Search

We propose a new method for solving Initial Value Problems (IVPs). Our method is based on analog computing and has the potential to almost eliminate traditional switching time in digital computing. The approach can be used to simulate large systems longer, faster, and with higher accuracy. Many algorithms for Model-Based Diagnosis use numerical integration to simulate physical systems. The numerical integration process is often either computationally expensive or imprecise. We propose a new method, based on Field-Programmable Analog Arrays (FPAAs) that has the potential to overcome many practical problems. We envision a software/hardware framework for solving systems of simultaneous Ordinary Differential Equations (ODEs) in fraction of the time of traditional numerical algorithms. In this paper we describe the solving of an IVP with the help of an Analog Computing Unit (ACU). To do this we build a special calculus based on operational amplifiers (op-amps) with local feedback. We discuss the implementation of the ACU on an Integrated Circuit (IC). We analyze the working if the IC and simulate the dynamic Lotka-Volterra system with the de-facto standard tool for electrical simulation: Spice.

Certain market forces play a decisive role in stimulating urban development and growth: economies of scale and agglomeration; factor mobility and migration; and transport and specialization costs. There is a fundamental, common idea for all decision-makers who are trying to promote solid and sustainable growth: cities are not state constructions but the result of dynamic markets. In this sense, the most effective policies do not seek to change the existing economic structure of cities, but to support, allowing local economies to do what they know best and to take advantage of the savings generated by agglomerations. Indeed, the data suggest that as economies face the transition from agriculture to industry and subsequently to service-based services, the concept of density, distance and division is of greater importance in supporting economic growth. It should not be deduced from this that governments have no role in urban planning and management. On the contrary, planners and decision-makers have a leverage to support development by understanding market forces by accepting them and accelerating their impact. However, in many cases, policies are designed to actively oppose market dynamics and overturn; such attempts are usually doomed to failure and generate a huge waste of resources. Classification-JEL: F63, J08, O18.

Deep learning platforms have become hugely popular due to their successes in natural language processing and image processing. Our objective is to show how deep learning platforms can be used for control problems. We do not make judgments about their performance as compared to traditional control approaches. We show that the main challenge when using deep learning platforms for learning control policies for nonlinear systems is ensuring the stability of the learning algorithm that depends on the stability of the closed loop system during the learning process. We discuss two approaches for overcoming the potential instability of the optimization algorithm, and showcase them in the context of learning a stabilizing controller for an inverted pendulum.

We introduce a novel machine learning-based fusion model termed as PI-LSTM (Physics-Infused Long Short-Term Memory Networks) that integrates first principle Physics-Based Models and Long Short-Term Memory (LSTM) network. Our architecture aims at combining equation-based models with data-driven machine learning models to enable accurate predictions of complex dynamic systems. In this hybrid architecture, recurrency aids the temporal memory of the inputs and output of the partial physics model, in a way that facilitates generalization with scarce data sets. We illustrate the application of PI-LSTM on two dynamical systems namely Inverted Pendulum and Tumor Growth. Empirical results on both test problems stand witness to the effectiveness of using physics in guiding machine learning models and the superiority of the outlined hybrid model over purely data-driven models.

Machine learning methods based on classifiers are more robust to system complexity, but they ignore the relations that exist in the data due to the physical laws governing the behavior of the system. In this paper we discuss how (partial) knowledge about the physical system can be integrated in the machine learning process. We focus on classification based diagnosis. We show how the partially known model is integrated in the classification algorithm, and how the new algorithm differs from a machine learning classifier based on neural networks. We demonstrate that by integrating the partial system knowledge, the cross-entropy optimization problem used for learning a classifier can be expressed as a set of regression problems in terms of the parameters of the model representing the unknown behavior, followed by a simpler classifier learning. We showcase our approach when diagnosis faults for a rail switch system.

Hybrid modeling architectures seek to combine a machine learning model with a computationally efficient (simplified or partial) physics model to predict the behavior of physical systems. Existing sequential or parallel approaches to hybrid modeling do not typically exploit the potential relationship between the input or latent features of the partial and full physics. In addition, very few existing architectures take advantage of the provision to generously or on-demand sample the partial physics model. To address these gaps, we have developed a novel neural network-based hybrid architecture called “Opportunistic Physics-mining Transfer Mapping Architecture” or OPTMA. The goal of the OPTMA architecture is to facilitate greater exploitation of input space correlations between partial and full physics where they exist. To this end, a transfer neural network is used to transform the original inputs into modified inputs or latent features, where the partial physics operates on these artificially transformed features to produce the final prediction. An extended back-propagation approach and a Particle Swarm Optimization (to deal with multimodal loss functions) are used to train the network weights. The new architecture is first tested on a simple regression problem for analysis. It is then used to predict the behavior of more complex dynamic systems — an Inverted pendulum and the motion of an unmanned aerial vehicle, both under wind effects. Subsequent tests on unseen samples demonstrate OPTMA’s competitive performance compared to pure ANN and sequential hybrid models and provide empirical validation of the transfer concepts underlying OPTMA.

Learning a model or learning a policy that optimizes some objective function relies on data-sets that describe the behavior of the system. When such sets are unavailable or insufficient, additional data may be generated through new experiments (if feasible) or through simulations (if an accurate model is available). In this paper we describe a third alternative that is based on the availability of a qualitative model of the physical system. In particular, we show how the number of experiments used in reinforcement learning can be reduced by leveraging geometric properties of the system. The geometric properties are independent of any particular instantiation of the qualitative model. As an illustrative example, we apply our approach to a cart-pole system.

A microassembly system based on electric field manipulation is described. Optical addressing of a photo-transistor array controls the position and orientation of micro-objects using dielectrophoretic and electrophoretic forces. Open-loop control is used to rapidly direct many micro-objects in parallel. Closed-loop control is used to position micro-objects precisely. Demonstrations with beads (10 μm diameter) and silicon chips (150 μm scale) are presented. We have achieved coarse assembly at a rate of 1000 micro-objects per second, fine assembly of beads at micrometer scales, and arrays of silicon chips with orientation control. The technology aims to enable a generalized printer system for building engineered microstructure and electronics.

We address the problem of position control of micro-chips (chiplets) immersed in dielectric fluid. An electric field, shaped by controlling the voltages of spiral shaped electrodes, is used to reliably and accurately transport and position chiplets using dielectrophoretic forces. A lumped, capacitive based (nonlinear) motion model is used to generate an open loop control policy. The open loop policy is generated using a one step model predictive control approach. By exploiting the spatial symmetry and periodicity of the open loop control solution, a real-time control scheme is designed by applying simple algebraic operations to a base function defined on a finite domain. The chiplet position is tracked using image processing algorithms. We demonstrate the validity of our approach by describing an experimental result, where real-time control is used to move a chiplet for 1000µm in a controlled manner.

It usually can take more than ten years from the time a new drug is discovered, until can be launched on the market. Regulatory requirements are part of the process of drug discovery and drug development. It acts at every developmental stage. Regulatory affairs works to establish an effective and uniform balance between voluntary and regulatory compliance and agency responsiveness to consumer needs. It evaluates and coordinates all proposed legal actions to ascertain compliance with regulatory policy. The ontology presented for regulatory affairs and drug research and development gives us the possibility to correlate information from different levels and to discover new relationships between the legal aspects. In addition, the transparency of the information is affected by the inability of existing integration strategies to organize and apply the available knowledge to the range of real scientific and business issue in critical safety and regulatory applications. Therefore, the semantic technologies based on ontologies make the knowledge reusable by several applications across business, from discovery to corporate affairs.

This paper describes the philosophy behind the remote level control of a didactical platform. Issues related to the hardware and software architecture that allow remote access to the experimental platform using a simple Internet browser and some concepts about the control strategy used are presented. The virtual laboratory application is built primarily as an educational tool for distance learning, but these types of applications can be extended, with some considerations, to industrial process control.

To minimize the design cost of a complex system and maximize performance, a design team ideally must be able to quantify reliability and mitigate risk at the earliest phases of the design process, where 80% of the cost is committed. This paper demonstrates the capabilities of a new System Reliability Exploration Tool based on the improved simulation capabilities of a system called Fault-Augmented Modelica Extension (FAME). This novel tool combines concepts from FMEA, traditional Reliability Analysis, and Quality Engineering to identify, gain insight, and quantify the impact of component failure modes through time evolution of a system’s lifecycle. We illustrate how to use the FAME System Reliability Exploration Tool through a vehicle design case study.Copyright © 2014 by ASME

We propose a model-based diagnosis framework in which Modelica models of faulted behavior are used in combination with a Bayesian approach. The fault augmented models are automatically generated through a process developed as part of our Fault Augmented Model Extension (FAME) work. Fault diagnosis using a Bayesian approach is based on computing a set of probability density functions, a process that is usually intractable for any reasonably complex system. We use Approximate Bayesian Computation (ABC) to bound the numerical and computational complexity. The basic idea is to use fault augmented Modelica models to create probability distributions of possible outcomes and then compare those distributions against actual observations to perform parameter estimation. The detection of faults is treated as a model selection problem and the inference of their severity levels is treated as parameter estimation. The diagnostic precision of this approach is evaluated on a Modelica vehicle drive line model.

A comparative study of tyrosine phosphorylation was performed on peripheral blood lymphocytes from systemic lupus erythematosus (SLE) patients and from healthy donors. Freshly isolated SLE lymphocytes presented an elevated tyrosine phosphorylation level when compared to healthy donors lymphocytes (p = 0.005). Among all phosphorylated proteins, those called p120, p110, p80 and p55-p60 were more phosphorylated. The level of tyrosine phosphorylation of p120 and p110 proteins discriminated significantly (p = 0.0048, respectively, p = 0.02) between SLE patients and healthy donors. Lymphocytes form SLE patients and healthy donors were then stimulated by cross-linking T cell antigens (CD3, CD4, CD8) to further distinguish the signal transduction between normal and pathologic lymphocytes. No statistical differences in the tyrosine phosphorylation pattern, following CD4 or CD8 cross-linking, were observed between SLE patients and healthy donors lymphocytes. CD3 cross-linking induced an effect on tyrosine phosphorylation different in SLE patients versus healthy donors lymphocytes. Thus, the lymphocytes of SLE patients were refractile in anti-CD3 stimulation in comparison with the healthy donors lymphocytes. Chi-square analysis demonstrated that a significantly larger number of healthy donors responded to anti-CD3 stimulation compared to SLE patients (p = 0.03). The high frequency of tyrosine phosphorylation of p110 and p80 proteins, following CD3 stimulation, in normal versus SLE lymphocytes, suggested that these proteins could be involved in abnormal signal transduction in SLE cells.

Computer-interpretable representations of system structure and behavior are at the center of designing today’s complex systems. Engineers create and review such representations using graphical modeling languages that support specification, analysis, design, verification and validation of systems that include hardware, software, data, personnel, procedures, and facilities, in particular the Systems Modeling Language (SysML), an extension of the Unified Modeling Language. However, SysML’s constructs are insufficient to capture all details necessary for dynamic (solver-based) simulation and must be enhanced with domain specific tools for this purpose. SysML modeling tools and simulation tools are often used separately and sequentially, which reduces the efficiency of the engineering process. As a result, there is an increasing need for integrating modeling constructs specific to simulation tools into SysML. In this report, we first analyze if SysML possesses constructs that match the constructs used in simulation tools. We conclude that such constructs exist only partially and propose extensions of SysML to accommodate modeling dynamical systems. In addition, we show through an example, how the newly proposed extensions can be used to model an electrical circuit in SysML.

We address the problem of state estimation of the power system for the Smart Grid. We assume that the monitoring of the electrical grid is done by a network of agents with both computing and communication capabilities. We propose a security mechanism aimed at protecting the state estimation process against false data injections originating from faulty equipment or cyber-attacks. Our approach is based on a multi-agent filtering scheme, where in addition to taking measurements, the agents are also computing local estimates based on their own measurements and on the estimates of the neighboring agents. We combine the multi-agent filtering scheme with a trust-based mechanism under which each agent associates a trust metric to each of its neighbors. These trust metrics are taken into account in the filtering scheme so that information transmitted from agents with low trust is disregarded. In addition, a mechanism for the trust metric update is also introduced, which ensures that agents that diverge considerably from their expected behavior have their trust values lowered.

We consider a sensor network monitoring a stochastic process. The sensors exchange information over wireless communication channels affected by noise. The goal of the sensors is to compute accurate estimates of the state of the stochastic process under limited energy available for communications. We introduce a distributed algorithm which computes a power allocation scheme aimed at ensuring accurate state estimates and at saving communication energy. The distributed algorithm arise from solving an approximate convex optimization problem with constraints. We present how the power allocation scheme can be used for performing distributed estimation in the case where measurements are shared between sensors.