摘要：Constructing metamodel with global high-fidelity in design space is significant in engineering design. In this paper, a double-stage metamodel （DSM） which integrates advantages of both interpolation metamodel and regression metamodel is constructed. It takes regression model as the first stage to fit overall distribution of the original model, and then interpolation model of regression model approximation error is used as the second stage to improve accuracy. Under the same conditions and with the same samples, DSM expresses higher fidelity and represents physical characteristics of original model better. Besides, in order to validate DSM characteristics, three examples including Ackley function, airfoil aerodynamic analysis and wing aerodynamic analysis are investigated, In the end, airfoil and wing aerodynamic design optimizations using genetic algorithm are presented to verify the engineering applicability of DSM.

摘要：Robust optimization approach for aerodynamic design has been developed and applied to supercritical wing aerodynamic design. The aerodynamic robust optimization design system consists of genetic optimization algorithm, improved back propagation （bp） neural network and deformation grid technology. In this article, the bp neural network has been improved in two major aspects to enhance the training speed and precision. Uniformity sampling is adopted to generate samples which will be used to establish surrogate model. The testing results show that the prediction precision of the improved bp neural network is reliable. On the assumption that the law of Mach number obeys normal distribution, supercritical wing configuration considering fuselage interfering of a certain aerobus has been taken as a typical example, and five design sections and twist angles have been optimized. The results show that the optimized wing, which considers robust design, has better aerodynamic characteristics. What＇s more, the intensity of shock wave has been reduced.

摘要：The existing methods for extracting the arrival time and amplitude of ultrasonic echo cannot eff ectively avoid the local interference of ultrasonic signals while drilling,which leads to poor accuracy of the echo arrival time and amplitude extracted by an ultrasonic imaging logging-while-drilling *** this study,a demodulation algorithm is used to preprocess the ultrasonic simulation signals while drilling,and we design a backpropagation neural network model to fit the relationship between the waveform data and time and *** ultrasonic imaging logging model is established,and the finite element simulation software is used for forward *** response under diff erent measurement conditions is simulated by changing the model parameters,which are used as the input layer of the neural network model;The ultrasonic echo signal is considered as a low-frequency signal modulated by a high-frequency carrier signal,and a low-pass fi lter is designed to remove the high-frequency signal and obtain the low-frequency envelope *** the amplitude of the envelope signal and its corresponding time are extracted as an output layer of the neural network *** comparing the application eff ects of the various training methods,we fi nd that the conjugate gradient descent method is the most suitable method for solving the neural network *** performance of the neural network model is tested using 11 groups of simulation test data,which verify the eff ectiveness of the model and lay the foundation for further practical application.

摘要：Subcellular location is one of the key biological characteristics of proteins. Position-specific profiles （PSP） have been introduced as important characteristics of proteins in this article. In this study, to obtain position-specific profiles, the Position Specific lterative-Basic Local Alignment Search Tool （PSI-BLAST） has been used to search for protein sequences in a database. Position-specific scoring matrices are extracted from the profiles as one class of characteristics. Four-part amino acid compositions and lst-7th order dipeptide compositions have also been calculated as the other two classes of characteristics. Therefore, twelve characteristic vectors are extracted from each of the protein sequences. Next, the characteristic vectors are weighed by a simple weighing function and inputted into a bp neural network predictor named PSP-Weighted neural network （PSP-WNN）. The Levenberg-Marquardt algorithm is employed to adjust the weight matrices and thresholds during the network training instead of the error back propagation algorithm. With a jackknife test on the RH2427 dataset, PSP-WNN has achieved a higher overall prediction accuracy of 88.4% rather than the prediction results by the general bp neural network, Markov model, and fuzzy k-nearest neighbors algorithm on this dataset. In addition, the prediction performance of PSP-WNN has been evaluated with a five-fold cross validation test on the PK7579 dataset and the prediction results have been consistently better than those of the previous method on the basis of several support vector machines, using compositions of both amino acids and amino acid pairs. These results indicate that PSP-WNN is a powerful tool for subcellular localization prediction. At the end of the article, influences on prediction accuracy using different weighting proportions among three characteristic vector categories have been discussed. An appropriate proportion is considered by increasing the prediction accuracy.

摘要：In order to construct objective relatively mapping relationship model between customer requirements and product technical characteristics, a novel approach based on customer satisfactions information digging from case products and satisfaction information of expert technical characteristics was put forward in this paper. Technical characteristics evaluation values were expressed by rough number, and technical characteristics target sequence was determined on the basis of efficiency, cost type and middle type in this method. Use each calculated satisfactions of customers and technical characteristics as input and output elements to construct bp network model. And we use MATLAB software to simulate this bp network model based on the case of electric bicycles.

摘要：The attitude adjustment of rope system faces the challenging due to the difficulty in obtaining accurate three-dimensional(3D)mathematical model and solving by traditional methods.A set of adjustment systems is designed and used to investigate the automatic control for level or preset attitude adjustment of unknown weights and eccentric *** system principle and characteristics are *** 3D model is decomposed into two two-dimensional(2D)subsystems,and an adaptive fuzzy controller based on bp neural network and least squares(LSE)is *** simulation experiment uses MATLAB to train the level-adjustment data for testing algorithm,and a small load is used to verify the effectiveness of the *** experimental results show that precise attitude adjustment can be achieved within the system load range,and the response speed is *** adjustment method provides a fast and effective method for precise adjustment of the load attitude.