Material properties are often determined by conducting uniaxial tests on a variety of materials. The precision of the material testing system is critical to the reliability of the mechanical properties of the tested materials. Among the significant factors affecting the performance of the material testing system is its loading coaxiality. Since predicting misalignments involves high border nonlinearity, it is challenging to quantify misalignments. To assess the alignment deviations of a tensile test system., on a thick-flat specimen, a feed-forward neural network was employed for the first time to develop a mapped relation involving misalignments and strain distribution. Additionally, we utilized the genetic algorithm to optimize the essential hyperparameters (learning rate, batch size, and the nodes in the hidden layer) of the created deep feed-forward network. Simulation results indicated that the optimized neural network could reach an accuracy of 99.89% in the prediction of misalignments. Predicted misalignments can measure the test errors associated with the material testing system to avoid incorrect conclusions from being drawn because of systematic errors. The proposed method could coordinate the location of the uniaxial test equipment during manufacturing and assembly.