Design Of Multivariable Fractional Order Pid Controller Using Covariance Matrix Adaptation Evolution Strategy

Sivananaithaperumal, S.; Baskar, S.

doi:10.2478/acsc-2014-0014

Cited by 11 publications

(6 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Here are presented the results of the developed system. Since evolution strategies are comparable to reinforcement learning methods (Salimans et al, 2017), we will compare this method to reinforcement learning methods using the negative amplitude of lateral and angular errors as its rewards, and a constant gain method trained using CMA-ES on the same objective function as the proposed method, similarly to existing methods (Wakasa et al, 2010;Sivananaithaperumal and Baskar, 2014;Marova, 2016). Tests are focused on the trained system and not on the training phase.…”

Section: Resultsmentioning

confidence: 99%

Neuroevolution with CMA-ES for Real-time Gain Tuning of a Car-like Robot Controller

Hill

Lucet

Lenain

2019

Proceedings of the 16th International Conference on Informatics in Control, Automation and Robotics

View full text Add to dashboard Cite

This paper proposes a method for dynamically varying the gains of a mobile robot controller that takes into account, not only errors to the reference trajectory but also the uncertainty in the localisation. To do this, the covariance matrix of a state observer is used to indicate the precision of the perception. CMA-ES, an evolutionary algorithm is used to train a neural network that is capable of adapting the robot's behaviour in real-time. Using a car-like vehicle model in simulation. Promising results show significant trajectory following performances improvements thanks to control gains fluctuations by using this new method. Simulations demonstrate the capability of the system to control the robot in complex environments, in which classical static controllers could not guarantee a stable behaviour.

show abstract

Section: Resultsmentioning

confidence: 99%

Neuroevolution with CMA-ES for Real-time Gain Tuning of a Car-like Robot Controller

Hill

Lucet

Lenain

2019

Proceedings of the 16th International Conference on Informatics in Control, Automation and Robotics

View full text Add to dashboard Cite

show abstract

“…The appropriateness of the proposed variants of the SOA in designing the FOPI controller is also compared with other global optimization techniques (DE [11] and CMAES [12]) and popular classical local search techniques (Nelder-Mead simplex search (NM-SS) [9] and interior point method (IPM) [10]) in Table 3. It can be observed that the CSOA outperforms the other algorithms under consideration, while the ESOA is better than the other algorithms except for the CSOA and is comparable to DE.…”

Section: Soamentioning

confidence: 99%

“…Obtain the personal best position, neighborhood best position and population best position Determine the search direction and step size of individual seeker by adopting chaotic behavior using (19) Based on search direction & step size, update the position of individual seeker using (12) Evaluate the objective function for each of the updated seekers Evaluate and update position for personal best, subpopulation best and over-all best…”

Section: Start Random Initialization Of Seeker Populationmentioning

confidence: 99%

“…Random initialization of seeker populationDivide the seeker population into subpopulations Evaluate the objective function for each seeker Obtain the personal best position, neighborhood best position and population best position Determine the search direction and step size of individual seeker by adopting chaotic behavior using(19) Based on search direction & step size, update the position of individual seeker using(12) …”

mentioning

confidence: 99%

“…However, their performance was sensitive to the selection of the initial feed point and the solutions obtained were prone to getting trapped into local optima. In this context, a number of works have adopted population-based evolutionary algorithms such as particle swarm optimization [8], differential evolution (DE) [11], and the covariance matrix adaptation evolution strategy (CMAES) [12] to obtain global solutions for fractional controller design problems.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Fractional-order controller design in frequency domain using an improved nonlinear adaptive seeker optimization algorithm

Kumar¹,

Deepak²,

Ghosh³

2017

Turk J Elec Eng & Comp Sci

View full text Add to dashboard Cite

Abstract:Two nonlinear adaptive versions of the conventional seeker optimization algorithm (SOA) have been proposed for the design of fractional-order controllers using frequency domain specifications. The highly nonlinear and undetermined nature of equations resulting from controller design specifications rules out obtaining a closed-form solution. In this regard, the controller design task has been formulated as an optimization problem and solved using modified variants of the SOA. With the nonlinear adaptation of tuning parameters, the proposed variants of the SOA increase the probability of finding global optima along with improved convergence speed. This has been achieved by incorporating exponential weighting and chaotic behavior in the search process. The validation of the proposed techniques on a set of fractional-order controller design problems clearly exhibits their superiority over other algorithms and controller design techniques. The hardware implementation of the controllers on a DSP TMS320F2812 board ensures applicability for real-time applications.

show abstract

An extensive critique of sliding mode control and adaptive neuro‐fuzzy inference system for nonlinear system

George

Mani

Stonier³

2021

Asian Journal of Control

View full text Add to dashboard Cite

Nonlinear control hypothesis is the region of control hypothesis which manages frameworks that are nonlinear, time‐variant, or both. Nonlinear control structures have extended high prevalence fundamentally because of the wide use of hypothetical intends to deal with credible issues in different zones of engineering and medical aspects. Sliding mode control (SMC) is a nonlinear control strategy applied when the deterministic model of the nonlinear system is discrepant with actual plant in terms of plant parameters and undetermined external disturbances. SMC is a sort of nonlinear control that accentuates zero error convergence modeling error, demonstrating nonlinearities and outside aggravations. Adaptive neuro‐fuzzy inference system (ANFIS) captures neural associations and soft reason guidelines in a single framework. ANFIS causes fewer errors and is more obvious to the customer isolated from the ANN. The paper centers that neural organizations accept a critical capacity in controlling nonlinear frameworks. This review aims an overview of existing methods for nonlinear control and application of SMC for nonlinear system based on ANFIS.

show abstract

Design Of Multivariable Fractional Order Pid Controller Using Covariance Matrix Adaptation Evolution Strategy

Cited by 11 publications

References 13 publications

Neuroevolution with CMA-ES for Real-time Gain Tuning of a Car-like Robot Controller

Neuroevolution with CMA-ES for Real-time Gain Tuning of a Car-like Robot Controller

Fractional-order controller design in frequency domain using an improved nonlinear adaptive seeker optimization algorithm

An extensive critique of sliding mode control and adaptive neuro‐fuzzy inference system for nonlinear system

Contact Info

Product

Resources

About