This paper is concerned with the fault detection problem for a class of networked multi-rate systems nonuniform sampling and dynamic quantization. The interval measurements allowed to be that governed by time-homogenous Markov process partly unknown uncertain transition probabilities. measured output quantized quantizer then transmitted through communication network subject data missing. main purpose under consideration design sampling-interval-dependent filters such that, in simultaneous...

Motivated by the successful application for feedback control, this study extends of reinforcement learning techniques to design two-degree-of-freedom controllers in data-driven environment. Based on residual generator based form Youla parameterisation, all stabilising are first interpreted feedback–feedforward situation with a Kalman filter-based acting as core part. For reference tracking problem, further discussions conducted from regulatory perspective and using Q learning, recursive...

Abstract Remaining useful life (RUL) prediction of bearings is important to guarantee their reliability and formulate a maintenance strategy. Recently, deep graph neural networks (GNNs) have been applied predict the RUL bearings. However, they usuallylack dynamic features, use manual stage identification, experience over-smoothing problem, which will negative effect on accuracy. This paper proposes new framework for bearing based spatial-temporal multi-scale convolutional network (STMSGCN),...

This paper presents fault estimation scheme for one-sided Lipschitz and quasi-one-sided systems with disturbance. One-sided are much more widely used nonlinear than the well-known systems. An combination between adaptive observer sliding mode is developed, which avoids restrictions of these two types observer. The can be estimated here by while utilized to compensate effects example given illustrate effectiveness this construction approach.

Soft sensing technology has been proved to be an effective tool for the online estimation of unmeasured or variables that are difficult directly measure. The performance a soft sensor depends heavily on its convergence speed and generalization ability great extent. Based this idea, we propose new model, Isomap-SVR. First, sample data set is divided into training testing by using self-organizing map (SOM) neural network ensure fairness symmetry segmentation. Isometric feature mapping (Isomap)...

To improve SPECT reconstruction using spatially-correlated magnetic resonance (MR) images as a source of side information, one must account for mismatch between MRI anatomical information and functional information. The authors investigate an approach which incorporates the into by region labels representing regions extracted from MRI. Each pixel corresponds to label. Both mean intensities are jointly estimated penalized Maximum-Likelihood criterion iterative Space-Alternating Generalized EM...

Large lag is one of the main prombems that existed in furnace temperature process control system. The large problem makes system disturbance can't be received timely response, so it causes big overshoot, continuous shock, poor dynamic quality, and may even make unstable. In this paper, to overcome problem, Wincc Matlab as platform, through OPC technology exchange data, combined with powerful calculate ability real-time collect applied double value matrix algorithm (DDMC) based on state space...

Single photon emission computed tomographic images (SPECT) have relatively poor resolution. In an attempt to improve SPECT image quality, many methods been developed for including anatomic information, extracted from higher resolution, structurally correlated magnetic resonance (MRI), into the reconstruction process. These provide improved accuracy if information is perfectly with functional information. However there exist mismatches between MRI anatomical structures and due different...

Sensor prediction models in ESP system are constructed with support vector machines (SVMs) regression algorithm. Thus SVMs used as residual generator via analytical redundancy of the sensors. DAGSVM classification algorithm fulfills sensor fault isolation. The research's result shows application to diagnosis is effective and feasible.

Fault diagnosis and location of distribution network is a difficult problem. With the integration distributed generation, multi-source fault current makes problem more complex. This paper proposes method based on synchronized measurement information matrix, which includes two procedures: 1) to detect occurrence fault; 2) locate section determine type. Firstly, local covariance matrix defined by introducing sliding window synchronous phasor information. Then, constructed taking each PMU's...

Remaining useful life (RUL) prediction is one of the most important technologies to implement health management and predictive maintenance rotating machinery. To predict precisely RUL, a three-stage strategy proposed. Firstly, twenty-four basic characteristics are extracted from vibration signal, which reconstructed by combining those with complete ensemble empirical mode decomposition adaptive noise (BC-CEEMDAN), then trend curves reduce fluctuation. Next, sensitive features selected...

Accurate Story visualization requires several necessary elements, such as identity consistency across frames, the alignment between plain text and visual content, a reasonable layout of objects in images. Most previous works endeavor to meet these requirements by fitting text-to-image (T2I) model on set videos same style with characters, e.g., FlintstonesSV dataset. However, learned T2I models typically struggle adapt new scenes, styles, often lack flexibility revise synthesized This paper...

As one of the most classical supervised learning algorithms, KNN algorithm is not only easy to understand but also can solve classification problems very well. Nevertheless, has a serious drawback:The voting principle used predict category samples be classified too simple and does take into account proximity number contained in each k near-neighbor samples. To this problem, paper proposes novel decision strategy based on probability iterative value improve algorithm. By constantly adjusting...

Despite its excellent performance in path tracking control, the model predictive control (MPC) is limited by computational complexity practical applications. The neural network (NNC) another attractive solution learning historical driving data to approximate optimal law, but a concern that NNC lacks security guarantees when encountering new scenarios it has never been trained on. Inspired prediction process of MPC, deviation sequence (DS-NNC) separates vehicle dynamic from approximation and...

Abstract With the increasing number and types of data center equipment, use centralized monitoring has some disadvantages, such as network bandwidth limit excessive processing pressure server. The equipment grouping according to running status trend can meet dynamic changes with similar is divided into groups for monitoring, which greatly improves sensitivity fault detection. Therefore, this paper designs a multidimensional time series device clustering algorithm based on feature extraction....

The unmanned aerial vehicle system (UAV) with automatic recovery function has been widely used because of its intelligence and automation in which precise landing is essential. This paper developed a series solutions to improve the accuracy through analysis mathematical model aerodynamic characteristics. Firstly, active disturbance rejection controller (ADRC) introduced solve interference by nonlinear changes power system, so as control robustness anti-interference ability. Then, coupling...

Aiming at fault detection rate (FDR) and isolation (FIR), testability demonstration technology is relatively mature. As an extension of technology, PHM (Prognostics Health Management) provides a more sophisticated form to quantify predict equipment health state remaining useful life (RUL). Therefore, many researches attach importance the prognostic validation verification during past two decades. However, there are fewer studies made in test plan universal method test. From point statistical...

Aiming at the problem that current diagnostic methods cannot identify effectively output source of electronic transformer fault, in this paper, wavelet transform is used to extract abnormal signal from terminal under different fault conditions, mutation point and abrupt change moment signals are obtained utilizing multi-scale modulus maxima. Specially, entropy theory introduced quantitatively calculate energy decomposition scales, support vector machine (SVM) method chosen classify forms...

In this paper, a simple yet robust closed-loop identification method is proposed, and an auto-tuner then constructed. Using the fast Fourier transform (FFT), process frequency response first calculated from recorded input output time responses to test. The step constructed using inverse FFT. A general second-order plus dead model thus obtained our newly developed technique on of transfer function responses. Simulation examples real test show its effectiveness.