Abstract: The RFID reader collision problem imposes great influence on RFID system performance.A solution for the problem is to assign time slots or frequencies to readers reasonably by graph coloring method.However，graph coloring problem is a NP difficult problem.Fortunately，with the good non-linearity approach ability，the neural network can find solution for the problem. Therefore，an anti-collision algorithm based on neural network graph coloring was proposed.The reader collision types，solutions were analyzed，and the anti-collision algorithm was presented in detail，formula deduction and the energy function of the algorithm were proposed.The simulation result has demonstrated that the proposed algorithm is effective.

摘要： 阅读器冲突问题严重影响了RFID系统的性能，降低了识别率。使用图着色方法将频率或时隙等资源合理分配，可以防止阅读器冲突的发生。但是图着色问题是一个NP难题，利用神经网络良好的非线性逼近能力，提出基于神经网络图着色的阅读器防冲突算法。分析了阅读器冲突类型及解决方法，给出了算法的详细步骤、公式推导和能量函数，并通过计算机仿真验证了算法的有效性。