Research On The Two-Dimensional PSD Nonlinear Correction Based On Genetic BP Neural Network

At present the two-dimensional PSD(Position Sensitive Detector, photoelectric position sensitive detector) non-linear correction method is relatively more, but most remain at the stage of qualitative analysis. Although aspects of the two-dimensional PSD structure and use of materials has been improved, it is still difficult to overcome the nonlinear two-dimensional PSD.BP neural network algorithm(Back Propagation Neutral Network Algorithm) has great advantages in solving nonlinear problems or nonlinear structural problems, it can make use of the input and output variables of the neural network training network, to achieve non-linear calibration purposes. Based on the study of the extensive literature on the proposed genetic algorithm(Genetic Algorithm, GA) and BP neural network algorithm used in the combination of two-dimensional nonlinear calibration and PSD to study it. The main contents are as follows:1. Applied to neural network algorithm for the two-dimensional PSD Nonlinear Correction: Based on the analysis of domestic and foreign research background, the significance, the two-dimensional PSD works, performance parameters and research of the status quo, through the non-existing two-dimensional PSD linearity correction method of analysis and comparison of the various ways in which the strengths and weaknesses, will be applied to the neural network algorithm for nonlinear correction of the two-dimensional PSD.2. Analysis on the two-dimensional PSD nonlinear correction based on BP neural network: the nonlinear two-dimensional PSD factors were analyzed in detail; BP neural network algorithm nonlinear mapping ability is applied to the two-dimensional PSD nonlinear correction; It focuses on the implementation process of the nonlinear correction;correction of deficiencies exist in the process were analyzed.3. Research on the two-dimensional PSD nonlinear correction based on genetic BP neural network: PSD proposes a two-dimensional non-linear correction method based on genetic BP neural network. This method is the use of genetic algorithm to optimize BP neural network algorithm for the initial connection weights and thresholds, BP neural network algorithm in accordance with the error function gradient fell method, by continuously adjusting the connection weights and thresholds to reduce the error between the actual output and the desired output, complete Nonlinear correction of the two-dimensional PSD.4. The two-dimensional PSD nonlinear correction method validation: Based on MATLAB programming and simulation, genetic algorithm optimization BP neural network and BP neural network algorithm is applied to two-dimensional PSD nonlinear correction,from convergence, accuracy and smooth were compared in terms of analysis, the optimization method is feasible and superiority.Comparative analysis of different graphic simulation results show that the two-dimensional non-linear PSD based on genetic BP neural network can be corrected in a shorter period of time, to achieve faster convergence, higher precision, accuracy and stability greatly outperforms PSD dimensional nonlinear correction BP neural network,which verifies the combination of two algorithms is feasible, the effect is significant.