Research On Motion Modeling And Maneuvering Control Technology Of A Hovering Submarine

Submarine is one of the most important platforms for carrying the weapons, which play an irreplaceable important strategic role in modern warfare. Maneuvering performance is one of the most important performances of submarines. In the implementation of some special missions, a submarine needs to control depth and pitch attitude in parking state, which is known as the submarine hovering control. Good hovering performance can reduce the noises of a submarine, and reduce the electricity consumption, which has an important strategic significance. However, research of the automatic control for submarine hovering is still in its initial stage, so there are a lot of basic researches need to be done. In this paper, based on a in-depth study of the development status of the motion modeling and maneuvering control strategy, a systematic study of submarine hovering movement mechanism and identification modeling technology and hovering maneuvering control strategy is carried on. Major research jobs in the paper are as followed:First of all, the underwater motion modeling reference coordinates system and the transforming law of the coordinates was introduced, and the motion equation of a submarine moving in the vertical plane was derved; the mechanism of the submarine hovering tanks was analized to establish a mathematical model for the buoyancy regulating tanks in the submarine hovering control system. Finally, the movement kinematics of a hovering submarine was analized to establish the hovering motion model of a submarine, and the model was validated.In order to get better control performance and building the simulation environment, the models of the interferences was derived, including the initial imbalances, the hull volume compression, the density changes of the seawater, the surface wave interferences and the ocean currents. The design of environment simulation platform for the maneuvering control of a submarine was described on the base of the modeling of the environmental interferences.Taking into account the shortcomings of the mechanism modeling approaches, a submarine hovering motion identification modeling technique based on an extended random decrement technique was proposed. First, the limitation of the classic random decrement technique was researched, and an extended random decrement technique with more liberal application conditions was proposed, and the extended random decrement equation of a hovering submarine was derived. Based on the extended random decrement technique, a hybrid network system identification technique based on weighted fitting linear regression algorithm and multilayer perceptron neural networks was designed, which is named Weighted Fitting Linear Regression Neural Network algorithm (WFLRNN). The input samples of the identification system was obtained with the extended random decrement technique without measuring the real-time input datas of the recognized system, the identification of the submarine hovering motion was divided into the identification of the damping parameters, recovery parameters and the identification of the coupling parameters, these two parts of identification jobs were finished by the weighted fitting linear regression algorithm and a MLP neural network independently. Experimental results show that both in the case of broadband or narrowband excitation, WFLRNN method can provide identificaed submarine hovering models with errors no more than5%.Submarine hover is essentially a weak motor, slow time-varying dynamic process, which provides sufficient basis for the application of the linear control strategies which are based on the system models. On the base of the decoupling of the system, a fuzzy PID submarine hovering controller whose subsystems in the two dimensions of the hovering motion can make self-tuning by identifing the PID parameters online was designed. For taking the real-time requirements in practical applications into account, a fast TS fuzzy modeling technology called FTFM was proposed. After all these preparing jobs, a decoupling fuzzy PID control system was designed based the FTFM for the hovering control of a submarine. Experimental results show that the designed decoupling fuzzy PID controller showed good properties not only in control accuracy, timeliness, but also in the size of fuzzy control rules.Considering complexity of the modeling of the motion and the interferences during the process, take the control methods which are not rely on the system modeling into consideration to manipulate the submarine hovering controller. A fuzzy logic control system with a neural network-based presentated structure whose structure and parameters are adaptive was designed, which was called Fuzzy Adaptive Neuro-Networks Control system (FANC). The FANC system was connected with a5-layer feedforward neural network to implement the fuzzy variables mapping from input to output. The system is a collection of FLS knowledge representation and reasoning and ANN knowledge acquisition, learning and adaptability. Meanwhile a mixed learning algorithm for the learning of FANC system was proposed, which is divided into a self-organized stage and a supervised learning stage. Finally a FANC system with the corresponding training algorithm for the submarine hovering controlling was designed. Experiments show that the FANC system applied to submarine hovering control can achieve good control performance.