This work is concerned with designing two types of controllers, a PID and a Fuzzy PID, to be used
for flying and stabilizing a quadcopter. The designed controllers have been tuned, tested, and
compared using two performance indices which are the Integral Square Error (ISE) and the Integral
Absolute Error (IAE), and also some response characteristics like the rise time, overshoot, settling
time, and the steady state error. To try and test the controllers, a quadcopter mathematical model has
been developed. The model concentrated on the rotational dynamics of the quadcopter, i.e. the roll,
pitch, and yaw variables. The work has been simulated with “MATLAB”. To make testing the
simulated model and the controllers more realistic, the testing signals have been applied by a user
through a joystick interfaced to the computer. The results obtained indicated a general superiority in
performance for the Fuzzy PID controller over the PID controller used in this work. This conclusion
is based by the following figures: lesser ISA for the roll, pitch, and yaw
consequently, lesser IAE for the roll, pitch, and yaw consequently,
lesser rise time and settling time for the roll and pitch consequently, and
lesser settling time for the yaw. Moreover, the FPID gave zero overshoot versus , ,
and in the PID case for the roll, pitch, and yaw consequently. Both controllers gave zero steady
state error with close rise times for the yaw. This superiority of the FPID controller is gained as the
fuzzy part of it continuously and online adapts the parameters of the PID part.
(3)
(4)
(3)
(4)
(24)
(23)
A three-stage learning algorithm for deep multilayer perceptron (DMLP) with effective weight initialisation based on sparse auto-encoder is proposed in this paper, which aims to overcome difficulties in training deep neural networks with limited training data in high-dimensional feature space. At the first stage, unsupervised learning is adopted using sparse auto-encoder to obtain the initial weights of the feature extraction layers of the DMLP. At the second stage, error back-propagation is used to train the DMLP by fixing the weights obtained at the first stage for its feature extraction layers. At the third stage, all the weights of the DMLP obtained at the second stage are refined by error back-propagation. Network structures an
... Show More
(1)
In this work, a novel technique to obtain an accurate solutions to nonlinear form by multi-step combination with Laplace-variational approach (MSLVIM) is introduced. Compared with the traditional approach for variational it overcome all difficulties and enable to provide us more an accurate solutions with extended of the convergence region as well as covering to larger intervals which providing us a continuous representation of approximate analytic solution and it give more better information of the solution over the whole time interval. This technique is more easier for obtaining the general Lagrange multiplier with reduces the time and calculations. It converges rapidly to exact formula with simply computable terms wit
... Show More
(1)
This work is concerned with the vibration attenuation of a smart beam interacting with fluid using proportional-derivative PD control and adaptive approximation compensator AAC. The role of the AAC is to improve the PD performance by compensating for unmodelled dynamics using the concept of function approximation technique FAT. The key idea is to represent the unknown parameters using the weighting coefficient and basis function matrices/vectors. The weighting coefficient vector is updated using Lyapunov theory. This controller is applied to a flexible beam provided with surface bonded piezo-patches while the vibrating beam system is submerged in a fluid. Two main effects are considered: 1) axial stretching of the vibrating beam that leads
... Show More
(3)
(4)
Generalized Additive Model has been considered as a multivariate smoother that appeared recently in Nonparametric Regression Analysis. Thus, this research is devoted to study the mixed situation, i.e. for the phenomena that changes its behaviour from linear (with known functional form) represented in parametric part, to nonlinear (with unknown functional form: here, smoothing spline) represented in nonparametric part of the model. Furthermore, we propose robust semiparametric GAM estimator, which compared with two other existed techniques.
The Dagum Regression Model, introduced to address limitations in traditional econometric models, provides enhanced flexibility for analyzing data characterized by heavy tails and asymmetry, which is common in income and wealth distributions. This paper develops and applies the Dagum model, demonstrating its advantages over other distributions such as the Log-Normal and Gamma distributions. The model's parameters are estimated using Maximum Likelihood Estimation (MLE) and the Method of Moments (MoM). A simulation study evaluates both methods' performance across various sample sizes, showing that MoM tends to offer more robust and precise estimates, particularly in small samples. These findings provide valuable insights into the ana
... Show More