摘要： Additive manufacturing (AM) techniques such as selective laser melting (SLM) have many advantages over traditional manufacturing methods. However, the quality of SLM products is critically dependent on the process parameters, e.g., the laser power, scanning speed, powder layer thickness, hatch space, and scan length. Determining the parameter settings which optimize the product quality is a challenging, but extremely important problem for manufacturers. In a previous study, the present group determined the optimal values of the laser power and scanning speed for 316L stainless steel powder beds. The present study extends this work to investigate the effects of the hatch space and scan length on the melting pool characteristics in a double-scanning track SLM process. A three-dimensional finite element model is constructed to predict the features of the scan track melt pool for various values of the hatch space and scan length. A circle packing design method is then used to select a representative set of hatch space and scan length parameters to train artificial neural networks (ANNs) to predict the melt pool temperature, melt pool depth, and overlap rate between adjacent the trained ANNs are used to create process maps relating the scan track features to the hatch space and scan length. The optimal hatch space and scan length region of the temperature process map is then determined based on a joint consideration of the peak temperature (less than 3300 K), the difference in depth of adjacent melt pools (less than 10 μm), and the overlap rate of adjacent scan tracks (25~35%). The results indicate that the optimal hatch space is equal to 61% of the laser spot size given an SLM system with a laser power of 180 W, a scanning speed of 680 mm/s, a laser spot size of 120 μm, and a 316L SS powder layer thickness of 50 μm.