其他摘要 | With the rapid development of airborne and spaceborne applications, portable military investigation, positioning, command weapon system, higher demand were required for the angular accuracy and resolution of small photoelectric encoders. Since the quality of Moire fringe photoelectric signals constrains high-resolution interpolation of small photoelectric encoder, and the interpolation error and the long-period error are main factors that affect the accuracy of photoelectric encoders, it is very important to carry out research on the moire fringe signals calibration, new interpolation methods, and long-period error correction to improve the resolution and accuracy of small photoelectric encoders and thus follow the international advanced level.Based on large reference at home and abroad, starting from the principle of Moire fringe photoelectric signals generation, the factors which affected the accuracy of photoelectric signals were analyzed thoroughly, including the Moire fringe signal’s quality for high-resolution interpolation, reasons of interpolation error generation, and long-period errors for the accuracy of photoelectric encoders. Much effort was paid on the methods of improving the resolution and accuracy of small photoelectric encoders.The error calibration method was proposed for Moire fringe photoelectric signals in the form of approximate triangular waves. The parameter equation of Moire fringe photoelectric signal waveform was built firstly, which contained direct current errors, amplitude errors and waveform distortions, and then Foruier transform was applied to solve the waveform parameters. The high spatial harmonics in the signal waveform were transformed into higher order components using the multiple angle formula, and photoelectric signals were calibrated to standard sine and cosine ones using the Newton iteration method. After that, the phase errors between signals were calibrated by the least squares fitting method. The calibration from two channels sample signals to the standard orthogonal sine and cosine signals was realized, which improved the interpolation accuracy of small photoelectric encoders.A novel high-resolution interpolation method based on CORDIC algorithm was proposed. Simple shifting and addition operations were used to resolve the interpolation phases of moire fringe photoelectric signals, which could get rid of conventional “computed interpolation” error cased by the deviation of sample signals from the ideal ones. For the contradiction of calculation speed and accuracy existing in CORDIC algorithm, two sorts of improvement were proposed including approximation of rotation angles and prediction of rotation direction. The resolution of small photoelectric encoder was thus improved.The Fourier neural network model to correct long-period error of small photoelectric encoder was established. Output values of a high accuracy benchmark encoder were set as the learning reference, and orthogonal trigonometric basis functions served as activation function of nodes in intermediate hidden layers. The differential evaluation algorithm combined with simulated annealing strategy was applied for training to solve the error correction parameters. The correction of the long-period error improved the accuracy of small photoelectric encoder effectively.The methods proposed above were applied to a small photoelectric encoder which was made by CIOMP. After actual measurement, the resolution of the photoelectric encoder was improved from 16-bit to 18-bit and the root mean square error was reduced from 60″ to 20″. The experiment result shows that, the methods could effectively improve the resolution and accuracy of small photoelectric encoders, and it is beneficial for development of miniaturized, high-resolution and high-accuracy small photoelectric encoders. |
修改评论