Analysis Of Multi-sensor Data Research Articles

Development of a multisensor biologging collar and analytical techniques to describe high-resolution spatial behavior in free-ranging terrestrial mammals.

Biologging has proven to be a powerful approach to investigate diverse questions related to movement ecology across a range of spatiotemporal scales and increasingly relies on multidisciplinary expertise. However, the variety of animal-borne equipment, coupled with little consensus regarding analytical approaches to interpret large, complex data sets presents challenges and makes comparison between studies and study species difficult. Here, we present a combined hardware and analytical approach for standardizing the collection, analysis, and interpretation of multisensor biologging data. Here, we present (i) a custom-designed integrated multisensor collar (IMSC), which was field tested on 71 free-ranging wild boar (Sus scrofa) over 2 years; (ii) a machine learning behavioral classifier capable of identifying six behaviors in free-roaming boar, validated across individuals equipped with differing collar designs; and (iii) laboratory and field-based calibration and accuracy assessments of animal magnetic heading measurements derived from raw magnetometer data. The IMSC capacity and durability exceeded expectations, with a 94% collar recovery rate and a 75% cumulative data recording success rate, with a maximum logging duration of 421 days. The behavioral classifier had an overall accuracy of 85% in identifying the six behavioral classes when tested on multiple collar designs and improved to 90% when tested on data exclusively from the IMSC. Both laboratory and field tests of magnetic compass headings were in precise agreement with expectations, with overall median magnetic headings deviating from ground truth observations by 1.7° and 0°, respectively. Although multisensor equipment and sophisticated analyses are now commonplace in biologging studies, the IMSC hardware and analytical framework presented here provide a valuable tool for biologging researchers and will facilitate standardization of biologging data across studies. In addition, we highlight the potential of additional analyses available using this framework that can be adapted for use in future studies on terrestrial mammals.

A Robust Deep Feature Extraction Method for Human Activity Recognition Using a Wavelet Based Spectral Visualisation Technique.

Human Activity Recognition (HAR), alongside Ambient Assisted Living (AAL), are integral components of smart homes, sports, surveillance, and investigation activities. To recognize daily activities, researchers are focusing on lightweight, cost-effective, wearable sensor-based technologies as traditional vision-based technologies lack elderly privacy, a fundamental right of every human. However, it is challenging to extract potential features from 1D multi-sensor data. Thus, this research focuses on extracting distinguishable patterns and deep features from spectral images by time-frequency-domain analysis of 1D multi-sensor data. Wearable sensor data, particularly accelerator and gyroscope data, act as input signals of different daily activities, and provide potential information using time-frequency analysis. This potential time series information is mapped into spectral images through a process called use of 'scalograms', derived from the continuous wavelet transform. The deep activity features are extracted from the activity image using deep learning models such as CNN, MobileNetV3, ResNet, and GoogleNet and subsequently classified using a conventional classifier. To validate the proposed model, SisFall and PAMAP2 benchmark datasets are used. Based on the experimental results, this proposed model shows the optimal performance for activity recognition obtaining an accuracy of 98.4% for SisFall and 98.1% for PAMAP2, using Morlet as the mother wavelet with ResNet-101 and a softmax classifier, and outperforms state-of-the-art algorithms.

Multi-sensor Data Analysis for Aerial Image Semantic Segmentation and Vectorization

Abstract. One of the urgent and constantly in demand problems is updating maps. Maps, representing geo-information in vector form, have undoubted advantages in compactness and ”readability” compared to aerial photographs. The issue of maps actuality is critically important for rational urban planning, precision farming, the relevance of the cadastre and other geospatial applications. Various sources of data are used for maps updating, with aerial imagery being the main and rich source of information. Automatic processing of aerial photographs makes it possible to efficiently extract vector information, providing operational monitoring and accounting for changes that have appeared. The presented study addresses the problem of multi sensor information fusion in order to obtain accurate vector information. We use aerial images as a main data source and additionally the data of laser scanning and ground survey to increase performance of automatic image semantic segmentation and vectorization. The proposed framework is demonstrated on the task of forest monitoring.

Time Series Analysis of Multisensor Data for Precision Viticulture—Assessing Microscale Variations in Plant Development with Respect to Irrigation and Topography

Time Series Analysis of Multisensor Data for Precision Viticulture—Assessing Microscale Variations in Plant Development with Respect to Irrigation and Topography

PREDICTING FIRE ALARMS USING MULTI SENSOR DATA: A BINARY CLASSIFICATION APPROACH

Fires pose significant threats to human life, property, and the environment. Early detection of fire incidents is crucial to prevent extensive damage and to ensure the safety of occupants. Traditional fire alarm systems typically rely on a single type of sensor, such as smoke detectors or heat sensors, to detect specific fire indicators. These systems operate based on predefined thresholds and triggers. However, they can be prone to false alarms triggered by non-fire-related events (e.g., cooking fumes or dust) and may not provide early warning signs in certain scenarios. To address these limitations, researchers and engineers have turned to advanced technologies, such as multi-sensor data analysis and machine learning algorithms, to develop more reliable and efficient fire alarm prediction systems. On the other hand, the need for a more robust and accurate fire alarm prediction system stems from the shortcomings of traditional methods. False alarms not only lead to wasted resources but also desensitize occupants, potentially leading them to ignore genuine alarms. Additionally, a delayed response to a fire incident can result in severe consequences, making it essential to develop an intelligent system that can effectively and timely predict fire events. Therefore, this work presents the utilization of multi-sensor data and binary classification to develop a more reliable fire alarm prediction system. The experiments are conducted using a dataset collected from various sensor inputs, including air temperature, humidity, CO2 concentration, molecular hydrogen, ethanol gas, and air pressure etc. Then applied binary classification algorithm to learn patterns from the data and classify fire-related events accurately. The results showed promising improvements in prediction accuracy, reduced false alarm rates, and early detection of fire incidents.

Research on transformer fire alarm threshold based on multi-sensor data analysis

Based on literature, on-site investigation and accident case analysis, this paper selects multi-sensor for transformer fire alarm, builds the transformer fire test platform, and carries out the transformer fire test and interference test to obtain the training and test data of neural network data fusion. This paper determines the fire alarm threshold of multi-sensor signal. It provides a benchmark for comparing the alarm results of multi-sensor data simulation fusion with those of single sensor. This paper studies the preprocessing multi-sensor data, compiles the corresponding algorithm, and obtains the multi-sensor preprocessed data after dimension reduction and noise reduction. It provides data support for multi-sensor data fusion.

A distributed land cover classification of FP and CP SAR observation using MapReduce-based multi-layer perceptron algorithm over the Mumbai mangrove region of India

ABSTRACT Globally, the rapid loss of mangrove forests often creates long-lasting environmental damage alongside coastlines, mudflats, and river banks. All-weather, physical monitoring of such areas is almost impossible because of inaccessibility to swampy areas and a hostile substrate. Subsequently, conventional field surveys are relatively unavailable to monitor human encroachment in coastal areas like the Mumbai mangrove region of India. In this context, the polarimetric synthetic aperture radar (PolSAR) remote sensing tool becomes a potential candidate for mangrove conservation and management. Since the Mumbai mangrove region of India exists over the extensive land cover, the large-scale classification can articulate the continuous encroachment because of the location of human settlements or activities across this metropolitan coastal city. For this, the traditional algorithms need the essential improvisation to apply for large-scale data analysis, aiming simultaneously at getting the highest decisive and time efficiency. Here, we introduce a shallow learning model of MapReduce-based Multi-Layer Perception (MLP) algorithm to classify the hybrid compact polarimetric (CP) and fully polarimetric (FP) feature space. Even though a shallow learning model of the automated method is easily scalable, it requires a distinctive shallow learnable feature set for better land cover classification. In this effort, this paper investigates the efficacy of derived feature space compared to direct polarimetric measurements of sensors and shows that the shallow learnable feature set is more effective with both CP & FP observations. Simultaneously, the relevancy of the proposed distributed model of MLP is also justified compared to distributed extreme learning machine (DELM) algorithm and provides a practically implementable scaled-MLP algorithm of shallow learning model. Ultimately, this paper comes up with a better polarimetric signature of land types for both the CP and FP datasets, which can be used in an alternative manner as per data availability for a multi-sensor data analysis.

Software platform for imaging diagnostic exploitation applied to edge plasma physics and real-time PFC monitoring

In current tokamaks, imaging diagnostics (visible/infrared) have become increasingly used for real-time control of the temperature of the Plasma Facing Components (PFC) and for off-line physical analysis. Developing a full acquisition and control system based on a network of multiple cameras is a complex task requiring advanced software tools. Furthermore, such systems produce a large quantity of data encouraging end users to share tools and codes for data access and analysis.In this paper, we introduce a software platform, named Thermavip, dedicated to these tasks. The main contribution of this software is to gather under the same framework different functionalities for (1) firm real-time video acquisition and PFC monitoring, (2) offline data access and display, (3) signal and video processing applied to plasma physics, (4) video annotation and thermal events database management. This software platform is widely used on the WEST (W Environment in Steady-state Tokamak) tokamak (France) for offline study of infrared (IR) and almost all WEST diagnostics data, and online display of IR videos and temperature time traces for critical components. Thermavip platform is also used on the HADES (High heAt LoaD tESt) facility (France) and the EAST (Experimental Advanced Superconducting Tokamak) tokamak (China) for offline signal analysis, and on the Wendelstein 7-X (W7-X) stellarator (Germany) for the IR diagnostic acquisition, PFC real-time (RT) monitoring, online display and offline video analysis. In addition, Thermavip provides a full featured video annotation tool used on both WEST and W7-X to build and manage thermal events datasets. These annotations are used as input dataset to train deep-learning models, like Region Based Convolutional Neural Networks, for automatic detection and classification of thermal events based on IR movies.The Thermavip framework is composed of a unique C++ Software Development Kit (SDK) providing high-level classes for offline multi-sensor data analysis, firm real-time processing and online visualization of these sensor data. Its strength comes from its unique component block architecture, allowing to build multicore and distributed processing pipelines for both offline and firm real-time applications. Thermavip architecture relies on a versatile plugin mechanism to extend its functionalities using C++ code or Python scripts, facilitating the integration on new machines.

A Dual-Branch Deep Learning Architecture for Multisensor and Multitemporal Remote Sensing Semantic Segmentation

Multi-sensor data analysis allows exploiting heterogeneous data regularly acquired by the many available Remote Sensing (RS) systems. Machine- and deep-learning methods use the information of heterogeneous sources to improve the results obtained by using single-source data. However, the State-of-the-Art (SoA) methods analyze either the multi-scale information of multi-sensor multi-resolution images or the time component of image time series. We propose a supervised Deep-Learning (DL) classification method that jointly performs a multi-scale and multi-temporal analysis of RS multi-temporal images acquired by different sensors. The proposed method processes Very-High-Resolution (VHR) images using a Residual Network (ResNet) with a wide receptive field that handles geometrical details and multi-temporal High-Resolution (HR) image using a 3D Convolutional Neural Network (3D-CNN) that analyzes both the spatial and temporal information. The multi-scale and multi-temporal features are processed together in a decoder to retrieve a land-cover map. We tested the proposed method on two multi-sensor and multi-temporal datasets. One is composed of VHR orthophotos and Sentinel-2 multi-temporal images for pasture classification, and another is composed of VHR orthophotos and Sentinel-1 multi-temporal images. Results proved the effectiveness of the proposed classification method.

Satellite Multi-Sensor Data Analysis of Unusually Strong Polar Lows over the Chukchi and Beaufort Seas in October 2017

Polar lows (PLs) are intense mesoscale weather systems that often cause severe storm winds in the Nordic Seas but were considered as being exceedingly rare in the Pacific Arctic region before sea ice decline. Here, we explore four PLs observed on 18–22 October 2017 in the Chukchi and Beaufort Seas—an area with an exceptionally sparse observation network. The study is based on the combined use of the satellite microwave measurements, as well as infrared imagery, the ERA5, MERRA-2 and NCEP-CFSv2 reanalysis data sets. An unusually strong PLs pair developed near the marginal ice zone during a marine-cold air outbreak in anomalously low sea ice extent conditions. PLs pair moved southward as a mesocyclonic system called the “merry-go-round”, under the upper-level tropospheric vortex with a cold core. Multi-sensor satellite measurements show that, in the mature stage, a PL pair had near-surface wind speeds (W) close to hurricane force—over 30 m/s. Comparison analysis of W distributions within the strongest PL showed that all reanalysis data sets reasonably reproduce the PL median wind speed but underestimate its extreme values by 15–23%. The reanalysis data sets detected only two PLs with horizontal scales of over 220 km. Tracks of identified PLs for all data sets are in good agreement with the ones obtained from satellite images capturing the main features of the mesoscale weather system propagation. For the track of the strongest PL event, ERA5 exhibited the most accordance with satellite observations with a tracking error of 50–60 km.

Multi-Sensor Data Analysis of an Intense Weather Event: The July 2021 Lake Como Case Study

A comprehensive analysis of the July 2021 event that occurred on Lake Como (Italy), during which heavy hailstorms and floods affected the surroundings of Lake, is presented. The study provides a detailed analysis of the event using different observation sources currently available. The employed techniques include both conventional (rain gauges, radar, atmospheric sounding) and non-conventional (satellite-based Earth observation products, GNSS, and lightning detection network) observations for hydro-meteorological analysis. The study is split in three main topics: event description by satellite-based observations; long-term analysis by the ERA5 model and ASCAT soil water index; and short-term analysis by lightning data, GNSS delays and radar-VIL. The added value of the work is the near-real-time analysis of some of the datasets used, which opens up the potential for use in alerting systems, showing considerable application possibilities in NWP modeling, where it can also be useful for the implementation of early warning systems. The results highlight the validity of the different techniques and the consistency among the observations. This result, therefore, leads to the conclusion that a joint use of the innovative techniques with the operational ones can bring reliability in the description of events.

A Novel Multi-Sensor Data-Driven Approach to Source Term Estimation of Hazardous Gas Leakages in the Chemical Industry

Source term estimation (STE) is crucial for understanding and addressing hazardous gas leakages in the chemical industry. Most existing methods basically use an atmospheric transport and dispersion (ATD) model to predict the concentrations of hazardous gas leakages from different possible sources, compare the predicted results with multi-sensor data, and use the deviations to search and derive information on the real sources of leakages. Although performing well in principle, complicated computations and the associated computer time often make these methods difficult to apply in real time. Recently, many machine learning methods have also been proposed for the purpose of STE. The idea is to build offline a machine-learning-based STE model using data generated with a high-fidelity ATD model and then apply the machine learning model to multi-sensor data to perform STE in real time. The key to the success of a machine-learning-based STE is that the machine-learning-based STE model has to cover all possible scenarios of concern, which is often difficult in practice because of unpredictable environmental conditions and the inherent robust problems with many supervised machine learning methods. In order to address challenges with the existing STE methods, in the present study, a novel multi-sensor data-driven approach to STE of hazardous gas leakages is proposed. The basic idea is to establish a multi-sensor data-driven STE model from historical multi-sensor observations that cover the situations known as the independent hazardous-gas-leakage scenarios (IHGLSs) in a chemical industry park of concern. Then the established STE model is applied to online process multi-sensor data and perform STE for the chemical industry park in real time. The new approach is based on a rigorous analysis of the relationship between multi-sensor data and sources of hazardous gas leakages and derived using advanced data science, including unsupervised multi-sensor data clustering and analysis. As an example of demonstration, the proposed approach is applied to perform STE for hazardous gas-leakage scenarios wherein a Gaussian plume model can be used to describe the atmospheric transport and dispersion. Because of no need of ATD-model-based online optimization and supervised machine learning, the new approach can potentially overcome many problems with existing methods and enable STE to be literally applied in engineering practice.

Data fusion system for monitoring water quality: Application to chlorophyll-a in Baltic sea coast

We present an operational system for multi-sensor data fusion implemented at the Finnish Environment Institute. The system uses Ensemble Kalman filter and smoother algorithms, which are often used for probabilistic analysis of multi-sensor data. Uncertainty and spatial and temporal correlations present in the available observation data are accounted for to obtain accurate and realistic results. To test the data fusion system, daily chlorophyll-a concentration has been modelled across northern shoreline of Gulf of Finland over the period of August 1st – October 31st 2011. Chlorophyll-a data from routine monitoring stations, ferrybox measurements, and data derived from Medium Resolution Imaging Spectrometer (MERIS) instrument on board the ENVISAT satellite has been used as input. The data fusion system demonstrates the use of existing and well-known Ensemble Kalman filtering and smoothing methods for improving water quality monitoring programs and for ensuring compliance with ecological standards.

Assessing the performance of machine learning algorithms for soil salinity mapping in Google Earth Engine platform using Sentinel-2A and Landsat-8 OLI data

• Earth Engine platform facilitated the analysis of multi-sensor data for soil salinity mapping. • Landsat- 8 and Sentinel-2A derived soil salinity maps were compared. • Machine learning regression methods provided good results to map soil salinity. Soil salinization caused by natural and anthropogenic factors is an important environmental hazard especially in arid and semi-arid regions of the world. Accumulation of salts in the soil is a major threat to crop production and global agriculture; therefore, rapid and precise detection of salt-affected lands is highly critical for preserving soil sustainability and supporting food production. Advancement in remote sensing techniques and machine-learning algorithms has started to contribute to fast and large-scale monitoring and mapping of soil salinization throughout the world. This paper aims to analyze the performance of three different machine-learning algorithms to map soil salinity using Landsat-8 OLI, Sentinel-2A satellite images, and ground-based electrical conductivity (EC) measurements with the aid of Google Earth Engine (GEE) platform. Classification and regression trees (CART), random forest (RF), and support vector regression (SVR) methods are implemented to create a correlation between ground measurements and satellite-derived environmental variables or spectral indices. After selecting the optimum five variables including wetness band, three soil salinity indices, and one vegetation index, soil salinity maps are generated in three machine-learning algorithms. The output soil salinity map in RF algorithm demonstrated the most reliable spatial distribution of various soil salinity classes in the selected study area. Despite CART provided slightly better prediction of soil salinity with R-squared (R 2 ) of 0.98 for Sentinel-2A data, and 0.96 for Landsat-8 OLI data in comparison with accuracy results of RF technique with R 2 of 0.96 for Sentinel-2A data and 0.94 for Landsat-8 OLI data, the output map of RF model estimated more reliable salinity levels in salt crusts, agricultural lands, drainage areas, and swamps. The corresponding result highly matched with visual interpretation. Soil salinity maps derived from SVR algorithms by using various combinations of input variables displayed relatively poor estimation of soil EC values compared to the other two methods.

Multi-Sensor fusion and data analysis for operating conditions of low power transmission lines

The research of the safety distance among drones and cable attracted wide attention with the rising deployment of unmanned aerial vehicle (UAV) to check for high-voltage overhead power systems. To increase the inspection operation's dependability and guarantee a safe and stable functioning of the transmission grid and inspections equipment, it is essential to determine the safety distance among the UAV and the driveway. Due to UAV patrol safety distances from overhead power lines, it is difficult to offer precise surfing information because of the lack of quantitative assistance. The author of the relevant study uses multi-sensor fusion data analysis (MFDA-LPTL). There have been discussions on the properties of large data sets, smart networks, and gigantic data sets before the emergence of low-cost power transmission lines and the benefits they may give. For example, using the adaptive weighted fusional method, which combines first-level data on homogenous sensor input based on critical UAV-influencing factors such as maximum inspection speed, wind speed, positioning error, and drone size, can help achieve this goal. As a secondary benefit, the theory makes use of more robust evidence than before. However, the use of big data analytics in present smart grids must be expanded due to numerous challenges, such as the need for new technologies and increased public awareness. The experimental analysis shows that the proposed MFDA-LPTL model increases the performance ratio of 98.9%, efficiency ratio of 97.2%, reduces the detection failure analysis of 10.2%, processing time of 7.8%, positioning error rate of 12.3% compared to other existing methods.

Data-driven thinking for measuring the human experience in the built environment

This article introduces a methodology to implement Data-driven Thinking in the context of urban design. We present the results of a case study based on a 7-day workshop with 10 participants with landscape design and architecture background. The goal of the workshop was to expose participants to Data-driven Thinking through experimental design, multi-sensor data collection, data analysis, visualization, and insight generation. We evaluate their learning experience in designing an experimental setup, collecting real-time immediate environmental and physiological body reactions data. Our results from the workshop show that participants increased their knowledge about measuring, visualizing and understanding data of the surrounding built environment.

An Interval Temporal Logic for Time Series Specification and Data Integration

The analysis of temporal series—in particular, analysis of multisensor data—is a complex problem. It depends on the application domain, the way the data have to be used, and sensors available, among other factors. Various models, algorithms, and technologies have been designed for this goal. Temporal logics are used to describe temporal properties of systems. The properties may specify the occurrence and the order of events in time, recurring patterns, complex behaviors, and processes. In this paper, a new interval logic, called duration calculus for functions (DC4F), is proposed for the specification of temporal series corresponding to multisensor data. DC4F is a natural extension of the well-known duration calculus, an interval temporal logic for the specification of process duration. The adequacy of the proposed logic is analyzed in the case of multisensor data concerning volcanic eruption monitoring. It turns out that the relevant behavior concerns time intervals, not only accumulated history as it is described in other kinds of temporal logics. The examples analyzed demonstrate that a description language is required to specify time series of various kind relative to time intervals. The duration calculus cannot be successfully applied for this task. The proposed calculus allows one to specify temporal series and complex interval-dependent behaviors, and to evaluate the corresponding data within a unifying logical framework. It allows to formulate hypotheses concerning volcano eruption phenomena. However, the expressivity of DC4F comes at the cost of its decidability.

OC-SMART: A machine learning based data analysis platform for satellite ocean color sensors

We introduce a new platform, Ocean Color - Simultaneous Marine and Aerosol Retrieval Tool (OC-SMART), for analysis of data obtained by satellite ocean color sensors. OC-SMART is a multi-sensor data analysis platform which supports heritage, current, and possible future multi-spectral and hyper-spectral sensors from US, EU, Korea, Japan, and China, including SeaWiFS, Aqua/MODIS, SNPP/VIIRS, ISS/HICO, Landsat8/OLI, DSCOVR/EPIC, Sentinel-2/MSI, Sentinel-3/OLCI, COMS/GOCI, GCOM-C/SGLI and FengYun-3D/MERSI2. The products provided by OC-SMART include spectral normalized remote sensing reflectances (Rrs values), chlorophyll_a (CHL) concentrations, and spectral in-water inherent optical properties (IOPs) including absorption coefficients due to phytoplankton (aph), absorption coefficients due to detritus and Gelbstoff (adg) and backscattering coefficients due to particulates (bbp). Spectral aerosol optical depths (AODs) and cloud mask results are also provided by OC-SMART.The goal of OC-SMART is to improve the quality of global ocean color products retrieved from satellite sensors, especially under complex environmental conditions, such as coastal/inland turbid water areas and heavy aerosol loadings. Therefore, the atmospheric correction (AC) and ocean IOP algorithms in OC-SMART are driven by extensive radiative transfer (RT) simulations in conjunction with powerful machine learning techniques.To simulate top of the atmosphere (TOA) radiances, we solve the radiative transfer equation pertinent for the coupled atmosphere-ocean system. For each sensor, we have created about 13 million RT simulations and comprehensive training datasets to support the development of the machine learning AC and in-water IOP algorithms. The results, as demonstrated in this paper, are very promising. Not only does OC-SMART improve the quality of the retrieved water products, it also resolves the negative water-leaving radiance problem that has plagued heritage AC algorithms. The comprehensive training datasets created using multiple atmosphere, aerosol, and ocean IOP models ensure global and generic applicability of OC-SMART.The use of machine learning algorithms makes OC-SMART roughly 10 times faster than NASA's SeaDAS platform. OC-SMART also includes an advanced cloud screening algorithm and is resilient to the contamination by weak to moderate sunglint and cloud edges. It is therefore capable of recovering large amounts of data that are discarded by other algorithms (such as those implemented in NASA's SeaDAS package), especially in coastal areas. OC-SMART is currently available as a standalone Python package or as a plugin that can be installed in ESA's Sentinel Application Platform (SNAP).

Multisensor Data Fusion in Testability Evaluation of Equipment

The multisensor data fusion method has been extensively utilized in many practical applications involving testability evaluation. Due to the flexibility and effectiveness of Dempster–Shafer evidence theory in modeling and processing uncertain information, this theory has been widely used in various fields of multisensor data fusion method. However, it may lead to wrong results when fusing conflicting multisensor data. In order to deal with this problem, a testability evaluation method of equipment based on multisensor data fusion method is proposed. First, a novel multisensor data fusion method, based on the improvement of Dempster–Shafer evidence theory via the Lance distance and the belief entropy, is proposed. Next, based on the analysis of testability multisensor data, such as testability virtual test data, testability test data of replaceable unit, and testability growth test data, the corresponding prior distribution conversion schemes of testability multisensor data are formulated according to their different characteristics. Finally, the testability evaluation method of equipment based on the multisensor data fusion method is proposed. The result of experiment illustrated that the proposed method is feasible and effective in handling the conflicting evidence; besides, the accuracy of fusion of the proposed method is higher and the result of evaluation is more reliable than other testability evaluation methods, which shows that the basic probability assignment of the true target is 94.71%.

다중 센서 데이터 분석을 위한 설명 가능한 합성곱신경망 모델

In this paper, we propose a new convolutional neural networks (CNNs) structure to overcome the interpretation problem posed by a series of the nonlinear transformation processes in CNNs. Convolutional neural networks (CNNs) have been successfully used in various applications including images, signals, and texts. Despite their superior performance in various tasks, the main limitation of CNNs is the lack of interpretability. Recently, deep neural network models including CNNs have been effectively used for the analysis of multi-sensor data. Although the original characteristics of deep neural network models offer high accuracy in classification for activity recognition, they are difficult to identify the critical sensors that play a significant role in discrimination of various equipment conditions. The proposed CNNs offers good performance and interpretability simultaneously. The effectiveness and applicability of the proposed method were demonstrated by the real sensor data collected from construction equipment for activity recognition.