Pub Date : 2023-06-01DOI: 10.53106/199115992023063403029
Boqing Feng Boqing Feng, Xiaolei Xu Boqing Feng, Congxu Li Xiaolei Xu, Wenbin Liu Congxu Li, Mohan Liu Wenbin Liu
With the increasing investment in railway construction, China’s railway transport network is now very sound, the number of operating miles is growing, and the operating speed has also made a qualitative leap. At the same time, the safety and reliability of the operation of railway signal cables and other electrical equipment has also put forward higher requirements. At the present stage, the management of railway electrical services equipment mainly relies on manual management, which is cumbersome, inefficient and unsuitable for multi-user sharing. At the same time, the structure of railway electrical equipment is complex, and the components of the equipment are prone to aging, which can easily cause equipment failure. How to professionally manage electrical service equipment and improve the safety and reliability of electrical service equipment has become an urgent problem for railway electrical service departments. Geographic Information System (GIS) architecture uses spatial data layering technology to achieve multi-level and proportional display of equipment and facilities, which can provide visual display of professional facilities such as railway engineering, electricity and power supply, and carry out multi-source and multi-temporal intelligent analysis of data, provide geographical information service interface for various professions of engineering and electricity to meet their own functional requirements. Knowledge mapping is a key technology for acquiring knowledge and building a knowledge database in the era of big data. In order to explore the hidden information between railway electrical resources, integrate seemingly independent data into the knowledge base and apply them. In this paper, we design a GIS-based electric service resource management system in combination with knowledge mapping that can make data complete and well-structured after processing scattered and redundant information, and analyze and discuss the system’s architecture, functional requirements, key technologies and development prospects.
{"title":"GIS-Based Electric Service Resource Management System","authors":"Boqing Feng Boqing Feng, Xiaolei Xu Boqing Feng, Congxu Li Xiaolei Xu, Wenbin Liu Congxu Li, Mohan Liu Wenbin Liu","doi":"10.53106/199115992023063403029","DOIUrl":"https://doi.org/10.53106/199115992023063403029","url":null,"abstract":"\u0000 With the increasing investment in railway construction, China’s railway transport network is now very sound, the number of operating miles is growing, and the operating speed has also made a qualitative leap. At the same time, the safety and reliability of the operation of railway signal cables and other electrical equipment has also put forward higher requirements. At the present stage, the management of railway electrical services equipment mainly relies on manual management, which is cumbersome, inefficient and unsuitable for multi-user sharing. At the same time, the structure of railway electrical equipment is complex, and the components of the equipment are prone to aging, which can easily cause equipment failure. How to professionally manage electrical service equipment and improve the safety and reliability of electrical service equipment has become an urgent problem for railway electrical service departments. Geographic Information System (GIS) architecture uses spatial data layering technology to achieve multi-level and proportional display of equipment and facilities, which can provide visual display of professional facilities such as railway engineering, electricity and power supply, and carry out multi-source and multi-temporal intelligent analysis of data, provide geographical information service interface for various professions of engineering and electricity to meet their own functional requirements. Knowledge mapping is a key technology for acquiring knowledge and building a knowledge database in the era of big data. In order to explore the hidden information between railway electrical resources, integrate seemingly independent data into the knowledge base and apply them. In this paper, we design a GIS-based electric service resource management system in combination with knowledge mapping that can make data complete and well-structured after processing scattered and redundant information, and analyze and discuss the system’s architecture, functional requirements, key technologies and development prospects. \u0000","PeriodicalId":345067,"journal":{"name":"電腦學刊","volume":"42 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124339193","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-06-01DOI: 10.53106/199115992023063403019
Dong-Liang Fan Dong-Liang Fan, Jian Wang Dong-Liang Fan, Qian-Han Zhang Jian Wang, Jin-Ping Du Qian-Han Zhang, Rui Yuan Jin-Ping Du
As the future direction of power development, microgrids are particularly important for rational and efficient energy management. This article establishes an optimization model with multiple uncertainties as parameters for the microgrid energy system, with the objective function of minimizing operating costs. Then, an improved harmony algorithm was used to solve for the optimal solution of the model parameters. Finally, a microgrid system consisting of wind and thermal power units established in a certain area of Hebei Province was used for solution analysis. After experimental verification, the proposed method in this paper achieved significant improvements in both operational cost reduction and microgrid efficiency.
{"title":"Artificial Intelligence Assisted Energy Optimization and Control Method for Microgrids","authors":"Dong-Liang Fan Dong-Liang Fan, Jian Wang Dong-Liang Fan, Qian-Han Zhang Jian Wang, Jin-Ping Du Qian-Han Zhang, Rui Yuan Jin-Ping Du","doi":"10.53106/199115992023063403019","DOIUrl":"https://doi.org/10.53106/199115992023063403019","url":null,"abstract":"\u0000 As the future direction of power development, microgrids are particularly important for rational and efficient energy management. This article establishes an optimization model with multiple uncertainties as parameters for the microgrid energy system, with the objective function of minimizing operating costs. Then, an improved harmony algorithm was used to solve for the optimal solution of the model parameters. Finally, a microgrid system consisting of wind and thermal power units established in a certain area of Hebei Province was used for solution analysis. After experimental verification, the proposed method in this paper achieved significant improvements in both operational cost reduction and microgrid efficiency.\u0000 \u0000","PeriodicalId":345067,"journal":{"name":"電腦學刊","volume":"48 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114686691","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-06-01DOI: 10.53106/199115992023063403026
Ziyang Chen Ziyang Chen, Yang Zhang Ziyang Chen
Dealing with high conflict evidence, traditional evidence theory sometimes has certain limitations, and results in fusion results contrary to common sense. In order to solve the problem of high conflict evidence fusion, this paper analyzes traditional evidence theory and proposes an evidence fusion method that combines cosine distance and information entropy. Cosine distance can measure the directionality between two vectors. The better the directionality, the more similar the two vectors are. Therefore, this article uses cosine distance to determine the similarity between evidences, and then calculates the credibility of each piece of evidence. Information entropy can calculate the amount of information for each evidence. The greater the information entropy, the greater the uncertainty of the evidence. Therefore, this article uses information entropy to measure the uncertainty of the evidence. Then, the credibility and uncertainty of the evidence are fused to calculate the weight of the evidence. Then we use d-s evidence theory for evidence fusion. The numerical example shows that the method is feasible and effective in dealing with conflict evidence.
{"title":"Conflict Evidence Fusion Algorithm Based on Cosine Distance and Information Entropy","authors":"Ziyang Chen Ziyang Chen, Yang Zhang Ziyang Chen","doi":"10.53106/199115992023063403026","DOIUrl":"https://doi.org/10.53106/199115992023063403026","url":null,"abstract":"\u0000 Dealing with high conflict evidence, traditional evidence theory sometimes has certain limitations, and results in fusion results contrary to common sense. In order to solve the problem of high conflict evidence fusion, this paper analyzes traditional evidence theory and proposes an evidence fusion method that combines cosine distance and information entropy. Cosine distance can measure the directionality between two vectors. The better the directionality, the more similar the two vectors are. Therefore, this article uses cosine distance to determine the similarity between evidences, and then calculates the credibility of each piece of evidence. Information entropy can calculate the amount of information for each evidence. The greater the information entropy, the greater the uncertainty of the evidence. Therefore, this article uses information entropy to measure the uncertainty of the evidence. Then, the credibility and uncertainty of the evidence are fused to calculate the weight of the evidence. Then we use d-s evidence theory for evidence fusion. The numerical example shows that the method is feasible and effective in dealing with conflict evidence. \u0000","PeriodicalId":345067,"journal":{"name":"電腦學刊","volume":"27 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121914045","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-06-01DOI: 10.53106/199115992023063403025
Xi-Lin Li Xi-Lin Li, Jie Yu Xi-Lin Li, Shi-Ming Zhao Jie Yu, Ya-Min Wang Shi-Ming Zhao, Hui-Hua Zhang Ya-Min Wang
In order to study common faults in motors and motor transmission systems, this article uses a 5kW motor system as an experimental platform to establish a virtual prototype model. The prototype model includes the following five parts: motor unit, 6-degree of freedom loading mechanism, transmission gearbox, loading spindle, and AC excitation converter. Then, the BP neural network is used to identify typical faults in the virtual prototype. The final recognition time for vibration changes, temperature changes, and current disturbances does not exceed 45 seconds, with an average accuracy rate of over 99%. Overall, the algorithm can accurately diagnose typical faults in a relatively short time.
{"title":"Virtual Prototyping Modeling and Fault Diagnosis Technology for Mechanical and Electrical Equipment","authors":"Xi-Lin Li Xi-Lin Li, Jie Yu Xi-Lin Li, Shi-Ming Zhao Jie Yu, Ya-Min Wang Shi-Ming Zhao, Hui-Hua Zhang Ya-Min Wang","doi":"10.53106/199115992023063403025","DOIUrl":"https://doi.org/10.53106/199115992023063403025","url":null,"abstract":"\u0000 In order to study common faults in motors and motor transmission systems, this article uses a 5kW motor system as an experimental platform to establish a virtual prototype model. The prototype model includes the following five parts: motor unit, 6-degree of freedom loading mechanism, transmission gearbox, loading spindle, and AC excitation converter. Then, the BP neural network is used to identify typical faults in the virtual prototype. The final recognition time for vibration changes, temperature changes, and current disturbances does not exceed 45 seconds, with an average accuracy rate of over 99%. Overall, the algorithm can accurately diagnose typical faults in a relatively short time. \u0000","PeriodicalId":345067,"journal":{"name":"電腦學刊","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130305719","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-06-01DOI: 10.53106/199115992023063403014
Yun Wu Yun Wu, Dan-Nan Zhang Yun Wu, Jie-Ming Yang Dan-Nan Zhang, Zhen-Hong Liu Jie-Ming Yang, Xing-Yu Pan Zhen-Hong Liu, Yi-Fan Huang Xing-Yu Pan, Wei Zheng Yi-Fan Huang
Traditional household power dispatching methods are difficult to deal with the complexity of dispatching environment and the randomness of power consumption behavior, and the QLearning algorithm is prone to fall into local optimal solutions and slow convergence when solving problems, this paper proposes a new method based on SA-α-QLearning’s home electricity scheduling strategy solution method. Firstly, a multi-intelligent Markov decision process model is established based on household electrical equipment; then the learning rate of a single value in the QLearning algorithm is replaced by a linear iterative learning rate; finally, a simulated annealing (SA) is used to optimize the QLearning algorithm to solve the model, by taking the Q value change difference as the new solution acceptance probability of Metropoils criterion and the dynamic adjustment temperature reduction coefficient, it alleviates the problem that the QLearing algorithm is easy to fall into the local optimal solution and the convergence speed is slow. Through a large number of comparative experiments, it is proved that the proposed method has a significant improvement in the solution of household electricity dispatching strategy.
{"title":"Household Electricity Scheduling Strategy Solution Based on SA-α-QLearning","authors":"Yun Wu Yun Wu, Dan-Nan Zhang Yun Wu, Jie-Ming Yang Dan-Nan Zhang, Zhen-Hong Liu Jie-Ming Yang, Xing-Yu Pan Zhen-Hong Liu, Yi-Fan Huang Xing-Yu Pan, Wei Zheng Yi-Fan Huang","doi":"10.53106/199115992023063403014","DOIUrl":"https://doi.org/10.53106/199115992023063403014","url":null,"abstract":"\u0000 Traditional household power dispatching methods are difficult to deal with the complexity of dispatching environment and the randomness of power consumption behavior, and the QLearning algorithm is prone to fall into local optimal solutions and slow convergence when solving problems, this paper proposes a new method based on SA-α-QLearning’s home electricity scheduling strategy solution method. Firstly, a multi-intelligent Markov decision process model is established based on household electrical equipment; then the learning rate of a single value in the QLearning algorithm is replaced by a linear iterative learning rate; finally, a simulated annealing (SA) is used to optimize the QLearning algorithm to solve the model, by taking the Q value change difference as the new solution acceptance probability of Metropoils criterion and the dynamic adjustment temperature reduction coefficient, it alleviates the problem that the QLearing algorithm is easy to fall into the local optimal solution and the convergence speed is slow. Through a large number of comparative experiments, it is proved that the proposed method has a significant improvement in the solution of household electricity dispatching strategy.\u0000 \u0000","PeriodicalId":345067,"journal":{"name":"電腦學刊","volume":"5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125557160","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-06-01DOI: 10.53106/199115992023063403002
Jian Wu Jian Wu, Honghui Deng Jian Wu, Fei Cheng Honghui Deng, Hongjun Wang Fei Cheng
There are often residual images of the camera tripod in panoramic images, which may reduce the image quality and deteriorate the post-processing speed. To address this problem, a camera tripod removal network (TRNet) based on generative adversarial network is proposed. As an end-to-end model, the generator is designed to include recognition and reconstruction branches, which reduce the number of parameters and improve the training efficiency by sharing the encoder and correspond to scaffold recognition and texture reconstruction respectively. The recognition branch based on the U-Net structure can effectively identify the tripod area, while the reconstruction branch can brilliantly reconstruct the texture details through an intermediate layer formed by stacking dilated convolution residual blocks. Furthermore, spectral normalized Markov discriminator and multiple combined loss function are adopted to promote global texture consistency and thus result in a better texture filling effect. Finally, a data set of 400 panoramic images is constructed and experimental results on this data set demonstrate the better repair ability of TRNet against other state-of-the-art methods.
{"title":"Camera Tripod Removal Model in Panoramic Images Based on Generative Adversarial Networks","authors":"Jian Wu Jian Wu, Honghui Deng Jian Wu, Fei Cheng Honghui Deng, Hongjun Wang Fei Cheng","doi":"10.53106/199115992023063403002","DOIUrl":"https://doi.org/10.53106/199115992023063403002","url":null,"abstract":"\u0000 There are often residual images of the camera tripod in panoramic images, which may reduce the image quality and deteriorate the post-processing speed. To address this problem, a camera tripod removal network (TRNet) based on generative adversarial network is proposed. As an end-to-end model, the generator is designed to include recognition and reconstruction branches, which reduce the number of parameters and improve the training efficiency by sharing the encoder and correspond to scaffold recognition and texture reconstruction respectively. The recognition branch based on the U-Net structure can effectively identify the tripod area, while the reconstruction branch can brilliantly reconstruct the texture details through an intermediate layer formed by stacking dilated convolution residual blocks. Furthermore, spectral normalized Markov discriminator and multiple combined loss function are adopted to promote global texture consistency and thus result in a better texture filling effect. Finally, a data set of 400 panoramic images is constructed and experimental results on this data set demonstrate the better repair ability of TRNet against other state-of-the-art methods.\u0000 \u0000","PeriodicalId":345067,"journal":{"name":"電腦學刊","volume":"81 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126224310","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-06-01DOI: 10.53106/199115992023063403007
Shuan-Jun Song Shuan-Jun Song, Long-Guang Peng Shuan-Jun Song, Jie Zhang Long-Guang Peng, Zhen Liu Jie Zhang
Aiming at the influence of AGV without considering the working state on task assignment decision in multi-AGV system task assignment, a dynamic task assignment decision method with task completion prediction based on genetic algorithm. When assigning the arrived tasks at each stage, the decision method brings the working AGVs and the idle AGVs into the set of schedulable vehicles at the same time, which expands the scope of the optimal decision, makes the available AGV resources more fully mobilized in the dynamic scheduling process, and improves the efficiency of the whole scheduling system. First, this paper establishes a prediction model for task completion. On this basis, the task assignment decision model of multi-AGV system based on task completion prediction is established, and the coding, fitness function and genetic operation of the genetic algorithm suitable for this problem are designed. Finally, a univariate factor analysis is carried out on the task assignment time interval and the number of AGVs by using an example, which verifies the effectiveness of the task assignment strategy of the multi-AGV system based on task completion prediction. The results show that the genetic algorithm can better solve the task assignment problem with task completion prediction, and can schedule the available AGV resources to a greater extent, which effectively increase the number of tasks completed by the multi-AGV system in one production cycle.
{"title":"A Dynamic Task Assignment Optimization Method for Multi-AGV System Based on Genetic Algorithm","authors":"Shuan-Jun Song Shuan-Jun Song, Long-Guang Peng Shuan-Jun Song, Jie Zhang Long-Guang Peng, Zhen Liu Jie Zhang","doi":"10.53106/199115992023063403007","DOIUrl":"https://doi.org/10.53106/199115992023063403007","url":null,"abstract":"\u0000 Aiming at the influence of AGV without considering the working state on task assignment decision in multi-AGV system task assignment, a dynamic task assignment decision method with task completion prediction based on genetic algorithm. When assigning the arrived tasks at each stage, the decision method brings the working AGVs and the idle AGVs into the set of schedulable vehicles at the same time, which expands the scope of the optimal decision, makes the available AGV resources more fully mobilized in the dynamic scheduling process, and improves the efficiency of the whole scheduling system. First, this paper establishes a prediction model for task completion. On this basis, the task assignment decision model of multi-AGV system based on task completion prediction is established, and the coding, fitness function and genetic operation of the genetic algorithm suitable for this problem are designed. Finally, a univariate factor analysis is carried out on the task assignment time interval and the number of AGVs by using an example, which verifies the effectiveness of the task assignment strategy of the multi-AGV system based on task completion prediction. The results show that the genetic algorithm can better solve the task assignment problem with task completion prediction, and can schedule the available AGV resources to a greater extent, which effectively increase the number of tasks completed by the multi-AGV system in one production cycle.\u0000 \u0000","PeriodicalId":345067,"journal":{"name":"電腦學刊","volume":"22 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123860123","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-06-01DOI: 10.53106/199115992023063403017
Fei Qi Fei Qi, Chen-Qing Wang Fei Qi
For image deblurring, multi-scale approaches have been widely used as deep learning methods recently. In this paper, a novel multi-scale conditional generative adversarial network (CGAN) is proposed to make full use of image features, which outperforms most state-of-the-art methods. We define a generator network and a discriminator network. First of all, we use the multi-scale residual modules proposed in this paper as main feature extraction blocks, and add skip connections to extract multi-scale image features at a finer granularity in the generator network. Secondly, we construct PatchGAN as the discriminator network to enhance the local feature extraction capability. In addition, we combine the adversarial loss based on Wasserstein GAN with gradient penalty (WGAN-GP) theory with the content loss defined by perceptual loss as the total loss function, which is conducive to improving the consistency between the generated images and the ground-truth sharp images in content. The experimental results show that the method in this paper outperforms the state-of-the-art methods in visualization and quantitative results.
{"title":"An End-to-End Multi-Scale Conditional Generative Adversarial Network for Image Deblurring","authors":"Fei Qi Fei Qi, Chen-Qing Wang Fei Qi","doi":"10.53106/199115992023063403017","DOIUrl":"https://doi.org/10.53106/199115992023063403017","url":null,"abstract":"\u0000 For image deblurring, multi-scale approaches have been widely used as deep learning methods recently. In this paper, a novel multi-scale conditional generative adversarial network (CGAN) is proposed to make full use of image features, which outperforms most state-of-the-art methods. We define a generator network and a discriminator network. First of all, we use the multi-scale residual modules proposed in this paper as main feature extraction blocks, and add skip connections to extract multi-scale image features at a finer granularity in the generator network. Secondly, we construct PatchGAN as the discriminator network to enhance the local feature extraction capability. In addition, we combine the adversarial loss based on Wasserstein GAN with gradient penalty (WGAN-GP) theory with the content loss defined by perceptual loss as the total loss function, which is conducive to improving the consistency between the generated images and the ground-truth sharp images in content. The experimental results show that the method in this paper outperforms the state-of-the-art methods in visualization and quantitative results.\u0000 \u0000","PeriodicalId":345067,"journal":{"name":"電腦學刊","volume":"84 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121442925","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-06-01DOI: 10.53106/199115992023063403010
Li Cui Li Cui, Ting-Xuan Chen Li Cui, Ying-Qing Xia Ting-Xuan Chen, Xia Cao Ying-Qing Xia, Ling Wu Xia Cao
Handwritten digit recognition is an active research field. These recognition systems are faced with many challenges, including accuracy, speed and automatic extraction of complex handwriting features. In this paper, a Stacking ensemble learning model based on fusion optimized CNN is proposed, which can be effectively used for handwritten digit recognition. To better extract the features of complex handwritten digital images and maximize the reliability of the model, the Bagging strategy combined with six CNNs is used for feature extraction for the first time, and SVM is used for classification. This not only improves the accuracy and stability of the model, but also effectively avoids over-fitting. In addition, a fusion optimization algorithm based on Adam and SGD is proposed to solve the problem that CNN falls into local optimum due to a large number of iterations. During the process of training, ASCNN can not only speed up the convergence rate in the early stage, but also reduce the oscillation phenomenon in the late stage. Extensive experimental results on the well-known MNIST and USPS handwriting image datasets demonstrate the effectiveness of the proposed model.
{"title":"Ensemble Learning Network for Handwritten Digit Recognition Based on Fusion Optimized CNN","authors":"Li Cui Li Cui, Ting-Xuan Chen Li Cui, Ying-Qing Xia Ting-Xuan Chen, Xia Cao Ying-Qing Xia, Ling Wu Xia Cao","doi":"10.53106/199115992023063403010","DOIUrl":"https://doi.org/10.53106/199115992023063403010","url":null,"abstract":"\u0000 Handwritten digit recognition is an active research field. These recognition systems are faced with many challenges, including accuracy, speed and automatic extraction of complex handwriting features. In this paper, a Stacking ensemble learning model based on fusion optimized CNN is proposed, which can be effectively used for handwritten digit recognition. To better extract the features of complex handwritten digital images and maximize the reliability of the model, the Bagging strategy combined with six CNNs is used for feature extraction for the first time, and SVM is used for classification. This not only improves the accuracy and stability of the model, but also effectively avoids over-fitting. In addition, a fusion optimization algorithm based on Adam and SGD is proposed to solve the problem that CNN falls into local optimum due to a large number of iterations. During the process of training, ASCNN can not only speed up the convergence rate in the early stage, but also reduce the oscillation phenomenon in the late stage. Extensive experimental results on the well-known MNIST and USPS handwriting image datasets demonstrate the effectiveness of the proposed model.\u0000 \u0000","PeriodicalId":345067,"journal":{"name":"電腦學刊","volume":"54 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122799317","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-06-01DOI: 10.53106/199115992023063403015
Hao Xu Hao Xu, Xian-Bin Wan Hao Xu, Hui Liu Xian-Bin Wan
With the advent of the industrial Internet era and rapid traffic growth, network optimization is increasingly needed, and network optimization starts with knowing QoS-related metrics. In this paper, we use a machine learning approach in a theoretical SDN architecture, using traffic as the input to a machine learning model, to predict network QoS metrics, focusing on network jitter and packet loss rate. We built a LAN and deployed a time server on the LAN in order to make the time of the devices on the LAN highly consistent. Experiments were conducted under this LAN to obtain data sets about traffic and QoS metrics. Then, we used the completed trained machine learning model to predict the network jitter and packet loss rate using traffic as the input to the machine learning model. The highest R² values for the prediction of network jitter and packet loss reached 0.9996 and 0.939, respectively. The experiments show that a suitable machine learning model is able to predict network jitter and packet loss rate relatively accurately for a specific network topology.
{"title":"A Machine Learning Based Approach to QoS Metrics Prediction in the Context of SDN","authors":"Hao Xu Hao Xu, Xian-Bin Wan Hao Xu, Hui Liu Xian-Bin Wan","doi":"10.53106/199115992023063403015","DOIUrl":"https://doi.org/10.53106/199115992023063403015","url":null,"abstract":"\u0000 With the advent of the industrial Internet era and rapid traffic growth, network optimization is increasingly needed, and network optimization starts with knowing QoS-related metrics. In this paper, we use a machine learning approach in a theoretical SDN architecture, using traffic as the input to a machine learning model, to predict network QoS metrics, focusing on network jitter and packet loss rate. We built a LAN and deployed a time server on the LAN in order to make the time of the devices on the LAN highly consistent. Experiments were conducted under this LAN to obtain data sets about traffic and QoS metrics. Then, we used the completed trained machine learning model to predict the network jitter and packet loss rate using traffic as the input to the machine learning model. The highest R² values for the prediction of network jitter and packet loss reached 0.9996 and 0.939, respectively. The experiments show that a suitable machine learning model is able to predict network jitter and packet loss rate relatively accurately for a specific network topology.\u0000 \u0000","PeriodicalId":345067,"journal":{"name":"電腦學刊","volume":"8 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114311285","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：