基于概率模型的实时修正IMM目标跟踪算法

针对传统的IMM算法采用固定测量噪声协方差矩阵和Markov转移概率矩阵导致模型切换缓慢，跟踪精度下降的问题，提出了一种具有模型概率实时修正的IMM机动目标跟踪算法。该算法在监控区域上建立无线电指纹库，利用支持向量回归算法训练得到观测模型。引入模糊神经网络，在模型交互输出阶段自适应地调整测量误差协方差矩阵。根据IMM子模型中连续时间点之间的模型概率的比值，对Markov转移概率进行修正。仿真结果表明，提出的方法在实时性、跟踪精度方面具有良好的性能。

Real-Time Correction of IMM Target Tracking Algorithm Based on Probability Model

This paper addresses the problem of maneuvering target tracking in wireless sensor networks. In most cases, the covariance matrix of the measurement noise and Markov parameters are assumed constant, so it results in slow model switching and decreases tracking accuracy. To overcome this problem, this paper proposes an IMM maneuver target tracking algorithm based on model probability real-time correction. The algorithm collects the radio-fingerprint of the received signal strength indicators（RSSIs） in the monitoring area, and then utilizes the support vector regression algorithm to train the observation model. The fuzzy neural network is introduced to adaptively adjust the measurement error covariance matrix during the multiple model interacting output stage. The Markov probability transition matrix is adjusted by the probability ratio between two continuous time points in the IMM sub-model. The simulation results show that the proposed method has good performance in real-time and tracking accuracy.