Explore chapters and articles related to this topic
Design of Actuator Servo Controller
Published in Abdullah Al Mamun, GuoXiao Guo, Chao Bi, Hard Disk Drive, 2017
Abdullah Al Mamun, GuoXiao Guo, Chao Bi
Among those non-gradient based optimization methods, Random Neighborhood Search (RNS) ([54]), Genetic Algorithm (GA) ([147], [62]), neural networks ([169]) and some other random optimization methods incorporated with statistical techniques ([147], [199]) are employed for their well known robustness property to the error of objective function and ability of global search. Simplex method ([204]) and Sequential Quadratic Programming (SQP) ([162]) have been used to find the optimal controller within a convex sub-region of performance surface.
A Simulation Based Meta-heuristic Approach for the Inbound Container Housekeeping Problem in the Automated Container Terminals
Published in Maritime Policy & Management, 2023
Hu Qin, Xinxin Su, Guoxin Li, Xin Jin, Mingzhu Yu
The simulation-based optimization (SO) is mainly applied to the case where the objective function, constraints or model parameters cannot be expressed by explicit functional relationships or values as in traditional optimization problems. The evaluation of the optimization object can only be achieved by the statistical indicators obtained by the simulation. For complex random optimization problems, simulation-based optimization is an ideal choice. In recent years, simulation-based optimization methods are applied in complex engineering systems, supply chain and logistics systems, manufacturing systems and socio-economic systems at home and abroad (Clarke, McLay, and McLeskey Jr. 2014; Dragović, Škurić, and Kofjač 2014; Ferrara et al. 2014; Nguyen, Reiter, and Rigo 2014; Sahay and Ierapetritou 2014; Wang et al. 2017, 2017; Yu et al. 2017; Cao and Lam 2019). In theory, Gosavi (2015) introduced the simulation-based optimization algorithm in detail from the aspects of origin, goal, limitation, simulation basis, applicable problem and specific algorithm. This book details the functions of computer simulation in simulation-based optimization methods.
Fabric defect detection algorithm using RDPSO-based optimal Gabor filter
Published in The Journal of The Textile Institute, 2019
Yueyang Li, Haichi Luo, Miaomiao Yu, Gaoming Jiang, Honglian Cong
It is obvious that the optimal Gabor filter approach has some outstanding advantages over Gabor filter bank method. However, the selection of the optimized Gabor filter parameters is very crucial to the performance of the detection approach. In recent years, population-based random optimization algorithms, such as genetic algorithm (GA) (Goldberg, 1989), differential evolution (DE) (Storn & Price, 1997), and particle swarm optimization (PSO) (Kennedy & Eberhart, 1995), have been wildly applied for global optimization. Based on a random drift model, a new version of PSO, which is called the random drift particle swarm optimization (RDPSO) is presented by Sun, Wu, Palade, Fang, and Shi (2015). Inspired by the free electron model in metal conductors placed in an external electric field, the RDPSO employs a novel set of evolution equations that can enhance the global search ability of the algorithm. Since it was introduced by Sun, it has been successfully applied in diverse application areas to handle complicated optimization problems (Sun, Palade, Cai, Fang, & Wu, 2014a; Sun, Palade, Wu, & Fang, 2014b; Sun, Palade, Wu, Fang, & Wang, 2014c; Yuan, Sun, & Zhou, 2016).
An automatic COVID-19 diagnosis from chest X-ray images using a deep trigonometric convolutional neural network
Published in The Imaging Science Journal, 2023
Regardless of algorithmic differences, all population-based random optimization methods have two phases: exploration and exploitation [48]. In the exploration stage, a randomized algorithm mixes stochastic replies at a high rate in order to discover possible areas inside the search space [49]. At the exploitation step, slight modifications are made to random replies, and outputs are computed [50]. Trigonometric functions have been widely applied to various complex problems [51–54]. Eqs. (1) and (2) illustrate the procedure for calculating these outputs after modifying random solutions [55]. In which is the location of the current response in the i-th dimension and t-th iteration. Also, are random numbers, are the destination's location in the i-th dimension and represents absolute value. Eqs. (1) and (2) can be combined to generate Eq. (3) [56]. In which is a random number in a range of . As shown in Eq. (3), there are four main parameters in the algorithm. The parameter shows the next location area (or direction of motion) between the source and destination (or outside of it). The parameter defines the amount of movement toward the goal or in the opposite direction. The parameter determines the size of the random weight to reach the destination (which may have a value as or) [55]. Eventually, changes equally between the components of the sinus and cosine, as shown in Eq. (1). However, it should be noted that Eqs. (1) and (2) can be extended to higher dimensions. If the parameter in Eq. (3) is defined as a random number in the range , then the existing mechanism guarantees to explore the search area. A suitable method should balance exploratory operations and identify potential search areas before eventually converging to a general optimum. The domain of the trigonometric functions in Eqs. (1) to (3) varies according to Eq. (4) in order to create a compromise among the exploitation and exploration stages. Where t is the current step, T is the maximum number of steps, and a is also a fixed number. Figure 1 shows the reduction in the range of the sinus and cosine functions during iterations’ execution.