Abstract: Focused on the issue that the precision of infrared gas sensor is affected greatly by temperature, a new method is put forward for sensor temperature compensation based on Adaptive Levy mutation Immune Particle Swarm Optimization-
Least Square Support Vector Machine（ALIPSO-LSSVM）. Levy flight is introduced in the adaptive mutation of offspring to ensure the diversity, and opposition-based learning is used to initialize the particle swarm to improve the convergence speed in the ALIPSO algorithm. Performance comparison with other PSOs is made through 5 benchmark test functions. Based on the ALIPSO, the optimum parameter selection of Least Squares SVM（LS-SVM） is studied, and the temperature compensation model of infrared gas sensor is established, the numerical simulation results show the relative error can be controlled within 6%.

Key words: Levy flight, adaptive, particle swarm optimization, infrared gas sensor, temperature compensation

摘要： 针对红外气体传感器测量精度受环境温度影响较大的问题，提出了一种基于嵌入自适应Levy变异免疫粒子群-最小二乘支持向量机（ALIPSO-LSSVM）的温度补偿算法。ALIPSO算法引入Levy flight对子代粒子进行自适应变异，确保粒子多样性，并在每次迭代之前，采用相对基学习方法初始化粒子群，提高算法的收敛速度。通过5个基准测试函数对ALIPSO算法进行性能评价，仿真结果表明该算法收敛速度较快、精度高，且具有较强的全局搜索能力。利用ALIPSO算法对LS-SVM的参数进行优化，并将该混合算法应用于红外气体传感器温度补偿，数值仿真结果表明采用该算法可将补偿结果的相对误差控制在6%范围内。

关键词: Levy flight, 自适应, 粒子群优化, 红外气体传感器, 温度补偿