In industrial manufacturing, a production process usually consists of multiple manufacturing steps during the transformation raw material into complete product. Depending on varying product specifications, may differ within performed and definition "normality", which refers to multi-dimensional sensor data captured process. Process Mining offers excellent discovery monitoring capabilities; however, it relies well-defined event logs. Industry 4.0 this requirement poses significant challenge...

Abstract Traditional controllers have limitations as they rely on prior knowledge about the physics of problem, require modeling dynamics, and struggle to adapt abnormal situations. Deep reinforcement learning has potential address these problems by optimal control policies through exploration in an environment. For safety-critical environments, it is impractical explore randomly, replacing conventional with black-box models also undesirable. Also, expensive continuous state action spaces,...

Friction is responsible for several servomechanism limitations, and their elimination always a challenge control engineers. In this paper, model-based feedback compensation studied tracking tasks. Several kinetic friction models are employed parameters identified experimentally. The effects of on system response examined using describing function analysis. A number laws including classical laws, rigid body motion models, compared experimentally in large-displacement Results show that the...

Defect detection methods have greatly assisted human operators in various fields, from textiles to surfaces and mechanical components, by facilitating decision-making processes reducing visual fatigue. This area of research is widely recognized as a cross-industry concern, particularly the manufacturing sector. Nevertheless, each specific application brings unique challenges that require tailored solutions. paper presents novel framework for leveraging latent space representations defect...

Electricity load consumption may be extremely complex in terms of profile patterns, as it depends on a wide range human factors, and is often correlated with several exogenous such the availability renewable energy weather conditions. The first goal this paper to investigate performance large selection different types forecasting models predicting electricity within short time horizon day or few hours ahead. Such forecasts rather useful for management individual residential buildings small...

A linear-programming (LP) based path planning algorithm is developed for deriving optimal paths a group of autonomous vehicles in an adversarial environment. In this method, both friendly and enemy are modelled as different resource types arena sectors, the problem viewed allocation problem. Simple model simplifications introduced to allow use linear programming conjunction with receding horizon implementation multi-vehicle planning. Stochastic models on current position opposing used...

We consider the problem of deriving optimal advertising policies for spread innovations in a social network. seek to compute that account i) endogenous network influences, ii) presence competitive firms, also wish influence network, and iii) possible uncertainties model. Contrary prior work advertising, which accounts we assume dynamic model preferences either finite or infinite horizons. robust case where evolution is affected by external disturbances. Finally, firm, Stackelberg Nash solutions.

Friction is responsible for several servomechanism problems, and their elimination always a challenge control engineers. In this paper, feedback model-based compensation of friction used set point tracking tasks. Basic models are tested influence on system response examined using describing function analysis. Analytical predictions compared to simulations experimental results. Various laws experimentally. Results showed that both types tasks, the best obtained by law with general kinetic model.

The management of resources among competing QoS-aware applications is often solved by a resource manager (RM) that assigns both the and application service levels. However, this approach requires all to inform RM available Then, has maximize "overall quality" comparing levels different which are not necessarily comparable. In paper we describe Linux implementation game-theoretic framework decouples two distinct problems assignment quality setting, solving them in domain where they naturally...

We consider the problem of distributed convergence to efficient outcomes in coordination games through dynamics based on aspiration learning. Our first contribution is characterization asymptotic behavior induced Markov chain iterated process terms an equivalent finite-state chain. then characterize explicitly proposed learning a generalized version games, examples which include network formation and common-pool games. In particular, we show that generic frequency at action profile played...

We consider the problem of network formation in a distributed fashion. Network is modeled as strategic-form game, where agents represent nodes that form and sever unidirectional links with other derive utilities from these links. Furthermore, can only limited set neighbors. Agents trade off benefit links, which determined by distance-dependent reward function, cost maintaining When each agent acts independently, trying to maximize its own utility we characterize “stable” networks through...

This paper presents a reinforcement learning algorithm and provides conditions for global convergence to Nash equilibria. For several schemes, including the ones proposed here, excluding action profiles which are not equilibria may be trivial, unless step-size sequence is appropriately tailored specifics of game. In this paper, we sidestep these issues by introducing new class schemes where strategy each agent perturbed state-dependent perturbation function. Contrary prior work on...

This paper considers a class of reinforcement-based learning (namely, perturbed automata) and provides stochastic-stability analysis in repeatedly played, positive-utility, finite strategic-form games. Prior work this dynamics primarily analyzes asymptotic convergence through stochastic approximations, where can be associated with the limit points an ordinary-differential equation (ODE). However, analyzing global ODE-approximation requires existence Lyapunov or potential function, which...