China
Africa
Explore chapters and articles related to this topic
Damage Assessment Algorithms for Structural Health Monitoring
Published in Jayantha Ananda Epaarachchi, Gayan Chanaka Kahandawa, Structural Health Monitoring Technologies and Next-Generation Smart Composite Structures, 2016
A correct selection of the GA parameters is crucial as they affect the solution and the algorithm runtime [37]. However, there is no general rule to select them; the right decision depends on the encoding, the number of genes, the objective function, and the application. The choice of the optimal parameters, such as population size and crossover and mutation probabilities, has been debated in several investigations. The codependence among the GA operators and their strong dependence with the nature of the problem are the main difficulties to define an optimum set of parameters. However, we can outline some general rules: Increasing the crossover probability increases the recombination, but it also increases the disruption of good chromosomes.Increasing the mutation probability helps to introduce new information, but transforms the search in a random search.Increasing the population size increases diversity and reduces the probability of premature convergence, but it also increases the convergence time.
Multi-objective nodes placement problem in large regions wireless networks
Published in Amir Hussain, Mirjana Ivanovic, Electronics, Communications and Networks IV, 2015
Mahmoud Gamal, Ehab Morsy, Ahmad Salah
After that, mutation operator is applied to individual solutions to get a new solution by complementing some genes randomly. Mutation is generally performed as a bit flip to maintain diversity in the population and inhibit premature convergence. In this algorithm, we perform a mutation with rate Pm=1/ (Parameters dimension), i.e., each bit has a probability of Pm to be flipped.
Introduction to Coverage Optimization in Wireless Sensor Networks
Published in Huynh Thi Thanh Binh, Nilanjan Dey, Soft Computing in Wireless Sensor Networks, 2018
Huynh Thi Thanh Binh, Nguyen Hai Nam
However, the framework of a general PSO has some weaknesses in its design so that it suffers from premature convergence (Kaveh, 2014; Nakisa, 2014). Democratic PSO (DPSO) appears to overcome this weakness, i.e., to solve the premature convergence problem and indirectly improve the quality of solution founded by PSO. Premature convergence is caused by the whole population following the best individual, leading to a possibility to get stuck in local maxima. This problem is solved in DPSO by taking bad individuals’ experiences into consideration when moving the population so that all individuals are treated equally, which widens the search space significantly.
The stochastic multi-modal hub location problem with direct link strategy and multiple capacity levels for cargo delivery systems
Published in Transportmetrica A: Transport Science, 2021
Xiaoting Shang, Kai Yang, Bin Jia, Ziyou Gao
Mutation operator assures diversity in the population and prevents premature convergence. For each chromosome in the population, it is checked if the mutation could be applied according to the mutation probability . Similarly, the mutation procedure is also randomly implemented on Part A or Part C independently. Owing to the non-repetition of genes in the first row of Part A, we take a swap mutation operator. Specifically, randomly choose two mutation positions, one is from and another is from , then swap the genes of these two positions. For Part C, one-column mutation procedure is taken, where gene 0 is replaced by 1 and vice versa, noted that the genes in the same row will be changed correspondingly. Lastly, the capacity level of hubs will be updated by Algorithm 2.
A parcel network flow approach for joint delivery networks using parcel lockers
Published in International Journal of Production Research, 2021
Shichang Pan, Lele Zhang, Russell G. Thompson, Hadi Ghaderi
For each iteration, the GA implements two crossover operations and one mutation operation in order. Firstly, it partitions the whole population into 4 groups with equal size. It then chooses one parent from group 1 and the other from group 2 and applies the first crossover operation (Co1, see Table 4). Similarly, it applies the second crossover operation (Co2, see Table 4) to the parents chosen from groups 3 and 4. The offspring produced by the crossover operators are added to the population, which increases the population size to , and their fitness values are evaluated. Then the best individuals are chosen for mutation (see Table 4), where is the mutation rate. This operation produces mutant solutions and increases the population size to . Finally, the best solutions are kept for the next generation. The mutation operation aims to increase/maintain diversity in the population and to prevent premature convergence. The GA terminates if one of the following conditions is satisfied: (i) it has bred generations; and (ii) the best fitness value has not changed (with some predefined small threshold) for a number of generations, which indicates convergence.
Minimizing transmission error with neural networks and genetic algorithm with quadratic mutation rates
Published in Mechanics Based Design of Structures and Machines, 2023
Ekansh Chaturvedi, Corina Sandu
The next step is to further add diversity and tweak the offspring generation by mutation. In a real coded GA, the mutation is achieved by adding a random number to the regularized variable. In this problem, the same was achieved by adding a random floating-point number between −0.1 and 0.1. This method is called uniform mutation rate. However, this method applied on a highly multi-modal design space often leads to premature convergence. The premature convergence is observed when the mean fitness value of every iteration converges extremely fast with no. of iterations but again begins to diverge as the simulation proceeds. The loss of diversity in population set is a major contributor to premature convergence. This observation is discussed in detail in Sec. 3.