The production optimization in the closed-loop reservoir management is generally empirical,and challenged by the issues such as low precision,low efficiency,and difficulty in solving constrained optimization problems....The production optimization in the closed-loop reservoir management is generally empirical,and challenged by the issues such as low precision,low efficiency,and difficulty in solving constrained optimization problems.This paper outlines the main principles,advantages and disadvantages of commonly used production optimization methods/models,and then proposes an intelligent integrated production optimization method for waterflooding reservoirs that considers efficiency and precision,real-time and long-term effects,and the interaction and synergy between a variety of optimization models.This method integrates multiple optimization methods/models,such as reservoir performance analysis,reduced-physics models,and reservoir numerical models,with these model results and insights organically coupled to facilitate model construction and matching.This proposed method is elucidated and verified by field examples.The findings indicate that the optimal production optimization model varies depending on the specific application scenario.Reduced-physics models are conducive to short-term real-time optimization,whereas the simulator-based surrogate optimization and streamline-based simulation optimization methods are more suitable for long-term optimization strategy formulation,both of which need to be implemented under reasonable constraints from the perspective of reservoir engineering in order to be of practical value.展开更多
A surrogate based particle swarm optimization （SBPSO） algorithm which combines the surrogate modeling technique and particle swarm optimization is applied to the reliability- based robust design （RBRD） of composit...A surrogate based particle swarm optimization （SBPSO） algorithm which combines the surrogate modeling technique and particle swarm optimization is applied to the reliability- based robust design （RBRD） of composite pressure vessels. The algorithm and efficiency of SBPSO are displayed through numerical examples. A model for filament-wound composite pressure vessels with metallic liner is then studied by netting analysis and its responses are analyzed by using Finite element method （performed by software ANSYS）. An optimization problem for maximizing the performance factor is formulated by choosing the winding orientation of the helical plies in the cylindrical portion, the thickness of metal liner and the drop off region size as the design variables. Strength constraints for composite layers and the metal liner are constructed by using Tsai-Wu failure criterion and Mises failure criterion respectively. Numerical examples show that the method proposed can effectively solve the RBRD problem, and the optimal results of the proposed model can satisfy certain reliability requirement and have the robustness to the fluctuation of design variables.展开更多
For high-purity distillation processes,it is difficult to achieve a good direct product quality control using traditional proportional-integral-differential(PID)control or multivariable predictive control technique du...For high-purity distillation processes,it is difficult to achieve a good direct product quality control using traditional proportional-integral-differential(PID)control or multivariable predictive control technique due to some difficulties,such as long response time,many un-measurable disturbances,and the reliability and precision issues of product quality soft-sensors.In this paper,based on the first principle analysis and dynamic simulation of a distillation process,a new predictive control scheme is proposed by using the split ratio of distillate flow rate to that of bottoms as an essential controlled variable.Correspondingly,a new strategy with integrated control and on-line optimization is developed,which consists of model predictive control of the split ratio,surrogate model based on radial basis function neural network for optimization,and modified differential evolution optimization algorithm. With the strategy,the process achieves its steady state quickly,so more profit can be obtained.The proposed strategy has been successfully applied to a gas separation plant for more than three years,which shows that the strategy is feasible and effective.展开更多
For the safety protection of passengers when train crashes occur, special structures are crucially needed as a kind of indispensable energy absorbing device. With the help of the structures, crash kinetic-energy can b...For the safety protection of passengers when train crashes occur, special structures are crucially needed as a kind of indispensable energy absorbing device. With the help of the structures, crash kinetic-energy can be completely absorbed or dissipated for the aim of safety. Two composite structures(circumscribed circle structure and inscribed circle structure) were constructed. In addition, comparison and optimization of the crashworthy characteristic of the two structures were carried out based on the method of explicit finite element analysis(FEA) and Kriging surrogate model. According to the result of Kriging surrogate model, conclusions can be safely drawn that the specific energy absorption(SEA) and ratio of specific energy absorption to initial peak force(REAF) of circumscribed circle structure are lager than those of inscribed circle structure under the same design parameters. In other words, circumscribed circle structure has better performances with higher energy-absorbing ability and lower initial peak force. Besides, error analysis was adopted and the result of which indicates that the Kriging surrogate model has high nonlinear fitting precision. What is more, the SEA and REAF optimum values of the two structures have been obtained through analysis, and the crushing results have been illustrated when the two structures reach optimum SEA and REAF.展开更多
The ever-increasing demand for training and inferring with larger machine-learning models requires more efficient hardware solutions due to limitations such as power dissipation and scalability.Optics is a promising c...The ever-increasing demand for training and inferring with larger machine-learning models requires more efficient hardware solutions due to limitations such as power dissipation and scalability.Optics is a promising contender for providing lower power computation,since light propagation through a nonabsorbing medium is a lossless operation.However,to carry out useful and efficient computations with light,generating and controlling nonlinearity optically is a necessity that is still elusive.Multimode fibers(MMFs)have been shown that they can provide nonlinear effects with microwatts of average power while maintaining parallelism and low loss.We propose an optical neural network architecture that performs nonlinear optical computation by controlling the propagation of ultrashort pulses in MMF by wavefront shaping.With a surrogate model,optimal sets of parameters are found to program this optical computer for different tasks with minimal utilization of an electronic computer.We show a remarkable decrease of 97%in the number of model parameters,which leads to an overall 99%digital operation reduction compared to an equivalently performing digital neural network.We further demonstrate that a fully optical implementation can also be performed with competitive accuracies.展开更多
基金Supported by the Major Scientific and Technological Special Project of CNPC(2023ZZ04)。
文摘The production optimization in the closed-loop reservoir management is generally empirical,and challenged by the issues such as low precision,low efficiency,and difficulty in solving constrained optimization problems.This paper outlines the main principles,advantages and disadvantages of commonly used production optimization methods/models,and then proposes an intelligent integrated production optimization method for waterflooding reservoirs that considers efficiency and precision,real-time and long-term effects,and the interaction and synergy between a variety of optimization models.This method integrates multiple optimization methods/models,such as reservoir performance analysis,reduced-physics models,and reservoir numerical models,with these model results and insights organically coupled to facilitate model construction and matching.This proposed method is elucidated and verified by field examples.The findings indicate that the optimal production optimization model varies depending on the specific application scenario.Reduced-physics models are conducive to short-term real-time optimization,whereas the simulator-based surrogate optimization and streamline-based simulation optimization methods are more suitable for long-term optimization strategy formulation,both of which need to be implemented under reasonable constraints from the perspective of reservoir engineering in order to be of practical value.
基金supported by the Natural Science Foundation of China(No.10772070)National Basic Research Program of China(No.2011CB013800)
文摘A surrogate based particle swarm optimization （SBPSO） algorithm which combines the surrogate modeling technique and particle swarm optimization is applied to the reliability- based robust design （RBRD） of composite pressure vessels. The algorithm and efficiency of SBPSO are displayed through numerical examples. A model for filament-wound composite pressure vessels with metallic liner is then studied by netting analysis and its responses are analyzed by using Finite element method （performed by software ANSYS）. An optimization problem for maximizing the performance factor is formulated by choosing the winding orientation of the helical plies in the cylindrical portion, the thickness of metal liner and the drop off region size as the design variables. Strength constraints for composite layers and the metal liner are constructed by using Tsai-Wu failure criterion and Mises failure criterion respectively. Numerical examples show that the method proposed can effectively solve the RBRD problem, and the optimal results of the proposed model can satisfy certain reliability requirement and have the robustness to the fluctuation of design variables.
基金Supported by the National High Technology Research and Development Program of China（2007AA04Z193） the National Natural Science Foundation of China（60974008 60704032）
文摘For high-purity distillation processes,it is difficult to achieve a good direct product quality control using traditional proportional-integral-differential(PID)control or multivariable predictive control technique due to some difficulties,such as long response time,many un-measurable disturbances,and the reliability and precision issues of product quality soft-sensors.In this paper,based on the first principle analysis and dynamic simulation of a distillation process,a new predictive control scheme is proposed by using the split ratio of distillate flow rate to that of bottoms as an essential controlled variable.Correspondingly,a new strategy with integrated control and on-line optimization is developed,which consists of model predictive control of the split ratio,surrogate model based on radial basis function neural network for optimization,and modified differential evolution optimization algorithm. With the strategy,the process achieves its steady state quickly,so more profit can be obtained.The proposed strategy has been successfully applied to a gas separation plant for more than three years,which shows that the strategy is feasible and effective.
基金Projects(51405516,U1334208)supported by the National Natural Science Foundation of ChinaProject(2013GK2001)supported by the Science and Technology Program for Hunan Provincial Science and Technology Department,ChinaProject(2013zzts040)supported by the Graduate Degree Thesis Innovation Foundation of Central South University,China
文摘For the safety protection of passengers when train crashes occur, special structures are crucially needed as a kind of indispensable energy absorbing device. With the help of the structures, crash kinetic-energy can be completely absorbed or dissipated for the aim of safety. Two composite structures(circumscribed circle structure and inscribed circle structure) were constructed. In addition, comparison and optimization of the crashworthy characteristic of the two structures were carried out based on the method of explicit finite element analysis(FEA) and Kriging surrogate model. According to the result of Kriging surrogate model, conclusions can be safely drawn that the specific energy absorption(SEA) and ratio of specific energy absorption to initial peak force(REAF) of circumscribed circle structure are lager than those of inscribed circle structure under the same design parameters. In other words, circumscribed circle structure has better performances with higher energy-absorbing ability and lower initial peak force. Besides, error analysis was adopted and the result of which indicates that the Kriging surrogate model has high nonlinear fitting precision. What is more, the SEA and REAF optimum values of the two structures have been obtained through analysis, and the crushing results have been illustrated when the two structures reach optimum SEA and REAF.
文摘The ever-increasing demand for training and inferring with larger machine-learning models requires more efficient hardware solutions due to limitations such as power dissipation and scalability.Optics is a promising contender for providing lower power computation,since light propagation through a nonabsorbing medium is a lossless operation.However,to carry out useful and efficient computations with light,generating and controlling nonlinearity optically is a necessity that is still elusive.Multimode fibers(MMFs)have been shown that they can provide nonlinear effects with microwatts of average power while maintaining parallelism and low loss.We propose an optical neural network architecture that performs nonlinear optical computation by controlling the propagation of ultrashort pulses in MMF by wavefront shaping.With a surrogate model,optimal sets of parameters are found to program this optical computer for different tasks with minimal utilization of an electronic computer.We show a remarkable decrease of 97%in the number of model parameters,which leads to an overall 99%digital operation reduction compared to an equivalently performing digital neural network.We further demonstrate that a fully optical implementation can also be performed with competitive accuracies.
