Pub Date : 2023-08-01DOI: 10.53106/199115992023083404013
Hai-Jun Shen Hai-Jun Shen, Qing-Hong Wang Hai-Jun Shen, Rui Fan Qing-Hong Wang, Wei-Min Liu Rui Fan
In the process of photovoltaic power generation, maximum power point tracking is an important method to improve the efficiency of photovoltaic power generation. Under the actual local shadow condition, the maximum power point of Photovoltaic system fluctuates. For this reason, this paper establishes the mathematical model and output characteristic equation of photovoltaic cells according to the actual application, and then uses the adaptive inertia weight Particle Swarm Optimization algorithm to solve the problem of slow search speed and low accuracy in the process of maximum power point tracking. After optimization, the method proposed in this paper can significantly improve the tracking effect efficiency, and the optimization results in real operation scenarios can improve the photovoltaic cell power generation efficiency by 21.3%, which proves the effectiveness of the algorithm.
{"title":"Intelligent Optimization Control Method for Photovoltaic Power Generation Systems Under Shadow Occlusion Conditions","authors":"Hai-Jun Shen Hai-Jun Shen, Qing-Hong Wang Hai-Jun Shen, Rui Fan Qing-Hong Wang, Wei-Min Liu Rui Fan","doi":"10.53106/199115992023083404013","DOIUrl":"https://doi.org/10.53106/199115992023083404013","url":null,"abstract":"\u0000 In the process of photovoltaic power generation, maximum power point tracking is an important method to improve the efficiency of photovoltaic power generation. Under the actual local shadow condition, the maximum power point of Photovoltaic system fluctuates. For this reason, this paper establishes the mathematical model and output characteristic equation of photovoltaic cells according to the actual application, and then uses the adaptive inertia weight Particle Swarm Optimization algorithm to solve the problem of slow search speed and low accuracy in the process of maximum power point tracking. After optimization, the method proposed in this paper can significantly improve the tracking effect efficiency, and the optimization results in real operation scenarios can improve the photovoltaic cell power generation efficiency by 21.3%, which proves the effectiveness of the algorithm.\u0000 \u0000","PeriodicalId":345067,"journal":{"name":"電腦學刊","volume":"4 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130522840","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 : 2023-08-01DOI: 10.53106/199115992023083404004
Chuang Ma Chuang Ma, Helong Xia Chuang Ma
Since multiple people share the same name in the real world, this will cause performance degradation to academic search systems and lead to misattribution of publications. The author name disambiguation algorithm has not yet to be well solved. In this paper, we propose a disambiguation method that combines heterogeneous graph-based and improved label propagation, first we construct a publication heterogeneous graph network, then graph neural networks is applied to aggregate the nodes representation and relation types, finally combined with the improved label propagation algorithm to realize clustering. The task of author name disambiguation is completed to improve the retrieval performance. Experimental results on two public datasets show that our method was improved by 2.8% and 4.9% over the suboptimal method, respectively. Our method can effectively reduce the number of publications returning the wrong author and improve the performance of the academic retrieval system.
{"title":"Author Name Disambiguation Based on Heterogeneous Graph","authors":"Chuang Ma Chuang Ma, Helong Xia Chuang Ma","doi":"10.53106/199115992023083404004","DOIUrl":"https://doi.org/10.53106/199115992023083404004","url":null,"abstract":"\u0000 Since multiple people share the same name in the real world, this will cause performance degradation to academic search systems and lead to misattribution of publications. The author name disambiguation algorithm has not yet to be well solved. In this paper, we propose a disambiguation method that combines heterogeneous graph-based and improved label propagation, first we construct a publication heterogeneous graph network, then graph neural networks is applied to aggregate the nodes representation and relation types, finally combined with the improved label propagation algorithm to realize clustering. The task of author name disambiguation is completed to improve the retrieval performance. Experimental results on two public datasets show that our method was improved by 2.8% and 4.9% over the suboptimal method, respectively. Our method can effectively reduce the number of publications returning the wrong author and improve the performance of the academic retrieval system.\u0000 \u0000","PeriodicalId":345067,"journal":{"name":"電腦學刊","volume":"208 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131401425","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 : 2023-08-01DOI: 10.53106/199115992023083404001
Zhan-Peng Cui Zhan-Peng Cui
In contrast enhancement of fuzzy defect image, details loss and noise expansion are east to occur, which brings difficulties to subsequent image analysis and defect recognition. Therefore, a fuzzy defect image restoration and enhancement method based on neural network is proposed. A double fusion neural network composed of a depth generation network and a discrimination network is designed. The residual of the denoised fuzzy image and the real image is output by the network, which is input into the discrimination network together with the real image, and the difference between the two is judged by the total loss function. To solve the problem of pixel coordinate value of fuzzy defect image, neural network is used to build a fast correction algorithm. Therefore, a fuzzy image restoration and enhancement method based on neural network is proposed to improve the image quality. By reconstructing the resolution of fuzzy defect image, a hierarchical enhancement method of fuzzy defect image region is constructed to achieve fuzzy defect image restoration and enhancement. The results show that the proposed method has high image processing ability in restoration and enhancement of fuzzy defect images. The fitting value of neural network is 0.92, which is significantly higher than that of the other two methods, indicating that the image restoration and enhancement method based on neural network has higher accuracy. Therefore, the restoration and enhancement method of fuzzy defect image based on neural network has a good restoration and enhancement effect, and can effectively meet the actual needs of people for high-quality images.
{"title":"Restoration and Enhancement of Fuzzy Defect Image Based on Neural Network","authors":"Zhan-Peng Cui Zhan-Peng Cui","doi":"10.53106/199115992023083404001","DOIUrl":"https://doi.org/10.53106/199115992023083404001","url":null,"abstract":"\u0000 In contrast enhancement of fuzzy defect image, details loss and noise expansion are east to occur, which brings difficulties to subsequent image analysis and defect recognition. Therefore, a fuzzy defect image restoration and enhancement method based on neural network is proposed. A double fusion neural network composed of a depth generation network and a discrimination network is designed. The residual of the denoised fuzzy image and the real image is output by the network, which is input into the discrimination network together with the real image, and the difference between the two is judged by the total loss function. To solve the problem of pixel coordinate value of fuzzy defect image, neural network is used to build a fast correction algorithm. Therefore, a fuzzy image restoration and enhancement method based on neural network is proposed to improve the image quality. By reconstructing the resolution of fuzzy defect image, a hierarchical enhancement method of fuzzy defect image region is constructed to achieve fuzzy defect image restoration and enhancement. The results show that the proposed method has high image processing ability in restoration and enhancement of fuzzy defect images. The fitting value of neural network is 0.92, which is significantly higher than that of the other two methods, indicating that the image restoration and enhancement method based on neural network has higher accuracy. Therefore, the restoration and enhancement method of fuzzy defect image based on neural network has a good restoration and enhancement effect, and can effectively meet the actual needs of people for high-quality images.\u0000 \u0000","PeriodicalId":345067,"journal":{"name":"電腦學刊","volume":"11 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131376871","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 : 2023-08-01DOI: 10.53106/199115992023083404016
Jian-Ming Yu Jian-Ming Yu, Ke Zhang Jian-Ming Yu, Jian-Zhong Zhang Ke Zhang, Feng Xue Jian-Zhong Zhang, Wei Liu Feng Xue
This article mainly focuses on the damage assessment of buildings after earthquakes. Firstly, a structural damage model was established based on most reinforced concrete buildings and described using a function. Then, a BP neural network was used to solve the function. Traditional neural networks are prone to falling into local optima. Therefore, in order to improve the performance of neural networks, cross fusion with genetic algorithms is used to avoid falling into local optima, Improve the efficiency of the algorithm. Finally, through experimental verification, the proposed method can quickly evaluate the damage of building structures, with an accuracy rate of 97%.
{"title":"Method for Predicting and Evaluating Post Earthquake Damage of Urban Buildings Based on Artificial Intelligence Algorithms","authors":"Jian-Ming Yu Jian-Ming Yu, Ke Zhang Jian-Ming Yu, Jian-Zhong Zhang Ke Zhang, Feng Xue Jian-Zhong Zhang, Wei Liu Feng Xue","doi":"10.53106/199115992023083404016","DOIUrl":"https://doi.org/10.53106/199115992023083404016","url":null,"abstract":"\u0000 This article mainly focuses on the damage assessment of buildings after earthquakes. Firstly, a structural damage model was established based on most reinforced concrete buildings and described using a function. Then, a BP neural network was used to solve the function. Traditional neural networks are prone to falling into local optima. Therefore, in order to improve the performance of neural networks, cross fusion with genetic algorithms is used to avoid falling into local optima, Improve the efficiency of the algorithm. Finally, through experimental verification, the proposed method can quickly evaluate the damage of building structures, with an accuracy rate of 97%.\u0000 \u0000","PeriodicalId":345067,"journal":{"name":"電腦學刊","volume":"52 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133595597","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}
With the development of digital education, colleges and universities are facing challenges in managing and utilizing digital resources. As an effective way of knowledge organization and representation, Knowledge graph can transform digital resources into structured entities, attributes and relationship forms, and provide personalized learning support and resource recommendation. By analyzing the characteristics, construction status and existing problems of digital resources in vocational colleges, this paper puts forward the overall plan and methods of digital resources knowledge graph construction, including data collection and sorting, knowledge graph construction, verification and optimization, and gives the general process design of knowledge graph construction; Then, this paper discusses the typical application scenarios of digital resources knowledge graph in vocational colleges, such as learning resource recommendation, path planning, teaching assistance and resource sharing; Finally, the challenges in this field, including data quality, knowledge graph updating and maintenance, privacy and security, are discussed, and the future trends and research directions are prospected. The research results are of great significance for promoting the informatization and intelligent development of vocational college education.
{"title":"Research on the Construction and Application of Knowledge Graph of Digital Resources in Vocational Colleges","authors":"Yongjun Wei Yongjun Wei, Qiumi Qin Yongjun Wei, Caisen Chen Qiumi Qin, Xiaoyu Liu Caisen Chen","doi":"10.53106/199115992023083404017","DOIUrl":"https://doi.org/10.53106/199115992023083404017","url":null,"abstract":"\u0000 With the development of digital education, colleges and universities are facing challenges in managing and utilizing digital resources. As an effective way of knowledge organization and representation, Knowledge graph can transform digital resources into structured entities, attributes and relationship forms, and provide personalized learning support and resource recommendation. By analyzing the characteristics, construction status and existing problems of digital resources in vocational colleges, this paper puts forward the overall plan and methods of digital resources knowledge graph construction, including data collection and sorting, knowledge graph construction, verification and optimization, and gives the general process design of knowledge graph construction; Then, this paper discusses the typical application scenarios of digital resources knowledge graph in vocational colleges, such as learning resource recommendation, path planning, teaching assistance and resource sharing; Finally, the challenges in this field, including data quality, knowledge graph updating and maintenance, privacy and security, are discussed, and the future trends and research directions are prospected. The research results are of great significance for promoting the informatization and intelligent development of vocational college education.\u0000 \u0000","PeriodicalId":345067,"journal":{"name":"電腦學刊","volume":"35 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128157658","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}
In order to solve the problems of missing detection due to overlap and occlusion of contraband in X-ray images and low accuracy of small object detection, we propose a single-stage object detection framework based on PP-YOLO. Compared with the traditional prohibited item detection algorithm, it adds CBAM module on the basis of ResNet50 feature extraction network to enhance the feature extraction ability; For increasing the detail features of the detection layer, MSF module is introduced into FPN, which fuses the feature map with accurate position information in the lower layer and the feature map with strong semantic information in the higher layer; The partial convolution of backbone is improved to CompConv to accelerate the processing speed of the model, which compresses the network structure and improves the inference speed without losing performance. The results show that the mAP of the improved network for prohibited item detection is 94.67%, and the processing speed reaches 45 FPS, which means that the recognition accuracy and reasoning speed of this method have been improved to some extent.
{"title":"X-ray Image Prohibited Item Detection Algorithm Based on Improved PP-YOLO","authors":"Ji-kai Zhang Ji-kai Zhang, Yue Liu Ji-Kai Zhang, Xiao-Qi Lv Yue Liu, Yong Liang Xiao-Qi Lv","doi":"10.53106/199115992023083404005","DOIUrl":"https://doi.org/10.53106/199115992023083404005","url":null,"abstract":"\u0000 In order to solve the problems of missing detection due to overlap and occlusion of contraband in X-ray images and low accuracy of small object detection, we propose a single-stage object detection framework based on PP-YOLO. Compared with the traditional prohibited item detection algorithm, it adds CBAM module on the basis of ResNet50 feature extraction network to enhance the feature extraction ability; For increasing the detail features of the detection layer, MSF module is introduced into FPN, which fuses the feature map with accurate position information in the lower layer and the feature map with strong semantic information in the higher layer; The partial convolution of backbone is improved to CompConv to accelerate the processing speed of the model, which compresses the network structure and improves the inference speed without losing performance. The results show that the mAP of the improved network for prohibited item detection is 94.67%, and the processing speed reaches 45 FPS, which means that the recognition accuracy and reasoning speed of this method have been improved to some extent.\u0000 \u0000","PeriodicalId":345067,"journal":{"name":"電腦學刊","volume":"18 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121534974","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}
Due to the large changes in dynamic image sequence frames and the complex detection scene, it is difficult to accurately detect moving objects. Therefore, the study proposes a moving target detection algorithm based on artificial neural network. First, the algorithm performs standardized grayscale processing and gamma correction processing on the dynamic image to eliminate the noise interference of the dynamic image. After that, the model calculates the gradient of the dynamic image in order to complete the feature extraction of the dynamic image. Then, according to the result of hog feature extraction, the study adopts the inter-frame calculation method to update the background of the dynamic image. Finally, the principle and structure of the neural network are analyzed experimentally, and a channel attention mechanism is introduced to train dynamic image sequences to obtain MTD results. Experimental results show that the proposed algorithm achieves higher accuracy in MTD than conventional detection algorithms. The calculation efficiency of the algorithm in this paper has significant advantages, and the average detection time is 3.69515ms, which can meet the real-time requirements of MTD.
{"title":"Moving Target Detection Algorithm for Dynamic Image Sequences on The Basis of Artificial Neural Network","authors":"Jia-Min Zhang Jia-Min Zhang, Yan-Xia Chen Jia-Min Zhang","doi":"10.53106/199115992023083404006","DOIUrl":"https://doi.org/10.53106/199115992023083404006","url":null,"abstract":"\u0000 Due to the large changes in dynamic image sequence frames and the complex detection scene, it is difficult to accurately detect moving objects. Therefore, the study proposes a moving target detection algorithm based on artificial neural network. First, the algorithm performs standardized grayscale processing and gamma correction processing on the dynamic image to eliminate the noise interference of the dynamic image. After that, the model calculates the gradient of the dynamic image in order to complete the feature extraction of the dynamic image. Then, according to the result of hog feature extraction, the study adopts the inter-frame calculation method to update the background of the dynamic image. Finally, the principle and structure of the neural network are analyzed experimentally, and a channel attention mechanism is introduced to train dynamic image sequences to obtain MTD results. Experimental results show that the proposed algorithm achieves higher accuracy in MTD than conventional detection algorithms. The calculation efficiency of the algorithm in this paper has significant advantages, and the average detection time is 3.69515ms, which can meet the real-time requirements of MTD.\u0000 \u0000","PeriodicalId":345067,"journal":{"name":"電腦學刊","volume":"23 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125729519","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}
Estimating human motion posture can provide important data for intelligent monitoring systems, human-computer interaction, motion capture, and other fields. However, the traditional human motion posture estimation algorithm is difficult to achieve the goal of fast estimation of human motion posture. To address the problems of traditional algorithms, in the paper, we propose an estimation algorithm for human motion posture using improved deep reinforcement learning. First, the double deep Q network is constructed to improve the deep reinforcement learning algorithm. The improved deep reinforcement learning algorithm is used to locate the human motion posture coordinates and improve the effectiveness of bone point calibration. Second, the human motion posture analysis generative adversarial networks are constructed to realize the automatic recognition and analysis of human motion posture. Finally, using the preset human motion posture label, combined with the undirected graph model of the human, the human motion posture estimation is completed, and the precise estimation algorithm of the human motion posture is realized. Experiments are performed based on MPII Human Pose data set and HiEve data set. The results show that the proposed algorithm has higher positioning accuracy of joint nodes. The recognition effect of bone joint points is better, and the average is about 1.45%. The average posture accuracy is up to 98.2%, and the average joint point similarity is high. Therefore, it is proved that the proposed method has high application value in human-computer interaction, human motion capture and other fields.
{"title":"Estimation on Human Motion Posture using Improved Deep Reinforcement Learning","authors":"Wenjing Ma Wenjing Ma, Jianguang Zhao Wenjing Ma, Guangquan Zhu Jianguang Zhao","doi":"10.53106/199115992023083404008","DOIUrl":"https://doi.org/10.53106/199115992023083404008","url":null,"abstract":"\u0000 Estimating human motion posture can provide important data for intelligent monitoring systems, human-computer interaction, motion capture, and other fields. However, the traditional human motion posture estimation algorithm is difficult to achieve the goal of fast estimation of human motion posture. To address the problems of traditional algorithms, in the paper, we propose an estimation algorithm for human motion posture using improved deep reinforcement learning. First, the double deep Q network is constructed to improve the deep reinforcement learning algorithm. The improved deep reinforcement learning algorithm is used to locate the human motion posture coordinates and improve the effectiveness of bone point calibration. Second, the human motion posture analysis generative adversarial networks are constructed to realize the automatic recognition and analysis of human motion posture. Finally, using the preset human motion posture label, combined with the undirected graph model of the human, the human motion posture estimation is completed, and the precise estimation algorithm of the human motion posture is realized. Experiments are performed based on MPII Human Pose data set and HiEve data set. The results show that the proposed algorithm has higher positioning accuracy of joint nodes. The recognition effect of bone joint points is better, and the average is about 1.45%. The average posture accuracy is up to 98.2%, and the average joint point similarity is high. Therefore, it is proved that the proposed method has high application value in human-computer interaction, human motion capture and other fields.\u0000 \u0000","PeriodicalId":345067,"journal":{"name":"電腦學刊","volume":"51 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133273327","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 : 2023-08-01DOI: 10.53106/199115992023083404014
Xue-Jun Liu Xue-Jun Liu, Wen-Hui Wang Xue-Jun Liu, Yong Yan Wen-Hui Wang, Zhong-Ji Cui Yong Yan, Yun Sha Zhong-Ji Cui, Yi-Nan Jiang Yun Sha
In the monitoring the safety status of hazardous chemical warehouses by three-dimensional re-construction of deep camera point clouds, there are classification difficulties such as large space, sparse distribution of point clouds in cargo images, and similar distribution in low dimensions. Based on the above problem, a point cloud recognition method based on multi-head attention mechanism is proposed. The algorithm first normalizes the distribution of the point cloud data set through the affine transformation algorithm to solve the problem of sparse distribution. Then, the high-dimensional feature map is obtained by fusing the data down-sampling and curve feature aggregation algorithms to solve the problem of low-dimensional distribution approximation. The feature map is then encoded using a multi-head self-attention encoder to obtain features under different heads, which are then merged into a feature map. Finally, a multi-layer fully connected neural network is used as the decoder to decode the feature map into the final object classification. Comparative experiments were performed on the ModelNet40 dataset and the self-built dataset of warehouse goods, and the results showed that the accuracy of this paper was improved by 0.5% to 7.8% compared with that of other classification algorithms.
{"title":"A Point Cloud Classification Method and Its Applications Based on Multi-Head Self-Attention","authors":"Xue-Jun Liu Xue-Jun Liu, Wen-Hui Wang Xue-Jun Liu, Yong Yan Wen-Hui Wang, Zhong-Ji Cui Yong Yan, Yun Sha Zhong-Ji Cui, Yi-Nan Jiang Yun Sha","doi":"10.53106/199115992023083404014","DOIUrl":"https://doi.org/10.53106/199115992023083404014","url":null,"abstract":"\u0000 In the monitoring the safety status of hazardous chemical warehouses by three-dimensional re-construction of deep camera point clouds, there are classification difficulties such as large space, sparse distribution of point clouds in cargo images, and similar distribution in low dimensions. Based on the above problem, a point cloud recognition method based on multi-head attention mechanism is proposed. The algorithm first normalizes the distribution of the point cloud data set through the affine transformation algorithm to solve the problem of sparse distribution. Then, the high-dimensional feature map is obtained by fusing the data down-sampling and curve feature aggregation algorithms to solve the problem of low-dimensional distribution approximation. The feature map is then encoded using a multi-head self-attention encoder to obtain features under different heads, which are then merged into a feature map. Finally, a multi-layer fully connected neural network is used as the decoder to decode the feature map into the final object classification. Comparative experiments were performed on the ModelNet40 dataset and the self-built dataset of warehouse goods, and the results showed that the accuracy of this paper was improved by 0.5% to 7.8% compared with that of other classification algorithms.\u0000 \u0000","PeriodicalId":345067,"journal":{"name":"電腦學刊","volume":"254 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133115290","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}
This article focuses on the current situation of large assembly errors, easy omissions and errors in the mechanical assembly process. Computer vision is introduced in the assembly process, and visual images are used to estimate assembly errors, thereby improving assembly accuracy. To this end, through improvements to the neural network, the addition of attention and measurement mechanisms, the network’s ability to extract and distinguish features from assembly images has been improved. Finally, deep learning algorithms are used to estimate assembly features in the image. Finally, simulation experiments have shown that the algorithm proposed in this paper can achieve 94.7% improvement in assembly accuracy and error estimation accuracy.
{"title":"A Method for Assembly Accuracy Detection and Intelligent Error Estimation Based on Computer Vision","authors":"Dan-Dan Cui Dan-Dan Cui, Chao Xu Dan-Dan Cui, Hong-Chao Zhou Chao Xu","doi":"10.53106/199115992023083404010","DOIUrl":"https://doi.org/10.53106/199115992023083404010","url":null,"abstract":"\u0000 This article focuses on the current situation of large assembly errors, easy omissions and errors in the mechanical assembly process. Computer vision is introduced in the assembly process, and visual images are used to estimate assembly errors, thereby improving assembly accuracy. To this end, through improvements to the neural network, the addition of attention and measurement mechanisms, the network’s ability to extract and distinguish features from assembly images has been improved. Finally, deep learning algorithms are used to estimate assembly features in the image. Finally, simulation experiments have shown that the algorithm proposed in this paper can achieve 94.7% improvement in assembly accuracy and error estimation accuracy.\u0000 \u0000","PeriodicalId":345067,"journal":{"name":"電腦學刊","volume":"18 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116997813","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：