Traffic flow prediction is an essential part of the intelligent transport system. This is the accurate estimation of traffic flow in a given region at a particular interval of time in the future. The study of traffic forecasting is useful in mitigating congestion and make safer and cost-efficient travel. While traditional models use shallow networks, there has been an exponential growth in the number of vehicles in recent times and these traditional machine learning models fail to work in current scenarios. In our paper, we review some of the latest works in deep learning for traffic flow prediction. Many deep learning architectures include Convolutional Neural Network (CNN), Recurrent Neural Network (RNN), Long Short-Term Memory (LSTM), Restricted Boltzmann Machines (RBM), and Stacked Auto Encoder (SAE). These deep learning models use multiple layers to extract higher level of features from raw input progressively. The latest deep learning models developed to tackle this very problem are reviewed and due to the complexity of transport networks, this review gives the reader information about how various factors influence these models and what models work best in different scenarios.
7 March 2022 in Peer-to-Peer Networking and Applications
With the advancement of sensor technologies, Wireless Sensor Networks (WSN) are envisioning a rich variety of promising services in many fields. WSN is formed by the deployment of sensor nodes in the regions of interest using a deterministic or random deployment strategy. The random deployment strategy is more suitable in large monitoring areas and harsh environments. But, in this type of deployment, coverage holes and disconnected networks can exist. Important events may get unnoticed reducing the reliability of the networks. Since the locations of nodes are unknown in a randomly deployed WSN, it is difficult to locate the holes. For this, we are proposing a localization and deployment model. The localization algorithm uses Arithmetic Optimization Algorithm (AOA) and the results of this algorithm are further used to develop a deployment model to achieve a completely connected network. This algorithm is tested in various fields. The algorithm is able to localize nodes accurately and identify the coverage holes with an error rate of less than 0.27% when the Average Localization Error (ALE) is within 5m.
A control valve plays a very significant role in the stable and efficient working of a control loop for any process. In a fluid flow process, the probability of failure of a control valve may increase for many reasons pertaining to a flow process such as high pressures at the inlet, different properties of the liquid flowing through the pipe, mechanical issue related to a control valve, ageing, etc. A method to detect faults in the valve can lead to better stability of the control loop. In the proposed work, a technique is developed to determine the fault in a pneumatic control valve by analyzing the vibration data at the outlet of the valve. The fault diagnosis of the valve is carried out by analyzing the change in vibration of the pipe due to the change in flow pattern induced by the control valve. The faults being considered are inflow and insufficient supply pressure faults. Vibration data obtained is processed using a signal processing technique like amplification, Fourier transform, etc. The support vector machine (SVM) algorithm is used to classify the vibration data into two classes, one normal and the other faulty. The designed algorithm is trained to identify faults and subjected to test with a practical setup; test results show an accuracy of 97%.