Abstract: Considering that the traditional method of solving policy anomalies in firewall is inefficient in cloud environment, the paper proposed an optimized method based on risk value of firewall rules and heap sort algorithm. One risk evaluation model is established to calculate the risk value of rule, using the combination of CVSS and BP neural network self-learning methods. According to risk value of firewall rules, the conflict abnormality will be solved, and the heap sort algorithm will be introduced to solve the redundant abnormality. The comparative experiments show that the improved method can well deal with conflict abnormality and redundant abnormality, improving the efficiency of the firewall