本文通过引入新息相关法，提出了一种基于新息相关性自调整的交互多模型算法。该方法首先利用模型概率识别出最可能近似真实模式的模型，而后针对该模型应用新息相关法滤波，在系统的数学模型和噪声的统计模型不精确的情况下，通过监测残差是否为白噪声，来相应地调整卡尔曼增益，从而使卡尔曼滤波达到最佳。关键词：卡尔曼滤波 交互多模型算法 新息自调整Abstract: a new IMM algorithm based on the auto-adjustable correlation of the residual isprovided. The algorithm first recognize the most probable model that near real mode by usingmodel probability, then filter for the recognized model using the correlation of the residual. The lgorithm adjust Kalman gain ,by monitor whether the residual is white noise when the system’s athmetical model and the statistical model of the noise are not accurate, so that Kalman filter is ptimal.Keywords: kalman filter ;interacting multiple algorithm; the auto-adjustable of the residual