目前三维成型技术被世界各国人所接受，并且这种将数字化模型文件快速地转化成生物材料或光敏树脂等的技术也迅速发展，前景十分广阔。桌面级的三维成型机具有耗材广泛、设备要求低、操作方便等优点。但是三维成型机的硬件主要包括电机、成型喷头、主控器等，其内部结构复杂，电机的稳定运行和温度的精确控制是确保成型物品精度的重要环节。所以，对如何提高三维成型机的成型精度的研究具有重要的意义。 本文对自主研发的三维成型机展开研究，首先，设计出三维成型机的整体系统方案及硬件结构，侧重研究系统的软硬件平台和部分内部资源在三维成型机中的具体应用，同时给出三维成型系统的整体工作链和控制原理。然后重点分析步进电机驱动系统，包括细分驱动技术的概念、特点和原理。除此之外，针对步进电机步进电机失步或跳步、和易产生共振等不足，本文提出电流矢量恒幅均匀旋转的细分驱动方法，该方法采用保持电流合成矢量的幅值不变，均匀地变化旋转的角度的思路，使得电流矢量恒幅且均匀细分驱动。经实验仿真验证，并与常用的细分驱动方法对比，该方法在三维成型系统上的应用是可行性，不仅能使得系统中的五轴电机协调稳定运行，而且成型精度有所提高。 其次，成型喷头温度的高低直接影响熔融耗材的成型实物效果。针对三维成型机的成型喷头温度波动大对成型效果甚至喷头寿命的影响，提出基于QPSO-BP神经网络建立成型喷头温度预测模型，该方法采用取长补短的优化思路，利用量子粒子群优化BP神经网络的初始权值和阈值，来提前预测温度趋势以达到对温度精确控制的目的。通过Matlab仿真证明了应用该算法的控制效果。 最后，在软件设计及实验性测试方面，对三维成型过程中所用的建模软件、切片软件进行分析，并设计操控界面。成型的上下位机进行调试后，在自主研制的平台上实际加工猎人头像实物，对本文提出的两种方法进行实验验证。实验测试及分析表明，优化后比优化前的成型实物的粗糙度降低了0.3mm。本文提出的方法可以为精细加工的成型系统控制提供参考。

At present, the three-dimensional molding technology is accepted by people around the world, and the technologyit of digital model file quickly into biological materials or photosensitive resin is developing rapidly, promising prospects.Desktop level of three-dimensional molding machines have a material widely, low equipment requirement, convenient operation and so on.But the hardware of three-dimensional molding machine, mainly includes the motor, molding nozzle, the host controller and so on, its internal structure is complex, the stable operation of the machine and the accurate control of temperature are important links to ensure the forming accuracy.Therefore, how to improve the forming precision of the three-dimensional molding machine research has the vital significance. In this paper, Independent research and development of the three-dimensional molding machine is researched , first, the whole system and hardware structure of three-dimensional molding machine are designed, system software and hardware platform and part of the internal resources in the application of three-dimensional molding machine are majorly researched,and the whole work of 3d molding system chain and control principle are given.Then analysis of stepping motor driven system, including the concept, characteristics and principle of subdivided driving technology are focused on.In addition, in view of the stepper motor stepper motor step or insufficient leapfrog, and easy to produce resonance, In this paper, Subdivided driving method of current vector of The constant amplitude current vector even rotation is put forward, the method using ideas of maintaining current synthetic vector amplitude constant, changing evenly rotating angle to make the constant amplitude current vector and even subdivided driving.Verified by simulation experiment, and compared with commonly used subdivided driving method, the method in the application of three-dimensional molding system is feasible, and not only can make the system five axis motor coordination and stable operation, but also forming precision is improved. Secondly, molding nozzle temperature directly affects the forming of the molten material physical effect. For three-dimensional molding machine，molding nozzle temperature fluctuations in molding effect even the influence of nozzle lifetime， BP neural network based on quantum particle swarm optimization is put forward to establish molding nozzle temperature prediction model,the method adopting the optimization thinking complement each other, using the quantum particle swarm optimization to the BP neural network's initial weights and thresholds, to predict the temperature trend in order to achieve the purpose of precise control of temperature. Matlab simulation proves that the application of the control effect of the algorithm. Finally, in terms of software design and experimental test, the three-dimensional modeling software and slicing software used in the forming process are analyzed, and the control interface is designed. After forming the upper and lower position machine debugging,in the self-developed platform of actual processing hunter head material, the two methods proposed in this paper is verified by experiment. Test and analysis show that the optimized molding material than before optimization of roughness was reduced by 0.3 mm. The proposed method can provide reference for fine processing of forming system control.