Study On Active Control Of Structural Responses For Helicopter In Frequency Domain

As an innovative and high technology, active control technique of structural responses has caught worldwide attention. With the background of the active vibration control of cockpit in a helicopter and using helicopter global model, a series of topics are studied based on the control approaches and procedures of this technique in frequency domain.Firstly, convergence, robustness and controllability of the control algorithm as well as the cautious controller are theoretically deducted and analyzed. Sequentially several explicit formulations are obtained and their validations are verified by simulations. Meanwhile, H∞Optimality of two online parameter identification methods (Least Mean Square and Kalman filter) is discussed. Secondly, using true flight measured data and measured frequency responses of control path of a typical helicopter, the whole processes between frequency and time domains are simulated after taking into consideration of the problems that will likely be encountered. Results of two parameter identification methods are compared. Thirdly, as the core part of the digital signal processing (DSP) control software, programs of the studied method in frequency domain are written. Fourthly, on the basis of simulation, single input/single output and multi input/multi output experimental studies are conducted on a free-free beam and a scale dynamics model of a helicopter airframe, respectively, using self- developed DSP control software which based on embedded SEED-DEC33 controller, and the effects of weighting matrix and parameter settings in identification methods etc. on control performance are emphatically explored. Finally, aiming at the different targets such as reducing computation amount, effectively smoothing control output, decreasing sensitivity to initial values of identification variables etc., three new modified methods are proposed according to traditional control technique. The first one is Dual LMS method, by which frequency response matrix of control path and the amplitude of external excitation response are identified, the second one is normalization of Dual LMS method, and the last is the one that the identification variables are identified by Kalman filtering after primary identification of LMS method. Theoretical deductions and simulations of all these modified methods are performed. All beneficial conclusions from above simulation analysis, experiment study and method modification can provide reference and basis for application of real projects.