arXiv subjects

Pramod P. Khargonekar

Publications and source records attributed to Pramod P. Khargonekar.

34 records · Page 2Linked to original sources

arXiv2021

Memory Augmented Neural Network Adaptive Controllers: Performance and Stability

In this paper, we propose a novel control architecture, inspired from neuroscience, for adaptive control of continuous-time systems. The proposed architecture, in the setting of standard Neural Network (NN) based adaptive control, augments an external working memory to the NN. The controller, through a write operation, writes the hidden layer feature vector of the NN to the external working memory and can also update this information with the observed error in the output. Through a read operation, the controller retrieves information from the working memory to modify the final control signal. First, we consider a simpler estimation problem to theoretically study the effect of an external memory and prove that the estimation accuracy can be improved by incorporating memory. We then consider a model reference NN adaptive controller for linear systems with matched uncertainty to implement and illustrate our ideas. We prove that the resulting controller leads to a Uniformly Bounded (UB) stable closed loop system. Through extensive simulations and specific metrics, such as peak deviation and settling time, we show that memory augmentation improves learning significantly. Importantly, we also provide evidence for and insights on the mechanism by which this specific memory augmentation improves learning.

eess.SY↗

arXiv2021

Graph Learning for Cognitive Digital Twins in Manufacturing Systems

Future manufacturing requires complex systems that connect simulation platforms and virtualization with physical data from industrial processes. Digital twins incorporate a physical twin, a digital twin, and the connection between the two. Benefits of using digital twins, especially in manufacturing, are abundant as they can increase efficiency across an entire manufacturing life-cycle. The digital twin concept has become increasingly sophisticated and capable over time, enabled by rises in many technologies. In this paper, we detail the cognitive digital twin as the next stage of advancement of a digital twin that will help realize the vision of Industry 4.0. Cognitive digital twins will allow enterprises to creatively, effectively, and efficiently exploit implicit knowledge drawn from the experience of existing manufacturing systems. They also enable more autonomous decisions and control, while improving the performance across the enterprise (at scale). This paper presents graph learning as one potential pathway towards enabling cognitive functionalities in manufacturing digital twins. A novel approach to realize cognitive digital twins in the product design stage of manufacturing that utilizes graph learning is presented.

cs.LG↗

arXiv2021

Neuroscience-Inspired Algorithms for the Predictive Maintenance of Manufacturing Systems

If machine failures can be detected preemptively, then maintenance and repairs can be performed more efficiently, reducing production costs. Many machine learning techniques for performing early failure detection using vibration data have been proposed; however, these methods are often power and data-hungry, susceptible to noise, and require large amounts of data preprocessing. Also, training is usually only performed once before inference, so they do not learn and adapt as the machine ages. Thus, we propose a method of performing online, real-time anomaly detection for predictive maintenance using Hierarchical Temporal Memory (HTM). Inspired by the human neocortex, HTMs learn and adapt continuously and are robust to noise. Using the Numenta Anomaly Benchmark, we empirically demonstrate that our approach outperforms state-of-the-art algorithms at preemptively detecting real-world cases of bearing failures and simulated 3D printer failures. Our approach achieves an average score of 64.71, surpassing state-of-the-art deep-learning (49.38) and statistical (61.06) methods.

cs.NE↗

arXiv2020

Cognitive Preadaptation for Resilient Adaptive Control

In this paper, we investigate a novel control architecture and algorithm for incorporating preadaption functions. We propose a preadaptation mechanism that can augment any adaptive control scheme and improve its resilience. Through simulations of a flight control system we illustrate the effectiveness of the preadaptation mechanism in improving the adaptation. We show that the preadaptation mechanism we propose can reduce the peak of the response by as much as $50\%$. The scenarios we present also show that the preadaptation mechanism is effective across a wide range of scenarios suggesting that the mechanism is reliable.

eess.SY↗

arXiv2020

Scene-Graph Augmented Data-Driven Risk Assessment of Autonomous Vehicle Decisions

Despite impressive advancements in Autonomous Driving Systems (ADS), navigation in complex road conditions remains a challenging problem. There is considerable evidence that evaluating the subjective risk level of various decisions can improve ADS' safety in both normal and complex driving scenarios. However, existing deep learning-based methods often fail to model the relationships between traffic participants and can suffer when faced with complex real-world scenarios. Besides, these methods lack transferability and explainability. To address these limitations, we propose a novel data-driven approach that uses scene-graphs as intermediate representations. Our approach includes a Multi-Relation Graph Convolution Network, a Long-Short Term Memory Network, and attention layers for modeling the subjective risk of driving maneuvers. To train our model, we formulate this task as a supervised scene classification problem. We consider a typical use case to demonstrate our model's capabilities: lane changes. We show that our approach achieves a higher classification accuracy than the state-of-the-art approach on both large (96.4% vs. 91.2%) and small (91.8% vs. 71.2%) synthesized datasets, also illustrating that our approach can learn effectively even from smaller datasets. We also show that our model trained on a synthesized dataset achieves an average accuracy of 87.8% when tested on a real-world dataset compared to the 70.3% accuracy achieved by the state-of-the-art model trained on the same synthesized dataset, showing that our approach can more effectively transfer knowledge. Finally, we demonstrate that the use of spatial and temporal attention layers improves our model's performance by 2.7% and 0.7% respectively, and increases its explainability.

cs.CV↗

arXiv2020

Online Algorithms for Dynamic Matching Markets in Power Distribution Systems

This paper proposes online algorithms for dynamic matching markets in power distribution systems, which at any real-time operation instance decides about matching -- or delaying the supply of -- flexible loads with available renewable generation with the objective of maximizing the social welfare of the exchange in the system. More specifically, two online matching algorithms are proposed for the following generation-load scenarios: (i) when the mean of renewable generation is greater than the mean of the flexible load, and (ii) when the condition (i) is reversed. With the intuition that the performance of such algorithms degrades with increasing randomness of the supply and demand, two properties are proposed for assessing the performance of the algorithms. First property is convergence to optimality (CO) as the underlying randomness of renewable generation and customer loads goes to zero. The second property is deviation from optimality, is measured as a function of the standard deviation of the underlying randomness of renewable generation and customer loads. The algorithm proposed for the first scenario is shown to satisfy CO and a deviation from optimal that varies linearly with the variation in the standard deviation. But the same algorithm is shown to not satisfy CO for the second scenario. We then show that the algorithm proposed for the second scenario satisfies CO and a deviation from optimal that varies linearly with the variation in standard deviation plus an offset.

eess.SY↗

arXiv2020

Improved Attention Models for Memory Augmented Neural Network Adaptive Controllers

We introduced a {\it working memory} augmented adaptive controller in our recent work. The controller uses attention to read from and write to the working memory. Attention allows the controller to read specific information that is relevant and update its working memory with information based on its relevance. The retrieved information is used to modify the final control input computed by the controller. We showed that this modification speeds up learning. In the above work, we used a soft-attention mechanism for the adaptive controller. Controllers that use soft attention or hard attention mechanisms are limited either because they can forget the information or fail to shift attention when the information they are reading becomes less relevant. We propose an attention mechanism that comprises of (i) a hard attention mechanism and additionally (ii) an attention reallocation mechanism. The attention reallocation enables the controller to reallocate attention to a different location when the relevance of the location it is reading from diminishes. The reallocation also ensures that the information stored in the memory before the shift in attention is retained which can be lost in both soft and hard attention mechanisms. We illustrate through detailed simulations of various scenarios for two link robot and three link robot arm systems we illustrate the effectiveness of the proposed attention mechanism.

eess.SY↗

arXiv2020

Memory Augmented Neural Network Adaptive Controller for Strict Feedback Nonlinear Systems

In this work, we consider the adaptive nonlinear control problem for strict feedback nonlinear systems, where the functions that determine the dynamics of the system are completely unknown. We assume that certain upper bounds for the functions $g_i$s of the system are known. The objective of the control design is to design an adaptive controller that can adapt to changes in the unknown functions that are even abrupt. We propose a novel backstepping memory augmented NN (MANN) adaptive control method for the control of strict feedback non-linear systems. Here, each NN, in the backstepping NN adaptive controller, is augmented with an external working memory. The NN can write relevant information to its working memory and later retrieve them to modify its output, thus providing it with the capability to leverage past learned information effectively and improve its speed of learning. We propose a specific design for this external memory interface and show that the proposed control design achieves bounded stability for the closed loop system. We also provide substantial numerical evidence showing that the proposed memory augmentation improves the speed of learning by a significant margin.

eess.SY↗

arXiv2020

Incentive Design in a Distributed Problem with Strategic Agents

In this paper, we consider a general distributed system with multiple agents who select and then implement actions in the system. The system has an operator with a centralized objective. The agents, on the other hand, are selfinterested and strategic in the sense that each agent optimizes its own individual objective. The operator aims to mitigate this misalignment by designing an incentive scheme for the agents. The problem is difficult due to the cost functions of the agents being coupled, the objective of the operator not being social welfare, and the operator having no direct control over actions being implemented by the agents. This problem has been studied in many fields, particularly in mechanism design and cost allocation. However, mechanism design typically assumes that the operator has knowledge of the cost functions of the agents and the actions being implemented by the operator. On the other hand, cost allocation classically assumes that agents do not anticipate the effect of their actions on the incentive that they obtain. We remove these assumptions and present an incentive rule for this setup by bridging the gap between mechanism design and classical cost allocation. We analyze whether the proposed design satisfies various desirable properties such as social optimality, budget balance, participation constraint, and so on. We also analyze which of these properties can be satisfied if the assumptions of cost functions of the agents being private and the agents being anticipatory are relaxed.

cs.GT↗

arXiv2019

A Minimal Incentive-based Demand Response Program With Self Reported Baseline Mechanism

In this paper, we propose a novel incentive based Demand Response (DR) program with a self reported baseline mechanism. The System Operator (SO) managing the DR program recruits consumers or aggregators of DR resources. The recruited consumers are required to only report their baseline, which is the minimal information necessary for any DR program. During a DR event, a set of consumers, from this pool of recruited consumers, are randomly selected. The consumers are selected such that the required load reduction is delivered. The selected consumers, who reduce their load, are rewarded for their services and other recruited consumers, who deviate from their reported baseline, are penalized. The randomization in selection and penalty ensure that the baseline inflation is controlled. We also justify that the selection probability can be simultaneously used to control SO's cost. This allows the SO to design the mechanism such that its cost is almost optimal when there are no recruitment costs or at least significantly reduced otherwise. Finally, we also show that the proposed method of self-reported baseline outperforms other baseline estimation methods commonly used in practice.

eess.SY↗

arXiv2018

Analysis of Solar Energy Aggregation under Various Billing Mechanisms

Ongoing reductions in the cost of solar photovoltaic (PV) systems are driving their increased installations by residential households. Various incentive programs such as feed-in tariff, net metering, net purchase and sale that allow the prosumers to sell their generated electricity to the grid are also powering this trend. In this paper, we investigate sharing of PV systems among a community of households, who can also benefit further by pooling their production. Using cooperative game theory, we find conditions under which such sharing decreases their net total cost. We also develop allocation rules such that the joint net electricity consumption cost is allocated to the participants. These cost allocations are based on the cost causation principle. The allocations also satisfy the standalone cost principle and promote PV solar aggregation. We also perform a comparative analytical study on the benefit of sharing under the mechanisms favorable for sharing, namely net metering, and net purchase and sale. The results are illustrated in a case study using real consumption data from a residential community in Austin, Texas.

eess.SY↗

arXiv2017

Sharing Storage in a Smart Grid: A Coalitional Game Approach

Sharing economy is a transformative socio-economic phenomenon built around the idea of sharing underused resources and services, e.g. transportation and housing, thereby reducing costs and extracting value. Anticipating continued reduction in the cost of electricity storage, we look into the potential opportunity in electrical power system where consumers share storage with each other. We consider two different scenarios. In the first scenario, consumers are assumed to already have individual storage devices and they explore cooperation to minimize the realized electricity consumption cost. In the second scenario, a group of consumers is interested to invest in joint storage capacity and operate it cooperatively. The resulting system problems are modeled using cooperative game theory. In both cases, the cooperative games are shown to have non-empty cores and we develop efficient cost allocations in the core with analytical expressions. Thus, sharing of storage in cooperative manner is shown to be very effective for the electric power system.

eess.SY↗

arXiv2014

A Global Identifiability Condition for Consensus Networks with Tree Graphs

In this paper we present a sufficient condition that guarantees identifiability of linear network dynamic systems exhibiting continuous-time weighted consensus protocols with acyclic structure. Each edge of the underlying network graph $\mathcal G$ of the system is defined by a constant parameter, referred to as the weight of the edge, while each node is defined by a scalar state whose dynamics evolve as the weighted linear combination of its difference with the states of its neighboring nodes. Following the classical definitions of identifiability and indistinguishability, we first derive a condition that ensure the identifiability of the edge weights of $\mathcal G$ in terms of the associated transfer function. Using this characterization, we propose a sensor placement algorithm that guarantees identifiability of the edge weights. We describe our results using several illustrative examples.

eess.SY↗

arXiv2013

Signal Reconstruction via H-infinity Sampled-Data Control Theory: Beyond the Shannon Paradigm

This paper presents a new method for signal reconstruction by leveraging sampled-data control theory. We formulate the signal reconstruction problem in terms of an analog performance optimization problem using a stable discrete-time filter. The proposed H-infinity performance criterion naturally takes intersample behavior into account, reflecting the energy distributions of the signal. We present methods for computing optimal solutions which are guaranteed to be stable and causal. Detailed comparisons to alternative methods are provided. We discuss some applications in sound and image reconstruction.

cs.IT↗

arXiv2013

Computational Modeling of Channelrhodopsin-2 Photocurrent Characteristics in Relation to Neural Signaling

Channelrhodopsins-2 (ChR2) are a class of light sensitive proteins that offer the ability to use light stimulation to regulate neural activity with millisecond precision. In order to address the limitations in the efficacy of the wild-type ChR2 (ChRwt) to achieve this objective, new variants of ChR2 that exhibit fast mono-exponential photocurrent decay characteristics have been recently developed and validated. In this paper, we investigate whether the framework of transition rate model with 4 states, primarily developed to mimic the bi-exponential photocurrent decay kinetics of ChRwt, as opposed to the low complexity 3 state model, is warranted to mimic the mono-exponential photocurrent decay kinetics of the newly developed fast ChR2 variants: ChETA (Gunaydin et al., Nature Neurosci, 13:387-392, 2010) and ChRET/TC (Berndt et al., PNAS, 108:7595-7600, 2011). We begin by estimating the parameters for the 3-state and 4-state models from experimental data on the photocurrent kinetics of ChRwt, ChETA and ChRET/TC. We then incorporate these models into a fast-spiking interneuron model (Wang and Buzsaki., J Neurosci, 16:6402-6413,1996) and a hippocampal pyramidal cell model (Golomb et al., J Neurophysiol, 96:1912-1926, 2006) and investigate the extent to which the experimentally observed neural response to various optostimulation protocols can be captured by these models. We demonstrate that for all ChR2 variants investigated, the 4 state model implementation is better able to capture neural response consistent with experiments across wide range of optostimulation protocol. We conclude by analytically investigating the conditions under which the characteristic specific to the 3-state model, namely the mono-exponential photocurrent decay of the newly developed variants of ChR2, can occurs in the framework of the 4-state model.

q-bio.NC↗

arXiv2013

Fast SVM training using approximate extreme points

Applications of non-linear kernel Support Vector Machines (SVMs) to large datasets is seriously hampered by its excessive training time. We propose a modification, called the approximate extreme points support vector machine (AESVM), that is aimed at overcoming this burden. Our approach relies on conducting the SVM optimization over a carefully selected subset, called the representative set, of the training dataset. We present analytical results that indicate the similarity of AESVM and SVM solutions. A linear time algorithm based on convex hulls and extreme points is used to compute the representative set in kernel space. Extensive computational experiments on nine datasets compared AESVM to LIBSVM \citep{LIBSVM}, CVM \citep{Tsang05}, BVM \citep{Tsang07}, LASVM \citep{Bordes05}, $\text{SVM}^{\text{perf}}$ \citep{Joachims09}, and the random features method \citep{rahimi07}. Our AESVM implementation was found to train much faster than the other methods, while its classification accuracy was similar to that of LIBSVM in all cases. In particular, for a seizure detection dataset, AESVM training was almost $10^3$ times faster than LIBSVM and LASVM and more than forty times faster than CVM and BVM. Additionally, AESVM also gave competitively fast classification times.

cs.LG↗