperparameter tuning procedure, the JOA can be used for optimum hyperparameter selection of the EfficientNet model. Furthermore, the long short-term memory (LSTM) technique can be exploited for the identification and classification of consumer interest. To establish the enhanced outcome of the JOADL-CIMAFA system, a widespread of simulations can be implemented. The experimental values highlighted that the JOADL-CIMAFA technique illustrates superior performance over other models in terms of different measures.
INTRODUCTION: The article discusses the key steps in digital visual design reengineering, with a special emphasis on the importance of information decoding and feature extraction for flat cultural heritage. These proc...
详细信息
INTRODUCTION: The article discusses the key steps in digital visual design reengineering, with a special emphasis on the importance of information decoding and feature extraction for flat cultural heritage. These processes not only minimize damage to the aesthetic heritage itself but also feature high quality, efficiency, and recyclability. OBJECTIVES: The aim of the article is to explore the issues of gene extraction methods in digital visual design reengineering, proposing a visual gene extraction method through an improved K-means clustering algorithm. METHODS: A visual gene extraction method based on an improved K-means clustering algorithm is proposed. Initially analyzing the digital visual design reengineering process, combined with a color extraction method using the improved JSO algorithm-based K-means clustering algorithm, a gene extraction and clustering method for digital visual design reengineering is proposed and validated through experiments. RESULT: The results show that the proposed method improves the accuracy, robustness, and real-time performance of clustering. Through comparative analysis with Dunhuang murals, the effectiveness of the color extraction method based on the K-means-JSO algorithm in the application of digital visual design reengineering is verified. The method based on the K-means-GWO algorithm performs best in terms of average clustering time and standard deviation. The optimization curve of color extraction based on the K-means-JSO algorithm converges faster and with better accuracy compared to the K-means-ABC, K-means-GWO, K-means-DE, K-means-CMAES, and K-means-WWCD algorithms. CONCLUSION: The color extraction method of the K-means clustering algorithm improved by the JSO algorithm proposed in this paper solves the problems of insufficient standardization in feature selection, lack of generalizationability, and inefficiency in visual gene extraction methods.
暂无评论