为了提高智能座椅坐姿识别的准确性,提出基于改进BP神经网络算法的坐姿评价模型。采用大面积阵列分布式薄膜压力传感器进行原始坐姿样本采集,建立BP神经训练网络,通过优化其学习率对网络进行改进,并编写程序提取训练结果。实验表明,该方案的准确性和实用性都比较高。 In order to improve the accuracy of sitting posture recognition, a sitting posture evaluation model based on improved BP neural network algorithm is proposed. A large area array of distributed film pressure sensors was used to collect the original sitting samples, BP neural training network was established, the network was improved by optimizing its learning rate, and the program was written to extract the training results. Experiments show that the program is more accurate and practical.