摘 要
隨著船舶電力推進技術的快速發(fā)展和廣泛應用，交流電力推進已經取代直流電力推進成為主流。同時，推進系統相關設備需求日益增大，但國產設備的裝船率并不高，制約著我國船舶行業(yè)的發(fā)展。研究船舶電力推進關鍵技術有助于增強我國船舶行業(yè)的綜合競爭力。因此，本課題對船舶電力推進系統直接轉矩控制（DTC）相關技術進行研究。
本文以傳統DTC為理論基礎，分析DTC的原理和結構，在MATLAB/Simulink中對系統各個組成部分仿真建模，分析電機的速度、磁鏈和轉矩特性。針對傳統DTC系統磁鏈和轉矩脈動較大，本文首先采用空間矢量脈寬調制技術與DTC系統相結合來改善磁鏈和轉矩的脈動，仿真表明SVM-DTC系統的性能得到全面的提高。其次，定子磁鏈的估測精度直接影響到系統的性能，而傳統U-I磁鏈觀測模型精度低，本文采用新型基于PI鎖相原理的定子磁鏈觀測器代替?zhèn)鹘y的U-I觀測器，磁鏈觀測的精度和系統的性能得到改善。
無速度傳感器技術在船舶電力推進中有著廣闊的前景，介紹了幾種常用的速度估計方法后，本文重點分析了模型參考自適應系統（MRAS）速度估計方法和原理，并應用到SVM-DTC系統中，仿真表明MRAS速度估計的精度和動態(tài)性有待提高。由于MRAS系統中由Popov超穩(wěn)理論推導出的自適應率為PI控制器，PI參數對速度估計精度影響較大，因此，本文采用對角遞歸神經網絡（DRNN）算法與MRAS相結合，動態(tài)地整定MRAS系統的PI參數，DRNN-MRAS速度估計的精度和動態(tài)性顯著提高。
本文針對實際船舶，建立船舶-螺旋槳仿真模型，分析船舶的速度、推力和扭矩特性。并分別將上述的傳統DTC、SVM-DTC和無速度DTC與實際船舶-螺旋槳模型結合，建立船舶電力推進DTC綜合仿真系統。對比分析當施加螺旋槳負載時DTC系統電機的響應特性，以及施加螺旋槳負載擾動時對DTC系統的影響。經過驗證，在船舶電力推進系統中，SVM-DTC具有較好的穩(wěn)態(tài)和動態(tài)性能，傳統DTC動態(tài)性能優(yōu)于穩(wěn)態(tài)，而無速度DTC穩(wěn)態(tài)性能優(yōu)于其動態(tài)性能。

關鍵字：船舶電力推進；直接轉矩控制；空間矢量脈寬調制；鎖相環(huán)；模型參考自適應系統；對角遞歸神經網絡；船槳模型











































Abstract
With the rapid development and widespread application of the ship electric propulsion technology, AC electric propulsion has replaced the DC electric propulsion and becomes the mainstream. What’s more, demands for the associated equipment of propulsion system are growing increasingly. However, the shipment rate of the domestic equipment is very low, which restricts the development of the shipping industry of our country. So it is helpful to research key technology of the marine electric propulsion in order to enhance the competitive ability of our country shipping industry. The DTC technology of marine electric propulsion is researched in this thesis.
This thesis analyses the principle and structure of direct torque control based on the traditional direct torque control, and establishes simulation models of each component of the system in the Simulink platform in MATLAB, then analyses the curves in which AC motor runs in constant speed state and constant torque state. For the traditional direct torque control possess great flux and torque, this thesis firstly adopts SVM combined with direct torque control system to improve the magnetic chain and torque, and the simulation results show that SVM-DTC system performance is overall improved. Secondly, estimation accuracy of stator flux directly affects the performance of the system, tradition U-I flux observer has low accuracy, this thesis adopt the improved stator flux observer based on the principle of PI phase lock loop to take the place of the traditional U-I observer, as a result, the flux observer accuracy and the performance of the system are improved greatly.
Since there is a broad prospect to apply without speed sensor technology in ship electric propulsion, this thesis analyzes the speed estimation method which is based on model reference adaptive (MRAS) after introducing some commonly used speed estimation, then which is applied in the SVM-DTC system. The simulation result showed that the precision and dynamics of MRAS speed estimation is awaited with improvement. Because in the MRAS system the adaptive rate deduced by Popov stability theory is PI controller, PI parameters have a big effect on the speed estimation precision, therefore this thesis combines diagonal recurrent neural network (DRNN) algorithm with MRAS to dynamically set PI parameters, the precision and dynamics of DRNN-MRAS speed estimation improved greatly.
In the end this thesis builds the ship-motor-propeller simulation model, analyzing the speed, thrust and torque characteristics of ships, and applies the traditional DTC, SVM-DTC and speed sensorless DTC to ship electric propulsion system respectively, establishes ship electric propulsion DTC comprehensive simulation system. What’s more, it is comparatively analyzed on the motor characteristics of the DTC system when the propellers are loaded on and the effect on the DTC system when the propellers are put load disturbance. According to verification, in the ship electric propulsion system SVM-DTC has good steady state and dynamics performance, dynamics performance of traditio..