Real-time Control Applications Research Articles

Rainfall-runoff modelling using octonion-valued neural networks

ABSTRACT Rainfall-runoff modelling is at the core of any hydrological forecasting system. High spatio-temporal variability of precipitation patterns, complexity of the physical processes, and large quantity of parameters to characterize a watershed make the prediction of runoff rates quite difficult. In this study, a hyper-complex Artificial Neural Network (ANN) in the form of an Octonion-Valued Neural Network (OVNN) is proposed to estimate runoff rates. Evaluation of the proposed model is performed using a rainfall time series from a rain gauge near a Canadian watershed. Results of the AI-generated runoff rates illustrate its capacity to produce more computationally efficient runoff rates when compared to those obtained using a physically-based model. In addition, training the data using the proposed OVNN versus a real-valued neural network shows less space-complexity (1*3*1 vs. 8*10*8, respectively) and more accurate results (0.10% vs. 0.95%, respectively), that accounts for the efficiency of the OVNN model for real-time control applications.

Development and validation of nonlinear dynamic engine airpath models for real-time advanced control application

This paper proposes model reduction techniques for reducing engine airpath models in real-time (RT), for a control oriented application in an engine equipped with high pressure exhaust gas recirculation (EGR-HP) and single stage turbocharger. There are two major challenges addressed by authors: First, reducing the order without compromising the performance in terms of accuracy by decoupling the nonlinear coupled differential equations of airpath system in-order to use them in real-time processor-in-loop control application. A model reduction technique based on different dynamic characteristics between thermodynamic states followed by semi-implicit Euler (SIE) numerical method to solve coupled dynamic multi-input multi-output (MIMO) differential equation models is demonstrated. Second, the authors have proposed a novel method to calculate gas mass flow via compressor, coupled with engine airpath model in real-time. The proposed models of airpath system coupled with turbocharger models for a diesel engine is validated with experimental data to evaluate performance of pressures, temperatures, mass flows at relevant components. The developed airpath model is used for calculating thermodynamic properties in real-time for state of art engine control unit (ECU) in production engine and become basis for feedforward as well as closed loop control of airpath variables for real-time system. Authors further propose to use this modeling approach for calculating airpath system variables for exhaust aftertreatment system, injection system, and for virtualization of sensor values in airpath systems.

Real Time Control Application of the Robotic Arm Using Neural Network Based Inverse Kinematics Solution

Robotic arms are widely used in many industrial applications at present. The control of robotic arms involves position coordination Cartesian space by a forward/inverse kinematics solution method. The inverse kinematics is difficult for real-time control applications, computational requirements are intensive and the run-time is high. The traditional solution methods used geometric, algebraic, and numerical iterative techniques are inadequate and slow in the inverse kinematics solution. Recently, alternative solution methods based on artificial intelligence techniques have been developed to solve the inverse kinematics problem. In this study, a multi-layered feed-forward Artificial Neural Network model was developed to solve the inverse kinematics of the 5 degrees of freedom robotic arm. Using the Proportional-Integral control algorithm combined with this Artificial Neural Network model, the real-time position control of the robotic arm was accomplished. The obtained results were compared with the PI control supported by an analytical inverse kinematics solution in real-time. The results showed that the PI control combined with Artificial Neural Network has superior tracking ability, smaller control error, and better absolute fit to the reference.

Plasma current profile reconstruction for EAST based on Bayesian inference

Determining the distribution of plasma current in the equilibrium state is one of the most important steps to realize effective and safe operation of tokamak. In this study, a novel reconstruction code based on Bayesian inference is developed to infer the plasma current distribution for experiment analysis of EAST. Without iteratively solving Grad-Shafranov (G-S) equation to find an optimal fit for the external magnetic diagnostic measurements, the distribution of toroidal current density and poloidal flux can be rapidly derived in a probabilistic manner. The reconstructed results are consistent with the results based on equilibrium fitting (EFIT) code, and the execution time is less than 1 ms for each time slice, which indicates its potential for application in future real-time plasma feedback control.

Improving the accuracy and efficiency of online calibration for simulation-based Dynamic Traffic Assignment

Simulation-based Dynamic Traffic Assignment models have important applications in real-time traffic management and control. The efficacy of these systems rests on the ability to generate accurate estimates and predictions of traffic states, which necessitates online calibration. A widely used solution approach for online calibration is the Extended Kalman Filter (EKF), which – although appealing in its flexibility to incorporate any class of parameters and measurements – poses several challenges with regard to calibration accuracy and scalability, especially in congested situations for large-scale networks. This paper addresses these issues in turn so as to improve the accuracy and efficiency of EKF-based online calibration approaches for large and congested networks. First, the concept of state augmentation is revisited to handle violations of the Markovian assumption typically implicit in online applications of the EKF. Second, a method based on graph-coloring is proposed to operationalize the partitioned finite-difference approach that enhances scalability of the gradient computations.Several synthetic experiments and a real world case study demonstrate that application of the proposed approaches yields improvements in terms of both prediction accuracy and computational performance. The work has applications in real-world deployments of simulation-based dynamic traffic assignment systems.

Back Cover: Advanced Real‐Time Process Analytics for Multistep Synthesis in Continuous Flow (Angew. Chem. Int. Ed. 15/2021)

Advanced process analytics are becoming an increasingly important part of continuous flow chemistry. In their Research Article on page 8139, Jason D. Williams, C. Oliver Kappe and co-workers describe the use of four different process analytical technologies, controlled by one central system, in a multistep continuous flow process. Advanced data processing methods, including artificial intelligence, are used to accurately quantify up to nine process species in real time, toward self-optimization and model predictive control applications.

Nature-Based Solutions and Real-Time Control: Challenges and Opportunities

Nature-based solutions (NBS) as green infrastructures to urban drainage are an effective mitigation strategy both in terms of quantity and quality of runoff. Real-time control (RTC) can complement both flood mitigation and improvement of water quality by controlling elements of the drainage and sewage system. This study assessed the improvement opportunities with RTC of three NBS-related techniques commonly applied in urban drainage with different spatial scales: green roof, bioretention and detention basin and the remaining challenges to integrate both methods. Additionally, our investigations showed that the main difficulties reported involve the planning and monitoring stages of the RTC system. All of the studied devices can benefit from RTC. It is possible to observe that, despite the good results reported in the literature, the application of RTC to NBS studies on urban drainage are very recent. There are several opportunities that can be explored to optimize the performance.

Short-term water demand predictions coupling an artificial neural network model and a genetic algorithm

Abstract The application of artificial neural network (ANN) models for short-term (15 min) urban water demand predictions is evaluated. Optimization of the ANN model's hyperparameters with a genetic algorithm (GA) and use of a growing window approach for training the model are also evaluated. The results are compared to those of commonly used time series models, namely the Autoregressive Integrated Moving Average (ARIMA) model and a pattern-based model. The evaluations are based on data sets from two Canadian cities, providing 15 min water consumption records over respectively 5 years and 23 months, with a respective mean water demand of 14,560 and 887 m3/d. The GA optimized ANN model performed better than the other models, with Nash–Sutcliffe Efficiencies of 0.91 and 0.83, and relative root mean square errors of 6 and 16% for City 1 and City 2, respectively. The results of this study indicate that the optimization of the hyperparameters of an ANN model can lead to better 15 min urban water demand predictions, which are useful for many real-time control applications, such as dynamic pressure control.

Communication Scheduling for Control Performance in TSN-Based Fog Computing Platforms

In this paper we are interested in real-time control applications that are implemented using Fog Computing Platforms consisting of interconnected heterogeneous Fog Nodes (FNs). Similar to previous research and ongoing standardization efforts, we assume that the communication between FNs is achieved via the IEEE 802.1 Time Sensitive Networking (TSN) standard. We model the control applications as a set of real-time flows, and we assume that the messages are transmitted using scheduled traffic that is using the Gate Control Lists (GCLs) in TSN. Given a network topology and a set of control applications, we are interested to synthesize the GCLs for messages such that the Quality-of-Control (QoC) of control applications is maximized and the deadlines of real-time messages are satisfied. We have proposed a Constraint Programming (CP)-based solution to this problem, and developed an accurate analytical model for QoC, which, together with a metaheuristic search employed in the CP solver can drive the search quickly towards good quality solutions. We have evaluated the proposed strategy on several test cases including realistic test cases and also validate the resulted GCLs on a TSN hardware platform and via simulations in OMNET++.

An IoT LoRaWAN Network for Environmental Radiation Monitoring

A reliable and highly scalable Internet of Things (IoT) end-to-end data infrastructure has been developed for environmental radiation monitoring at the European Organization for Nuclear Research (CERN) based on a low-power wide-area network (LPWAN). The proposed system, called Waste radiation MONitoring (W-MON), consists of an interconnected network of thousands of highly sensitive and ultralow-power gamma radiation sensors acting as long range (LoRa) transceivers. The aim of the system is to improve and automatize the radiological controls of conventional waste containers. The end devices measure the radiation levels in the waste containers on a continuous basis and send the data periodically to the LoRaWAN network server. The network has been deployed in an outdoor environment covering hundreds of hectares. A set of web-based user applications for real-time monitoring, data visualization, and status control of the devices have been designed based on open-source tools. The data pipeline infrastructure has been designed to allow an easy integration into the overall CERN Radiation and Environment Monitoring Unified Supervision service (REMUS).

Hybrid Machine Learning and Estimation-Based Flight Trajectory Prediction in Terminal Airspace

For air traffic management, trajectory prediction plays an important role as the predicted trajectory information is used in crucial tasks for the safety and efficiency of air traffic operations, such as conflict detection and resolution, scheduling, and sequencing. In this paper, we propose a framework for trajectory prediction in terminal airspace by combining a machine learning-based method and a physics-based estimation method. A trajectory prediction model based on machine learning is trained from historical surveillance data to represent the collective behavior of a set of flight trajectories, from which the data-driven prediction can be obtained as the expected future behavior of an incoming flight. A physics-based estimation algorithm called Residual-Mean Interacting Multiple Models (RM-IMM) then incorporates the machine learning prediction as a pseudo-measurement to account for the current motion of the aircraft. The proposed framework is tested, with real air traffic surveillance data, by predicting the future state information of the flights for real-time air traffic control applications. The results show that the proposed framework produces a greatly improved prediction accuracy compared to the two existing machine learning-based algorithms.

Advanced-multi-step Moving Horizon Estimation

Abstract Moving Horizon Estimation (MHE) is an important optimization-based approach for state estimation and parameter updates, because of its capabilities in dealing with nonlinearity and state constraints. In addition, one of the applications is to provide the full state information for Model Predictive Controller (MPC) to control the process in either setpoint tracking or economic control purposes. However, the computational burden of MHE could deteriorate the control performance if the feedback delay caused by computation is too long, leading to potential safety issues or process damage. In this paper, we propose a fast moving horizon estimation algorithm to overcome the long computational time of MHE for real-time control applications, especially for fast dynamics or large-scale systems. We exploit the nonlinear programming (NLP) sensitivity and make use of efcient NLP solvers, IPOPT and k_aug, to reduce the on-line computational costs. This new approach is demonstrated on a CSTR process, where results are compared to ideal MHE and advanced-step MHE (asMHE).

Recovering Software for the Whirlwind Computer

The late 1940s were seminal years in the development of electronic digital computing machines, as researchers began to realize the enormous potential held by these machines. One of these new generations of computing devices was the vacuum-tube Whirlwind computer, designed in the late 1940s at the Massachusetts Institute of Technology. Designed from the start for real-time control applications, Whirlwind evolved to be a key element in a proof of concept for national air defense. The article starts with a review of Whirlwind and its evolving software environment, goes on to describe the current effort to decode some of the Whirlwind software artifacts remaining in museum archives, and then describes some of the material the work has made accessible, allowing us to further study some of the software developed during that formative decade.

Combined Application of Real-Time Control and Green Technologies to Urban Drainage Systems

The increase in waterproof surfaces, a typical phenomenon of urbanization, on the one hand, reduces the volume of rainwater that naturally infiltrates the subsoil and, on the other, it determines the increase in speeds, flow rates, and outflow volume surface; at the same time, it causes a qualitative deterioration of the water. This study researched the optimal management of urban drainage systems via the combined application of real-time control and green technologies. A hydraulic model of the sewer system of the suburbs of Bologna (Italy) was set up using the Environmental Protection Agency (EPA) Storm Water Management Model (SWMM) to evaluate the reduction in water volume and the masses of pollutants discharged in water bodies. The combined application of these technologies allows significantly reducing both the pollutants released into the receiving water bodies and the overflow volumes, while optimizing the operation of the treatment plants. Green technologies cause an average reduction equal to 45% in volume and 53% of total suspended solids (TSS) sent to the receiver. The modeled cases represent only some of the possible configurations achievable on urban drainage systems; the combined use of different solutions could lead to further improvements in the overall functioning of the drainage system.

Resource-Optimal Fault-Tolerant Scheduler Design for Task Graphs Using Supervisory Control

Real-time control applications are highly parallelizable and can be used to effectively harness the capacity of a given computing platform when appropriately scheduled. Given a multicore platform for executing a set of parallelizable applications, it is necessary to ensure proper functioning of the system even in the presence of transient processor faults. However, most existing scheduling approaches for parallel applications have been heuristic schemes which are often based only on the satisfaction of a set of sufficiency conditions and cannot take into consideration of all necessary schedulability requirements. Consequently, such schemes lead to suboptimal usage of resources resulting in higher design costs. Formal model-based safe design methodologies such as supervisory control are often desirable in the design of correct-by-construction fault-tolerant schedulers. This article proposes a supervisory control-based fault-tolerant scheduler synthesis scheme for real-time tasks modeled as precedence-constrained task graphs, executing on multicores. Further, we devise search strategies to obtain schedules that 1) maximize fault-tolerance and 2) minimize peak-power (MPP) dissipation. Conducted experiments using real-world benchmarks reveal the efficacy of our scheme.

Brain–computer interface‐based single trial P300 detection for home environment application

P300 speller-based brain–computer interface (BCI) is an immediate correspondence between the human brain and computer that depends on the translation of mind reactions produced by the stimulus of a subject utilising the P300 speller. No muscle movements are required for this communication. As a P300 paradigm, a novel 2 × 3 matrix consisting of visual home appliances is proposed, which helps disabled people ease their lives by accessing mobile, light, fan, door, television, electric heater etc. In most of the current P300-based BCIs, 5–15 trials work better and the low information transfer rate (ITR) is a major issue in its adaptation in real-time. The objective of this Letter is to improve accuracy as well as an ITR for real-time home appliance control applications. To address this, the authors proposed a single trial weighted ensemble of compact convolution neural network and obtained an ITR of 46.45 bits per minute and an average target appliance accuracy of 93.22% for the BCI-based home environment system. The experimental findings confirmed the feasibility of the proposed method and thus can provide guidance for future use of the system for paralysed patients.

Optimally designed Lyapunov–Krasovskii terminal costs for robust stable–feasible model predictive control of uncertain time-delay nonlinear dynamical systems

This article details a new Model Predictive Control algorithm ensuring robust stability and control feasibility for uncertain nonlinear multi-input multi-output dynamical systems considering uncertain time-delay effects. The proposed control algorithm is based on construction of a Lyapunov–Krasovskii functional as terminal cost. Incorporation of this terminal cost into the Model Predictive Control optimization problem and calculation of the associated admissible set result in robust feasibility and robust stability of closed-loop system in presence of uncertain time-delay effects and bounded disturbance signals. The Lyapunov–Krasovskii functional term is constructed with respect to predicted sliding functions over the prediction horizon and considers the effects of dynamical variations over the prediction horizon in generation of control inputs. As dynamical variations are investigated in a sample-to-sample basis, feasible sliding regions are updated at each sample as well. Finally, based on expression of sliding functions as a combination of dynamical variations and input-based terms, required control inputs are calculated in the admissible bound by the optimization algorithm. Construction of control scheme on this basis permits straightforward calculation of robust stability and feasibility conditions for a general class of uncertain nonlinear system in finite prediction horizon whereas in the previous works, often-restrictive conditions were considered for the investigated dynamical systems. Numerical illustrations indicate precision and efficiency of control algorithm and improved stability and convergence rate for multivariable nonlinear dynamical systems considering uncertain time-delay effects. Finally, hardware-in-the-loop implementation indicates applicability of the proposed scheme in real-time control applications particularly in case appropriate compromises between optimality and calculation speed are considered.

Joint Stochastic Spline and Autoregressive Identification Aiming Order Reduction Based on Noisy Sensor Data.

This article introduces the spline approximation concept, in the context of system identification, aiming to obtain useful autoregressive models of reduced order. Models with a small number of poles are extremely useful in real time control applications, since the corresponding regulators are easier to design and implement. The main goal here is to compare the identification models complexity when using two types of experimental data: raw (affected by noises mainly produced by sensors) and smoothed. The smoothing of raw data is performed through a least squares optimal stochastic cubic spline model. The consecutive data points necessary to build each polynomial of spline model are adaptively selected, depending on the raw data behavior. In order to estimate the best identification model (of ARMAX class), two optimization strategies are considered: a two-step one (which provides first an optimal useful model and then an optimal noise model) and a global one (which builds the optimal useful and noise models at once). The criteria to optimize rely on the signal-to-noise ratio, estimated both for identification and validation data. Since the optimization criteria usually are irregular in nature, a metaheuristic (namely the advanced hill climbing algorithm) is employed to search for the model optimal structure. The case study described in the end of the article is concerned with a real plant with nonlinear behavior, which provides noisy acquired data. The simulation results prove that, when using smoothed data, the optimal useful models have significantly less poles than when using raw data, which justifies building cubic spline approximation models prior to autoregressive identification.

Photoacoustic method for real-time assessment of salt content in aqueous solutions

In 2004, the Food and Drug Administration established the foundations for the application of process analytical technologies (PAT) in real-time control of the drug manufacturing process, where progress has been essentially directed to solid formulations. In order to enlarge the application of PAT principles to injectable drug products, the development of appropriate manufacturing process control tools is mandatory. Photoacoustics is a non-invasive technique with the potential for application in real-time control of the manufacturing process of injectable drug products. Herein, we applied a photoacoustic method for the determination of the concentration of salts (sodium chloride) in mono-salt formulations by measuring the changes induced in the speed of sound by density changes. This method was explored using two modes of generating the photoacoustic wave and two detectors with central frequencies of 10 MHz and 100 MHz. The results were analyzed using a 2k full-factorial design, considering the generation mode and detection as independent variables. The optimized method was subsequently validated according to the International Council for Harmonisation (ICH) standards. The method showed good linearity, precision, and accuracy, with a lower limit of quantification of 0.05% (w/v) of NaCl and a limit of detection of 0.02% (w/v) of NaCl. Due to its simplicity and high throughput, this method has potential applicability as PAT in the manufacturing of injectable drug products.

Timely updates in MEC-assisted status update systems: Joint task generation and computation offloading scheme

Fresh status updates are vital to the efficient operation of network monitoring and real-time control applications. In this paper, we consider a mobile edge computing (MEC)-assisted status update system, where smart devices extract valuable status updates from sensed data to achieve timely awareness of the surroundings by exploiting computational resources at the device and edge server. To quantify the freshness of status updates obtained by executing computation tasks, we employ the concept of age of information (AoI) to characterize the timeliness of status updates. To cope with the limited energy at devices, we investigate a joint task generation and computation offloading scheme under a given energy budget for minimizing the age of obtained status updates. The age minimization problem is modeled as a constrained Markov decision process (CMDP). To obtain the optimal policy, we derive the structural properties of the optimal deterministic policy and propose a lightweight structure-based status update algorithm in the case of known channel statistics. Moreover, we consider a more realistic scenario without prior knowledge of channel statistics, and propose a Q-learning-based status update algorithm to make online decisions. Simulation results show that the performance of our proposed algorithms is competitive when compared with existing schemes.