Abstract:

Objective Due to the lightweight and damage tolerance of composite materials, more and more large-sized shaped structural parts in important fields such as aerospace are using composite materials instead of traditional metal materials. This paper presents a computational solution for a dual robot cooperative braiding process, as well as a model for predicting yarn trajectories for robot-tracted mandrel braiding, taking into account the robot trajectory and the geometric characteristics of complex mandrels. This method facilitates improved braiding accuracy, shortens the composite design cycle and provides a basis for mechanical analysis of the composite at a later stage.

Method The spatial geometry of the discrete mandrel and the braiding plane was the research focus, and the trajectory of the robot end was solved using a rotational and translational transformation to obtain position and attitude information. Based on the spatial relationship between the yarn and the mandrel during the braiding process, a yarn prediction was made for this robot trajectory, which was used todetermine whether the yarn has been deposited on the surface of the mandrel based on the geometric relationship between the yarn and the surface of the mandrel at different moments and to predict the fabric construction.

Results According to the calculation model established in this paper, the trajectory of the dual robots was obtained. The master robot trajectory ensured the mandrel to pass vertically through the braiding plane. In the bending part, the robot end-effector away from the braiding plane demonstrated a larger movement stroke, and the slave robot trajectory was constrained by the initial position relationship of the dual robot end-effectors, while the dual robot end-effectors were guaranteed to be relatively stationary when moving to any point. The experimental results showed that the error between the braiding angle of the equal section part of the mandrel and the expected value could be within ±3 degrees and that of the bending unequal section part could be within ±7 degrees by using the robot trajectory solved in this paper. The large error at the bend was due to the large curvature of the mandrel bend, the large difference between the surface area of the outer and inner surface of the mandrel led to a large variation of the inner and outer braiding angles when the number of yarns was equal. The results of the predicted model and the actual measured braiding angle in equal section part could be within ±3 degrees, while the errors of the bending unequal section part could be within ±5 degrees, the main cause of error in the prediction model is the large effect of yarn interactions on the prediction of the fabric due to the non-circular cross-section of the mandrel, where yarn interactions affected the deposition position of the braiding fall points. The trajectory calculation and yarn prediction model had some errors, but it still served as a guide for actual production and improved production accuracy (Fig. 6 and Fig. 8).

Conclusion The dual robot trajectory solving method proposed in this paper can solve the braiding problem of large size shaped structure mandrels, while the yarn trajectory prediction model can make accurate prediction of shaped structure mandrels, improve braiding efficiency and enhance the mechanical properties of composite materials under the same conditions. Due to the assumption that the mandrel is always braided perpendicular to the braiding plane and that yarn interactions are not taken into account in this paper, the prediction results are subject to some errors on complex mandrels. Therefore, optimization of the braiding position of the mandrel and the addition of yarn interactions to the prediction model are considered in subsequent studies.

Key words: composite material, large size mandrel, circular braiding, braiding angle, take-up trajectory, yarn trajectory prediction