Optimal Sensor Placement Based On Monkey Algorithm

Optimal Sensor Placement (OSP) is one of the most important steps in Structural Health Monitoring (SHM) system. As modern architectural structures have become more and more complicated, the traditional OSP methods cannot meet the application requirement. In this paper, a new intelligent algorithm, i.e., Monkey Algorithm (MA) is proposed to study OSP. Based on different structure features, this article comes up with different kinds of OSP methods and is mainly divided into seven chapters, introducing as follows:In the first chapter, SHM system is introduced as well as the importance of OSP. The article reviews related research results in China and abroad and establishes the content of the study according to the correspondent problem; the article puts forward three-dimensional finite element model of three different structures that studied in the article. At the same time, the article results also include the optimal model based on OSP in the paper.In the second chapter, an OSP method based on a new intelligent algorithm, i.e. monkey algorithm (MA) is proposed. Considering the characteristics of the OSP, the. location of the monkeys is given by integer encoding, which overcomes the problem that the MA can only solve the optimization of variables with continuity. The diversity of the monkeys is increased by introducing the hamming distance in the initial location, in order to improve the capacity of global search, and the random disturbance mechanism of the Harmony search is introduced in the process of climbing to improve the capacity of local search. Finally,taking the Dalian World Trade Building as an example, the parametric sensitivity is analyzed and the OSP schemes are chosen. The results show that the improved MA can better solve the problem of OSP, which has obvious advantages compared with the classical sequential OSP algorithm.In the third chapter, the distributed monkey algorithm (DMA) is established for the OSP of multi-degree-of-freedom large structures. Firstly, the problem that the original MA can only solve the continuous variables is overcome by introducing the dual-structure coding method. Then, a new method is proposed that assigning the large number of individual monkeys generated by the initialization to multi-group of monkeys in a specific way and performing the parallel search. Considering that the original MA is able to step out local optimum and the harmony search algorithm has better local search capability, the two-step search algorithm is advanced based on the basic harmony search algorithm by collecting the preliminary optimum gotten by each group of monkeys as the original harmony memory to obtain the final sensor placement scheme. At last, the parametric sensitivity analysis and the selection of sensor placement schemes are performed on the Dalian International Trade Mansion. The results show that the DMA has better global search capability and it’s applicable to the sensor placement of multiple-degree-of-freedom large structures.In the fouth chapter, a novel asynchronous climbing monkey algorithm (ACMA) for OSP is presented to solve the problems of sightless search and low efficiency in the key climb process of the MA. The search pattern is improved by the information of global optimal solution and previous best solution during the search process of monkey population. Meanwhile, the asynchronous variable learning factor is introduced into the search pattern to maintain the balance of global and local search by adjusting the effect of monkeys’own and social experiences during the climb process, which greatly improve the search efficiency of the algorithm. The results show that the ACMA is efficient and effective for sensor placement problem compared to the monkey algorithm.In the fifth chapter an adaptive monkey algorithm (AMA) used for OSP is proposed to solve the problem that the search methods of the climb process and watch-jump process are mechanic and the pattern of the somersault process is single. Then, the climb process and watch-jump process are updated in order to adaptively select the two search methods, which can improve the local search ability and efficiency of the algorithm. In addition, the two new somersault processes, i.e., reflection somersault process and variation somersault process, are introduced to strengthen the global search ability of the algorithm. The results show that the AMA can better work out the OSP and the search efficiency has been greatly improved compared to the original algorithm.In the sixth chapter, the immune monkey algorithm (IMA) for OSP is proposed by introducing the immune mechanism of biosphere. The chaotic search is adopted to initialize the monkey’s location for ensuring the uniform distribution of monkey, which can improve the global search capability. Besides, the deep climb is introduced in the climb process to enhance the local search capability of the algorithm. Alter the end of the climb process, the first selection by density-dependent mechanism on the monkey is performed and the immune clone is done on the monkeys with best location to guarantee the diversity of monkey. While after the end of the watch process, the second selection is added based on the fitness and the immune vaccination is carried out on the monkeys with poor location to improve the convergence of the algorithm. Finally, the parametric sensitivity analysis and OSP is done on the Dalian world trade building. The results show that the search efficiency of the IMA greatly increases compared with the original MA, which can better solve the OSP problem.