第2期 ZHAO Shi-gang,et al:Fiber optic strain twirr sensor-array for smart structural health monitoring ·179· where =n-2 0.8 ▣Signals intensity represents the equivalent refractive index of the fi- 0.4 ber core.For silica materials at wavelengthA= 1 550 nm,the parameters are n=1.46.=0.25, pu≈0.l2,pn≈0.27iol,and the equivalent re- g Fiber optic sensor number fractive index can be calculated as nequivalent1.19. This means that the peak shift AXe only de- Fig.3 Normalized signal intensity for rth pends on the changes of the fiber optic sensing sensor in the twim sensor-array gauge length l,and the refractive index due to co- axial strain,and it is independent of the optical 4 Experimental results path variations caused by the elevated tempera- A three twim sensor-array was demonstrated ture.Therefore,the changes due to the environ- in our experiments.In the sensing system,the mental temperature fluctuation of the fiber path LED light source power is 30 uW with drive cur- can automatically be compensated. rent 50 mA,and the insertion losses of moving Thus,the distributed strain can be measured GRIN lens part is in the range of 4 dB to 8 dB as by the gap distance from 3 mm to 70 mm (correspond- g=△X:△X4 (8) ing the optical path change within the range 6 mm nequivalent i to 140 mm).And each individual sensor's gauge 3 Analysis of signals intensity length is about 1 000 mm(1 meter single mode op- tical fiber patchcord). The signals intensity is different for the The PIN-detector output when the value of X schemes shown in Fig.1.The signal intensity from is varied from 15 mm to 21.5 mm is shown in sensor pair j that is due to the coherent mixing be- Fig.4.The four major peaks correspond,respec- tween the reflected signals from the two partial re- tively,to the sensor pair being matched of the flectors that define the sensor may be expressed as three twim sensor-array. Pa=gAx[Π7BJ',9 1.2 0.9 where the 2 X2 coupler is assumed to be 3 dB cou- 0.6 0.3 pler and the insertion losses are neglected.B re- 0 presents the excess loss associated with sensor j A/indino -0.3 -0.6 because connection loss between the sensing seg- -0.9 ments.T,and R,are respectively the transmission -13 0.15 0.651.151.652.14x10 and reflection coefficient of the fth partial reflec- Scanning displacement/gm tor.T,is in general smaller than 1-R,because of the loss factor B.nX)is the loss associated with Fig.4 Three twimfibersensor array experimental the moving GRIN lenses systems and is a function scanning peak signals of Xj. Theoretical simulations were conducted for In order to show the compensation efficiency typical parameters:B =0.9(j=1,2...N+1), of the sensor system,the experiments were per- R,=1 %T=0.89.The average attenuation of formed under the two different circumstances of the moving GRIN lens part is taken as 6 dB,i.e. temperature conditions.The test coupon is depic- X)=1/4.The power coupled into the input fi- ted in Fig.5.The load was supplied from a load ber is Po.The normalized signal intensity for each cell to the test coupon and introduced a uniform sensor in the 10 sensors array is shown in Fig.3. stress field o.Then the corresponding strain will 1994-2008 China Academie Journal Electronic Publishing House.All rights reserved.hup://www.cnki.netwhere nequivalent = n - 1 2 n 3 [ (1 - μ) p12 - μp11 ] (7) represents the equivalent refractive index of t he fi2 ber core. For silica materials at wavelengt h λ= 1 550 nm , t he parameters are n = 1. 46 , μ= 0. 25 , p11≈0. 12 , p12 ≈0. 27 [10 ] , and t he equivalent re2 fractive index can be calculated as nequivalent≈1. 19 . This means t hat t he peak shift ΔXk only de2 pends on t he changes of t he fiber optic sensing gauge lengt h li and t he refractive index due to co2 axial strain , and it is independent of t he optical path variations caused by t he elevated tempera2 t ure. Therefore , t he changes due to the environ2 mental temperat ure fluctuation of the fiber pat h can automatically be compensated. Thus , the distributed strain can be measured by εi = ΔXi - ΔXi- 1 nequivalent li . (8) 3 Analysis of signals intensity The signals intensity is different for the schemes shown in Fig. 1. The signal intensity from sensor pair j t hat is due to t he coherent mixing be2 tween the reflected signals from the two partial re2 flectors that define t he sensor may be expressed as PD1 ( j) = 1 8 P0η( Xj) Rj+1 ∏ j i =1 ( Tβi i) 2 , (9) where t he 2 ×2 coupler is assumed to be 3 dB cou2 pler and t he insertion losses are neglected. βj re2 presents t he excess loss associated with sensor j because connection loss between t he sensing seg2 ments. Tj and Rj are respectively t he transmission and reflection coefficient of the j2th partial reflec2 tor. Tj is in general smaller t han 1 - Rj because of t he loss factorβj . η( Xj) is the loss associated wit h t he moving GRIN lenses systems and is a f unction of Xj . Theoretical simulations were conducted for typical parameters:βj = 0. 9 ( j = 1 , 2 , …, N + 1) , Rj = 1 % , Tj = 0. 89. The average attenuation of t he moving GRIN lens part is taken as 6 dB , i. e. η( Xj) = 1/ 4. The power coupled into t he inp ut fi2 ber is P0 . The normalized signal intensity for each sensor in t he 10 sensors array is shown in Fig. 3. Fig. 3 Normalized signal intensity for i2th sensor in the twin2sensor2array. 4 Experimental results A t hree twin2sensor2array was demonstrated in our experiments. In t he sensing system , t he L ED light source power is 30 μW with drive cur2 rent 50 mA , and t he insertion losses of moving GRIN lens part is in t he range of 4 dB to 8 dB as t he gap distance from 3 mm to 70 mm (correspond2 ing t he optical pat h change wit hin the range 6 mm to 140 mm) . And each individual sensor’s gauge lengt h is about 1 000 mm (1 meter single mode op2 tical fiber patchcord) . The PIN2detector outp ut when t he value of X is varied from 15 mm to 21. 5 mm is shown in Fig. 4. The four major peaks correspond , respec2 tively , to the sensor pair being matched of t he t hree twin2sensor2array. Fig. 4 Three twin2fiber2sensor array experimental scanning peak signals In order to show t he compensation efficiency of t he sensor system , t he experiments were per2 formed under t he two different circumstances of temperat ure conditions. The test coupon is depic2 ted in Fig. 5. The load was supplied from a load cell to the test coupon and introduced a uniform stress field σ. Then t he corresponding strain will 第 2 期 ZHAO Shi2gang ,et al :Fiber optic strain twin2sensor2array for smart structural health monitoring · 971 · © 1994-2008 China Academic Journal Electronic Publishing House. All rights reserved. http://www.cnki.net