Adaptive neural network observer for proton-exchange membrane fuel cell system

This paper develops an adaptive neural network(NN)observer for proton-exchange membrane fuel cells(PEMFCs).Indeed,information on the oxygen excess ratio(OER)value is crucial to ensure optimal management of the durability and reliability of the PEMFC.The OER indicator is computed from the mass of oxygen and nitrogen inside the PEMFC cathode.Unfortunately,the measurement process of both these masses is difficult and costly.To solve this problem,the design of a PEMFC state observer is attractive.However,the behaviour of the fuel cell system is highly non-linear and its modelling is complex.Due to this constraint,a multilayer perceptron neural network(MLPNN)-based observer is proposed in this paper to estimate the oxygen and nitrogen masses.One notable advantage of the suggested MLPNN observer is that it does not require a database to train the NN.Indeed,the weights of the NN are updated in real time using the output error.In addition,the observer parameters,namely the learning rate and the damping factor,are online adapted using the optimization tools of extremum seeking.Moreover,the proposed observer stability analysis is performed using the Lyapunov theory.The observer performances are validated by simulation under MATLAB®/Simulink®.The supremacy of the proposed adaptive MLPNN observer is highlighted by comparison with a fixed-parameter MLPNN observer and a classical high-gain observer(HGO).The mean rela-tive error value of the excess oxygen rate is considered the performance index,which is equal to 1.01%for an adaptive MLPNN and 3.95%and 9.95%for a fixed MLPNN and HGO,respectively.Finally,a robustness test of the proposed observer with respect to measurement noise is performed.

Adaptive Dynamic Programming-Based Attitude Optimal Tracking Control for a Quadrotor with Unmeasured Velocities and Model Uncertainties

This paper proposes an optimal output feedback tracking control scheme of the quadrotor unmanned aerial vehicle(UAV)attitude system with unmeasured angular velocities and model uncertainties.First,neural network(NN)is used to approximate the model uncertainties.Then,an NN velocity observer is established to estimate the unmeasured angular velocities.Further,a quadrotor output feedback attitude optimal tracking controller is designed,which consists of an adaptive controller designed by backstepping method and an optimal compensation term designed by adaptive dynamic programming.All signals in the closed-loop system are proved to be bounded.Finally,numerical simulation example shows that the quadrotor attitude tracking scheme is effective and feasible.

Robust Load Frequency Control for Isolated Microgrids Based on Double-loop Compensation

Frequency is an important indicator for the oper-ation of microgrids.However,the randomness and uncertainty of renewable energy and load variability may lead to frequency undulation.So,a robust load frequency control(LFC)is pro-posed for isolated wind-diesel microgrids considering time delay and parameter uncertainty.The control strategy can suppress frequency fuctuation and optimize frequency dynamic response.First,the double compensation loop,including feedforward control and integral sliding mode control(SMC),is devised to provide anti-disturbance compensation for the diesel generator system and ameliorate the frequency stability of independent microgrids.Secondly,a dynamic fuzzy controller,composed of wind speed and load demand,is designed to provide real-time response reference power for doubly fed induction generator systems(DFIGs),which can promote the effective participation of a wind turbine system for frequency regulation.Then,the proportional differential(PD)parameters of a dynamic fuzzy controller and the frequency adjustment compensation of DFIGs can be obtained by using a particle swarm optimization(PSO)algorithm.Thirdly,load demand is an important index of the robust dynamic load frequency control method;the radial basis function(RBF)neural network observer(NNO)based on the LFC model is presented to obtain more accurate load deviations and improve the control precision of LFC.The performance of the proposed LFC method is tested under different operation cases.Index Terms-Load frequency control,microgrid,neural network observer,sliding mode,time delay and parameter uncertainty.