شناسایی مدل های rotorcraft هوایی برای کسب و کارهای کوچک بدون سرنشین بر ​​اساس الگوریتم مورچگان تطبیقی

This paper proposes a model identification method to get high performance dynamic model of a small unmanned aerial rotorcraft. With the analysis of flight characteristics, a linear dynamic model is constructed by the small perturbation theory. Using the micro guidance navigation and control module, the system can record the control signals of servos, the state information of attitude and velocity information in sequence. After the data preprocessing, an adaptive ant colony algorithm is proposed to get optimal parameters of the dynamic model. With the adaptive adjustment of the pheromone in the selection process, the proposed model identification method can escape from local minima traps and get the optimal solution quickly. Performance analysis and experiments are conducted to validate the effectiveness of the identified dynamic model. Compared with real flight data, the identified model generated by the proposed method has a better performance than the model generated by the adaptive genetic algorithm. Based on the identified dynamic model, the small unmanned aerial rotorcraft can generate suitable control parameters to realize stable hovering, turning, and straight flight.

With the ability of taking off and landing vertically,as well as hovering, Small Unmanned Aerial Rotorcraft (SUAR) plays an irreplaceable role in civil applications, including road traffic monitoring, city building surveillance, etc.[1–5]. As a complex Multi-In put Multi-Output (MIMO) system, a SUAR has the characteristics of nonlinear, multivariable and complex coupled[6–8]. Therefore, a precise dynamic model is the basis for high performance attitude and position control[9]. Since parameter identi- fication method is relatively simple, it has become a common way for the model construction of a SUAR system [10]. With the experimental input-output data, parameter identification method can produce a mathematical representation of the system dynamics quickly.Many researchers have used time-domain system iden- tification and frequency-domain identification methods to get optimal parameters of the dynamic model[11].Based on straight steady flight data, Nino et al.[12] used frequency-domain identification method to construct lateral and longitudinal Single Input Single Output (SISO) models for micro air vehicle. With different fre- quencies data, Wu et al . [13] constructed two autoregressive models to represent the attitude characteristics of ahomemade 1-m sized aircraf[13]. Frequency-domain identification method needs a lot of flight data regarding different frequencies[14] . However, pilots could not ac- curately finish the whole frequency range control due to the constraint of visual delay. Although the autopilot can replace human pilot to generate different frequency control commands, there exist certain flight dangers for SUAR system in data collection process. Time-domain system identification method is often used in the dynamic model identification process. Park et al. [15]used the Least-Squares Estimation (LSE) to develop a new autoregressive model for helicopters [With the adaptive genetic algorithm, Lei et al.[16]proposed a linear dynamic model for a small unmanned aerial vehicle to realize stable hovering motion. Salman et al.[17] used a nonlinear mapping method to get anon-linear state space model for a MUAV system. Since there are so many variables in the dynamic model of SUAR system, the selection of optimal parameters of the dynamic model is computationally expensive. With the positive feedback strategy, Ant Colony (AC) has already shown high searching perform- ance in large ranges. Hence, AC was used to get optimalparameters for the dynamic model of SUAR system. For the above purposes, the objective of this paper is to propose a model identification method for SUAR system. With the Adaptive Ant Colony (AAC) algorithm, the dynamic model of SUAR system can be identified with flight data.The remainder of the paper is organized as follows: in Section 2, the dynamic model of SUAR is analyzed.In Section3, an AAC method is developed to realizeprecise parameter identification for the proposed dynamic model based on the test data. A series of flight tests for hovering and straight flight validate the effectiveness of the dynamic model in section4, followed bythe conclusion in section5

This paper proposes a new system identification method to construct high performance dynamic model of a SUAR. Based on the AAC, system can get optimal parameters of the dynamic model from preprocessing flight data. Compared with dynamic model generated by the AGA and LS, the identified model generated by the AAC has a better estimation performance for the real system. With the identified model, system can generate suitable control parameters to realize stable hovering and straight flight under wind disturbance. The mean trajectory following error is less than 1.2 m. Due to the wind disturbance in the flight data collection process, there exists a certain estimation error of the dynamic model. Therefore, the future work is tostudy the robust control method to improve flight performance.