Ebook: Intelligent Computing Technology and Automation

Artificial Intelligence is a rapidly developing field of computer science that integrates multiple disciplines, such as computer science, psychology, and philosophy. It is a technology that can study and develop theories, methods, technologies, and application systems to simulate, extend, and expand human intelligence by attempting to understand the essence of intelligence, producing a new, intelligent machine that can respond in a similar way to human intelligence. Research in this field includes robots, language recognition, image recognition, natural language processing, and expert systems.

Artificial intelligence plays an increasingly important role in the development of global industries and economies, such as the information industry, agriculture, bio-informatics, education, and the economy. It is currently changing our world significantly, and its development depends heavily on artificial intelligence theory and technological progress, making research in AI a hot topic all over the world.

The 16th th International Conference On Intelligent Computing Technology and Automation (ICICTA 2023) was held in Xi’an, on 24 & 25 October 2023. It is an annual forum dedicated to emerging and challenging topics in artificial intelligence and its applications. The aim of the conference is to bring together an international community of researchers and practitioners in the field of AI to share the latest research achievements, discuss recent advances, influence future direction, and promote the diffusion of the discipline throughout the scientific community at large.

ICICTA 2023 received 322 submissions, and all papers underwent a rigorous peer-review procedure with each receiving at least 2 review reports. Based on these, 141 high quality papers were ultimately accepted and are included in these proceedings. We hope that these contributions will provide effective channels for peer-to-peer communication.

On behalf of the conference chairs, we thank the members of the organizing committees and the program committees. We would also like to express our heartfelt appreciation to the keynote speakers, reviewers, editors and to those students who assisted with the conference. Last but not least, we thank all authors and participants for the contributions that made this conference possible and all the hard work worthwhile.

We hope to see you at the next ICICTA in 2024.

Yihu WU

Changsha University of Science and Technology

ICICTA2023 General Chair

To better analyze the power energy optimization problem in the integrated energy system, this paper uses PSO algorithm with good global search ability as the scheduling model solution. Firstly, we propose the joint optimal scheduling model of power integrated energy system with the participation of user side, and provide corresponding algorithm objectives and constraints. Then, PSO algorithm using inertia weight to control the learning factor is adopted to improve its convergence speed and obtain the optimal solution of equipment output and performance index. Finally, through specific examples, the model is verified and the optimal scheduling scheme of the integrated energy system under the objectives of economy and environmental protection is acquired. The results show that the scheme can effectively avoid particles falling into local optimization, and accurately calculate the reliability quantitative index of the comprehensive energy system in the target area, which ensures the coordinated interaction of endogenous, storage and load in the power system.

Considering the fluctuation of predicted output power of new energy units such as wind power, a strategy to strengthen the dispatching optimization of new energy distribution network is proposed. Taking the constraints such as operation cost, voltage deviation and network loss of distribution network as a reference, we provide the description of wind power active power control. Then, combined with stochastic chance constrained programming, an improved particle swarm optimization algorithm is proposed to establish the optimal dispatching model of “source-network-load” distribution network. Finally, the effectiveness of the models and methods proposed in this chapter is verified by a simulation example of IEEE-33 node distribution system. The results show that the comprehensive optimization of the two is better than the reconstruction or optimization separately, which can facilitate the dispatcher to select the best optimal scheduling strategy.

To realize the optimal allocation and target controllability of automatic generation control (AGC) regulation capacity, an AGC allocation method based on genetic algorithm(GA) is proposed. Firstly, the basic principle of AGC is analyzed, and the mathematical model of unit allocation to find the optimal solution under constraints is provided. Then, GA is tentatively applied to the allocation of a AGC unit, and the crossover and mutation are performed according to the capacity segment and state segment respectively. Combined with the priority method, the continuous variables in the model are processed to improve the convergence performance of AGC. The example of the actual system shows that our method can overcome the disadvantage that the integer programming method may not acquire the optimal solution, and achieve a certain economy, which also provides a new and effective strategy for the allocation of AGC units.

To effectively evaluate the aging and moisture status of high-voltage bushings, a method for evaluating the insulation status of high-voltage bushings is proposed. This article analyzes the time-domain dielectric properties of oil-paper composite insulation in different aging states, and proposes the use of time-domain dielectric spectroscopy method to solve the equivalent circuit parameters of transformer oil-paper insulation under uncertain relaxation terms. In order to expand the application range of time-domain dielectric spectroscopy, Morlet wavelet is adopted to transform such signal and reduce background interference in high-frequency current. In the simulation section, a simulation model of high-voltage bushing oil-paper insulation structure is established. By analyzing the relaxation process of oil-water mixture, the insulation status of high-voltage bushings with different degrees of aging is tested. The experimental results show that time-domain dielectric spectroscopy analysis based on wavelet transform can acquire overall spectral information reflecting physical properties, which can be used for insulation detection and condition evaluation of oil-paper high-voltage bushings.

Aiming at the diverse types and complex judgments of electrical faults, an improved BP neural network model is proposed for fault diagnosis of CNC machine tools. By integrating the strong global search ability and fast optimization speed of particle swarm optimization(PSO) algorithm, the premature phenomenon that often occurs in neural network algorithms in diagnosis is improved to enhance the diagnostic ability of RBF algorithm. The algorithm corrects the problem of individual particle actions and the tendency of standard PSO and neural networks to fall into local minima. The experiment and simulation results based on MATLAB indicate that compared with traditional BP neural networks and fuzzy neural networks, PSO optimizes neural networks which have higher accuracy in fault identification and stronger generalization ability. The application of this scheme in fault diagnosis of CNC machine tools can effectively improve the efficiency of fault diagnosis.

In order to meet the needs of large-scale traffic control and traffic flow guidance coordination under the background of urban expansion, a regional traffic flow coordination control scheme is proposed by using agent and fuzzy control method. Firstly, the basic structure of Agent in intelligent traffic control is analyzed, and on this basis, an intelligent decision support system model for traffic flow prediction is established. Then, aiming at the shortcomings of agent in micro traffic simulation, it is proposed to combine other cellular automata for modeling. The fuzzy control strategy is applied to design the coordination controller for the execution strategy of the decision-making unit. The simulation results in MATLAB environment prove the effectiveness of the intelligent vehicle flow coordination control model, which can carefully describe the characteristics of traffic flow and the micro behavior of traffic entities.

To recommend video resources in education for better results, this article attempts to use a deep learning model to construct a system suitable for mobile devices. According to the specific design principles and requirements analysis, the overall logical framework of the system and the main technology of development are provided. Then the composition of server and client is explained by B/S system architecture, and the function realization scheme of key modules of the system is provided by video coding and compression technology, MSSQL database technology and ASP.NET. The recommendation module adopts the deep learning lightweight model MobileNet to improve detection speed and accuracy, and reduce memory, which achieves real-time detection and pushing effects. The empirical analysis of the system shows that it has the characteristics of simple and fast interface, safe and reliable. It also realizes the mining and matching of massive potential information in videos, so it has an important auxiliary role for the ideological and political teaching.

Aiming at the problem of improving the navigation accuracy of ships, a Kalman filter algorithm based on optimization is proposed for navigation optimization. On the basis of studying the dead reckoning knowledge, we establish the dimensional dynamic equation of the ship integrated navigation system. Then the extended Kalman filter is adjusted by using fuzzy logic adaptive controller (FLAC) and it is applied to GPS/INS information fusion to enable it to have adaptive ability to deal with environment disturbance. Finally, the method is used to simulate the GPS/INS integrated navigation system on MATLAB platform. The results show that the improved model can effectively use part of the system data to update and iterate, and improve the stability and reliability of the filter, so it is also better suitable for ship navigation and target tracking.

Aiming at the current problems of single feature extraction of motor imagery EEG signals and low accuracy of classification and recognition, a feature extraction method of motor imagery EEG signals based on the fusion of PSD and CSP (PSD-CSP) is proposed. Firstly, the FastICA algorithm is employed for artifact signal removal from the raw EEG data. Subsequently, features are extracted from the Power Spectral Density (PSD) and Common Spatial Pattern (CSP), followed by their serial fusion. Finally, a Support Vector Machine (SVM) classifier is used for classification. In binary motor imagery classification experiments on the BCI Competition IV Dataset I, the average classification accuracy reaches 91.43%, and comparative analysis with other methods demonstrates the feasibility of the proposed fusion algorithm.

In order to improve the accuracy of human motion data acquisition and reduce the time-consuming of data acquisition, an intelligent motion Bracelet human motion data acquisition system based on human-computer interaction is proposed and designed. Firstly, the overall architecture of the system is designed, mainly including application layer, transmission layer and perception layer. Secondly, the system hardware is designed, mainly including the motion data acquisition module, temperature acquisition module and ZigBee wireless transmission module of human-computer interaction intelligent sports bracelet, so as to ensure the realization of the system function. Finally, in the part of system software design, wavelet transform is used to discretize the collected human motion data and divide the attributes, so as to improve the availability of the data. The experimental results show that compared with the traditional motion data acquisition system, the designed system has higher acquisition accuracy and shorter time-consuming.

In order to solve the problems of low estimation accuracy and long estimation time of traditional construction project cost estimation methods, a new fast estimation method of green construction project cost based on support vector regression machine is proposed in this paper. Firstly, according to the principle of support vector regression machine, the nonlinear classification function of green building cost data is constructed to complete the classification of cost data. Secondly, based on the classification results, the principal components of the sample data series of cost estimation are calculated, and the constraint parameters of cost estimation are constructed. Finally, according to the relationship between green building construction cycle and overall project cost, a fast estimation model of green building project cost is constructed. The experimental results show that compared with the traditional estimation model, the estimation accuracy of this method is higher and the estimation time is shorter.

In order to provide an efficient and reliable way to address recruitment challenges in talent resource management and improve the efficiency and quality of the entire recruitment process, a big data-based intelligent recommendation method for talent resources is proposed. Firstly, using the principles of big data mining, construct an adjacency matrix to complete the mining of talent resource data. Secondly, based on the excavated talent resource data, describe the characteristics of the talent resource data and calculate the similarity of talent resource scores and labels. Finally, based on the predicted results of talent resource ratings, complete the recommendation of talent resources. The test results show that the method proposed in this paper can reduce the root mean square error between the actual and predicted scores of talent resources, and improve the accuracy of talent resource recommendation.

In order to improve the convergence of the algorithm and the utilization of human resources, a new human resources allocation optimization method based on improved hybrid genetic algorithm is proposed in this paper. Firstly, according to the relationship between post demand and human resources, the objective function of human resources allocation optimization is constructed. Secondly, the principle of traditional hybrid genetic algorithm is analyzed, and tabu search algorithm is introduced to improve the traditional hybrid genetic algorithm. Finally, the improved hybrid genetic algorithm is used to solve the objective function to complete the optimization of human resource allocation. The experimental results show that compared with the traditional optimization methods, the convergence of this method is better, and the utilization rate of human resources is significantly improved, which is always maintained at the level of more than 92%.

This article proposes a design scheme for intelligent equipment for remote search and rescue in the event of a mine disaster. This design uses STM32 and ESP32 as the core for data processing and transmission. Multiple sensors are used to intelligently detect harmful gases, distinguish the presence of human voices, and automatically avoid obstacles. Remote control is achieved through Bluetooth technology, and MIME technology is used for real-time image transmission on site. By detecting signal strength, relative positioning is achieved using hyperbolic positioning method. The intelligent remote search and rescue vehicle based on STM32 can detect the harmful gas content and presence of vital signs at the rescue site while ensuring normal rescue, and collect real-time images of the scene.

When the temperature of the cable core is too high, it will accelerate the aging of the insulation material, affect the safe operation of the cable, and shorten the service life of the cable. Therefore, a simple and effective way to monitor the cable core temperature in real time is an important measure to improve operation and maintenance efficiency and reduce the risk of failure. There are many problems with the traditional cable temperature measurement methods, such as the large influence of ambient temperature, low measurement accuracy, and high application cost. In order to solve these problems, this paper designs a set of low-power non-invasive high-voltage cable core temperature monitoring system based on the non-invasive high-voltage cable core temperature measurement principle of temperature field construction and sensor array. In this paper, through theoretical analysis and with the help of experiments, we obtain the temperature gradient law of 10kV true high-voltage cable after local wrapping of heat-resistant medium and derive the analytical calculation formula of the cable core temperature.

In order to improve the accuracy of identifying overheating faults in electrical equipment and shorten the response time, a method for identifying overheating faults in electrical equipment based on infrared technology is proposed. Firstly, infrared images of electrical equipment are collected through an infrared scanner. Secondly, in order to improve the quality of infrared images, feature extraction is performed on the collected infrared images. Finally, the infrared image is enhanced through nonlinear NSCT transformation, and the identification of overheating faults and defects in electrical equipment is completed by calculating the relative temperature difference of the tested electrical equipment. The experimental results show that compared to existing fault identification methods, this method can accurately identify overheating faults of various electrical equipment, and the identification time response is significantly shortened.

In order to improve the accuracy of risk recognition in power operation site, an intelligent remote monitoring method for power operation site based on video recognition technology is proposed. First, according to the video recognition architecture, the pictures of the power operation site are collected through the video camera. Secondly, in order to improve the quality of the image, histogram enhancement, smooth denoising and gradient sharpening are performed on the collected image. Finally, the preprocessed image is input into BP neural network, and the risk recognition of power operation site is obtained by adjusting and calculating the threshold and weight. Through the risk recognition results, the monitoring terminal personnel can take targeted measures. The test results show that this method can accurately monitor the risk of power operation site, and can improve the query processing times per second, which can meet the requirements of intelligent remote monitoring of power operation site.

In order to meet the mileage requirements of new energy vehicles, it is necessary to improve the capacity of power batteries. Therefore, this paper proposes a new optimization method of energy storage capacity of power batteries of new energy vehicles based on fuzzy control. Firstly, according to the control block diagram of power battery, the fuzzy control method is used to control the power of power battery, and the relationship between the sum of power fluctuation out of limit and the change of battery rated energy storage capacity is obtained. Secondly, the objective function of energy storage capacity optimization of new energy vehicle power battery is constructed, and the constraints of energy storage capacity optimization of new energy vehicle power battery are constructed based on the constraints of battery charge and discharge power and SOC. The optimization of energy storage capacity can be realized by solving the objective function under constraints. Finally, a comparative verification experiment is carried out. The experimental results show that this method can effectively control the power fluctuation of power battery and improve the energy storage capacity of battery.

The classification of weeds and crops is key to precision agriculture. To achieve the challenge of recognizing and classifying weeds and crops with high accuracy, this paper uses a deep learning approach. We propose the SE-ResNext network to optimize ResNext using the channel attention mechanism. Embedding the channel attention mechanism in the backbone network of ResNext, this deep learning model achieves 97.51% classification accuracy on the V2 Plant Seedlings dataset. The training accuracy of the SE-ResNext model is 98.49% and the MSE of the SE-ResNext model is 0.044, which are the best values. Compared to AlexNet and GoogleNet, this model has higher classification accuracy. The results can provide technical support for the autonomous operation of field-weeding robots. Enables accurate classification of weeds and crops.

In order to properly solve the problems of scattered data and difficult information updating in urban road information management, the cluster search scheme is used to improve the management efficiency of road planning. This paper takes the urban disaster prevention as example, and reveals the contradiction between the existing urban form in several cases. To shorten the average waiting time and working distance of stored vehicles, the corresponding mathematical model of cluster search is established by investigating the collected data and extracting the data. Then combining with the actual garage layout and taking the distance of stacker, the average waiting time of customers and the average waiting queue length as the measurement index, the simulation program is compiled with MATLAB, and the effects of different garage layout modes on the overall operation efficiency under a certain storage capacity are analyzed and compared. The results show that cluster search can shorten the running distance by about 50% and the average waiting time and queue length by about 31% and 76% respectively.

Changes in the membrane potential of neurons in the cerebral cortex produce measurable electrical brain activity. Electroencephalograms (EEGs) measure and record these electrical activity signals through the montage of the brain electrical activity signal channels formed by non-invasive sensing electrode arrays attached to the cortex. As a non-invasive detection method, EEG is widely used in the research fields of neuroscience and cognitive psychology. With the rapid development of electronic and computer technology, the development of Brain Computer Interface (BCI) technology has been promoted. This paper proposed a self-adaptive coupled motion control strategy based on the current working pose and micro-torque feedback. Compared with the conventional control way, no matter the coherence or the stability, this self-adaptive coupled motion control strategy manifests the better performance than conventional non-feedback mechanism.

At present, the industrial park has problems such as complex and unbalanced electricity consumption and low utilization rate of electricity resources. When the load is peak, the power supply of the electrical equipment is insufficient, which seriously affects the production quality of the park. In order to solve the problems of peak-to-valley difference and insufficient power resources, this paper proposes a power resource allocation method based on edge computing. This method adjusts the power scheduling by the power consumption duration of the flexible load, establishes the power resource allocation model, and solves it with the improved particle swarm algorithm (I-PSO), that is, the roulette algorithm is introduced into the particle swarm algorithm. Through simulation verification, it is concluded that the industrial park based on edge computing has higher utilization of power resources, reduces electricity consumption cost, and achieves the goals of efficiency reallocation power resources, reducing carbon emissions.

In order to reduce the relative error of carbon emission prediction and improve the prediction efficiency, a new prediction method of coal combustion carbon emission of power generation enterprises based on rough set and grey SVM is proposed in this paper. Firstly, according to the rough set theory, the carbon emission data of power generation enterprises are deeply mined to enrich the data sources of carbon emission prediction. Secondly, grey GM model and SVM model are constructed according to the mining results. Finally, combined with the grey GM model and SVM model, the prediction model of coal combustion carbon emission of power generation enterprises is constructed to complete the prediction of carbon emission. The experimental results show that compared with the traditional prediction methods, the prediction relative error of this method is significantly reduced, and the prediction time is significantly reduced.

in order to improve the security performance of multi-level residual network in the presence of DDoS attacks, a multi-level residual network DDoS attack detection method based on random forest is proposed. The random forest method is used to classify the multi-level residual network DDoS attacks, the Stirling approximation is used to obtain the entropy change rate, and the multi-level residual network DDoS attack detection is realized based on the comprehensive correlation of mutual information between features. The experimental results show that when using the improved method, the detection accuracy is 98.71%, the detection time is 36.87s, the stability is high, and has certain advantages.