Said Kerrache, Hafida Benhidour, Sahar Al-Homaidhi
The effectiveness of crowd evacuation is an important public safety issue. Crowd related incidents are frequent and often result in serious material and human loss, hence the increasing interest in developing new techniques to assist in crowd evacuation. In this paper, we introduce a new crowd evacuation method, where the evacuation plan is computed based on optimal transport. The use of the optimal transport formulation results in the minimization of the kinetic energy necessary to perform the evacuation, thus reducing the effort exercised by the crowd members. To ensure the safety and the feasibility of the computed plan, a safety model is introduced as an additional factor in the optimization. The proposed model allows avoiding dangerous places, spreading out the crowd, limiting the crowd density and imposing one-way circulation. A numerical method adapted to the resulting optimization problem is presented. The efficiency of the proposed approach is evaluated through simulation. The experimental results show that the proposed method computes a solution that strikes a balance between the different considered factors.
{"title":"An Optimal Transport-Based Approach for Crowd Evacuation","authors":"Said Kerrache, Hafida Benhidour, Sahar Al-Homaidhi","doi":"10.1109/CIMSIM.2013.31","DOIUrl":"https://doi.org/10.1109/CIMSIM.2013.31","url":null,"abstract":"The effectiveness of crowd evacuation is an important public safety issue. Crowd related incidents are frequent and often result in serious material and human loss, hence the increasing interest in developing new techniques to assist in crowd evacuation. In this paper, we introduce a new crowd evacuation method, where the evacuation plan is computed based on optimal transport. The use of the optimal transport formulation results in the minimization of the kinetic energy necessary to perform the evacuation, thus reducing the effort exercised by the crowd members. To ensure the safety and the feasibility of the computed plan, a safety model is introduced as an additional factor in the optimization. The proposed model allows avoiding dangerous places, spreading out the crowd, limiting the crowd density and imposing one-way circulation. A numerical method adapted to the resulting optimization problem is presented. The efficiency of the proposed approach is evaluated through simulation. The experimental results show that the proposed method computes a solution that strikes a balance between the different considered factors.","PeriodicalId":249355,"journal":{"name":"2013 Fifth International Conference on Computational Intelligence, Modelling and Simulation","volume":"27 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-09-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125294994","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}
A new scheme for adaptive neural networks for nonlinear dynamic system identification is proposed in this paper. The network of structure multi-layer perceptron with external recurrence is trained offline at first to get the initial network parameters. The parameters of the network are classified into short-term memory part and long-term memory part. The short-term memory part includes the parameters which are linear to the network output. In the implementation, the network is validated in each sampling time using a set of new measurement data. Training procedure will be executed if the model error exceeds a specified value and the short-term memory part will be adjusted. The application in modelling of room thermal behaviour demonstrates the performance of the proposed scheme.
{"title":"Adaptive Neural Networks for Nonlinear Dynamic Systems Identification","authors":"Erwin Sitompul","doi":"10.1109/CIMSIM.2013.10","DOIUrl":"https://doi.org/10.1109/CIMSIM.2013.10","url":null,"abstract":"A new scheme for adaptive neural networks for nonlinear dynamic system identification is proposed in this paper. The network of structure multi-layer perceptron with external recurrence is trained offline at first to get the initial network parameters. The parameters of the network are classified into short-term memory part and long-term memory part. The short-term memory part includes the parameters which are linear to the network output. In the implementation, the network is validated in each sampling time using a set of new measurement data. Training procedure will be executed if the model error exceeds a specified value and the short-term memory part will be adjusted. The application in modelling of room thermal behaviour demonstrates the performance of the proposed scheme.","PeriodicalId":249355,"journal":{"name":"2013 Fifth International Conference on Computational Intelligence, Modelling and Simulation","volume":"19 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-09-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115478295","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 paper presents the simulation study of magneto-rheological (MR) damper and hydraulic actuator for suspension system using intelligent PID controller with iterative learning algorithm. The MR damper is an intelligent damper filled with particle magnetic polarizable and suspended into a liquid form. This actuator was installed to the semi-active suspension system as a variable damper. The Bouc Wen model of MR damper was used to determine the required damper force based on the force-displacement and force velocity characteristic. For the purpose of comparison of performance, a hydraulic actuator, working as an additional damper, was installed within an active suspension system. Two different disturbances namely bump and random disturbances were introduced as the road profile. The performances of theproposed actuators were investigated in term of body displacement, velocity and acceleration. The results indicated the active system based on hydraulic actuator was better than semi-active based on MR damper and passive system in term of the body displacement, velocity and acceleration.
{"title":"Self-Tuning PID Controller with MR damper and Hydraulic Actuator for Suspension System","authors":"M. H. Ab Talib, I. Darus","doi":"10.1109/CIMSIM.2013.27","DOIUrl":"https://doi.org/10.1109/CIMSIM.2013.27","url":null,"abstract":"This paper presents the simulation study of magneto-rheological (MR) damper and hydraulic actuator for suspension system using intelligent PID controller with iterative learning algorithm. The MR damper is an intelligent damper filled with particle magnetic polarizable and suspended into a liquid form. This actuator was installed to the semi-active suspension system as a variable damper. The Bouc Wen model of MR damper was used to determine the required damper force based on the force-displacement and force velocity characteristic. For the purpose of comparison of performance, a hydraulic actuator, working as an additional damper, was installed within an active suspension system. Two different disturbances namely bump and random disturbances were introduced as the road profile. The performances of theproposed actuators were investigated in term of body displacement, velocity and acceleration. The results indicated the active system based on hydraulic actuator was better than semi-active based on MR damper and passive system in term of the body displacement, velocity and acceleration.","PeriodicalId":249355,"journal":{"name":"2013 Fifth International Conference on Computational Intelligence, Modelling and Simulation","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-09-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121665827","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}
We define a human motion data format for providing a character model with motion so that any motion capture data can be used to generate exchangeable human animation sequences through networks of heterogeneous computer systems. We have extended the ISO/IEC 19774 H-Anim specification to include motion definition in order to meet this objective. A character model can be prepared with any general graphics tool according to the hierarchical structure defined in the extended H-Anim specification. We can provide the character model with any captured motion data using the interfaces for motion parameters defined in the extended H-Anim specification, even when the data is not relevant to the character model initially. Several kinds of motion capture data can be applied to generate animation sequences for a character model; conversely, one kind of motion capture data can be applied to several different character models. New Joint and Motion nodes are defined for animation exchange in the H-Anim specification. Schema expansion and validation are also included.
{"title":"Human Motion Data Definition for 3D Characters","authors":"Myeong-Won Lee, Chul-Hee Jung, C. Park","doi":"10.1109/CIMSIM.2013.23","DOIUrl":"https://doi.org/10.1109/CIMSIM.2013.23","url":null,"abstract":"We define a human motion data format for providing a character model with motion so that any motion capture data can be used to generate exchangeable human animation sequences through networks of heterogeneous computer systems. We have extended the ISO/IEC 19774 H-Anim specification to include motion definition in order to meet this objective. A character model can be prepared with any general graphics tool according to the hierarchical structure defined in the extended H-Anim specification. We can provide the character model with any captured motion data using the interfaces for motion parameters defined in the extended H-Anim specification, even when the data is not relevant to the character model initially. Several kinds of motion capture data can be applied to generate animation sequences for a character model; conversely, one kind of motion capture data can be applied to several different character models. New Joint and Motion nodes are defined for animation exchange in the H-Anim specification. Schema expansion and validation are also included.","PeriodicalId":249355,"journal":{"name":"2013 Fifth International Conference on Computational Intelligence, Modelling and Simulation","volume":"17 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-09-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126920528","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 paper proposes a method for assessing the subjective classifications of traditional Chinese medicine (TCM) and investigating the influence of attributes on them, while these attributes are extracted from multi-sensors and represented by different modes. In TCM, a person's health states can be represented by 13 Zhengs that are not entirely independent, while the diagnosis data given by TCM doctors are subjective. Accordingly, the influence of the modes and the attributes extracted from the multimodal sensor data on the Zheng's classification is validated by a defined aggregation function called aas. Moreover, the conditions of removing the weak modes are proposed based on the correlation between the attributes of modes and the number of the attributes close to the Zhengs. The simulation results verify the adequacy of the above aas and conditions in evaluating the effect of attributes on the classification performance.
{"title":"Evaluating the Effect of Different Mode's Attributes on the Subjective Classification in the Case of TCM","authors":"Ying Dai","doi":"10.1109/CIMSIM.2013.35","DOIUrl":"https://doi.org/10.1109/CIMSIM.2013.35","url":null,"abstract":"This paper proposes a method for assessing the subjective classifications of traditional Chinese medicine (TCM) and investigating the influence of attributes on them, while these attributes are extracted from multi-sensors and represented by different modes. In TCM, a person's health states can be represented by 13 Zhengs that are not entirely independent, while the diagnosis data given by TCM doctors are subjective. Accordingly, the influence of the modes and the attributes extracted from the multimodal sensor data on the Zheng's classification is validated by a defined aggregation function called aas. Moreover, the conditions of removing the weak modes are proposed based on the correlation between the attributes of modes and the number of the attributes close to the Zhengs. The simulation results verify the adequacy of the above aas and conditions in evaluating the effect of attributes on the classification performance.","PeriodicalId":249355,"journal":{"name":"2013 Fifth International Conference on Computational Intelligence, Modelling and Simulation","volume":"22 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-09-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126150040","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 this paper we compare the performance of back propagation and resilient propagation algorithms in training neural networks for spam classification. Back propagation algorithm is known to have issues such as slow convergence, and stagnation of neural network weights around local optima. Researchers have proposed resilient propagation as an alternative. Resilient propagation and back propagation are very much similar except for the weight update routine. Resilient propagation does not take into account the value of the partial derivative (error gradient), but rather considers only the sign of the error gradient to indicate the direction of the weight update. We show that resilient propagation yields faster convergence and higher accuracy on the UCI Spambase dataset.
{"title":"Comparison of Back Propagation and Resilient Propagation Algorithm for Spam Classification","authors":"Navneel Prasad, Rajeshni Singh, S. Lal","doi":"10.1109/CIMSIM.2013.14","DOIUrl":"https://doi.org/10.1109/CIMSIM.2013.14","url":null,"abstract":"In this paper we compare the performance of back propagation and resilient propagation algorithms in training neural networks for spam classification. Back propagation algorithm is known to have issues such as slow convergence, and stagnation of neural network weights around local optima. Researchers have proposed resilient propagation as an alternative. Resilient propagation and back propagation are very much similar except for the weight update routine. Resilient propagation does not take into account the value of the partial derivative (error gradient), but rather considers only the sign of the error gradient to indicate the direction of the weight update. We show that resilient propagation yields faster convergence and higher accuracy on the UCI Spambase dataset.","PeriodicalId":249355,"journal":{"name":"2013 Fifth International Conference on Computational Intelligence, Modelling and Simulation","volume":"92 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-09-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115172094","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}
For implementation of Cyber-Physical Systems (CPS) simulation, interoperability between physical and cyber system are required. IEEE High Level Architecture (HLA) is standard for distributed simulation interoperability, and ETSI Machine-to-Machine (M2M) is standard for integration between physical machines. In this paper, we propose M2MHLA adaptor to integrate HLA-based simulations and M2M physical machines. M2M-HLA adaptor is located between HLA simulation systems and M2M systems. This adaptor transforms HLA data to M2M data and M2M data to HLA data. We propose sequence structure between M2M-HLA APIs for interconnection, data type definition and transformation method and time management method for design and implementation M2M-HLA adaptor. Because of M2M-HLA adaptor allows interoperability between different cyber systems and physical systems, this would extend the scope of CPS simulation system and increase its utilization.
{"title":"Design and Implementation of M2M-HLA Adaptor for Integration of ETSI M2M Platform and IEEE HLA-Based Simulation System","authors":"Yunjung Park, D. Min","doi":"10.1109/CIMSIM.2013.57","DOIUrl":"https://doi.org/10.1109/CIMSIM.2013.57","url":null,"abstract":"For implementation of Cyber-Physical Systems (CPS) simulation, interoperability between physical and cyber system are required. IEEE High Level Architecture (HLA) is standard for distributed simulation interoperability, and ETSI Machine-to-Machine (M2M) is standard for integration between physical machines. In this paper, we propose M2MHLA adaptor to integrate HLA-based simulations and M2M physical machines. M2M-HLA adaptor is located between HLA simulation systems and M2M systems. This adaptor transforms HLA data to M2M data and M2M data to HLA data. We propose sequence structure between M2M-HLA APIs for interconnection, data type definition and transformation method and time management method for design and implementation M2M-HLA adaptor. Because of M2M-HLA adaptor allows interoperability between different cyber systems and physical systems, this would extend the scope of CPS simulation system and increase its utilization.","PeriodicalId":249355,"journal":{"name":"2013 Fifth International Conference on Computational Intelligence, Modelling and Simulation","volume":"7 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-09-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125534644","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 this paper we propose a methodology for investigating the impact of basic Network-on-Chip (NoC) parameters and self-adaptive scheme in the context of the Field Programmable Gate Array (FPGA). With our proposed methodology that is based on the Bayesian networking model we examined the effects of flit buffer depth, flit data width and virtual channel parameters through an extensive experimentation and simulation for scalable and adaptive NoC on Xilinx Virtex7 FPGA device. To demonstrate the flexibility and extensible design space coverage of our methodology, we design and present hardware synthesis results of 96 different NoCs configurations. We used a cycle accurate simulation system and drive the NoCs with four different traffic patterns and varying number of virtual channels (VCs) and show the resulting load-delay curves. Our results show that, for scalable and adaptive NoC, the flit data width and flit buffer depth parameters have the largest impact on FPGA area and clock frequency. We show that these parameters need to be properly adjusted for better run-time performance of the FPGA. Moreover, the neighbor traffic pattern with 4 VCs offer the best performance with 95% throughput, low latency and efficient silicon area in both Mesh and Torus networks.
{"title":"Examining the Performance Impact of NoC Parameters for Scalable and Adaptive FPGA-Based Network-on-Chips","authors":"S. Abba, Jeong-A Lee","doi":"10.1109/CIMSIM.2013.65","DOIUrl":"https://doi.org/10.1109/CIMSIM.2013.65","url":null,"abstract":"In this paper we propose a methodology for investigating the impact of basic Network-on-Chip (NoC) parameters and self-adaptive scheme in the context of the Field Programmable Gate Array (FPGA). With our proposed methodology that is based on the Bayesian networking model we examined the effects of flit buffer depth, flit data width and virtual channel parameters through an extensive experimentation and simulation for scalable and adaptive NoC on Xilinx Virtex7 FPGA device. To demonstrate the flexibility and extensible design space coverage of our methodology, we design and present hardware synthesis results of 96 different NoCs configurations. We used a cycle accurate simulation system and drive the NoCs with four different traffic patterns and varying number of virtual channels (VCs) and show the resulting load-delay curves. Our results show that, for scalable and adaptive NoC, the flit data width and flit buffer depth parameters have the largest impact on FPGA area and clock frequency. We show that these parameters need to be properly adjusted for better run-time performance of the FPGA. Moreover, the neighbor traffic pattern with 4 VCs offer the best performance with 95% throughput, low latency and efficient silicon area in both Mesh and Torus networks.","PeriodicalId":249355,"journal":{"name":"2013 Fifth International Conference on Computational Intelligence, Modelling and Simulation","volume":"16 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-09-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125225101","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 paper presents a novel Genetic Algorithms with Parameter Exchanger and its application to system identification for a single-link flexible manipulator system. A simulation environment characterizing the dynamic behavior of the flexible manipulator system was first developed using finite difference method to acquire the input-output data of the system. In this study, system identification scheme is developed to obtain a dynamic model of the manipulator in parametric form using Genetic Algorithms. A novel methodology of Genetic Algorithms namely as Genetic Algorithms with Parameter Exchanger (GAPE) was proposed and its performance is assessed in comparison to a standard Genetic Algorithms in characterizing the flexible manipulator structure. Results demonstrate the advantages of Genetic Algorithm with Parameter Exchanger over their standard counterpart in parametric identification.
{"title":"Modeling of Flexible Manipulator Structure Using Genetic Algorithm with Parameter Exchanger","authors":"H. Yatim, I. Darus, M. S. Hadi","doi":"10.1109/CIMSIM.2013.15","DOIUrl":"https://doi.org/10.1109/CIMSIM.2013.15","url":null,"abstract":"This paper presents a novel Genetic Algorithms with Parameter Exchanger and its application to system identification for a single-link flexible manipulator system. A simulation environment characterizing the dynamic behavior of the flexible manipulator system was first developed using finite difference method to acquire the input-output data of the system. In this study, system identification scheme is developed to obtain a dynamic model of the manipulator in parametric form using Genetic Algorithms. A novel methodology of Genetic Algorithms namely as Genetic Algorithms with Parameter Exchanger (GAPE) was proposed and its performance is assessed in comparison to a standard Genetic Algorithms in characterizing the flexible manipulator structure. Results demonstrate the advantages of Genetic Algorithm with Parameter Exchanger over their standard counterpart in parametric identification.","PeriodicalId":249355,"journal":{"name":"2013 Fifth International Conference on Computational Intelligence, Modelling and Simulation","volume":"36 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-09-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124956995","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 this paper, we present a weight learning method introduced to learn weights on each individual classifier to construct an ensemble. Genetic algorithm is applied to search for an optimal combination of weights for each individual classifier on which classifier ensemble is expected to give best performance. Our proposed ensemble approach can combine heterogeneous classifiers and/or classifier ensembles to enhance the overall classification performance of a given classifier system. We have evaluated our proposed ensemble approach on variety of real life datasets. The proposed approach is compared with existing state-of-the art ensemble techniques such as Adaboost, Bagging and RSM to demonstrate the superiority of proposed work as compared to the competitors.
{"title":"Framework for Constructing Hybrid Classifier Using Weight Learning to Combine Heterogeneous Classifiers","authors":"S. Khalid, S. Arshad","doi":"10.1109/CIMSIM.2013.34","DOIUrl":"https://doi.org/10.1109/CIMSIM.2013.34","url":null,"abstract":"In this paper, we present a weight learning method introduced to learn weights on each individual classifier to construct an ensemble. Genetic algorithm is applied to search for an optimal combination of weights for each individual classifier on which classifier ensemble is expected to give best performance. Our proposed ensemble approach can combine heterogeneous classifiers and/or classifier ensembles to enhance the overall classification performance of a given classifier system. We have evaluated our proposed ensemble approach on variety of real life datasets. The proposed approach is compared with existing state-of-the art ensemble techniques such as Adaboost, Bagging and RSM to demonstrate the superiority of proposed work as compared to the competitors.","PeriodicalId":249355,"journal":{"name":"2013 Fifth International Conference on Computational Intelligence, Modelling and Simulation","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-09-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130967966","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：