Bayesian Information Fusion Research Articles

Reliability analysis and evaluation model for space phased‐mission system

AbstractA reliability analysis and evaluation model based on event chain and Bayesian information fusion is proposed for the characteristics of complex space phased‐mission systems (SPMSs), the implementation process of integrated reliability modeling is detailed, and the identification of weak links and reliability evaluation is carried out with the entry, descent, and landing process of Tianwen‐1 Mars Probe as an example. The application results show that the reliability analysis and evaluation model based on event chain and Bayesian information fusion has good engineering applicability, and the method can be extended and applied to the reliability modeling, analysis, and evaluation of complex SPMSs.

A probabilistic framework for product health monitoring in multistage manufacturing using Unsupervised Artificial Neural Networks and Gaussian Processes

The emergence of highly instrumented manufacturing systems has enabled the paradigm of smart manufacturing that provides high levels of prognostics functionality. Of particular interest is to precisely determine geometric conformance or non-conformance of workpieces during manufacturing. This paper presents a new dimensional product health monitoring system that learns from in-process sensor data and updates the prediction of the product quality as the product is manufactured. The system uses data from multiple manufacturing stages, unlike from a single stage at a time, to predict the dimensional quality of the finished product that is updated with subsequent measurements such as On-Machine Measurements (OMMs), in on-line incremental learning fashion. It is based on self-supervised neural networks for dimensionality reduction, Gaussian Process Regression (GPR) models for probabilistic prediction about the end product condition and the associated uncertainty, and Bayesian information fusion for updating the conditional probability distribution of the end product quality in the light of new information. The monitoring approach was tested on the prediction of diameter deviations with validation results showing its ability to achieve an average accuracy better than 5 μm in terms of the Root Mean Squared Error (RMSE). Having obtained a Probability Density Function (PDF) for the measurand of interest, the conformance and non-conformance probabilities given the tolerance specifications are computed to support the principle of inspection by exception. This ability to construct a conformance probability-based product quality monitoring system using probabilistic machine learning methods constitute a step change to manufacturing prognostics.

Infrasound Source Localization of Distributed Stations Using Sparse Bayesian Learning and Bayesian Information Fusion

The precise localization of the infrasound source is important for infrasound event monitoring. The localization of infrasound sources is influenced by the atmospheric propagation environment and infrasound measurement equipment in the large-scale global distribution of infrasound arrays. A distributed infrasound source localization method based on sparse Bayesian learning (SBL) and Bayesian information fusion is proposed to reduce the localization error. First, the arrival azimuth of the infrasound source is obtained based on the SBL algorithm. Then, the infrasound source localization result is obtained by the Bayesian information fusion algorithm. The localization error of the infrasound source can be reduced by this infrasound source method, which incorporates the uncertainty of the infrasound propagation environment and infrasound measurement equipment into the infrasound source localization results. The effectiveness of the proposed algorithm was validated using rocket motor explosion data from the Utah Test and Training Range (UTTR). The experimental results show that the arrival azimuth estimation error can be within 2° and the localization distance error is 3.5 km.

Supervised sparse preserving projections model based on distributed principal component analysis for chemical process monitoring

AbstractFor the monitoring of large‐scale chemical processes, the distributed method is often used to extract local feature information and model the extracted local feature information to obtain a process monitoring model. But the distributed process monitoring model often contains more process variables, which makes the local information of the process data flooded. To make up for the insufficient extraction of local information in traditional distributed process monitoring, supervised sparse preserving projections model based on distributed principal component analysis (DPCA‐SSPP) is proposed in this paper. First, the process data are decomposed by the PCA algorithm, and the principal component space and residual space are obtained. Second, the variables of each sub‐block are selected according to the maximum correlation criterion, and the SSPP process monitoring model is established for each sub‐block. Finally, the monitoring results of each sub‐block are combined together to form a global monitoring result through the Bayesian information fusion strategy. The proposed scheme can be proved to be effective through the simulation on a nonlinear numerical example and the Tennessee Eastman benchmark (TE) process.

A Bayesian information fusion approach for end product quality estimation using machine learning and on-machine probing

There is an increasing demand for manufacturing processes to improve product quality and production rates while minimising the costs. The quality of the products is influenced by several sources of errors introduced during the series of manufacturing operations. These errors accumulate over these multiple stages of manufacturing. Therefore, monitoring systems for product health utilising data and information from different sources and manufacturing stages is a key factor to meet these growing demands. This paper addresses the process of combining new measurement data or information with machine learning-based prediction information obtained as each product goes through a series of manufacturing steps to update the conditional probability distribution of the end product quality during manufacturing. A Bayesian approach is adopted in obtaining an updated posterior distribution of the end product quality given new information from subsequent measurements, and, in particular, On-Machine Probing (OMP). Following the steps of heat treatment, machining, and OMP, the posterior distribution of the previous step can be considered as the new prior distribution to obtain an updated posterior distribution of the product condition as new metrological information becomes available. It is demonstrated that the resulting posterior estimates can lead to more efficient product condition monitoring in multistage manufacturing.

Bayesian information fusion method for reliability analysis with failure‐time data and degradation data

AbstractDegradation data and failure‐time data are two types of data commonly used in reliability analysis. Both types of data are collected from different sources for reliability analysis of complex products. For highly reliable products, however, it is often difficult to collect sufficient useful data for reliability analysis with high accuracy, which poses the challenge for small sample size problems, that is, single‐type data with few samples. In this paper, three novel Bayesian information fusion models are first proposed to characterize the inherent relationship between the failure‐time data and the degradation data, and further to integrate the heterogeneous data to obtain accurate reliability analysis results under small sample size. Then, a model selection method is developed to choose appropriate model from the Wiener process, gamma process, and IG process models. Finally, the reliability analysis is completed based on the parameter estimation of the Bayesian information fusion model with the aid of the MCMC method. An industrial example is presented to demonstrate the effectiveness of the proposed Bayesian information fusion method.

Automatic Docking Optimization Method of Bed and Chair Based on Multi-sensor Information Fusion

<p>Automatic docking is a basic function of intelligent wheelchair bed system. In order to achieve high-precision docking between intelligent wheelchair and U-shaped bed, this paper proposes an optimization method of automatic docking between bed and chair based on multi-sensor information fusion. By installing a variety of sensors, such as depth vision camera and lidar ranging, there is usually cross sensitivity when multiple sensors work together. Based on real sensing technology，wireless sensor network technology and lidar ranging equipment l, a dynamic Bayesian multi-mode information fusion method with asynchronous constraints is proposed. Wireless sensors are used to collect wheelchair pose information in real time. By analyzing the characteristics of the observation information of various sensors, the complementary and redundant information of various sensors in space and time is combined according to optimization criteria. The data fitting between multi-sensor information is carried out efficiently and quickly by surface fitting method. Based on the extended Kalman filter algorithm, the information fusion of IMU and combined sensing unit is realized, and the analysis model is established. Finally, the feasibility of the method is verified by Matlab simulation and the use of Gazebo in ROS to build a simulation experiment environment for fusion algorithm verification.</p> <p> </p>

Bayesian Inference Approach for Information Fusion in Distribution System State Estimation

This paper presents a three-phase Distribution System State Estimator (DSSE) based on a Bayesian inference approach to manage different sampling rates of typical sources of information present in distribution networks. Such information comes from smart meters, supervisory control and data acquisition (SCADA) measurements, phasor measurement units and typical load profiles from pseudo measurements. The temporal aspect of the measurement set is incorporated in the estimation process by using a sampling layer concept, dealing separately with each group of measurements according to the respective updating rate. A Bayesian information fusion procedure provides the final estimation. The proposed DSSE consists in a multiple stage estimator that combines a prior model for the state variables, updated by new observations from measured values in each sampling layer, through Maximum a Posteriori estimation. This work also introduces an orthogonal method for the information fusion numerical solution, to tackle the severe ill-conditioning associated with practical distribution systems. Simulations with IEEE distribution test feeders and a Brazilian real distribution feeder illustrate the features of the proposed DSSE and its applicability. By exploring the concept of credibility intervals, the method is able to detect events on the grid, such as subtle load variation and contingencies, while maintaining accuracy.

A Bayesian Network-Based Information Fusion Combined with DNNs for Robust Video Fire Detection

Fire is an abnormal event that can cause significant damage to lives and property. Deep learning approach has made large progress in vision-based fire detection. However, there is still the problem of false detections due to the objects which have similar fire-like visual properties such as colors or textures. In the previous video-based approach, Faster Region-based Convolutional Neural Network (R-CNN) is used to detect the suspected regions of fire (SRoFs), and long short-term memory (LSTM) accumulates the local features within the bounding boxes to decide a fire in a short-term period. Then, majority voting of the short-term decisions is taken to make the decision reliable in a long-term period. To ensure that the final fire decision is more robust, however, this paper proposes to use a Bayesian network to fuse various types of information. Because there are so many types of Bayesian network according to the situations or domains where the fire detection is needed, we construct a simple Bayesian network as an example which combines environmental information (e.g., humidity) with visual information including the results of location recognition and smoke detection, and long-term video-based majority voting. Our experiments show that the Bayesian network successfully improves the fire detection accuracy when compared against the previous video-based method and the state of art performance has been achieved with a public dataset. The proposed method also reduces the latency for perfect fire decisions, as compared with the previous video-based method.

Application of information fusion techniques and satellite products in the optimal redesign of rain gauge networks

Rain gauge networks are usually redesigned to improve the accuracy of spatial and temporal estimates and reduce the monitoring costs. In addition to the rain-gauge data, satellite-based data can be used for the optimal redesign of rainfall monitoring networks. Despite high spatial and temporal resolutions of products of meteorological satellites, they must be calibrated using ground-based data. In this paper, a new methodology is presented for an optimal redesign of rain-gauge networks. In this methodology, after the bias correction of some satellite-based rainfall products (i.e., PERSIANN-CCS, TRMM3B42RTV7, and CMORPH), they are combined using a Bayesian information fusion technique. The spatiotemporal rainfall maps, which are developed using the fusion model, are used in an optimization model. The optimization model provides the optimal locations of the rain-gauge stations, considering the objectives of minimizing both the monitoring cost and the average standard deviation of estimation errors. To illustrate the applicability and efficiency of the proposed methodology, it is applied to the Khorasan Razavi Province in Iran. The results show that by using the satellite data and the Bayesian information fusion technique, the summation of standard deviations of estimation errors is significantly reduced from 18,787.9 to 778.28 mm.

Bayesian information fusion and multitarget tracking for maritime situational awareness

The goal of maritime situational awareness (MSA) is to provide a seamless wide-area operational picture of ship traffic in coastal areas and the oceans in real time. Radar is a central sensing modality for MSA. In particular, oceanographic high-frequency surface-wave (HFSW) radars are attractive for surveying large sea areas at over-the-horizon distances, due to their low environmental footprint and low power requirements. However, their design is not optimal for the challenging conditions prevalent in MSA applications, thus calling for the development of dedicated information fusion and multisensor-multitarget tracking algorithms. In this study, the authors show how the multisensor-multitarget tracking problem can be formulated in a Bayesian framework and efficiently solved by running the loopy sum-product algorithm on a suitably devised factor graph. Compared to previously proposed methods, this approach is advantageous in terms of estimation accuracy, computational complexity, implementation flexibility, and scalability. Moreover, its performance can be further enhanced by estimating unknown model parameters in an online fashion and by fusing automatic identification system (AIS) data and context-based information. The effectiveness of the proposed Bayesian multisensor-multitarget tracking and information fusion algorithms is demonstrated through experimental results based on simulated data as well as real HFSW radar data and real AIS data.

An adaptive variational Bayesian filter for nonlinear multi-sensor systems with unknown noise statistics

An improved adaptive variational Bayesian cubature information fusion algorithm for nonlinear multi-sensor systems with uncertain noise statistics is proposed in this paper. Aiming to estimate uncertain process and measurement noise covariances in nonlinear systems, the variational Bayesian theory is combined with the inverse Wishart distribution. System states and uncertain noise covariances are jointly estimated for nonlinear systems by means of cubature sampling, deriving the variational Bayesian cubature Kalman filter (VBCKF-QR). In addition, a variational Bayesian Cubature Information filter (VBCIF-QR) is proposed, and a distributed information feedback fusion algorithm is also derived for multi-sensor systems with unknown noise statistics. Simulation results demonstrate that the proposed VBCKF-QR/VBCIF-QR outperform conventional cubature Kalman/information filters.

Bayesian prior information fusion for power law process via evidence theory

The power law process (PLP) is widely used to analyze the failures of repairable systems, and the PLP parameter estimation is the primary concern for reliability assessment or maintenance decision making. Although the Bayesian estimation of the PLP has been studied in existing research, little attention has been paid to how to obtain its prior distribution, especially when the prior information is coming from multiple sources. To address this problem, a framework for Bayesian prior information fusion using evidence theory is proposed in this paper. This framework first uses evidence theory to represent the prior information from multiple sources or experts and then combines them into fused information. Based on the belief and plausibility functions of the fused information, the prior distribution is bounded by an upper and lower probability density functions which are derived by moment equivalence. A case study is also carried out to verify and illustrate the proposed method. The results show that this proposed approach is beneficial for the Bayesian estimation of the power law process.

Research on Fault Detection Method of Mineral Powder Production Line Based on Information Fusion

By analysing the uncertainty of information in the production process of the mine, and the diversity of the fault detection components of the production equipment. Combining the trend of big data development, comparing and analysing various fault diagnosis methods, a Bayesian information fusion algorithm based on improved learning process is proposed. The traditional Bayesian information fusion algorithm is combined with the learning idea of migration algorithm to obtain an optimized Bayesian fault detection algorithm. This algorithm can minimize the waste and loss of data and improve the detection accuracy.

Optimisation estimation of uncertainty integrated with production information based on Bayesian fusion method

The new generation Geometrical Product Specifications require consideration of the effects of measurement uncertainty in the product inspection. This study estimated the measurement results and the uncertainty by integrating the statistical production information into the product detection results to rationally and fairly narrow the uncertainty area of qualification determination. Based on the Bayesian information fusion and statistical inference principle, the model of uncertainty evaluation is established. The Bayesian information fusion model integrated measuring information with manufacturing information was built, with which the uncertainty of product inspection was reappraised based on posteriori distribution function. The validity of the proposed method and theory was demonstrated by the example analysis.

Bayesian uncertainty quantification and information fusion in CALPHAD-based thermodynamic modeling

Calculation of phase diagrams is one of the fundamental tools in alloy design—more specifically under the framework of Integrated Computational Materials Engineering. Uncertainty quantification of phase diagrams is the first step required to provide confidence for decision making in property- or performance-based design. As a manner of illustration, a thorough probabilistic assessment of the CALPHAD model parameters is performed against the available data for a Hf–Si binary case study using a Markov Chain Monte Carlo sampling approach. The plausible optimum values and uncertainties of the parameters are thus obtained, which can be propagated to the resulting phase diagram. Using the parameter values obtained from deterministic optimization in a computational thermodynamic assessment tool (in this case Thermo-Calc) as the prior information for the parameter values and ranges in the sampling process is often necessary to achieve a reasonable cost for uncertainty quantification. This brings up the problem of finding an appropriate CALPHAD model with high-level of confidence which is a very hard and costly task that requires considerable expert skill. A Bayesian hypothesis testing based on Bayes’ factors is proposed to fulfill the need of model selection in this case, which is applied to compare four recommended models for the Hf–Si system. However, it is demonstrated that information fusion approaches, i.e., Bayesian model averaging and an error correlation-based model fusion, can be used to combine the useful information existing in all the given models rather than just using the best selected model, which may lack some information about the system being modelled.

Bayesian information fusion for degradation analysis of deteriorating products with individual heterogeneity

Degradation analysis is a popular and effective method for reliability analysis of long-life and high-reliability products. However, for newly developed products, especially for highly customized products with small sample size, the challenge of sparse degradation observations with product heterogeneity is still an open issue deserving further research. In this article, Bayesian degradation analysis is presented for reliability analysis of products with heterogeneity. The degradation process is modeled by a Gamma process. Random effects are incorporated in the Gamma process model for characterizing the individual heterogeneity. To improve the precision of parameter estimation and degradation analysis, a Bayesian information fusion is presented to leverage degradation information from multiple sources. The proposed model is demonstrated through degradation-based reliability analysis of heavy-duty machine tool’s spindle system, which is characterized as degradation analysis with individual heterogeneity and information fusion.

Analysis and Comparison of Bayesian Methods for Measurement Uncertainty Evaluation

Based on the Bayesian principle, the modern uncertainty evaluation methods can fully integrate prior and current sample information, determine the prior distribution according to historical information, and deduce the posterior distribution by integrating prior distribution and the current sample data with the Bayesian model. As such, it is possible to evaluate uncertainty, updating in real time the uncertainty of the measuring instrument according to regular measurement, and timely reflect the latest information on the accuracy of the measurement system. Based on the Bayesian information fusion and statistical inference principle, the model of uncertainty evaluation is established. The maximum entropy principle and the hill-climbing search optimization algorithm are introduced to determine the prior distribution probability density function and the sample information likelihood function. The probability density function of posterior distribution is obtained by the Bayesian formula to achieve the optimization estimation of uncertainty. Three methods of measurement uncertainty evaluation based on Bayesian analysis are introduced: the noninformative prior, the conjugate prior, and the maximum entropy prior distribution. The advantages and limitations of each method are discussed.

Bayesian Network inference for probabilistic strength estimation of aging pipeline systems

A novel approach for reliability assessment of aging gas pipeline systems based on a Bayesian network methodology is proposed in this paper with a focus on the improvement of the pipeline strength prediction. A multimodal diagnosis is performed by assessing the variation in the mechanical property (e.g., yield strength) within the pipe in terms of material property measurements, such as microstructure, composition, and hardness. The multimodality measurements are then integrated with the Bayesian network information fusion model. Prototype testing is carried out for model verification, validation and demonstration. The model updating scheme employs a Markov Chain Monte Carlo algorithm to infer the posterior distribution of the pipe strength using the multimodality measurements, whereas, the priors are derived from the literature knowledge of such systems. Moreover, through-thickness studies of pipe cross-sections are performed to demonstrate the mechanical property variation from the surface to bulk. Finally, data training of the model is employed to obtain a more accurate measure of the probabilistic pipe strength. Discussions on the observations and future work are provided.

Bayesian information fusion for probabilistic health index of power transformer

This study proposes a Bayesian information fusion approach for determining the probabilistic health index of power transformer. The proposed approach integrates a variety of data obtained from transformer measurements, maintenance records, and failure statistics. By making use of these data, an inference model is constructed using Bayesian belief network (BBN). In the inference model, the significance of each individual measurement on the corresponding component in the BBN is decided by both principal component analysis and expert's experience, and subsequently quantified with a score-probability transform. Finally, the inference model is used to derive a probabilistic health index. Case studies are provided to demonstrate the applicability of the proposed approach for evaluating transformer condition.