Optimization of traffic signal controls within equilibrium route choice models : final report to Illinois Department of Transpotation/

Cover title.

System for foreign exchange trading using genetic algorithms and reinforcement learning

Foreign exchange trading has emerged recently as a significant activity in many countries. As with most forms of trading, the activity is influenced by many random parameters so that the creation of a system that effectively emulates the trading process will be very helpful. A major issue for traders in the deregulated Foreign Exchange Market is when to sell and when to buy a particular currency in order to maximize profit. This paper presents novel trading strategies based on the machine learning methods of genetic algorithms and reinforcement learning.

Parameter Identification of a Fed-Batch Cultivation of S. Cerevisiae using Genetic Algorithms

Fermentation processes as objects of modelling and high-quality control are characterized with interdependence and time-varying of process variables that lead to non-linear models with a very complex structure. This is why the conventional optimization methods cannot lead to a satisfied solution. As an alternative, genetic algorithms, like the stochastic global optimization method, can be applied to overcome these limitations. The application of genetic algorithms is a precondition for robustness and reaching of a global minimum that makes them eligible and more workable for parameter identification of fermentation models. Different types of genetic algorithms, namely simple, modified and multi-population ones, have been applied and compared for estimation of nonlinear dynamic model parameters of fed-batch cultivation of S. cerevisiae.

Traffic congestion control in communication networks

Next generation networks are characterized by ever increasing complexity, intelligence, heterogeneous technologies and increasing user expectations. Telecommunication networks in particular have become truly global, consisting of a variety of national and regional networks, both wired and wireless. Consequently, the management of telecommunication networks is becoming increasingly complex. In addition, network security and reliability requirements require additional overheads which increase the size of the data records. This in turn causes acute network traffic congestions. There is no single network management methodology to control the various requirements of today's networks, and provides a good level of Quality of Service (QoS), and network security. Therefore, an integrated approach is needed in which a combination of methodologies can provide solutions and answers to network events (which cause severe congestions and compromise the quality of service and security). The proposed solution focused on a systematic approach to design a network management system based upon the recent advances in the mobile agent technologies. This solution has provided a new traffic management system for telecommunication networks that is capable of (1) reducing the network traffic load (thus reducing traffic congestion), (2) overcoming existing network latency, (3) adapting dynamically to the traffic load of the system, (4) operating in heterogeneous environments with improved security, and (5) having robust and fault tolerance behavior. This solution has solved several key challenges in the development of network management for telecommunication networks using mobile agents. We have designed several types of agents, whose interactions will allow performing some complex management actions, and integrating them. Our solution is decentralized to eliminate excessive bandwidth usage and at the same time has extended the capabilities of the Simple Network Management Protocol (SNMP). Our solution is fully compatible with the existing standards.

NUMERICAL SIMULATION STUDY OF VIBRATION MITIGATION EFFECTIVENESS OF TUNED MASS DAMPERS FOR TRAFFIC SIGNAL MAST ARM STRUCTURES

Fatigue damage in the connections of single mast arm signal support structures is one of the primary safety concerns because collapse could result from fatigue induced cracking. This type of cantilever signal support structures typically has very light damping and excessively large wind-induced vibration have been observed. Major changes related to fatigue design were made in the 2001 AASHTO LRFD Specification for Structural Supports for Highway Signs, Luminaries, and Traffic Signals and supplemental damping devices have been shown to be promising in reducing the vibration response and thus fatigue load demand on mast arm signal support structures. The primary objective of this study is to investigate the effectiveness and optimal use of one type of damping devices termed tuned mass damper (TMD) in vibration response mitigation. Three prototype single mast arm signal support structures with 50-ft, 60-ft, and 70-ft respectively are selected for this numerical simulation study. In order to validate the finite element models for subsequent simulation study, analytical modeling of static deflection response of mast arm of the signal support structures was performed and found to be close to the numerical simulation results from beam element based finite element model. A 3-DOF dynamic model was then built using analytically derived stiffness matrix for modal analysis and time history analysis. The free vibration response and forced (harmonic) vibration response of the mast arm structures from the finite element model are observed to be in good agreement with the finite element analysis results. Furthermore, experimental test result from recent free vibration test of a full-scale 50-ft mast arm specimen in the lab is used to verify the prototype structure’s fundamental frequency and viscous damping ratio. After validating the finite element models, a series of parametric study were conducted to examine the trend and determine optimal use of tuned mass damper on the prototype single mast arm signal support structures by varying the following parameters: mass, frequency, viscous damping ratio, and location of TMD. The numerical simulation study results reveal that two parameters that influence most the vibration mitigation effectiveness of TMD on the single mast arm signal pole structures are the TMD frequency and its viscous damping ratio.

Optimisation of a class of transportation problems using genetic algorithms

This paper presents a Genetic Algorithms (GA) approach to search the optimized path for a class of transportation problems. The formulation of the problems for suitable application of GA will be discussed. Exchanging genetic information in the sense of neighborhoods will be introduced for generation reproduction. The performance of the GA will be evaluated by computer simulation. The proposed algorithm use simple coding with population size 1 converged in reasonable optimality within several minutes.

Dynamic optimization of migration topology in internet-based distributed genetic algorithms

Distributed Genetic Algorithms (DGAs) designed for the Internet have to take its high communication cost into consideration. For island model GAs, the migration topology has a major impact on DGA performance. This paper describes and evaluates an adaptive migration topology optimizer that keeps the communication load low while maintaining high solution quality. Experiments on benchmark problems show that the optimized topology outperforms static or random topologies of the same degree of connectivity. The applicability of the method on real-world problems is demonstrated on a hard optimization problem in VLSI design.

Filter and control performance bounds in the presence of model uncertainties with aerospace applications

This thesis establishes performance properties for approximate filters and controllers that are designed on the basis of approximate dynamic system representations. These performance properties provide a theoretical justification for the widespread application of approximate filters and controllers in the common situation where system models are not known with complete certainty. This research also provides useful tools for approximate filter designs, which are applied to hybrid filtering of uncertain nonlinear systems. As a contribution towards applications, this thesis also investigates air traffic separation control in the presence of measurement uncertainties.

Digital signal processing and artificial intelligence for the automated classification of food allergy

As a by-product of the ‘information revolution’ which is currently unfolding, lifetimes of man (and indeed computer) hours are being allocated for the automated and intelligent interpretation of data. This is particularly true in medical and clinical settings, where research into machine-assisted diagnosis of physiological conditions gains momentum daily. Of the conditions which have been addressed, however, automated classification of allergy has not been investigated, even though the numbers of allergic persons are rising, and undiagnosed allergies are most likely to elicit fatal consequences. On the basis of the observations of allergists who conduct oral food challenges (OFCs), activity-based analyses of allergy tests were performed. Algorithms were investigated and validated by a pilot study which verified that accelerometer-based inquiry of human movements is particularly well-suited for objective appraisal of activity. However, when these analyses were applied to OFCs, accelerometer-based investigations were found to provide very poor separation between allergic and non-allergic persons, and it was concluded that the avenues explored in this thesis are inadequate for the classification of allergy. Heart rate variability (HRV) analysis is known to provide very significant diagnostic information for many conditions. Owing to this, electrocardiograms (ECGs) were recorded during OFCs for the purpose of assessing the effect that allergy induces on HRV features. It was found that with appropriate analysis, excellent separation between allergic and nonallergic subjects can be obtained. These results were, however, obtained with manual QRS annotations, and these are not a viable methodology for real-time diagnostic applications. Even so, this was the first work which has categorically correlated changes in HRV features to the onset of allergic events, and manual annotations yield undeniable affirmation of this. Fostered by the successful results which were obtained with manual classifications, automatic QRS detection algorithms were investigated to facilitate the fully automated classification of allergy. The results which were obtained by this process are very promising. Most importantly, the work that is presented in this thesis did not obtain any false positive classifications. This is a most desirable result for OFC classification, as it allows complete confidence to be attributed to classifications of allergy. Furthermore, these results could be particularly advantageous in clinical settings, as machine-based classification can detect the onset of allergy which can allow for early termination of OFCs. Consequently, machine-based monitoring of OFCs has in this work been shown to possess the capacity to significantly and safely advance the current state of clinical art of allergy diagnosis

Akt signal transduction dysfunction in the 3xTg-AD mouse model of Alzheimer's disease

Alzheimer’s disease (AD) is an incurable neurodegenerative disorder, accounting for over 60% of all cases of dementia. The primary risk factor for AD is age, however several genetic and environmental factors are also involved. The pathological characteristics of AD include extracellular deposition of the beta-amyloid peptide (Aβ) and intraneuronal accumulation of neurofibrillary tangles (NFTs) made of aggregated paired helical filaments (PHFs) of the hyperphosphorylated tau protein, along with synaptic loss and neuronal death. There are numerous biochemical mechanisms involved in AD pathogenesis, however the reigning hypothesis points to toxic oligomeric Aβ species as the primary causative factor in a cascade of events leading to neuronal stress and dyshomeostasis that initiate abnormal regulation of tau. The insulin and IGF-1 receptors (IR, IGF-1R) are the primary activators of PI3- K/Akt through which they regulate cell growth, development, glucose metabolism, and learning and memory. Work in our lab and others shows increased Akt activity and phosphorylation of its downstream targets in AD brain, along with insulin and insulin-like growth factor-1 signalling (IIS) dysfunction. This is supported by studies of AD models in vivo and in vitro. Our group and others hypothesise that Aβ activates Akt through IIS to initiate a negative feedback mechanism that desensitises neurons to insulin/IGF-1, and sustains activation of Akt. In this study the functions of endogenous Akt, IR, and the insulin receptor substrate (IRS-1) were examined in relationship to Aβ and tau pathology in the 3xTg-AD mouse model, which contains three mutant human transgenes associated with familial AD or dementia. The 3xTg-AD mouse develops Aβ and tau pathology in a spatiotemporal manner that best recapitulates the progression of AD in human brain. Western blotting and immunofluorescent microscopy techniques were utilised in vivo and in vitro, to examine the relationship between IIS, Akt, and AD pathology. I first characterised in detail AD pathology in 3xTg-AD mice, where an age-related accumulation of intraneuronal Aβ and tau was observed in the hippocampal formation, amygdala, and entorhinal cortex, and at late stages (18 months), extracellular amyloid plaques and NFTs, primarily in the subiculum and the CA1 layer of the hippocampal formation. Increased activity of Akt, detected with antibody to phosphoSer473-Akt, was increased in 3xTg-AD mice compared to age-matched non-transgenic mice (non-Tg), and in direct correlation to the accumulation of Aβ and tau in neuronal somatodendritic compartments. Akt phosphorylates tau at residue Ser214 within a highly specific consensus sequence for Akt phosphorylation, and phosphoSer214-tau strongly decreases microtubule (MT) stabilisation by preventing tau-MT binding. PhosphoSer214-tau increased concomitantly with this in the same age-related and region-specific fashion. Polarisation of tau phosphorylation was observed, where PHF-1 (tauSer396/404) and phosphoSer214-tau both appeared early in 3xTg-AD mice in distinct neuronal compartments: PHF-1 in axons, and phosphoSer214-tau in neuronal soma and dendrites. At 18 months, phosphoSer214-tau strongly colocalised with NFTs positive for the PHF- 1 and AT8 (tauSer202/Thr205) phosphoepitopes. IR was decreased with age in 3xTg-AD brain and in comparison to age-matched non-Tg, and this was specific for brain regions containing Aβ, tau, and hyperactive Akt. IRS-1 was similarly decreased, and both proteins showed altered subcellular distribution. Phosphorylation of IRS-1Ser312 is a strong indicator of IIS dysfunction and insulin resistance, and was increased in 3xTg-AD mice with age and in relation to pathology. Of particular note was our observation that abberant IIS and Akt signalling in 3xTg-AD brain related to Aβ and tau pathology on a gross anatomical level, and specifically localised to the brain regions and circuitry of the perforant path. Finally, I conducted a preliminary study of the effects of synthetic Aβ oligomers on embryonic rat hippocampus neuronal cultures to support these results and those in the literature. Taken together, these novel findings provide evidence for IIS and Akt signal transduction dysfunction as the missing link between Aβ and tau pathogenesis, and contribute to the overall understanding of the biochemical mechanisms of AD.

Optimisation of viscoelastic treatments using genetic algorithms

Viscoelastic treatments are one of the most efficient treatments, as far as passive damping is concerned, particularly in the case of thin and light structures. In this type of treatment, part of the strain energy generated in the viscoelastic material is dissipated to the surroundings, in the form of heat. A layer of viscoelastic material is applied to a structure in an unconstrained or constrained configuration, the latter proving to be the most efficient arrangement. This is due to the fact that the relative movement of both the host and constraining layers cause the viscoelastic material to be subjected to a relatively high strain energy. There are studies, however, that claim that the partial application of the viscoelastic material is just as efficient, in terms of economic costs or any other form of treatment application costs. The application of patches of material in specific and selected areas of the structure, thus minimising the extension of damping material, results in an equally efficient treatment. Since the damping mechanism of a viscoelastic material is based on the dissipation of part of the strain energy, the efficiency of the partial treatment can be correlated to the modal strain energy of the structure. Even though the results obtained with this approach in various studies are considered very satisfactory, an optimisation procedure is deemed necessary. In order to obtain optimum solutions, however, time consuming numerical simulations are required. The optimisation process to use the minimum amount of viscoelastic material is based on an evolutionary geometry re-design and calculation of the modal damping, making this procedure computationally costly. To avert this disadvantage, this study uses adaptive layerwise finite elements and applies Genetic Algorithms in the optimisation process.