Feature Extraction for Traffic Incident Detection Using Wavelet Transform and Linear Discriminant Analysis

To eliminate false alarms, an effective traffic incident detection algorithm must be able to extract incident-related features from the traffic patterns. A robust feature-extraction algorithm also helps reduce the dimension of the input space for a neural network model without any significant loss of related traffic information, resulting in a substantial reduction in the network size, the effect of random traffic fluctuations, the number of required training samples, and the computational resources required to train the neural network. This article presents an effective traffic feature-extraction model using discrete wavelet transform (DWT) and linear discriminant analysis (LDA). The DWT is first applied to raw traffic data, and the finest resolution coefficients representing the random fluctuations of traffic are discarded. Next, LDA is employed to the filtered signal for further feature extraction and reducing the dimensionality of the problem. The results of LDA are used as input to a neural network model for traffic incident detection.     

Enhancing Neural Network Traffic Incident-Detection Algorithms Using Wavelets

Researchers have presented freeway traffic incident-detection algorithms by combining the adaptive learning capability of neural networks with imprecision modeling capability of fuzzy logic. In this article it is shown that the performance of a fuzzy neural network algorithm can be improved through preprocessing of data using a wavelet-based feature-extraction model. In particular, the discrete wavelet transform (DWT) denoising and feature-extraction model proposed by Samant and Adeli (2000) is combined with the fuzzy neural network approach presented by Hsiao et al. (1994). It is shown that substantial improvement can be achieved using the data filtered by DWT. Use of the wavelet theory to denoise the traffic data increases the incident-detection rate, reduces the false-alarm rate and the incident-detection time, and improves the convergence of the neural network training algorithm substantially.     

Wavelet-Clustering-Neural Network Model for Freeway Incident Detection

Abstract:   An improved freeway incident-detection model is presented based on speed, volume, and occupancy data from a single detector station using a combination of wavelet-based signal processing, statistical cluster analysis, and neural network pattern recognition. A comparative study of different wavelets (Haar, second-order Daubechies, and second- and fourth-order Coifman wavelets) and filtering schemes is conducted in terms of efficacy and accuracy of smoothing. It is concluded that the fourth-order Coifman wavelet is more effective than other types of wavelets for the traffic incident detection problem. A statistical multivariate analysis based on the Mahalanobis distance is employed to perform data clustering and parameter reduction to reduce the size of the input space for the subsequent step of classification by the Levenberg–Marquardt backpropagation (BP) neural network. For a straight two-lane freeway using real data, the model yields an incident detection rate of 100%, false alarm rate of 0.3%, and detection time of 35.6 seconds.     

Data Fusion of Fixed Detector and Probe Vehicle Data for Incident Detection

An important feature of many advanced traveler information systems (ATIS) is real-time information about incidents on the street network. This paper describes a system for automatically detecting incidents for such an ATIS developed using artificial neural networks and statistical prediction methods. The system monitors traffic conditions using two types of data: inductive loop detectors (ILDs) and vehicle probes. For both neural network and statistical methods, incident detection is accomplished using two approaches: by processing traffic input data directly and by processing the output of specialized algorithms that detect incidents using information from each data source. Analysis data generated from a simulation of a typical suburban signalized major arterial street are used. Different model configurations are examined and tested to identify the input variables and methods that are the best predictors of incident occurrence. The neural network approaches consistently perform at least as well as the discriminant analysis models, especially when results are adjusted to avoid false alarms.     

A deep marked graph process model for citywide traffic congestion forecasting

Forecasting citywide traffic congestion on large road networks has long been a nontrivial research problem due to the challenge of modeling complex evolution patterns of congestion in highly stochastic traffic environments. Arguing that purely data-driven methods may not perform well for congestion forecasting, we propose a deep marked graph process model for predicting the congestion indices and the occurrence time of traffic congestion events for complex signalized road networks. Traffic congestion is considered as a nonrigorous spatiotemporal extreme event. We extend the traditional point process model by integrating a specially designed spatiotemporal graph convolutional network. This hybrid strategy takes advantage of the simple form of the point process model as well as the ability of graph neural networks to emulate the evolution of congestion. Experiments on real-world congestion data sets show that the proposed method outperforms state-of-the-art baseline methods, yielding satisfactory prediction results on a large signalized road network with superior computational efficiency.     

A Linear Model for the Continuous Network Design Problem

Abstract:   This article is concerned with the continuous network design problem on traffic networks, assuming system optimum traffic flow conditions and time-dependent demand. A linear programming formulation is introduced based on a dynamic traffic assignment (DTA) model that propagates traffic according to the cell transmission model. The introduced approach is limited to continuous link improvements and does not provide for new link additions. The main contribution of the article is to provide an analytical formulation for network design that accounts for DTA conditions that can be used for further analysis and extensions. The model is tested on a single destination example network, resembling a freeway corridor, for various congestion levels, loading patterns and budget sizes, to demonstrate the simplicity and effectiveness of the approach.     

An Approach to Dynamic Classification of Traffic Flow Patterns

Abstract: Analyses on the dynamics of traffic flow, ranging from intersection flows to network‐wide flow propagation, require accurate information on time‐varying local traffic flows. To effectively determine the flow performance measures and consequently the congestion indicators of segmented road pieces, the ability to process such data in real time is out of the question. In this article, a dynamic approach to specify flow pattern variations is proposed mainly concentrating on the incorporation of neural network theory to provide real‐time mapping for traffic density simultaneously in conjunction with a macroscopic traffic flow model. To deal with the noise and the wide scatter of raw flow measures, a filtering is applied prior to modeling processes. Filtered data are dynamically and simultaneously input to processes of neural density mapping and traffic flow modeling. The classification of flow patterns over the fundamental diagram, which is dynamically plotted with the outputs of the flow modeling subprocess, is obtained by considering the density measure as a pattern indicator. Densities are mapped by selected neural approximation method for each simulation time step considering explicitly the flow conservation principle. Simultaneously, mapped densities are matched over the fundamental diagram to specify the current corresponding flow pattern. The approach is promising in capturing sudden changes on flow patterns and is open to be utilized within a series of intelligent management strategies including especially nonrecurrent congestion effect detection and control.     

Neural Networks for Evaluating CPT Calibration Chamber Test Data

Abstract: The feasibility of using neural network models for evaluating CPT calibration chamber test data is investigated. The backpropagation neural network algorithm was used to analyze the data. After learning from a set of randomly selected patterns, the neural network model was able to produce reasonably accurate predictions for patterns not included in the training set. The neural network performance was found to be simpler and more effective than regression analysis for modeling the CPT test data. Correlations between the cone measurements and the engineering properties of sand can be developed using the generalization capabilities of the neural network.     

Dynamic origin‐destination flow estimation using automatic vehicle identification data: A 3D convolutional neural network approach

Dynamic origin‐destination (OD) flow estimation is one of the most fundamental problems in traffic engineering. Despite numerous existing studies, the OD flow estimation problem remains challenging, as there is large dimensional difference between the unknown values to be estimated and the known traffic observations. To meet the needs of active traffic management and control, accurate time‐dependent OD flows are required to understand time‐of‐day traffic flow patterns. In this work, we propose a three‐dimensional (3D) convolution‐based deep neural network, “Res3D,” to learn the high‐dimensional correlations between local traffic patterns presented by automatic vehicle identification observations and OD flows. In this paper, a practical framework combining simulation‐based model training and few‐shot transfer learning is introduced to enhance the applicability of the proposed model, as continuously observing OD flows could be expensive. The proposed model is extensively tested based on a realistic road network, and the results show that for significant OD flows, the relative errors are stable around 5%, outperforming several other models, including prevalent neural networks as well as existing estimation models. Meanwhile, corrupted and out‐of‐distribution samples are generated as real‐world samples to validate Res3D's transferability, and the results indicated a 60% improvement with few‐shot transfer learning. Therefore, this proposed framework could help to bridge the gaps between traffic simulations and empirical cases.     

Traffic Prediction and Management via RBF Neural Nets and Semantic Control

The objective of this work has been to develop layers of control and optimization modules for the purpose of urban traffic management. We utilize the semantic control paradigm to model both the macrolevel (traffic control) and the microlevel (vehicle path planning and steering control). A semantic controller consists of three modules for identification, goal selection, and adaptation, respectively. This hierarchical structure has been used successfully at the Center for Optimization and Semantic Control to solve complex, nonlinear, and time-varying problems. In our previous work we have used a judicious combination of artificial intelligence, optimization, and control systems.
The focus of this paper is the identifier module, which performs "system identification," i.e., determines the road network congestion level. Traffic flow can be characterized as a nonlinear stochastic process where linear prediction models such as linear regression are not suitable. However, neural network techniques may provide an effective tool for data-based modeling and system identification. The radial basis function neural network (RBFNN) is an attractive tool for nonlinear time-series modeling and traffic-flow prediction. The goal selector module that finds the shortest path is also discussed in some detail.
A model of the highway system, based on historical data provided by Missouri Highway and Transportation Department (MoHTD), has been developed. The prediction and planning system is evaluated using the traffic-flow data from nine sensors located on the highway in the St. Louis metropolitan area.     

Burst detection using hydraulic data from water distribution systems with artificial neural networks

This paper presents research into the application of artificial neural networks (ANNs) for analysis of data from sensors measuring hydraulic parameters (flow and pressure) of the water flow in treated water distribution systems. Two neural architectures (static and time delay) are applied for time series pattern classification from the perspective of detecting leakage. Results are presented using data from an experimental site in a distribution system of a UK water company in which bursts were simulated by hydrant flushing. Field trials have shown how ANNs can be used effectively for a leakage detection task. Both static and time delay ANNs learned patterns of leaks/bursts. The time delay neural network showed improved performance over the static network. It is concluded that the effectiveness of an ANN in discovering relationships within the data is dependent upon two key factors: availability of sufficient exemplars and data quality.     

基于Elman经网络的深基坑变形智能监测

Elman神经网络是一种典型的动态递归神经网络，它既可以学习空域模式，又可以学习时域模式，能使训练好的网络具有非线性和动态特性。文章采用11个输入单元和1个输出单元Elman神经网络。利用Maflab提供的神经网络工具箱编写了Elman神经网络程序，可以通过几种监测数据来推测另一种监测数据。最后以一幢超高层建筑的深基坑工程为例．说明了Elman神经网络方法用于深基坑变形的预测具有较好的可靠性，通过不断加入新的数据，Elman神经网络程序所能判断的数据类型和精度均不断提高。     

Using Machine Learning, Neural Networks, and Statistics to Predict Corporate Bankruptcy

Recent literature strongly suggests that machine learning approaches to classification outperform "classical" statistical methods. We make a comparison between the performance of linear discriminant analysis, classification trees, and neural networks in predicting corporate bankruptcy. Linear discriminant analysis represents the "classical" statistical approach to classification, whereas classification trees and neural networks represent artificial intelligence approaches. A proper statistical design is used to be able to test whether observed differences in predictive performance are statistically significant. The data set consists of a collection of 576 annual reports from Belgian construction companies. We use stratified 10–fold cross–validation on the training set to choose "good" parameter values for the different learning methods. The test set is used to obtain an unbiased estimate of the true prediction error. Using rigorous statistical testing, we cannot conclude that in the case of the data set studied, one learning method clearly outperforms the other methods.     

Pipe crack identification based on the energy method and committee of neural networks

A crack identification method using an equivalent bending stiffness for cracked beam and committee of neural networks is presented. The equivalent bending stiffness is constructed based on an energy method for a straight thin-walled pipe, which has a through-the-thickness crack, subjected to bending. Several numerical analysis for a steel cantilever pipe using the equivalent bending stiffness are carried out to extract the natural frequencies and mode shapes of the cracked beam. The extracted modal properties are used in constructing a training patterns of a neural network. The input to the neural network consists of the modal properties and the output is composed of the crack location and size. Multiple neural networks are constructed and each individual network is trained independently with the different initial synaptic weights. Then, the estimated crack locations and sizes from different neural networks are averaged. Crack detection is carried out for 16 damage cases using the proposed method, and the identified crack locations and sizes agree reasonably well with the exact values.     

A neuro-wavelet technique for seismic damage identification of cantilever structures

A method based on artificial neural networks and wavelet transform is proposed for identifying seismic-induced damage of cantilever structures. In the proposed method, response accelerations are measured at strategically selected locations. To extract damage-induced sharp transitions from the measured signals, they are decomposed by continuous wavelet transform. The size of the decomposed signals is reduced by principal component analysis (PCA). Principal components obtained from PCA are fed to a set of neural networks to identify damage. The proposed algorithm is applied to a tall airport traffic control tower by means of numerical simulations. The obtained results show that the proposed method effectively identifies seismic-induced damage, and the noise intensity has a negligible effect on the predicted results. Moreover, the trained neural network system is able to predict the seismic-induced damage of unseen samples well.     

Using artificial neural networks for calculation of temperatures in timber under fire loading

Artificial neural networks have been used in recent years as a tool to model properties and behavior of materials in many areas of civil engineering applications. Because of their ability to learn and adapt they can be used to find complex relations between different properties. In the present paper artificial neural networks are used for predicting the temperatures in timber under fire loading. The artificial neural network model has been trained and tested using available numerical results obtained using design methods of Eurocode 5 for the calculation of temperatures in timber under fire loading. A multilayer feed forward network has been used with input data arranged in a format of three input parameters that cover the density of timber, the time of fire exposure and the distance from exposed side and the output parameter being the temperature in timber. The training and testing results in the neural network model have shown that neural networks can accurately calculate the temperature in timber members subjected to fire.     

A hybrid neural network strategy for identification of structural parameters

This paper presents a multistage identification scheme for structural damage detection using modal data. Previous studies of damage assessment using neural networks mostly involved training a backpropagation neural network (BPN) to learn damage patterns that were obtained either experimentally or by simulation for different damage cases. Damage identification for large structures, especially those involving multiple member damage, could result in large training data sets that require a large BPN and consequently greater computational effort. The proposed scheme involves using a counterpropagation neural network (CPN) in the first stage for sorting the training data into clusters and giving an approximate guess of the damage extent within a very short time. After an approximate estimate of the damage is obtained, a new set of training patterns of reduced size is generated using the CPN prediction. In the second stage, a BPN trained with the Levenberg–Marquardt algorithm is used to learn the new training data and predict a more accurate result. A superior convergence and a substantial decrease in central processing unit (CPU) time are observed for three numerical examples.     

Patrol Route Planning for Incident Response Vehicles under Dispatching Station Scenarios

Traffic incidents often contribute to major safety concerns, impose additional congestion in the neighboring transportation networks, and induce indirect costs to economy. As roughly a third of traffic crashes are secondary accidents, effective incident management activities are critical, especially on roadways with high traffic volume, to detect, respond to, and clean up incidents in a timely fashion, which supports safety constraints and restores traffic capacity in the transportation network. Hence, it is beneficial to simultaneously plan for first respondents’ dispatching station location and patrol route design to mitigate congestion. This article presents an optimal route planning for patrolling vehicles to facilitate quick response to potential accidents. A mixed‐integer nonlinear program is proposed that minimizes the respondents’ patrolling travel cost based on the expected maximum response time from each arbitrary location to all incident locations (a.k.a. hotspots) with various incident occurrence probabilities. We have developed a column generation‐based solution technique to solve the route optimization model under different station design scenarios. To investigate the impact of dispatching station design on the routing cost, an integrated genetic algorithm framework with embedded continuous approximation approach is developed that reduces the complexity of the hybrid location design and route planning problem. Numerical experiments on hypothetical networks of various sizes are conducted to indicate the performance of the proposed algorithm and to draw managerial insights. The models and solution techniques, developed in this article, are applicable to a number of network problems that simultaneously involve routing and facility location choices.     

An Implementation of the Mode-Split Traffic-Assignment Method

A software called Optimal Traffic Signal Control System (OTSCS) was developed by us for testing the feasibility of dynamically controlling a traffic signal by finding optimal signal timing to minimize delay at signalized intersections. It also was designed as a research tool to study the learning behavior of artificial neural networks and the properties of heuristic search methods. It consists of a level-of-service evaluation model that is based on an artificial neural network and a heuristic optimization model that interacts with the level-of-service evaluation model. This article discusses the latter model, named the Optimal Traffic Signal Timing Model (OTSTM). The OTSTM was applied to determine optimal signal timing of two-phase traffic signals to evaluate the model's performance. Two search methods were employed: a depth-first search method (an enumeration method) and a direction-search method that the authors developed. It was found that the OTSTM with the direction search resulted in "optimal" signal timings similar to the depth-first search, which would always produce a global optimal timing. Yet the cost of the direction search, as measured by the CPU time of the computer used for analysis, was found to be much less than the cost of obtaining an optimal solution by the depth-first search cases—more than 10 times less. The study showed that once the artificial neural network is properly trained, heuristic optimal signal timing combined with artificial networks can be used as a decision-support tool for dynamic signal control. This article demonstrates how OTSTM can quickly find an optimal signal-timing solution for two-phase traffic signals.     

Multiscale computation on feedforward neural network and recurrent neural network

Homogenization methods can be used to predict the effective macroscopic properties of materials that are heterogenous at micro- or fine-scale. Among existing methods for homogenization, computational homogenization is widely used in multiscale analyses of structures and materials. Conventional computational homogenization suffers from long computing times, which substantially limits its application in analyzing engineering problems. The neural networks can be used to construct fully decoupled approaches in nonlinear multiscale methods by mapping macroscopic loading and microscopic response. Computational homogenization methods for nonlinear material and implementation of offline multiscale computation are studied to generate data set. This article intends to model the multiscale constitution using feedforward neural network (FNN) and recurrent neural network (RNN), and appropriate set of loading paths are selected to effectively predict the materials behavior along unknown paths. Applications to two-dimensional multiscale analysis are tested and discussed in detail.