Random Step Changes Research Articles

Dynamic Control of a Fiber Manufacturing Process Using Deep Reinforcement Learning

This article presents a model-free deep reinforcement learning (DRL) approach for controlling a fiber drawing system. The custom DRL-based control system predictively regulates fiber diameter and produces a fiber with a desired, constant or nonconstant, diameter trajectory, i.e., diameter variation along the fiber length. Physical models of the system are not used. The system was trained and tested on a compact fiber drawing system, which has nonlinear delayed dynamics and stochastic behaviors. For a reference trajectory with random step changes, after 1 h of training, the DRL controller showed the same root-mean-squared error (RMSE) as an optimized PI controller; after 3 h of training, it achieved the performance of a quadratic dynamic matrix controller (QDMC). While the PI feedback controller showed 3.5 s of time lag in a step response, the DRL controller showed less than a second of time lag. Controller performance tests on trajectories not used in the training process are conducted; for a sine sweep reference trajectory, the DRL controller maintained an RMSE under 40 <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$\mu \text{m}$</tex-math></inline-formula> up to a frequency of 45 mHz, compared to 25 mHz for QDMC.

Discharge performance and dynamic behavior of refuellable zinc-air battery

Zinc-air batteries (ZABs) are considered a promising energy storage system. A model-based analysis is one of the effective approaches for the study of ZABs. This technique, however, requires reliable discharge data as regards parameter estimation and model validation. This work, therefore, provides the data required for the modeling and simulation of ZABs. Each set of data includes working time, cell voltage, current, capacity, power, energy, and temperature. The data can be divided into three categories: discharge profiles at different constant currents, dynamic behavior at different step changes of discharge current, and dynamic behavior at different random step changes of discharge current. Constant current discharge profile data focus on the evolution of voltage through time. The data of step changes emphasize the dynamic behavior of voltage responding to the change of discharge current. Besides, the data of random step changes are similar to the data of step changes, but the patterns of step changes are random. Such data support the modeling of a zinc-air battery for both theoretical and empirical approaches.

Compensatory mechanisms of balance to the scaling of arm-swing frequency

The present study investigated the contribution of the Hof (2007) mechanism 1 (M1-moving the center of pressure (COP) with respect to the vertical projection of the center of mass (COMTotal)); and mechanism 2 (M2-rotating the trunk and upper limbs around the COMTotal) to postural control and the stability of COP–COMTotal cophase as a function of lateral arm-swing frequency. Young adults were instructed to stand still on a force platform while alternating their arm swinging from above the head to the side of their thigh to create perturbations to postural control. Scaling the frequency of arm-swing (random step changes of 0.2Hz within a bandwidth of 0.2 to 1.6Hz) increased the SD of COP but decreased the SD of COMTotal. Increments in arm-swing frequency induced a progressive increase in M1 and decrease in M2 in terms of their relative contribution to postural stability. The cophase between COP and COMTotal became more tightly in-phase over increments of arm-swing frequency. These findings show an adaptive compensatory role of M1 and M2 within the stability of COP–COMTotal coupling in the regulation of human balance control.

Genetic Optimization based Adaptive Fuzzy Logic Control of a pH Neutralization Process

Presence of severe nonlinearity in pH neutralization process makes the problem of pH control a difficult and challenging one. Further, optimized control of such highly nonlinear system requires tuning of various parameters with a suitable evolutionary algorithm (EA). This paper presents genetic algorithm (GA) based unconstrained, continuous and single-objective optimization of conventional and adaptive nonlinear fuzzy logic control (FLC) of pH neutralization process for servo and regulatory operations. The GA optimized adaptive FLC scheme changes the universe of discourse of input-output membership functions based on the region of operation and amount of step change in set-point and load. The performance of controllers are evaluated using integral of squared errors (ISE) and tested for random step changes in set-point and load.

Control of indirect matrix converter with bidirectional output stage for micro-turbine

In this study, a new AC-DC-AC matrix converter is developed. This matrix converter is designed to connect a micro-turbine to the grid, to replace the traditional system used in micro-turbines, which is based on an inverter-rectifier system (VSR/VSI) with energy storage elements. A specific power circuit is introduced. The circuit consists of two stages, where the second stage is bidirectional. Also, a new switching technique is presented for the circuit mentioned above, which works with two voltage space vectors. The first vector corresponds to the input voltages (variable frequency) and second one to the output voltages (fixed frequency). This technique switching allowing to replace the VSR/VSI regarding operational concerns, with less loss, at lower cost and lower weight. For verifying the control algorithm, a laboratory prototype has been developed, using among other elements a Spectrum Digital eZdspTM TMS320F2812 card. In addition, the model predictive control (MPC) is used to control the electrical power of the micro-turbine generator by manipulating the fuel flow. The suggested method provided the optimal functioning not only for the small step changes but also for large random step changes.

Non-linear Control of Heave for an Unmanned Helicopter Using a Neural Network

This paper describes a new non-linear control technique applied to the heave control of an unmanned rotorcraft. First a hybrid plant model consisting of exactly known dynamics is combined with a black-box representation of the unknown dynamics. Desired trajectories are calculated to smoothly achieve a sequence of random step changes in desired height according to certain optimal criterion and plant limitations. Control inputs are then determined using the MATLAB® optimisation toolbox to achieve those desired trajectories for the plant heave model. Finally, a neural network is trained to mimic the control inputs resulting from the optimisation process. The neural network controller produces trajectories closely resembling the results from the optimisation process but with a much reduced computation time. Flight test results of control of the heave dynamics of a helicopter confirm the neural network controller's ability to operate in high disturbance conditions and outperform a proportional-derivative (PD) controller.

Design and Performance Analysis of the Exponentially Weighted Moving Average Mean Estimate for Processes Subject to Random Step Changes

The exponentially weighted moving average (EWMA) is a well-known and popular statistic used for smoothing and forecasting time series and as a process mean estimator, due to its simplicity and ability to capture nonstationarity. The EWMA statistic has been shown to be an optimal mean estimator for a certain disturbance process and an effective estimator for various other processes. In this article we focus on a practical disturbance process—relatively small random step changes that are difficult to distinguish from white noise and usually overlooked by practitioners. We propose an optimal EWMA parameter for step-change disturbance processes, as well as methodologies to identify and estimate the process models. The EWMA estimator's performance is then evaluated analytically. We demonstrate that a well-designed EWMA control scheme can effectively reduce the process variation even for processes subject to infrequent, small step changes. A semiconductor process example illustrates the design and analysis.

The effect of random step changes of system parameters on the stability of steady vibration of nonlinear systems

The author proposes a method of analysing the effect of random step changes of system parameters on the stability of steady solution of a nonlinear system in case this steady solution is not the only one. The method is based on the assumption that the time between successive step changes is comparatively long against the period of steady vibration so that transients become stabilized after each parameter change. The parameter is free to vary on a finite interval of values.