There is currently no trustworthy method in place to provide continuous monitoring of all process conditions and leather qualities inside closed reactors during leather production, implying that there is no genuine control. Only by pausing the reactor and sampling the solution and leather can conditions inside the reactors be checked. The leather industry's automation systems are based on a dated centralised architecture, which creates a critical point of failure and an operational bottleneck. To increase the efficiency of the leather manufacturing process, the paper offers a fault-tolerant multi-agent system (MAS) architecture that delivers the high flexibility and agility required by the leather industry's turbulent environment. The quality of leather can be determined by image structure, and the quality can be estimated using image processing.

In an attempt to enhance solar light photon to electron transformation proficiency of copper indium gallium selenide (CIGS) solar cells, computational exploration has been accomplished through numerical simulation. The SCAPS program was utilized to simulate enactment of CIGS. The electrical, optical properties of CIGS such as band diagram, current density, recombination current, IPCE and current - voltage efficiency was analyzed. The electrical, physical properties, thicknesses of individual layers comprising CIGS, CdS and ZnO were optimized along with their operating temperature. The CIGS solar cell efficiency analysis was executed and analyzed in the AM1.5 spectrum. The depth of CIGS, CdS and ZnO layers in CIGS solar cell determines the efficiency. The simulated optimization of CIGS properties is encouraging for enhancing the CIGS solar cell proficiency.

The forecast of river water temperature has been carried out based on a variety of weather reports and reports on water bodies on earth. Currently, a range of numerical modeling procedures is used for forecast of river water temperature. For this, the range of input parameters like hourly temperature of river water (Th) and atmospheric temperature (TA) relating to rivers in India is considered. Thus, to develop an indigenous river water prediction model, certain autoregressive inputs based on meteorological and hydrological models are considered. Each neural network type like feed forward neural network (FFNN) with backpropagation algorithm (BPA) and convolution neural network (CNN) is calibrated independently for 1000 iterations and the mean, median and standard deviation are computed and used for the comparison. Finally, all the models are collectively tested. The results demonstrate that CNN in majority cases outperformed the results obtained from the FFNN trained with BPA. The selection of artificial neural network (ANN) models relies on the method by which the river models are evaluated. Hence, one should consider this constraint so as to propose any other equivalent river models. The optimal results are obtained when mean, maximum and minimum daily atmospheric temperatures for the preceding days are used as inputs. The mean squared error is minimized by this collective aggregation technique by 98.2-% for specific cases, and obviously, it reduces the divergence in modeling performance obtained by various ANN models. The optimally efficient model serves as a soft sensor offering inputs to Internet of things (IoT) system. A wireless area network (WAN) is a collection of many sensor nodules, each competent of sampling, dispensation and corresponding one or more environmental parameters like hourly temperature of river water (Th) and atmospheric temperature (TA) relating to rivers in India.

Screening of nanosized pancreatic tumors, both benign and malignant, is a very important issue in the medical field because it is directly influences the digestive system, which has an effect on human health. Nanobiosensor is the application of nanotechnology in the medical field. This is a multidisciplinary field that currently involves nanotechnology and biomedical applications. The malfunctioning of the pancreas is a health concern because it helps maintain the blood glucose level at a nominal value. Day-to-day food habits have driven the need to develop rapid, responsive, and reliable methods to detect pancreatic tumors. The rapid development of nanosensors that have an advantage to detect variations in the texture of the pancreas in nanometers has paved way to diagnose a malfunctioning pancreas at the onset stage linking nanosensors with modern Information and Communication Technologies (ICTs) enabling novel and online ways of detection accompanied with high accuracy. Various types of nanosensors are being developed to meet different requirements in the field of medicine for detection of various abnormalities related to the organs of the human body. Detection of nanosized pancreatic tumors is the focus of this work. The existence of nanosized pancreatic tumors in the patient leads to early diagnosis. If the tumor is identified in the chronic stage, the chances of survival of the patient are very less. Detection of nanosized tumors will enhance the analysis, diagnosis, and prognosis at the onset stage leading to suitable and timely medication. Currently, this work depends on feature analysis of Magnetic Resonance Imaging (MRI) images obtained from the database to identify the nanosized pancreatic tumors at the onset stage. A distinct diagnosis method is proposed for identification of pancreatic tumors using image texture characters, which were statistically evaluated using MATLAB. Diagnosis was done using Deep Wavelet Neural Networks (DWNN). Using DWNN method, combined with intelligent and pattern recognition algorithms, nearly 99% of sensitivity in the detection of nanosized pancreatic tumors is achieved.

Endometrial cancer causes death among women community worldwide and that the content-based image retrieval (CBIR) is used to locate images in vast databases using machine vision techniques. Effective decision making is the only solution the physicians can offer to detect the occurrence of heart disease so that the patients can be relaxed and can undergo treatment at an appropriate time. The CBIR model on receipt of query on receipt of query, extracts the same set of features of a query which matches against indexed features index and retrieves similar images from the database. Further, these are facilitated by medical images stored in distributed and centralized servers are referred for knowledge, teaching, information, and diagnosis. Abnormal vaginal bleeding serves as the primary symptom of endometrial cancer. Thus, the system performance mainly depends on the features extracted from the computed tomography (CT) images that are adopted for indexing. Conversely, there is a shortage of ineffective scrutiny tools to identify the concealed relationship in the data pattern. This proposed technique intends to provide a study of existing techniques to form the knowledge base which will guide the gynecologists to take an effective decision. Nine features with radial basis function network were used to compute the formation of endometrial cancer at early stages. The feature selected must require lesser storage, retrieval time, cost of retrieval model and must support different classifier algorithms. Feature set adopted should support to increase the early diagnosis of the disease. This work summarizes the strength of local binary patterns (LBP) and its variants for indexing medical images. The efficacy of the LBP is verified using medical images from OASIS. The results prove good prospects of LBP and its variants which is due to the presence of unique binary patterns for normal and abnormal uterus conditions. The LBP features are used to train the intelligent classifier where feed-forward neural networks (FFNN) are trained with radial basis function network (RBFN). The RBFN facilitates the retrieval of images corresponding to normal and abnormal conditions.

Scrutiny of combustion quality and its equivalent NOx emissions from flame images in thermal and gas turbine power plants is of immense significance in the realm of climate change. A remote monitoring scheme using image processing, Artificial Intelligence (AI) and Internet of Things (IoT) to efficiently minimize the flue gas emissions can be carried out. The principal goal is in detection, recognition and understanding of combustion conditions in power plants ensuring low green house or flue gas emissions which contribute to climate change. In this work, smart sensors using feed forward neural network with Ant Colony Optimisation (ACO) and Particle Swarm Optimization (PSO) are used for estimation of various flue emissions. This scheme uses the information from the colour of the flame images in the combustion chamber at power plants, which is the foundation for obtaining high combustion quality and low flue gas emissions. The initial gait is to describe a facet vector for each flame image including 10 feature elements. Image Enhancement is done to obtain distinctive attributes from the captured images. The perception of object (flame feature) recognition and classification of the flame image is conceded out to measure the combustion quality and flue gas emissions from the flame colour. The samples including some flame images, parts of which are used to train and test the model. Finally, the entire samples are recognized and classified. Experiments prove that flame image classification to be an effective monitoring scheme for reducing the flue gas emissions.

Worldwide, the cardiac stenosis is the number one cause of death among human community. The countries which are below the poverty line and having mediocre income experience nearly 80% of deaths due to cardiovascular disease. If the present status continues, approximately, about 20 million people will die due to coronary artery block by 2030. Effective decision making is the only solution the physicians can offer to detect the occurrence of heart disease so that the patients can be relaxed and can undergo treatment at appropriate time. The heart attack mainly occurs due to decreased blood and oxygen supply to the heart due to deposition of cholesterol on the sidewalls of the arteries causing heart disease. Conversely, there is a shortage in effective scrutiny tools to identify the concealed relationship in the data pattern. This proposed technique intends to provide a study of existing techniques to form the knowledge base which will guide the cardiologists to take an effective decision. The objective of this scheme is to diagnose the presence of various levels of block in coronary artery using the features extracted from the computed tomography (CT) angiogram. Seven features with radial basis function network were used to measure the percentage of blocks in the coronary artery. This investigation proved that the data mining approaches like RBFN were used to measure the blocks in the coronary artery.

X-ray diffractometry is a unique technique and that the X-ray diffraction patterns which depict the structure of the steel sheets during processing with the features extracted, that they serve directly as a signature which is very complicated. X-ray diffraction (XRD) techniques are a type of non-destructive method of investigation to identify the flaws during the fabrication of steel sheets. X-ray diffraction is comparatively simple and can be effectively used for the examination and identification of flaws during the rolling process of steel sheets. XRD technique finds application in various fields like textile industry, forensic, qualitative and quantitative phase analysis of poly crystalline material, to infer overall properties of the fiber and measure the degree of crystalline nature. It is extensively used to explore areas like material science, chemistry and in industry for research and quality control. This effective method gains novelty by combining the signal processing algorithms like multiple threshold based Fast Fourier Transform (FFT) and Artificial neural network (ANN) trained with Back Propagation Algorithm (BPA) thereby offering an automated system for online monitoring during fabrication of flawless metal sheets. The hot rolled steel sheets for three categories namely, flawless, moderate flaw and extreme flaw conditions are obtained from the XRD pattern. Then multiple thresholds are incorporated to identify the peak position, peak width and peak intensity. The FFT algorithm computes the power spectrum which is used as features to identify the flaws in the steel sheets during cold rolling process. The extracted features are used as inputs to train the ANN with BPA whose performance is evaluated to be 90% efficient. (C) 2020 The Authors. Published by Elsevier Ltd.

Breast cancer is the lethal form of cancers as it affects the adjacent organs like lungs, liver and heart very easily. The tumors in the nodules of mammary glands serve as the cause for the development of a malignant tumor. Magnetic Resonance Imaging (MRI) images are used to detect breast cancer. The objective of this work is to develop a structured scheme to analyze and evaluate the probabilities of breast cancer with the help of a typical user friendly image processing algorithms. The novelty of this work is that it has a well-developed strategy for breast cancer detection using high performance image based machine learning algorithms to extract the variations in intensity levels at the pretreatment stage followed by segmentation and feature extraction from the region of interest, present in the breast nodules. This method uses images from the open source database like The Cancer Imaging Archive (TCIA). The pretreatment of MR images processing comprises filtering for noise removal and edge detection to extract the Region of Interest (RoI). The power spectrum of the MR images is evaluated and they play an important role to increase the sensitivity in identifying the breast tumors. These power spectrum coefficients extracted using Discrete Fourier Coefficients are used as distinct input feature set for training the Radial Basis Function Network (RBFN) to detect, identify and cross validate the activity of the drug (ethanolic extract) prepared from the traditional plant from the leaves of Excoecaria agallocha (EEEA), tested on a laboratory scale to investigate the probability of cytotoxic activity along with the anti-progressive nature to treat the apoptosis initiation and Cell cycle detention in the breast cancer MCF-7 cell lines causing malignancy in the nodules and ducts of mammary glands. The Apoptosis assessment is found to be nearly 99%.

The aim is to suggest a control scheme for the wind mills which convert wind energy to electrical energy. The functioning of the governing scheme is characterised by incorporating it to a doubly fed induction generator (DFIG). The stationary part of the DFIG is unswervingly linked to the electric network. The rotating part is allied to this electric network all the way through a back-to-back AC-DC-AC PWM converter. Fuzzy logic is used to acquire features using decision making logic which as human-like flexibility. The FLC provides a crisp and smooth control action. The governing process of the converter on rotating part is comprehended by stationary magnetic flux to adjust the performance of the fuzzy logic controller (FLC). The FLC is opted to have an intelligent speed control. To enable a level direct current voltage and to guarantee a pure sine wave for the current in the grid side a Grid Side Converter (GSC) is used which is controlled using FLC. The accuracy of the FLC used for the control of DFIG has a quick vibrant retort with almost unsteady error value once evaluated with the scheme using conformist proportional integral (PI) controller. Image processing algorithms are used to track the blade sweep and angular velocity. The entire monitoring is implemented using ATmega processor and incorporated in cloud service for online monitoring.

The focus of this research would be efficient monitoring of windmills using Image processing and Artificial Neural Network (ANN) based wind speed estimation. The related parameters to wind speed include humidity, atmospheric temperature and pressure apart from wind velocity. The identification of potential windmills plays an important role with respect to grid integration. The camera as a vision sensor is used to capture the video of the rotating windmills. The features extracted from captured images which are initially preprocessed. The preprocessing includes noise removal using median filter. Various features are used to test and train the Feed Forward Neural Network (FFNN) trained with Back Propagation Algorithm (BPA) which identifies the potential windmills. If minimum output is not obtained from windmills then the wind velocity prediction will be used to predict the wind speed, angular velocity and blade sweep area which is of great importance for wind power output maximization. To yield the targeted wind power an appropriate control strategy using Fuzzy Logic Control (FLC) is used. Now that the output of this integrated scheme is compared with Weibull model for achieving optimal wind energy output.

Wind energy has become a main challenge of conventional relic fuel energy, chiefly with the flourishing operation of multi-megawatt sized wind turbines. Though, wind with sensible speed is not sufficiently sustainable all over to construct an inexpensive wind farm. The probable site has to be systematically investigated at least with respect to wind speed profile and air density. Modelling and forecast of wind speed are indispensable rudiments in the setting and sizing of wind power applications. In this study, the sketch of wind speed in Mediterrean Sea of Turkey is modelled using artificial neural network (ANN). The aim of this project work is to show that a feed forward neural network can symbolize a practical tool to cautiously monitor the wind speed output. Simulation results can be reported, showing that the estimated wind speed values can be in good concurrence with the investigational values.