Dissertation > Excellent graduate degree dissertation topics show

Research of Freight Volume Forecast Model Based on RBF Neural Network

Author: WeiYanQiang
Tutor: NingHongYun
School: Tianjin University of Technology
Course: Applied Computer Technology
Keywords: Cargo Forecasts RBF Neural Network Extended matrix Dynamic Clustering
CLC: U492.313
Type: Master's thesis
Year: 2008
Downloads: 482
Quote: 4
Read: Download Dissertation

Abstract


Basis for investment decisions as the construction of transport infrastructure, the cargo forecasts has a very important significance in the national and regional economic development planning. Cargo Forecasts goal of this study is how in the transport system excavated effective prediction model of these data and information processing, which provides accurate and efficient cargo forecasts for the traffic management department and transport enterprises, to facilitate the relevant departments and enterprises the unit reasonable arrangements capacity, the freight process control. There is a long history of research on Cargo Forecasts quantitative prediction method is generally used in practice. Time series analysis and regression analysis are based on classical statistical method, they constructed the model has been very mature, in certain cargo forecasts; gray prediction method and neural network is a hot the current prediction research areas in , there is a lot of research space. Radial basis function (Radial Basis Function, RBF) neural network, it has the best function approximation performance and the global optimal characteristics, is currently unmatched by other methods. In this paper, based on the analysis of the research status, RBF neural network for cargo forecasts. Since the traditional RBF neural network learning through empirical formulas or experimenter artificially determine the network structure, so that the trained network is often not optimal. In this regard, the paper improved the RBF neural network learning algorithm, proposed the idea of ??a dynamic clustering-based self-generate hidden layer nodes. It is a two-stage learning algorithm by clustering center and expansion constant, after output weight matrix was obtained by the method of least squares. While studying network structure in the network parameters are also adjusted so that the error is continuously reduced. Programming MATLAB7.0 platform to achieve the learning algorithm, and verified through two sets of instances of the function approximation ability of dynamic clustering algorithm learning efficiency and extrapolation. Twice in a predictable manner, for a number of application examples using dynamic clustering algorithm is applied to cargo forecasts instance, defines a continuation into the time domain information matrix. The basic idea is an expansion of the matrix to improve the structure of the input and output data corresponding to simplify the forecasting process, learning neural network data can be utilized to a greater extent, thereby strengthening the RBF neural network learning, to eliminate the accumulated error . Finally, the dynamic clustering algorithm based on the extended matrix applied to the highway freight traffic forecasts and predictive models of the total freight Experimental results show that the extended matrix-based dynamic clustering algorithm is effective.

Related Dissertations

  1. Linear guide system Elevator single electromagnetic levitation RBF neural network sliding mode control,TP273
  2. The Pattern Recognition Design of Bus Passenger Count System,U491.17
  3. The Research on PID Predictive Control of Full Hydrogen Bell-Type Annealing Furnace Temperature Based on RBF Neural Network,TP273
  4. Based on RBF neural network of rural roads Concrete Old Bridge Damage Assessment Methods,TP18
  5. Characteristic States Based Supervision of Penicllin Fermentation Batch Processes,TP301.1
  6. Study on Freeway Traffic Safety Level of Service Classification Method,U492.8
  7. A Model for Tred of Stock Based on Rough Sets and Network of Radical Basis Function,TP183
  8. Development and Research of Remote Monitoring of Car Running Status and Fault Analysis System,U472.42
  9. Research on Motor Flux Observer Base on Neural Network,TP183
  10. Research on the Microwave Radiation Characteristic of Insulator Contamination Based on Working Condition,TM216
  11. On-Line Measurement of Ceramic Paste Inner Stress Based on the Soft-Measurement Technology,TQ174.6
  12. Study on Process Parameters Optimization of Gas-Assisted Injection Molding for Complex Shell Parts,TQ320.662
  13. The locomotive energy consumption calculation method,U260.1
  14. Study on Fuzzy RBF Neural Network Based on RPROP-DE Algorithm,N945.14
  15. Mine Seismic Activity and Chaotic Time Series Analysis,P315
  16. The Research of Environmental Early-warning System for Foreign Trade of China,F205;F224
  17. Based on adaptive RBF neural network commercial bank credit risk early warning assessment study,F224
  18. Study of the Strip Thickness Intelligent Control Methods,TP273.5
  19. Research and Application in Particle Size Soft-Sensor of Radial Basis Function Neural Network,TP183
  20. Improvement and Application of Ensemble Learning Algorithm,TP181
  21. Reach Based on RBF Neural Network of Pneumatic Position Servo System,TP273

CLC: > Transportation > Road transport > Technical management of traffic engineering and road transport > Operation Technology > Transport of goods and commercial work > Organization and management of cargo transport > Investigation and prediction of the flow of goods and freight traffic
© 2012 www.DissertationTopic.Net  Mobile