Multi-sensor Data Fusion Approach Research Articles

Analysis of the Motion Postures in Equestrian Sports Based on Multi-Sensor Data Fusion

Horse riding, in its essence, is both an art and a captivating sport. In these events, riders showcase their refined skills and deep bond with their horses, most notably through changes in physical positioning. This research aims to accurately capture and analyze the spatial postures of riders in equestrian sports, exploring the dynamic differences between professional and amateur riders. To achieve this, we employed a multi-sensor data fusion approach based on human kinetics theories, enhanced by an extended Kalman filter. This method, combined with an optical tracking system, enabled us to intricately compare and analyze the 3D postures of riders during walking and trotting phases. Our research captured these nuanced postural shifts using an inertial sensor network, yielding nine-axis motion data of riders during their performance. The accuracy of our fusion technique was validated using the optical tracking system. By analyzing the motion data, we discerned posture variations among the riders of differing expertise levels, even when executing the same riding technique. The insights from this research offer a quantifiable metric for refining equestrian training and contribute to understanding the complex dynamics of rider–horse interaction.

Multi-branch convolutional attention network for multi-sensor feature fusion in intelligent fault diagnosis of rotating machinery

Multi-sensor data fusion approaches based on deep learning are widely used for fault diagnosis of rotating machinery. Massive sensor data bring not only abundant information but also technical challenges for the fault diagnosis. The main challenge is how to discriminate the discrepancies between multi-sensor data and efficiently fuze these sensor data to improve diagnostic performance. To overcome the challenge, a novel multi-branch convolutional attention network (MBCAN) is proposed for fault diagnosis of the rotating machinery. In this method, a feature extractor with attention mechanism is established to extract fault-related features from the sensor data and reduce irrelevant noise interference hidden in the sensor data. Meanwhile, a multi-branch convolutional pathway is designed to enrich the features. Furthermore, a feature fusion module is constructed by defining an adaptive weighted fusion rule to fuze the extracted features. The performance of the MBCAN is verified on gearbox and centrifugal pump under different noise environments. The experimental results show that the proposed MBCAN has more excellent anti-noise ability and reliable diagnostic performance than other existing approaches.

Accelerating Crop Yield: Multisensor Data Fusion and Machine Learning for Agriculture Text Classification

Sensors are now used by farmers and agronomists to help them improve their operations. They use sensor data transmitted via IoT to remotely monitor their crops. Farmers today manage crops in a controlled environment to increase yields in the name of modern farming. Crop productivity, on the other hand, is influenced by the severity of the weather and disease variations. The primary objective of this paper is to present a novel Multisensor Machine-Learning Approach (MMLA) for classifying multisensor data. The fusion strategy supports high-quality data analysis in agricultural contexts for cultivation recommendations. Based on the proposed recommendation system, eight crops were classified: cotton, gram, groundnut, maize, moong, paddy, sugarcane, and wheat. Crop species were classified using three machine learning algorithms: J48 Decision Tree, Hoeffding Tree, and Random Forest. To evaluate the performance of the proposed multi-text classifier, only the top eight classes were investigated. The classifier’s performance is measured in terms of precision, recall, F-measure, MCC, ROC Area, and PRC Area class, and the results are compared with the state-of-the-art classifiers. The Random forest algorithm has the lowest error measure of RMSE at 13%, RAE at 38.67%, and RRSE at 44.21%, demonstrating effectiveness in classifying the agriculture text. Thus, the use of a multisensor data fusion approach based on crop recommendation provides greater precision in prediction, resulting in a significant increase in crop yield while also creating awareness in the condition-based environmental monitoring system.

Wearable multi-sensor data fusion approach for human activity recognition using machine learning algorithms

Wearable sensor based human activity recognition (HAR) has a broad range of applications in healthcare, fitness, smart home, and surveillance. In spite of the substantial amount of computational research on HAR, the open challenges in multi-sensor based activity recognition including complex temporal data, identifying discriminative feature vectors from multi-modal data and dimensionality reduction require considerable research attention. Hence, in this paper, using multi-resolution time-frequency analysis of received signal strength (RSS) between the wearable sensors, we present a supervised machine learning (ML) based activity recognition framework. The multi-sensor activity data acquired using the wireless sensor network (WSN) nodes and inertial-sensor embedded in a smartphone are decomposed using the discrete wavelet transform (DWT) and the empirical mode decomposition (EMD) techniques for extracting the prominent feature vector. Using the discriminative statistical features from DWT along with the entropy features from EMD, the four ML classifier models such as support vector machine (SVM), K-nearest neighbour (KNN), ensemble classifier (EC), and decision tree (DT) are trained to classify various human activities. The efficacy of the proposed design is assessed on a publicly available dataset from UCI. The experimental results assessed using the confusion matrix and parallel coordinate plot (PCP) substantiate that the proposed ML based HAR framework can achieve a maximum classification accuracy of 99.63% and is superior to several state-of-the-art ML techniques.

Multi-Sensor Data Fusion Approach for Kinematic Quantities

A theoretical framework to implement multi-sensor data fusion methods for kinematic quantities is proposed. All methods defined through the framework allow the combination of signals obtained from position, velocity and acceleration sensors addressing the same target, and improvement in the observation of the kinematics of the target. Differently from several alternative methods, the considered ones need no dynamic and/or error models to operate and can be implemented with low computational burden. In fact, they gain measurements by summing filtered versions of the heterogeneous kinematic quantities. In particular, in the case of position measurement, the use of filters with finite impulse responses, all characterized by finite gain throughout the bandwidth, in place of straightforward time-integrative operators, prevents the drift that is typically produced by the offset and low-frequency noise affecting velocity and acceleration data. A simulated scenario shows that the adopted method keeps the error in a position measurement, obtained indirectly from an accelerometer affected by an offset equal to 1 ppm on the full scale, within a few ppm of the full-scale position. If the digital output of the accelerometer undergoes a second-order time integration, instead, the measurement error would theoretically rise up to 12n(n+1) ppm in the full scale at the n-th discrete time instant. The class of methods offered by the proposed framework is therefore interesting in those applications in which the direct position measurements are characterized by poor accuracy and one has also to look at the velocity and acceleration data to improve the tracking of a target.

Latent heat flux variability and response to drought stress of black poplar: A multi-platform multi-sensor remote and proximal sensing approach to relieve the data scarcity bottleneck

High-throughput mapping of latent heat flux (λET) is critical to efforts to optimize water resources management and to accelerate forest tree breeding for improved drought tolerance. Ideally, investigation of the energy response at the tree level may promote tailored irrigation strategies and, thus, maximize crop biomass productivity. However, data availability is limited and planning experimental campaigns in the field can be highly operationally complex. To this end, a multi-platform multi-sensor observational approach is herein developed to dissect the λET signature of a black poplar (Populus nigra) breeding population (“POP6”) at the canopy level. POP6 comprised more than 4600 trees representing 503 replicated genotypes, whose parents were derived from contrasting environmental conditions. Trees were trialed in two adjacent plots where different irrigation treatments (moderate drought [mDr] and well-watered [WW]) were applied. Data collected from satellite and unmanned aerial vehicles (UAVs) remote sensing as well as from ground-based proximal sensors were integrated at consistent spatial aggregation and combined to compute the surface energy balance of the trees through a modified Priestley-Taylor method. Here, we demonstrated that λET response was significantly different between WW and mDr trees, whereby genotypes in mDr conditions exhibited larger standard deviations. Importantly, genotypes classified as drought tolerant based on the stress susceptibility index (SSI) presented λET values significantly higher than the rest of the population. This study confirmed that water limitation in mDr settings led to reduced soil moisture in the tree root zone and, thus, to lower λET. These results pave the way to breeding poplar and other bioenergy crops with this underexploited trait for higher λET. Most notably, the illustrated work demonstrates a multi-platform multi-sensor data fusion approach to tackle the global challenge of monitoring landscape-scale ecosystem processes at fine resolution.

Urban tree species classification using UAV-based multi-sensor data fusion and machine learning

ABSTRACT Urban tree species classification is a challenging task due to spectral and spatial diversity within an urban environment. Unmanned aerial vehicle (UAV) platforms and small-sensor technology are rapidly evolving, presenting the opportunity for a comprehensive multi-sensor remote sensing approach for urban tree classification. The objectives of this paper were to develop a multi-sensor data fusion technique for urban tree species classification with limited training samples. To that end, UAV-based multispectral, hyperspectral, LiDAR, and thermal infrared imagery was collected over an urban study area to test the classification of 96 individual trees from seven species using a data fusion approach. Two supervised machine learning classifiers, Random Forest (RF) and Support Vector Machine (SVM), were investigated for their capacity to incorporate highly dimensional and diverse datasets from multiple sensors. When using hyperspectral-derived spectral features with RF, the fusion of all features extracted from all sensor types (spectral, LiDAR, thermal) achieved the highest overall classification accuracy (OA) of 83.3% and kappa of 0.80. Despite multispectral reflectance bands alone producing significantly lower OA of 55.2% compared to 70.2% with minimum noise fraction (MNF) transformed hyperspectral reflectance bands, the full dataset combination (spectral, LiDAR, thermal) with multispectral-derived spectral features achieved an OA of 81.3% and kappa of 0.77 using RF. Comparison of the features extracted from individual sensors for each species highlight the ability for each sensor to identify distinguishable characteristics between species to aid classification. The results demonstrate the potential for a high-resolution multi-sensor data fusion approach for classifying individual trees by species in a complex urban environment under limited sampling requirements.

Studying drought-induced forest mortality using high spatiotemporal resolution evapotranspiration data from thermal satellite imaging

Drought can have pervasive and wide-spread impacts to forest health, as evidenced in several severe events occurring over the recent decades. Extensive forest die-off due to drought can impair the ecological functioning of forests, impacting habitat, water yield and quality from forested lands, and altering forest fire dynamics and intensity. Satellite remote sensing provides an effective means for detecting and monitoring spatial patterns of forest mortality over large areas, exploiting free and open long-term image archives available at a range in spatial and temporal resolutions. While remotely sensed surface reflectances and vegetation indices have been widely used to study optical response of forest canopies to drought events, retrievals of evapotranspiration (ET) derived from thermal satellite imagery – particularly at resolutions approaching crown scale - can provide insights into cumulative tree stresses that can incite disease and trigger mortality. In this study, we applied a multi-sensor satellite data fusion approach to estimate daily 30-m resolution ET and an associated Evaporative Stress Index (ESI) to study drought-induced mortality in a temperate forest at the Missouri Ozark AmeriFlux (MOFLUX) site, located in the central United States. The study covered the period from 2010 to 2014, including an exceptional drought year of 2012. Modeled ET agreed well with eddy flux measurements from the MOFLUX tower, with average monthly relative errors of 15%. Plot-scale ESI, describing temporal anomalies in the ratio of actual-to-reference ET, was used as an index of relative forest health to investigate relationships between forest mortality and drought severity. ESI showed good agreement with observed predawn leaf water potential, especially during the drought year. Furthermore, plot-scale ESI was also correlated with the subsequent year's tree mortality, suggesting the importance of considering the forest health condition prior to drought when studying drought-induced forest impacts. This study demonstrates the utility of multi-year ET remote sensing data at the stand or plot scale as an indicator of forest health and as a predictor of future mortality due to drought.

Actual Evapotranspiration from UAV Images: A Multi-Sensor Data Fusion Approach

Multispectral imaging using Unmanned Aerial Vehicles (UAVs) has changed the pace of precision agriculture. Actual evapotranspiration (ETa) from the very high spatial resolution of UAV images over agricultural fields can help farmers increase their production at the lowest possible cost. ETa estimation using UAVs requires a full package of sensors capturing the visible/infrared and thermal portions of the spectrum. Therefore, this study focused on a multi-sensor data fusion approach for ETa estimation (MSDF-ET) independent of thermal sensors. The method was based on sharpening the Landsat 8 pixels to UAV spatial resolution by considering the relationship between reference ETa fraction (ETrf) and a Vegetation Index (VI). Four Landsat 8 images were processed to calculate ETa of three UAV images over three almond fields. Two flights coincided with the overpasses and one was in between two consecutive Landsat 8 images. ETrf was chosen instead of ETa to interpolate the Landsat 8-derived ETrf images to obtain an ETrf image on the UAV flight. ETrf was defined as the ratio of ETa to grass reference evapotranspiration (ETr), and the VIs tested in this study included the Normalized Difference Vegetation Index (NDVI), Soil Adjusted Vegetation Index (SAVI), Enhanced Vegetation Index (EVI), Normalized Difference Water Index (NDWI), and Land Surface Water Index (LSWI). NDVI performed better under the study conditions. The MSDF-ET-derived ETa showed strong correlations against measured ETa, UAV- and Landsat 8-based METRIC ETa. Also, visual comparison of the MSDF-ET ETa maps was indicative of a promising performance of the method. In sum, the resulting ETa had a higher spatial resolution compared with thermal-based ETa without the need for the Albedo and hot/cold pixels selection procedure. However, wet soils were poorly detected, and in cases of continuous cloudy Landsat pixels the long interval between the images may cause biases in ETa estimation from the MSDF-ET method. Generally, the MSDF-ET method reduces the need for very high resolution thermal information from the ground, and the calculations can be conducted on a moderate-performance computer system because the main image processing is applied on Landsat images with coarser spatial resolutions.

Fusing simulated GEDI, ICESat-2 and NISAR data for regional aboveground biomass mapping

Accurate mapping of forest aboveground biomass (AGB) is critical for better understanding the role of forests in the global carbon cycle. NASA's current GEDI and ICESat-2 missions as well as the upcoming NISAR mission will collect synergistic data with different coverage and sensitivity to AGB. In this study, we present a multi-sensor data fusion approach leveraging the strength of each mission to produce wall-to-wall AGB maps that are more accurate and spatially comprehensive than what is achievable with any one sensor alone. Specifically, we calibrate a regional L-band radar AGB model using the sparse, simulated spaceborne lidar AGB estimates. We assess our data fusion framework using simulations of GEDI, ICESat-2 and NISAR data from airborne laser scanning (ALS) and UAVSAR data acquired over the temperate high AGB forest and complex terrain in Sonoma County, California, USA. For ICESat-2 and GEDI missions, we simulate two years of data coverage and AGB at footprint level are estimated using realistic AGB models. We compare the performance of our fusion framework when different combinations of the sparse simulated GEDI and ICEsat-2 AGB estimates are used to calibrate our regional L-band AGB models. In addition, we test our framework at Sonoma using (a) 1-ha square grid cells and (b) similarly sized irregularly shaped objects. We demonstrate that the estimated mean AGB across Sonoma is more accurately estimated using our fusion framework than using GEDI or ICESat-2 mission data alone, either with a regular grid or with irregular segments as mapping units. This research highlights methodological opportunities for fusing new and upcoming active remote sensing data streams toward improved AGB mapping through data fusion.

Prediction of human odour assessments based on hedonic tone method using instrument measurements and multi-sensor data fusion integrated neural networks

A Cyranose 320 (eNose) and a Fast Gas Chromatograph (CG) analyser (zNose™) were used to measure the headspace odour of solid samples from dairy operations. The measurements of both sensors were trained by Levenberg–Marquardt Back-propagation Neural Network (LMBNN) to match human assessments. A trained human panel was used to assess the odours based on hedonic tone method and provide the model targets. A multi-sensor data fusion approach was developed and applied to integrate the eNose and zNose readings for higher predictive accuracy compared to each sensor alone. Principle Component Analysis, Forward Selection, and Gamma Test were applied to reduce the model input dimensions. Measurement fusion models and information fusion model approaches were applied. The information fusion prediction models were shown to be more accurate than all other models, including single instrument models. The information fusion model based on eNose with Gamma Test data reduction + zNose showed the best results of all cases in validation mean square error (0.34 odour units), R value (0.92), probability of the prediction falling within 10% of the target (96%), and probability of the prediction falling within 5% of the target (63%).

A Dynamic multi-sensor data fusion approach based on evidence theory and WOWA operator

Multi-sensor data fusion (MSDF) problems have attracted widespread attention recently. However, it is still an open issue about how to make the fusion process effectively even if the collected data conflict due to several unpredictable reasons. Moreover, most existing approaches mainly concentrated on the distinction of evidence sources, which cannot well consider the feature of individual belief degree and the associated preference of decision-makers. To address such an issue, a dynamic MSDF method based on evidence theory and weighted ordered weighted averaging (WOWA) operator is proposed in this study. A numerical example is analyzed to demonstrate its whole calculation procedure. Two simulation experiments, composed of a motor rotor fault diagnosis and an insulator string target recognition application, are also mentioned to illustrate its effectiveness and applied value. The results show that the proposed methodology can enhance the fusion accuracy in the constrained scenarios with the consideration of preference relation.

A Multisensor Data Fusion Approach for Predicting Consumer Acceptance of Food Products.

Sensory experiences play an important role in consumer response, purchase decision, and fidelity towards food products. Consumer studies when launching new food products must incorporate physiological response assessment to be more precise and, thus, increase their chances of success in the market. This paper introduces a novel sensory analysis system that incorporates facial emotion recognition (FER), galvanic skin response (GSR), and cardiac pulse to determine consumer acceptance of food samples. Taste and smell experiments were conducted with 120 participants recording facial images, biometric signals, and reported liking when trying a set of pleasant and unpleasant flavors and odors. Data fusion and analysis by machine learning models allow predicting the acceptance elicited by the samples. Results confirm that FER alone is not sufficient to determine consumers’ acceptance. However, when combined with GSR and, to a lesser extent, with pulse signals, acceptance prediction can be improved. This research targets predicting consumer’s acceptance without the continuous use of liking scores. In addition, the findings of this work may be used to explore the relationships between facial expressions and physiological reactions for non-rational decision-making when interacting with new food products.

Using High-Spatiotemporal Thermal Satellite ET Retrievals for Operational Water Use and Stress Monitoring in a California Vineyard

In viticulture, deficit irrigation strategies are often implemented to control vine canopy growth and to impose stress at critical stages of vine growth to improve wine grape quality. To support deficit irrigation scheduling, remote sensing technologies can be employed in the mapping of evapotranspiration (ET) at the field to sub-field scales, quantifying time-varying vineyard water requirements and actual water use. In the current study, we investigate the utility of ET maps derived from thermal infrared satellite imagery over a vineyard in the Central Valley of California equipped with a variable rate drip irrigation (VRDI) system which enables differential water applications at the 30 × 30 m scale. To support irrigation management at that scale, we utilized a thermal-based multi-sensor data fusion approach to generate weekly total actual ET (ETa) estimates at 30 m spatial resolution, coinciding with the resolution of the Landsat reflectance bands. Crop water requirements (ETc) were defined with a vegetative index (VI)-based approach. To test capacity to capture stress signals, the vineyard was sub-divided into four blocks with different irrigation management strategies and goals, inducing varying degrees of stress during the growing season. Results indicate derived weekly total ET from the thermal-based data fusion approach match well with observations. The thermal-based method was also able to capture the spatial heterogeneity in ET over the vineyard due to a water stress event imposed on two of the four vineyard blocks. This transient stress event was not reflected in the VI-based ETc estimate, highlighting the value of thermal band imaging. While the data fusion system provided valuable information, latency in current satellite data availability, particularly from Landsat, impacts operational applications over the course of a growing season.

Health risk assessment and decision-making for patient monitoring and decision-support using Wireless Body Sensor Networks

This paper proposes a generalized multi-sensor fusion approach and a health risk assessment and decision-making (Health-RAD) algorithm for continuous and remote patient monitoring purposes using a Wireless Body Sensor Network (WBSN). Health-RAD determines the patient’s health condition severity level routinely and each time a critical issue is detected based on vital signs scores. Hence, a continuous health assessment and a monitoring of the improvement or the deterioration of the state of the patient is ensured. The severity level is represented by a risk variable whose values range between 0 and 1. The higher the risk value, the more critical the patient’s health condition is and the more it requires medical attention. Moreover, we calculate the score of a vital sign using its past and current value, thus assessing its status based on its evolution during a period of time and not only on sudden deviations. We propose a generalized multi-sensor data fusion approach regardless of the number of monitored vital signs. The latter is employed by Health-RAD to find the severity level of the patient’s health condition based on his/her vital signs scores. It is based on a fuzzy inference system (FIS) and early warning score systems (EWS). This approach is tested with a previously proposed energy-efficient data collection approach, thus forming a complete framework. The proposed approach is evaluated on real healthcare datasets and the results are compared with another approach from the literature in terms of data reduction, energy consumption, risk assessment of vital signs, the patient’s health risk level determination and accuracy. The results show that both approaches have coherently assessed the health condition of different Intensive Care Unit (ICU) patients. Yet, our proposed approach overcomes the other approach in terms of energy consumption (around 86% less energy consumption) and data reduction (around 70% for sensing and more than 90% for transmission). Additionally, contrary to our proposed framework, the approach taken from the literature requires an offline model building and depends on available patient datasets.

A Novel Evidence Theory and Fuzzy Preference Approach-Based Multi-Sensor Data Fusion Technique for Fault Diagnosis.

The multi-sensor data fusion technique plays a significant role in fault diagnosis and in a variety of such applications, and the Dempster–Shafer evidence theory is employed to improve the system performance; whereas, it may generate a counter-intuitive result when the pieces of evidence highly conflict with each other. To handle this problem, a novel multi-sensor data fusion approach on the basis of the distance of evidence, belief entropy and fuzzy preference relation analysis is proposed. A function of evidence distance is first leveraged to measure the conflict degree among the pieces of evidence; thus, the support degree can be obtained to represent the reliability of the evidence. Next, the uncertainty of each piece of evidence is measured by means of the belief entropy. Based on the quantitative uncertainty measured above, the fuzzy preference relations are applied to represent the relative credibility preference of the evidence. Afterwards, the support degree of each piece of evidence is adjusted by taking advantage of the relative credibility preference of the evidence that can be utilized to generate an appropriate weight with respect to each piece of evidence. Finally, the modified weights of the evidence are adopted to adjust the bodies of the evidence in the advance of utilizing Dempster’s combination rule. A numerical example and a practical application in fault diagnosis are used as illustrations to demonstrate that the proposal is reasonable and efficient in the management of conflict and fault diagnosis.

A Multi-Sensor Data Fusion Approach for Atrial Hypertrophy Disease Diagnosis Based on Characterized Support Vector Hyperspheres.

Disease diagnosis can be performed based on fusing the data acquired by multiple medical sensors from patients, and it is a crucial task in sensor-based e-healthcare systems. However, it is a challenging problem that there are few effective diagnosis methods based on sensor data fusion for atrial hypertrophy disease. In this article, we propose a novel multi-sensor data fusion method for atrial hypertrophy diagnosis, namely, characterized support vector hyperspheres (CSVH). Instead of constructing a hyperplane, as a traditional support vector machine does, the proposed method generates “hyperspheres” to collect the discriminative medical information, since a hypersphere is more powerful for data description than a hyperplane. In detail, CSVH constructs two characterized hyperspheres for the classes of patient and healthy subject, respectively. The hypersphere for the patient class is developed in a weighted version so as to take the diversity of patient instances into consideration. The hypersphere for the class of healthy people keeps furthest away from the patient class in order to achieve maximum separation from the patient class. A query is labelled by membership functions defined based on the two hyperspheres. If the query is rejected by the two classes, the angle information of the query to outliers and overlapping-region data is investigated to provide the final decision. The experimental results illustrate that the proposed method achieves the highest diagnosis accuracy among the state-of-the-art methods.

Extending lifetime of wireless sensor networks using multi-sensor data fusion

In this paper a multi-sensor data fusion approach for wireless sensor network based on bayesian methods and ant colony optimization techniques has been proposed. In this method, each node is equipped with multiple sensors (i.e., temperature and humidity). Use of more than one sensor provides additional information about the environmental conditions. The data fusion approach based on the competitive-type hierarchical processing is considered for experimentation. Initially the data are collected by the sensors placed in the sensing fields and then the data fusion probabilities are computed on the sensed data. In this proposed methodology, the collected temperature tand humidity data are processed by multi-sensor data fusion techniques, which help in decreasing the energy consumption as well as communication cost by fusing the redundant data. The multiple data fusion process improves the reliability and accuracy of the sensed information and simultaneously saves energy, which was our primary objective. The proposed algorithms were simulated using Matlab . The executions of proposed arnd low-energy adaptive clustering hierarchy algorithms were carried out and the results show that the proposed algorithms could efficiently reduce the use of energy and were able to save more energy, thus increasing the overall network lifetime.

A Weighted Belief Entropy-Based Uncertainty Measure for Multi-Sensor Data Fusion.

In real applications, how to measure the uncertain degree of sensor reports before applying sensor data fusion is a big challenge. In this paper, in the frame of Dempster–Shafer evidence theory, a weighted belief entropy based on Deng entropy is proposed to quantify the uncertainty of uncertain information. The weight of the proposed belief entropy is based on the relative scale of a proposition with regard to the frame of discernment (FOD). Compared with some other uncertainty measures in Dempster–Shafer framework, the new measure focuses on the uncertain information represented by not only the mass function, but also the scale of the FOD, which means less information loss in information processing. After that, a new multi-sensor data fusion approach based on the weighted belief entropy is proposed. The rationality and superiority of the new multi-sensor data fusion method is verified according to an experiment on artificial data and an application on fault diagnosis of a motor rotor.

Multi Sensor Data Fusion based Gear Fault Diagnosis using Complete Ensemble Empirical Mode Decomposition with Adaptive Noise

Objective: To develop a methodology based on multi sensor data fusion approach which combined vibration and sound signals to identify the gearbox faults using Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN). Method: The vibration and sound signals acquired from a gearbox are decomposed into a number of IMFs using CEEMDAN. Best IMF is selected based on proposed fault index and the statistical parameters are extracted from the IMFs of vibration and sound signals for different simulated faults. Principal Component Analysis (PCA) is implemented in order to select the best features. K-nearest neighbor classifier is used to demonstrate the classification accuracy. Findings: Initially statistical parameters were extracted for raw vibration and sound signals in order to obtain the fault severity. But due to uneven trend these were failed to reveal the fault information with higher accuracy. CEEMDAN based feature sets provide good diagnosis results due to its capability to decompose signal into different higher to lower frequency modes called IMFs. Hence, it is concluded that the proposed method has the ability to extract the gearbox fault characteristic and diagnose the severity of fault. Sensitive IMF selection is a paramount to swift the overall performance of a diagnosis system. Improvement: The results show that the data fusion approach and combination of CEEMDAN and PCA techniques employs the advantages traits of one or the other technique to generate overall improvement in diagnosing severity of local faults.