Objective The tracking accuracy of CCD-based photoelectric tracking systems is directly restricted by the control bandwidth. In practical operation, image sensors require a certain period to complete target detection and image processing, which introduces inherent time delay into the system, reduces the control bandwidth, and degrades the tracking performance. Traditional feedforward control can compensate for the influence of time delay, but it highly relies on additional sensors and fixed trajectory prediction models. In real application scenarios, the target motion state is unknown and complex, and a single model can hardly describe its motion characteristics accurately, resulting in limited performance of conventional feedforward control. Meanwhile, traditional physical experimental platforms suffer from high cost, long development cycle, inflexible parameter adjustment, and difficulty in reproducing extreme working conditions, which seriously restrict the rapid iteration and verification of control algorithms. To overcome the above theoretical and engineering limitations, this paper aims to propose an error-based adaptive feedforward control method without requiring additional sensors, and develop a high-fidelity photoelectric tracking digital twin system to realize efficient verification of the proposed algorithm in a virtual environment close to real working conditions. Ultimately, the research is expected to improve the tracking accuracy and operational adaptability of photoelectric tracking systems for targets with complex motion states under time-delay conditions.

Methods Firstly, this paper synthesizes the relative position of the target using the driving quantity and line-of-sight error. Considering the delay existing in the synthesized position and the complexity of the target motion state, a trajectory prediction method based on multi-motion model fusion is designed to predict it and form an adaptive feedforward. Secondly, to break through the limitations of physical testing, a photoelectric tracking digital twin system integrating image vision and servo control functions is developed. A modular design scheme is adopted to realize the closed-loop coordination of visual information and control feedback. This system constructs an equivalent servo module based on the characteristics of the real platform, supports the independent configuration of target motion and environment, and realizes the efficient virtual verification of control algorithms. Finally, digital twin experiments are carried out to verify the effectiveness of the proposed method.

Results and Discussions Simulation experiments are carried out based on the established photoelectric tracking digital twin system, which can accurately reproduce the working state of a real photoelectric tracking system, including image processing delay, complex target motion characteristics, and servo control response. Targets with different motion states are set in the experiments to conduct comparative tests on tracking performance. The results show that the proposed error-based adaptive feedforward control method can effectively compensate for the adverse effects caused by system time delay without relying on additional sensors or accurate trajectory prediction models. Compared with single feedback control, the proposed adaptive feedforward control reduces the tracking error by 88.5% and 88.8% under 0.3 Hz sinusoidal motion and 0.8 Hz sinusoidal motion, respectively, achieving a significant improvement in tracking accuracy under time-delay conditions. In the low-maneuvering scenario of 0.3 Hz low-frequency sinusoidal motion, the adaptive feedforward control effectively inherits the stability advantage of the CA model feedforward and improves the tracking accuracy by 70.8% compared with the CS model feedforward control, solving the problem of insufficient adaptability of a single CS model to low-maneuvering targets. In the high-maneuvering scenario of 0.8 Hz high-frequency sinusoidal motion, the proposed method gives full play to the strong tracking advantage of the CS model feedforward and improves the accuracy by 83.5% compared with the CA model feedforward control, greatly alleviating the prediction lag of a single CA model for high-maneuvering targets. Meanwhile, the developed digital twin system realizes efficient virtual verification of the control algorithm, effectively avoiding the high cost and long cycle of physical tests, and has high engineering application value. In addition, the multi-motion model fusion trajectory prediction method effectively addresses the difficulty that a single model is difficult to describe the complex motion characteristics of targets, further enhancing the operational adaptability and tracking accuracy of the adaptive feedforward control.

Conclusions Aiming at the problems that the bandwidth of the CCD-based photoelectric tracking system is limited by image processing time delay, and the traditional feedforward control has high dependence on trajectory prediction and additional sensors, this paper proposes an error-based adaptive feedforward control method and designs a corresponding photoelectric tracking digital twin system. The method synthesizes the target relative position through driving quantity and line-of-sight error, and adopts multi-motion model fusion for trajectory prediction to form adaptive feedforward, which effectively compensates for time delay and adapts to complex target motion states. The digital twin system realizes efficient virtual verification of the control algorithm, which is more efficient and economical than physical testing. Experimental results verify that the proposed method can significantly improve the tracking accuracy of targets with complex motion states under time-delay conditions, providing a feasible technical solution for improving the performance of photoelectric tracking systems. Future work will focus on optimizing the multi-motion model fusion strategy to further improve the prediction accuracy, and expanding the digital twin system to adapt to more complex environmental interference scenarios, so as to promote the practical application of the proposed method in engineering practice.