Research On Multifunctional Intelligent Access Control Terminal And Its Network Completeness

The access control terminal is the first device for the customer experience in the security field.With the rapid change of customer requirements in the application scenario,the requirements for the access control terminal are also increasing.It requires not only diversified access functions,but also installation and maintenance of the terminal access engineering.The completeness of topology scalability and the intelligent maintenance of the network have all brought new technical requirements.The research work of this thesis is based on the security system integrated platform of AirLink Intelligent Technology Co.,Ltd.The research and development of user-oriented componentized and scalable key technologies of the nextgeneration security system.The main research and development work is as follows: This paper designs a multifunctional intelligent access terminal from the hardware and the software.The intelligence of the access control and the completeness of its network are studied algorithmically.First of all,for the problem of extended access control and imperfect database of access control,the paper designs a multi-signal intelligent access control terminal from hardware and simulates the access control network to acquire and establish a database.Extensibility means that the access control can be compatible with input of multiple signals.The man-car of the paper designs a multi-channel signal intelligent access terminal.Its processing center uses an STM32 MCU and can simultaneously receive signals for handling credit card and vehicle communication.Data is an important part of data analysis algorithms.At present,there is no established database for access control.The paper simulates a 3*4 level 2 access control network to collect data.The first level consists of 3 STM32 MCUs.The second level consists of 3*4 sensors.In the collection process,the host computer written in python is used to display data.Secondly,for the problem of imbalanced access control dataset,the paper studies it from the data level,discusses the SMOTE algorithm and the Borderline-SMOTE algorithm,and analyzes the defects in the practical application with noise,and proposes a central radiationbased SMOTE algorithm.This algorithm takes advantage of the fact that SMOTE can effectively avoid overfitting and Borderline-SMOTE can improve the recognition rate of edge samples.By using the center point of small samples,the distance between each small class sample to the center sample is calculated and the sample at a longer distance is discarded,while keeping the safety samples and all boundary samples within a certain distance.In this thesis,two imbalanced ratios(0.05 and 0.1)and three noise ratios sets(0.1,0.2 and 0.3)for each ratio are used for experiments.The results show that the proposed algorithm has obvious classification advantages in classifying minimal-class samples with noise.Finally,aiming at the problem of imbalanced access control dataset,the paper studies it from the aspect of algorithm and proposes an example cost-sensitive method based on multiclass integration.The access control data used in this paper are both class sensitive(the cost of the same access control brand under different circumstances is different)and category sensitive(different access control brands with different cost matrix).Obviously,using only one of these methods can not achieve the scenario of this paper.The paper analyzes the CSDT algorithm and the instance-based cost-sensitive decision tree method firstly,then proposes an instancebased cost-sensitive approach based on multi-class integration.The thesis carries out experiments from single-class and multi-class data respectively.The results show that the performance of CSDT in multi-class scenes drops sharply;however,the classification method proposed in this paper can make up for the short board of CSDT algorithm in multi-class scenarios.