量子粒子群演算法在三維無人飛機任務巡航規劃

Abstract

大多數關於最佳路徑規劃論文文獻中，多關注於從啟航點飛行至不同於啟航點的另一點之研究。在一般的情況下，路徑規劃的要求，是要能避免被雷達探測到，以及環繞飛行其需要執行的偵察區域。除此之外，無人飛機（UAV）在飛航過程中，包括最大爬升率，飛行高度，以及最小迴轉半徑等約束的適航性，都是必須在最佳路徑規劃過程中加以限制的。在本研究中，無人飛機在真實環境的最佳路徑規劃中，限制啟航點與終點兩者是相同的。現在，使用基因演算法（GA），差分進化演算法（DE），粒子群演算法（PSO）和量子粒子群演算法（QPSO），這四種啟發式演算法，來解決無人飛機最佳路徑規劃問題。 在二維最佳路徑規劃問題中，通過研究各種不同的成本函數、限制條件，以及巡航點的設置。在最佳化的過程中，我們對於每一個參數，識別其最穩定的執行數值。並從上述的四個演算法結果中，選取其最好的兩個，用來模擬三維最佳路徑規劃的問題。最後，在數值結果中，表明不論在二維，或是在三維的最佳化路徑規劃問題中，QPSO演算法都優於其它三種演算法。

Most of the studies about optimal path planning problems concern with the path from the point of departure to another point of arrival. In general, the requirements of planning the path are to be able to avoid the radar detection and to encircle the reconnaissance area. Moreover, during the navigation, airworthiness of the unmanned aerial vehicle (UAV) including the constraint of maximum climb rate, flying altitude and least radius of gyration, etc., must be considered as constrains in optimization process as well. In this study, UAV in this optimal path planning departed and arrived at the same point in real environment. Now, the following four heuristic algorithms, Genetic Algorithms (GA), Differential Evolution (DE), Particle Swarm Optimization (PSO), and Quantum-Behaved Particle Swarm Optimization (QPSO) are methods to solve the UAV path planning problems. By studying various treatments for the cost functions, constrains and waypoints on 2D path planning problem, we identify the most robust implementation for each element in the optimization. Two of the best algorithms among the above mentioned four heuristic methods are employed to simulate the 3D path planning problem. Our numerical results show that the QPSO algorithm outperforms the other methods in both 2D and 3D path planning problem.