With the development of flexible HVDC technology, fault diagnosis MMC-HVDC becomes a new research direction. Based on theory, this paper proposes robust method to study problem systems. By optimizing gain matrix in observer, detection with good sensitivity and robustness disturbance is achieved. In system, because inherently uncertain system presence various random disturbances, particularly important. Simulation studies during AC faults have been carried out based 61-level mathematical...

With its capabilities for virtual-real symbiosis, fusion, and interaction, digital twin (DT) technology has become a pivotal tool the transformation of communication equipment testing evaluation (T&E). Unlike typical mechanical equipment, physical activities encompass transmission exchange information energy across physics, electromagnetism, domains. This paper explores unique challenges encountered in T&E introduces concept network driven by technology. Building on this, reference...

State-of-health (SOH) is a health index (HI) that directly reflects the performance degradation of lithium-ion batteries in engineering, but SOH Li-ion difficult to measure directly. In this paper, novel data-driven method proposed estimate accurately and explore relationship-like mechanism. First, features battery should be extracted from data. Next, by using evolution genetic programming reflect change SOH, mathematical model describing relationship between constructed based on...

Residual life estimation is an important problem in reliability engineering. Traditional methods, which are based on time-to-failure distribution, have limitations for components of on-orbit satellites characterized as high with small sample size. Various types information can be collected during test and operation, including historical lifetime data, degradation similar expert information, etc. Therefore, making full use multi-source meaningful improving precision. However, research...

Querying IP geolocation databases directly with a host's address is common and convenient way to determine the geographic location of that host. However, there are usually multitude unsatisfactory "null" replies data inconsistencies in query results. Thus, this paper applies novel approach calibrate databases. We use variety classifiers employ time delays as features automatically reveal relation between distances delays. To begin with, target Internet hosts authentic locations obtained by...

Abstract In the traditional switching policy of a standby system, unit is switched to normal operation state only after active fails. contrast, in satellite engineering, new periodic strategy used by technicians gyroscope system. Due performance degradation long‐term and process, satellites may fail on orbit. For this problem, an optimization model established for two‐unit warm system considering strategy. First, process units described multiphase Wiener process. Second, based first‐hitting...

In the contemporary digital age, an intensified emphasis has been placed on research of book recommendation systems.Historically, these systems predominantly focused readers' past preferences, overlooking inherent characteristics book's content and design.To address this gap, a novel algorithm, leveraging both multimodal image processing deep learning, was designed.Features from cover images were extracted using VGG16 model, while textual attributes discerned through combination Word2Vec...

Multiple Type-I censoring represents that all the test units are terminated at different times. Hence, it is general form of censoring, which common in life testing experiments due to simplicity. Besides, two-parameter exponentiated Weibull (EW) distribution could describe both nonmonotone and monotone failure rates. Obviously, extremely meaningful estimate EW parameters with multiply censored data. The problem studied this paper. First, point estimates presented using a maximum likelihood...

Residual life prediction for complex system is attracting increasing attentions in practice. However, it usually difficult to obtain analytical solution due complexity. In this paper, based on Simpson formula, a numerical method proposed simplify the computation, by which point estimate of system's residual can be derived according structure characteristics system. Four typical reliability subsystems, including series, parallel, voting and cold standby are studied, analytic relationships...

Most previous studies need to learn a complex object model for parsing specific instance. This paper directly learns the general patterns from set of parsed objects and formalizes as series templates instead learning model. Moreover, novel hierarchical structure is presented represent an by using templates, which implicitly contains multi-scale parts their relationships. For single object, process equivalent establishing its representation determining template each node. We combine top-down...

Residual life estimation is significant in reliability engineering. In this paper, a Bayes model proposed to estimate the residual of components by fusing expert knowledge, lifetime and degradation data, which provides new method for characterized as small sample, high long life. The linear Wiener process used data described inverse Gauss distribution. joint maximum likelihood function can be obtained integrating data. With Maximum Entropy Method (MEM), prior distribution determined...

In the course of face feature match, many classifiers have been designed. The neural network is usually selected as a classifier because its validity and universality, whereas training time, epochs, convergence, all are not satisfied to us. It often influenced by author’s experience. case, collaborative genetic algorithm presented new recognition classifier. one thing train NN weights GA until stopping criterion met, next use BP continue network. time epochs improved in experiment on ORL...

Residual life estimation is increasingly concerned in reliability engineering. For conventional methods, problem arises when products are characterized as high and small sample size. However, practical engineering, experts can provide valuable information, which apparently improve the precision of residual make up above defects. In this paper, a method proposed by fusing expert information based on Bayesian theory. Usually, could take different forms, including point confidence interval for...

Abstract For large complex systems, it is of great importance to study effective approaches for reliability evaluation. In this paper, a system evaluation method based on Bayesian theory and multi-source information fusion proposed. Firstly, model the network, including establishment block diagram determination components’ life distributions. Then, posterior distributions moments component are determined. With moments, failure components can be converted prior moment according structure,...

摘要: 产品的退化数据和寿命数据是两类重要的可靠性信息,通过融合利用两类数据可显著提高产品可靠性评估结果的精度。现有融合方法常利用Bayes理论,但计算过程繁琐复杂,需要通过提高计算量来保证结果精度,且两类样本数据量的不均衡性会导致出现"数据淹没"的问题。考虑这一问题,提出了基于退化和寿命数据分步融合的可靠性评估方法,首先利用产品的退化数据,建立随机过程退化模型,对参数进行初始估计,随后结合寿命数据对寿命样本的失效概率进行更新,分两步将两类信息进行折合计算,从而达到数据融合的目的。最后通过寿命分布曲线拟合计算退化模型参数,并给出产品的可靠性评估结果。仿真试验和算例分析表明,与传统Bayes融合方法相比,在新的数据融合思路下,该方法能够简化可靠性评估工作实施步骤,同时提高评估精度和运算效率,还有效避免了"数据淹没"问题。

Nowadays, as the system has been becoming more and complex in engineering practice, residual useful life (RUL) prediction of is an important research content to project management. In this paper, order cope with fact that components' characteristics are diverse system, a method considering component follows different exponential distribution proposed predict RUL system. This describes analytical relationship between reliability function. And according structures, expressions series parallel...

Vehicle target recognition technology is an important in the auxiliary safe driving system, which greatly improves safety assist driving. This paper proposes a method to identify vehicle with lidar point cloud data and machine learning; first establishes ROI (region of interest), uses voxel grid filter downsample this area reduce amount processed data, then use RANSAC (random sampling consensus) remove ground points that are useless for process, Euclidean clustering clustering. A rough...