Simultaneous perturbation stochastic approximation （SPSA） belongs to the class of gradient-free optimization methods that extract gradient information from successive objective function evaluation. This paper descri...Simultaneous perturbation stochastic approximation （SPSA） belongs to the class of gradient-free optimization methods that extract gradient information from successive objective function evaluation. This paper describes an improved SPSA algorithm, which entails fuzzy adaptive gain sequences, gradient smoothing, and a step rejection procedure to enhance convergence and stability. The proposed fuzzy adaptive simultaneous perturbation approximation （FASPA） algorithm is particularly well suited to problems involving a large number of parameters such as those encountered in nonlinear system identification using neural networks （NNs）. Accordingly, a multilayer perceptron （MLP） network with popular training algorithms was used to predicate the system response. We found that an MLP trained by FASPSA had the desired accuracy that was comparable to results obtained by traditional system identification algorithms. Simulation results for typical nonlinear systems demonstrate that the proposed NN architecture trained with FASPSA yields improved system identification as measured by reduced time of convergence and a smaller identification error.展开更多
无线监测网络中多电台监测节点通过捕捉和分析无线用户的通信数据,可以达到监测网络行为、诊断网络故障和管理网络资源的目的,而为多电台监测节点优化选择工作信道、最大化捕获数据量、获得最佳网络监测质量(quality of monitoring,QoM...无线监测网络中多电台监测节点通过捕捉和分析无线用户的通信数据,可以达到监测网络行为、诊断网络故障和管理网络资源的目的,而为多电台监测节点优化选择工作信道、最大化捕获数据量、获得最佳网络监测质量(quality of monitoring,QoM)是一个关键问题。文章研究了一种基于同步微扰随机近似(SPSA)的信道选择算法。该算法在迭代过程中以随机扰动策略得到目标函数的近似梯度,引导搜索过程逐步逼近最优解;适合于复杂的多维优化问题求解,收敛速度快、复杂度低。实验结果表明,该算法可以实现无线监测网络中多电台监测节点的信道优化选择,并且性能优良。展开更多
In past years,growing efforts have been made to the rapid interpretation of magnetic field data acquired by a sparse synthetic or real magnetic sensor array.An appealing requirement on such sparse array arranged withi...In past years,growing efforts have been made to the rapid interpretation of magnetic field data acquired by a sparse synthetic or real magnetic sensor array.An appealing requirement on such sparse array arranged within a specified survey region is that to make the number of sensor elements as small as possible,meanwhile without deteriorating imaging quality.For this end,we propose a novel methodology of arranging sensors in an optimal manner,exploring the concept of information capacity developed originally in the communication society.The proposed scheme reduces mathematically the design of a sparse sensor array into solving a combinatorial optimization problem,which can be resolved efficiently using widely adopted Simultaneous Perturbation and Statistical Algorithm(SPSA).Three sets of numerical examples of designing optimal sensor array are provided to demonstrate the performance of proposed methodology.展开更多
VISCAL （VISSIM calibration） is an automated calibration tool for microscopic simulation parameters in VISSIM environment, based on three heuristic optimization algorithms： （a） genetic algorithm （GA）; Col simult...VISCAL （VISSIM calibration） is an automated calibration tool for microscopic simulation parameters in VISSIM environment, based on three heuristic optimization algorithms： （a） genetic algorithm （GA）; Col simultaneous perturbation stochastic approximation （SPSA）; （c） simulated annealing （SA）. It is developed with a goal to automate and ease the tedious process of calibration, offering greater flexibility to the users by providing control on every aspect of the calibration process. It includes multiple features for a generic application tool with the ability to test the significance of the appropriate decision parameter set for a particular network, to determine the most suitable objective function to reflect network characteristics, and to check the suitability of any of the three heuristic optimization al- gorithms for a particular network. VISCAL also offers four objective function choices into the system： （1） speed, （2） flow, （3） delay, and （4） multi-objective criteria. It is able to calibrate all the driving behavior parameters for any type （urban, rural） and extent of network （small or large network）. However, for this study, the operation of the tool is tested by a dataset obtained from a 3.26 km freeway of Dhaka, Bangladesh.展开更多
文摘Simultaneous perturbation stochastic approximation （SPSA） belongs to the class of gradient-free optimization methods that extract gradient information from successive objective function evaluation. This paper describes an improved SPSA algorithm, which entails fuzzy adaptive gain sequences, gradient smoothing, and a step rejection procedure to enhance convergence and stability. The proposed fuzzy adaptive simultaneous perturbation approximation （FASPA） algorithm is particularly well suited to problems involving a large number of parameters such as those encountered in nonlinear system identification using neural networks （NNs）. Accordingly, a multilayer perceptron （MLP） network with popular training algorithms was used to predicate the system response. We found that an MLP trained by FASPSA had the desired accuracy that was comparable to results obtained by traditional system identification algorithms. Simulation results for typical nonlinear systems demonstrate that the proposed NN architecture trained with FASPSA yields improved system identification as measured by reduced time of convergence and a smaller identification error.
文摘无线监测网络中多电台监测节点通过捕捉和分析无线用户的通信数据,可以达到监测网络行为、诊断网络故障和管理网络资源的目的,而为多电台监测节点优化选择工作信道、最大化捕获数据量、获得最佳网络监测质量(quality of monitoring,QoM)是一个关键问题。文章研究了一种基于同步微扰随机近似(SPSA)的信道选择算法。该算法在迭代过程中以随机扰动策略得到目标函数的近似梯度,引导搜索过程逐步逼近最优解;适合于复杂的多维优化问题求解,收敛速度快、复杂度低。实验结果表明,该算法可以实现无线监测网络中多电台监测节点的信道优化选择,并且性能优良。
文摘In past years,growing efforts have been made to the rapid interpretation of magnetic field data acquired by a sparse synthetic or real magnetic sensor array.An appealing requirement on such sparse array arranged within a specified survey region is that to make the number of sensor elements as small as possible,meanwhile without deteriorating imaging quality.For this end,we propose a novel methodology of arranging sensors in an optimal manner,exploring the concept of information capacity developed originally in the communication society.The proposed scheme reduces mathematically the design of a sparse sensor array into solving a combinatorial optimization problem,which can be resolved efficiently using widely adopted Simultaneous Perturbation and Statistical Algorithm(SPSA).Three sets of numerical examples of designing optimal sensor array are provided to demonstrate the performance of proposed methodology.
基金supported by the Committee for Advanced Studies and Research (CASR)Bangladesh University of Engineering and Technology (BUET)
文摘VISCAL （VISSIM calibration） is an automated calibration tool for microscopic simulation parameters in VISSIM environment, based on three heuristic optimization algorithms： （a） genetic algorithm （GA）; Col simultaneous perturbation stochastic approximation （SPSA）; （c） simulated annealing （SA）. It is developed with a goal to automate and ease the tedious process of calibration, offering greater flexibility to the users by providing control on every aspect of the calibration process. It includes multiple features for a generic application tool with the ability to test the significance of the appropriate decision parameter set for a particular network, to determine the most suitable objective function to reflect network characteristics, and to check the suitability of any of the three heuristic optimization al- gorithms for a particular network. VISCAL also offers four objective function choices into the system： （1） speed, （2） flow, （3） delay, and （4） multi-objective criteria. It is able to calibrate all the driving behavior parameters for any type （urban, rural） and extent of network （small or large network）. However, for this study, the operation of the tool is tested by a dataset obtained from a 3.26 km freeway of Dhaka, Bangladesh.
