Deterministic and probabilistic ship pitch prediction using a multi-predictor integration model based on hybrid data preprocessing,reinforcement learning and improved QRNN

The deterministic and probabilistic prediction of ship motion is important for safe navigation and stable real-time operational control of ships at sea. However, the volatility and randomness of ship motion, the non-adaptive nature of single predictors and the poor coverage of quantile regression pose serious challenges to uncertainty prediction, making research in this field limited. In this paper, a multi-predictor integration model based on hybrid data preprocessing, reinforcement learning and improved quantile regression neural network (QRNN) is proposed to explore the deterministic and probabilistic prediction of ship pitch motion. To validate the performance of the proposed multi-predictor integrated prediction model, an experimental study is conducted with three sets of actual ship longitudinal motions during sea trials in the South China Sea. The experimental results indicate that the root mean square errors (RMSEs) of the proposed model of deterministic prediction are 0.0254°, 0.0359°, and 0.0188°, respectively. Taking series #2 as an example, the prediction interval coverage probabilities (PICPs) of the proposed model of probability predictions at 90%, 95%, and 99% confidence levels (CLs) are 0.9400, 0.9800, and 1.0000, respectively. This study signifies that the proposed model can provide trusted deterministic predictions and can effectively quantify the uncertainty of ship pitch motion, which has the potential to provide practical support for ship early warning systems.     

On the numerical solution of Fredholm integral equations utilizing the local radial basis function method

The current investigation describes a computational technique to solve one- and two-dimensional Fredholm integral equations of the second kind. The method estimates the solution using the discrete collocation method by combining locally supported radial basis functions (RBFs) constructed on a small set of nodes instead of all points over the analysed domain. In this work, we employ the Gauss–Legendre integration rule on the influence domains of shape functions to approximate the local integrals appearing in the method. In comparison with the globally supported RBFs for solving integral equations, the proposed method is stable and uses much less computer memory. The scheme does not require any cell structures, so it is meshless. We also obtain the error analysis of the proposed method and demonstrate that the convergence rate of the approach is high. Illustrative examples clearly show the reliability and efficiency of the new method.     

基于VMD瞬时能量与GA- RBF的滚动轴承故障诊断

针对滚动轴承早期故障的有效识别,提出了一种基于VMD瞬时能量与GA优化的RBF神经网络的滚动轴承故障诊断方法,可以有效对滚动故障做出诊断。首先,VMD将滚动轴承振动信号进行分解成合适数目的本证模态函数;其次,计算本证模态函数分量的瞬时能量并组成特征向量;最后,将特征向量输入到GA优化的RBF神经网络实现轴承故障识别。通过滚动轴承故障诊断实验对该方法进行验证。结果表明,该方法识别滚动轴承故障的准确率为96.43%,较默认参数的RBF神经网络和EEMD瞬时能量与GA-RBF神经网络有明显的提高,证明了所提方法的可行性。     

Prediction of process parameters of water-assisted injection molding based on inverse radial basis function neural network

Because of the introduction of new processing parameters in water-assisted injection molding (WAIM), processes control has become more difficult. First, design of experiment (DOE) was carried out by using optimized Latin hypercubes (Opt LHS). On the basis of this, computational fluid dynamics (CFD) method was used to simulate and calculate hollowed core ratios and wall thickness differences of cooling water pipe at different positions. Then inverse radial basis function (RBF) neural network model reflecting the fitting relationship between processing parameters and molding quality was established, and accuracy of the model was detected by cross validation. Finally, expected molding quality was applied to predict processing parameters, and the obtained molding quality under the predicted processing parameters was verified by computer aided engineering (CAE) simulation and experimental methods. The results showed that mean relative precisions of processing parameters such as melt temperature, delay time, short shot size, water pressure, and mold temperature for inverse RBF model were 98.6%, 93.6%, 98.5%, 93.9%, and 97.9%, respectively, which met the accuracy requirements. Furthermore, compared with expected values of hollowed core ratios and wall thickness differences, the average errors of CAE and experiment were 2.3% and 4.9%, respectively.     

Stabilisation of Schrödinger equation in dynamic boundary feedback with a memory-typed heat equation

In this work, we study the dynamic behaviour for a heat equation with exponential polynomial kernel memory to be a controller for a Schrödinger system. By introducing some new variables, the time-variant system is transformed into a time-invariant one. Remarkably, the resolvent of the closed-loop system operator is not compact anymore. The residual spectrum is shown to be empty and the continuous spectrum consisting of finite isolated points are obtained. It is shown that the sequence of generalised eigenfunctions forms a Riesz basis for the Hilbert state space. This deduces the spectrum-determined growth condition for the C ₀-semigroup, and the exponential stability is then established.     

Comparison of approaches to the numerical modelling of pure metals solidification using the control volume method

The thermal processes proceeding in the casting-mould system are analysed. Solidification process can be described in the different ways. One of them consists in the application of the Fourier–Kirchhoff equation in which the source function controlling the phase change is introduced. In this paper, two approaches to the modelling of pure metals solidification, i.e. the solidification proceeding at the constant temperature and in the artificially introduced interval of temperature are considered. Numerical model used at the stage of computations is constructed using the Control Volume Method in which the principle of conservation of energy (taking also into account the release of latent heat of solidification) is applied to each control volume ensuring the correctness of the model. The use of Voronoi/Thiessen tessellation to the discretization of the casting-mould system can accurately reproduce the shape of each sub-domain. In the final part of the paper, the examples of computations are shown.     

Prediction and sensitivity analysis of CNTs-modified asphalt’s adhesion force using a radial basis neural network model

Abstract

The expected longer service life of modified asphalt can be jeopardized by different environmental factors, such as moisture, oxidation, etc. which affect the desired properties by altering the adhesive property. An insight into knowledge of the adhesive property of the asphalt can help in providing more durable asphalt pavement. The study attempted to develop different models of adhesive properties of polymers and carbon nanotubes (CNTs) modified asphalt binders. The polymer-CNT modified asphalt is processed to prepare different types of samples, by simulating the damage due to moisture and oxidization, following the corresponding standard method. An Atomic Force Microscopy (AFM) was employed to assess the nanoscale adhesion force of the tested samples following the existing functional group in asphalt. Finally, the study has developed Radial Basis Function Neural Network (RBFNN) as a function of different parameters including; asphalt chemistry (i.e. AFM tip type and constant), type and percentages of polymers and CNTs and different environmental exposures (oxidation, moisture, etc.) to predict the nano adhesion force of asphalt. It is observed that the adhesive property of the Styrene–Butadiene modified asphalt is more consistent compared to the Styrene–Butadiene–Styrene modified asphalt, while the presence of Single-Wall Nanotubes (SWNT) is observed to affect the adhesive properties of asphalt significantly as compared to Multi-Wall Nanotubes (MWNT). The higher accuracy level of RBFNN model also indicates that the functional group (tip-type) adding with the percentages and types of polymers and CNTs significantly affect the adhesive properties of asphalt.     

基扩展算法在信道参数估计中的应用

针对如何减少信道参数估计所需要的已知数据,文中在基于扩展算法的基础上,采用复指数基函数,通过获取部分信道响应数据来计算得到其他信道数据,从而减少信道的实际测量次数。重点研究了基扩展算法在不同基函数阶数和不同传输符号块采样点数的情况下,模型准确度的变化,给出了要使模型具有佳性能的参数设定方案。     

Small oscillation fault detection for a class of nonlinear systems with output measurements using deterministic learning

Early detection of small faults is an important issue in the literature of fault diagnosis. In this paper, for a class of nonlinear systems with output measurements, an approach for rapid detection of small oscillation faults is presented. Firstly, locally accurate approximations of unknown system dynamics and fault functions are achieved by combining a high gain observer and a deterministic learning (DL) theory. The obtained knowledge of system dynamics for both normal and fault modes is stored in constant RBF networks. Secondly, a bank of dynamical estimators are constructed for all the normal mode and oscillation faults. The knowledge obtained through DL is reused with a nonhigh-gain design. The occurrence of a fault can be detected if one of residual norms of a fault estimator becomes smaller than that of the normal estimator in a finite time. A rigorous analysis of the detectability properties of the proposed fault detection scheme is also given, which includes the fault detectability condition and the fault detection time. The attractions of the paper lie in that with output measurements, the knowledge of modeling uncertainty and nonlinear faults is obtained and then is utilized to enhance the sensitivity to small faults.     

Classification of the cardiotocogram data for anticipation of fetal risks using machine learning techniques

The aim of the research is evaluating the classification performances of eight different machine-learning methods on the antepartum cardiotocography (CTG) data. The classification is necessary to predict newborn health, especially for the critical cases. Cardiotocography is used for assisting the obstetricians’ to obtain detailed information during the pregnancy as a technique of measuring fetal well-being, essentially in pregnant women having potential complications. The obstetricians describe CTG shortly as a continuous electronic record of the baby's heart rate took from the mother's abdomen. The acquired information is necessary to visualize unhealthiness of the embryo and gives an opportunity for early intervention prior to happening a permanent impairment to the embryo. The aim of the machine learning methods is by using attributes of data obtained from the uterine contraction (UC) and fetal heart rate (FHR) signals to classify as pathological or normal. The dataset contains 1831 instances with 21 attributes, examined by applying the methods. In the paper, the highest accuracy displayed as 99.2%.