Showing papers presented at "International Conference on Measuring Technology and Mechatronics Automation in 2019"

A Research on IoT Based Smart Home

Abstract: Smart Home is a safe, networked and intelligent home control system integrated with automation control, network communication and Internet of things (IoT). The IoT based Smart Home, by application of sensors, collects indoor environmental parameter information such as temperature, humidity and gas concentration, etc., as well as information on various household appliances. The information collected will then be transmitted by the IoT gateway to the server on the Internet and users can monitor the operation of each subsystem of the Smart Home after login. In this paper, a systematic framework of IoT based Smart Home has been designed, studies on home safety precaution, smart lighting, environmental control and home appliance control have been conducted and the system functions have also been tested.

Three-Dimensional Localization Algorithm for WSN Nodes Based on RSSI-TOA and LSSVR Method

Abstract: Aiming at the problem that traditional single ranging method is difficult to ensure the accuracy of wireless sensor networks (WSN) node localization, a least squares support vector regression (LSSVR) node three-dimensional localization algorithm based on received signal strength index (RSSI) and time of arrival (TOA) ranging information is proposed. Firstly, by introducing a single mobile anchor node, three-dimensional localization model of WSN nodes is established based on LSSVR by RSSI and TOA sampling values of virtual nodes. Then, based on the idea of using RSSI ranging in the short distance and TOA ranging in the longer distance, the ranging of nodes in different locations is realized. Finally, the ranging information of RSSI and TOA are used as the input vector of LSSVR localization model to complete the location estimation of unknown nodes. The simulation results of MATLAB show that the proposed algorithm has higher localization accuracy than LSSVR-based or RSSI-TOA-based localization method in the three-dimensional environment with randomly distributed nodes.

Traffic Prediction and Attack Detection Approach Based on PSO Optimized Elman Neural Network

Abstract: In view of the characteristics of non-linearity, multi-variable and time-varying of network system, PSO algorithm is proposed to optimize Elman neural network and obtain the optimal network parameters to improve the prediction accuracy and adaptability of the algorithm. Then, the anomalous traffic detected by the anomalous traffic detection model is used as input, and the samples projected into the high-dimensional feature space are classified by KNN algorithm. An attack detection classifier is established to identify the types of network attacks. The experimental results show that, compared with traditional attack detection methods, the proposed network attack detection model based on abnormal traffic analysis improves the accuracy of intrusion detection, which can effectively identify the types of network attacks and ensure the security of information systems.

Research in Fast Modular Exponentiation Algorithm Based on FPGA

Abstract: Modular exponentiation of large number is widely applied in public-key cryptosystem, also the bottleneck in the computation of public-key algorithm. Modular multiplication is the key calculation in modular exponentiation. An improved Montgomery algorithm is utilized to achieve modular multiplication and converted into systolic array to increase the running frequency. A high efficiency fast modular exponentiation structure is developed to bring the best out of the modular multiplication module and enhance the ability of defending timing attacks and power attacks. For 1024-bit key operands, the design can be run at 170MHz and finish a modular exponentiation in 4,402,374 clock cycles.

A Teaching System of English Online Course Based on Artificial Intelligence

Abstract: In comparison with common teaching auxiliary system, this paper applies artificial intelligence module combining with knowledge recommendation to the system and designs an online English course teaching system. The core modules in system adopt domain knowledge base and teaching management and they separates organizing management of knowledge from processing mechanism of knowledge. Through interface accessing knowledge base, processing mechanism promotes that the change of content organizing in knowledge base will not lead to large change in processing mechanism. Meanwhile, the realization of adaptive teaching planning is also put forward so that the teaching can be diversified. Test application shows the system can assist students to improve learning efficiency and learning contents will be more pertinent. Furthermore, the system provides similar methods with an example model so it also has referential significance.

Design and Implementation of Intelligent Control System for Green Building

Abstract: To meet the demands of users in green building and to reduce the learning cost by intelligent control system, this paper establishes the overall design framework of intelligent control system for green building, and formulates the specific implementation scheme of system hardware and software, as well as the overall control strategy of the system. We use Kingview and Zigbee ad hoc network technology to establish wireless sensor network, to transmit and store all kinds of sensor data. RS485 bus is adopted to realize switch control of household equipment and irrigation solenoid valve, and Modbus-RTU protocol is used to realize system monitoring and control. Then, the temperature control subsystem based on fuzzy-PID control algorithm is established according to the relevant theory of the fuzzy-PID control algorithm, and the simulation experiment is completed in MATLAB, which achieves better control effect. Finally, through the functional test of the acquisition and the control part of the system, it is proved that the scheme achieves the expected design requirements.

A Traffic Sign Image Recognition and Classification Approach Based on Convolutional Neural Network

Abstract: In this paper, the traffic sign recognition method based on CNN is studied by using image processing and machine vision processing technology combined with the application of in-depth learning in target classification. A traffic sign recognition method with high recognition efficiency and high efficiency is proposed. The scheme uses the method of training cascade classifiers based on HOG features to detect traffic signs. At the same time, to ensure that all traffic signs are detected, an improved CNN model is trained to determine whether there are traffic signs in the candidate area, and the final decision is made as the last level of the cascade classifier. The simulation results show that the scheme can accurately locate the location of traffic signs, greatly reduce the false detection rate, improve the accuracy rate, and lay the foundation for subsequent traffic sign recognition.

Research on OMR Recognition Based on Convolutional Neural Network Tensorflow Platform

Abstract: Aiming at optical mark reader (OMR) recognization problems and the complexity of virtual operating that has resulted in much increased human labor, the model architecture based on convolution neural network(CNN) and TensorFlow platform is proposed. In order to improve OMR recognition technology, we uses deep learning tools in TensorFlow to construct the neural network model. For the sake of training the model we uses TensorFlow to preprocess the OMR input data, obtains the TFRecord file, and then train model to reach the requisite OMR recognition level in test paper image and obtain the standard parameters. The view tool TensorBoard is used to display the calculated flow graph of the TensorFlow model.

Integrated Load Forecasting Model of Multi-Energy System Based on Markov Chain Improved Neural Network

Abstract: Aiming at the uncertain relationship between renewable energy power generation or functional system output and energy consumption load in multi-energy system, a multi-energy comprehensive load forecasting model is studied to solve the problem of source-load matching. This paper studies the load characteristic analysis method of multi-energy system after cooling, heating and electricity load integration, establishes a comprehensive load data processing model based on membership function, and establishes a comprehensive load neural network forecasting model according to the time characteristics of comprehensive load. The non-linear correlation of thermal and cooling load in time is studied, and a modified method of comprehensive load neural network prediction model based on Markov chain is proposed. Finally, taking a typical multi-energy system as an example, a comprehensive load forecasting simulation system model is established. The simulation results show that the accuracy of the proposed algorithm can meet the multi-energy scheduling requirements of multi-energy system.

An Acoustic Recognition Model for English Speech Based on Improved HMM Algorithm

Abstract: This paper presents an English speech recognition system based on the improved HMM model. Firstly, this paper integrates the process including speech acoustic analysis, acoustic HMM model establishment and recognition through the analysis of the system framework and implementation of continuous speech recognition system. To enable HMM to assume that the time distribution of continuous segments obeys geometric distribution and its real distribution law is not consistent, we adopt HMM-based segment distribution to describe the time correlation of speech signals more accurately. The experiments show that the algorithm improves the recognition rate and shows better adaptability. Therefore, the recognition ability of acoustic target recognition system can also be effectively improved.

A Novel Algorithmic Trading Approach Based on Reinforcement Learning

Abstract: Algorithmic trading has gained great popularity due to the rapidly increasing computing power of modern day computers. In order to reduce trading latency, market participants and academic researchers are constantly looking for better novel and successful approaches to help them to achieve greater success. In this paper, a novel reinforcement learning approach is proposed which defines the algorithmic trading problem under the framework of the classic reinforcement learning problem, aiming to optimize the agent's performance in an unknown environment. By using state-of-the-art techniques based on least-squares temporal difference learning, an algorithmic trading system is built to support the reinforcement learning process. Evaluation of the approach is done with data from the foreign exchange market and results shows that it is profitable, easy to be expanded in the future.

Sports Video Motion Target Detection and Tracking Based on Hidden Markov Model

Abstract: Sports video moving object detection and tracking is a hot topic in computer vision research. The hidden Markov model is applied to the detection and tracking of moving objects in sports video. The acquired student motion is used as input to interact with the virtual scene. Firstly, the virtual and real difference measurement selection algorithm based on statistics is used to select the sports students with strong ability. Then, the classifier confidence method is used to select the students with high confidence level from the sports students who have not been marked, and classify them into the marked sports students to promote the generalization ability of the model. Raise. The experimental results show that this method can effectively assist physical education teaching activities and provide objective and effective data analysis for sports video target detection and tracking.

Design and Implementation of Virtual Education Laboratory

Abstract: Aiming to improve the design effect of distributed 3D virtual education laboratory, a design method of distributed 3D virtual education laboratory based on virtual reality technology is proposed. The 3D virtual scene reconstruction method is used to construct the scene simulation model designed by the distributed 3D virtual education laboratory, and the edge feature segmentation and information fusion of the distributed 3D virtual educational laboratory spatial distribution image are carried out. The RGB color decomposition method is used to decompose the color pixel feature of 3D virtual education laboratory spatial distribution image, and the color space reconstruction of distributed 3D virtual education laboratory is realized by combining the 3D point cloud feature recombination method. The optimization combination of spatial distribution of distributed 3D virtual education laboratory design is realized. On the basis of image and color processing algorithm design, based on virtual reality and scene simulation technology, the development and design of virtual education laboratory system is carried out, and 3D modeling of virtual education laboratory design is carried out by using 3DStudio MAX. The spatial combination design of 3D virtual education laboratory is realized on the Multigen Creator modeling software. The simulation results show that the visual reconstruction ability and the space combination ability of the virtual education laboratory design with this method are better.

Interactive Teaching Path Analysis of Higher Vocational Ideological and Political Courses Based on Artificial Intelligence Algorithms

Abstract: In today's society, science and technology are the primary productive forces, and the reform of any mode can not be separated from science and technology, especially in education and teaching. Most vocational college students lack good learning habits, poor understanding of classroom knowledge, and are not good at solving all kinds of difficulties encountered in learning independently. In order to solve these problems, an interactive teaching path analysis method for ideological and political courses in Higher Vocational Colleges Based on artificial intelligence algorithm is proposed. The objective function of teaching path analysis is established and the constraints are determined. Analyse and use artificial intelligence algorithm to establish interactive teaching path research model. The experimental results show that the interactive teaching path analysis method based on artificial intelligence algorithm has higher practicability than the traditional interactive teaching path research method of Ideological and political course in Higher Vocational colleges, the interactive teaching path of Ideological and political courses in higher vocational colleges can be accurately analyzed.

Research on Financial Risk Prediction and Prevention Countermeasures Based on Big Data

Abstract: In this era of "data is king", continuous strengthening of networking and digitization makes big data and finance increasingly integrated, and big data has become the core asset of finance. Financial management is inherently risk management. It is of great significance to study how to use big data to better predict and prevent financial risks. In this paper, it firstly introduces the concept and characteristics of big data. Then it discusses the financial risks in the big data environment, combines the success cases of big data in financial risk prevention and control, and analyzes the advantages of using big data for financial risk prediction and prevention. Finally it summarizes specific and effective prediction systems for financial risks and countermeasures.

Comparison of the Different Sampling Techniques for Imbalanced Classification Problems in Machine Learning

Abstract: Imbalanced class distribution is a scenario where the number of observations belonging to one class is significantly lower than those belonging to the other ones. Machine learning algorithms are often designed to improve accuracy by reducing the errors. Thus, they do not consider the class distribution proportion or the balance of classes. In this paper, firstly, we describes the various approaches for solving such class imbalance problems, using various sampling techniques. Then we weigh each technique for its pros and cons. Finally, an approach purpose is revealed in which you can create a balanced class distribution and apply ensemble learning technique designed especially for imbalanced class distribution.

Design And Simulation of Naval Gun Servo Controller Based on ADRC

Abstract: In order to improve the precision and anti-interference ability of the Naval Gun servo system and make up for the shortcomings of the traditional control strategy, the ADRC technology is applied to the Naval Gun Servo System. Firstly, the mathematical model of Servo System is analyzed and its mathematical model is established. The disturbance characteristics of Servo System are briefly analyzed. Then the ADRC controller is designed and the parameters are adjusted according to the system characteristics. The MATLAB/Simulink simulation platform compares the control precision and system anti-interference ability of Servo System under the control of ADRC and PID. The simulation results show that the Naval Gun Servo System based on ADRC control strategy can effectively improve tracking accuracy and system anti-interference ability.

Training Method for Amateur Sports in Schools Dependent on Time Series Data

Abstract: At present, the levels of amateur sports training in domestic schools are uneven. Many schools rely excessively on the traditional sports training methods. This kind of sports training method where the empty talks on the stratagems are carried out on paper often gains half the results with double the effort in the cultivation of the high-quality sports talents. Therefore, it is imperative to explore the training methods for the amateur sports in the schools. In this paper, a kind of training method for the amateur sports in the schools that is dependent on the time series data is put forward. Firstly, the training time series data is applied and processed to achieve the effective recognition of the sports training items; and then, for the training knowledge and the sports items thus obtained, sports training is carried out accordingly. Finally, the examples show that the method put forward in this paper can be used to improve the sports level of the students, so as to accomplish the purpose of amateur sports training in the schools.

English Pronunciation Recognition and Detection Based on HMM-DNN

Abstract: A hybrid speech recognition system based on deep neural network(HMM-DNN) is constructed with different acoustic modeling units in this paper. Using the idea of multitask learning, the input layer and hidden layer of DNN are shared, and the accuracy of modeling is improved by joint training. Then, the traditional classification method is improved from two aspects of input feature selection and classifier design. Using high-dimensional phonetics as the basic input feature, the binary classifier of each phoneme is embedded in the same neural network, and the multi-phoneme error detection problem is transformed into a multi-task learning problem. The experimental results show that the improved model can achieve better recognition performance than the single task learning model, and the output of the two systems is integrated to further reduce the word error rate.

A Combined Control Method of Temperature and Humidity Inside the Museum Cabinet

Abstract: This paper analyzes the coupling of temperature and humidity in microenvironment of museum display cabinet. Based on the original temperature and humidity control system with STM32F103ZET6 as the core, the system structure was improved and a set of working mode was designed according to experience. In addition, the temperature and humidity control algorithm is improved. Based on the traditional PID algorithm, fuzzy table lookup control is introduced. Meanwhile, the refrigerating capacity (dehumidification capacity) is corrected and controlled by self-optimizing operation. After the test run, the method improves the performance of the system and reduces the waste of resources.

An Acoustic Model for English Speech Recognition Based on Deep Learning

Abstract: As one of the core modules of English speech recognition system, the performance of acoustic model directly affects the recognition effect of the final system. Aiming at the problems of acoustic modeling, model optimization and training efficiency in English speech recognition system, this paper focuses on the acoustic modeling method based on deep learning technology. By analyzing the basic theory of deep neural network and HMM, the structure and parameter configuration of DNN-HMM are intensively studied. A new hybrid network model is proposed by using clustered state instead of single factor state as the output unit of the neural network, and it is applied to the acoustic model of English speech recognition. The simulation results show that the improved scheme can get better recognition effect by increasing the enhancement of speech features under the condition of three-phoneme structure. In addition, the experimental results also verify that the acoustic modeling method based on DNN-HMM is superior to traditional GMM-HMM method.

Research on the Construction of E-commerce Security Risk Assessment Model Based on Cloud Computing

Abstract: The use and popularity of cloud computing in e-commerce has injected new power into the fast-growing e-commerce industry. However, the popularity and development of e-commerce still faces a series of problems. The most prominent of these problems is the security issue. Risk assessment is the basis for risk control and guaranteeing transaction security. However, due to the particularity of the e-commerce security system skeleton and security requirements in the cloud computing environment, the assessment of e-commerce security risks faces new challenges. Based on the hierarchical structure of e-commerce security system, this paper analyzes the security requirements of e-commerce and proposes a quantitative e-commerce risk assessment model based on cloud computing.

Volleyball Footwork Movement Teaching Mode Based on the Fusion of Frontal Double Handed Overhead Pass Technical Action Model

Abstract: In view that the footwork movement in volleyball may be in line with the features of the dynamic discrete process, in accordance with the characteristics of the frontal double handed overhead pass technique, comprehensive study and observation is carried out on the steadiness in the footwork movement of the athletes at the venue of the match to explore the effect of irregularity on the volleyball motion path of the athletes at the venue of the match and understand the relationship between the footwork movement and the steadiness optimization of the athletes. In addition, the underlying conditions and theoretical formulas for the parameters related to the selected athletes are provided to carry out the exploitation capacity. From the experiment, it can be concluded that the method put forward in this paper is conducive to the pre-judgment and adjustment of the footwork movement of the volleyball players in the actual volleyball matches.

Design and Implementation of Electronic Archives Information Management Under Cloud Computing Platform

Abstract: This paper designs a Hadoop cloud platform for electronic archives information management, and designs its functional modules. The pre-processing and eigenvalue extraction of electronic archives data can be realized through Mapreduce programming model, and the data can be stored in an infinitely expanded base for call at any time. During the process of experiment, the experimental environment is established and configured to analyze the relevant data and perform data pre-processing experiments. The processed data is stored in HBase instance, which basically achieves the expected functions. At the same time, the experiment shows that the archive-Cloud platform based on Hadoop can meet the basic requirements of electronic archives data preprocessing and data storage.

An Efficient Semantic Retrieval Method for Network Education Information Resources

Abstract: To improve the performance of network education information resources system, this paper proposes ontologybased similarity algorithm and semantic query expansion method. We mainly study ontology reasoning and define reasoning rules based on knowledge point ontology. Combining with various conceptual relationships in knowledge point ontology, we define corresponding rules and verify them by experiments, which lays a foundation for semantic extension. Then, a domain ontology-based information retrieval model is proposed based on the above content. By separating the relationship between concepts, the semantic similarity and correlation are calculated by using the conceptual hierarchy and association relationship and considering various factors, the semantic similarity and correlation are synthetically quantified into the semantic association degree for semantic retrieval. Finally, the prototype system of semantic retrieval is designed and validated, and the simple retrieval of resources is realized.

Coordination Scheduling Model for Minimum Energy Consumption of Cogeneration of Cool, Heat and Electricity

Abstract: In today's world, there is an energy crisis. In order to save energy, energy must be used to the maximum extent according to the final requirements of the user. In this paper, the combined cold, heat and electricity (CCHP) plant is modeled and optimized, in which the minimization of coal consumption is set as the objective function. Power and refrigeration requirements are initially set by the factory and, depending on the circumstances, variables can be changed every 24 hours after the optimization process is implemented. CCHP systems have thermal, electrical, mechanical and chemical components. This means that there are multiple energy domains. This article also takes these shadows into account when setting objective functions Sound, and in the optimization process to carry out a full study.

Design and Implementation of Core Modules of WEB Application Vulnerability Detection Model

Abstract: In view of the shortcomings of existing vulnerability detection tools, an efficient Web application vulnerability detection system model based on crawler and feature recognition is proposed in this paper. The model is optimized from two aspects: scanning information collection and vulnerability detection efficiency. The network crawler module proposes a multi-threaded crawler mechanism based on breadth-first strategy. It extracts and scans a large number of duplicate and useless links through Hash Table duplicate checking, URL duplicate checking and parameter duplicate links, to improve the efficiency of scanning information collection and detection. The feature recognition module reduces the large amount of information exchange process needed to scan the website through two phases, namely, the identification of known websites and the screening of test objects based on VSM, which improves the scanning efficiency and reduces the false alarm rate to a certain extent.

Improved Particle Swarm Optimization Image Registration Based on Mutual Information

Abstract: An improved particle swarm optimization (PSO) image registration algorithm based on mutual information is proposed. Firstly, the mutual information value of the image is calculated, and the mutual information is used as the fitness function of PSO to seek the global optimal value of the image and realize image registration. Then, the registration and the floating image are fused. The results show that the algorithm not only improves the ability of optimization, speeding up the speed and accuracy of the algorithm, but also makes the results of image registration more accurate.

DVC Decoding Algorithm Based on Wireless Transmission

Abstract: This letter presents an improved distributed video coding (DVC) decoding algorithm is proposed to improve the performance of a multipath fading wireless channel. Considering the influence of channel error on Wyner-Ziv and key frame bit stream, a new noise model is proposed and the algorithm is improved accordingly. Using the W-CDMA wireless channel to simulate the simulation, and analyzing the simulation results, to determine the effect of each change. The simulation results indicate that the proposed improvement has a significant improvement in the performance of the DVC decoder under the adverse channel conditions.

Non-Intrusive Load Monitoring Method Based on Improved Differential Evolution Algorithm

Abstract: Non-intrusive load monitoring is an efficient way to identify the load and get the details of the user’s power consumption information. This paper proposes a new nonintrusive load monitoring method based on improved differential evolution algorithm. Firstly, the advantages of non-intrusive load monitoring are briefly introduced. Secondly, the detailed procedure of load identification of household appliances by improved differential evolution algorithm is described. At last, an experimental platform is established to test the effectiveness of the proposed nonintrusive load monitoring method. By introducing the signal-to-noise ratio and load identification rate, the performance of improved differential evolution algorithm is studied.