Research On Tracking Control Technology Of Photoelectric System Based On Predictive Filtering

Photoelectric tracking systems are widely used in the fields of laser communications,aerospace,astronomical observation,and military guidance.Due to the increased randomness and mobility of tracking targets,many problems are caused.such as the closed loop of the system caused by the time delay of the image sensor that detects the target.For example,the time delay of the image sensor that detects the target severely limits the system closed-loop bandwidth.The strong mobility and weak motion law of the target seriously reduces the tracking accuracy.The target is easily suffered occlusion by various environmental objects when moving in low altitude.In order to overcome the above problems,this article is bent on the research of photoelectric tracking technology based on predictive filtering,focusing on the improvement and innovation of predictive filtering technology from motion model,filtering principle,input signal and application scheme,thereby improving the system tracking performance and robustness.The main research works are divided into the following four parts.First of all,from the perspective of predictive filter algorithm's motion model,this research innovatively proposes the idea of combining machine learning and motion models.Specifically,in view of the contradiction between the computational complexity and accuracy of the current mainstream interactive multi-model algorithm,this research proposes to use the naive Bayes classifier to classify the model set used by the interactive multi-model algorithm in real time,and switch the target motion model used by the filter according to the classification result.When the classification is relative accurate,this method can ensure the accuracy of the algorithm while reducing the computational complexity.Secondly,from the perspective of the filtering principle of the predictive filter algorithm,this research proposes two robust predictive filter algorithms that consider the uncertainty of the target motion model and the measurement delay of the target motion trajectory simultaneously.For the time delay problem of the fusion target trajectory unique to the photoelectric tracking system,the measurement delay is introduced into the existing robust state estimation method.The state vector of the system is augmented by a special dimension expansion method,and the system with measurement delay is equivalently converted into a delay-free augmented system.The iterative filtering algorithm and its convergence and boundedness conditions are further derived.Finally,simulation experiments verify the effectiveness of the proposed robust predictive filtering algorithm.Next,from the perspective of predictive filtering algorithm's input signal,this research proposes an acceleration frequency-domain fusion method based on the combination of model output and sensor data.As the accelerometer sensor measuring the attitude of the system itself is severely affected by drift and noise at low frequencies,the fusion target trajectory is inaccurate,and finally the accuracy of the predicted target state information is reduced.Therefore,the system model of controlled object is established.The mid-high-frequency information of the accelerometer is used to further identify the controlled object.Then an open-loop model of system acceleration with known parameters can be obtained.The acceleration information of the system can be calculated by using the drive input signal and the known acceleration open-loop model.The low-frequency signal of the system acceleration model and the mid-high-frequency signal of the accelerometers are used respectively to perform an open-loop fusion so as to obtain a new acceleration.The new acceleration overcomes low-frequency drift and noise,and is accurate in all frequency ranges.Therefore,the new acceleration can be used to obtain a more accurate synthetic target trajectory.Finally,the predictive filtering technology can be improved from the perspective of the input terminal.Finally,from the perspective of predictive filtering algorithms' application,this research proposes an anti-occlusion servo control scheme based on the cooperation of predictive filtering technology and image processing.For the problem that the target is completely occluded in the short-term when passing through various environmental objects,this research proposes to select the control mode and prediction mode of the servo control unit according to the target occlusion state information provided by image processing.At the same time,the location of the target in the image is predicted to improve the speed and accuracy of image processing to recapture the target.Based on the scheme proposed in this study,a verification scheme of the actual experimental system is designed,and the final experimental results verified the feasibility and effectiveness of the scheme proposed in this study.