Dempster-Shafer Research Articles

A comparison of imprecise Bayesianism and Dempster–Shafer theory for automated decisions under ambiguity

Abstract Ambiguity occurs insofar as a reasoner lacks information about the relevant physical probabilities. There are objections to the application of standard Bayesian inductive logic and decision theory in contexts of significant ambiguity. A variety of alternative frameworks for reasoning under ambiguity have been proposed. Two of the most prominent are Imprecise Bayesianism and Dempster–Shafer theory. We compare these inductive logics with respect to the Ambiguity Dilemma, which is a problem that has been raised for Imprecise Bayesianism. We develop an agent-based model comparison that isolates the difference between the two inductive logics in their updating methods. We find that Dempster–Shafer theory does not avoid the Ambiguity Dilemma. We discuss the implications of this result.

A comprehensive framework for evaluating and predicting investment risks in renewable energy projects: a case study of Maynak hydropower station

ABSTRACT This study introduces a holistic analysis framework designed to evaluate and predict the investment risks associated with foreign renewable energy initiatives. This framework aims to reduce the uncertainties inherent in renewable energy projects. To achieve this, we utilize the variable weight matter-element extension model to establish the project’s fundamental reliability function. We then refine this model by incorporating evidence theory to assess the project’s risk level and derive precise risk index measurements. We utilize the GM model to forecast potential future risks associated with the project. In addition, a case study is presented on the risk assessment and prediction for the Maynak Hydropower Station. Our findings reveals that the project encountered significant investment risks in 2008, 2014, 2020, and 2022. Key risk indicators include political turmoil，policy change，imperfect legislation，cultural risk，exchange rate changes，technical risks and management risks. Furthermore, the investment risk level declined from 2023 to 2027, with our risk measurement outcomes closely aligning with actual conditions, thereby validating the effectiveness and applicability of our model.

A compressor blade crack detection method based on the multilevel information fusion of acoustic and vibration signals

Single-signal crack detection methods perform unsatisfactorily in the presence of noise, prompting the need for alternative approaches. Leveraging multisource signals, which contain rich and complementary information, is a promising direction. In light of this, a compressor blade crack detection method based on the multilevel information fusion of acoustic and vibration signals is proposed. First, a multiscale data-level fusion convolutional neural network is designed, which fuses multisource homogeneous signals for crack detection. Second, multiple networks are trained with acoustic and vibration signals as inputs, respectively. Softmax layer of each network provides preliminary probabilities for each category, while the precision for each category in the validation set is calculated. Finally, an improved Dempster–Shafer theory approach is proposed to derive the final probability for each category according to the preliminary probabilities and precision, after which crack detection is realized. The proposed method is validated with experimental data from five categories of blades.

Small sample composite fault diagnosis of hydraulic bearings based on improved VME algorithm and mRVM

The working environment of rolling bearings is complex. Once a fault occurs, various parts will affect each other and produce a compound fault. Traditional methods often use signal separation algorithms to separate different types of signals for fault diagnosis, but it is difficult to analyze specific faults efficiently and accurately. To solve this problem, this paper combines variational mode decomposition (VMD), Laplace energy index (LE) and variational mode extraction (VME) for signal extraction. Multi-class relevance vector machine (mRVM) and DS evidence theory are used for intelligent fault diagnosis, focusing on the context of small sample data. First, the VMD-LE-VME method is used to extract effective fault information from the fault signal and obtain multi-domain features. Then, the multi-domain features are input into mRVM for fault identification. Finally, the classification results are fused through DS evidence theory to obtain the final classification results. The effectiveness and superiority of this method in processing small sample data are verified by experiments.

Knowledge-based and Expert Systems in Prognostics and Health Management: a Survey

Prognostics and Health Management (PHM) has become increasingly popular in recent years, and data-driven methods and artificial intelligence have emerged as dominant tools within the PHM field. This trend is mainly due to the increasing use of sensors and the ability of machine learning techniques to leverage condition monitoring data. However, despite their utility and effectiveness, these techniques are not without drawbacks. One major issue is that data-driven methods often lack transparency in their reasoning, which is crucial for understanding fault occurrences and diagnostics. Additionally, the availability of data can be a challenge. In some cases, data are scarce or hard to obtain, either due to the cost of installing necessary sensors or the rarity of the required information. Lastly, the insights derived from data can sometimes diverge from those obtained through expert analysis and established norms. This contrasts with knowledge-based approaches such as expert systems, which formally organize the knowledge acquired from norms and experts, and then deduce the desired conclusion. While research is increasingly exploring data-driven techniques, industry tends to still frequently employ knowledge-based methods. To fill this gap, this paper offers a detailed survey of knowledge-based and expert systems in PHM, examining methodologies such as propositional logic, fuzzy logic, Dempster-Shafer theory and Bayesian networks. It assesses the integration and impact of these techniques in PHM for fault detection, diagnosis and prognosis, highlighting their strengths, limitations, and potential future developments. The study provides a thorough evaluation of current developments and contributes significant insights into the current capabilities and future directions of knowledge-based techniques in enhancing decision-making processes in PHM.

A study of multi-model identification and fusion control algorithms for rotary limit devices

In order to improve the control performance of rotating limit device on the motion range of rotating parts of mechanical equipment, the multi model identification and fusion control algorithm of rotating limit device is studied. The angle sensor and speed sensor are used to collect the rotation angle and speed of rotating machinery equipment. Taking the collected data as the input of multiple support vector machines, the D-S evidence theory is selected to fuse the status of rotating machinery equipment identified by multiple models and determine the final status of rotating machinery equipment. Set the deviation between the target angle and the actual angle of the rotating mechanical equipment as the input of the PID controller. The PID controller determines the limit control quantity of the rotating limit device to the rotating machinery equipment, and realizes the rotating limit control of the rotating machinery equipment. The rotary table is selected as the mechanical equipment controlled by the rotary limit device. The experimental results show that the rotary limit device can control the rotation of the rotary table in the ideal angle range, achieve smooth tracking of the rotation angular velocity, and has excellent control performance.

A fault diagnosis method for bearings and gears in rotating machinery based on data fusion and transfer learning

Abstract Rotating machinery is a crucial component of industrial equipment, and the fault diagnosis of bearings and gears, as vital elements of rotating machinery, is essential since they often fail under harsh working conditions, leading to significant property losses and serious personal safety problems. However, fault data for gears and bearings are often sparse in actual condition, and it is a challenge to ensure the reliability and stability of fault diagnosis results by extracting the features of a single data. To solve the above problems, this paper proposes a fault diagnosis method that combines Transfer Learning and data fusion techniques. Firstly, in this method, two kinds of fault signals are transformed into Gramian Angular Difference Fields and Recurrence Plot. Next, a U-shaped feature fusion dual discriminator generative adversarial network is used to fuse two-dimensional images from multiple sensor data. Its feature fusion module deeply integrates the features of the two images, thereby solving the impact of single data on the reliability and stability of fault diagnosis. Moreover, open-source datasets are used for Transfer Learning training to tackle the small sample problem. Finally, a decision-level information fusion classifier, the Dual-Branch Dempster-Shafer Classifier (DB-DSC), classifies the fused images. This classifier incorporates an improved soft threshold function and D-S evidence theory to achieve adaptive gradient changes and improve the robustness and accuracy of classification results. The experimental results show the effectiveness and stability of the proposed method, and the generated images get high score in several metrics. The average classification accuracy of the classification network reaches 93% and 92.5% on the two datasets, Therefore, the proposed method exhibits strong fault diagnosis capabilities under the small sample conditions of bearings and gears.

Research on the Structural Design of a Pressurized Cabin for Civil High-Speed Rotorcraft and the Multi-Dimensional Comprehensive Evaluation Method

For civil high-speed rotorcraft designed to operate at specific cruising altitudes, this study proposes nine structural design schemes for pressurized cabins. These schemes integrate commonly used materials and processing technologies in the aviation industry with advanced PRSEUS (Pultruded Rod Stitched Efficient Unitized Structure) technology. An analysis of the structural composition reveals that frames constitute 8–19% of the total structural weight, while stringers and beams make up 15–50%, and skins account for 11–25%, with thicknesses ranging from 1.0 mm to 2.0 mm. The separating interface of the pressurized cabin contributes 4–29% of the total structural weight. The weight distribution of each component in the pressurized cabin structure varies significantly depending on the chosen materials and processing technologies. Utilizing the Analytic Hierarchy Process (AHP), along with Gray Relational Analysis (GRA) and Dempster–Shafer (D-S) evidence theory, this study compares the simulation results of the nine schemes across multiple dimensions. The findings indicate that the configuration combining 7075 aluminum alloy and T300 composite material has the greatest advantages in terms of the high structural reliability of the configuration, light weight, mature processing technology, and low production cost. This comprehensive evaluation method quantitatively analyzes the factors influencing the structural configuration design of the pressurized cabin for civil high-speed rotorcraft, offering a valuable reference for the design of similar structures in related fields.

The use of electronic evidence in the criminal process of foreign countries

It is emphasized that electronic evidence has become indispensable for criminal investigations and trials around the world. The importance of legal regulation of the criminal procedural form of digital technologies is emphasized. The improper nature of such regulation entails certain legal consequences, including the recognition of electronic evidence as improper or inadmissible. The experience of some foreign countries in the field of recognition and use of electronic evidence (the USA, Great Britain, the People’s Republic of China and the countries of the European Union) is analyzed. It has been proven that each state builds its own way and approach to incorporating such evidence into its legal systems depending on legal traditions, technological infrastructure and privacy considerations. From the analysis of procedural codes, it was noted that procedural science develops a theory of electronic evidence according to the industry principle. It is emphasized that they are based on common digital technologies, therefore this provides an opportunity to develop uniform cross­industry rules for the use of electronic evidence in any process. It was determined that in the Anglo-Saxon legal system there is no clear division of evidence into types. Evidence law is based on typical problem situations. The US legislation, which regulates the use of electronic evidence in the criminal process, is analyzed. It is noted that the condition for their acceptance is its authenticity and relevance to the case. It is noted that in the UK, all electronic evidence is subject to the same rules and laws that apply to documentary evidence. The legislation of the European Union, which regulates electronic evidence and allows to strengthen cooperation between member states in the field of using electronic evidence, is analyzed. It is emphasized that as states continue to adapt to the age of digital technologies, the development of a reliable legal framework and international cooperation will be essential to ensure the effective and fair use of electronic evidence in the pursuit of justice.

Coal-Mine Water-Hazard Risk Evaluation Based on the Combination of Extension Theory, Game Theory, and Dempster–Shafer Evidence Theory

Due to the complex hydrogeological conditions and water hazards in coal mines, there are multiple indexes, complexities, incompatibilities, and uncertainty issues in the risk evaluation process of coal-mine water hazards. To accurately evaluate the risk of coal-mine water hazards, a comprehensive evaluation method based on extension theory, game theory, and Dempster–Shafer (DS) evidence theory is proposed. Firstly, a hierarchical water-hazard risk-evaluation index system is established, and then matter-element theory in extension theory is used to establish a matter-element model for coal-mine water-hazard risk. The membership relationship between various evaluation indexes and risk grades of coal-mine water-hazard risk is quantified using correlation functions of extension set theory, and the quantitative results are normalized to obtain basic belief assignments (BBAs) of risk grades for each index. Then, the subjective weights of evaluation indexes are calculated using the order relation analysis (G1) method, and the objective weights of evaluation indexes are calculated using the entropy weight (EW) method. The improved combination weighting method of game theory (ICWMGT) is introduced to determine the combination weight of each evaluation index, which is used to correct the BBAs of risk grades for each index. Finally, the fusion of DS evidence theory based on matrix analysis is used to fuse BBAs, and the rating with the highest belief fusion result is taken as the final evaluation result. The evaluation model was applied to the water-hazard risk evaluation of Sangbei Coal Mine, the evaluation result was of II grade water-hazard risk, and it was in line with the actual engineering situation. The evaluation result was compared with the evaluation results of three methods, namely the expert scoring method, the fuzzy comprehensive evaluation method, and the extension method. The scientificity and reliability of the method adopted in this paper were verified through this method. At the same time, based on the evaluation results, in-depth data mining was conducted on the risk indexes of coal-mine water hazards, and it was mainly found that 11 secondary indexes are the focus of coal-mine water-hazard risk prevention and control, among which seven indexes are the primary starting point for coal-mine water-hazard risk prevention and control. The groundwater index in particular has the most prominent impact. These results can provide a theoretical basis and scientific guidance for the specific water-hazard prevention and control work of coal mines.

A dynamic system reliability analysis model on safety instrumented systems

This paper introduces a novel hybrid dynamic model for complex systems reliability assessment. The model synergizes expert knowledge elicitation and an enhanced Dempster-Shafer Theory (DST) with Dynamic Bayesian Networks (DBNs) modeling, aiming to surmount the limitations such as uncertainty and static modeling inherent in traditional methods. The proposed model is deployed on a Safety Instrumented System (SIS) designed to prevent runaway reactions within a Continuously Stirred Tank Reactor (CSTR), considering factors such as system degradation, human interventions, and proof testing on system reliability. The analysis pinpointed the logic solver subsystem as the principal vulnerability within the assessed SIS, leading to targeted recommendations to bolster system reliability. The outcomes offer insights for a wide range of safety-critical systems aiming to augment the safety and efficacy of SISs, thereby advancing safety and resilience management across various complex engineering systems, particularly in contexts where field data is scant.

A Tunnel Fire Detection Method Based on an Improved Dempster-Shafer Evidence Theory

Tunnel fires are generally detected using various sensors, including measuring temperature, CO concentration, and smoke concentration. To address the ambiguity and inconsistency in multi-sensor data, this paper proposes a tunnel fire detection method based on an improved Dempster-Shafer (DS) evidence theory for multi-sensor data fusion. To solve the problem of evidence conflict in the DS theory, a two-level multi-sensor data fusion framework is adopted. The first level of fusion involves feature fusion of the same type of sensor data, removing ambiguous data to obtain characteristic data, and calculating the basic probability assignment (BPA) function through the feature interval. The second-level fusion derives basic probability numbers from the BPA, calculates the degree of evidence conflict, normalizes the BPA to obtain the relative conflict degree, and optimizes the BPA using the trust coefficient. The classical DS evidence theory is then used to integrate and obtain the probability of tunnel fire occurrence. Different heat release rates, tunnel wind speeds, and fire locations are set, forming six fire scenarios. Sensor monitoring data under each simulation condition are extracted and fused using the improved DS evidence theory. The results show that there is a 67.5%, 83.5%, 76.8%, 83%, 79.6%, and 84.1% probability of detecting fire when it occurs, respectively, and identifies fire occurrence in approximately 2.4 s, an improvement from 64.7% to 70% over traditional methods. This demonstrates the feasibility and superiority of the proposed method, highlighting its significant importance in ensuring personnel safety.

Graph-Structure-Based Multigranular Belief Fusion for Human Activity Recognition.

The belief functions (BFs) introduced by Shafer in the mid of 1970s are widely applied in information fusion to model epistemic uncertainty and to reason about uncertainty. Their success in applications is however limited because of their high-computational complexity in the fusion process, especially when the number of focal elements is large. To reduce the complexity of reasoning with BFs, we can envisage as a first method to reduce the number of focal elements involved in the fusion process to convert the original basic belief assignments (BBAs) into simpler ones, or as a second method to use a simple rule of combination with potentially a loss of the specificity and pertinence of the fusion result, or to apply both methods jointly. In this article, we focus on the first method and propose a new BBA granulation method inspired by the community clustering of nodes in graph networks. This article studies a novel efficient multigranular belief fusion (MGBF) method. Specifically, focal elements are regarded as nodes in the graph structure, and the distance between nodes will be used to discover the local community relationship of focal elements. Afterward, the nodes belonging to the decision-making community are specially selected, and then the derived multigranular sources of evidence can be efficiently combined. To evaluate the effectiveness of the proposed graph-based MGBF, we further apply this new approach to combine the outputs of convolutional neural networks + attention (CNN + Attention) in the human activity recognition (HAR) problem. The experimental results obtained with real datasets prove the potential interest and feasibility of our proposed strategy with respect to classical BF fusion methods.

Evidence combination with multi-granularity belief structure for pattern classification

Belief function (BF) theory provides a framework for effective modeling, quantifying uncertainty, and combining evidence, rendering it a potent tool for tackling uncertain decision-making problems. However, with the expansion of the frame of discernment, the increasing number of focal elements processed during the fusion procedure leads to a rapid increase in computational complexity, which limits the practical application of BF theory. To overcome this issue, a novel multi-granularity belief structure (MGBS) method was proposed in this study. The construction of MGBS reduced the number of focal elements and preserved crucial information in the basic belief assignment. This effectively reduced the computational complexity of fusion while ensuring the highest possible classification accuracy. We applied the proposed MGBS algorithm to a human activity recognition task and verified its effectiveness using the University of California, Irvine mHealth, PAMAP2, and Smartphone datasets.

EMNet: An ensemble deep learning approach for geological condition detection in tunnel excavation

This paper established a novel ensemble deep learning method for the identification of the geological condition encountered during tunnel boring machine (TBM) operation. By integrating multiple MobileNet base models with information fusion through the Dempster-Shafer theory (DST), the proposed method is able to provide reliable identification of the geological condition based on the images of the excavated mucks on TBM’s conveyor belt. Shapley additive explanations (SHAP) analysis is further presented for model interpretation and understanding of the key features in the images. A case study using data collected from Singapore’s Circle Line 6 tunnel construction projects is conducted to verify the effectiveness of the proposed method. The results indicate that (1) The proposed method detects the encountered geological condition with high accuracy; (2) EMNet outperforms all the base MobileNet models; (3) SHAP analysis demonstrates that surface texture could be the key feature that assists the model for the classification. The developed method contributes to the state of knowledge in proposing a novel ensemble method for reliable image identification and the state of practice in proposing a useful tool to identify the encountered geological condition automatically.

An Evidential Multi-Target Domain Adaptation Method Based on Weighted Fusion for Cross-Domain Pattern Classification.

For cross-domain pattern classification, the supervised information (i.e., labeled patterns) in the source domain is often employed to help classify the unlabeled target domain patterns. In practice, multiple target domains are usually available. The unlabeled patterns (in different target domains) which have high-confidence predictions, can also provide some pseudo-supervised information for the downstream classification task. The performance in each target domain would be further improved if the pseudo-supervised information in different target domains can be effectively used. To this end, we propose an evidential multi-target domain adaptation (EMDA) method to take full advantage of the useful information in the single-source and multiple target domains. In EMDA, we first align distributions of the source and target domains by reducing maximum mean discrepancy (MMD) and covariance difference across domains. After that, we use the classifier learned by the labeled source domain data to classify query patterns in the target domains. The query patterns with high-confidence predictions are then selected to train a new classifier for yielding an extra piece of soft classification results of query patterns. The two pieces of soft classification results are then combined by evidence theory. In practice, their reliabilities/weights are usually diverse, and an equal treatment of them often yields the unreliable combination result. Thus, we propose to use the distribution discrepancy across domains to estimate their weighting factors, and discount them before fusing. The evidential combination of the two pieces of discounted soft classification results is employed to make the final class decision. The effectiveness of EMDA was verified by comparing with many advanced domain adaptation methods on several cross-domain pattern classification benchmark datasets.

Computer Simulation and Simulation of Mechanical and Electrical Equipment based on Artificial Intelligence Algorithms

The computer in mechanical and electrical equipment can now detect equipment faults through simulation thanks to the advancement of artificial intelligence (AI) technology, which makes it convenient to monitor mechanical and electrical equipment. This paper begins with a quick introduction to Agent and the Agent system, explains how Agent is applied in the current context, then analyzes the hierarchical fault diagnostic model, identifies its flaws, and suggests improvement techniques. To increase the model’s accuracy and speed of operation, the contract net model and D-S (Dempster-Shafer) evidence theory are then incorporated. Finally, simulation experiments are used to confirm the accuracy and consistency of this model. According to the experimental findings, this optimization model runs at a pace that is noticeably faster than that of other models when subjected to the same workload, demonstrating the model’s efficacy. Models 1, 2, and 3 are put side by side to demonstrate how clearly multi-task processing may cut down on the model’s running time. In the second experiment, samples of mechanical and electrical equipment defect data are taken from two groups. The results of the comparative experiments demonstrate that the optimized model in this work can be reliable up to a maximum of 0.91 and a minimum of 0.63. It is demonstrated that the model in this work is rational by the fact that among the four types of fault prediction, the optimized model’s reliability is significantly greater than the traditional model’s. The research described in this publication therefore has some reference value for computer simulation of electromechanical equipment.

Landslide susceptibility assessment along highways (SH-12 and NH-717A) in Darjeeling Himalayas

Landslides are the utmost damaging natural along with anthropogenic hazards which cause huge damage to assets and human lives in hilly environments all over the world. The key aim of this study is to explore landslide susceptibility (LS) maps along highways (SH-12 and NH-717A) in the Darjeeling Himalayas, India. Four data-driven bivariate statistical models (BSMs) were used to develop LS maps based on geospatial techniques. A total of 90 (100%) polygons with different size of landslide locations were identified to develop landslide inventory/distribution (LI/LD) map by incorporating Google Earth and satellite imageries, and filed investigation through the global positioning system (GPS). The LI/LD map is distributed into testing 30% (27 landslide locations), and training 70% (63 landslide locations) of the models. Eighteen landslide causative factors (LCFs) were selected to develop LS maps. Multicollinearity analysis showed that there is no collinearity problem in the LCFs. The receiver operating characteristics (ROC) curve and frequency ratio (FR) value techniques were applied for validation. The relative importance of each LCF for each model has been measured using Jack-knife test. The LS maps were divided into six landslide susceptibility zones (LSZs). The ROC curve showed the accuracy of the LSZ maps is 93.50% for FR, 85.90% for evidential belief functions (EBFs), 81.80% for fuzzy membership (FM) using FR, and 77.20% for FM using cosine amplitude (CA) approaches respectively. Among all models, the FR model showed the highest percentage of accuracy for LS assessment and prediction. The FR values of six LSZs rise from very low to very high LSZs which indicate a positive correlation between LSZs and FR values. The results of Jack-knife test showed that terrain ruggedness index (TRI) factor has the highest relative importance for all models in LSZ mappings. The study findings will help to the planner, and policymakers for spatial planning, implement slope management strategies, land use management in mountainous environments.

Deceptive evidence detection of belief functions based on reinforcement learning in partial label environment

Counter-deception evidence fusion is a critical issue in the application of Dempster-Shafer Theory (DST). Effectively detecting deceptive evidence poses a significant challenge in DST-based information fusion. Existing research on this topic is limited and often lacks a clear distinction between deceptive and trustworthy evidence. Recently, two explicit definitions of deceptive evidence have been proposed to address different scenarios: one for cases with label information and another for cases without. However, these definitions are somewhat counter-intuitive and do not address situations where partial label information is available.To address this gap, our paper introduces a new, explicit definition of deceptive evidence that considers both the characteristics of the evidence and the fusion system. This definition encompasses cases including with label information, without label information, and with partial label information. It extends the two previously mentioned definitions and, in certain circumstances, aligns with them.Based on our new definition, we propose a mathematical model for counter-deception evidence fusion across these three scenarios and apply reinforcement learning to solve it. We present several numerical simulations, a data-driven counter-deception test, and a practical application to demonstrate that our method outperforms previous approaches in both detecting deceptive evidence and in practical applications, showcasing superior effectiveness and robustness.

A Knowledge Distillation Method Based on Evidence Theory to Prevent Catastrophic Forgetting

Abstract Incremental learning is an emerging machine-learning approach designed to prevent catastrophic forgetting while learning a new task with knowledge retained from previous tasks. However, existing methods often must fully utilize the distributional information of the old task model’s outputs. To alleviate this, we propose a class incremental learning method grounded in evidence theory to leverage the distributional information of outputs from the old task model, which integrates uncertainty estimation into the knowledge distillation process to ensure the new task model effectively learns from the old task model’s distribution information. Specifically, we incorporate uncertainty estimation based on evidence theory to calculate knowledge distillation loss during training. We compute the uncertainty estimates of outputs from both the new and old task models, and then they are used in the knowledge distillation loss calculation. We propose a novel classification strategy that considers outputs’ probability and uncertainty estimates, which could determine the sample category without requiring an additional training phase or supplementary models. Experiments on the CIFAR-100 and ImageNet datasets demonstrate the effectiveness of the proposed method, showing that it outperforms existing methods by improving accuracy by 1%-2%. Results suggest that leveraging the distribution information of model outputs can effectively mitigate catastrophic forgetting in deep learning models within incremental learning scenarios.