Expert Control Research Articles

Business process improvement with AB testing and reinforcement learning: grounded theory-based industry perspectives

AbstractIn order to better facilitate the need for continuous business process improvement, the application of DevOps principles has been proposed. In particular, the AB-BPM methodology applies AB testing—a DevOps practice—and reinforcement learning to increase the speed and quality of business process improvement efforts. In this paper, we provide an industry perspective on this approach, assessing prerequisites, suitability, requirements, risks, and additional aspects of the AB-BPM methodology and supporting tools. Our qualitative study follows the grounded theory research methodology, including 16 semi-structured interviews with BPM practitioners. The main findings indicate: (1) a need for expert control during reinforcement learning-driven experiments in production, (2) the importance of involving the participants and aligning the method culturally with the respective setting, (3) the necessity of an integrated process execution environment, and (4) the long-term potential of the methodology for effective and efficient validation of algorithmically (co-)created business process variants, and their continuous management.

Optimal load distribution control for airport terminal chiller units based on deep reinforcement learning

The building sector consumes substantial energy, with HVAC systems accounting for nearly half of the total energy use. Optimizing chiller unit operations is crucial for reducing energy consumption. This research introduces a novel model-free control method for optimizing central chiller load distribution using deep reinforcement learning (DRL). The aim is to address the challenges posed by the high-dimensional complexity and parameter tuning in traditional meta-heuristic algorithms. The proposed method leverages deep neural networks combined with reinforcement learning to optimize chiller control based on real-world data. Case studies conducted in a large airport terminal demonstrate that this approach achieves a 7.71 % energy savings, closely matching the performance of model-based control methods with only a 0.26 % difference. The results indicate that the DRL method not only outperforms traditional expert control systems but also offers a feasible alternative for optimal control in complex, variable environments with limited data. This study contributes to the field by filling a research gap in applying DRL for optimal chiller load (OCL) control and demonstrates its potential for large-scale public building applications. The novelty of this research lies in its model-free approach, which provides a robust solution for energy optimization in dynamic and uncertain environments.

Knowledge Base Prototype Creating with Using Interdisciplinary Metathesaurus.

This article presents our experience in development an ontological model can be used in clinical decision support systems (CDSS) creating. We have used the largest international biomedical terminological metathesaurus the Unified Medical Language System (UMLS) as the basis of our model. This metathesaurus has been adapted into Russian using an automated hybrid translation system with expert control. The product we have created was named as the National Unified Terminological System (NUTS). We have added more than 33 million scientific and clinical relationships between NUTS terms, extracted from the texts of scientific articles and electronic health records. We have also computed weights for each relationship, standardized their values and created symptom checker in preliminary diagnostics based on this. We expect, that the NUTS allow solving task of named entity recognition (NER) and increasing terms interoperability in different CDSS.

Intelligent Control of Pre-Chamber Pressure Based on Working Condition Identification for the Coke Dry Quenching Process

The pre-chamber pressure is an important control parameter that affects the coke dry quenching process. It often fluctuates violently and is detrimental for the safe operation of the coke dry quenching process. This study proposes an intelligent control method for the pre-chamber pressure based on working condition identification for the coke dry quenching process to realize stable control of the pre-chamber pressure. First, by describing the coke dry quenching process and analyzing the factors affecting the pre-chamber pressure, an intelligent control strategy was developed. Then, the K-means clustering algorithm was used to identify the working conditions of pre-chamber, and the working conditions were divided into two categories: stable and fluctuating. For stable conditions, a fuzzy proportional-integral-derivative controller was designed to improve the pressure control accuracy. For fluctuating conditions, an expert controller was designed to rapidly adjust the pressure. Finally, experiments based on actual data were performed and the results showed that the proposed method can effectively improve the control accuracy of pressure under different conditions. This satisfies the requirements for a continuous coke dry quenching process.

Dissipative imitation learning for discrete dynamic output feedback control with sparse data sets

AbstractImitation learning enables synthesis of controllers for systems with complex objectives and uncertain plant models. However, ensuring an imitation learned controller is stable requires copious amounts of data and/or a known plant model. In this paper, we explore an input–output (IO) stability approach to imitation learning, which achieves stability with sparse data sets while only requiring coarse knowledge of the energy characteristics of the plant. A constrained optimization problem is developed, in which the controller learns to mimic expert data while maintaining stabilizing energy characteristics induced by the plant. While the learning objective is nonconvex, iterative convex overbounding (ICO) and projected gradient descent (PGD) are explored as methods to learn the controller. In numerical examples, it is shown that with little knowledge of the plant model and a small data set, the dissipativity constrained learned controller achieves closed loop stability and successfully mimics the behavior of the expert controller, while other methods often fail to maintain stability and achieve good performance.

A novel model predictive controller for the drifting vehicle to track a circular trajectory

ABSTRACT In rally racing, a high sideslip cornering technique known as drifting is employed to minimise the lap time. However, for autonomous vehicles, tracking the trajectory while drifting is a challenging control problem due to the conflict between the two objectives. This paper proposes a novel model predictive controller (MPC) with two operation modes to address this conflict. In Mode 1, fuzzy expert strategies combined with the lateral error are designed to calculate the drift equilibrium point. Subsequently, the linear MPC controller is utilised to achieve quick adjustments of the lateral error. In Mode 2, terminal state constraints and error constraints are added to the MPC optimisation problem. The stability of the MPC controller is guaranteed to keep the vehicle within the expected error range. Finally, a comparison between the existing expert controller and the proposed controller is carried out on both the CarSim platform and a 1:10 scale race car. The results demonstrate that the proposed controller outperforms the existing expert controller in terms of tracking performance and drift stability.

COMPLEX MATHEMATICAL MODEL OF THE PROCESS OF LOADING A MULTICOMPONENT CHARGE INTO A BLAST FURNACE

The rational mode of loading the blast furnace is the most important condition for its highly efficient operation. The selection of rational values of the control parameters of the load mode is carried out on the basis of information obtained with the help of instrumental tools and mathematical models. Mathematical models of the process of blast furnace loading are necessary components of expert (intelligent) control systems for blast furnace melting. The presented complex mathematical model of the process of loading a multicomponent charge into a blast furnace differs from known developments taking into account the redistribution of components into the volume of loaded portions during movement along the path "charge feed - blast furnace", which provides the possibility of obtaining calculated characteristics of the distribution of each component of the charge and determining the composition of the formed mixtures of charge materials in any given zone of the furnace. The existence of such a model opens up new opportunities in the management of the blast furnace smelting process, as well as in conducting analytical studies of the conditions of slag formation and the distribution of properties of melts across the section of the blast furnace.

Enhancing UAV Aerial Docking: A Hybrid Approach Combining Offline and Online Reinforcement Learning

In our study, we explore the task of performing docking maneuvers between two unmanned aerial vehicles (UAVs) using a combination of offline and online reinforcement learning (RL) methods. This task requires a UAV to accomplish external docking while maintaining stable flight control, representing two distinct types of objectives at the task execution level. Direct online RL training could lead to catastrophic forgetting, resulting in training failure. To overcome these challenges, we design a rule-based expert controller and accumulate an extensive dataset. Based on this, we concurrently design a series of rewards and train a guiding policy through offline RL. Then, we conduct comparative verification on different RL methods, ultimately selecting online RL to fine-tune the model trained offline. This strategy effectively combines the efficiency of offline RL with the exploratory capabilities of online RL. Our approach improves the success rate of the UAV’s aerial docking task, increasing it from 40% under the expert policy to 95%.

What We Do in the Shadows: How expert workers reclaim control in digitalized and centralized organizations through ‘stealth work’

Organizations often depend on experts to carry out complex tasks that require specialized or tacit knowledge. Yet, organizations often want to increase their control over how tasks are performed and thus reduce the autonomy of experts. In the past, scholars have argued that experts had the ability to rebuff organizational attempts to control them. However, in an era with increased digitalization and centralization in organizations, experts risk losing control. How experts react when facing this increased centralization and digitalization is not well understood. Thus, this study seeks to improve knowledge on how experts react as organizations digitalize and centralize control over tasks. To do so, we studied a large energy company, which sought to increase its control over tasks and reduce the autonomy of its expert engineers by implementing an organizational change that included centralization and digital control. Using in-depth interviews, we portray how the expert workers reclaimed control using three micro-level tactics – strategic compliance and workaround, using legacy to reclaim control and concealing expert control. Based on these findings, our paper makes three contributions to the literature on experts and control. First, we provide the concept of ‘stealth work’, outlining how experts can reclaim control when centralization and digitalization have otherwise stripped them of status and power. Second, we highlight how expert control may be nested in organizations as a legacy, which experts can use when facing centralization and digitalization, and finally, we highlight how experts can engage in small hacks that curb the usefulness of digital control systems.

Research and Simulation of Fuzzy Expert Anti-surge Control System for Compressor

In order to solve the problem of compressor surge caused by flow fluctuations that cannot be effectively solved by conventional PID control, a fuzzy expert PID controller was designed based on the analysis of the working principles of fuzzy controllers and expert controllers. This controller has the advantages of good stability and high control accuracy of fuzzy control, as well as the advantage of fast response time of expert controllers. The MATLAB simulation results show that the fuzzy expert PID control algorithm has the characteristics of high stability, fast response speed, and small overshoot compared to conventional PID.

Data-Driven Inverse Reinforcement Learning Control for Linear Multiplayer Games.

This article proposes a data-driven inverse reinforcement learning (RL) control algorithm for nonzero-sum multiplayer games in linear continuous-time differential dynamical systems. The inverse RL problem in the games is solved by a learner reconstructing the unknown expert players' cost functions from demonstrated expert's optimal state and control input trajectories. The learner, thus, obtains the same control feedback gains and trajectories as the expert, only using data along system trajectories without knowing system dynamics. This article first proposes a model-based inverse RL policy iteration framework that has: 1) policy evaluation step for reconstructing cost matrices using Lyapunov functions; 2) state-reward weight improvement step using inverse optimal control (IOC); and 3) policy improvement step using optimal control. Based on the model-based policy iteration algorithm, this article further develops an online data-driven off-policy inverse RL algorithm without knowing any knowledge of system dynamics or expert control gains. Rigorous convergence and stability analysis of the algorithms are provided. It shows that the off-policy inverse RL algorithm guarantees unbiased solutions while probing noises are added to satisfy the persistence of excitation (PE) condition. Finally, two different simulation examples validate the effectiveness of the proposed algorithms.

Development of a human-analogue, 3-symptom domain Dog ADHD and Functionality Rating Scale (DAFRS)

The family dog, in its natural environment, exhibits neuropsychological deficits redolent of human psychiatric disorders, including behaviours that are similar to human attention-deficit/hyperactivity disorder (ADHD) symptoms. Based on standard questionnaire methods in humans, we aimed to develop and validate a detailed, psychometrically improved tool to assess owner views on relevant dog behaviours. We modified available questionnaires by adding items that allow for separate analysis of impulsivity, and items on functional impairment. We collected data from 1168 owners for different validation steps of the new questionnaire and, similarly to assessment of humans where teachers also evaluate as an expert control, we collected data from dog trainers. Exploratory and confirmatory factor analysis revealed 3 factors: inattention (IA), hyperactivity (H) and impulsivity (I), corresponding to all three human symptom dimensions in dogs. Test–retest analyses showed excellent agreement between measurements for all factors. Similarly to findings with humans, trainer-owner rating comparisons showed fair (IA) to moderate (H, I) agreement. As in humans, greater ADHD scores were associated with greater functional impairment scores. We suggest that in dogs, similarly to humans, parallel examination of (extreme) ADHD and functional impairment scores could help distinguish diagnosable individuals, after further validation of the questionnaire using a relevant behaviour test.

Resilient Event Triggered Interval Type‐2 Fuzzy Sliding Mode Control for Connected and Autonomous Vehicles Subjected to Multiple Cyber Attacks

Connected and autonomous vehicles (CAVs) are considered a hot area of research in the field of intelligent transportation systems. However, over the past few years, cybersecurity threats have posed significant challenges to such systems, given the ever‐evolving automotive industry. Thus, there is a growing need to design resilient control strategies to address the issue of cyberattacks. This article proposes the design of a distributed multiagent expert control scheme for cyberattack‐resilient control of CAVs. The study implemented an event‐triggered consensus‐based attack detection scheme capable of distinguishing between replay (RA), denial‐of‐service (DoS), and false data injection (FDI) attacks. The attacks in this study occur randomly, are bounded and time‐varying, and can overlap with each other. It was demonstrated that by considering an estimator error boundεfor the attacked signal reconstruction, the SMC controller in feedback with a vehicle remains stable in the sense of Lyapunov. Conditions were then provided that guarantee global asymptotic stability for a minimum dwell‐time constraint , and the platoon was shown to be string stable for the minimum distance between vehicles, denoted as . Finally, the performance of the control strategy was evaluated using multiple performance indices, considering platoons of varying sizes.

Multiport Energy Management System Design for a 150 kW Range-Extended Towing Vessel

This paper proposes a multiport energy management system (EMS) and its rule-based expert control strategy for a 150 kW range-extended towing vessel (RETV). The system integrates a diesel generator system, a permanent magnet synchronous motor, a lithium battery, and supercapacitors. To verify its feasibility and effectiveness, the proposed multiport EMS was modelled and tested through MATLAB/Simulink. Simulation results demonstrate that the designed multiport EMS works efficiently under the five typical operating conditions of the 150 kW RETV. In addition, two case studies were conducted and compared to investigate the impact of the battery’s initial state of charge (SoC) on the system’s energy efficiency. It was found that an overall 85% energy efficiency can be achieved for the RETV when the initial SoC is either 75% or 15%. The battery consistently operates within the optimal SoC range of 20% to 80%, and the supercapacitors effectively meet the instantaneous high-power demand.

End-to-end decentralized formation control using a graph neural network-based learning method.

Multi-robot cooperative control has been extensively studied using model-based distributed control methods. However, such control methods rely on sensing and perception modules in a sequential pipeline design, and the separation of perception and controls may cause processing latencies and compounding errors that affect control performance. End-to-end learning overcomes this limitation by implementing direct learning from onboard sensing data, with control commands output to the robots. Challenges exist in end-to-end learning for multi-robot cooperative control, and previous results are not scalable. We propose in this article a novel decentralized cooperative control method for multi-robot formations using deep neural networks, in which inter-robot communication is modeled by a graph neural network (GNN). Our method takes LiDAR sensor data as input, and the control policy is learned from demonstrations that are provided by an expert controller for decentralized formation control. Although it is trained with a fixed number of robots, the learned control policy is scalable. Evaluation in a robot simulator demonstrates the triangular formation behavior of multi-robot teams of different sizes under the learned control policy.

Hierarchical Intelligent Control Method for Mineral Particle Size Based on Machine Learning

Mineral particle size is an important parameter in the mineral beneficiation process. In industrial processes, the grinding process produces pulp with qualified particle size for subsequent flotation processes. In this paper, a hierarchical intelligent control method for mineral particle size based on machine learning is proposed. In the machine learning layer, artificial intelligence technologies such as long and short memory neural networks (LSTM) and convolution neural networks (CNN) are used to solve the multi-source ore blending prediction and intelligent classification of dry and rainy season conditions, and then the ore-feeding intelligent expert control system and grinding process intelligent expert system are used to coordinate the production of semi-autogenous mill and Ball mill and Hydrocyclone (SAB) process and intelligently adjust the control parameters of DCS layer. This paper presents the practical application of the method in the SAB production process of an international mine to realize automation and intelligence. The process throughput is increased by 6.05%, the power consumption is reduced by 7.25%, and the annual economic benefit has been significantly improved.

Reward inference of discrete-time expert's controllers: A complementary learning approach

Uncovering the reward function of optimal controllers is crucial to determine the desired performance that an expert wants to inject to a certain dynamical system. In this paper, a reward inference algorithm of discrete-time expert's controllers is proposed. The approach is inspired by the complementary mechanisms of the striatum, neocortex, and hippocampus for decision making and experience transference. These systems work together to infer the reward function associated to expert's controller using the complementary merits of data-driven and online learning methods. The proposed approach models the neocortex system as two independent learning algorithms given by a Q-learning algorithm and a gradient identification rule. The hippocampus is modelled by a least-squares update rule that extracts the relation from the states and control inputs of the expert's data. The striatum is modelled by an inverse optimal control algorithm which iteratively finds the hidden reward function. Lyapunov stability theory is used to show the stability and convergence of the proposed approach. Simulation studies are given to demonstrate the effectiveness of the proposed complementary learning algorithm.

Learning Deep Neural Network Controller for Path Following of Unicycle Robots

This letter investigates the scope of deep neural network (DNN) based controller in the path following task for unicycle mobile robots. A DNN-based controller is trained to follow paths with arbitrary curvature in two-dimensional space. The training process does not require initialization or supervision from any other known expert controller. Rather, the training of the DNN controller is guided by another predictive neural network that represents a path following error dynamics which is exponentially stable at the origin. The two DNNs are trained jointly in a simulated environment. The learned DNN controller is then employed as a standalone controller in a real unicycle robot for the tasks of following various linear and curved paths.

EQGP concept: Principles and models of expert quality control of the government policy

The article studies the problems of quality and efficiency of public administration. The article describes foundations of a new concept of public administration – the concept of expert quality control of the government policy (EQGP), which is an effective way to improve the quality and efficiency of government in post-industrial (information) society. In order to improve the efficiency of government administration, it is necessary to improve the mechanisms for forming public bodies and to implement monitoring mechanisms into the system of public bodies formation. To achieve this goal, it is necessary to activate expert communities (“expert factories”), to enhance expert and analytical activities both at the stage of formation of public bodies (examination of election programs) and monitoring of their activities (examination of reports). Depending on the positions occupied by experts in the decision-making system, four models of expert control have been identified. The features of each model are presented. The article also describes the basic principles of EQGP.

Heavy Metal Evaluation of Overused Commercial Fertilizers and Their Interactions with Soil Properties

Şanlıurfa province has 36% of the agricultural areas in Turkey and 64.1% of the agricultural areas in the GAP Region. With approximately 600 000 tons of chemical fertilizer consumption in 2021, it is the province with the most fertilizer consumption in Turkey. This causes some negative and high environmental risks such as salinization in the soil, heavy metal (HM) accumulation, deterioration of nutrient balance, damage to microorganism activity, and formation of eutrophication in the region. The objective of this study was to determine the HMs (Zn, Ni, Mn, Cu, Mo, Pb, Cd) concentrations of the soils and some commercial fertilizers overused, and evaluate their interactions with soil properties. The average values of HM concentration of the soil are as follows; 32.65 and 46.88 mg kg-1 Zn; 649.03 and 730.58 mg kg-1 Mn; 79.86 and 95.54 mg kg-1 Ni; 0.15 to 0.27 mg kg-1 Cd; 0.26 and 0.97 mg kg-1 Mo; 8.54 and 18.67 mg kg-1 Pb; 19.45-25.37 mg kg-1 Cu. HM contents of some fertilizers were found to be very high in this study. This causes an increase in the HM concentration in the soil. HMs concentrations of several soil samples exceeded the threshold level of Europe standards except for Ni, Mo, and Mn. Study results can help the authorities to develop effective fertilizer management strategies for the Harran Plain, which has once again revealed the necessity of applying agricultural activities such as fertilization with a fertilization program prepared under expert control according to the results of soil analysis.