Segmenting the vertebrae is an important part of diagnosing spine diseases and planning surgery. The vertebra's complex anatomy, featuring similar shapes of neighboring bones and vague boundaries, presents substantial challenges for precise semantic segmentation. Traditional methods often depend heavily on local features, making it difficult to capture global context and resulting in less accurate recognition. Consequently, automated and efficient segmentation approaches are crucial to overcoming these issues. This study introduces an innovative Swin-BFB-UNet architecture derived from UNet to enhance vertebra segmentation. We include the Swin Transformer into the encoder to capture both global and local information, hence enhancing vertebra segmentation accuracy. The bottleneck fusion block (BFB) fuses multi-scale features and semantic information, enhancing the model's feature representation. Furthermore, we employ both Dice Loss and Cross-Entropy Loss to alleviate the adverse effects of foreground-background area imbalance. We conducted annotations and experiments on the lateral X-ray images in the BUU-LSPINE dataset. The experimental findings indicate that our model exceeds other leading methodologies, attaining a mDice of 90.3%, hence evidencing its proficiency in precise vertebra segmentation.

针对四旋翼无人机控制超调量大和响应速度慢的问题,在基于自耦PID的四旋翼无人机姿态控制的基础上对自耦PID控制算法进行了优化.首先建立四旋翼无人机的非线性数学模型,进而得出横滚角、俯仰角、偏航角的传递函数;然后结合自耦PID的控制律和飞行控制律,分析原有速度因子的自适应模型;创新点在于将速度因子中的参数与模糊控制相结合来减少系统响应时间和超调量.数字仿真实验结果表明:优化后的自耦PID控制算法,具有更加良好的可控性能.

In-band full-duplex is an efficient technology to alleviate the shortage of spectrum resources in wireless communication system. To ensure the information security of the full-duplex communication system, where a Full Duplex Access Point (FD-AP) receives information from the uplink users and transmits information to the downlink users simultaneously, an Intelligent Reflecting Surface (IRS) assisted physical layer security scheme is proposed in this paper. A multi-variable coupling non-convex optimization problem is formulated to maximize the secrecy rate of downlink users, subject to constraints of the maximum transmit power, the minimum Signal-to-Interference and Noise Ratio(SINR) required at the AP and unit modulus of IRS phase shift. To solve the multi-variable coupling optimization problem, the Alternating Optimization (AO) algorithm is adopted to optimize the AP transmit beamforming and IRS reflection phase shift iteratively, and a Riemannian manifold optimization based on exact penalty method is proposed to deal with the unit modulus constraint and transform the phase shift optimization sup-problem into an unconstrained minimization problem on Riemannian manifold. Simulation results show that the proposed scheme can significantly improve the security performance of full-duplex communication system. In addition, compared with the positive SemiDefinite Relaxation (SDR) algorithm, the proposed algorithm has much lower computational complexity.

针对跳点搜索(JPS,jump point search)算法在障碍物位置随机的栅格地图中路径规划时间较长的问题,提出了并行-交替式双向跳点搜索(PA-BJPS,parallel alternate bidirectional jump point search)算法;首先,在起始点与目标点间确定一个中心热点区域;其次,采用改进了预计代价函数的并行式双向跳点搜索算法,分别规划从起始点抵达中心热点区域以及目标点抵达中心热点区域的路径;然后,采用交替式双向跳点搜索算法,规划中心热点区域内部的路径;最后,提出迭代式路径修正方法来改良危险路径,并采用3次B-样条曲线替代拐角来平滑路径;仿真结果表明,并行-交替式双向跳点搜索算法有效地缩短了路径规划时间,同时提高了路径的安全性和平滑性.

线结构光三维视觉测量技术最关键的一步是提取出结构光图像中的激光条纹中心线;针对动态测量环境下激光条纹图像存在复杂背景信息、激光光强分布不均、光带各部分宽度差别大、激光条纹断裂等问题,文章研究了一种适用于动态测量环境的激光条纹中心线提取方法;首先通过图像预处理以及自适应裁剪算法提取出感兴趣区域(ROI,region of interest);其次通过改进型伽马校正(IGC,improved gamma correction)以及改进型变阈值大津阈值分割算法(IVT Ostu,improved variable threshold)分割出激光条纹区域;然后使用二维灰度重心法(TD-GBM,two-dimensional gray baryeentric method)提取激光条纹的初始中心线;最终使用二次优化算法对初始中心线进行优化,精确地提取出激光条纹中心线;实验结果表明,相比于灰度重心法、Steger法等算法,文章所提方法受背景干扰以及激光条纹质量的影响较小,能够在多种复杂情况下更精确地提取激光条纹中心线,满足准确性高、稳定性强以及实时性好的要求.

针对不规则堆料的体积难以直接测量的问题,提出一种基于双目视觉的不规则堆料体积测量方法;首先,使用双目相机采集堆料的俯视图像,对图像进行预处理;接着通过立体匹配求出堆料图像的视差图,并采用中值滤波和最邻近插值算法对于视差图中的空洞进行填充,同时计算出视差图对应的三维点云坐标;再采用改进的OTSU阈值分割算法将视差图中的堆料区域提取出来,计算出堆料区域的平均高度和面积,最后计算出堆料的体积;实验结果表明,体积测量误差小于5％,可以快速有效地测量出堆料的体积,能够满足工业检测领域的需求,同时为准确快速地测量堆料体积提供了一种新的思路.

随着工业化程度的提高,设备的故障预测的重要性日趋提高;提出了一种基于循环神经网络(RNN)的故障预测算法,通过数据训练,充分发掘了RNN对时间序列数据的拟合能力;RNN故障预测模型由数据处理模块和神经网络识别模块组成;在数据处理模块中,采用数学函数分配的方法建立了RNN模型的训练样本和测试样本;在神经网络识别模块中,针对当前故障预测技术中异常点难以确定的问题,应用了一种逐步逼近的神经网络训练方法;最后利用气体绝缘开关(GIS)故障数据对该算法进行了验证,结果表明,该方法可以在故障发生前检测到故障发生趋势,进而实现故障预测,并且能在逐步训练中确定异常点的位置.

With the increase of the operating time of sulphur hexafluoride (SF6) electrical equipment, the different degrees of discharge may occur inside the equipment. It makes the insulation performance of the equipment decline and will cause serious damage to the equipment. Therefore, it is of practical significance to diagnose fault and assess state for SF6 electrical equipment. In recent years, the frequency of monitoring data acquisition for SF6 electrical equipment has been continuously improved and the scope of collection has been continuously expanded, which makes massive data accumulated in the substation database. In order to quickly process massive SF6 electrical equipment condition monitoring data, we built a two-level fault diagnosis model for SF6 electrical equipment on the Hadoop platform. And we use the MapReduce framework to achieve the parallelization of the fault diagnosis algorithm, which further improves the speed of fault diagnosis for SF6 electrical equipment.

A vehicle type image recognition algorithm based on improved capsule network (CapsNet) is proposed. The algorithm mainly includes three steps. Firstly, the coefficients used to weight each vote between capsules of different layers in CapsNet are iteratively updated by using the Expectation-Maximization algorithm (EM). Secondly the spread loss is optimized by cross-entropy to make the training less sensitive to the initialization and hyper-parameters of the model, then avoid the occurrence of gradient dispersion resulting in a slow learning rate. Finally, the improved CapsNet is trained and tested by using Comprehensive Cars (CompCars) dataset. Experimental results show that the proposed method can effectively improve the recognition rate by 5.08%.

In this paper, an unsupervised learning algorithm of spiking neural network (SNN) by using biologically plausible mechanisms is presented for achieving handwritten digital recognition. Firstly, the spike-timing-dependent plasticity ( STDP) model was established based on pre- and postsynaptic trace learning rule to determine the connection between neurons. Secondly, genetic algorithm (GA) was used to optimize the initial presynaptic weights and axon delay in the neural network. Finally, the Mixed National Institute of Standards and Technology (MNIST) dataset was trained and tested. Experimental results show that the proposed method can effectively improve the recognition rate of handwritten numbers and realize unsupervised learning of handwritten digital recognition. The accuracy in the MNIST benchmark test is increased by using this improved unsupervised SNN learning scheme. The computational complexity is greatly reduced; therefore, the calculation speed is increased. Simulation result is better than the implementation of the previous unsupervised SNNs.

Today, the world comes into a booming information age. With the rapid development of computer and sensors, various bits of data are continuously being generated as time goes by. The amount of data to be processed has entered into the big data category. Data mining is widely used to discover hidden information in the large amounts of data. However, data mining applied to medical databases is a challenging process. The unavailability of large raw data and data complexity are some of the difficulties encountered. This research work proposes a way of dealing with big medical data with a case study of gout disease. Gout is common chronic disease caused by the most important risk factor hyperuricemia. There is no drug to completely cure the gout, the patient is suffering a lot of pain. It is important to control the occurrence of gout and to study the association of gout with other metabolic diseases. This paper discusses methods for the analysis of this complex dataset of this disease, to help get more understanding of the disease and associated diseases. Association Rule is used as the mining algorithm for the data processing. An associated relation is proposed according to the experiments, which applies auxiliary support for doctor's clinical diagnosis and disease research in the local.

Heavy rainfall and aging urban drainage system make the low-lying road waterlogged seriously and have an important effect on the safety and property of citizens,roadbed and urban traffic.In order to solve this problem,the wireless sensor / actuator network (WSAN) technology is used to design a monitoring system.The system is divided into three layers:sensor / actuator layer,network layer and application layer.Ultrasonic sensors are deployed to collect the information of water level in sensor nodes of sensor / actuator layer,actuator pump is used to drain away the water in actuator nodes of sensor / actuator layer.The ZigBee network and GPRS network are configured for two-way data transmission between sensor / actuator layer and application layer in the gateway and sink node of network layer.The server of application layer stores the information of road waterlogging,calls Baidu map API to display the real-time information on the web,and sends commands to achieve closed-control by controlling the actuator pump in the road waterlogging area action through the network layer.The system not only shares the real-time information of the road waterlogging simply and effectively,but also realizes the timely drainage.So it has practical application value.

In this paper, a high voltage switchgear fault diagnosis system is designed with cloud computing technology, which is used for distributed storage and distributed processing of large-scale switching device status information data. This paper takes the high voltage switch cabinet as an example, to design a high voltage switchgear fault diagnosis system by cloud computing technology. In this paper, the three sets of characteristic parameters of the switchgear are analyzed, and the working status of the high voltage switchgear is judged. To speed up the fault diagnosis, the Hadoop cloud platform is built. The Hadoop distributed configuration is completed. A probabilistic neural network algorithm based on MapReduce is designed, and parallel processing of diagnostic data is realized in the cluster environment. The operating speed of fault diagnosis system in a stand-alone environment and the cluster environment are compared. The simulation results show that system running time can be reduced by using Hadoop clusters in the case of large data size. As well, the obvious advantage of cloud computing technology in processing large-scale fault diagnosis data is revealed.

Prefabricated substation is composed of four modules,which are insulated switchgear,transformers,low voltage switch cabinet and secondary equipment,making the whole substation become a product,so it's possible to assess the overall status of it.In this paper,an evaluation method based on information fusion is proposed.The multilayer fuzzy comprehensive evaluation method,BP neural network and grey correlation method are used to assess the four modules to get the evaluation results.Then,the DS combination rule is used to fuse the evaluation results.Finally,the fuzzy comprehensive evaluation method is used again to assess the status of the prefabricated substation based on the four modules evaluation results.Simulation results show that the proposed method can correctly evaluate the operating status of the preinstalled substation.

In the wireless sensor actuator network (WSAN), in order to make the sensor nodes (S) and the actuator nodes (A) work together more efficiently and obtain more accurate assignment information of actuators, a novel data fusion model of actuators assignment were constructed. In this paper, the weights and thresholds of BP neural network (BPNN) were optimized by genetic algorithm (GA), and the GA-BPNN model was applied to prefabricated substation. In order to verify the characteristics of the model, the simulation and analysis of GA-BPNN were carried out comparing with the traditional BPNN data fusion model. The results demonstrate that the running time of the whole system can greatly be reduced, and the efficiency of operation and the correctness of the actuator nodes task assignment information can also be improved by using GA-BPNN task assignment data fusion model.

In order to deal with the large amount of monitoring data of SF 6 electrical equipment, the two-level fault diagnosis model of SF 6 electrical equipment was constructed. Based on SF6 gas decomposition products, the fault diagnosis was made by using the decision tree and the improved neural network algorithm. The decision tree model is used to deal with the original data, so that the equipment can be determined whether it's faulty preliminarily. It can improve the processing speed of the mass data. Then, the fault data is put into the improved neural network model to judge the fault type accurately. The improved neural network model not only can recognize the existing fault, but also can identify the new fault types. Firstly, the reliability of the results is calculated. The fault data those reliability is less than the threshold is selected. Secondly, the selected data is evaluated by experts. If the data is defined as the new fault type, then the network structure and weights will be updated.

Today, data mining is widely used to discover hidden information in the large amounts of data. However, data mining applied to medical databases is a challenging process. The unavailability of data in the source and data complexity are some of the difficulties in medical data research. This research work proposes a way of dealing with big medical data with a case study of gout patients disease. Gout is common chronic disease caused by the most important risk factor hyperuricemia. There is no drug to completely cure the gout, the patient is suffering a lot of pain. It is important to control the occurrence of gout and to study the association of gout with other metabolic diseases This paper discusses methods for the analysis of this complex dataset of this disease, to help get more understanding of the disease and associated diseases. Association Rule is used as the mining algorithm for the data processing. An associated relation is proposed according to the experiments, which applies auxiliary support for doctor's clinical diagnosis and disease research.

As one of the AC and DC switching devices in power system, high voltage switchgear is mainly used for the control and protection of power systems. In order to meet the demand of large amount of data, many types, fast processing speed and high quality of fault diagnosis in large data environment, a parallel processing framework based on a data fusion fault diagnosis algorithm designed by D-S Evidence Theory is introduced, taking SF6 high voltage circuit breaker as an example in this paper. SF6 high voltage circuit breaker trip(closing) coil current, voltage and current time are selected as input of the diagnosis system, and six main fault types are selected as output of the diagnosis system in this paper. Considering the multi-level, multi-layer and multi-faceted advantages of multi-sensor data fusion, D-S evidence theory based on MapReduce framework is designed. The simulation shows that the requirements of mass rapid diagnosis of high voltage switch equipment can be satisfied. Compared with the traditional serial processing method, processing time can be reduced by 95 percent under situation of hundreds of megabytes data.

Reliability is an important research issue in wireless sensor networks(WSNs). Many literatures about how to improve network reliability in WSNs have been proposed, however the research on establishing evaluation models of reliability for WSNs is yet insufficient. In this paper, a model to evaluate the reliability of wireless sensor networks (WSNs) is proposed. This network state based model can effectively evaluate the probability of successful transmission of data packets in WSNs with the consideration of the work states of both the sensor node's different modules and the quality of the communication link. We evaluated the reliability of WSNs in single task and multi-task mode. At the same time, how different network topologies, such as flat and clustering structures, affect the reliability of WSNs is also discussed respectively. Furthermore, some typical reliable strategies, such as retransmission, multi-path transmission and multiple sink nodes, are analyzed and evaluated. The simulation attempts to evaluate the reliability of WSNs quantitatively and the results have shown the impacts of network state on reliability and the effectiveness in different reliable mechanisms.

In order to solve the tasks assignment of actuators in wireless sensor and actuator network (WSAN), an improved distributed auction algorithm (IDAA) is proposed.Utility of each task and cost of each actuator are taken into account to obtain the optimal assignment.The construction method of the response tree is improved, and the matching degree is introduced in the calculation of utility.Simulation results show that the energy consumption has been more balanced, and the numbers of data packets and time of tasks assignment have been reduced.