HYPERSTATIC STRUCTURE MAPPING MODEL BUILDING AND OPTIMIZING DESIGN

Hyperstatic structure plane model being built by structural mechanics is studied. Space model precisely reflected in real stress of the structure is built by finite element method (FEM) analysis commerce software. Mapping model of complex structure system is set up, with convenient calculation just as in plane model and comprehensive information as in space model. Plane model and space model are calculated under the same working condition. Plane model modular construction inner force is considered as input data; Space model modular construction inner force is considered as output data. Thus specimen is built on input data and output data. Character and affiliation are extracted through training specimen, with the employment of nonlinear mapping capability of the artificial neural network. Mapping model with interpolation and extrapolation is gained, laying the foundation for optimum design. The steel structure of high-layer parking system (SSHLPS) is calculated as an instance. A three-layer back-propagation (BP) net including one hidden layer is constructed with nine input nodes and eight output nodes for a five-layer SSHLPS. The three-layer structure optimization result through the mapping model interpolation contrasts with integrity re-analysis, and seven layers structure through the mapping model extrapolation contrasts with integrity re-analysis. Any layer SSHLPS among 1~8 can be calculated with much accuracy. Amount of calculation can also be reduced if it is applied into the same topological structure, with reduced distortion and assured precision.     

FUZZY NEURAL NETWORK CONTROL FOR VIBRATION WAVEFORM SYSTEM OF MOLD

Combining with the characteristic of the fuzzy control and the neural network control(NNC), a new kind of the fuzzy neural network controller is proposed, and the synthesis design method of the control law and fast speed learning algorithm of the parameters of networks are put forward. The output of the controller is composed of two parts, part one is derived on basis of the principle of sliding control, the lower order model and the estimated parameters of the plant are only required, part two is derived on basis FNN, it is used to compensate the uncertainties of the systems. Because new type of FNN controller extracts from the advantages of the intelligent control and model based sliding mode control, the numbers of adjusting parameters and the structure of FNN are simplified at large, and the practical significance and variation range are attached to each layer of the network and its connected weights, the control performance and learning speed are increased at large. The lightness of the conclusions     

Fault diagnosis for manifold absolute pressure sensor(MAP) of diesel engine based on Elman neural network observer

Intake system of diesel engine is a strong nonlinear system, and it is difficult to establish accurate model of intake system; and bias fault and precision degradation fault of MAP of diesel engine can’t be diagnosed easily using model-based methods. Thus, a fault diagnosis method based on Elman neural network observer is proposed. By comparing simulation results of intake pressure based on BP network and Elman neural network, lower sampling error magnitude is gained using Elman neural network, and the error is less volatile. Forecast accuracy is between 0.015-0.017 5 and sample error is controlled within 0-0.07. Considering the output stability and complexity of solving comprehensively, Elman neural network with a single hidden layer and with 44 nodes is presented as intake system observer. By comparing the relations of confidence intervals of the residual value between the measured and predicted values, error variance and failures in various fault types. Then four typical MAP faults of diesel engine can be diagnosed: complete failure fault, bias fault, precision degradation fault and drift fault. The simulation results show: intake pressure is observable and selection of diagnostic strategy parameter reasonably can increase the accuracy of diagnosis;the proposed fault diagnosis method only depends on data and structural parameters of observer, not depends on the nonlinear model of air intake system. A fault diagnosis method is proposed not depending system model to observe intake pressure, and bias fault and precision degradation fault of MAP of diesel engine can be diagnosed based on residuals.     

Electromagnetic Torque Estimation of a Three-Phase Induction Motor for Setting the Feedrate of a Milling Machine

In this paper, it presents a project of a fuzzy controller and a neural estimator to control a coordinate table powered by three-phase induction motor, aiming to implement an intelligent milling system. The position/speed control is performed using vector techniques of three-phase induction machines. The estimation of the motor electromagnetic torque is used for setting the feedrate of the table. The speed control is developed using TS （Takagi-Sugeno） fuzzy logic model and electromagnetic torque estimation using neural network type LMS （least mean square） algorithm. The induction motor is powered by a frequency inverter driven by a DSP （digital signal processor）. Control strategies are implemented in DSP. Simulation results are presented for evaluating the performance of the system.     

NONLINEAR MODELING AND CONTROLLING OF ARTIFICIAL MUSCLE SYSTEM USING NEURAL NETWORKS

The pneumatic artificial muscles are widely used in the fields of medical robots, etc. Neuralnetworks are applied to modeling and controlling of artificial muscle system. A single-joint artificialmuscle test system is designed. The recursive prediction error (RPE) algorithm which yields fasterconvergence than back propagation (BP) algorithm is applied to train the neural networks. Therealization of RPE algorithm is given. The difference of modeling of artificial muscles using neuralnetworks with different input nodes and different hidden layer nodes is discussed. On this basis thenonlinear control scheme using neural neworks for artificial muscle system has been introduced. Theexperimental results show that the nonlinear control scheme yields faster response and higher controlaccuracy than the traditional linear control scheme.     

Nonlinear dynamic analysis of coupled gear-rotor-bearing system with the effect of internal and external excitations

Extensive studies on nonlinear dynamics of gear systems with internal excitation or external excitation respectively have been carried out. However, the nonlinear characteristics of gear systems under combined internal and external excitations are scarcely investigated. An eight-degree-of-freedom(8-DOF) nonlinear spur gear-rotor-bearing model, which contains backlash, transmission error, eccentricity, gravity and input/output torque, is established, and the coupled lateral-torsional vibration characteristics are studied. Based on the equations of motion, the coupled spur gear-rotor-bearing system(SGRBS) is investigated using the Runge-Kutta numerical method, and the effects of rotational speed, error fluctuation and load fluctuation on the dynamic responses are explored. The results show that a diverse range of nonlinear dynamic characteristics such as periodic motion, quasi-periodic motion, chaotic behaviors and impacts exhibited in the system are strongly attributed to the interaction between internal and external excitations. Significantly, the changing rotational speed could effectively control the vibration of the system. Vibration level increases with the increasing error fluctuation. Whereas the load fluctuation has an influence on the nonlinear dynamic characteristics and the increasing excitation force amplitude makes the vibration amplitude increase, the chaotic motion may be restricted. The proposed model and numerical results can be used for diagnosis of faults and vibration control of practical SGRBS.     

Prediction of Wing Aerodynamic Performance in Rain Using Neural Net

A new method for prediction of wing aerodynamic performance in rain condition was presented.Three-and four-layer artificial neural networks based on improved algorithm for error Back Propagation(BP)network were respectively built.Detailed approaches to determine the optical parameters for network model were introduced and the specific steps for applying BP network model to predict wing aerodynamic performance in rain were given.On this basis,the established optimal three-and four-layer BP network model was used for this prediction.Results indicate that both of the network models are appropriate for predicting wing aerodynamic performance in rain.The sum of square error level produced by two models is less than 0.2%,and the prediction accuracy by four-layer network model is higher than that of three-layer network.     

考虑风速相关性和出力不确定性的风电场运行风险评估

Aiming at the problem that the operation risk assessment of wind farm was not accurate,the wind speed dependence,the wind power output forecasting model,and the operational risk assessment were studied. The matrix transform method was used to simulate the wind speed conditions between different wind farm,and quantity the wind speed conditions with certain correlation. The wind power output model was used to calculate the predicted values of short-termwind power,and the Beta distribution was used to model the conditional probability of a given output. At the same time,in order to quantify the impact of weather factors especially the bad weather on the operation risk assess­ment of wind farm,the two state weather model was used. And an operation risk assessment method of wind farm operation was presented, which took into account the correlation of wind speed and the output uncertainty and the weather factor. The risk assessment process based on non-sequential Monte Carlo method was given,and the improved IEEE-RTS79 system with wind farm was used for the simulation. The results indicate that the method can provide more accurate and reasonable risk assessment results,and provide reference for the operation risk-assessment of wind farms. [ABSTRACT FROM AUTHOR]     

CONTROL SCHEMES FOR CMAC NEURAL NETWORK-BASED VISUAL SERVOING

In IBVS (image based visual servoing), the error signal in image space should be transformed into the control signal in the input space quickly. To avoid the iterative adjustment and complicated inverse solution of image Jacobian, CMAC (cerebellar model articulation controller) neural network is inserted into visual servo control loop to implement the nonlinear mapping. Two control schemes are used. Simulation results on two schemes are provided, which show a better tracking precision and stability can be achieved using scheme 2.     

Nonlinear modeling and identification of the electro-hydraulic control system of an excavator arm using BONL model

Electro-hydraulic control systems are nonlinear in nature and their mathematic models have unknown parameters. Existing research of modeling and identification of the electro-hydraulic control system is mainly based on theoretical state space model, and the parameters identification is hard due to its demand on internal states measurement. Moreover, there are also some hard-to-model nonlinearities in theoretical model, which needs to be overcome. Modeling and identification of the electro-hydraulic control system of an excavator arm based on block-oriented nonlinear（BONL） models is investigated. The nonlinear state space model of the system is built first, and field tests are carried out to reveal the nonlinear characteristics of the system. Based on the physic insight into the system, three BONL models are adopted to describe the highly nonlinear system. The Hammerstein model is composed of a two-segment polynomial nonlinearity followed by a linear dynamic subsystem. The Hammerstein-Wiener（H-W） model is represented by the Hammerstein model in cascade with another single polynomial nonlinearity. A novel Pseudo-Hammerstein-Wiener（P-H-W） model is developed by replacing the single polynomial of the H-W model by a non-smooth backlash function. The key term separation principle is applied to simplify the BONL models into linear-in-parameters struc~tres. Then, a modified recursive least square algorithm（MRLSA） with iterative estimation of internal variables is developed to identify the all the parameters simultaneously. The identification results demonstrate that the BONL models with two-segment polynomial nonlinearities are able to capture the system behavior, and the P-H-W model has the best prediction accuracy. Comparison experiments show that the velocity prediction error of the P-H-W model is reduced by 14%, 30% and 75% to the H-W model, Hammerstein model, and extended auto-regressive （ARX） model, respectively. This research is helpful in controller design, system monitoring and diagnosis.     

Research on Power Control of Wind Power Generation Based on Neural Network Adaptive

For the characteristics of wind power generation system is multivariable,nonlinear and random,in this paper the neural network PID adaptive control is adopted.The size of pitch angle is adjusted in time to improve the performance of power control.The PID parameters are corrected by the gradient descent method,and Radial Basis Function(RBF)neural network is used as the system identifier in this method.Simulation results show that by using neural network adaptive PID controller the generator power control can inhibit effectively the speed and affect the output power of generator.The dynamic performance and robustness of the controlled system is good,and the performance of wind power system is improved.     

Research on Power Control of Wind Power Generation Based on Neural Network Adaptive Control

-For the characteristics of wind power generation system is multivariable,nonlinear and random,in this paper the neural network PID adaptive control is adopted.The size of pitch angle is adjusted in time to improve the performance of power control.The PID parameters are corrected by the gradient descent method,and Radial Basis Functinn(RBF)neural network is used as the system identifier in this method.Simulation results shaw that by using neural adaptive PID controller the generator power control can inhibit effectively the speed and affect the output power of generator.The dynamic performance and robustness of the controlled system is good,and the performance of wind power system is improved.     

基于结构自适应径向基神经网络的油样光谱数据建模

基于光谱分析数据的机械磨损状态预测有利于发现机械系统的早期磨损故障。由于神经网络对于非线性模型的辨识和非平稳信号的预测，与传统预测模型相比具有明显的优势，将神经网络预测方法运用于光谱分析，提出了基于神经网络预测的光谱分析监测技术。在预测模型中采用了在函数逼近、分类能力和学习速度均优于BP网络的径向基函数（RBF）神经网络模型，针对RBF网络的结构对于信号预测或模型辨识的精度具有影响很大的问题，提出了结构自适应RBF网络预测模型。利用遗传算法，对神经网络输入节点数、径向基函数分布系数及网络训练误差进行了优化，得到了最优的RBF网络预测模型。最后，对某航空发动机实际的光谱分析数据进行了预测和分析，并与ARMA模型进行了比较，结果充分表明了文中方法的有效性和优越性。     

应用径向基函数神经网络的经纬仪跟踪误差建模

提出了一种基于径向基函数(RBF)神经网络建立光电经纬仪等效跟踪误差模型的方法来评价光电经纬仪的跟踪性能.分析了光电经纬仪存在的非线性因素,说明了采用理论建模方法难以准确描述其全部过程的原因.然后,介绍了RBF神经网络和靶标系统,基于一组靶标参数建立了RBF神经网络模型,并更换靶标参数进行模型验证.最后,对更换后的靶标参数进行重新训练建模,并改变参数周期,对模型进行了验证.实验结果表明:所建的神经网络模型精度与靶标参数有关,当动态靶标的半椎角a为21.2°,倾角b为43.8°,靶标匀速运行周期T为8.5s时,网络模型在靶标速度最大时误差也达到最大为3.18′,其它时刻均小于0.6′.当a为16.6°,b为37.5°,T为13 s时,模型最大误差为1.8′左右,在此模型下真实输出与网络模型输出的最大偏差为2.4′左右,其它时刻均小于1.2′.结果表明,采用RBF神经网络所建立的跟踪误差模型能够反应真实系统的情况,是可行实用的,且具有较高的精度和泛化能力.     

轴对称件拉深成形智能化控制过程中参数实时识别的GA-ENN建模

材料性能参数和摩擦系数的实时识别是实现拉深过程智能化控制的关键。建立了遗传算法与神经网络相结合的识别模型（GA-ENN），利用遗传算法进行网络权系的训练和优化。给出了网络输入层，输出层和稳层的确定方法以及GA-ENN模型的学习算法。验证结果表明，与BP网络模型比较，GA-ENN模型的学习效率和学习精度均有明显的提高，是一种有效的识别模型，为实现拉深成形过程的智能化控制奠定了基础。     

汽车驾驶机器人车速跟踪神经网络控制方法

为了实现汽车驾驶机器人对给定车速的准确跟踪,提出了一种驾驶机器人车速跟踪神经网络控制方法。网络模型输入层变量为驾驶机器人油门和制动器、离合器机械腿、换挡机械手的位移;中间层为隐层,节点数为5,神经元传递函数为正切传递函数;输出层变量为试验车辆车速,神经元传递函数为线性传递函数。结果表明,该方法的收敛速度明显高于梯度下降法的收敛速度,且达到的控制精度更高,车速跟踪误差满足国家汽车试验标准的要求。     

大量程柔性铰六维力传感器静态解耦的研究

为提高大量程六维力传感器的测量精度,提出了一种新型的六维力传感器非线性静态解耦方法,该方法结合混合递阶遗传算法和小波神经网络的优点,采用递阶遗传算法与最小二乘法分别对小波神经网络隐层结构参数以及输出层权值进行优化,再将优化后的小波神经网络模型用于六维力传感器非线性解耦.建立了基于混合递阶遗传算法和优化小波神经网络的六维力传感器非线性解耦模型,设计了基于混合递阶遗传算法的小波神经网络结构及参数优化算法,给出了六维力传感器非线性解耦的具体实现流程.以最新研制的6-UPUR大量程柔性铰六维力传感器为对象进行实验,结果表明,采用该方法六维力传感器的Ⅰ类误差和Ⅱ类误差分别为1.25％和2.59％,比采用BP和RBF神经网络方法的测量精度高.     

基于神经网络的振动响应趋势预测研究

将Levenberg-M arquardt BP人工神经网络应用于复杂的非线性振动响应的趋势预测,避免了时序分析复杂的数据预处理、模型识别、参数估计和模型适用性检验过程。通过对样本预测效果的比较,全面考虑了网络的输入层、隐层和输出层的神经元节点数和各层之间的传递函数对预测精度的影响,引入Box-Cox变换改善了网络的收敛性并加快了网络的收敛速度,同时采用重复训练法来提高网络的稳定性和预测精度。预测实例表明,相比于传统的时间序列分析方法,这种预测方法能对振动响应的趋势进行更准确的预测。     

基于径向基函数神经网络的压电式六维力传感器解耦算法

针对四点支撑结构的压电式六维力传感器线性度差,维间耦合严重的问题,提出了基于径向基函数(RBF)神经网络的解耦算法。分析了耦合产生的主要原因,建立了RBF神经网络模型。通过对六维力传感器进行标定实验获取解耦所需的实验数据,并对实验数据进行处理。然后采用RBF神经网络优化传感器输出系统的多维非线性解耦算法,解耦出传感器的输入输出映射关系,得到解耦后的传感器输出数据。对传感器解耦后的数据分析表明:采用RBF神经网络的解耦算法得到的最大Ⅰ类误差和Ⅱ类误差分别为1.29%、1.56%。结果显示:采用RBF神经网络的解耦算法,能够更加有效地减小传感器的Ⅰ类误差和Ⅱ类误差,满足了传感器两类误差指标均低于2%的要求。该算法有效地提高了传感器的测量精度,基本解决了传感器解耦困难的难题,     

量子BP神经网络在发动机故障诊断中的应用

为了解决普通BP神经网络收敛速度慢、分类正确率低等问题,提出一种量子BP神经网络算法。该算法在普通BP神经网络中引入了量子算法,量子BP神经网络结构由输入层、隐含层和输出层组成,其中,量子神经元的输入和传递函数均由量子比特表示,输出结果为实数。首先,该算法将实数值训练样本变换为量子态训练样本,从而作为算法的输入。然后,通过传递函数,计算量子态权值并更新网络参数以达到训练效果。最后,利用训练好的网络进行故障诊断,并将结果以实数值输出。将该方法应用于发动机故障诊断,实验结果表明,与普通BP神经网络相比,量子BP神经网络算法在收敛速度、分类正确率和执行时间等方面具有明显的优势。