Abstract：At present, artificial intelligence which stands as a new quality productive force in the IT industry, is a major strategic focus among the major economies for enhancing the national competitiveness and safeguarding the national security. However, the scarcity of computing power has emerged as a critical bottleneck impeding China’s AI development. To cope with the issue of ecological fragmentation in self-developed computing power, this paper puts forward building an intelligent computing operating system based on the general server operating system with AI enhancement and enabled by the intelligent computing platform. By better supporting the development and operation of AI applications, the intelligent computing operating system can meet the computing power demand for development of AI in China. Focusing on the important parts of the intelligent computing platform, the domestic and international development status of heterogeneous resource scheduler and AI programming framework are elaborated in details. The development of the management and scheduling of heterogeneous computing power and the distributed training are also analyzed. Based on the description of the domestic and international AI server market and the reality of heterogeneous computing resource management, the current situation that the capability of chip is weak and the industrial ecology is scattered in the development of AI computing power in China is pointed out. Through systematically sorting out the relevant supporting policies of the development of operating systems in China, the feasibility and necessity of developing the operating system are further confirmed. Subsequently, this paper presents the AI enhancement of the general server operating system, which is one of the two major components of t

Abstract：Based on the 0.18 μm BCD process, a higher-order curvature compensation bandgap reference with low temperature coefficient is proposed. On the basis of the basic Kuijk bandgap reference structure, the VBE linearization method is used to perform high-order curvature compensation, while ensuring that the current flowing through the compensating bipolar transistor is also compensated for higher-order. The two higher-order compensations make the temperature coefficient of the bandgap reference source lower. According to simulation results, the bandgap reference voltage has a maximum voltage difference of 0.376 6 mV over a temperature range of -55°C ~ 125°C, a temperature coefficient of 1.762 ppm/°C, and a power supply rejection ratio (PSRR) of -73.9 dB at low frequencies. This bandgap reference source has been applied in a high-precision low-dropout linear regulator (LDO).

Abstract：In recent years, with rapid development of technology, the functionality of chips has become increasingly complex and their integration levels have continuously risen. System-level design has emerged as a critical aspect of chip development, requiring the integration of numerous subsystems such as CPUs, buses, memory, and various peripherals. These different components must be seamlessly interconnected and work in concert, a task which is both tedious and prone to error at the system's top level. To reduce the inefficiencies and high risks associated with traditional manual management, this paper introduces an automated design tool developed using Python that is applied to system integration. It also explores the significant advantages presented by the tool during the automation of the integration process.

Abstract：WDT counter has become a necessary part in modern MCU system. But with the development of the function of MCU system, the traditional WDT counters that cannot process asynchronous clock and cannot work during low power mode（suspend system clock or peripheral clock）, are not suitable for MCU with various function. The WDT counter that is described in the passage optimized asynchronous clock and signal generated from it, which makes the WDT could be driven by asynchronous or synchronous clock and work during low power mode.

Abstract：A high precision LED current detection circuit is designed in order to meet the functional safety requirements of ISO 26262. Based on the traditional LED constant-current detection circuit, a negative feedback loop is added in the current mirror module to achieve accurate copy of the detection current, avoid the current sampling error in the traditional current detection circuit, and improve the accuracy of the current detection. At the same time, the new circuit adopts the structure of sampling block and power block in series, which reduces the occurrence of current miss detecting and expands the coverage of current detection. The LED current detection circuit is designed by SMIC 0.18 μm CMOS process. The test results show that the detection accuracy can reach ±1% in the range of 5 mA ~ 60 mA LED current.

Abstract：Wind power generation, as an important component of national electricity energy, is developing rapidly in China. The safety of wind power systems is of great significance to the security of national basic energy facilities. The core controller of the wind turbine control system relies on imports, mainly including Germany’s BECKHOFF, Austria’s BACHMANN, etc. The underlying CPU chips are all imported products. The critical core components are severely constrained by others, posing a serious threat to the stable development of the wind power industry and the safe operation of important national power infrastructure. As an independent and secure industrial control core product, CHAOYU Programmable Logic Controller (PLC) builds a new type of power system with new energy as the main body and solves key technical bottleneck problems, which is a key measure to implement the new development concept and the national dual carbon goals. The research and design of a wind turbine control system based on CHAOYU PLC is a key measure to implement national energy facility safety and achieve autonomous security.

Abstract：In the field of nuclear power, any unforeseen situation may lead to serious consequences. Therefore, the development of scientific and effective emergency response management system is an indispensable part of guaranteeing the safety of nuclear power. Model-Based Systems Engineering (MBSE), as an advanced systems engineering method, provides new ideas and methods for the design and implementation of nuclear power emergency response management system with its advantages of systematization, modeling and visualization. The preliminary application of MBSE in nuclear power plant emergency response management system is explored. The key elements of the nuclear power emergency response system are analyzed for its fitness with the MBSE methodology, and the MBSE methodology is utilized to undertake the top-level requirements. The interaction model between the components of the nuclear power emergency response system is constructed for the emergency response system interface problem. The results indicate that MBSE can enhance the convenience of designing emergency response management systems for nuclear power plants and improve the development efficiency.

Abstract：For high-speed data transmission in IEEE 802.15.4, a customized MAC protocol with high-throughput and low-latency characteristic is proposed. It is concluded with in-depth analysis of the data transmission model and throughput of HRP UWB PHY, combined with the characteristics of MAC protocol’s Carrier Sense Multiple Access with Collision Avoidance (CSMA-CA) and Guaranteed Timeslot (GTS) for real-time transmission, then the effective management process of the mechanism is designed. Finally, hardware design and verification of the MAC controller is conducted using Verilog HDL in Vivado. Simulation results demonstrate that compared to the classic IEEE 802.15.4 MAC protocol, the proposed MAC controller improves its throughput by 26% and reduces data transmission latency by 17%.

Abstract：To solve the problem of high false alarm rate in transmission corridor image fire warning, a transmission corridor image mountain fire warning method based on spatiotemporal diffusion is proposed. Firstly, the camera is powered by a pulse width modulation rectifier induction power transformer. Secondly, the collected images are preprocessed to eliminate the impact of image noise on model fire recognition, image features of power transmission corridors are extracted, and a normal sample library is established. On this basis, abnormal samples are screened out through normal sample comparison, and the spatiotemporal diffusion method is combined to analyze the morphological changes of smoke in the time dimension, thereby identifying fires. Finally, in the practical application of a certain power supply company, the accuracy of fire identification using this method is 98.7%. The proposed method can effectively distinguish the characteristics of clouds and fire smoke, achieving accurate identification of forest fires crossing power transmission corridors.

Abstract：Wired sensor networks have high cost, complex wiring, difficult maintenance and expansion, and most wireless sensor networks are difficult to balance between power consumption and transmission distance. To solve the above problems, this paper proposes a remote wireless sensor network based on LoRa. Based on spread spectrum technology, LoRa has strong anti-interference penetration ability and wide network coverage. Combined with the anti-collision mechanism of MAC layer and the composite networking protocol of power adaptive algorithm, it can achieve reliable wireless communication within 1 km with a delay of 0.01 s in the case of multiple terminals.

Abstract：With the rapid development of database technology, the requirements for data processing capabilities and efficiency have rendered many legacy security techniques ineffective, especially in the field of data security. The importance of data security risks in database usage has become extremely critical. A new database security control method is proposed in this paper, which includes three steps: zero trust user context identification, fine-grained data access authentication control, and network-wide traffic-based data access auditing, to identify and intercept risky data access in the database environment. This fine-grained database security method based on zero trust model provides robust support for enhancing database security.

Abstract：As China’s novel power system construction is continuously accelerating, the reform on the electricity sales side continues to deepen, and the competition in the retail market has become increasingly fierce. As a market entity with a large scale of electricity consumption, large-scale interactive users need to formulate reasonable strategies when making power purchase decisions to maximize benefits. This paper proposes an optimization model for power purchase decisions based on the improved gray wolf optimization algorithm. Firstly, reasonable design of power purchase decisions is carried out from the aspects of green electricity consumption rights and responsibilities, temporary electricity demand, controllable electricity plans, and minimization of total electricity bills. Then, based on the combination of different retail packages, an improved grey wolf optimization algorithm is constructed to iteratively solve the optimal strategy for the power purchase decision model. The results of the example show the effectiveness and rationality of this power purchase strategy, which can effectively release the dividends of power system reform.

Abstract：Computer vision has a good engineering application prospect for semantic recognition of small target objects in non-structural environments, but due to the changeable structure, small samples and complex environment of small target parts, the accuracy of traditional methods for small target segmentation recognition is low. To solve this problem, an improved Mask R-CNN model based on attention mechanism is designed in this paper. Based on the original network structure, an A-FNP module is proposed by combining FNP and fusion attention module. Taking the space satellite as the unstructured environment test scene, the satellite sails and explosive bolts are tested. The experimental results show that the accuracy of the proposed method is improved by 5.36%, which is significantly superior to the original model.

Abstract：Facing global computing power competition driven by the new technological revolution and industrial transformation, this paper studies the design of a secure and dependable computing power infrastructure architecture. It integrates cloud computing, secure and dependable chips, operating systems and a comprehensive end-to-end security protection system to build a green, advanced and secure heterogeneous big data foundation. The system deeply integrates high-performance computing, edge computing and artificial intelligence to achieve resource pool and high-performance computing power sharing, providing flexible, automated and high-performance support for computing tasks. In terms of security architecture, it constructs a comprehensive security defense system for the computing power platform through compliance defense based on Classified Protection 2.0, secure and dependable systems and trusted computing systems, emphasizing data security and privacy protection. This infrastructure provides sustainable, secure and dependable high-performance computing support for the country and enterprises in global computing power competition, promoting the rapid development of the digital economy and industrial transformation.

Abstract：Addressing the issues of high cost and complexity inherent in traditional vector frequency converters, a new design based on the TX8M2260 microcontroller has been proposed. This design takes full advantage of the TX8M2260's high performance, low power consumption, and its multiplication division unit (MDU). It employs a highly integrated hardware architecture that is both space-efficient and cost-effective. By implementing control algorithms based on field-oriented control (FOC) theory and optimizing the hardware design, the proposed vector frequency converter has been realized and experimentally validated on the TX8M2260 platform. The experimental results indicate that the designed converter successfully fulfills all required functions and demonstrates excellent dynamic performance and stability, confirming the efficacy and convenience of this vector frequency conversion solution.

Abstract：This paper proposes an SSI communication method deployed on FPGA devices for application in the field of data transmission. The SSI communication controller adopts the principle of high-frequency clock synchronization edge detection to sample the input clock signal, achieving automatic compatibility of communication rate during data reception. By measuring the hardware interface timing parameters at a communication rate of 10 Mb/s, it was verified that the controller IP function is normal and the signal quality is good. Simulation testing was used to analyze that the effective communication bandwidth of the controller IP during read and write operations reaches over 99.9% of the set communication bandwidth, verifying that the controller performance can meet the communication interface requirements.

Abstract：Cryptographic chip is an important carrier for cryptographic algorithms, which implements functions such as encryption, decryption, signature, and authentication of information system. Side channel analysis is an important method to verify the security of cryptographic chips. In the current industry, post-silicon side channel analysis with special equipment is a common method, which is too late and expensive in making any changes to the design to solve the leakage issue. This paper proposes a simulation-based power trace acquisition and side channel analysis method. EDA tools are used to perform functional simulation on the RTL code of the cryptographic chip during the design phase, and we collect the simulated power trace by analyzing the waveform record file. By using Welch t test, KL divergence and correlation energy analysis, leakage can be located in time and space dimensions. Through the side channel analysis experiment on AES-128 RTL design, we proved that the method proposed in this paper can correctly reflect the power leakage, which can detect the side channel leakage risk in the early stage of the cryptographic chip design without the help of special hardware equipment.

Abstract：In the testing and measurement of high-precision data acquisition systems, high-quality bandpass filters are crucial. However, using passive components to achieve low-frequency and high-quality filters requires large inductors, and the size and cost are daunting. This design uses the high-performance fully differential audio amplifier OPA1632D to design a sixth order active 1kHz bandpass filter. The bandpass filter adopts a multi-level feedback design, with a simple principle structure and low design cost. The simulation test results using TINA-TI are good, and it is actually used for front-end filtering testing of high-performance analog-to-digital converter AD7961. The SNR and SFDR dynamic indicators are close to the theoretical values in the manual. Simulation and practice have shown that the design of this filter meets the needs of high-precision data acquisition front-end filtering.

Abstract：RapidIO bus is a high-performance interconnect bus which is widely used in embedded systems. It has the characteristics of high bandwidth, low latency, and multiple processors supported. In response to the problem that most processors on the market currently don’t support the RapidIO bus, a PCIe add-in card based on domestic PCIe to RapidIO controller was designed. The hardware design of each module of the PCIe add-in card was introduced in detail, and a testing environment was built to test the eye diagram and DMA transmission rate of the RapidIO bus. By testing, when the transfer rate of the RapidIO bus is configured as 5 Gb/s, the DMA read and DMA write rates of the RapidIO bus are 1 677 MB/s and 1 711 MB/s, respectively.