Pub Date : 2022-10-21DOI: 10.1109/INMIC56986.2022.9972959
Shuja Ali, Muhammad Hanzla, A. Rafique
Traffic monitoring plays a vital role in the current world. Previously, stationary data collectors such as video cameras and induction loops were employed for this task. However, the availability of unmanned aerial vehicles (UAV) has opened up new horizons for this task and numerous research projects are being conducted in this field. But object detection and tracking become a challenging task in the case of aerial images due to the presence of high density of objects, challenging view angles, different illumination changes, and varying altitudes of the drone. In this paper, we propose a method for detecting vehicles and also tracking them through the use of cascade classifier and centroid tracking. We have also incorporated georeferencing and coregistration of acquired images and then proceeded on to extract lanes. After segmenting out the region of interest, we proceeded with the detection and tracking tasks.
{"title":"Vehicle Detection and Tracking from UAV Imagery via Cascade Classifier","authors":"Shuja Ali, Muhammad Hanzla, A. Rafique","doi":"10.1109/INMIC56986.2022.9972959","DOIUrl":"https://doi.org/10.1109/INMIC56986.2022.9972959","url":null,"abstract":"Traffic monitoring plays a vital role in the current world. Previously, stationary data collectors such as video cameras and induction loops were employed for this task. However, the availability of unmanned aerial vehicles (UAV) has opened up new horizons for this task and numerous research projects are being conducted in this field. But object detection and tracking become a challenging task in the case of aerial images due to the presence of high density of objects, challenging view angles, different illumination changes, and varying altitudes of the drone. In this paper, we propose a method for detecting vehicles and also tracking them through the use of cascade classifier and centroid tracking. We have also incorporated georeferencing and coregistration of acquired images and then proceeded on to extract lanes. After segmenting out the region of interest, we proceeded with the detection and tracking tasks.","PeriodicalId":404424,"journal":{"name":"2022 24th International Multitopic Conference (INMIC)","volume":"5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-10-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125334778","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 : 2022-10-21DOI: 10.1109/INMIC56986.2022.9972908
N. Perwaiz, M. Shahzad, M. Fraz
Person re-identification (Re-ID) is a process to re-identify a person if he has been already seen by a camera network. Since start the convolutional neural networks (CNNs) are dominantly being used to solve the person Re-ID problem. The default limitation of CNNs i.e., local receptive field, prohibits the network to learn the distinctive global dependencies at initial layers. This study proposes a self-attention based deep architecture that learns global dependencies at each network layer to address CNN's limitation. Additionally, the introduction of a novel contextual learning module called Attention Drop Block (ADB) supports learning of less attentive areas of an image as well. The proposed model is evaluated on two public Re-ID benchmarks Market1501 and DukeMTMC-ReID, and outperformed all CNN baseline Re-ID models. The implementation and trained models are made publicly available at https://git.io/JYRE3.
{"title":"Unveiling the Potential of Vision Transformer Architecture for Person Re-identification","authors":"N. Perwaiz, M. Shahzad, M. Fraz","doi":"10.1109/INMIC56986.2022.9972908","DOIUrl":"https://doi.org/10.1109/INMIC56986.2022.9972908","url":null,"abstract":"Person re-identification (Re-ID) is a process to re-identify a person if he has been already seen by a camera network. Since start the convolutional neural networks (CNNs) are dominantly being used to solve the person Re-ID problem. The default limitation of CNNs i.e., local receptive field, prohibits the network to learn the distinctive global dependencies at initial layers. This study proposes a self-attention based deep architecture that learns global dependencies at each network layer to address CNN's limitation. Additionally, the introduction of a novel contextual learning module called Attention Drop Block (ADB) supports learning of less attentive areas of an image as well. The proposed model is evaluated on two public Re-ID benchmarks Market1501 and DukeMTMC-ReID, and outperformed all CNN baseline Re-ID models. The implementation and trained models are made publicly available at https://git.io/JYRE3.","PeriodicalId":404424,"journal":{"name":"2022 24th International Multitopic Conference (INMIC)","volume":"67 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-10-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128753508","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 : 2022-10-21DOI: 10.1109/INMIC56986.2022.9972951
Rabia Fazal Dad, S. Saleem
Because of increasing energy demand and environmental concerns, use of renewable energy resources has increased during the past two decades. Wind, solar, hydro power, biomass, and hydrogen fuel cells are some common renewable energy resources. Due to their complementary nature and inherit intermittency, renewable energy resources are often combined along with a battery backup to form an off-grid or grid-connected hybrid system, also known as renewable micro grid. Due to the cost and reliability concerns, proper sizing of such system is very crucial at the design stage. This paper reviews various optimization techniques for the optimal sizing of a renewable micro grid. Moreover, based on this review a hybrid strategy that combines various optimization techniques is recommended to optimally size and increase overall efficiency of a renewable micro grid.
{"title":"A Comparative Study of Various Optimization Techniques to Size a Hybrid Renewable Energy System","authors":"Rabia Fazal Dad, S. Saleem","doi":"10.1109/INMIC56986.2022.9972951","DOIUrl":"https://doi.org/10.1109/INMIC56986.2022.9972951","url":null,"abstract":"Because of increasing energy demand and environmental concerns, use of renewable energy resources has increased during the past two decades. Wind, solar, hydro power, biomass, and hydrogen fuel cells are some common renewable energy resources. Due to their complementary nature and inherit intermittency, renewable energy resources are often combined along with a battery backup to form an off-grid or grid-connected hybrid system, also known as renewable micro grid. Due to the cost and reliability concerns, proper sizing of such system is very crucial at the design stage. This paper reviews various optimization techniques for the optimal sizing of a renewable micro grid. Moreover, based on this review a hybrid strategy that combines various optimization techniques is recommended to optimally size and increase overall efficiency of a renewable micro grid.","PeriodicalId":404424,"journal":{"name":"2022 24th International Multitopic Conference (INMIC)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-10-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129214734","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 : 2022-10-21DOI: 10.1109/INMIC56986.2022.9972928
Muhammad Mahad Khaliq, R. Mumtaz
Normalized Difference Vegetation Index (NDVI) has been one of the key scales for monitoring multiple plant parameters, but satellite imagery is never up to date, which makes it difficult to get readings for the recent situation of field crops. Doing so with Unmanned Aerial System, drone, in this case, is an intricate task, but with its advantages which include timely and effective measurements with the least errors to be fixed in post-processing of data. Before this, NDVI has been calculated using an Unmanned Aerial System, but the problem of the low resolution of the imagery always lingers. With the recent advancement of generated adversarial networks, the up-scaling of images has been made possible, which, if done with the right model, rules out the need for upgrading the camera hardware that is never cost-effective. We have come up with the solution of calculating the vegetation index of field crops by implementing Enhanced Super-Resolution Generated Adversarial Networks with drone imagery to calculate the vegetation index of crop fields. A simple near-infrared spectrum camera is usually not capable of producing a higher resolution image, by implementing the aforementioned generated adversarial network, we have been able to calculate vegetation index for a comparably much higher resolution image without upgrading with sophisticated hardware. We were able to perform the calculations for more pixels (12952) against the same area yielded an output value of 0.829 as compared to 0.828 in the case of low-resolution imagery (546416 pixels). The averaged values for red and near-infrared pixels showed changes from 32.337 to 30.264 for red, and from 189.168 to 182.1656 for near-infrared pixels. The results produced with this technique are different from those generated using original images which account for a new gateway in the calculation of the NDVI.
{"title":"Enhancing NDVI Calculation of Low-Resolution Imagery using ESRGANs","authors":"Muhammad Mahad Khaliq, R. Mumtaz","doi":"10.1109/INMIC56986.2022.9972928","DOIUrl":"https://doi.org/10.1109/INMIC56986.2022.9972928","url":null,"abstract":"Normalized Difference Vegetation Index (NDVI) has been one of the key scales for monitoring multiple plant parameters, but satellite imagery is never up to date, which makes it difficult to get readings for the recent situation of field crops. Doing so with Unmanned Aerial System, drone, in this case, is an intricate task, but with its advantages which include timely and effective measurements with the least errors to be fixed in post-processing of data. Before this, NDVI has been calculated using an Unmanned Aerial System, but the problem of the low resolution of the imagery always lingers. With the recent advancement of generated adversarial networks, the up-scaling of images has been made possible, which, if done with the right model, rules out the need for upgrading the camera hardware that is never cost-effective. We have come up with the solution of calculating the vegetation index of field crops by implementing Enhanced Super-Resolution Generated Adversarial Networks with drone imagery to calculate the vegetation index of crop fields. A simple near-infrared spectrum camera is usually not capable of producing a higher resolution image, by implementing the aforementioned generated adversarial network, we have been able to calculate vegetation index for a comparably much higher resolution image without upgrading with sophisticated hardware. We were able to perform the calculations for more pixels (12952) against the same area yielded an output value of 0.829 as compared to 0.828 in the case of low-resolution imagery (546416 pixels). The averaged values for red and near-infrared pixels showed changes from 32.337 to 30.264 for red, and from 189.168 to 182.1656 for near-infrared pixels. The results produced with this technique are different from those generated using original images which account for a new gateway in the calculation of the NDVI.","PeriodicalId":404424,"journal":{"name":"2022 24th International Multitopic Conference (INMIC)","volume":"160 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-10-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116159942","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 : 2022-10-21DOI: 10.1109/INMIC56986.2022.9972938
Muhammad Farjad Ali Raza, M. Naeem
Word prediction, spelling error correction and finding similarity between words are very useful features in any language. The Saraiki is one of the popular languages spoken in Pakistan. To the best of our knowledge, very little work has been done in the literature for word prediction, spell correction and finding similar words for the Saraiki language. In this paper we address these issues by presenting a novel approach for word prediction, finding similar words, and spell correction in the Saraiki language. To achieve this, we used CBOW and Skip-Gram for the vectorization of the Saraiki language. From our results, we achieved word prediction accuracy of 24 % in case of word2vec while 29 % in case of the fastText. In case of word similarity, we achieved similarity score equal to 0.35, and 0.39 for word2vec CBOW and word2vec Skip-Gram respectively and similarity score of 0.35 and 0.41 for the fastText CBOW and the fastText Skip-Gram respectively. Our spell correction results show that as we increase wrong characters in words, the accuracy gets decreased. For sentence-level word prediction, we achieved accuracy of 63% in case of RoBERTa and 58% for distilled respectively.
{"title":"Saraiki Language Word Prediction And Spell Correction Framework","authors":"Muhammad Farjad Ali Raza, M. Naeem","doi":"10.1109/INMIC56986.2022.9972938","DOIUrl":"https://doi.org/10.1109/INMIC56986.2022.9972938","url":null,"abstract":"Word prediction, spelling error correction and finding similarity between words are very useful features in any language. The Saraiki is one of the popular languages spoken in Pakistan. To the best of our knowledge, very little work has been done in the literature for word prediction, spell correction and finding similar words for the Saraiki language. In this paper we address these issues by presenting a novel approach for word prediction, finding similar words, and spell correction in the Saraiki language. To achieve this, we used CBOW and Skip-Gram for the vectorization of the Saraiki language. From our results, we achieved word prediction accuracy of 24 % in case of word2vec while 29 % in case of the fastText. In case of word similarity, we achieved similarity score equal to 0.35, and 0.39 for word2vec CBOW and word2vec Skip-Gram respectively and similarity score of 0.35 and 0.41 for the fastText CBOW and the fastText Skip-Gram respectively. Our spell correction results show that as we increase wrong characters in words, the accuracy gets decreased. For sentence-level word prediction, we achieved accuracy of 63% in case of RoBERTa and 58% for distilled respectively.","PeriodicalId":404424,"journal":{"name":"2022 24th International Multitopic Conference (INMIC)","volume":"17 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-10-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126430552","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 : 2022-10-21DOI: 10.1109/INMIC56986.2022.9972915
Abdul Rehman, L. Fahad
Daily gun and knife related incidents are increasing due to lack of security check. In most of the places CCTV cameras are being installed however they require surveillance all the time. It is difficult due to limitations of humans in vigilant monitoring of the surveillance videos. The need of automated weapon detection is evident to limit and reduce these types of incidents. The proposed approach is mainly focused on developing an automated weapon detection system to detect different types of firearms and knives. In order to detect these types of incidents, we used a YOLOv5 deep learning model on a self collected dataset. The evaluation of the proposed approach shows its ability in the accurate detection of these weapons with an F1 score of 0.95 in CCTV video.
{"title":"Real-Time Detection of Knives and Firearms using Deep Learning","authors":"Abdul Rehman, L. Fahad","doi":"10.1109/INMIC56986.2022.9972915","DOIUrl":"https://doi.org/10.1109/INMIC56986.2022.9972915","url":null,"abstract":"Daily gun and knife related incidents are increasing due to lack of security check. In most of the places CCTV cameras are being installed however they require surveillance all the time. It is difficult due to limitations of humans in vigilant monitoring of the surveillance videos. The need of automated weapon detection is evident to limit and reduce these types of incidents. The proposed approach is mainly focused on developing an automated weapon detection system to detect different types of firearms and knives. In order to detect these types of incidents, we used a YOLOv5 deep learning model on a self collected dataset. The evaluation of the proposed approach shows its ability in the accurate detection of these weapons with an F1 score of 0.95 in CCTV video.","PeriodicalId":404424,"journal":{"name":"2022 24th International Multitopic Conference (INMIC)","volume":"19 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-10-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125759247","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 : 2022-10-21DOI: 10.1109/INMIC56986.2022.9972968
Mamoona Tasadduq
Roman Urdu is an informal form of writing the Urdu language which is written in Latin script. It is the language most widely used on the internet, social media, and text messaging by native Urdu speakers. The problem that arises with Roman Urdu is an inconsistent way of writing by different people. No standard rules are defined for writing Roman Urdu which makes it very difficult to perform Natural Language Processing. To overcome this issue, the text needs to be normalized to perform effective analysis. Therefore, this work provides a Roman Urdu dictionary that works as the foundation for processing Roman Urdu. It also proposes a model for the lexical normalization of Roman Urdu text.
{"title":"Lexical Normalization of Roman Urdu","authors":"Mamoona Tasadduq","doi":"10.1109/INMIC56986.2022.9972968","DOIUrl":"https://doi.org/10.1109/INMIC56986.2022.9972968","url":null,"abstract":"Roman Urdu is an informal form of writing the Urdu language which is written in Latin script. It is the language most widely used on the internet, social media, and text messaging by native Urdu speakers. The problem that arises with Roman Urdu is an inconsistent way of writing by different people. No standard rules are defined for writing Roman Urdu which makes it very difficult to perform Natural Language Processing. To overcome this issue, the text needs to be normalized to perform effective analysis. Therefore, this work provides a Roman Urdu dictionary that works as the foundation for processing Roman Urdu. It also proposes a model for the lexical normalization of Roman Urdu text.","PeriodicalId":404424,"journal":{"name":"2022 24th International Multitopic Conference (INMIC)","volume":"38 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-10-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124822154","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 : 2022-10-21DOI: 10.1109/INMIC56986.2022.9972989
Afaq Ahmad Khan, A. Hassan, Muhammad Talha Jahangir
Machine learning (ML) has no doubt virtually helped in nearly all fields of life, including medical sciences. ML models are now being trained, tested and developed with the help of information gained from Electroencephalogram (EEG) Signals. Neural Networks (NN) are being used specifically in this regard to exploit their image classification ability. A special class of NN called Transfer Learning (TL), is used to enhance the capability of NNs. In this paper, EEG signals are extracted and used to classify Left or Right Motor Images of the brain using Inception V3 and VGG 16 models. We try to enhance the accuracy of these TL Models by exploiting a different methodology as compared to other available statistical methods available in the research community. For the said purpose, a dataset from Brain-Computer Interface (BCI) Competition IV 2b was used. EEG signals are extracted and transformed into Short Time Fourier Transform (STFT) images. These STFT images are labeled with either Left or Right Motor Imagery (MI) Class. The transfer learning models are trained using these STFT images and results are also compared with a state-of-the art research, implementing Capsule Networks.
{"title":"Subject Wise Motor Imagery Classification from EEG Data Using Transfer Learning","authors":"Afaq Ahmad Khan, A. Hassan, Muhammad Talha Jahangir","doi":"10.1109/INMIC56986.2022.9972989","DOIUrl":"https://doi.org/10.1109/INMIC56986.2022.9972989","url":null,"abstract":"Machine learning (ML) has no doubt virtually helped in nearly all fields of life, including medical sciences. ML models are now being trained, tested and developed with the help of information gained from Electroencephalogram (EEG) Signals. Neural Networks (NN) are being used specifically in this regard to exploit their image classification ability. A special class of NN called Transfer Learning (TL), is used to enhance the capability of NNs. In this paper, EEG signals are extracted and used to classify Left or Right Motor Images of the brain using Inception V3 and VGG 16 models. We try to enhance the accuracy of these TL Models by exploiting a different methodology as compared to other available statistical methods available in the research community. For the said purpose, a dataset from Brain-Computer Interface (BCI) Competition IV 2b was used. EEG signals are extracted and transformed into Short Time Fourier Transform (STFT) images. These STFT images are labeled with either Left or Right Motor Imagery (MI) Class. The transfer learning models are trained using these STFT images and results are also compared with a state-of-the art research, implementing Capsule Networks.","PeriodicalId":404424,"journal":{"name":"2022 24th International Multitopic Conference (INMIC)","volume":"68 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-10-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129951214","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 : 2022-10-21DOI: 10.1109/INMIC56986.2022.9972919
Muhammad Burhan, Ahmad Arsalan, R. A. Rehman
The future Internet architecture, like Named-Data Network (NDN), converts host-based network structures into content-based network structures. Through this transformation, the overall network performance and efficiency are increased. Furthermore, every router in the NDN uses a caching mechanism. Thus, the cache replacement policy used by the NDN routers is also a significant determinant of the NDN's overall performance. Therefore, several types of research have been conducted about the NDN's cache replacement policy. In this article, a light-weighted cache replacement strategy is proposed that overcomes the limitations and drawbacks of the Least Frequently Used (LFU) cache policy. The proposed strategy applies variables, based on real-time producer popularity. Additionally, it can be observed through extensive simulations that the proposed strategy provides better results and shows a higher cache hit ratio as compared to the existing cache policies.
{"title":"EPP-LFU: An Efficient Producer Popularity-based LFU Policy for the Applications of Named-Data Network","authors":"Muhammad Burhan, Ahmad Arsalan, R. A. Rehman","doi":"10.1109/INMIC56986.2022.9972919","DOIUrl":"https://doi.org/10.1109/INMIC56986.2022.9972919","url":null,"abstract":"The future Internet architecture, like Named-Data Network (NDN), converts host-based network structures into content-based network structures. Through this transformation, the overall network performance and efficiency are increased. Furthermore, every router in the NDN uses a caching mechanism. Thus, the cache replacement policy used by the NDN routers is also a significant determinant of the NDN's overall performance. Therefore, several types of research have been conducted about the NDN's cache replacement policy. In this article, a light-weighted cache replacement strategy is proposed that overcomes the limitations and drawbacks of the Least Frequently Used (LFU) cache policy. The proposed strategy applies variables, based on real-time producer popularity. Additionally, it can be observed through extensive simulations that the proposed strategy provides better results and shows a higher cache hit ratio as compared to the existing cache policies.","PeriodicalId":404424,"journal":{"name":"2022 24th International Multitopic Conference (INMIC)","volume":"22 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-10-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127507935","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 : 2022-10-21DOI: 10.1109/INMIC56986.2022.9972975
Noman Khan, Sohail Ahmed Khan, Muhammad Osama Afridi, Tanveer Abbas
Resonant converter based power supplies are widely deployed in many industrial applications for their high efficiency and high power density. For control and regulation of the output voltage in resonant converters, frequency modulation (FM) is an intuitive and preferred method. However, the control-input-to-output transfer function is non-linear function of the operating frequency. The mean operating frequency determines the DC gain and bandwidth of the system for small signal perturbations. In this regard, third order non-linear model for LLC resonant converters is proposed in the literature. At a fixed operating point (i.e. a fixed output voltage corresponding to a particular operating frequency), the model can be treated as fairly linear for small perturbations. It is observed that the DC gain and bandwidth at different operating points is quite different, so deciding an appropriate operating point for a particular application is an important design choice. This research considers the small signal model of an 8.8kV/2A LLC resonant converter designed with resonant frequency ($f_{r}$) of 22.7kHz and quality factor (Q)=4 for an industrial magnetron as a specific load. The model is validated through simulations and hardware experiments using the LLC resonant converter. For our specific system DC gain at $f_{r}$ is - 65dB and bandwidth is 310Hz. DC gain and bandwidth of the system at different operating frequencies give an insight to decide the appropriate operating point. Hence, this work offers a strong foundation for a controller design for the LLC resonant converter under consideration.
{"title":"Validation of Small Signal Model of an LLC Resonant Converter Based HVDC Modulator","authors":"Noman Khan, Sohail Ahmed Khan, Muhammad Osama Afridi, Tanveer Abbas","doi":"10.1109/INMIC56986.2022.9972975","DOIUrl":"https://doi.org/10.1109/INMIC56986.2022.9972975","url":null,"abstract":"Resonant converter based power supplies are widely deployed in many industrial applications for their high efficiency and high power density. For control and regulation of the output voltage in resonant converters, frequency modulation (FM) is an intuitive and preferred method. However, the control-input-to-output transfer function is non-linear function of the operating frequency. The mean operating frequency determines the DC gain and bandwidth of the system for small signal perturbations. In this regard, third order non-linear model for LLC resonant converters is proposed in the literature. At a fixed operating point (i.e. a fixed output voltage corresponding to a particular operating frequency), the model can be treated as fairly linear for small perturbations. It is observed that the DC gain and bandwidth at different operating points is quite different, so deciding an appropriate operating point for a particular application is an important design choice. This research considers the small signal model of an 8.8kV/2A LLC resonant converter designed with resonant frequency ($f_{r}$) of 22.7kHz and quality factor (Q)=4 for an industrial magnetron as a specific load. The model is validated through simulations and hardware experiments using the LLC resonant converter. For our specific system DC gain at $f_{r}$ is - 65dB and bandwidth is 310Hz. DC gain and bandwidth of the system at different operating frequencies give an insight to decide the appropriate operating point. Hence, this work offers a strong foundation for a controller design for the LLC resonant converter under consideration.","PeriodicalId":404424,"journal":{"name":"2022 24th International Multitopic Conference (INMIC)","volume":"24 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-10-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128330804","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：