Although the phase-shift seismic processing method has characteristics of high accuracy, good stability, high efficiency, and high-dip imaging, it is not able to adapt to strong lateral velocity variation. To overcome...Although the phase-shift seismic processing method has characteristics of high accuracy, good stability, high efficiency, and high-dip imaging, it is not able to adapt to strong lateral velocity variation. To overcome this defect, a finite-difference method in the frequency-space domain is introduced in the migration process, because it can adapt to strong lateral velocity variation and the coefficient is optimized by a hybrid genetic and simulated annealing algorithm. The two measures improve the precision of the approximation dispersion equation. Thus, the imaging effect is improved for areas of high-dip structure and strong lateral velocity variation. The migration imaging of a 2-D SEG/EAGE salt dome model proves that a better imaging effect in these areas is achieved by optimized phase-shift migration operator plus a finite-difference method based on a hybrid genetic and simulated annealing algorithm. The method proposed in this paper is better than conventional methods in imaging of areas of high-dip angle and strong lateral velocity variation.展开更多
This study provides insights into the distillation sequence optimization of refinery system in a methanol to propylene plant with extractive distillation under multiple conditions. The simulated annealing algorithm(SA...This study provides insights into the distillation sequence optimization of refinery system in a methanol to propylene plant with extractive distillation under multiple conditions. The simulated annealing algorithm(SA) with relative cost function was used to solve a meaningful optimization problem. It was observed that different conditions had differed on the flowsheet. Case study shows the effectiveness of the proposed method.展开更多
Recent studies employing deep learning to solve the traveling salesman problem(TSP)have mainly focused on learning construction heuristics.Such methods can improve TSP solutions,but still depend on additional programs...Recent studies employing deep learning to solve the traveling salesman problem(TSP)have mainly focused on learning construction heuristics.Such methods can improve TSP solutions,but still depend on additional programs.However,methods that focus on learning improvement heuristics to iteratively refine solutions remain insufficient.Traditional improvement heuristics are guided by a manually designed search strategy and may only achieve limited improvements.This paper proposes a novel framework for learning improvement heuristics,which automatically discovers better improvement policies for heuristics to iteratively solve the TSP.Our framework first designs a new architecture based on a transformer model to make the policy network parameterized,which introduces an action-dropout layer to prevent action selection from overfitting.It then proposes a deep reinforcement learning approach integrating a simulated annealing mechanism(named RL-SA)to learn the pairwise selected policy,aiming to improve the 2-opt algorithm's performance.The RL-SA leverages the whale optimization algorithm to generate initial solutions for better sampling efficiency and uses the Gaussian perturbation strategy to tackle the sparse reward problem of reinforcement learning.The experiment results show that the proposed approach is significantly superior to the state-of-the-art learning-based methods,and further reduces the gap between learning-based methods and highly optimized solvers in the benchmark datasets.Moreover,our pre-trained model M can be applied to guide the SA algorithm(named M-SA(ours)),which performs better than existing deep models in small-,medium-,and large-scale TSPLIB datasets.Additionally,the M-SA(ours)achieves excellent generalization performance in a real-world dataset on global liner shipping routes,with the optimization percentages in distance reduction ranging from3.52%to 17.99%.展开更多
A software defined networking(SDN) system has a logically centralized control plane that maintains a global network view and enables network-wide management, optimization, and innovation. Network-wide management and o...A software defined networking(SDN) system has a logically centralized control plane that maintains a global network view and enables network-wide management, optimization, and innovation. Network-wide management and optimization problems are typicallyvery complex with a huge solution space, large number of variables, and multiple objectives. Heuristic algorithms can solve theseproblems in an acceptable time but are usually limited to some particular problem circumstances. On the other hand, evolutionaryalgorithms(EAs), which are general stochastic algorithms inspired by the natural biological evolution and/or social behavior of species, can theoretically be used to solve any complex optimization problems including those found in SDNs. This paper reviewsfour types of EAs that are widely applied in current SDNs: Genetic Algorithms(GAs), Particle Swarm Optimization(PSO), Ant Colony Optimization(ACO), and Simulated Annealing(SA) by discussing their techniques, summarizing their representative applications, and highlighting their issues and future works. To the best of our knowledge, our work is the first that compares the tech-niques and categorizes the applications of these four EAs in SDNs.展开更多
基金the Open Fund(PLC201104)of State Key Laboratory of Oil and Gas Reservoir Geology and Exploitation (Chengdu University of Technology)the National Natural Science Foundation of China(No.61072073)the Key Project of Education Commission of Sichuan Province(No.10ZA072)
文摘Although the phase-shift seismic processing method has characteristics of high accuracy, good stability, high efficiency, and high-dip imaging, it is not able to adapt to strong lateral velocity variation. To overcome this defect, a finite-difference method in the frequency-space domain is introduced in the migration process, because it can adapt to strong lateral velocity variation and the coefficient is optimized by a hybrid genetic and simulated annealing algorithm. The two measures improve the precision of the approximation dispersion equation. Thus, the imaging effect is improved for areas of high-dip structure and strong lateral velocity variation. The migration imaging of a 2-D SEG/EAGE salt dome model proves that a better imaging effect in these areas is achieved by optimized phase-shift migration operator plus a finite-difference method based on a hybrid genetic and simulated annealing algorithm. The method proposed in this paper is better than conventional methods in imaging of areas of high-dip angle and strong lateral velocity variation.
文摘This study provides insights into the distillation sequence optimization of refinery system in a methanol to propylene plant with extractive distillation under multiple conditions. The simulated annealing algorithm(SA) with relative cost function was used to solve a meaningful optimization problem. It was observed that different conditions had differed on the flowsheet. Case study shows the effectiveness of the proposed method.
基金Project supported by the National Natural Science Foundation of China(Grant Nos.72101046 and 61672128)。
文摘Recent studies employing deep learning to solve the traveling salesman problem(TSP)have mainly focused on learning construction heuristics.Such methods can improve TSP solutions,but still depend on additional programs.However,methods that focus on learning improvement heuristics to iteratively refine solutions remain insufficient.Traditional improvement heuristics are guided by a manually designed search strategy and may only achieve limited improvements.This paper proposes a novel framework for learning improvement heuristics,which automatically discovers better improvement policies for heuristics to iteratively solve the TSP.Our framework first designs a new architecture based on a transformer model to make the policy network parameterized,which introduces an action-dropout layer to prevent action selection from overfitting.It then proposes a deep reinforcement learning approach integrating a simulated annealing mechanism(named RL-SA)to learn the pairwise selected policy,aiming to improve the 2-opt algorithm's performance.The RL-SA leverages the whale optimization algorithm to generate initial solutions for better sampling efficiency and uses the Gaussian perturbation strategy to tackle the sparse reward problem of reinforcement learning.The experiment results show that the proposed approach is significantly superior to the state-of-the-art learning-based methods,and further reduces the gap between learning-based methods and highly optimized solvers in the benchmark datasets.Moreover,our pre-trained model M can be applied to guide the SA algorithm(named M-SA(ours)),which performs better than existing deep models in small-,medium-,and large-scale TSPLIB datasets.Additionally,the M-SA(ours)achieves excellent generalization performance in a real-world dataset on global liner shipping routes,with the optimization percentages in distance reduction ranging from3.52%to 17.99%.
文摘A software defined networking(SDN) system has a logically centralized control plane that maintains a global network view and enables network-wide management, optimization, and innovation. Network-wide management and optimization problems are typicallyvery complex with a huge solution space, large number of variables, and multiple objectives. Heuristic algorithms can solve theseproblems in an acceptable time but are usually limited to some particular problem circumstances. On the other hand, evolutionaryalgorithms(EAs), which are general stochastic algorithms inspired by the natural biological evolution and/or social behavior of species, can theoretically be used to solve any complex optimization problems including those found in SDNs. This paper reviewsfour types of EAs that are widely applied in current SDNs: Genetic Algorithms(GAs), Particle Swarm Optimization(PSO), Ant Colony Optimization(ACO), and Simulated Annealing(SA) by discussing their techniques, summarizing their representative applications, and highlighting their issues and future works. To the best of our knowledge, our work is the first that compares the tech-niques and categorizes the applications of these four EAs in SDNs.
