Abstract： In the operation process of the value chain of group-type manufacturing enterprises, there is a problem of insufficient power supply during peak periods, and the demand for production electricity cannot be effectively guaranteed. The group company urgently needs to comprehensively consider the synergistic relationship between the production demand of the production base and the safety of electricity consumption, complete the fine-grained upstream and downstream power distribution of the value chain, and ensure the safe and reliable operation of the value chain. This paper designs a blockchain-based upstream and downstream electricity security collaboration system in the value chain, adopts a method based on smart contracts to ensure the compliance of value chain data use, and effectively solves the risk of value chain data sharing of group-type manufacturing enterprises and conflict of values in each production base. The system provides a good reference and decision-making basis for the safety guarantee of electricity consumption in the upstream and downstream of the value chain by completely recording the order coordination data of the production base and the electricity load data of the production base, which is of great significance for maintaining the company's safe and stable operation and improving the company's economic benefits.

Abstract： The current industrial electricity prices in China have peak and trough periods in a day, and cement production is characterized by high electricity consumption. The optimization of cement production arrangements can improve the utilization rate of production equipment and use the optimization results to study the production regulation capacity, so that the production can be adjusted according to the peak and trough electricity prices to achieve cost savings and reduce the load on the power grid. The study uses state task network to model the cement production process with and without storage process, taking into account electrical safety constraints, and obtains the optimal production arrangement by solving the corresponding mixed integer linear programming model, from which it is concluded that the cement production process with storage process can provide more production regulation capability.

Abstract： Effective power maintenance schedule will significantly increase the reliability of power grid operation as a crucial component of assuring the normal operation of the power system. In order to achieve the optimization of the maintenance plan, a probabilistic maintenance model is created in this work using equipment condition categorization, equipment operation life, equipment operation cost, and other indicators. The Markov Chain Monte Carlo approach is utilized in the maintenance model to increase model accuracy, and the probabilistic maintenance plan optimization strategy is employed to improve the reliability and economic indicators of the power grid. Finally, simulation is used to assess and compare the probabilistic maintenance model and the conventional maintenance model. The superiority of the probabilistic maintenance model, which offers a theoretical foundation for the enhancement of the power maintenance link in the power industry value chain system, is highlighted while determining the best maintenance probability.

Abstract：Electricity safety inspection is an important way to ensure the normal operation of the power grid. Traditional electricity safety inspection mainly relies on manual inspection of places and equipment with safety hazards one by one. With the development of artificial intelligence technology, intelligent analysis based on image data can assist in timely identification of relevant safety hazards, reduce the experience requirements for inspectors, and improve efficiency while ensuring the accuracy of safety inspections. In order to better improve the accuracy of electricity safety inspection, the article proposes an electricity hazard identification algorithm based on YOLO neural network, which can dynamically identify the indicator lights of electrical equipment and compare them with normal states, and promptly issue alarm messages when abnormal states are found. Based on this algorithm, the article also designed and implemented an electricity safety inspection subsystem based on image recognition. Through actual data validation, the system can achieve a high level of inconsistent detection of equipment indicator status, meeting the demand for electricity safety inspection.

Abstract： With the increase of electric vehicle penetration, the coupling of power network and transportation network in time and space is closer. In order to reduce the challenges brought by the disorderly integration of electric vehicles to the safe and stable operation of power system, this paper fully considers the influence of power load demand and traffic condition uncertainty, and establishes a multi-agent interest hierarchical game model of EV users-charging station aggregators-grid operators. On the premise of satisfying the constraints, the multi-agent interest hierarchical optimization problem is solved based on improved bat algorithm. The simulation example shows that the grid operators can obtain higher operating benefits by actively adjusting the pricing strategy for different electric vehicle penetration rates. The proposed method can effectively improve the voltage quality of power grid to ensure the safe and stable operation of the system, and provide theoretical support for improving the value chain system of power industry.

Abstract： In the era of big data, accurate and timely identification of research frontiers is particularly important for the positioning of science and technology strategy and the deployment of scientific research direction, which can provide more comprehensive reference and basis for decision-making. The self-developed multi-source data fusion analysis method was used to conduct data mining and denoising from global journal papers, conference papers, patent literature, science and technology news, research reports, science and technology policies, standards and other data. The list of current research frontiers and technologies in the field of IC was effectively identified. At present, the research hotspot and development direction of global integrated circuit field are concentrated on semiconductor manufacturing equipment and high process manufacturing technology, and the focus is on high-end chips, power devices, sensors, memory and third-generation semiconductor related products. The localization of China's semiconductor industry is imminent. The results of frontier identification provide valuable decision support and reference for the government and enterprises.

Abstract：Human key point detection has important applications in intelligent video surveillance, human-computer interaction and other fields. Aiming at the problem that the human key point detection algorithm based on heatmap depends on high-resolution heatmap and consumes large computational resources, a lightweight algorithm combined with uncertainty estimation is proposed. The reliability of prediction results is improved by using low resolution heatmap and combining uncertainty to estimate the scale parameters of prediction error distribution. The scale parameter is used to monitor and constrain the heatmap to alleviate the gradient disappearance and enhance the robustness of the network. The experiments on COCO dataset show that the average accuracy of the improved algorithm is improved by 3.3% and the resource occupation is reduced compared with integral pose regression.

Abstract：The background of the power transmission line inspection image is complex, and the target detection is easy to be disturbed. Based on YOLOX neural network model, this paper proposes a method of power transmission line mountain fire detection. Firstly, the backbone feature extraction network framework of YOLOX is adopted, and the conventional convolution of the multi-scale feature extraction module is replaced by deformable convolution. Secondly, the fusion of channel attention and spatial attention modules is added in the enhanced feature extraction stage, which can adapt to the variable shape of flames, extract mountain fire features more effectively, and thus improve the accuracy of target detection. The experiment verifies the effectiveness of the proposed method.

Abstract：This paper presents a 10-bit Successive Approximation Register (SAR) analog-to-digital converter (ADC) with 20 MS/s sampling frequency. By using the piecewise capacitor array design, the settling time after large capacitor turnover during quantization is shortened and thus the quantization speed is improved. In addition, a novel and efficient comparator calibration method is proposed to reduce the offset voltage of the comparator at a lower cost. The chip is manufactured in 180 nm CMOS process with a core area of 0.213 5 mm2. The test results show that the ADC achieves 58.24 dB signal-to-noise/distortion ratio (SNDR) at 1.8 V supply voltage while sampling at 20 MS/s

Abstract：To reduce the high resource consumption and high data dependency of the CABAC entropy decoding module in the new generation of High Efficiency Video Coding (H.265/HEVC), a multiple parallel CABAC decoding hardware circuit architecture is designed. According to the characteristics of different types of grammar elements in the decoding process, grouped parallel data scheduling is used to reduce the waiting time and the number of internal memory accesses for data processing, while pipeline technology is used to achieve acceleration of the arithmetic module. The design results show that the entropy decoding module has a throughput of 1.84 bins/clock cycle, meeting the current requirements for real-time decoding of UHD video.

Abstract：An integrated computer system based on cloud architecture is constructed, the localization rate is 100%. And the application technology of space data multicast based on cloud platform is proposed. Thus， The technical optimization of operating system, cloud platform and virtual machine CPU resources is realized, which improves the performance of data multicast in cloud platform. Through the actual test, the improved performance can meet the business needs of the space mission center.

Abstract：Electronic nose is a bionic sensing system, which can identify many gases at the same time, so it is used in many fields. The gas concentration detection algorithm is the core part of the gas quantitative analysis by electronic nose. In order to improve the accuracy of the electronic nose concentration detection algorithm, a prediction model based on online sequential-extreme learning machine (OS-ELM) is proposed. The model uses one-dimensional convolutional neural network (1DCNN) to extract features, uses OS-ELM to predict gas concentration, and proposes an improved Particle Swarm Optimization(PSO) algorithm to overcome the problem that OS-ELM needs to manually adjust model parameters. The theoretical analysis shows that the improved algorithm has stronger search ability than the traditional PSO algorithm. Finally, the experimental results show that the proposed model has higher prediction accuracy and generalization ability compared with the traditional prediction model.

Abstract：Aiming at the problems of high authentication delay in the existing satellite-ground cooperative authentication mode, and the ground authentication server is vulnerable to failure when attacked, this scheme innovatively proposes a switchable and efficient multi-mode authentication scheme. The scheme adopts the authentication mode of primary use on the satellite and standby on the ground. Under the authentication mode on the satellite, it can effectively reduce the high communication delay caused by the traditional satellite-ground authentication transmission process, and achieve fast authentication without relying on the ground attribution network. When the on-satellite authentication mode is not available, the ground authentication mode can be switched quickly to ensure the stability and reliability of the authentication process and improve the system stability and survivability. At the same time, the scheme uses hash algorithm and symmetric encryption algorithm to complete the authentication core operation, effectively reducing the calculation cost of the authentication process, and has high application value in improving the access efficiency and security of satellite communication equipment.

Abstract：With the rapid development of automobile industry and communication technology, high quality traffic information in vehicle is of vital importance to the emergency situation. High-speed mobility of the vehicle in the car networking will lead to interrupt the risks to information transmission, at the same time with the increase of vehicle number, the limited spectrum resources connected to the power allocation poses challenges for car. In order to solve this problem, this paper uses the single frequency network technology, through resource allocation strategy to reduce the letter the interrupt probability of dry ratio, minimize the single frequency network transmission power of each side of the road in the unit. The optimization problem is modeled as a Markov decision process, and is solved by curiosity-driven DQN (C-DQN) resource optimization algorithm. A large number of simulation results show that the scheme can minimize transmission power on the premise of low interrupt probability, compared with the baseline algorithm, the adopted algorithm has a good performance in the learning speed and stability.

Abstract：Due to the comprehensive effect of environmental factors and growth mechanism, the sap flow often presents the characteristics of nonlinearity and high randomness, and it is often difficult to predict it accurately by a single prediction method. This paper proposes to introduce the empirical wavelet transform (EWT) method to decompose the nonlinear and highly random ginkgo sap flow data to obtain two sets of multi-resolution components, and the ARIMA model is used to predict the components respectively. According to the results, it is proposed that the EWT-ARIMA model can accurately predict the change trend of sap flow, and the model evaluation indicators MSE, MAE, MAPE, R2 are 11.05, 0.1640, 0.9599 and 0.9598, respectively, which are greatly improved compared with the single ARIMA model. In this paper, transfer entropy (TE) is also used to explore the causal reflection between environmental factors in time delay and ginkgo sap flow without model assumptions.

Abstract： Image data decompression is a kind of important data processing problem. Data feature learning has important research value in data compression research. This paper proposes a feature representation model of variational autoencoder based on cloud model, which takes the cloud model as the prior distribution of variational autoencoder and solves its limitation in feature representation. The encoder of the variational autoencoder is responsible for constructing the feature space of data, obtaining hidden variables by sampling in the space, and completing data compression. The decoder completes the generation from data features to raw data, that is the decompression of data. The correctness and effectiveness of the proposed method are verified by the experimental comparison with the original method on face dataset.

Abstract： Aiming at the inefficiency of multi-dimensional association rules mining algorithm and the existence of redundant rules, an algorithm of multi-Dimensional apriori with constraints is proposed. Based on the multi-dimensional Apriori algorithm, the algorithm controls the mining process with user constraints. According to the predicate constraint, the frequent predicate set that is of interest to the user is generated, and the transaction set is deleted based on the predicate constraint. On the one hand, the algorithm greatly reduces the generation of candidate predicate sets through user constraints. On the other hand, the reduced transaction set also reduces the scanning database overhead. Finally, the efficiency of mining is improved and the redundant rules are reduced. This algorithm is used to compare experiments on FPGA code defect transaction sets. The experimental results show that compared with the multi-dimensional Apriori algorithm, this algorithm has improved the search efficiency of frequent predicate sets and the accuracy of mining results.

Abstract： In IC verification,due to the complexity and multiplicity of the JTAG protocol,the code written separately in TestBench for verification is long and difficult to maintain.Sometimes some companies and groups put this part of verification in FPGA prototype verification.In prototype verification,some modules need to be replaced,which cannot be guaranteed to be the same as the RTL level netlist.It may lead to the failure of chip tog debugging after streaming.In view of this situation,this paper proposes an implementation method for joint verification of JTAG debugging protocol based on UVM and C language.Combining the universality of UVM methodology and the convenience of C language,UCM builds a framework for verification of JTAG protocol,C language verification of chip JTAG protocol is realized by calling the chip JTAG interface implemented on the UVM side to drive the timing.

Abstract： The quantitative remote sensing monitoring of Fengyun satellite for Marine environmental disasters (ulva prolifera, red tide and oil spill) should to be improved. This paper uses the MERSI data of FY-3D satellite data, as well as some high-resolution satellite data,to explore methods to solve the two technical difficulties of atmospheric correction and cloud detection. Based on the spectral characteristics of ulva prolifera, red tide and oil spill, relevant monitoring algorithms were established for identifications of ulva prolifera, red tide and oil spill, and there were good monitoring performances in corresponding cases.

Abstract：Chinese herbal medicines play an important role in the prevention and control of human diseases, but the general public's knowledge of Chinese medicinal materials is too little, which may bring uncontrollable consequences due to the abuse of Chinese medicinal materials. Therefore, the accurate identification of Chinese medicinal materials is an urgent task. In this paper, the lightweight neural network model is applied to the recognition of Chinese medicinal materials, and an image recognition system based on the MobileNetV3 model is proposed on a microcontroller. Firstly, the image dataset of Chinese medicinal materials is established, the recognition basic model is built according to MobileNetV3 in the eIQ machine learning software development environment, and the model is optimized by adjusting the model parameters, and finally the model file is deployed to i.MX RT1060. Image recognition of 30 kinds of Chinese medicinal materials was realized, and the accuracy rate in the verification set reached 86.79%. The results showed that the image recognition of Chinese medicinal materials on i.MX RT1060 has a good practical effect.

Abstract： Obstructive Sleep Apnea Hypopnea Syndrome (OSAHS) is a common respiratory sleep disease. It decreases the quality of people's sleep and causes fatigue, and more seriously, it harms people's physical and mental health. The study designed an ARM-based OSAHS detection system. The system uses i.MX6ULL as the hardware master control and the embedded Linux system as the software platform. It has functions such as snore collection and processing, detection and classification, transmission, etc. The system has established a complete OSAHS detection system with the cloud platform. By comparing with the standard polysomnography (PSG) device, the detection effect reaches 83.9%. It achieves the role of primary screening and has strong auxiliary diagnostic application value.

Abstract： In order to obtain stable electron beam output and improve the quality of electron beam selective melting, a set of electron beam selective melting beam control system based on DSP was developed. DSP's general timer and comparison register are used to generate PWM pulses with a frequency of 20 kHz and adjustable duty cycle and dead time; a feedback sampling circuit for high-voltage and beam current signals is designed; DSP uses integral separation PID control algorithm to achieve rapid beam current adjustment and closed-loop control. The test results show that the maximum beam current of the electron gun controlled by the control system reaches 150 mA under load. When the high voltage and the heating current of the filament are given, the beam current gradually increases with the decrease of the bias voltage，and a stable beam current output is realized.

Abstract：When the eigen-projection matrix preprocessing algorithms suppress the mainlobe interference, the sidelobe interferences will shift after preprocessing and the output performance is seriously degraded. To solve this problem, the mainlobe interference suppression algorithm based on eigen-projection preprocessing and reconstruction of modified covariance matrix is proposed. The method makes use of eigen-projection to remove the mainlobe interference in the received data. Then modified sidelobe interference plus noise covariance matrix (SINCM) is constructed for further beamforming after the power of the mainlobe interference is replaced by zero. The method eliminates the effect of mainlobe interference to the adaptive weight vector. And the beam pattern forms deep nulls at deviation positions of sidelobe interferences. Simulation results show that the proposed method can effectively suppress the mainlobe interference and sidelobe interferences when the training data contains the signal of interest (SOI), and it has good robustness.

Abstract：Based on the existing two-wheeled intelligent balance car in China, it has almost no autonomous obstacle avoidance and positioning functions. In order to improve its safety and flexibility, laser SLAM technology is applied to the traditional self-balancing car to realize autonomous mapping and path planning, positioning and obstacle avoidance. The Kalman filter is used to fuse the acceleration and tilt angle collected by the six-axis sensor. In terms of mapping and positioning, the Cartographer algorithm released by Google is used, and the Teb algorithm under move base integrated by the Navigation function package is used for path planning and obstacle avoidance. Since the speed of the vehicle is relatively slow and the jitter needs to be avoided as much as possible when mapping the lidar, it is very important to keep the vehicle model in a stable motion state. For this purpose, the data obtained by the sensor is firstly filtered, and then the PID parameters of the car are fine-tuned. At the same time, for more convenient control, the bluetooth function is added to control the movement of the car through bluetooth to achieve rapid map building. After adding SLAM technology, traditional self-balancing scooters can realize obstacle avoidance and positioning functions, detect static and dynamic obstacles in real time, and plan an optimal route around obstacles, realizing the function of unmanned driving.

Abstract： With the rapid development of the national network information industry, the application of domestic computers has become the mainstream trend of the domestic market, but also led to the development of domestic smart NIC(Network Interface Card) intelligent network card, but there is still a big gap between domestic smart NIC and domestic computer adaptation. In view of the problem that domestic smart NIC are rarely applied and studied on domestic computers, this paper carries out adaptation verification of domestic smart NIC based on PKS system, analyzes the performance rules of domestic smart NIC in different scenarios under PKS system, and finds the performance bottleneck of domestic smart NIC at the present stage, so as to promote the optimization of domestic smart NIC. At the same time, it is also convenient for enterprises and users to better understand the current performance status of domestic smart NIC, and make a reasonable choice.

Abstract： SilverLight is a browser plug-in developed for Windows-based systems and applied to web page programs, which currently cannot run on the PKS system. In order to solve this problem, this paper runs SilverLight plug-in in the system environment compatibility layer composed of binary translation tool and Wine, and uses Pipelight as the communication mechanism to combine the system environment compatibility layer with it. This solution helps SilverLight running efficiently on the PKS system.