-
基于嵌入式实时操作系统(Operating System, OS)μC/OS-III开发了主控ARM驱动层的程序。相较于传统的轮询系统和前后台系统,该设计具有更加优异的实时性和扩展性。驱动层程序主要完成如下操作:激光驱动器和TEC驱动器输出的激活与关闭,电流和温度的设置、输出与校正,LD状态检测及报警,输出限幅检测,零点飘移校正,硬件自检,485 (MODBUS-RTU协议)通讯,串口通讯,LCD显示,FLASH读写,CPU和内存占用率监测等;FPGA作为协处理器,程序采用Verilog语言编写,驱动模数转换器采集当前数据,通过串口发送给ARM处理。
-
遗传算法衍生自达尔文进化论中的“物竞天择,适者生存”法则,通过模拟遗传学机制和自然选择的过程寻找全局最优解,具有自适应性和群体搜索能力,并有避免需要优化参数在搜索期间因为落入局部某个单峰的极值点而将之作为最优解的优势[15-17]。图3(a)为基本遗传算法最优解求解过程。
传统PID控制起源于20世纪初,通过对设定值和实际值的偏差进行比例、积分、微分线性组合运算,输出反馈结果到被控对象使其达到动态平衡[18]。PID控制具有简洁、鲁棒性好、易实现等优点,对于大致呈线性、动态特性不随时间变化的系统具有优良的控制效果[19]。图3(b)为传统PID控制系统工作原理图,其位置式离散数学模型为:
$$ \begin{split} {{OUT}} = &({K_p} \times {E_k}) + \left({K_i} \times \sum\limits_{k = 0}^n {{E_k}} \right) + \\ &[{K_d} \times ({E_k} - {E_{k - 1}})] \end{split} $$ (1) 式中:
${K_p}$ 、${K_i}$ 、${K_d}$ 分别为比例、积分、微分系数;${E_k}$ 、${E_{k - 1}}$ 为本次和上次偏差;$OUT$ 为输出结果。但在嵌入式系统中应用时,该模型存在计算量大、占用内存多的缺点。为此,采用增量式PID算法,记上次的输出控制信号值为$OU{T_{k - 1}}$ ,当前输出值为$OU{T_k}$ ,两者相减得到增量式PID输出$\Delta OUT$ 的数学模型:$$ \begin{split} \Delta OUT =& OU{T_k} - OU{T_{k - 1}} = {K_p} \times ({E_k} - {E_{k - 1}}) + \\ &{K_i} \times {E_k} + {K_d} \times ({E_k} - 2{E_{k - 1}} + {E_{k - 2}}) \end{split} $$ (2) 式中:
${E_{k - 2}}$ 为上上次偏差。增量式PID算法中没有对历史偏差的累加,极大地弥补上述不足之处。但无论位置式还是增量式PID算法,对${K_p}$ 、${K_i}$ 、${K_d}$ 三个变量的整定是控制系统的关键,整定方式一般选取经验数值或者试凑数值,由于三个参数并非独立存在,经验法或试凑法往往需要耗费大量人力及时间,且在被控对象为参数时变或非线性时传统PID控制器经常存在控制效果不稳定的情况[20]。针对上述问题,文中结合遗传算法实现智能PID控制算法,更精准地进行电流和温度的校正。首先,参照PID参数整定的经验值,将比例、积分、微分三个参数的取值区间确定为:
${K_p} \in [0,1]$ ,${K_i} \in [0,0.1]$ ,${K_d} \in $ $ [0,5]$ ,依据参数区间初始化种群,并根据处理器性能将种群数量设置为合适值。评价PID算法控制效果的主要性能指标为超调量$\sigma $ 、上升时间$ {{t}}_{{r}} $ 及其稳态误差函数,为平衡系统的动态和稳态性能指标,设定适应度函数如下:$$\begin{split} \\ J = (1 - \phi ) \times \sigma + \phi \times {t_r} + \int_0^\infty {t\left| {e(t)} \right|{\rm{d}}t} \end{split} $$ (3) 式中:
$\phi $ 为加权值,取值区间为(0,1);遗传函数$ {{J}} $ 的值越小代表个体适应能力越强。将个体适应度从大到小依次排序,淘汰概率设置为50%,未被淘汰的个体进行概率为$\alpha $ 的随机交叉,由个体p、q产生新个体y的算法为:$$\begin{split} \\ \left[ {\begin{array}{*{20}{c}} {{K_{py}}} \\ {{K_{iy}}} \\ {{K_{dy}}} \end{array}} \right] = (1 - \alpha ) \times \left[ {\begin{array}{*{20}{c}} {{K_{pp}}} \\ {{K_{ip}}} \\ {{K_{dp}}} \end{array}} \right] + \alpha \times \left[ {\begin{array}{*{20}{c}} {{K_{pq}}} \\ {{K_{iq}}} \\ {{K_{dq}}} \end{array}} \right] \end{split}$$ (4) 为防止参数整定过程在早期就因陷入局部最优解而终止,设置变异概率为
$\;\beta $ ,对个体y有:$$\left[ {\begin{array}{*{20}{c}} {{K^{'}}_{py}} \\ {{K^{'}}_{iy}} \\ {{K^{'}}_{dy}} \end{array}} \right] = \left[ {\begin{array}{*{20}{c}} {{K_{py}}} \\ {{K_{iy}}} \\ {{K_{dy}}} \end{array}} \right] + (1 - \beta ) \times \left[ {\begin{array}{*{20}{c}} {{K_{py}}} \\ {{K_{iy}}} \\ {{K_{dy}}} \end{array}} \right]$$ (5) 完成选择交叉变异操作之后,再次计算个体的适应度。当进化的次数达到最大迭代次数或整定参数满足条件,整定寻优过程停止,否则继续进行选择交叉变异操作,直至得到全局最优参数
${K_p}$ 、${K_i}$ 、${K_d}$ 并存储,下次开机可直接调用,以节约前期迭代的时间。图3(c)为遗传算法优化的增量式PID控制系统结构图。 -
为验证研制的驱动系统性能,选取两个额定功率500 mW、中心波长980 nm的LD (Oclaro,型号LC96A74P)和三个额定功率650 mW、中心波长1480 nm的LD (Anritsu,型号AF4B),分别编号LD1~LD5。在室温环境下,将LD1电流设置为80 mA,LD2电流设置为200 mA,LD3~LD5电流均设置为1200 mA,所有LD温度均设定为25 ℃。使用六位半数字万用表(Agilent Technologies,型号34410A)测试五路LD的电流值,同时记录温度数据,每5 min一次,持续记录300 min,测试结果如图4所示。
根据稳定度计算公式[21]:
$${\rm{stability}}\; {\rm{value}} = \frac{{{\rm{standard}}\; {\rm{deviation}}}}{{{\rm{average}}\; {\rm{value}}}} \times 100\% $$ (6) 将测得数据代入公式(6),五路LD的输入电流稳定度分别为0.001%,0.0009%,0.0005%,0.0005%,0.0006%,温度稳定度分别为0.032%,0.031%,0.034%,0.033%,0.035%。
-
驱动系统是激光泵浦源的核心,为验证驱动系统稳定性,将文中设计的驱动系统集成到中心波长为1.5 μm的飞秒光纤激光器中驱动五路LD作为泵浦源,实验装置示意图如图5所示。
泵浦源采用波长为980 nm和1480 nm的LD,自左至右依次为LD1~LD5,其中LD1和LD2波长为980 nm,LD3~LD5波长为1480 nm。经由波分复用器(Wavelength Division Multiplexer, WDM)将泵浦光与信号光耦合进入谐振腔内,掺铒增益光纤(Erbium-Doped Fiber, EDF)作为激光器中的工作物质,通过吸收泵浦光的能量,在谐振腔内产生1.5 μm波段的种子激光。腔内加入光隔离器(Isolator, ISO)确保激光的运转方向,规避产生空间烧孔效应。加入碳纳米管可饱和吸收体器件(Carbon Nanotube Saturable Absorber, CNT SA)在激光腔内对入射激光起到调制作用,脉冲激光通过一个分束比为10∶90的输出耦合器(Output Coupler, OC)将10%的激光输出进入后级激光放大器,90%的激光反馈回腔内用于铒离子的受激辐射。种子源激光器输出的脉冲激光进入一级掺铒光纤放大器进行放大,由中心波长为980 nm的LD泵浦,一级放大器输出的脉冲激光进入二级掺铒光纤放大器进行放大。二级放大器由三个中心波长为1480 nm的LD泵浦。在一、二级放大器之间接入色散补偿光纤(Dispersion Compensation Fiber, DCF),以防止脉冲激光峰值功率增加太快而损坏仪器,或者过早出现非线性效应而影响对脉冲激光的功率放大。使用光谱仪(YOKOGAWA,型号AQ6375)、数字示波器(Tektronix,型号DPO70604C)、自相关仪(Aveata,型号AA-10DD)、频谱仪(Agilent,型号E4411B)和光功率计(Thorlabs,型号PM100D)测试激光器输出的光信号[22-24]。
在1.5 μm光纤激光器中,种子泵浦源的工作温度设置为25 ℃,提高LD1驱动电流至60 mA,激光器输出连续激光,继续提高LD1驱动电流至65 mA,激光器实现自启动锁模脉冲激光输出。测得种子源输出光谱如图6(a)所示,其3 dB带宽为2.2 nm,种子源输出的激光脉冲序列如图6(b)所示,可以看出种子激光器具有稳定的脉冲输出,其相邻脉冲时间间隔为23.2 ns,对应重复频率为43.1 MHz。种子源输出的激光单脉冲如图6(c)所示,脉冲宽度为100 fs。保持激光器持续工作,测得其射频谱如图6(d)所示,其基频位于43.47 MHz处,信噪比为50 dB,其中插图为大扫描范围(1000 MHz)下的输出激光射频谱图,可以看到锁模激光有稳定的脉冲输出。在测试长期稳定性时,每间隔36 min进行一次光谱采样,持续3 h,如图6(e)所示。测试结果表明,锁模激光输出始终保持稳定,说明驱动系统工作稳定性良好。
设置一级放大器的泵浦源LD2驱动电流为200 mA,二级放大器泵浦源LD3~LD5电流均为1200 mA,LD2~LD5工作温度均设置为25 ℃,测得一级放大器的输出光谱如图7(a)所示,其3 dB带宽为3.1 nm;二级放大器的输出光谱如图7(a)所示,在对超短脉冲激光进行放大时,信号光经一级掺铒光纤放大器后进入一段色散补偿光纤,由于自相位调制效应导致光谱变宽,但其中心波长依旧为1.5 μm。二级放大器的输出激光的脉冲序列如图7(b)所示,可以看出其输出脉冲稳定,间隔为22.9 ns,对应重复频率为43.67 MHz,其输出单脉冲如图7(c)所示,由于脉冲激光峰值功率的增加,其宽度被压缩为98.0 fs。监测泵浦源的注入电流与1.5 μm光纤激光器的输出光功率,得到其I-P关系如图7(d)所示。根据相关系数R的计算公式:
$$R = \frac{{\displaystyle\sum\limits_{i = 1}^N {{{({X_i} - \overline X)}^2}{{({Y_i} - \overline Y)}^2}} }}{{\sqrt {\displaystyle\sum\limits_{i = 1}^N {{{({X_i} - \overline X)}^2}} } \sqrt {\displaystyle\sum\limits_{i = 1}^N {{{({Y_i} - \overline Y)}^2}} } }}$$ (7) 式中:
$({X_i},{Y_i})$ 代表驱动电流与光功率的第i个数据点;$\overline X$ 和$\overline Y$ 为数据的平均值。当N=12时,得到R=0.997。由此可知,在该实验的注入电流与输出光功率范围内两者成线性关系[25],说明驱动系统温控效果良好。每隔5 min记录一次激光器的输出功率,持续300 min,测试结果如图7(e)所示。根据公式(6)计算得到功率稳定度为0.16%,说明驱动系统输出电流和泵浦源工作温度稳定度均良好。
Design and application of driving system for pump source of ultrashort pulsed laser
-
摘要: 设计并实现了一种用于超短脉冲激光器泵浦源的驱动系统。该系统可以同时为五路半导体激光器提供高精度、高稳定度、高灵活性的恒温控制和恒流驱动。为了提高驱动系统的灵活性和可集成度,硬件部分采用上位机、控制、驱动、供电分开的模块化设计方案;基于嵌入式实时操作系统μC/OS-III开发了系统驱动程序,提高了实时性和扩展性,输出校正部分采用遗传算法优化的增量式PID算法,减少了系统的超调量和到达动态平衡的时间;驱动系统具有完备的保护措施,如软启动和关闭、驱动回路检测、过温保护等功能,确保了泵浦源的稳定运行。在实际应用中,半导体激光器温度稳定度优于0.035%,驱动系统输出电流稳定度优于0.001%。将研制的驱动系统集成到1.5 μm飞秒光纤激光器中驱动五路半导体激光泵浦源,获得的种子激光、放大器的输出激光脉冲光谱、脉冲序列和单脉冲均可稳定运行;经过连续3 h的测试,1.5 μm飞秒光纤激光器的输出功率稳定度为0.16%。Abstract: A driving system for the pump source of the ultrashort pulse laser was developed, which could provide high precision, high stability and high flexibility of constant-temperature control and constant-current drive for the five semiconductor lasers simultaneously. In order to improve the flexibility and integration of the driving system, the hardware part adopted the modular design scheme that the host computer, control board, driving board and power supply module were separated. The driving system program was developed based on embedded real-time operating system μC/OS-III, which improved the real-time performance and expansibility of the system program. In the output correction part of the system, incremental PID algorithm optimized by genetic algorithm was adopted, the overshoot and the time to establish dynamic equilibrium of the system were reduced. The driving system had complete protection measures, such as soft-start and soft-shutdown, driving loop monitoring, overtemperature protection and other functions, to ensure the stable operation of the pump source. In practical application, the temperature stability of the semiconductor laser is better than 0.035%, and the output current stability of the driving system is better than 0.001%. The developed driving system is integrated into a 1.5 μm femtosecond fiber laser, the driving system drives five semiconductor lasers as the pump source, the emission spectrum, output pulse trains and single pulse profile of the seed laser and the laser amplifier are stable, The output power stability of the 1.5 μm femtosecond fiber laser is 0.16% after three hours of continuous testing.
-
[1] Ren Jun, Wu Sida, Cheng Zhaochen, et al. Mode-locked femtosecond erbium-doped fiber laser based on graphene oxide versus semiconductor satura absorber mirror [J]. Chinese Journal of Lasers, 2015, 42(6): 0602013. (in Chinese) [2] Ma Xiaoyu, Zhang Naling, Zhong Li, et al. Research progress of high power semiconductor laser pump source [J]. High Power Laser and Particle Beams, 2020, 32(12): 120-129. (in Chinese) [3] Yang D C, Chu S L, Wang Y F, et al. Frequency upconverted amplified spontaneous emission and lasing from inorganic perovskite under simultaneous six-photon absorption [J]. Optics Letters, 2018, 43(9): 2066. doi: 10.1364/OL.43.002066 [4] Li Jinyou, Wang Hailong, Yang Jin, et al. Voltage-temperature characteristics of InGaAs/GaAs/InGaP quantum well laser [J]. Chinese Journal of Luminescence, 2020, 41(8): 971-976. (in Chinese) doi: 10.37188/fgxb20204108.0971 [5] Zhang Peng. Analysis and study of the physical characteristics of a semiconductor laser [J]. Laser Journal, 2018, 39(12): 106-110. (in Chinese) [6] Pongrac B, Onlagic D, Njegovec M, et al. THz signal generator using a single DFB laser diode and the unbalanced optical fiber interferometer [J]. Sensors, 2020, 20(17): 4862. doi: 10.3390/s20174862 [7] Dong Ningning, Cui Jinjiang, Xu Jiangen, et al. Design of control system for 1470-nm high-power semiconductor laser lipolysis device [J]. Optics and Precision Engineering, 2018, 26(8): 1896-1903. (in Chinese) doi: 10.3788/OPE.20182608.1896 [8] Lin Xingchen, Zhang Yawei, Zhu Hongbo, et al. 10 kW CW diode laser cladding source and thermal effect [J]. Chinese Optics, 2019, 12(4): 820-825. (in Chinese) doi: 10.3788/co.20191204.0820 [9] Liu Xu, Wei Jingsong, Tan Chaoyong, et al. Theoretical analysis of multi-wavelength temperature-free-control pump source of laser [J]. Infrared and Laser Engineering, 2016, 45(5): 0505004. (in Chinese) [10] Cong Menglong, Li Li, Cui Yangsong, et al. Design of high stability digital control driving system for semiconductor laser [J]. Optics and Precision Engineering, 2010, 18(7): 1629-1635. (in Chinese) [11] Yan Wanhong, Zhou Yanwen, Yu Di, et al. Temperature control system of semiconductor device and application for infrared gas detection [J]. Acta Photonica Sinica, 2019, 48(3): 0312002. (in Chinese) [12] 我国激光技术与应用2035发展战略研究[J]. 中国工程科学, 2020, 22(3): 1-6. Research Group of Strategic Research on China's Laser Technology and Its Application by 2035. Strategic research on China's laser technology and its application by 2035[J]. Strategic Study of CAE, 2020, 22(3): 1-6. (in Chinese) [13] Li Xudong, Mei Feng, Yan Renpeng, et al. Review of burst-mode lasers for high-speed PLIF imaging diagnostics [J]. Optics and Precision Engineering, 2019, 27(10): 2116-2126. (in Chinese) doi: 10.3788/OPE.20192710.2116 [14] Quan Wei, Li Guanghui, Chen Xi, et al. Structural design and ANSYS thermal simulation for semiconductor laser system [J]. Optics and Precision Engineering, 2016, 24(5): 1080-1086. (in Chinese) doi: 10.3788/OPE.20162405.1080 [15] Wei Xiaochao, Ni Xiangdong, ZhaoXin, et al. Feedforward-feedback and PID control of hydraulic speed regulation system based on genetic algorithm [J]. Chinese Hydraulics & Pneumatics, 2020(11): 21-26. (in Chinese) doi: 10.11832/j.issn.1000-4858.2020.11.001 [16] Cuéllar M P, Gómez-Torrecillas J, Lobillo F J, et al. Genetic algorithms with permutation-based representation for computing the distance of linear codes [J]. Swarm and Evolutionary Computation, 2021, 60(6): 100797. [17] Ramos-Figueroa O, Quiroz-Castellanos M, Mezura-Montes E, et al. Variation operators for grouping genetic algorithms: A review [J]. Swarm and Evolutionary Computation, 2021, 60(6): 100796. [18] Hosseini S A, Shirani A S, Lotfi M, et al. Design and application of supervisory control based on neural network PID controllers for pressurizer system [J]. Progress in Nuclear Energy, 2020, 130: 103570. doi: 10.1016/j.pnucene.2020.103570 [19] Zhou Jinglong, Chen Jun, Song Laijian. PID control method of calender temperature control system optimized by genetic algorithm [J]. Plastics Science and Technology, 2020, 48(10): 115-118. (in Chinese) [20] 孟卓. 基于神经网络PID的水泥回转窑温度控制研究[D]. 西安科技大学, 2020. Meng Zhuo. Research on temperature control of cement rotary kiln based on neural network PID[D]. Xi'an: Xi'an University of Science and Technology, 2020. (in Chinese) [21] Dai Yuanyuan, Song Limin, Zhong Haiwen, et al. Design of control system of portable dual-wavelength laser therapeutic instrument [J]. Laser Journal, 2020, 41(11): 144-148. (in Chinese) [22] 康喆. 基于金纳米棒可饱和吸收体的锁模光纤激光器及其应用研究[D]. 吉林大学, 2015. Kang Zhe. Study on mode-locked fiber lasers based on gold nanorods saturable absorbers and their applications[D]. Changchun: Jilin University, 2015. (in Chinese) [23] 李楠. 2 μm波段宽调谐超短脉冲激光及其应用研究[D]. 吉林大学, 2017. Li Nan. Study on widely tunable ultrashort pulse fiber laser around 2 μm and its application[D]. Changchun: Jilin University, 2017. (in Chinese) [24] 刘嘉兴. 基于金纳米棒/D形光纤可饱和吸收体的飞秒光纤激光器[D]. 吉林大学, 2020. Liu Jiaxing. Femtosecond fiber lasers based on gold nanorods/D-shaped fiber saturable absorber[D]. Changchun: Jilin University, 2020. (in Chinese) [25] Tian Xiaojian, Shang Zuguo, Gao Bo, et al. Control system for 980 nm high stability laser pump source [J]. Optics and Precision Engineering, 2015, 23(4): 982-987. (in Chinese) doi: 10.3788/OPE.20152304.0982 [26] Zhou Zhen, Qi Zhongliang, Qin Yong. Design of driving method for low power semiconductor laser [J]. Infrared and Laser Engineering, 2012, 31(10): 135-139. (in Chinese)