DESIGN OF UNMANNED SHIP HEADING CONTROLLER BASED ON FCMAC-PID

Transcription

1 DESIGN OF UNMANNED SHIP HEADING CONTROLLER BASED ON FCMAC-PID Xinyu Hu*, Chao Li, Yuheng Yu, Xiaolong Wang, Daode Zhang School of Mechanical Engineering, Hubei University of Technology, Wuhan, China ABSTRACT: To solve the problems of inaccuracy of path tracking and untimely response of trajectory tracking for small unmanned ship, a FCMAC-PID course controller with fast response, good robustness and strong anti-jamming ability is designed. This method combines the advantages of Fuzzy-PID control and CMAC neural network, and overcomes the shortcomings of the Fuzzy-PID control which has steady-state error and weak anti-interference ability. On the basis of analyzing the kinematics equation of ship manufacturing, the model of ship hull is established according to the second-order wild model. Simulink simulation is carried out under the condition of no disturbance and comprehensive disturbance. The control effects of Fuzzy-PID and FCMAC-PID controllers are compared. The results show that FCMAC-PID path tracking controller has stronger self-adaptability, fast response speed, strong anti-interference ability and better tracking performance. KEYWORDS: Unmanned Ship manufacturing; Course Control; Simulink Simulation; Fuzzy-PID; FCMAC-PID 1 INTRODUCTION In recent years, the emergence of unmanned ships manufacturing has attracted wide attention of researchers from all over the world (Zheng Ruicong, 2017), (Song YJ, 2011). As a new type of transport platform, unmanned ships have been widely used in military and civilian fields (Fan Jiadong, 2017). Compared with the conventional unmanned platform, the mini unmanned ship has obvious advantages, such as low cost, flexible maneuverability, easy carrying and easy to carry out experiments. Its control system uses modular structure design, easy to quickly integrate; carrying different sensors can complete different tasks. The control technology of unmanned ship has become the focus and hot spot of scholars in various countries. In the control research of unmanned ship track tracking, the traditional PID control is mostly used, but the parameters of the traditional PID control algorithm are generally set to a constant value (Zhang Jingzhou 2009), which means that the control effect is very dependent on a stable environment; while the ship in the course of operation, there are currents, wind and other effects, easy to cause the PID control system. The robustness is not enough to achieve the desired control effect. With the development of intelligent control in theory and technology, fuzzy control, neural network algorithm and expert control are introduced into the classical PID control system (Lavalle,1998). The adaptive fuzzy PID control proposed in reference is a combination of fuzzy control and traditional PID control (Duan Mengxia, 2016), which meets the requirements of nonlinear and imprecise mathematical models of the controlled object, but has strong dependence on rule base and parameter setting experience. Reference presents a PID control algorithm based on cerebellar model network (CMAC) (Liu Weiwei, 2017), (Zhao Zu, 2017). Although the search performance has been improved, the structure is more complex and 53

2 difficult to implement. Cerebellar Model Network (CMAC) is a kind of learning algorithm with tutors(chang Ma,2017). It has fast convergence speed and no local minimum problem. It is suitable for real-time control. Fuzzy control does not require precise control model and has strong adaptability. Combining the advantages of CMAC and fuzzy PID control system, this paper designs a fuzzy PID controller based on CMAC, that is FCMAC-PID controller, and compares the control effects of Fuzzy-PID and FCMAC-PID control systems, so as to obtain better control effect and improve the adaptive ability of the controller to uncertain disturbance(zhang Rongjun,2000).At the same time, this method is applied to the autonomous cruise ship control system under the national water resources efficient utilization plan. of the surface unmanned craft is obtained through MEMS inertial navigation system and navigation positioning system. Finally, the deviation angle of the course is calculated. It is 0. This course deviation is the input of the course control system. The system outputs the corresponding PWM signal through the controller and converts it to the rudder angle output through the rudder. When the 0, the output of the rudder is 0, indicating that the surface unmanned ship is sailing on the desired course. 2 COMPOSITION AND WORKING PRINCIPLE OF UNMANNED SHIPCONTROL SYSTEM 2.1. Control system composition The UAV control system consists of four modules: information acquisition module, autonomous obstacle avoidance module, speed control module and course control module, as shown in Figure 1. The design of the controller in the course control module is the focus of this paper. Location information of obstacles Heading speed information Current speed Current heading Giving heading 0 Course control module Speed v Rudder angle a Fig. 1 structure diagram of control system for unmanned ship 2.2. working principle of heading controller The principle of course control for surface unmanned aerial vehicle is illustrated as follows: Firstly, adesired course 0 is set for surface Fig. 2 heading control schematic diagram 3. ESTABLISHMENT OF MOTION MODEL FOR UNMANNED SHIP For a ship in horizontal motion, only three directions of lateral, longitudinal and undulating motion can be considered, and the 6-DOF unmanned ship can be simplified to a 3-DOF plane motion [15]. A mathematical model is established as shown in Figure 3. In order to determine the nonlinear relationship between X, Y, Z and various motion parameters (J. Ghommem 2007), (Du Jialu, 2006). assuming that the constant-speed linear motion of an unmanned ship is considered as a balanced state, Abkowitz proposes a small perturbation and Taylor expansion method to study the expressions of X, Y and Z. The state-space model is transformed into a transfer function, i.e. the stern rudder of an unmanned ship under the condition of constant-speed direct navigation. The transfers function to the change of direction of the course: unmanned aerial vehicle. Then, the current heading 54

3 The ship has large inertia, so long as it pays attention to the dynamic characteristics of the low frequency band, the formula (3) is simplified, from the third order to the second order, so that, when x 0, there is and neglect the second and third order, we can get the second order form of Nomoto model, that is, the second order form of Nomoto model(shuren Yang,2011). Fig. 3 mathematical model of unmanned ship According to the momentum theorem along the center of mass and the moment of momentum theorem around the center of mass, the basic equations of ship plane motion are obtained as follows: The mass of the UAV is represented by m, the abscissa of the mass center of the UAV is represented by X, the moment of inertia of the Formula: denotes the merits and demerits of the ship's maneuverability, which is called the maneuverability index of unmanned ship, denotes the merits and demerits of the ship's maneuverability, which is called the maneuverability index of unmanned ship. Based on the experience of repeated tests on experimental ships, the second order Notomo model of unmanned ships is obtained by taking =0.926 and =0.075 as follows: UAV to the oz axis is represented by I z,and the external forces acting on the three axes are represented by X,Y and Z respectively. Assuming that the origin of the hull coordinate system is in the center of mass, then X =0 then the formula (1) will be simplified to: In order to determine the nonlinear relationship between X,Y, Z and various motion parameters, assuming that the constant-speed linear motion of an unmanned ship is considered as a balanced state, Abkowitz proposes a small perturbation and Taylor expansion method to study the expressions of X, Y and Z. The state-space model is transformed into a transfer function, i.e. the stern rudder of an unmanned ship under the condition of constantspeed direct navigation. The transfers function to the change of direction of the course: 4. FUZZY PID HEADING CONTROLLER DESIGN Because of its simple structure and good stability, traditional PID control mostly adopts PID control technology in actual navigation, but it has the disadvantages of poor adaptability and low control precision (Chen, Sui, 2012). Fuzzy control does not need precise control object; it can be directly summed up and optimized according to people's knowledge and experience, and then get the ideal control output. It has the characteristics of good anti-interference and strong adaptability. So fuzzy control and PID control are combined to form fuzzy PID control (Yuan Li, 2010). It can not only improve the response speed and adaptability of the system, but also reduce the overshoot of the system Fuzzy control principle 55 Fuzzy PID controller is based on the current deviation and deviation rate of change, the use of

4 three PID parameters and error E and error rate of change between the fuzzy reasoning, real-time K, K, K on-line adjustment, and fuzzy PID p i d controller structure as shown in the figure (Liu 2012). Fig. 5 fuzzy membership table and output surface graph Fig. 4 Schematic diagram of fuzzy PID control The control input is heading deviation and the rate of change of heading deviation, and the increment of three parameters of PID controller, Ki, Kd is taken as the control output. The K p basic field of the variables is this: The weights of e, ec, K p, Ki, Kd are K, K, K K,which should be determined e ec 1, K2, 3 according to the actual project and system debugging. The fuzzy subsets of fuzzy input and output variables are NB (negative large), NM (negative middle), NS (negative small), ZO (zero), PS (positive small), PM (positive middle), PB (positive large). The membership function of the system and the input-output surface view of fuzzy inference are shown in Figure Simulation model establishment and simulation results analysis of 3.2 Fuzzy- PID course control According to the control model and rudder model deduced from the 2 section, the simulation model of heading controller can be established. In the simulation experiment (Shuren Yang, 2011), the initial course of the unmanned ship is set to be 45 degrees, i.e. sailing along the Northeast direction, and the course controller model is designed with two algorithms: fuzzy adaptive tuning PID and classical PID. Secondly, the simulation disturbance of wind, wave and current is added to the simulation environment of the unmanned ship, and the wind and wave disturbance is equivalent to rudder angle disturbance. A short time square wave interference model is added to the control object, as shown in the interference model shown in Figure 7. The simulation models without disturbance and without disturbance are shown in Figure 6 and Figure 7 respectively, and the simulation results of fuzzy PID heading controller and PID heading controller are compared. 56

5 Fig. 6 Simulink simulation diagram of a fuzzy PID heading control system based on no interference. Fig. 7 Simulink simulation diagram of fuzzy PID heading control system based on wind and wave interference Fig. 8 angle chart of course without interference Fig. 9 helm angle diagram without interference 57

6 Fig.10 heading angle curveafter adding interference Fig.11 curve of rudder after adding interference Table 1: performance index of control system without interference Controller Adjustment time(s) Classic PID Fuzzy-PID Overshoot (%) Table 2: performance index of control system after adding interference Controller Adjustment time(s) Classic PID Fuzzy-PID Overshoot (%) As can be seen from Tables 1 and 2, the overshoot of classical PID is 40%, the system regulation time is 22.5s, the overshoot of Fuzzy- PID is 27%, and the system regulation time is 11.62s. Compared with the traditional PID controller, the fuzzy adaptive tuning PID controller can restrain overshoot and improve the response speed. In Fig. 10, the adjusting time of Fuzzy-PID and PID is 13 s and 15 s respectively after adding the disturbance of wind and wave in s, and it can be seen from Fig. 11 that the rudder angle of Fuzzy-PID varies greatly when the disturbance occurs. When the disturbance of fuzzy controller changes greatly, only some rules of fuzzy control are used, which results in slow response time and poor optimization performance. 5. DESIGN OF FCMAC-PID COURSE CONTROLLER CMAC neural network can learn, compare and analyze the input and output values of the system. It not only has the function of "self-learning", but also has the characteristics of short learning time and high real-time. Therefore, in view of the shortcomings of fuzzy PID controller and the advantages of CMAC neural network, a fuzzy CMAC-PID heading (i.e. FCMAC-PID) is designed. Controller, this control method to determine the deviation and deviation rate of change as input, through fuzzy reasoning on-line tuning PID parameters, through CMAC neural network parallel control PID output, to achieve realtime PID control adjustment. 5.1.Principle of CMAC control algorithm CMAC adopts tutorial learning algorithm. Each control cycle is updated. The output U n (k) is calculated by comparing with the total input of the system, updating the weight value, and then entering the learning process. Finally, the difference between the total input and the output of CMAC is minimized. After CMAC learning, the input of the master controller is all output by the CMAC controller. The control algorithm of the system is as follows: In the formula, ai is the binary selection vector; c is the generalization parameter of CMAC network; U (n) is the corresponding output of CMAC; n U (n) is the output of conventional p controller. The adjustment indicators of CMAC are: In the formula. is the learning rate. is momentum factor, ( 0, 1) ( 0, 1) 58

7 5.2. FCMAC-PID control principle FCMAC-PID control principle is to control CMAC neural network and fuzzy PID control in parallel. At the end of each control cycle, the output U 0 of CMAC is compared with the total input U of the control object, and the weight is modified to enter the learning stage, so that the difference between them is minimized. Finally, the total output of the system is controlled by CMAC instead of the output of fuzzy PID control.. The schematic diagram of the FCMAC-PID controller is shown in Figure 12. Fig. 14 response curve when adding interference Fig. 12 FCMAC-PID control schematic diagram 5.3. Simulation model establishment and simulation results analysis of FCMAC- PID course controller The traditional PID, fuzzy PID and FCMAC- PID control strategies are used to simulate the unmanned ship control model and steering gear model derived in Section 2, as shown in Figure 13. Firstly, the initial heading of the unmanned ship is set at 45 degrees, i.e. along the Northeast direction, and the same interference signal is given at 15s, and the simulation results are compared. Fig. 13 simulation model of FCMAC-PID system 59 Fig. 15 curve of rudder angle with interference added Controller Table 3 response index of controller Pre regulation time(s) Oversh oot(%) PID Fuzzy-PID FCMAC-PID Adjustment time after jamming(s) As shown in Fig. 14 and Table 3, the minimum overshoot of FCMAC-PID is 4.3%, the adjustment time is 3.2s, the overshoot of Fuzzy-PID is 27% and the adjustment time is 11.6s before the interference; as shown in Fig. 15, the adjustment time of FCMAC-PID is 4S after the interference is added in the 15-17s period, and the change of rudder angle is small, while the adjustment time of Fuzzy-PID is less. For 13s, the rudder angle varies greatly. Therefore, the control effect of FCMAC-PID controller designed in this paper is obviously better than that of fuzzy PID controller and classical PID controller in overshoot and adjustment time, and the optimization effect is obvious. 6. CONCLUSION In this paper, FCMAC-PID control method adjusts K, K and p i Kd on-line through fuzzy reasoning in the control process, and CMAC realizes parallel adaptive adjustment with fuzzy PID controller, so that the controller has better control

8 effect. The simulation results show that this method can give full play to the advantages of CMAC neural network and fuzzy PID control. It can achieve stability in a shorter time and has less overshoot. At the same time, it has strong adaptive ability, can better adapt to the impact of external disturbances on the system in the control process, strong anti-interference ability, and has a good prospect of Engineering application. 7. ACKNOWLEDGEMENT This work is supported by Wuhan science and technology support program No REFERENCE Chang MA. Simulation Studies of CMAC-PID Combined Control for Electro hydraulic Position Servo System[A]. Advanced Science and Industry Research Center. Proceedings of nd International Conference on Computer, Mechatronics and Electronic Engineering (CMEE 2017) [C]. Advanced Science and Industry Research Center, 2017:5. Chen, Sui., Liu, C., Huang, Z., et al.: AUV global path planning based on sparse A* algorithm. Torpedo Technol. 20(4), (2012) Du Jialu, Guo Chen, Yang Cheng en (2006). Adaptive nonlinear design of autopilot for ship course tracking. Journal of Applied Sciences, 24(1), DuanMengxia. Research on steering control system of unmanned ship based on fuzzy PID control [D]. Hainan University, Fan Jiadong, Ye Anqi. Solar energy wind powered unmanned ship manufacturing. Jiangsu water conservancy, 2017 (12): Ho HF, Wong YK, Rad AB (2009). Adaptive fuzzy sliding mode control with chattering elimination for nonlinear SISO systems. Simulation Modeling Practice and Theory, 17, J. Ghommem, F. Mnif, G. Poisson, "Nonlinear Formation Control of a Group of Underactuated Ships," in OCEANS Europe, pp.1-8, Lavalle, S.M.: Rapidly-exploring random trees: a new tool for path planning. Algorithmic Comput. Rob. New Dir., (1998) Liu Weiwei, Song Honggang, Liu Huifang. Design and Simulation of PID Controller Based on FCA-CMAC Fuzzy Neural Network. Mechanical Engineer, 2017 (10): 4-7. Liu, J.: Advanced PID Control and MATLAB Simulation, 2nd edn. Electronic Industry Press, Beijing (2012). pp ShurenYang ;Hongbo Wang(2011) ; Study of efficient ship heading controller International Conference on Electrical and Control Engineering:16-18 Song YJ, Woo JH, Shin JG (2011) Research on systematization and advancement of shipbuilding production management for flexible and agile response for high value offshore platform. International Journal of Naval Architecture and Ocean Engineering 3(3): Yuan Li, Wu HS (2010). Terminal sliding mode control fuzzy control based on multiple sliding surface for nonlinear ship autopilot system. Journal of Marine and Application, 9(4), Zhang Jingzhou, Yang Weijing, Zhang Anxiang.Research and Application Simulation of Fuzzy Adaptive PID Control.Computer Simulation, 2009, 26 (9). ZHANG Rong-jun, CHEN Yao-bin, SUN Zengqi et al., "Path control of a surface ship in restricted waters using sliding mode", IEEE Transactions on Control Systems Technology, vol. 8, no. 4, pp , Zhao Zu,Zhi P ZH,Chuan Y J. Simulation and experimental research of digital valve control servo system based on CMAC-PID control method[j]. High Technology Letters,2017,23(03): 306. Zheng Ruicong, QiuXiangyao. Open the door of fishing intelligent ship manufacturing. Guangdong shipbuilding, 2017, 36 (04):