Review on meta-heuristics approaches for airside operation research

K.K.H. Ng, C.K.M. Lee, Felix T.S. Chan, Yaqiong Lv
2018 Applied Soft Computing  
The number of publications related to airside operations research is increasing and gaining in popularity. This paper aims to provide researchers with a comprehensive and extensive overview of meta-heuristics application for aviation research, with a particular focus on the airside operations. The scope of airside operations research covers airspace and air traffic flow management, aircraft operation in the terminal manoeuvring area and surface traffic operation. Based on the recent
more » ... related to airside operations, the meta-heuristics approach is a promising approach to enhance the computational efficiency and achieve higher applicable in various decisions in airside operations. However, the literature on airside operations research is quite disjointed and disparate. Therefore, a general taxonomy framework for the airside information system is proposed in order to classify the research systematically and expedites related research and development of engineering applications in the aviation industry. To the best of our knowledge, this is the first review of the field using the meta-heuristics approach. The prominent findings of recent publication and the directions of future research are addressed throughout the review and analysis of the relevant studies. This is the Pre-Published Version. © 2018. This manuscript version is made available under the CC-BY-NC-ND 4.0 license order, complex and stochastic combinatorial problem, or even real-life CO applications [10] . Heuristics is regarded as basic approximate algorithms for providing near optimal results [9, 11] . However, the design of heuristics is problem-specific and problem-dependent methods. Meta-heuristics approach is a high-level problem-independent framework, which provides a trajectory of searching for close-to-optimal solutions from practical problems within satisfactory computation time [9, 12] . The design of a meta-heuristic algorithm includes two major concepts, which is exploitation and exploration. Exploitation refers to the ability of foraging around a promising candidate solution to reach the optimal solution, while exploration indicates the ability of terminating searching under the condition of local optimal trapping [13] . The selection of proper meta-heuristics was related to the complexities of exploitation and exploration on the CO problem. In general, meta-heuristics can be categorised as the single solution-based methods and population-based methods [14] . The single agent-based methods also called trajectory methods, by constructing a searching process regarding an individual solution. The trajectory methods include, but are not limited to Tabu Search [12], Greedy Search [15] and Iterative Local Search [16] algorithms. The existing population-based algorithms fall into three major categories: Evolutionary Algorithm (EA), physics-based algorithms and Swarm Intelligence. The typical examples of EAs involve the Genetic Algorithm (GA) [17], Memetic Algorithm [18] and Differential Evolution algorithm [19] . EAs deal with information exchange procedures among several candidates by continuously improving the solution quality by iterations, which are known as a blind search method that seldom exploits the domain knowledge and uses evolutionary operators iteratively from known solutions [20] . The simple mechanism of evolutionary operators can be effective in exploitation phase, but the balance of exploration and exploitation is often ignored in the design of the algorithms. By contrast, many naturally inspired meta-heuristic algorithms, including physics-based and SI-based algorithms, have gained increasing popularity because of their high efficiency, which involves specific controlling parameters to maintain the balance between exploitation and exploration. The physics-based meta-heuristics approach is a kind of discipline that aims to simulate the laws of natural science and knowledge of nature in algorithm design. The search agents perform searching according to the natural interaction between matter and energy. Despite the fact that the control parameters usually contain complex functions that lead to long computation time, certain physics-based algorithms are promising in achieving optimal or near-optimal solutions [21] . Representative examples are Big-Bang and Big-Crunch algorithm [22], Gravitational Search Algorithm [23], Ray Optimisation algorithm [24]. Swarm intelligence (SI) is a new type of bio-inspired meta-heuristics that emphasises distributing individual agents for solving hard CO problems. The philosophy of SI, which incorporates the collective behaviour of natural species, is a fascinating meta-heuristics research area in the contemporary evolutionary computation. Although SI for optimisation is still in the proof-ofconcept stage in industrial engineering, current publications recommend that SI is qualified to obtain good-quality solutions than single-based and evolutionary methods given a reasonable CPU time. Compared with physics-based algorithms, SI-based algorithms highlight the simple collective behaviour of individual agents rather than complex controlling mechanisms. During the era of SI, different SI-based algorithms have been introduced to CO applications such as the Artificial Bee Colony (ABC) algorithm [25], Ant Colony Optimisation (ACO) algorithm [26], Bat algorithm [27] and Particle Swarm Optimisation (PSO) [28]. Contribution of the research A large amount of meta-heuristics with different features and intrinsic characteristics have been proposed throughout the last four decades and found to be a promising technique for real-life application. The availability of periodic review and assessment becomes more important to guide the readers in understanding the meta-heuristics research progress and highlight the research potential in the domain of the airside operations system with a meta-heuristics approach. The comparison of the meta-heuristics techniques in airside operations research is crucial so as to explore the future research direction. Hence, this paper attempts to identify the concealed research field of meta-heuristics research in airport operation. Organisation of the paper The organisation of this paper is summarised as follows. After the background of the airside operations research and meta-heuristics in Section 1, section 2 presents the review framework and the selection criteria of the relevant articles. Section 3 summaries the classification and description of the meta-heuristics (See Section 3.1) and the research findings of operations research in airside activities (See Section 3.2). The statistical analysis of the studies are reported in Section 4. The trend analysis and the research potential of the field are illustrated in Section 5. Finally, the concluding remarks are raised in Section 6. Research methodology The primary objective of this paper is to present a taxonomic framework for outlining and consolidating the current research field of extant airside operations in the literature with reference to functionality, which indicates potential topics for future research and development in the aviation industry. The literature review approach necessarily contributes to the research progress to discover potential research and study in airport OR, which is a valid tool to synthesise and consolidate scattered knowledge systematically [29] . In Fig. 1 , the review process of this review article follows the four major steps proposed by Mayring for conducting content analysis [30]. (iii) Research methodology with non-meta-heuristics but using meta-heuristics as benchmark for comparison According to the above inclusion and exclusion criteria, 103 studies were extracted for the formulation of the taxonomy framework in airside activities using meta-heuristic approaches and the analysis of the trends in the research domain. Problem classification Classification scheme of meta-heuristics The solution methods for airside operations research can be categorised into two major groups, which are exact approach and approximate approach. Although the exact approaches are the frequent approaches to optimise CO models, it lacks the capability to handle practical cases within a reasonable time frame. The approximate approach, especially meta-heuristics, has become more favoured for searching for a good solution with a reasonable computation time. The approximate approach can be further divided into heuristics and meta-heuristics methods. The concept of the meta-heuristics was introduced by [12], aiming to encounter the problem of local optimum through the controlling mechanism during the recursive search method. The fundamental controlling mechanism in the meta-heuristics consists of trajectory method, control and memories, hybrid strategies, parallelism, and
doi:10.1016/j.asoc.2018.02.013 fatcat:fvpq3pblwjbghdtdxi5a6sh73e