A kind of second order algorithm--recursive approximate Newton algorithm was given by Karayiannis. The algorithm was simplified when it was formulated. Especially, the simplification to matrix Hessian was very relucta...A kind of second order algorithm--recursive approximate Newton algorithm was given by Karayiannis. The algorithm was simplified when it was formulated. Especially, the simplification to matrix Hessian was very reluctant, which led to the loss of valuable information and affected performance of the algorithm to certain extent. For multi layer feed forward neural networks, the second order back propagation recursive algorithm based generalized cost criteria was proposed. It is proved that it is equivalent to Newton recursive algorithm and has a second order convergent rate. The performance and application prospect are analyzed. Lots of simulation experiments indicate that the calculation of the new algorithm is almost equivalent to the recursive least square multiple algorithm. The algorithm and selection of networks parameters are significant and the performance is more excellent than BP algorithm and the second order learning algorithm that was given by Karayiannis.展开更多
This paper presents the forward displacement analysis of an 8-PSS(prismatic-spherical-spherical)redundant parallel manipulator whose moving platform is linked to the base platform by eight kinemtic chains consisting o...This paper presents the forward displacement analysis of an 8-PSS(prismatic-spherical-spherical)redundant parallel manipulator whose moving platform is linked to the base platform by eight kinemtic chains consisting of a PSS joint and a strut with fixed length.A general approximation algorithm is used to solve the problem.To avoid the extraction of root in the approximation process,the forward displacement analysis of the 8-PSS redundant parallel manipulator is transformed into another equivalent problem on the assumption that the strut is extensible while the slider is fixed.The problem is solved by a modified approximation algorithm which predicates that the manipulator will move along a pose vector to reduce the difference between the desired configuration and an instantaneous one,and the best movement should be with minimum norm and least quadratic sum.The characteristic of this modified algorithm is that its convergence domain is larger than that of the general approximation algorithm.Simulation results show that the modelified algorithm is general and can be used for the forward displacement analysis of the redundant parallel manipulator actuated by a revolute joint.展开更多
Aimed at the real-time forward kinematics solving problem of Stewart parallel manipulator in the control course, a mixed algorithm combining immune evolutionary algorithm and numerical iterative scheme is proposed. Fi...Aimed at the real-time forward kinematics solving problem of Stewart parallel manipulator in the control course, a mixed algorithm combining immune evolutionary algorithm and numerical iterative scheme is proposed. Firstly taking advantage of simpleness of inverse kinematics, the forward kinematics is transformed to an optimal problem. Immune evolutionary algorithm is employed to find approximate solution of this optimal problem in manipulator's workspace. Then using above solution as iterative initialization, a speedy numerical iterative scheme is proposed to get more precise solution. In the manipulator running course, the iteration initialization can be selected as the last period position and orientation. Because the initialization is closed to correct solution, solving precision is high and speed is rapid enough to satisfy real-time requirement. This mixed forward kinematics algorithm is applied to real Stewart parallel manipulator in the real-time control course. The examination result shows that the algorithm is very efficient and practical.展开更多
In this paper, a practical Werner-type continued fraction method for solving matrix valued rational interpolation problem is provided by using a generalized inverse of matrices. In order to reduce the continued fracti...In this paper, a practical Werner-type continued fraction method for solving matrix valued rational interpolation problem is provided by using a generalized inverse of matrices. In order to reduce the continued fraction form to rational function form of the interpolants, an efficient forward recurrence algorithm is obtained.展开更多
为实现井下落鱼高精度实时探测、保障油气田安全高效作业,攻克现有超声探测系统有线依赖、低信噪比下回波初至提取精度不足、高温适应性差的瓶颈,研制了基于现场可编程门阵列(Field-Programmable Gate Array,FPGA)的井下超声前视探测系...为实现井下落鱼高精度实时探测、保障油气田安全高效作业,攻克现有超声探测系统有线依赖、低信噪比下回波初至提取精度不足、高温适应性差的瓶颈,研制了基于现场可编程门阵列(Field-Programmable Gate Array,FPGA)的井下超声前视探测系统,集成超声发射、回波调理与数据采集电路;针对井下低信噪比、资源受限与高温工况,研究了面向边缘部署的赤池信息准则(Akaike Information Criterion,AIC)算法硬件加速架构,采用固定时窗约束缩小搜索范围、坐标旋转数字计算方法(COordinate Rotation DIgital Computer,CORDIC)迭代单元替代高开销对数运算、构建全流水并行方差计算通路,完成了算法轻量化与硬件协同优化。研究结果表明:①系统可清晰成像落鱼鱼顶形状,常温低信噪比环境下落鱼尺寸还原误差为3%,较阈值法精度提升1.33倍;②120℃高温条件下单点测距功能稳定,测距误差小于2.5 mm;③优化后的AIC硬件架构仅消耗9850个查找表(Look-Up Table,LUT)与116个数字信号处理器(Digital Signal Processor,DSP)资源,处理3001点数据延迟为6.8 ms,资源占用与处理效率显著优于传统查表法与平方法近似;④FPGA逻辑资源占用合理,LUT消耗占总资源39%,DSP消耗占78%,可满足井下实时处理需求。结论认为,该系统能够实现无缆化部署与实时成像,在精度、实时性与高温稳定性上达到井下探测要求,FPGA加速AIC算法可有效提升低信噪比下回波初至提取精度,为深井落鱼探测提供可靠技术方案。展开更多
The Rock-soil interface is a common geological interface.Due to mechanical differences between soil and rock,the stress waves generated by underground blasting undergo intense polarization when crossing the rock-soil ...The Rock-soil interface is a common geological interface.Due to mechanical differences between soil and rock,the stress waves generated by underground blasting undergo intense polarization when crossing the rock-soil interface,making propagation laws difficult to predict.Currently,the characteristics of the impact of the rock-soil interface on blasting stress waves remain unclear.Therefore,the vibration field caused by cylindrical charge blasting in elastic rock and partial-saturation poro-viscoelastic soil was solved.A forward algorithm for the underground blasting vibration field in rock-soil sites was proposed,considering medium damping and geometric diffusion effects of stress waves.Further investigation into the influence of rock and soil parameters and blasting source parameters revealed the following conclusions:stress waves in soil exhibit dispersion,causing peak particle velocity(PPV)to display a discrete distribution.Soil parameters affect PPV attenuation only within the soil,while blasting source parameters affect PPV attenuation throughout the entire site.Multi-wave coupling effects induced by the rocksoil interface result in zones of enhanced and attenuated PPV within the site.The size of the enhancement zone is inversely correlated with the distance from the blasting source and positively correlated with the blasting source attenuation rate and burial depth,providing guidance for selecting explosives and blasting positions.Additionally,PPV attenuation rate increases with distance from the rock-soil interface,but an amplification effect occurs near the interface,most noticeable at 0.1 m.Thus,a sufficient safety distance from the rock-soil interface is necessary during underground blasting.展开更多
In the paper, a method of building mathematic model employing genetic multilayer feed forward neural network is presented, and the quantitative relationship of chemical measured values and near-infrared spectral data ...In the paper, a method of building mathematic model employing genetic multilayer feed forward neural network is presented, and the quantitative relationship of chemical measured values and near-infrared spectral data is established. In the paper, quantitative mathematic model related chemical assayed values and near-infrared spectral data is established by means of genetic multilayer feed forward neural network, acquired near-infrared spectral data are taken as input of network with the content of five kinds of fat acids tested from chemical method as output, weight values of multilayer feed forward neural network are trained by genetic algorithms and detection model of neural network of soybean is built. A kind of multilayer feed forward neural network trained by genetic algorithms is designed in the paper. Through experiments, all the related coefficients of five fat acids can approach 0.9 which satisfies the preliminary test of soybean breeding.展开更多
摘要A kind of second order algorithm--recursive approximate Newton algorithm was given by Karayiannis. The algorithm was simplified when it was formulated. Especially, the simplification to matrix Hessian was very reluctant, which led to the loss of valuable information and affected performance of the algorithm to certain extent. For multi layer feed forward neural networks, the second order back propagation recursive algorithm based generalized cost criteria was proposed. It is proved that it is equivalent to Newton recursive algorithm and has a second order convergent rate. The performance and application prospect are analyzed. Lots of simulation experiments indicate that the calculation of the new algorithm is almost equivalent to the recursive least square multiple algorithm. The algorithm and selection of networks parameters are significant and the performance is more excellent than BP algorithm and the second order learning algorithm that was given by Karayiannis.
基金Funded by the National Natural Science Foundation of China(Grant No.50905102)the China Postdoctoral Science Foundation(Grant No.200801199)the Natural Science Foundation of Guangdong Province(Grant No.8351503101000001)
摘要This paper presents the forward displacement analysis of an 8-PSS(prismatic-spherical-spherical)redundant parallel manipulator whose moving platform is linked to the base platform by eight kinemtic chains consisting of a PSS joint and a strut with fixed length.A general approximation algorithm is used to solve the problem.To avoid the extraction of root in the approximation process,the forward displacement analysis of the 8-PSS redundant parallel manipulator is transformed into another equivalent problem on the assumption that the strut is extensible while the slider is fixed.The problem is solved by a modified approximation algorithm which predicates that the manipulator will move along a pose vector to reduce the difference between the desired configuration and an instantaneous one,and the best movement should be with minimum norm and least quadratic sum.The characteristic of this modified algorithm is that its convergence domain is larger than that of the general approximation algorithm.Simulation results show that the modelified algorithm is general and can be used for the forward displacement analysis of the redundant parallel manipulator actuated by a revolute joint.
摘要Aimed at the real-time forward kinematics solving problem of Stewart parallel manipulator in the control course, a mixed algorithm combining immune evolutionary algorithm and numerical iterative scheme is proposed. Firstly taking advantage of simpleness of inverse kinematics, the forward kinematics is transformed to an optimal problem. Immune evolutionary algorithm is employed to find approximate solution of this optimal problem in manipulator's workspace. Then using above solution as iterative initialization, a speedy numerical iterative scheme is proposed to get more precise solution. In the manipulator running course, the iteration initialization can be selected as the last period position and orientation. Because the initialization is closed to correct solution, solving precision is high and speed is rapid enough to satisfy real-time requirement. This mixed forward kinematics algorithm is applied to real Stewart parallel manipulator in the real-time control course. The examination result shows that the algorithm is very efficient and practical.
摘要In this paper, a practical Werner-type continued fraction method for solving matrix valued rational interpolation problem is provided by using a generalized inverse of matrices. In order to reduce the continued fraction form to rational function form of the interpolants, an efficient forward recurrence algorithm is obtained.
摘要为实现井下落鱼高精度实时探测、保障油气田安全高效作业,攻克现有超声探测系统有线依赖、低信噪比下回波初至提取精度不足、高温适应性差的瓶颈,研制了基于现场可编程门阵列(Field-Programmable Gate Array,FPGA)的井下超声前视探测系统,集成超声发射、回波调理与数据采集电路;针对井下低信噪比、资源受限与高温工况,研究了面向边缘部署的赤池信息准则(Akaike Information Criterion,AIC)算法硬件加速架构,采用固定时窗约束缩小搜索范围、坐标旋转数字计算方法(COordinate Rotation DIgital Computer,CORDIC)迭代单元替代高开销对数运算、构建全流水并行方差计算通路,完成了算法轻量化与硬件协同优化。研究结果表明:①系统可清晰成像落鱼鱼顶形状,常温低信噪比环境下落鱼尺寸还原误差为3%,较阈值法精度提升1.33倍;②120℃高温条件下单点测距功能稳定,测距误差小于2.5 mm;③优化后的AIC硬件架构仅消耗9850个查找表(Look-Up Table,LUT)与116个数字信号处理器(Digital Signal Processor,DSP)资源,处理3001点数据延迟为6.8 ms,资源占用与处理效率显著优于传统查表法与平方法近似;④FPGA逻辑资源占用合理,LUT消耗占总资源39%,DSP消耗占78%,可满足井下实时处理需求。结论认为,该系统能够实现无缆化部署与实时成像,在精度、实时性与高温稳定性上达到井下探测要求,FPGA加速AIC算法可有效提升低信噪比下回波初至提取精度,为深井落鱼探测提供可靠技术方案。
基金supported by the National Natural Science Foundation of China(Grant Nos.41972286 and 42102329).
摘要The Rock-soil interface is a common geological interface.Due to mechanical differences between soil and rock,the stress waves generated by underground blasting undergo intense polarization when crossing the rock-soil interface,making propagation laws difficult to predict.Currently,the characteristics of the impact of the rock-soil interface on blasting stress waves remain unclear.Therefore,the vibration field caused by cylindrical charge blasting in elastic rock and partial-saturation poro-viscoelastic soil was solved.A forward algorithm for the underground blasting vibration field in rock-soil sites was proposed,considering medium damping and geometric diffusion effects of stress waves.Further investigation into the influence of rock and soil parameters and blasting source parameters revealed the following conclusions:stress waves in soil exhibit dispersion,causing peak particle velocity(PPV)to display a discrete distribution.Soil parameters affect PPV attenuation only within the soil,while blasting source parameters affect PPV attenuation throughout the entire site.Multi-wave coupling effects induced by the rocksoil interface result in zones of enhanced and attenuated PPV within the site.The size of the enhancement zone is inversely correlated with the distance from the blasting source and positively correlated with the blasting source attenuation rate and burial depth,providing guidance for selecting explosives and blasting positions.Additionally,PPV attenuation rate increases with distance from the rock-soil interface,but an amplification effect occurs near the interface,most noticeable at 0.1 m.Thus,a sufficient safety distance from the rock-soil interface is necessary during underground blasting.
基金Heilongjiang Natural Science Foundation (F0318).
摘要In the paper, a method of building mathematic model employing genetic multilayer feed forward neural network is presented, and the quantitative relationship of chemical measured values and near-infrared spectral data is established. In the paper, quantitative mathematic model related chemical assayed values and near-infrared spectral data is established by means of genetic multilayer feed forward neural network, acquired near-infrared spectral data are taken as input of network with the content of five kinds of fat acids tested from chemical method as output, weight values of multilayer feed forward neural network are trained by genetic algorithms and detection model of neural network of soybean is built. A kind of multilayer feed forward neural network trained by genetic algorithms is designed in the paper. Through experiments, all the related coefficients of five fat acids can approach 0.9 which satisfies the preliminary test of soybean breeding.