Parameter Identification for a Quadrotor Helicopter Using Multivariable Extremum Seeking Algorithm

Weizhen Liu,Xin Huo,Jinkun Liu,Libin Wang 제어·로봇·시스템학회 2018 International Journal of Control, Automation, and Vol.16 No.4

Parameter identification for a quadrotor helicopter is challenging from a theoretical point of view. In this paper, a closed-loop multivariable extremum seeking algorithm (MESA) is proposed for a nonlinear quadrotor helicopter parameter identification with two groups of input data. The proposed scheme is universally applicable to the closed-loop identification for cross-coupling multivariable processes where the identification problem is formulated as a minimization of a cost function. As the gradients of the performance parameters are obtained by step response experiments, the whole system searches along the negative gradient of the cost function until the reference trajectory or point is derived. Since the cost function is treated as a mapping from the model parameters, then the parameters can be identified online and in a real-time manner. The procedure of the identification algorithm is presented, and its effectiveness is illustrated by numerical simulations.

Ant Colony Optimization Based Fixed Order Controller Design and System Simulations International Conference on Control, Automation and Systems

Faizullah Mahar,Saad Azhar Ali,Ayaz Hussain,Zohaibuddin Bhutto 제어로봇시스템학회 2011 제어로봇시스템학회 국제학술대회 논문집 Vol.2011 No.10

Thus far, researchers test several heuristic in order to produce the optimal solutions of the cost function and controller parameters. In this paper, an implementation of the Ant Colony Optimization (ACO) algorithm for the optimization of cost function and controller parameters has been discussed. To implement the ACO algorithm with a problem requires defining; cost function and the constraint. All simulation has been performed using a software program developed in the MATLAB environment. The simulation results show that ACO algorithm can be used to counterbalance the effect and improve the performance of any control system. Moreover, the proposed scheme overcomes the weaknesses of conventional fixed gain controller and improvement is accomplished in terms of settling time, oscillations and overshoot.

무인기(UAV)개발의 성능기반비용(PBC)분석모델 연구-무인기의 선행연구단계에서의 개발 비용구조를 중심으로-

최청호 ( Cheong-ho Choi ),방장규 ( Jang-kyu Bang ) 한국항공경영학회 2016 한국항공경영학회지 Vol.14 No.6

최근 무인기의 개발 사업이 매우 빠르게 진행되고 있다. 소형 저고도 무인기에서부터 대형 고고도 무인기에 이르는 다양한 성능과 비용을 포함한 개발 사업이 주어진다. 무인기 개발 획득을 위한 사업은 선행연구, 탐색개발 단계에서의 상세설계 및 시제기 제작, 체계개발 단계에서의 체계설계 및 양산을 통하여 요구부서에 배치하여 운영하게 된다. 여기서 사업 초기의 진행 사항으로 선행연구 단계에서 사업의 비용구조와 요구 성능을 분석 하는 업무는 개념설계 및 시제기 사업의 성공과 실패를 결정하는 중요한 요소가 된다. 선행 연구 단계에서는 무인기의 형상이 구체화 되지 않은 상태이다. 이러한 선행연구 단계는 운용자가 요구 하는 세부적인 요구 성능과 비용에 관한 사항이 개괄적으로 제시되는 단계이다. 이 단계에서 사업비용 구조와 요구 성능에 관한 합리적인 분석이 이루어 질 때 다음 단계인 개념설계와 시제기를 제작하는 업무수행을 가능하게 한다. 구체화된 시제기의 설계 및 제작은 체계설계 단계의 업무를 합리적으로 진행되도록 한다. 이러한 이유 때문에 무인기 사업의 비용구조를 선행연구 단계에서 분석하는 것은 매우 중요한 일이다. 그러나 선행연구 단계에서 요구 성능과 비용관계를 분석하는 것이 중요함에도 불구하고 접근 방법이 최근까지 제시되고 있지 않은 상태 이다. 일반적으로 설계 획득 과정에서 비용분석은 성능기반(PBC) 분석과 활동기반(ABC) 비용분석으로 구분하여 제시되고 있다. 성능기반비용분석은 선행연구단계에서와 같이 무인기의 형상과 구성품이 확정되지 않기 때문에, 단지 성능과 비용만이 관계되는 단계에서 적용 될 수 있으며, 무인기의 형상과 치구 및 구성품이 구체화되는 개념설계 단계부터는 활동기반으로 비용분석이 가능하게 된다. 본 연구는 무인기의 선행연구단계에 적용하는 성능기반비용(PBC)분석에 대하여 비용과 비용변수를 멱함수(power law) 로 표현하는 Moore의 식을 이용하고, 비용함수의 변수에 관한 멱함수의 비용변수에 대하여 미분하고 그 미분 변수의 비를 기술적 인비율(TRS) 변수로 정의하고 이를 이용하여 분석하는 Cobb-Douglas의 분석 기법 이론식을 적용하였다. 여기서 비용분석은 무인기의 기체 중량과 공허하중을 변수로 하는 2-변수 방법과, 2-변수에 체공시간을 포함한 3-변수 방법으로 수행하였다. 무인기 사업 구조의 비용과 비용변수의 관계는 2-변수 비용함수에 대하여 기체 중량, 공허하중으로 나타낼 수 있고, 3-변수 함수에 대하여 기체중량, 공허하중, 체공시간으로 나타낼 수 있다. Cobb-Douglas 식에서 정의 하는 기술적 인비율(TRS)을 변수로 하여 기체 중량과 공허하중에 대하여 2 개의 무인기 사업으로 제시되고 검토된 사업 비용구조에 대하여 사업비용 구조를 비교한 결과 사업의 비용을 최소화 하는 것은 사업의 기술적인비율의 최대화 또는 최소화에 관계없이, 사업을 이루고 있는 비용구조 자체에 의해서만 결정 될 수 있음을 보여 주었다. 비용구조를 가지고 있는 무인기의 중량에 대한 비용을 확인하기 위하여 Monte Carlo 시뮬레이션을 이용하여 비용구조에서의 단위 중량당의 비용을 통계적으로 확인 하였다. Recently unmanned air vehicle has been developed more fast. Acquisition process has more diversity from small UAV with lower altitude to bigger one with higher altitude. UAV(Unmanned Air Vehicle) acquisition consists of material solution phase, technical development with protype and capability development with initial one, and disposition to user. Where, on the material solution phase, it is important to get analysis for relation of cost structures since it to be in success or not for next phase of technical development. On the material solution phase, configuration of UAV is not in details. Moreover on this material solution phase, requirements as for key performance parameters and cost from user have been suggested roughly. On this phase, when resonable analysis of cost structures has been done, it is feasible to go into next phase as for technical development and prototype manufacturing. And also embodiment for prototype design and manufacturing can make it go through onto capability development phase timely. So it is important to have analysis of cost structures on material solution phase due to this reasons. However even though cost analysis is important, few research has been approached and reported until these days. In general design and acquisition process of UAV for cost analysis has been suggested as PBC and ABC method. PBC method can be used on material solution phase with performance and cost of UAV, but ABC mathod can be used on the capability development phase with configuration and component of UAV. This paper describes that cost analysis of UAV has been done by PBC(performance based cost) analysis method using Moore`s Law of Cost Model and modified Cobb-Douglas function and UAV unit cost estimation is done by Monte carlo simulation during the material solution phase. Unit cost of UAV have a relation to function of 2-parameters as weight and payload, as well as 3-parameters as weight, payload and endurance hours. The prepared data of UAV have been assessed using TRS(technical rate of substitution) of modified Cobb-Douglas function and it shows that is reasonable and preferable determination for pre-study and material solution phase. Estimations of unit cost of UAV are analysed by Monte Carlo simulation provide that the probability distribution of unit cost of UAV could be obtained by the random variable generation process. This cost model and cost analysis procedures for UAV could be applied to pre-developing phase.

Parameter Identification of Induction Motors using Variable-weighted Cost Function of Genetic Algorithms

A.C. Megherbi,H. Megherbi,K. Benmahamed,A.G. Aissaoui,A. Tahour 대한전기학회 2010 Journal of Electrical Engineering & Technology Vol.5 No.4

This paper presents a contribution to parameter identification of a non-linear system using a new strategy to improve the genetic algorithm (GA) method. Since cost function plays an important role in GA-based parameter identification, we propose to improve the simple version of GA, where weights of the cost function are not taken as constant values, but varying along the procedure of parameter identification. This modified version of GA is applied to the induction motor (IM) as an example of nonlinear system. The GA cost function is the weighted sum of stator current and rotor speed errors between the plant and the model of induction motor. Simulation results show that the identification method based on improved GA is feasible and gives high precision.

Parameter Identification of Induction Motors using Variable-weighted Cost Function of Genetic Algorithms

Megherbi, A.C.,Megherbi, H.,Benmahamed, K.,Aissaoui, A.G.,Tahour, A. The Korean Institute of Electrical Engineers 2010 Journal of Electrical Engineering & Technology Vol.5 No.4

This paper presents a contribution to parameter identification of a non-linear system using a new strategy to improve the genetic algorithm (GA) method. Since cost function plays an important role in GA-based parameter identification, we propose to improve the simple version of GA, where weights of the cost function are not taken as constant values, but varying along the procedure of parameter identification. This modified version of GA is applied to the induction motor (IM) as an example of nonlinear system. The GA cost function is the weighted sum of stator current and rotor speed errors between the plant and the model of induction motor. Simulation results show that the identification method based on improved GA is feasible and gives high precision.

Parameter Identification of Induction Motors using Variable-weighted Cost Function of Genetic Algorithms

Ahmed Chaouki Megherbi,Hassina Megherbi,Khier Benmahammed,Abdel Ghani Aissaoui,Ahmed Tahour 대한전기학회 2010 Journal of Electrical Engineering & Technology Vol.5 No.4

This paper presents a contribution to parameter identification of a non-linear system using a new strategy to improve the genetic algorithm (GA) method. Since cost function plays an important role in GA-based parameter identification, we propose to improve the simple version of GA, where weights of the cost function are not taken as constant values, but varying along the procedure of parameter identification. This modified version of GA is applied to the induction motor (IM) as an example of nonlinear system. The GA cost function is the weighted sum of stator current and rotor speed errors between the plant and the model of induction motor. Simulation results show that the identification method based on improved GA is feasible and gives high precision.

Various types of modelling for scale parameter in Weibull intensity function for two-dimensional warranty data

Jai Wook Baik,Jin Nam Jo 한국데이터정보과학회 2010 한국데이터정보과학회지 Vol.21 No.3

One-dimensional approach to two-dimensional warranty data involves modeling us-age as a function of time. Iskandar (1993) suggests a simple linear model for usage. However, simple linear form of intensity function is of limited value to model the situa-tion where the intensity varies over time. In this study Weibull intensity is considered where the scale parameter is expressed in terms of different models. We will find out how each parameter in the model affects the warranty cost and which model gives a bigger number of failures within the two-dimensional warranty region.

Various types of modelling for scale parameter in Weibull intensity function for two-dimensional warranty data

백재욱,조진남 한국데이터정보과학회 2010 한국데이터정보과학회지 Vol.21 No.3

One-dimensional approach to two-dimensional warranty data involves modeling us-age as a function of time. Iskandar (1993) suggests a simple linear model for usage. However, simple linear form of intensity function is of limited value to model the situa-tion where the intensity varies over time. In this study Weibull intensity is considered where the scale parameter is expressed in terms of different models. We will nd out how each parameter in the model affects the warranty cost and which model gives a bigger number of failures within the two-dimensional warranty region.

Various types of modelling for scale parameter in Weibull intensity function for two-dimensional warranty data

Baik, Jai-Wook,Jo, Jin-Nam The Korean Data and Information Science Society 2010 한국데이터정보과학회지 Vol.21 No.3

One-dimensional approach to two-dimensional warranty data involves modeling us- age as a function of time. Iskandar (1993) suggests a simple linear model for usage. However, simple linear form of intensity function is of limited value to model the situa-tion where the intensity varies over time. In this study Weibull intensity is considered where the scale parameter is expressed in terms of different models. We will nd out how each parameter in the model a ects the warranty cost and which model gives a bigger number of failures within the two-dimensional warranty region.

Estimation of parameter matrices based on measured data

Lee, Eun-Taik,Rahmatalla, Salam,Eun, Hee-Chang Elsevier 2011 Applied mathematical modelling Vol.35 No.10

<P><B>Abstract</B></P><P>Finite element structural updating based on measured data may inherent significant errors due to uncertainties in the updated physical parameter matrices. This study presents analytical equations to estimate the change in the physical parameter matrices based on the measured modal data of dynamic systems and the measured displacement data of static systems. The equations for the parameter estimation are derived by minimizing cost functions in the satisfaction of the eigenvalue equation, the mode shape orthogonality requirements for the dynamic system, and the satisfaction of the measured displacement data for the static systems. The proposed method utilizes the Moore–Penrose inverse for the inverse of the rectangular matrices without using Lagrange multipliers. Comparing the analytical results with Berman & Nagy’s method and Yang & Chen’s method, this study demonstrates that the derived equations take simpler forms and produce more accurate results. The proposed method can be widely utilized in predicting static or dynamic parameter matrices for the design and analysis of any structure.</P>