The article describes the development of two neural network algorithms for recognizing objects of the railway infrastructure in video images. Both algorithms are aimed at improving railway traffic safety. One algorith...
详细信息
The article describes the development of two neural network algorithms for recognizing objects of the railway infrastructure in video images. Both algorithms are aimed at improving railway traffic safety. One algorithm detects foreign objects on railway tracks and objects relating to the railway infrastructure. The other algorithm implements the semantic segmentation of main and auxiliary railway tracks, as well as trains within the visible range of the locomotive. The algorithms are implemented based on convolutional neuralnetworks (CNN) YOLO and U-Net. The CNN is trained and tested using the image database of the Research Institute of Information, Automation and Communications in Railway Transport. The experimental studies conducted are aimed at increasing the efficiency of algorithms for object detection and segmentation through the use of data augmentation methods and additional preprocessing, as well as selecting an architecture and optimal network hyperparameters. The detection algorithm works in real time, achieving an average accuracy of 64 % for 11 object classes according to the mAP metric. The operating speed of the semantic segmentation algorithm is 5 frames/s, the average accuracy for three classes of to the IoU metric is 92 %.
With the continuous advancement of computer technology and sensor technology, rehabilitation robots have shown great potential in the rehabilitation treatment of limb movement disorders. This paper designs a rehabilit...
详细信息
With the continuous advancement of computer technology and sensor technology, rehabilitation robots have shown great potential in the rehabilitation treatment of limb movement disorders. This paper designs a rehabilitation robot based on a neuralnetwork algorithm to improve the rehabilitation effect of patients with limb movement disorders. The robot consists of a four-axis robotic arm, a three-finger gripper, a wheeled chassis and a variety of sensors. The control program written in C language and the host computer program written in Python language realize the control of motion and force, and the neuralnetwork algorithm is used to accurately adjust the position and force. In the control system design, the back propagation algorithm is used to train the neuralnetwork, and it is optimized in combination with multiple data to ensure that the robot can accurately track and assist the patient's limb movement. In order to verify the effect of the rehabilitation robot, this paper conducted an experimental evaluation. The experimental group was patients who received rehabilitation training based on the neuralnetwork algorithm, and the control group was patients who received traditional physical therapy. Through the evaluation of multiple indicators such as the 10-meter walking test, Berg balance scale, JebsenTaylor hand function test and MOS SF-36 health questionnaire, the experimental results showed that the improvement of the experimental group in each test was significantly better than that of the control group. Especially in the Berg balance scale and Jebsen-Taylor hand function test, the scores of the experimental group were significantly improved, indicating that the rehabilitation robot plays an important supporting role in the rehabilitation training of patients with limb movement disorders and has broad application prospects.
This work discusses the predictable control of plasma-assisted physical vapor deposition (PVD) of coatings. The multiple process parameters and the instability of the nonequilibrium ion plasma system create substantia...
详细信息
This work discusses the predictable control of plasma-assisted physical vapor deposition (PVD) of coatings. The multiple process parameters and the instability of the nonequilibrium ion plasma system create substantial obstacles to the wide industrial application of promising multicomponent functional coatings. Here we propose a solution to this problem, which includes: creation of a database of diamond-like carbon (DLC) coatings to identify a limited set of adjustable process control parameters, determination of how these parameters affect the coating properties, analysis of the revealed effects using statistical methods and neural network algorithms, and use of the results for the predictable tuning of specified coating properties. The object of research is original DLC coatings whose structure is stabilized with nitrogen instead of conventionally used hydrogen. The experimental database of DLC coatings is created based on our previous studies and includes structural, morphological and architectural characteristics of coatings, various types of substrates, sublayers, physical, mechanical and tribological properties, and various combinations of coating deposition parameters. A specific problem is solved to determine the influence of deposition parameters such as chamber pressure P, stabilizer content (% nitrogen), ion flux rate (coil current lambda) and deposition time t on hardness H and elastic modulus E of coatings. Based on the results obtained, the deposition parameters are optimized so as to obtain predictable strength values of the formed carbon coating. The optimization procedure is developed using both classical statistical methods and modern algorithms of ridge regression, randomized trees (ExtraTrees), and a fully connected neuralnetwork (multilayer perceptron MLP).
A demonstration is presented of how to use neural network algorithms to schedule classes in an educational institute. Such a scheduling problem is basically a graph-coloring or graph-partitioning problem which belongs...
详细信息
A demonstration is presented of how to use neural network algorithms to schedule classes in an educational institute. Such a scheduling problem is basically a graph-coloring or graph-partitioning problem which belongs to the large class of NP (nondeterministic polynomial time)-complete problems, and it is difficult to solve. G.C. Fox and W. Furmanski (1988) proposed some neural network algorithms to decompose loosely synchronous problems onto parallel machines. The author adopts these algorithms to schedule timetables. The algorithms can be implemented on a digital computer or on analog neuralnetworks
In the traditional manual tool management methods used in the electric power field, efficiency is relatively low. Additionally, in small-scale and complex electric power operation environments, tool loss occurs freque...
详细信息
A method for evaluating the thermophysical characteristics of the torch is developed. Mathematically the temperature at the end of the zone of active combustion based on continuous distribution functions of particles ...
详细信息
A method for evaluating the thermophysical characteristics of the torch is developed. Mathematically the temperature at the end of the zone of active combustion based on continuous distribution functions of particles of solid fuels, in particular coal dust. The particles have different average sizes, which are usually grouped and expressed as a fraction of the total mass of the fuel. The authors suggest taking into account the sequential nature of the entry into the chemical reactions of combustion of particles of different masses. In addition, for the application of the developed methodology, it is necessary to divide the furnace volume into zones and sections. In particular, the initial section of the torch, the zone of intense burning and the zone of afterburning. In this case, taking into account all the thermophysical characteristics of the torch, it is possible to make a thermal balance of the zone of intense burning. Then determines the rate of expiration of the fuel-air mixture, the time of combustion of particles of different masses and the temperature at the end of the zone of intensive combustion. The temperature of the torch, the speed of flame propagation, and the degree of particle burnout must be controlled. The authors propose an algorithm for controlling the thermophysical properties of the torch based on neural network algorithms. The system collects data for a certain time, transmits the information to the server. The data is processed and a forecast is made using neural network algorithms regarding the combustion modes. This allows to increase the reliability and efficiency of the combustion process. The authors present experimental data and compare them with the data of the analytical calculation. In addition, data for certain modes are given, taking into account the system's operation based on neural network algorithms.
Background: Gestational diabetes mellitus (GDM) is globally recognized as a significant pregnancy-related condition, contributing to complex complications for both mothers and infants. Traditional glucose tolerance te...
详细信息
Background: Gestational diabetes mellitus (GDM) is globally recognized as a significant pregnancy-related condition, contributing to complex complications for both mothers and infants. Traditional glucose tolerance tests lack the ability to identify the risk of GDM in early pregnancy, hindering effective prevention and timely intervention during the initial stages. Objective: The primary objective of this study is to pinpoint potential risk factors for GDM and develop an early GDM risk prediction model using neuralnetworks to facilitate GDM screening in early pregnancy. Methods: Initially, we employed statistical tests and models, including univariate and multivariate logistic regression, to identify 14 potential risk factors. Subsequently, we applied various resampling techniques alongside a multi-layer perceptron (MLP). Finally, we evaluated and compared the classification performances of the constructed models using various metric indicators. Results: As a result, we identified several factors in early pregnancy significantly associated with GDM (p < 0.05), including BMI, age of menarche, age, higher education, folic acid supplementation, family history of diabetes mellitus, HGB, WBC, PLT, Scr, HBsAg, ALT, ALB, and TBIL. Employing the multivariate logistic model as the baseline achieved an accuracy and AUC of 0.777. In comparison, the MLP-based model using NearMiss exhibited strong predictive performance, achieving scores of 0.943 in AUC and 0.884 in accuracy. Conclusions: In this study, we proposed an innovative interpretable early GDM risk prediction model based on MLP. This model is designed to offer assistance in estimating the risk of GDM in early pregnancy, enabling proactive prevention and timely intervention.
To address the critical role of atmospheric temperature in climate change and disaster monitoring, enhancing measurement accuracy to 0.1 degrees C is essential. Current instruments are susceptible to radiation interfe...
详细信息
To address the critical role of atmospheric temperature in climate change and disaster monitoring, enhancing measurement accuracy to 0.1 degrees C is essential. Current instruments are susceptible to radiation interference, resulting in errors of approximately 1 degrees C. This study introduces a novel temperature sensor that improves accuracy by combining natural ventilation with forced ventilation. Silver-coated aluminum plates (95 % reflectivity) and white-coated deflectors (87 % reflectivity) minimize solar radiation errors. A neuralnetwork algorithm, along with CFD simulations, further corrects radiation errors under varying weather conditions. Field tests based on the 076B ventilation device demonstrate that this new sensor reduces the average radiation error to 0.02 degrees C, achieving a RMSE of 0.034 degrees C and a MAE of 0.028 degrees C. The correlation coefficient (r) with the reference temperature reached 0.999, demonstrating the sensor's high precision and providing an effective solution for reducing temperature measurement errors to below 0.1 degrees C. Um die entscheidende Rolle der atmosph & auml;rischen Temperatur im Klimawandel und bei der Katastrophen & uuml;berwachung zu adressieren, ist eine Verbesserung der Messgen & auml;uigkeit auf 0,1 degrees C unerl & auml;sslich. Aktuelle Instrumente sind anf & auml;llig f & uuml;r Strahlungsst & ouml;rungen, was zu Fehlern von etwa 1 degrees C f & uuml;hrt. Diese Studie stellt einen neuartigen Temperatursensor vor, der die Genauigkeit durch die Kombination von nat & uuml;rlicher Bel & uuml;ftung und erzwungener Bel & uuml;ftung verbessert. Silberbeschichtete Aluminiumplatten (95 % Reflektivit & auml;t) und wei ss beschichtete Abweiser (87 % Reflektivit & auml;t) minimieren Strahlungsfehler. Ein neuronales Netzwerk-Algorithmus sowie CFD-Simulationen korrigieren zus & auml;tzlich Strahlungsfehler unter variierenden Wetterbedingungen. Feldtests basierend auf dem 076B Bel & uuml;ftungsger & auml;t zeigen, d
The global use of freshwater has increased six-fold over the past 100 years and has been growing by about 1% per year since the 1980s. By 2021, 40% of the water bodies on which Spain depends were in poor condition due...
详细信息
The global use of freshwater has increased six-fold over the past 100 years and has been growing by about 1% per year since the 1980s. By 2021, 40% of the water bodies on which Spain depends were in poor condition due to overexploitation, pollution, or ecological deterioration. To address this challenging situation, a Monitoring and Mitigation Framework (MMF) was developed to minimize damage from contaminant spills, identify the location of contamination sources, and estimate the Chemical Oxygen Demand (COD) flow concentration rates. The framework is designed to detect environmental crimes, identify the companies responsible for contamination spills, and schedule mitigation actions in coordination with law enforcement authorities for future events. Although the framework includes real-time modules to identify contamination sources and estimate COD concentration rates, this manuscript focuses solely on the offline module, where Inverse Estimation (IE) algorithms form the core. The IE algorithms are based on neuralnetworks: a classification algorithm to locate the contamination source and a regression algorithm to estimate the COD flow concentration at the source location over time. These algorithms provide evidence to help local police specifically the Malaga City Council (Spain) identify potential companies involved in environmental crimes. The MMF was developed and tested at a Llobregat river reach (Barcelona, Spain), and it was considered 2621 test cases with different spill location and COD concentrations ranging from 0.1 to 80,000 mg/l. In those tests, IE algorithms were able to estimate COD concentration at source with a mean absolute error (MAE) lower than 2.6 mg/l or a Mean Square Relative Error (MSRE) lower than 10 mg/l, and it correctly identified the contamination source location in 85% of cases.
As the Internet of Things (IoT) technology develops rapidly, the interconnection of various devices and sensors has brought great convenience to people's lives and work. However, at the same time, IoT systems also...
详细信息
暂无评论