Quadrotor Unmanned Helicopter ( QUH) is a kind of under － actuated system with multiple inputs and strong coupling． It
has a broad application prospect，such as aerial photography，archaeology，patrolling border，anti － terrorism investigation． Dynamical
model can be established according to Euler＇s theorem and Newton＇s law，considering the effect of air resistance and rotating torque． A
dual － loop control system is designed based PID algorithm，and then another system is designed based on fuzzy self － tuning PID controller．
Simulation results of two controllers reveal that the QUH can arrive at the specified position and keep hovering and control effects
of fuzzy self － tuning PID controller are better than that of classic PID in the aspect of response time and stability．

The attitude control is the key of the automatic flight control for quadrotor UAV． In order to improve the robustness of quadrotor
UAV for various uncertainties，a robust adaptive backstepping controller is designed． The whole attitude kinematical model is obtained
and translated into a MIMO nonlinear system with generalized uncertainty． Because the design feature of the system is strict feedback，
the backstepping controller is designed． A robust adaptive function is designed to counteract the influence of the uncertainties，
which include external disturbance and interior parameters perturbation． The nonlinear tracking differentiator is introduced to estimate
the differential signal of the virtual control law and to reduce the“computer explosion”problem which is ubiquitous in the backtepping
controller． The closed － loop system is proved to be stable and converge exponentially through constructing Lyapunov function． Simulation
results are presented to corroborate the effectiveness and robustness of the proposed control strategy．

For the problems that Hough transform can not correct the license plate without frame border and Radon transform is computation-intensive and time-consuming, a fast algorithm combined Hough transform and Radon transform for license plate tilt correction was proposed. By setting effective detection areas and making full use of Krisch to get the horizontal edge, this method can correct both the rim license plates and the rimless ones fast and effectively.  Selected 200 artificial slanted pictures to verify the accuracy of angle inclination got from the algorithm, and selected 400 natural slanted pictures to compare the speed and accuracy of the three algorithms, namely Hough, Radon and the method in this paper. Performed in Matlab platform with the angle ranged in [-20°,20°], the average execution time of the proposed algorithm is 0.38s and its accuracy achieves more than 90%. Experimental results show the algorithm is fast, accurate and robust.

null

BP algorithm usually has slow convergence speed and is easy to fall into local minimum value． On the basis of the analysis
and study for the domestic market of air conditioning order． This article proposes a method of the model parameters of grey neural network
optimized by genetic algorithm． By utilizing the property of gray model that the randomness of data can be reduced and the strong
nonlinearity of neural network，the method establishes a non-linear prediction model for air conditioning demand，and the genetic algorithm
is used to optimize it so as to improve the accuracy of forecasting and speed up the degree of convergence． Simulation results indicate
that the algorithm can better solve the problem of air conditioner order forecasting and can be widely apply to similar prediction．

The traditional mutation strategy of differential evolution algorithm can not reach a good balance between the global search
and the local search and the operators are constant． The differential evolution algorithm leads to premature convergence and the low
search efficiency． Based on analysis of the performance of the optimization strategies，a hybrid mutation strategy is proposed in this paper．
The scheme attempts to balance the exploration and exploitation abilities． In this way，emphasis is laid on the global search at the
beginning，which results in maintaining the diversity of population． Later，contribution from the local search increases in order to converge
to the optimal faster． Meanwhile，the random normal scaling factor F and the time － varying crossover probability factor CR are
used synchronously to improve the performance of DE． Finally，the modified differential evolution algorithm is tested on benchmark
functions． The simulation results show that the modified algorithm can effectively avoid the premature convergence，as well as modified
the global convergence ability and the search efficiency remarkably．

null