Parameter Update Research Articles

A Digital-Twin Framework for Predicting the Remaining Useful Life of Piezoelectric Vibration Sensors with Sensitivity Degradation Modeling.

Piezoelectric vibration sensors (PVSs) are widely applied to vibration detection in aerospace engines due to their small size, high sensitivity, and high-temperature resistance. The precise prediction of their remaining useful life (RUL) under high temperatures is crucial for their maintenance. Notably, digital twins (DTs) provide enormous data from both physical structures and virtual models, which have potential in RUL predictions. Therefore, this work establishes a DT framework containing six modules for sensitivity degradation detection and assessment on the foundation of a five-dimensional DT model. In line with the sensitivity degradation mechanism at high temperatures, a DT-based RUL prediction was performed. Specifically, the PVS sensitivity degradation was described by the Wiener-Arrhenius accelerated degradation model based on the acceleration factor constant principle. Next, an error correction method for the degradation model was proposed using real-time data. Moreover, parameter updates were conducted using a Bayesian method, based on which the RUL was predicted using the first hitting time. Extensive experiments on distinguishing PVS samples demonstrate that our model achieves satisfying performance, which significantly reduces the prediction error to 8 h. A case study was also conducted to provide high RUL prediction accuracy, which further validates the effectiveness of our model in practical use.

Z-number based fuzzy neural network for system identification

In this paper, a novel Z-number based Fuzzy Neural Network (Z-FNN) based on the integration of Z-valued fuzzy logic and neural networks is proposed. Z-valued fuzzy rule base is presented and its inference process is described using interpolative approximate reasoning. Accordingly, the structure of the Z-FNN is proposed using a distance measure and interpolative approximate reasoning scheme. Based on presented architecture the learning algorithm of Z-FNN is designed. The updating of the unknown parameters of the network is carried out using Genetic Algorithms (GA). The proposed Z-FNN system is utilized for dynamic plant identification. The effectiveness of Z-FNN has been tested by comparing its performance with the performances of other fuzzy systems available in the literature. The proposed approach has been proven to be a suitable alternative for the identification of nonlinear systems characterized by uncertain and imprecise information.

Incorporating Effects of Time Accrued on the Waiting List into Lung Transplantation Survival Models.

Rationale: U.S. lung transplant mortality risk models do not account for patients' disease progression as time accrues between mandated clinical parameter updates. Objectives: To investigate the effects of accrued waitlist (WL) time on mortality in lung transplant candidates and recipients beyond those expressed by worsening clinical status and to present a new framework for conceptualizing mortality risk in end-stage lung disease. Methods: Using Scientific Registry of Transplant Recipients data (2015-2020, N = 12,616), we modeled transitions among multiple clinical states over time: WL, posttransplant, and death. Using cause-specific and ordinary Cox regression to estimate trajectories of composite 1-year mortality risk as a function of time from waitlisting to transplantation, we quantified the predictive accuracy of these estimates. We compared multistate model-derived candidate rankings against composite allocation score (CAS) rankings. Measurements and Main Results: There were 11.5% of candidates whose predicted 1-year mortality risk increased by >10% by day 30 on the WL. The multistate model ascribed lower numerical rankings (i.e., higher priority) than CAS for those who died while on the WL (multistate mean; median [interquartile range] ranking at death, 227; 154 [57-334]; CAS median [interquartile range] ranking at death, 329; 162 [11-668]). Patients with interstitial lung disease were more likely to have increasing risk trajectories as a function of time accrued on the WL compared with other lung diagnoses. Conclusions: Incorporating the effects of time accrued on the WL for lung transplant candidates and recipients in donor lung allocation systems may improve the survival of patients with end-stage lung diseases on the individual and population levels.

Distributed parameterized topology-independent noise reduction in acoustic sensor networks

Distributed noise reduction is a challenging task in acoustic sensor networks (ASNs). To address the problem, a distributed parameterized topology-independent noise reduction method is proposed in this paper. To be specific, the parameterized topology-independent distributed adaptive node-specific signal estimation (PTI-DANSE) is first presented in fully connected ASNs, where only one node uses its local signals and the compressed signals summation from the other nodes at a time to estimate signal statistics and update parameters. Then, to make all nodes estimate signal statistics and update parameters simultaneously, the relaxed simultaneous PTI-DANSE (rsPTI-DANSE) is proposed in a synchronous update fashion. Finally, through using in-network summation techniques, the distributed parameterized topology-independent noise reduction algorithm for any topology is obtained. The proposed algorithm is applicative for any network topology, and it is adjustable in terms of speech distortion and noise reduction. Simulation results and real-world experiments indicate that the proposed algorithm has good noise reduction performance.

Memory Clustering Autoencoder Method for Human Action Anomaly Detection on Surveillance Camera Video

Unsupervised deep-learning methods with a deep clustering model are widely used to detect anomaly human actions obtained by surveillance camera due to their powerful image feature learning abilities. These methods aim to optimize a model through a clustering induction target to provide useful cluster assignment, and usually use a forward propagation process: an autoencoder reconstructs an input sequence, a clustering model provides cluster allocation, and a scoring model evaluates distribution and provides scores for each sample. However, these methods have two main problems: one is that an autoencoder is difficult to handle human posture when abnormal and normal actions occur in crowd scenes captured by surveillance camera. The other is that network updating is interrupted by feature extraction and clustering in one epoch. To solve these problems, we design a deep memory storage clustering method based on graph convolution autoencoder to implement the real-time updating of pseudo-labels and network parameters. We also design a new loss function to express similarity between sample feature and centroid feature in a memory storage module and to constrain the parameter update of a network.We evaluate the unsupervised method for three important and representative datasets mainly composed of surveillance videos and use the AUC score as an experimental evaluation indicator.

A lightweight object detection framework for underwater imagery with joint image restoration and color transformation

Traditional underwater detectors struggle to match the demands of visual observation and icon detection because of the underwater optical pictures' color distortion, uneven illumination, and excessive noise levels. In this paper, a lightweight deep-water object detection network is proposed, which consists of three key components, namely image restoration, color transformation, and object detection. First, we propose a lightweight attention module for the problems of color distortion and uneven illumination in underwater images. This module pays more attention to the areas with a greater influence of underwater color when dealing with different types of poor-quality images. Second, to reduce the computational complexity of object detection, we propose a color conversion module based on the laplacian pyramid decomposition. This module incorporates multi-scale features to effectively address the loss of details in the brightest or darkest regions. Finally, we propose a multi-scale dense residual object detection network. The network applies a multi-scale kernel to an improved residual module to realize the extraction and fusion of feature information of multiple receptive fields. Furthermore, to deal with the problem of underwater small-sample object detection, we propose a mutually exclusive loss function. The mutual exclusion loss promotes the update of network parameters so that the classes with different characteristics of samples generated by the network are mutually exclusive. We ran tests on the Pycharm platform to show that the suggested lightweight joint underwater object identification algorithm outperformed the most sophisticated ones. The proposed system was put to the test in an intricate underwater setting, achieving 92.56 percent accuracy and 19.2 frames per second.

Remaining Useful Life Prediction Method of PEM Fuel Cells Based on a Hybrid Model

To predict the remaining useful life (RUL) of the proton exchange membrane fuel cell (PEMFC) in advance, a prediction method based on the voltage recovery model and Bayesian optimization of a multi-kernel relevance vector machine (MK-RVM) is proposed in this paper. First, the empirical mode decomposition (EMD) method was used to preprocess the data, and then MK-RVM was used to train the model. Next, the Bayesian optimization algorithm was used to optimize the weight coefficient of the kernel function to complete the parameter update of the prediction model, and the voltage recovery model was added to the prediction model to realize the rapid and accurate prediction of the RUL of PEMFC. Finally, the method proposed in this paper was applied to the open data set of PEMFC provided by Fuel Cell Laboratory (FCLAB), and the prediction accuracy of RUL for PEMFC was obtained by 95.35%, indicating that this method had good generalization ability and verified the accuracy of the method when predicting the RUL of PEMFC. The realization of long-term projections for PEMFC RUL not only improves the useful life, reliability, and safety of PEMFC but also reduces operating costs and downtime.

Offset-Free Quasi-Infinite Horizon Nonlinear Model Predictive Controller Design Using Parameter Adaptation

Abstract This paper proposes an offset-free Quasi-Infinite Horizon Nonlinear Model Predictive Controller (QIH-NMPC) using online parameter adaptation. In the proposed method, the adaptation law is modeled by a first order differential equation as a function of the tracking error and subsequently combined with a QIH-NMPC algorithm for online updating of the unknown parameter. The effectiveness of the proposed control scheme is demonstrated on a continuous stirred tank reactor (CSTR) and an experimental cascaded three-tank system with uncertain model parameters, structural plant/model mismatch and noisy measurements. For the purpose of comparison, the state-of-the-art online state and parameter estimators such as moving horizon estimation (MHE) and extended Kalman filter (EKF) were also incorporated into QIH-NMPC algorithm. The simulation and experimental results obtained showed the efficacy of the proposed adaptation scheme as it demonstrated a comparable performance to standard estimators (MHE and EKF) although with a lesser computational time.

MCRSpell: A metric learning of correct representation for Chinese spelling correction

Chinese spelling correction (CSC) is a difficult but gratifying work that not only helps individuals read and understand the material in their daily lives but also serves as pre-processing for numerous natural language processing (NLP) applications. With pre-trained language models like BERT, many recent works have had great success. However, these methods have three limitations. Firstly, they rely on the output of the last layer of the correction model for parameter updates, ignoring the guidance of intermediate features. This can prevent the model from being adequately trained and thus lead to overfitting. Secondly, they use a variety of data augmentations, which depend on expert knowledge. Some specific augmentation rules only improve the performance on the target dataset, and for out-of-set data, the performance decreases instead. Thirdly, due to the nature of statistical models, they tend to transform low-frequency expressions into more common high-frequency expressions, although correct expressions should be preserved as much as possible for error correction tasks. In this work, we propose a novel general framework for CSC that takes advantage of metric learning to learn semantic knowledge adaptively from multiple intermediate features. Our approach only feeds the input text and target text of the parallel corpus to the model separately for the spelling error correction task without any data augmentation. Moreover, we designed a replication mechanism to keep those low-frequency correct expressions instead of over-correcting them. Our suggested solution can greatly outperform the SOTA methods in the CSC problem, according to extensive trials on benchmarks. We will release the source code for further use by the community.

Efficient Newton‐multigrid FEM solver for multifield nonlinear coupled problems applied to thixoviscoplastic flows

AbstractThis note is concerned with efficient Newton‐multigrid finite element method (FEM) solver for multifield nonlinear flow problems. In our approach, for efficient FEM solver, we advantageously use the delicate symbiosis aspects of the problem settings for FEM approximations, and the algorithmic tools to obtain the numerical solutions. We concretize our ideas on thixoviscoplastic flow problems. It is a two‐field coupled nonlinear problem. And beside the integrated nonlinearity within momentum and microstructure equations, thixoviscoplastic problems induce a nonlinear two‐way coupling. As far as FEM numerical solutions are concerned, we set the problem in a suitable variational form to use the corresponding wellposedness analysis to develop FEM techniques for the solver. Indeed, the wellposedness study is not an intellectual exercise, rather it is the foundation for the approximate thixoviscoplastic problem as well as for the development of an efficient solver. We base our investigations for the solver on our wellposedness and error analysis results of thixoviscoplastic flow problems published in Proc. Appl. Math. Mech. We continue our series, and proceed to develop a monolithic Newton‐multigrid thixoviscoplastic solver. The solver is based on Newton's method and geometric multigrid techniques to treat the coupling of the problem. So, we use Local Pressure Schur Complement (LPSC) concept to solve the coupled problem on mesh's elements, and proceed with outer blocks Gauss‐Seidel iteration to update the global solutions. Furthermore, we handle the nonlinearity of the problem with the combined adaptive discrete Newton's and multigrid methods. The adaptivity within discrete Newton's method is based on the adaptive step‐length control for the discrete differencing in the Jacobian calculations, while the convergence of linear multigrid solver is made to match the convergence requirement of nonlinear solver, accordingly. And the solver's update parameters are solely dependent on the actual convergence rate of the nonlinear problem. We provide the numerical results of solver performance for thixoviscoplastic lid‐driven cavity flow.

Dynamic production bottleneck prediction using a data-driven method in discrete manufacturing system

In the complex discrete manufacturing system (DMS), the production bottleneck shifts in space as time goes on and constrains operational efficiency. Accurate proactive production bottleneck prediction provides a reliable basis for dynamic production decisions and helps to improve management timeliness and production efficiency. According to the production characteristics of DMS and the relationship between supply and demand, the production bottleneck is given a new quantification. A long and short-term memory network (LSTM) with dual attention mechanism and a dynamic updating method for the source model are proposed to predict production bottlenecks accurately. Firstly, feature and state attention mechanisms are designed to improve the feature extraction and prediction ability of LSTM. Secondly, as the applicability of the prediction model gradually declines over time, sliding time windows and fast Hoeffding concept detection are combined to trigger the update of model parameters. Then a competitive strategy is explored to choose the source model that is the most suitable for the current data distribution in the model library. Model-based transfer learning is adopted to update the source model parameters, making the prediction model highly adaptive. Subsequently, an elimination strategy is set to update the model library to ensure its timeliness. Finally, experiments demonstrate that the proposed method is effective in bottleneck prediction and superior to other methods.

Research on variable time-scale SOC and SOH asynchronous collaborative estimation strategy for electric vehicle power lithium iron phosphate batteries

Battery state parameters such as state-of-health (SOH) have slow time-varying characteristics, while state-of-charge (SOC) parameters have fast time-varying features. Using the same time scale algorithm framework to estimate the above state parameters simultaneously will waste BMS system calculation and cause severe parameter fluctuations due to frequent parameter updates, affecting the estimation accuracy. To address the above issues, this work proposes a variable time-scale SOC and SOH asynchronous collaborative estimation strategy for automotive power lithium iron phosphate batteries. Among them, the real-time changing battery state SOC is estimated through micro time scale filtering algorithms, the slowly changing battery parameter SOH is estimated through macro time scale filtering algorithms, and the time scale parameters that can minimize the cost function when maximizing noise are obtained through the robust optimization strategy of minimum maximization differential evolution. The results indicate that the asynchronous collaborative estimation strategy of variable time scale SOC and SOH proposed in this paper can effectively balance the accuracy of SOC estimation and SOH estimation.

The Elements of the Deterministic Prediction of Physico-Chemical Phenomena

This paper briefly explores the roles of empiricism and determinism in science and engineering. From this analysis, I conclude that the intellectual exercise that we call “science” is best described as the transition from empiricism (i.e., from what we can observe) to determinism, which is the philosophy that the future can be predicted from the past on the basis of the natural laws that are condensations of all previous scientific knowledge. Thus, “science” is enacted by formulating theories to explain the observations and models, based on those theories, are developed to predict new phenomena. Accordingly, models are the computational arms of theories. Importantly, all models must possess a theoretical basis but not all theories need to predict. The structure of a deterministic model is that it must contain an input, a model “engine”, and an output, that are all linked by a feedback loop that permits the continual updating of the model parameters and a means of assessing predictions against new observations. This latter feature, in essence, enables the application of the “scientific method” of cyclical modification/assessment that continues until the model no longer accounts for new observations. At that point, the model (and possibly the theory, as well) must be discarded and a new theory/model developed. Again, importantly, no amount of successful prediction can “prove” a theory/model to be “correct”, because theories and models are merely the figments of our imagination as developed through imperfect senses and imperfect intellect. Accordingly, all theories and models are wrong at some level of detail. Contrariwise, a single failure of a model to predict an observation invalidates the theory/model unequivocally. The principal impediment to model building is complexity and it is the reason why a compromise must always be made between physical reality and mathematical tractability.

Discovering Lin-Kernighan-Helsgaun heuristic for routing optimization using self-supervised reinforcement learning

Vehicle routing optimization is a crucial responsibility of transportation service providers, which can significantly reduce operating expenses and improve client satisfaction. Learning to tackle routing optimization problems automatically can be the next significant step forward in optimization technology. Despite recent advancements in automatically learned heuristics for routing optimization problems, state-of-the-art traditional methods such as Lin-Kernighan-Helsgaun (LKH) still outperform machine learning-based approaches. To narrow this gap, we propose a novel technique called self-supervised reinforcement learning (SSRL), which combines self-supervised learning with the LKH heuristic. We provide a node decoder and an edge decoder corresponding to reinforcement learning and self-supervised learning for learning node penalties and edge scores, respectively. The self-supervised part with cross-entropy loss offers strong gradient signals for parameter updates. At the same time, the reinforcement learning component functions as a regularizer to drive the supervised part, which focuses on particular rewards. SSRL learns and replicates all of the LKH’s significant components, improving the original LKH’s generalization and performance. Through experiments on multiple vehicle routing problems, SSRL has demonstrated superior accuracy and efficiency compared to existing methods. Our results provide empirical evidence of SSRL’s effectiveness and potential as a promising solution for optimizing complex routing problems.

Distributed online multi‐task sparse identification for multiple systems with asynchronous updates

AbstractSimultaneous online identification of the structure and parameters for multiple systems can use the multi‐task learning to share parameter sparse patterns across different identification tasks to improve the estimation accuracy. To reduce the communication and computational burden, improve the convergence rate and increase the reliability of the identification task, a distributed asynchronous multi‐task sparse identification algorithm is proposed. The alternating minimization algorithm is applied to design a distributed multi‐task loss function. Then by constructing the auxiliary matrix and transforming the sparse constraint on the parameter matrix to the constructed auxiliary matrix, the asynchronous parameter update and sparse model update can be realized. The algorithm uses a recursive form to complete online identification which is feasible to analyze. After proving the convergence of the identification algorithm, we present an example to support the theoretical analysis and show the advantages of the proposed algorithm.

On‐line instrumental variable‐based feedforward tuning for non‐resetting motion tasks

SummaryData‐driven feedforward control for tracking of varying and non‐resetting point‐to‐point references requires continuous updating of feedforward parameters instead of task‐by‐task updating. The aim of this paper is to develop an adaptive feedforward controller for non‐resetting point‐to‐point motion tasks by a data‐driven feedforward controller. An approximate optimal instrumental variable (IV) estimator with real‐time bootstrapping is employed in a closed‐loop setting to update the feedforward parameters. A case study on a wafer‐stage and experimental validation on a benchmark motion system show the performance benefit.

A novel adaptive parameter strategy differential evolution algorithm and its application in midcourse guidance maneuver decision-making

In this paper, a novel Adaptive Parameter Strategy Differential Evolution (APSDE) algorithm is proposed to overcome the parameters dependence and avoid local optima. The Parameter Update Mechanism (PUM), which has three different strategies, is used to reduce the dependence on parameters of DE. The Adaptive Proportion Adjustment Mechanism (APAM) is used to balance the proportion of PUM strategies in different development terms of exploitation and exploration, and the Random Restart Mechanism (RRM) is used to improve population diversity when exploitation is in stagnation. The proposed algorithm is verified in the CEC2018 test functions and the results show that APSDE has good abilities of exploitation, exploration, convergence, and stability. Secondly, Midcourse Guidance Maneuver Decision-making (MGMD) in Beyond Visual Range (BVR) air combat is studied and transformed into a single objective variational optimization problem, a MGMD system based on APSDE is established. Finally, the simulation of MGMD is carried out. The APSDE ranks first in the typical MGMD scenario experiment. In the adaptive Midcourse guidance confrontation, the winning rate of APSDE is 54%, and the statistical results show that the APSDE has an excellent MGMD ability.

Prescribed performance tracking control for spacecraft proximity operations with inertia property identification

This paper investigates the integrated position and attitude tracking control with prescribed performance and inertia property identification for spacecraft proximity operations subject to unknown inertia properties, external disturbances, and input saturation. To pursue this, the radical basis function neural network is utilized to approximate the external disturbances, where the involved weight matrix, together with the unknown inertia properties, is considered as parameter uncertainty. Then, a rank condition to guarantee the parameter identification is proposed within the framework of concurrent learning. Subsequently, an integrated position and attitude tracking controller is designed based on dynamic surface method in conjunction with certain performance functions and a saturation compensator, and concurrent learning adaptive laws are developed for parameter update. It is proven that the proposed controller can achieve position and attitude tracking with prescribed performance and, meanwhile, guarantee the uniform ultimate boundedness of all the closed-loop states. Additionally, the inertia properties, including the mass, inertia matrix, and center of mass location, can be identified with high accuracy. Finally, simulation experiments are carried out to evaluate the effectiveness of the proposed control scheme.

Adaptive optimized residual convolutional image annotation model with bionic feature selection model

The advancements in Internet and Web 2.0 technologies resulted in the easier generation and distribution of digital images. Many images are stored without any form of significant labelling. This makes searching for desired images tedious and time-consuming, resulting in an increased demand for an effective image retrieval technique. Image annotation provides one or more labels to an image that describes its content. These annotations greatly help to retrieve images from large databases using different related queries. Automatic Image Annotation (AIA) applications in computer vision, data mining and web search are increasing progressively. This work proposes a hybrid Deep Learning (DL) based optimized image annotation model for AIA. Initially, the pre-processing of an input image is carried out to remove the image noise using the triple Gaussian guided filter. In the second stage, the image's features like shape, colour and texture are extracted using Slantlet transform, YCrCb Colour space and Local Binary Pattern (LBP). The feature dimensionality is reduced by selecting features based on the adaptive red colobuses monkey bionic model. Finally, the adaptive image annotation is obtained using the deep-optimized convolutional residual image annotation model. The loss function reduction and updating of weight parameters are facilitated using the hunger game bionic model. The experiments are performed in Python using Corel-5 K, ESP-Game and MIR FLICKR datasets. The proposed model attains an overall precision of 56, 99.2 and 90%, respectively. Thus, the experimental evaluation results demonstrate the excellent performance of a proposed model over existing approaches.

Multi-Sensor Information Fusion and Multi-Model Fusion-Based Remaining Useful Life Prediction of Fan Slewing Bearings with the Nonlinear Wiener Process

Many factors affect the accuracy of the estimation of the remaining useful life (RUL) of the fan slewing bearings, thereby limiting the sustainable development of the wind power industry. More specifically, the traditional vibration data, which are easily disturbed by noises, cannot comprehensively characterize the health status; thus, the physical model is difficult to establish, and when the data-driven model analyzes the status, it results in unclear physical mechanisms. A new nonlinear Wiener degradation model was established based on the fusion of the physical models and the data-driven models, which was employed to characterize the degradation process of the slewing bearings in this work, and for the local temperature distribution, which has a strong anti-interference ability, the multi-sensor temperature data fusion was selected to analyze the RUL prediction. First, the multi-sensor temperature data were fused by performing a principal component analysis (PCA) to obtain the comprehensive health index (CHI), which represents the fan slewing bearings. Second, the Arrhenius Equation, which characterizes the degradation using temperature, was introduced into the nonlinear Wiener model, and a new degradation model was established. Moreover, considering the random change of the drift coefficients and the individual differences, the closed expression of the probability density function (PDF) of RUL was derived. Third, maximum likelihood estimation (MLE) was used to estimate the parameters. In addition, Bayesian analysis was used to update parameters to achieve real-time estimation. The results demonstrated that the proposed method can be used to significantly improve the fitting degree of the model and the accuracy of RUL estimation.