Search

Assessing the structural integrity of ageing structures that are affected by climate-induced stressors, challenges traditional engineering methods. The reason is that structural degradation often initiates and advances without any notable warning until visible severe damage or catastrophic failures occur. An example of this, is the conventional inspection methods for prestressed concrete bridges which fail to interpret large permanent deflections because the causes—typically tendon loss—are barely visible or measurable. In many occasions, traditional inspections fail to discern these latent defects and damage, leading to the need for expensive continuous structural health monitoring towards informed assessments to enable appropriate structural interventions. This is a capability gap that has led to fatalities and extensive losses because the operators have very little time to react. This study addresses this gap by proposing a novel machine learning approach to inform a rapid non-destructive assessment of bridge damage states based on measurable structural deflections. First, a comprehensive training dataset is assembled by simulating various plausible bridge damage scenarios associated with different degrees and patterns of tendon losses, the integrity of which is vital for the health of bridge decks. Second, a novel General Regression Neural Network (GRNN)-based cascade ensemble model, tailored for predicting three interdependent output attributes using limited datasets, is developed. The proposed cascade model is optimised by utilising the differential evolution method. Modelling and validation were conducted for a real long-span bridge. The results confirm the efficacy of the proposed model in accurately identifying bridge damage states when compared to existing methods. The model developed demonstrates exceptional prediction accuracy and reliability, underscoring its practical value in non-destructive bridge damage assessment, which can facilitate effective restoration planning.

This paper presents a parallel approach to the Levenberg-Marquardt algorithm (LM). The use of the Levenberg-Marquardt algorithm to train neural networks is associated with significant computational complexity, and thus computation time. As a result, when the neural network has a big number of weights, the algorithm becomes practically ineffective. This article presents a new parallel approach to the computations in Levenberg-Marquardt neural network learning algorithm. The proposed solution is based on vector instructions to effectively reduce the high computational time of this algorithm. The new approach was tested on several examples involving the problems of classification and function approximation, and next it was compared with a classical computational method. The article presents in detail the idea of parallel neural network computations and shows the obtained acceleration for different problems.

Background: Today, IoT-based systems are widely used in various applications. Intellectual analysis of the data collected by IoT devices is an important task for the efficient and successful functioning of such systems. In particular, the reliability of such kind of analysis has greatly influenced the ability to partially or fully automate certain processes or subsystems. However, imperfect devices of data collection, transportation errors, etc. lead to gaps in the data. Such incompetency makes an effective intellectual data analysis for the specific tasks impossible. That is why there must be a scientific solution to this problem of effectively filling the gaps in the data collected by the IoT sensors. Methods: The authors propose a new prediction method for missing data recovery tasks based on the use of General Regression Neural Networks (GRNN). Results: The possibility of approximation and partial elimination of the error of this computational intelligence tool has been analytically proved. A prediction method based on the use of GRNN was developed. The optimal parameters of the developed method were selected. The simulation of its work was performed to solve the problem of recovery of missing sensor data in the air pollution monitoring dataset. Conclusion: The highest accuracy of the developed method in comparison with other methods of this class was experimentally established. All advantages and disadvantages are described. Perspectives of further research are outlined.

Nowadays, the fast development of hardware for IoT-based systems creates appropriate conditions for the development of services for different application areas. As we know, the large number of multifunctional devices, which are connected to the Internet is constantly increasing. Today, most of the IoT devices just only collect and transmit data. The huge amount of data produced by these devices requires efficient and fast approaches to its analysis. This task can be solved by combining Artificial Intelligence and IoT tools. Essentially, AI accelerators can be used as a universal sensor in IoT systems, that is, we can create Artificial Intelligence of Things (AIoT). AIoT can be considered like a movement from data collection to knowledge aggregation. AIoT-based systems are being widely implemented in many high-tech industrial and infrastructure systems. Such systems are capable of providing not only the ability to collect but also analyse various aspects of data for identification, planning, diagnostics, evaluation, monitoring, optimization, etc., at the lower level in the entire system's hierarchy. That is, they are able to work more efficiently and effectively by generating the knowledge that is needed for real-time analytics and decision-making in some application areas.

This paper proposes mathematical models of the process of information interaction between a set of subjects (IoT devices) and a base station in the 5G-IoT ecosystem. The researched process, in this case, is represented by a queuing system with a flow of new incoming requests with needs for the desired volumes of system resources and a flow of service signals, the receipt of which initiates redefinition of allocated for received incoming requests volumes of system resources. A controlled parameter in the created queuing system is the acceptance or rejection of new incoming requests. The priority is to support active sessions of subject-system interaction initiated for the received incoming requests. We consider two functional scenarios, which differ asynchronously or synchronously when a service signal comes to the base station’s front-end interface. The receipt marks a partial or complete redefinition of system resources involved in supporting active sessions of subject-system interaction. Formalized in the basis of a single queuing system, mathematical models of development of these functional scenarios allow calculating the probability of rejection of a new incoming request and the percentage of occupied system resources, combined in the metrics of quality indicators, in the conditions of asynchronous or synchronous change of the spatial location relative to the base station of the end IoT devices, the received input requests from which are active in the 5G-IoT ecosystem.