Partial Data Loss Research Articles

Missing data recovery based on temporal smoothness and time-varying similarity for wireless sensor network

Wireless Sensor Networks (WSN) play a vital role in the Internet of Things (IoT) and show great potential in monitoring applications. However, due to harsh environmental conditions and unreliable communication links, WSN often encounter partial data loss during data collection, which inevitably affects the quality of service. To address this challenge, researchers have employed matrix completion techniques to recover missing data by exploiting the low-rank features in the data, but its accuracy is not satisfactory. This paper argues that the spatiotemporal characteristics of the data underlie its low-rank nature, enabling a more accurate capture of the intrinsic patterns within the data. Drawing on this insight, we propose a missing data recovery algorithm based on Temporal Smoothness and Time-Varying Similarity (TSTVS). Unlike traditional low-rank methods, the TSTVS algorithm directly utilizes the structural features of data in the spatiotemporal domain to establish a missing data completion model. Subsequently, the model is converted into an unconstrained optimization problem using the penalty function method, and the gradient descent method is applied to solve it, reconstructing the complete data matrix. Finally, simulation experiments were conducted on three real-world monitoring datasets, comparing the TSTVS with three low-rank methods, Efficient Data Collection Approach (EDCA), Matrix factorization with Smoothness constraints (MFS) and Data Recovery Based on Low Rank and Short-Term Stability(DRLRSS). The experimental results indicate that the proposed TSTVS algorithm consistently outperforms the three low-rank based algorithms in terms of recovery accuracy across different datasets and missing rate scenarios.

Resilient Dynamic State Estimation for Multi-Machine Power System With Partial Missing Measurements

Accurate tracking the dynamics of power system plays a significant role in its reliability, resilience and security. To achieve the reliable and precise estimation results, many advanced estimation methods have been developed. However, most of them are aiming at filtering the measurement noise, while the adverse affect of partial measurement missing is rarely taken into account. To deal with this issue, a discrete distribution in the interval [0,1] is introduced to depict mechanism of partial measurement data loss that caused by the sensor failure. Then, a resilient fault tolerant extended Kalman filter (FTEKF) is designed in the recursive filter framework. Eventually, extensive simulations are carried on the different scale test systems. Numerical experimental results illustrate that the resilience and robustness of the proposed fault tolerant EKF method against partial measurement data loss.

Assessment of the present capacity for Management of the Loco Concholepas concholepas (BRUGUIERE, 1789) (GASTROPODA, MURICIDAE) in Chile.

In recent years the fishery for loco (Concholepas concholepas) has undergone a number of changes in management strategy. The first was a change in the mínimum size limit from 95 mm in the north of Chile and 105 min in the south to a uniform 100 mm in 1981. An annual 3 month closed season was initiated in 1982, and. starting in September of 1985. There was a closure of nearly two years in all regions except the southernmost three. This two-year closure has been followed by two short open seasons, one of 45 days in 1987. and one with a national 5,000 ton quota in 1988. The response by fishermen to increasingly restrictive regtdation has been some clandestine fishing and the stockpiling of locos prior to the open season. These actions have resulted in i he partial loss of landings data and a nearly total loss of meaningful effort data. This loss of data will have serious impact on the ability of fishery managers to effectively evaluate the State of the fishery and determine future regulatory actions. There is considerable published background information on the basic biology of the species. and some Information on aspects of the population dynamics such as growth and mortality. There are. however, large discrepancies between authors which mav be due lo sub-populalion or regional differences.

Алгоритм вычисления температуры поверхности для восстановления потери данных Landsat 8–9 Collection 2 Level 2

The use of thermal images for the analysis of urban areas is presented in Proceedings of InterCarto. InterGIS, on the example of the city of Krasnoyarsk according to Landsat 8 data. New Landsat 8–9 collection 2 level 2 data became available in March 2021. The data has been supplemented with valuable scientific products such as atmospheric parameters (atmospheric spectral transmittance, up-going radiation and down-going radiation), earth surface temperatures, thermal radiation, quality assessment mask. A detailed study of the data showed that for many territories previously studied by the authors, data were not provided in full. There is partial or complete loss of data in the emissivity layer, and as a consequence the loss is repeated in the Earth’s surface temperature layers. The same situation with data loss is observed in a number of other large Russian cities. The current situation required a detailed study of the documentation describing the Landsat 8–9 algorithms. Having studied all the stages of the algorithm execution, the authors managed to restore the sequence of calculations, while retaining additional atmospheric parameters. The lost data in the emissivity channel has been replaced with a new layer computed via NDVI. As a result, the authors tested the operation of the proposed algorithm on the territory of the city of Krasnoyarsk. The result of the algorithm execution is layers with surface temperature values without data loss, except for scenes with high cloudiness. There is a slight discrepancy between the surface temperature values when using the described algorithm and initial data, due to the use of different sources to determine the emissivity.

DISNet: A sequential learning framework to handle occlusion in human action recognition with video acquisition sensors

Human action recognition (HAR) requires a clear line of sight for the video acquisition system to properly collect the data. However, the data acquisition is hampered due to presence of the obstructions in front of the action happening in real time environment. In recently reported HAR techniques, the widely used multi-stream sequential networks are having two major limitations. First, the performance of the existing algorithms degrades comprehensively in the presence of partial loss of action data due to obstruction. Second, the performance of the existing networks is restricted by not exploiting the dependency relationship between different streams of a multi-stream network. To handle these problems, a novel double input sequential network (DISNet) is proposed which takes care the HAR algorithm in the presence of partial loss of action data. The DISNet, which learns inter-stream information, is jointly trained on the normal data and the artificially created obstructed data of a single video to provide immunity to the HAR network against obstructions. The proposed DISNet is evaluated on publicly available datasets like KTH, UCF-sports, JHMDB and UCF101, which provides performance accuracy of 83.57%, 82.27%, 50.24% and 54.96% respectively.

Contactless medical equipment AI big data risk control and quasi thinking iterative planning

Research Background, the intelligent polymorphic system of heavy core clustering fitting iterative programming is constructed by using the edge lens of dual core heavy core. The tracking system of heavy core TANH equilibrium array is used to obtain the abnormal data range. The energy regular fluctuation of the edge lens with dual core and heavy core is used to obtain high-definition images. And build the complexity dependent parameter group from low-end equipment to high-end equipment. Heavy core clustering of hierarchical fuzzy clustering system based on differential incremental balance theory is applied to Contactless medical equipment AI big data risk control and quasi thinking iterative planning. At the same time, the mathematical model risk control is performed by fitting the TANH balance of the local nonlinear random regular micro-vibration diffusion curve. The CT/MR original data is subjected to hierarchical cross domain overlapping grid screening with the structure of fitting weakly nonlinear curve, which can capture the heavy core cluster analysis of the core layer of big data anomalies [1:10]. Successfully control the parameter group of CT/MR machine internal data, big data AI Mathematical model risk. The polar graph of high-dimensional heavy core clustering processing data is regular and scientific. The same time, it can prevent the dimension disaster caused by the construction of high-dimensional big data due to the partial loss of original data, and form a stable and predictable maintenance of CT/MR. Compared with the discrete characteristics of the polar graph of the original data. So as to correctly detect and control the dynamic change process of CT/MR in the entire life cycle. It provides help for predictive maintenance of early pre-inspection and orderly maintenance of the medical system, and developed standardized model software of automated unsupervised learning for medical big equipment big data AI Mathematical model risk control. Scientifically evaluated the exposure time and heat capacity MHU% of CT tubes, as well as the internal law of MR (nuclear magnetic resonance), and processed big data twice and three times in heavy nuclear clustering. After optimizing the algorithm, hundreds of thousands of nonlinear random vibrations are performed in the operation and maintenance database every second, and at least 30 concurrent operations are formed, which greatly improves and shortens the operation time (Yanwei et al. in J Complex 2017:1–9, 2017. https://doi.org/10.1155/2017/3437854). Finally, after adding micro-vibration quasi thinking iterative planning for the uncertain structure of AI operation, we can successfully obtain the scientific and correct results required by high-dimensional information and analyze images. This kind of AI big data risk control improves the intelligent management ability of medical institutions. Cross platform embedded web system for predictable maintenance of AI big data is established (Qi et al. in J IEEE Trans Ind Inf 99:1, 2020. https://doi.org/10.1109/tii.2020.3012157).

Information system for creating a generalized model of complex technical objects

Creation of complex technical objects (CTO, such as science-intensive engineering objects, which are characterized by the number of elements and connections equal to 106 and more) – a process containing long subprocesses, complex objects, models, and based on available standards and information technology (IT). It is known from practice that CTO models are created with the help of information systems - components of CALS and PLM-solutions. This process have the following shortcomings: models at different stages of creation of CTO are not completely interconnected; CTO modeling takes place using different components of IT CALS and PLM-solutions, which creates constant difficulties in data conversion and leads to partial or complete loss of model data; there is no automated communication with other CTO models.
 A generalized model of complex technical objects is proposed, which based on using the theoretical-multiple approach allows to establish an information connection between the models of the process of creating complex technical objects. The information system of creation of the generalized model of difficult technical objects that allowed automating process of processing and construction of models is developed; to supplement the technology of parallel PLM design and components of information technologies CALS and PLM-solutions for the task of creating models of complex technical objects.

GapFill-Recon Net: A Cascade Network for simultaneously PET Gap Filling and Image Reconstruction

PET image reconstruction from incomplete data, such as the gap between adjacent detector blocks generally introduces partial projection data loss, is an important and challenging problem in medical imaging. This work proposes an efficient convolutional neural network (CNN) framework, called GapFill-Recon Net, that jointly reconstructs PET images and their associated sinogram data. GapFill-Recon Net including two blocks: the Gap-Filling block first address the sinogram gap and the Image-Recon block maps the filled sinogram onto the final image directly. A total of 43,660 pairs of synthetic 2D PET sinograms with gaps and images generated from the MOBY phantom are utilized for network training, testing and validation. Whole-body mouse Monte Carlo (MC) simulated data are also used for evaluation. The experimental results show that the reconstructed image quality of GapFill-Recon Net outperforms filtered back-projection (FBP) and maximum likelihood expectation maximization (MLEM) in terms of the structural similarity index metric (SSIM), relative root mean squared error (rRMSE), and peak signal-to-noise ratio (PSNR). Moreover, the reconstruction speed is equivalent to that of FBP and was nearly 83 times faster than that of MLEM. In conclusion, compared with the traditional reconstruction algorithm, GapFill-Recon Net achieves relatively optimal performance in image quality and reconstruction speed, which effectively achieves a balance between efficiency and performance.

РАЗРАБОТКА НАДЕЖНОГО МЕТОДА СВЯЗИ РТК НА БАЗЕ ГРУППОВОГО МЕТОДА РАЗДЕЛЕНИЯ ДАННЫХ, ОСНОВАННОГО НА СИСТЕМЕ ОСТАТОЧНЫХ КЛАССОВ

The paper considers a reliable method of data transmission in commu-nication systems andcontrol of robotic complexes. Due to the fact that a change in part of the encoded informationtransmitted through com-munication channels can lead to partial or complete loss of data and, asa consequence, lead to loss of control over the robotic complex. There-fore, it is necessary to apply methods of protecting data transmitted over radio channels. The proposed method is intendedto ensure the protection of information in the communication channels of robotic complexes fromthe access of unauthorized users and to confirm the reliability of the information received.The article examines data protection methods designed to protect in-formation circulating in systemsformed by several interacting agents. The approach under consideration is based on themethods of infor-mation protection and error-correcting coding based on the system of residualclasses. The applied methods of error-correcting coding, based on the system of residual classes,are based on the idea of threshold data separation, in which the original information can be restoredfrom parts of the original information. This is due to the fact that redundant modulararithmetic, or redundant system of residual classes, has unique properties with re-spect to errordetection and correction. In addition, the system of resid-ual classes has such an advantage as lowcomputational complexity of data separation algorithms. To increase the reliability of communicationbetween robotic complex-es in multichannel communication systems, the paper proposes ameth-od for information protection and noise-resistant coding, based on a multistage thresholddata separation. As a result of the work, a system of noise-resistant information trans-mission wasobtained, which provides comprehensive protection for robotic systems.

An Adaptive Fault-Tolerant EKF for Vehicle State Estimation With Partial Missing Measurements

Some vehicle state information that cannot be measured by in-vehicle sensors is quite important for the active safety control of intelligent vehicles. To obtain these key information in real-time, many advanced estimation algorithms are proposed. However, the existing studies focus on the effect of sensor measurement noise on estimation accuracy and rarely consider the impact of sensor data loss. In this article, a novel adaptive fault-tolerant extended Kalman filter is proposed to estimate vehicle state in case of partial loss of sensor data. The randomness of the data loss is first defined by a discreet distribution in interval [0,1]. Then, the fault-tolerant extended Kalman filter is derived based on a recursive filter framework. Furthermore, a fading factor on the basis of the orthogonal theory is used to improve the adaptability of fault-tolerant extended Kalman filter. Experimental results demonstrate that the estimation performance of the proposed approach is better than the extended Kalman filter.

Fine-grained alignment of cryo-electron subtomograms based on MPI parallel optimization

BackgroundCryo-electron tomography (Cryo-ET) is an imaging technique used to generate three-dimensional structures of cellular macromolecule complexes in their native environment. Due to developing cryo-electron microscopy technology, the image quality of three-dimensional reconstruction of cryo-electron tomography has greatly improved.However, cryo-ET images are characterized by low resolution, partial data loss and low signal-to-noise ratio (SNR). In order to tackle these challenges and improve resolution, a large number of subtomograms containing the same structure needs to be aligned and averaged. Existing methods for refining and aligning subtomograms are still highly time-consuming, requiring many computationally intensive processing steps (i.e. the rotations and translations of subtomograms in three-dimensional space).ResultsIn this article, we propose a Stochastic Average Gradient (SAG) fine-grained alignment method for optimizing the sum of dissimilarity measure in real space. We introduce a Message Passing Interface (MPI) parallel programming model in order to explore further speedup.ConclusionsWe compare our stochastic average gradient fine-grained alignment algorithm with two baseline methods, high-precision alignment and fast alignment. Our SAG fine-grained alignment algorithm is much faster than the two baseline methods. Results on simulated data of GroEL from the Protein Data Bank (PDB ID:1KP8) showed that our parallel SAG-based fine-grained alignment method could achieve close-to-optimal rigid transformations with higher precision than both high-precision alignment and fast alignment at a low SNR (SNR=0.003) with tilt angle range ±60∘ or ±40∘. For the experimental subtomograms data structures of GroEL and GroEL/GroES complexes, our parallel SAG-based fine-grained alignment can achieve higher precision and fewer iterations to converge than the two baseline methods.

Precise capture of fish movement trajectories in complex environments via ultrasonic signal tag tracking

Ultrasonic tag tracking technology has been widely used in fish behavior research, but most existing studies obtain fish movement trajectories through use of supporting software or post-processing services provided by the manufacturers of the relevant equipment. Articles concerning fish positioning principles and optimization and processing methods suited to complex water environments, are rarely published. This paper proposes a set of solutions permitting precise capture of fish movement trajectories. The proposed methods can effectively mitigate the impact of difficult conditions such as partial loss of time synchronization signals and data containing gross error-bearing observations on the process of locating tagged fish. Test results show that: (1) Even if time synchronization signals for a continuous 2 h period are lost, the accuracy of time synchronization between the hydrophones remains consistent with the requirements of meter-level positioning. (2) The proposed method can effectively detect the existence of gross errors exceeding 10 m in the observation data; the success rate of the method in identifying unique gross error-bearing observations was 99.7%. (3) Based on existing hydrophones with millisecond-level observation precision, three-dimensional (3D) positioning of fish-borne signal tags to 2˜5 m precision is possible.

Переход с PAPI на CAPI: опыт интервьюеров и характеристики, влияющие на их ожидания

В работе представлен анализ оценки интервьюерами и супервайзерами пробного этапа перевода опроса, проводимого в рамках проекта «Российский мониторинг экономического положения и здоровья населения НИУ ВШЭ» (РМЭЗ НИУ ВШЭ), с методики бумажной анкеты (Pen and Pencil Interview — PAPI) на методику с использованием планшета (Computer Assisted Personal Interview — CAPI). В качестве данных для анализа послужил материал 13 интервью с полевыми интервьюерами и тремя интервью с супервайзерами, а также анкетирование 50 интервьюеров.
 Основными преимуществами CAPI являются: значительное снижение стоимости проведения полевого этапа (особенно в долгосрочной перспективе), повышение скорости передачи и обработки данных, а также в некоторых случаях повышение качества данных (например, увеличение доли содержательных ответов на сензитивные вопросы — потребление алкоголя, доход). Тем не менее, переход к сбору данных с использованием планшетов связан также с некоторыми рисками. Среди основных недостатков CAPI можно выделить увеличение воспринимаемой длительности анкеты, потерю гибкости в процессе заполнения анкеты, психологический дискомфорт (страх), а также возрастающую когнитивную нагрузку на интервьюеров. Использование планшетов также было сопряжено с такими техническими трудностями, как медленная работа устройства, полная или частичная потеря данных, невозможность определения местоположения.
 Кроме того, интервьюеры отмечают изменение коммуникативной составляющей интервью при опросе с использованием планшетов, что потенциально может оказывать значительное влияние на восприятие ситуации интервьюерами и респондентами и снижать удовлетворенность участием в проекте. С помощью моделирования структурными уравнениями на данных анкетирования 50 интервьюеров, отобранных для участия в пробном этапе, было выявлено, что влияние возраста на ожидания интервьюеров относительно CAPI полностью опосредуется их уверенностью пользования техническими средствами.
 Благодарность. Статья подготовлена в рамках работы по гранту Российского научного фонда «Разработка интерактивной методической инфраструктуры для повышения качества данных лонгитюдных социологических исследований» (№ 17-78-20172).
 Авторы статьи выражают благодарность за помощь в сборе данных и за ценные комментарии в процессе написания статьи И.Ф. Девятко и А.М. Мавлетовой.

An energy efficient flexible delay tolerant network with adaptive secured framework (ASF-DTN)

Delay tolerant networks are widely used in mobile communications because of network withstanding capability of delay. However, when the connectivity of nodes increases, data loss may occur while transmission. Due to the malicious behavior of nodes the data may get permanently lost, which is known as black hole attacks and there is a chance to partial data loss because of the lower energy level of a node, which is known as a gray hole attack. In existing work, the data capacity of a node is not limited, so most of the highly energy efficient node contains a huge amount of data, when compared to other nodes which may lead to random attack. To overcome this, we propose a subjective capability model (SCM) for each and every node to limit the capacity of each node. ASF-DTN prevents collision attack and injection attack by implementing Kalman filtering, which can statistically analyze the behavior of nodes while each and every transmission. Here, we propose an effective node optimization scheme using genetic algorithm with a fitness function to find out energy efficient nodes among the optimal path for effective communication and its performance.

АНАЛИЗ ПОМЕХОУСТОЙЧИВОСТИ НЕРАВНОМЕРНЫХ КОДОВ, ПРИМЕНЯЕМЫХ ДЛЯ ПЕРЕДАЧИ ИНФОРМАЦИИ ПО КАНАЛУ СВЯЗИ

The article meets the challenge of the synchronization detection in the code sequence of uneven codes, lost in the result of impact of errors in the communication channel when transmitting compressed messages. For this purpose, basing on statistical properties of the messages and digital communication channel source, the authors got the observation likelihood functions in the presence and in the absence of synchronization. In this situation, the frequencies of occurrence of volume-defined alphabetic letters in the decoded sequence of characters can be used as the observations. As a primary distribution, the authors used a polynomial distribution, which characterizes the frequencies of occurrence of alphabetic letters in the volume-defined selection depending on their probabilities. The peculiarity of this approach for the synchronization detection is the use of polynomial distribution for description of the function of synchronization unavailability likelihood on the basis of the theory of complex hypotheses recognition with a uniform averaging. In the result of such approach, the authors obtained the likelihood ratio depending on the probabilities of alphabetic letters occurrence using which it is possible to make the conclusion about the presence or absence of synchronization. The algorithm for synchronization detection in event of non-uniform encoding was developed on the base of the sequential analysis procedure. The value of average Bayes risk presented in the form of probability of synchronization detection error was selected to be an indicator for the algorithm effectiveness evaluation. The analysis of noise stability of uneven codes shows that the application of modern multimedia technologies in communication networks increases the amount of transmitted information and used alphabet of messages that leads to the loss of synchronization and the effect of error propagation. It causes partial or total loss of data. The developed algorithm will allow recovering of the compressed messages that were distorted during their transmission via communication channel.

Image Tampering Detection and Repairing

It is very crucial to design efficient methods to provide authentication to document images mainly for images whose security must be protected. In digital media the authentication, tampering detection and repairing of the tampered data is main concern. Nowadays many digital documents are transferred on the internet regularly e.g. circuit diagrams, scanned cheques, signed documents, drawings, design drafts etc. Here the method of image authentication, tampering detection and repairing is explained. The grayscale image document image authentication with self repair method is provided. The authentication signal is calculated from binary image of the original document image. Secret sharing scheme is used to calculate this authentication signal. The extra alpha channel is embedded in the original grayscale document image. The calculated authentication signal is then embedded in this alpha channel. This is one of the way of visual cryptography in which image is visible but authentication signal will not be recognized. Then this authentic image is encrypted for better security. this whole procedure is done at sender side. In second module i.e. receiver side the image is decrypted. Then the embedded authentication signal is extracted from the received authentic image. The new authentication signal is calculated from the binary image of the received authentic image. This new authentication signal is compared with the extracted signal and the integrity check is provided. If data modification is detected, then the method is proved to repair the tampered pixels using the reverse secret sharing method. The tampering of data is detected successfully in this method without any original image backup. This type of secret sharing scheme is helpful for reducing the risk of incidental partial data loss. This method is also applicable to the color image. Algorithm for color image is also same as grayscale images. In this paper the result is provided only for the grayscale images and supposed to expand for color images as well.

Authentication of grayscale document images using shamir secret sharing scheme.

This paper proposed a new blind authentication method based on the secret sharing technique with a data repair capability for grayscale document images .Shamir proposed threshold secret sharing scheme in which a secret message is transformed in to shares for keeping by participants, and when some of the shares, not necessarily all of them, are collected, the secret message can be lossless recovered.. This secret sharing scheme is useful for reducing the risk of incidental partial data loss. In additional image transformed in to a stego-image in the PNG format with an additional alpha channel for transmission on networks. For tampered block data repairing is applied by reverse Shamir scheme. It also presents the security issues and discuss on keeping high quality visual effect.

GREYSCALE IMAGE AUTHENTICATION AND REPAIRING

It is very necessary to design effective methods to solve image authentication problem, particularly f or images whose security must be protected. In digital documents it is necessary to provide data authentication and repairing of tamper ed data e.g. important documents such as certificates, scanned cheques, drawings, circuit diagrams, signed documents, desig n draft etc. In this paper, a technique for authentication of images with self-re pair capability for fixing tampered image data is eThe input image is assumed to be a binary-type greyscale image with 2 main gray values. Alpha channel is combined in the greyscale image .Using the binary image, authentication signal is calculated w hich is then embedded in the alpha channel to creat e an authentic image. After embedding the authentication signal, image is encry pted. If still Content modifications of the stego-image is detected, then data is repaired at the pixel level using reverse secrete s haring scheme. In case if the alpha channel is com pletely removed from the stegoimage, the integral resulting image is regarded as unauthentic, meaning that the integrity check of th e image fails. The proposed method is based on the generation of (k, n)-thresho ld secret sharing scheme proposed by Shamir. In thi s method a secret message is converted into n shares which is used as authentica tion signal for keeping by n participants; and when k of the n shares are collected, the secret image can be recovered without any loss[ 8]. This type of secret sharing scheme is helpful fthe risk of incidental partial data loss.

Application of a Fully Automatic Autofocusing Algorithm for Post-Processing of Synthetic Aperture Radar Images based on Image Entropy Minimization

In this paper an application of an autofocusing algorithm, previously developed by the authors for the case of Inverse Synthetic Aperture Radar (ISAR) [i.e. airborne radar targets], is presented here for the case of Synthetic Aperture Radar (SAR) geometry [i.e. ground radar targets]. Here, a new theory and methodology for generating SAR synthetic backscattered data is also developed. This algorithm is named ‘CPI-split-algorithm’, where CPI stands for ‘Coherent Processing Interval’. Moreover, two simulation scenarios are presented for a ship target, which is located on the sea surface. In the first simulation scenario, the ship is considered at first to be stationary, and subsequently an oscillatory movement is induced to its position along the vertical axis, due to sea surface motion. In the second simulation scenario, a partial loss of data is examined, caused by temporary accidental malfunctioning of radar transmitter or receiver, assuming that the target (ship) is stationary. Numerical results presented in this paper show the effectiveness of the proposed autofocusing algorithm for SAR image enhancement. DOI: http://dx.doi.org/10.5755/j01.eee.19.6.4573

Expectation maximization approach to data-based fault diagnostics

The data-based fault detection and isolation (DBFDI) process becomes more potentially challenging if the faulty component of the system causes partial loss of data. In this paper, we present an iterative approach to DBFDI that is capable of recovering the model and detecting the fault pertaining to that particular cause of the model loss. The developed method is an expectation–maximization (EM) based on forward–backward Kalman filtering. We test the method on a rotational drive-based electro-hydraulic system using various fault scenarios. It is established that the developed method retrieves the critical information about presence or absence of a fault from partial data-model with minimum time-delay and provides accurate unfolding-in-time of the finer details of the fault, thereby completing the picture of fault detection and estimation of the system under test. This in turn is completed by the fault diagnostic model for fault isolation. The obtained experimental results indicate that the developed method is capable to correctly identify various faults, and then estimating the lost information.