针对传统的最小二乘板形模式识别方法的抗干扰能力差、精度低和神经网络方法存在网络学习时间长、易陷入局部最小值等问题,把模糊理论和神经网络的优点融合在一起,通过三个自适应神经模糊推理系统的有效拟合,提出了一种基于自适应神经模糊推理系统的板形模式识别方法。研究结果表明,该方法能够很好地克服以上缺点,而且能够有效识别出常见的板形缺陷,识别速度和精度有所提高,识别结果跟板形仪的实测板形也非常接近。 In order to overcome the problems of poor anti-jamming capability, low precision and neural network method, which have long learning time and easy to fall into local minimum, the traditional least squares flat pattern recognition method combines the advantages of fuzzy theory and neural network, Three adaptive neuro-fuzzy inference system for the effective fitting, an adaptive neuro-fuzzy inference system based on the plate-shaped pattern recognition method. The results show that this method can well overcome the above shortcomings, and can effectively identify the common plate-shaped defects. The identification speed and accuracy are improved. The recognition results are very close to the measured shape of the plate-shaped meter.