Abstract: To solve the problem of multireaders’collision，an optimization algorithm of neural network based on Probabilistic Power Control was proposed． In the algorithm each reader’s transmitter power changed with the dynamic distribution of β， then the sensors were adopted to determine the similarity of readers’power on the basis of the Probabilistic Power Control， which can enable the readers work under best conditions． The experimental results show that the interference between multiple readers greatly reduced，and the distance between their identification labels can get the desired value．

Key words: RFID, neural network, reader, anti-collision problem

摘要： 为了解决多阅读器碰撞问题，提出了基于概率功率控制的神经网络优化算法。该算法中各阅读器的发射功率 按β分布动态发生变化，然后利用简单神经网络即感知器对多个阅读器功率的相似性进行判别并处理，使各阅读器 彼此之间工作在最佳状态。仿真实验结果表明:多阅读器之间的干扰大大减小，其识别标签的距离可以达到期望 值。

关键词: RFID, 神经网络, 阅读器, 防碰撞问题