针对线扫描视觉检测系统精度易受机械结构误差影响且具体影响机制不明确的问题，建立并分析了机械误差对系统成像误差影响的数学模型。基于多体运动学与齐次坐标变换理论推导了线扫描视觉检测系统机械系统误差传递模型，并结合线扫描成像特点建立了系统误差综合模型，阐明了机械误差与系统输出图像误差间的对应关系。利用多元函数全微分方法对模型进行了误差灵敏度分析，明确了对系统输出图像x、y两个维度误差影响显著的误差源。针对实际线扫描视觉检测系统进行了误差源验证实验，实验结果表明：所建立的系统误差综合模型可以准确识别出对线扫描视觉检测系统输出图像影响最大的关键误差源；模型对于关键误差源在不同位置灵敏度数值预测与实际偏差不超过2.38%，可以实现对系统关键误差源灵敏度的准确预测。

To address the problem that the accuracy of the line scan vision detection system is easily affected by the mechanical structure error and the specific influence mechanism is not clear, a mathematical model of the influence of mechanical error on the system imaging error was established and analyzed. Based on the theories of multi-system kinematics and homogeneous coordinate transformation, a mechanical system error transfer model of the line scan vision detection system was derived, and a system error comprehensive model was established with reference to the line scan imaging characteristics to clarify the correspondence between mechanical errors and system image output errors. The error sensitivity of the model was analyzed based on the complete differential-coefficient theory, and the error sources that had a great impact on the errors of the x andy dimensions of the output image were clarified. An experiment for verifying error sources is carried out and the result shows that the established system error comprehensive model can accurately identify the key error sources that have the greatest influence on the output image. The deviation between the numerical sensitivity prediction by the model and the actual value does not exceed 2.38%, which can achieve the accurate sensitivity prediction of the key error sources.