【作者】 廖天发；

【导师】 薛家祥；

【作者基本信息】 华南理工大学 ， 制造工程智能化检测及仪器， 2016， 博士

【摘要】 铝合金材料以其轻量化、耐磨耐腐蚀、抗冲击性能好、可回收利用等优异的综合性能,越来越多地应用于汽车、船舶以及航空航天等领域,而焊接在铝合金工业生产过程中占有非常重要的地位。铝合金焊接技术的关键点是对电流的精确控制,如果输入电流不适当将导致焊接过程不稳定,焊接接头性能下降等缺陷。论文对铝合金双脉冲熔化极惰性气体保护焊(Metal inert-gas welding,MIG)波形调制方法及工艺机理进行了研究,对促进智能弧焊电源技术、焊接控制策略技术的发展与应用,具有重要的学术价值和实际意义。研究得到了广东省自然科学基金(2016A030313117)资助、广东省省部产学研(2013B090600098、2013B090600030)资助。铝合金双脉冲MIG焊接电源是实现铝合金焊接技术的前提和基础,论文先研究了焊接电源的硬件系统:全桥移相软开关主电路、功率电路参数设计、DSP核心控制电路、模块化控制电路、IGBT驱动电路、送丝电路、人机交互系统等,为MIG焊电流波形调制和工艺机理研究打下基础。在焊接电源硬件电路设计的基础上,建立基于Simulink的系统仿真模型,包括主电路、控制电路、脉冲电流波形和PID控制建模等,并且进行了仿真分析。通过Simulink仿真,验证了硬件电路拓扑结构设计的合理性;优化了焊接电源关键元件的参数,为焊接工艺试验奠定了良好的基础;通过常规PID控制和基于模糊约束满意度函数的遗传优化PID控制的对比分析,验证了遗传优化PID控制具有良好的鲁棒性和自适应性,能够在焊接过程中更精细和准确地控制电流。针对铝合金双脉冲MIG焊焊接参数繁多、调节不便和焊接规范复杂等问题,分析了焊接电源的静特性和动特性,研究了基于脉冲电流变频调节的弧长控制,在对比拉格朗日插值、牛顿插值和三次样条插值等插值算法之后,建立了基于大步距标定和三次样条插值的参数一元化调节专家数据库。在硬件电路设计、Simulink仿真和专家数据库的基础上,成功地研制出铝合金双脉冲MIG焊接电源及其软件系统。通过自行搭建的焊接试验平台,进行了常规的铝合金双脉冲MIG焊焊接工艺性能试验,验证了强弱脉冲个数比率、强弱脉冲群脉冲个数、高频脉冲频率和低频调制频率对焊缝成形及焊接质量的影响。提出了利用神经网络算法对送丝速度进行预测的建模方法,通过给定焊接平均电流、高频脉冲频率和低频调制频率预测焊接过程的送丝速度,并进行了试验验证。试验结果表明,在实际焊接生产过程中,基于神经网络的送丝速度预测模型获得的送丝速度,能有效地匹配其他焊接参数,获得令人满意的焊缝。针对铝合金常规(矩形波)双脉冲MIG焊能量突变的问题,提出了改进型双脉冲MIG焊焊接工艺。在分析梯形波调制和正弦波调制工作原理的基础上,分别进行了梯形波与矩形波调制、正弦波与矩形波调制的双脉冲MIG焊接对比试验,试验结果表明梯形波调制和正弦波调制在铝合金MIG焊上比矩形波调制的焊接过程更稳定、焊缝更美观,验证了改进型双脉冲MIG焊接工艺对铝合金焊接的研究具有重要的参考意义。论文最后介绍了焊接过程电信号的特征信息量:概率密度比L、基于U-I图的周期重复率P和电流样本熵CSaEn,以及各自对焊接质量的定量评定。为了避免单一评定指标的不确定性、评定过程中人的主观随意性以及离线检测的滞后性,提出了一种能在线对焊接质量检测的模糊综合定量评定方法,通过观察各特征信息L、P和CSaEn的变化规律,获取专家经验,得到各自的隶属度函数,总结出模糊逻辑推理规则,根据最大隶属度原则对焊接质量进行定量评定。通过梯形波双脉冲MIG焊获得的焊缝验证,该模糊逻辑综合定量评定系统评定结果的正确率达83.3%,说明了该定量评定系统的有效性。更多还原

【Abstract】 Aluminum alloy is more and more widely applied in many fields, such as automobiles, vessels, aerospace, and so on, because of to its excellent comprehensive performance of light weight, wear-resistance, corrosion resistance, good anti-impact properties, recycleability, and so on. And in the manufacturing of aluminum alloy production, welding plays a very important role. One key point in the welding of aluminum alloy is the precise control of the welding current. If the input current is not appropriate, the welding process would be unstable, and there would be defects in welding points. The waveform modulation method and technological mechanism for the DPMIG welding of aluminum alloy is studied in this paper. This will be academically and practically valuable for the intelligent arc welding power supply technology and welding control strategy. The research work is supported by Guangdong Natural Science Foundation(2016A030313117), and Speacial Project on the Integration of Industry, Education and Research of Guangdong Province(2013B090600098、2013B090600030).The DPMIG welding power is the basis to realize aluminum alloy welding technology, so the hardware system of welding power is firstly studied in this paper, including the main circuit of phase-shifted soft switching, power circuit, DSP control core, modular circuit unit, IGBT driving circuit, wire-feed circuit, man-machine interactive system, and so on. This sets a basis for the subsequent waveform modulation and technological mechanism.Then the system simulation model with Simulink is built on the basis of the hardware design of aluminum alloy welding power, including the main circuit model, control current model, current waveform model and PID control model, as well as the simulation analysis is carried out. In the Simulink simulation, the correctness of the topologies of the hardwre is verified, the key element parameters of welding power is optimizied. In order to control the current more precisely, the conventional PID control and genetic optimized PID control based on the fuzzy function are compared. The results tells that the genetic optimized PID control based on the fuzzy function has good robustness and adaptability.Since that the DPMIG welding parameters of aluminum alloy are diverse and they are inconvenient to adjust. Moreover, the welding specification is complex, the expert database of DPMIG welding is studied based on the analysis for the static and dynamic properties of welding power source. Firstly, the Lagrange interpolation, Newton interpolation and the cubic spline interpolation are compared. Then arc-length control of frequency conversion for the pulse current is studied. Finally the unified adjustive and gradually changing database of parameters based on the large-step calibration and cubic spline interpolation is established.On the base of hardware circuit design, simulation of Simulink software and expert databse, the DPMIG of aluminium alloy and its software system are developed. Then the technological property tests of DPMIG welding are performed on the DPMIG welding power of aluminum alloy on the welding test platform constructed by ourselves. The influences of the ratio of the number of strong pulse to the number of weak pulse, the number of strong and weak pulse groups, high frequency fH, and low frequency fL on the welds, weld forming, and welding quality are verified. Moreover, the neural network algorithm is used for the predictive modeling of wire-feed speed. With all the parameters of DPMIG welding given, the wire-feed speed is predicted by the predictive model, and then the wire-feed speed is verified by the test. The results show that the wire-feed speed prediction model of neural network have better prediction performance and could effectively guide the selection of welding parameters in the process of practical production.In addition, the modified DPMIG welding technology is proposed for the energy mutation in the process of square-wave DPMIG welding of aluminum alloy. The work principles of trapezoid wave and sinusoid wave are analyzed, then the trapezoid wave and sinusoid wave DPMIG welding tests as well as the sinusoid wave and square wave DPMIG welding tests are contrasted, respectively. It is found that the process of trapezoid wave and sinusoid wave aluminum alloy welding are more stable than those of square wave aluminum alloy welding, the welds of trapezoid-wave aluminum alloy welding are more normative. This suggested that the modified DPMIG welding technology has an important reference significance for the study on aluminum alloy welding.Finally, electrical signal feature information of the welding process including the probability density ratio, the periodic repetition rate based on the working points of U-I graph and sample entropy, and their impact on the quality assessment of welding are discussed in this paper. In order to avoid the uncertainty of single evaluation index, as well as in the process of evaluation peoples’ s subjective randomness and delay of offline testing, A online welding quality inspection of fuzzy comprehensive evaluation method is proposed. By observing the feature information P, L and CSaEn to obtain expert experience, then get the membership functions and summarize the fuzzy logic reasoning rules. By the weld verification of TPMIG, the fuzzy logic reasoning evaluation system achieved an accuracy of 83.3%, demonstrating the effectiveness of the quantitative assessment system.更多还原