Apparatus for measuring optical signals from multiple optical fiber sensors

Abstract

There is described a sensor apparatus. It comprises an interrogator comprising a light source emitting pulses having a wavelength about an average wavelength; and a fiber Bragg grating (FBG) arrangement. The arrangement comprises a FBG sensor array comprising a plurality of FBG sensors on an optical fiber and being for reflecting the pulses, thereby producing reflected pulses at each one of the FBG sensors. FBG sensors of a given FBG sensor array have a spatial separation therebetween which is sufficient to allow, at a receiver, a temporal discrimination between the reflected pulses produced by each one of the FBG sensors. The FBG sensor array has a spectral reflection window which comprises the average wavelength.

Claims

1. A sensor apparatus comprising: an interrogator comprising light sources, each one of the light sources emitting pulses having either a first respective wavelength or a second respective wavelength about a respective average wavelength of the each one of the light sources; and FBG sensor arrays, each one of the FBG sensor arrays corresponding to one of the light sources and comprising a plurality of FBG sensors on an optical fiber and being for reflecting the pulses, wherein FBG sensors of a given FBG sensor array have a spatial separation therebetween which is sufficient to allow, at a receiver, a temporal discrimination between reflected pulses produced by each one of the FBG sensors of a given FBG sensor array; wherein each one of the FBG sensor arrays has a respective spectral reflection window having a reflection spectrum with sides, the respective spectral reflection window comprising the respective average wavelength of the corresponding one of the light sources, wherein the first respective wavelength and the second respective wavelength have a difference substantially smaller than the spectral width of the FBG sensors that allows both the first respective wavelength and the second respective wavelength to be on a same side of the reflection spectrum; wherein the respective average wavelengths of the light sources have a spectral separation therebetween which is sufficient to allow, at the receiver, a spectral discrimination between reflected pulses from each one of the FBG sensor arrays.

2. The sensor apparatus of claim 1, wherein the receiver comprises a processor adapted, for each one of the FBG sensor arrays corresponding to one of the light sources and based on prior knowledge of a reference reflection spectrum of the each one of the FBG sensors, to use the reflected pulses from the first respective wavelength and the second respective wavelength to unambiguously determine a shift of the peak of an actual reflection spectrum of the each one of the FBG sensors.

3. The sensor apparatus of claim 1, wherein the FBG sensor arrays are provided on a plurality of optical fibers, each one of the optical fibers holding a given number of FBG sensor arrays.

4. The sensor apparatus of claim 3, further comprising a multiplexer for connecting the plurality of optical fibers thereto, the interrogator being optically coupled to the multiplexer for sending the pulses to the plurality of optical fibers.

5. A sensor apparatus comprising: an interrogator comprising a light source emitting pulses having either a first respective wavelength and a second respective wavelength about an average wavelength; and a fiber Bragg grating (FBG) arrangement comprising a FBG sensor array comprising a plurality of FBG sensors on an optical fiber and being for reflecting the pulses, thereby producing reflected pulses at each one of the FBG sensors, wherein FBG sensors of a given FBG sensor array have a spatial separation therebetween which is sufficient to allow, at a receiver, a temporal discrimination between the reflected pulses produced by each one of the FBG sensors; wherein the FBG sensor array has a spectral reflection window having a reflection spectrum with sides, the respective spectral reflection window comprising the average wavelength, wherein the first respective wavelength and the second respective wavelength have a difference substantially smaller than the spectral width of the FBG sensors that allows both the first respective wavelength and the second respective wavelength to be on a same side of the reflection spectrum.

6. The sensor apparatus of claim 5, wherein the receiver comprises a processor adapted, based on prior knowledge of a reference reflection spectrum of the plurality of FBG sensors, to use the reflected pulses from the first respective wavelength and the second respective wavelength to unambiguously determine an actual reflection spectrum of each one of the plurality of FBG sensors.

7. The sensor apparatus of claim 1, wherein the interrogator is a diode laser.

8. The sensor apparatus of claim 5, wherein the interrogator is a diode laser.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) Further features and advantages of the present disclosure will become apparent from the following detailed description, taken in combination with the appended drawings, in which:

(2) FIG. 1A is a diagram illustrating the apparatus comprising a single module interrogating a serial array of FBG sensors located along a single optical fiber, according to an embodiment;

(3) FIG. 1B is a diagram illustrating the apparatus comprising a single module interrogating a parallel array of FBG sensors distributed along a plurality of optical fibers, according to an embodiment;

(4) FIG. 2A is a diagram illustrating the apparatus comprising multiple modules interrogating multiple serial arrays of FBG sensors located along multiple optical fibers, according to an embodiment;

(5) FIG. 2B is a diagram illustrating the apparatus comprising multiple modules interrogating multiple parallel arrays of FBG sensors located along multiple optical fibers, according to an embodiment;

(6) FIGS. 3A to 3C are graphs illustrating the intensity of emitted and reflected pulses with respect to time, according to an embodiment;

(7) FIGS. 4A-4C are graphs illustrating emitted and reflected pulses intensity with respect to time, according to another embodiment;

(8) FIGS. 5A and 5B are graphs illustrating examples of a reflection spectrum of a FBG; and

(9) FIGS. 6A-6C are diagrams illustrating the apparatus comprising a photodiode for monitoring the emitted pulses, according to various embodiments.

(10) FIGS. 7a-7C are diagrams illustrating various embodiments of FBG array configurations; and

(11) FIG. 8 is a diagram illustrating how add/drop multiplexers can be used to launch signals from one or more modules into separate branches, according to an embodiment.

(12) It will be noted that throughout the appended drawings, like features are identified by like reference numerals.

DETAILED DESCRIPTION

(13) This disclosure pertains to an apparatus for measuring multiple FBG sensors using a combination of WDM and TDM techniques.

(14) The present description describes embodiments which are based on a modular system, where each module is made of very low cost, standard components. Thus the user can build an instrument at a cost that is proportional to the number of sensors required. Furthermore, modularity coupled with standardization of the components allows for a reduction of manufacturing costs.

(15) There is described herein a technique to measure the central wavelength of the FBG sensors that is fundamentally different from known prior art techniques. Whereas most of the prior art aims to find the wavelength of peak reflectivity of the grating, this description uses the knowledge of the FBG spectral profile to determine the location of the FBG central wavelength relative to the fixed wavelength of the interrogating source.

(16) Now referring to FIG. 1A, there is shown an apparatus 1 which comprises a plurality of FBG sensors 50 located along one single optical fiber 30, or along multiple optical fibers 30 branching out from a main fiber (see FIGS. 1B, 2A and 2B), the plurality of FBG sensors 50 being comprised in one or multiple sensor arrays 55. Each sensor array 55 is composed of a number N of FBG sensors 50 operating within the same spectral window, separated by regular or irregular but known distances from the source along one optical fiber 30, as shown in FIG. 1A. Each sensor array 55 operates within a distinct spectral window. Each FBG sensor 50 has a known and well characterized reflection spectrum. Since all FBG sensors 50 from an array 55 are on the same optical fiber 30, the array 55 can be defined as a serial array, which is different from a parallel array, as described below.

(17) According to another embodiment shown in FIG. 1B, each sensor array 55 comprises FBG sensors 50 distributed on a plurality of optical fibers 30, defining a parallel array. Each parallel array is composed of a number N of FBG sensors (N=4 in FIG. 1B) located on separate optical fibers branching out from a single optical fiber, and located at incremental distances from the source. This increment is either regular, or irregular but known. The same principle applies in FIG. 2B, in which a plurality of modules is used, as described further below.

(18) The apparatus 1 further comprises: an interrogator 12 comprising one or multiple individual modules 10, each module 10 interrogating one of the sensor arrays 55 of FBG sensors 50 operating in one spectral window; an optical wavelength division multiplexer 60 combining the outputs of the multiple modules 10 into a single optical fiber 30, if more than one module 10 is used (see FIG. 2); optional spectral add/drop multiplexers 65 that can separate the outputs of the multiple modules 10 from the main optical fiber 30 into separate branches of optical fiber, either one channel at a time, or multiple channels at a time (see FIG. 2); and one or more optional reference FBG sensor (not shown) in each sensor array 55, such a reference FBG sensor being at known strain and temperature.

(19) The basic unit of the apparatus 1 consists of a single module 10, interrogating a sensor array 55 of N FBG sensors 50 operating in one spectral window, located along a single optical fiber 30, as illustrated in FIG. 1A. Alternative configurations for the embodiment that uses a reference photodiode, as will be described below, are shown in FIGS. 6A-6C. Multiple modules 10 can be used concurrently by multiplexing their output using a wavelength division multiplexer 60, as shown in FIG. 2.

(20) Each module 10 consists of a light source 11 with narrow spectral width (typically less than 1 GHz) made to repeatedly emit pulses of duration τ.sub.p, such that τ.sub.p is shorter than the minimum travel time of light τ between consecutive FBG sensors 50 in the sensor array 55. The central wavelength of the emitted pulses is made to alternate between a high value λ.sub.+ and a low value λ.sub.−, the difference between λ.sub.+ and λ.sub.− being much smaller than the spectral width of the FBG sensors 50 (so that pulses of wavelength λ.sub.+ or λ.sub.− can both be reflected by a FBG sensor), but typically larger than the spectral width of each pulse (so that pulses of wavelength λ.sub.+ or λ.sub.− can be spectrally distinguished). Thus one can also define an average wavelength λ.sub.av=(λ.sub.++λ.sub.−)/2. Both λ.sub.+, λ.sub.− and λ.sub.av should remain as stable as possible.

(21) The pattern of alternating λ.sub.+ and λ.sub.− pulses can be implemented in different ways. In one implementation, shown in the graph of FIG. 3A, pulses are emitted at time intervals T, such that T is larger than the total travel time of light between the first and last FBG sensor in the array, and alternate between λ.sub.+ and λ.sub.−.

(22) In another implementation (not shown), bursts of a number of pulses at λ.sub.+, emitted at intervals T, are followed by bursts of a number of pulses at λ.sub.−.

(23) In a third possible implementation, shown in the graph of FIG. 4A, two pulses are emitted within the time interval τ, one at λ.sub.+ followed by one at λ.sub.−. For the third implementation, τ.sub.p has to be smaller than τ/2. The two pulses can be distinct, but can also be merged into a single pulse having average wavelengths λ.sub.+ and λ.sub.− in its first and second half, respectively (i.e., the initial wavelength starts close to λ.sub.+ and ends close to λ.sub.−). For best results, the time between λ.sub.+ and λ.sub.− pulses or bursts of pulses should be much shorter than the time it takes for the measurand to change by a meaningful value.

(24) The light pulses are launched into an optical fiber 30. According to an embodiment, the light source 11 is a fiber coupled, temperature-stabilized DFB laser diode (LD in FIG. 1), driven by current pulses of alternating high and low peak current. Since it is known that the central wavelength of light emitted by such a DFB laser diode is related to the driving current, pulses with different peak currents will therefore have a different central wavelength. The DFB laser diode assembly may generally include an optical isolator to block the reflected light from getting back into the laser diode, and an internal photodiode that monitors the emitted light.

(25) Following the fiber-coupled source, an optical fiber coupler 15 or an optical circulator is inserted. It allows the reflected light to be redirected onto a photodetector 20 (PD in FIG. 1A). Various other optical fiber components (C used as a generic symbol in FIG. 1) may be inserted between the optical fiber coupler 15 and the sensor array 55, such as the wavelength division multiplexer 60, connectors, fiber optic rotary joints 70, add/drop multiplexers 65 (shown in FIG. 2).

(26) If an optical fiber coupler 15 is used instead of a circulator, the second branch of the optical fiber coupler 15 may contain a reference FBG sensor, as will be described below.

(27) The N FBG sensors 50 which make up the sensor array 55 are fiber Bragg gratings 50 with a reflection spectrum that falls within a spectral window, which is defined as the wavelength range covered by the FBG central wavelength over the range of possible values of the measurand. The average wavelength of the pulses λ.sub.av is located more or less at the center of that spectral window. As in other TDM schemes, the maximum reflectivity of the FBG sensors 50 is made to be small enough that multiple reflections (resulting from the reflection on more than one FBG sensor) are considered negligible. Typically, it means that the maximum reflectivity lies in the range of 1-2%. Thus the FBG sensors 50 reflect the light pulses, but the reflectivity for each sensor depends on the value of the measurand at each sensor location, since the measurand affects the central wavelength of the FBG reflection spectrum. Pulses of different wavelengths λ.sub.+ and λ.sub.− will also be reflected with different intensities, because their wavelength is at a different location within the FBG reflection spectrum. This is illustrated in FIGS. 5A and 5B, which show the FBG spectrum, as well as the location of λ.sub.+, λ.sub.− and λ.sub.av within the FBG spectrum, and the corresponding reflectivity, for two different values of λ.sub.B.

(28) For each emitted pulse, the reflected signal thus consists in a series of N pulses, N being the number of sensors in the array, with a delay τ between each pulse being equal to the travel time of light between each FBG sensor 50. Since that delay τ is always larger than τ.sub.p (pulse duration), the reflected pulses are temporally distinct. This is illustrated in FIGS. 3A-3C, in a case where, for each emitted pulse as seen in FIG. 3A, there would be 4 such reflections (N=4) as seen in FIG. 3B, and where the wavelength alternates from pulse to pulse (λ.sub.+,λ.sub.−), as well as the peak intensity (I.sub.+,I.sub.−). The graph of FIG. 4A shows a pattern of alternating λ.sub.+ and λ.sub.− pulses that also have different peak intensities, which is likely to be the case if one uses DFB laser diodes with different peak current in order to vary the wavelength of the pulses. The graph of FIG. 3B shows the reflected echoes for a case where there would be an array of 4 FBGs along the fiber.

(29) For the case where the two pulses are emitted within a time τ, illustrated in the graph of FIG. 4A, the reflected pulses are corresponding pairs of pulses, as shown in the graph of FIG. 4B.

(30) The reflected echoes from the N FBG sensors 50 pass through the optical fiber coupler 15 or circulator, and are redirected onto a photodetector 20, where the optical signal is converted into an electronic signal. The response time of the photodetector 20 and electronic amplifying circuit 25 is made to be substantially shorter than the pulse duration, so that the amplified electronic signal is a faithful temporal reproduction of the optical signal.

(31) The determination of the measurand value for each sensor in the FBG array is accomplished by an electronic and numerical processing of this electronic signal, as described below.

(32) For the purpose of this description, the FBG sensors 50 must have a specially designed and well characterized spectral response. Standard FBGs typically have a reflection spectrum with a peak reflection at a central wavelength λ.sub.B, and sidelobes on each side of the central peak. However, for the purpose of this description, the shape of the reflection spectrum according to an embodiment is a gaussian function with no sidelobes. Such gaussian-shaped FBGs can be fabricated by a number of techniques known to a person skilled in the art. For such a function, a spectral bandwidth Δλ.sub.B can be defined as the range of wavelengths where reflectivity is greater than 50% of the maximum reflectivity.

(33) The effect of the measurand is to shift the central wavelength λ.sub.B. The measurand can be temperature, strain, or any other environmental condition which modifies the reflection spectrum of FBG sensors. Over the measurement range targeted by the sensor, that wavelength will shift by a maximum amount Δλ.sub.MAX. Given that the light source 11 used to interrogate a given sensor has a narrow optical spectrum centered on a wavelength λ.sub.av, one must design the central wavelength of the FBG sensors 50 and the width of their reflection spectrum in such a way that at both extremities of the measurement range, the FBG reflectivity is large enough to result in an acceptable signal-to-noise ratio. As a rule of thumb, one may design the FBG sensors 50 such that reflectivity is always greater than 50% of the maximum over the entire measurement range, in which case one will chose λ.sub.B to be equal to λ.sub.av at the center of the measurement range, and Δλ.sub.MAX to be equal to Δλ.sub.B. Nevertheless, it will be recognized that the reflected signal of the apparatus presently described will in general be much higher than with the use of a broadband source, which facilitates detection with a high signal-to-noise ratio.

(34) Once the reflected signal has been converted to an electronic signal, the electronic and numerical processing that is performed to extract the value of the measurand for each sensor can be accomplished in a variety of ways, all known in the art. Therefore, details of all possible ways to perform such processing will not be given herein. All such methods rely on measuring the peak intensity or integrated energy of each individual echo pulse, and applying algorithms to such measured values. According to an embodiment, the sequences of N pulses are digitized by an ADC chip (analog-to-digital converter). Once digitized, the integrated energy of each echo pulse can be calculated by integrating the signal over the temporal window of duration τ associated with each echo. The N values thus obtained for each emitted pulse are then processed by a microprocessor, which can be a fast CPU chip or an FPGA chip, programmed to perform treatment comprising a series of steps to extract the values of the measurands for each FBG sensor 50 in the array.

(35) This mathematical treatment is based on the following method. As described above, the pulses launched into the optical fiber 30 are of two types: those of higher and lower wavelengths, respectively λ.sub.+ and λ.sub.−. If a DFB laser diode is used, with different peak driving current for each type of pulse, then the peak optical intensity of each pulse will also be different, and is labeled as I.sub.+ and I.sub.−, respectively. Assuming that the round-trip transmission loss α between the light source 11 and the FBG sensors 50 remains constant or nearly constant between the time of emission of both types of pulse, and also that the value of the measurand does not change in a significant way between each pulse, then the reflected pulse intensities relative to emitted pulse intensity for both types of pulses will differ only by the difference in reflectivity due to different central wavelengths λ.sub.+ and λ.sub.− of the first and second pulse types, as illustrated in FIGS. 5A-5B. This is due to the fact that the FBG reflectivity is a function of wavelength. If the reflectivity of the FBG as a function of wavelength is given by the function R(λ), then the difference D between the peak reflected intensity of the first and second pulse types is:
D=α*[I.sub.+*R(λ.sub.+)−I.sub.−*R(λ.sub.−)]  (1)

(36) On the other hand, the sum S of the intensities is:
S=α*[I.sub.+*R(λ.sub.+)+I.sub.−*R(λ.sub.−)]  (2)

(37) The ratio A between these two quantities, defined as A=D/S, is:
A=[I.sub.+*R(λ.sub.+)−I.sub.−*R(λ.sub.−)]/[I.sub.+*R(λ.sub.+)+I.sub.−*R(λ.sub.−)]  (3)

(38) Defining the ratio C=I.sub.−/I.sub.+, then equation (3) can be rewritten as:
A=[R(λ.sub.+)−C*R(λ.sub.−)]/[R(λ.sub.+)+C*R(λ.sub.−)]  (4)

(39) It can be seen that this ratio is independent of the transmission loss α. For constant laser driving conditions, the ratio C is constant and, in principle, is a known quantity. Therefore, the quantity A is strictly a function of the shape of the reflectivity spectrum R(λ) of the FBG, and the value of A depends on the difference between λ.sub.av and the central wavelength of the FBG sensor λ.sub.B. If the shape R(λ) is such that the function given by eq. (4) is single-valued over a range of wavelengths corresponding to the measurement range (the spectral window), then the value A can be uniquely attributed to the difference between the central wavelength of the FBG, λ.sub.B, and the average wavelength of the two types of pulses, λ.sub.av. Since λ.sub.av is kept constant, A only varies with λ.sub.B and can therefore be related to the measurand.

(40) Having a knowledge of the ratio C, one can also calculate a normalized value B, given by:
B=[R(λ.sub.+)−R(λ.sub.−)]/[R(λ.sub.+)+R(λ.sub.−)]  (5)

(41) As will be shown below, the quantity B can be directly related to the derivative of the spectral shape of the FBG reflection spectrum.

(42) Either quantity A or B may be used to relate the measurement of the reflected pulse intensities to the measurand. For reasons that are explained below, the quantity B is preferable.

(43) To illustrate how this is achieved in practice, let's take the simple example where the shape of the grating spectrum is a gaussian function, expressed as:
R(λ)=R.sub.maxexp(−4 ln(2)(λ−λ.sub.B).sup.2/Δλ.sub.B.sup.2)  (6)

(44) R.sub.max is the peak reflectivity, λ.sub.B is the peak wavelength, and Δλ.sub.B is the full width at half maximum of the gaussian spectrum. Since the difference between λ.sub.+ and λ.sub.− is much smaller than the width of the spectrum Δλ.sub.B, the difference between R(λ.sub.+) and R(λ.sub.−) can be approximated as the derivative of the function R(λ) at a wavelength λ.sub.av, multiplied by the wavelength difference between λ.sub.+ and λ.sub.−, expressed as δλ. On the other hand, the denominator in eq.(5) can be approximated to be twice the reflectivity at the average wavelength λ.sub.av, i.e., R(λ.sub.+)+R(λ.sub.−)≈2R(λ.sub.av). Using the function of eq. (5), one then gets:
B=−4 ln(2)R.sub.maxδλ(λ.sub.av−λ.sub.B)/Δλ.sub.B.sup.2  (7)

(45) As can be seen, B is then a linear function of (λ.sub.av−λ.sub.B) over the entire spectrum. Other shapes than gaussian for R(λ) can be used, but they do not result in a linear function for B. A linear response simplifies calibration, and the fact that B is single-valued for any value of λ.sub.B prevents ambiguous measurements. Therefore, the gaussian shape is one which works well. However, should the shape deviate from an ideal gaussian shape, prior knowledge of the spectral shape of each grating can be used to generate calibration curves.

(46) The quantity B can be calculated either using the peak intensity of the pulses or their integrated energy. However, the latter is preferable because it is less prone to noise. Furthermore, B can be averaged over a number of pulses to further reduce noise.

(47) In practice, the ratio C may vary either from pulse to pulse, or due to a slow drift in the driving electronics. To ensure that the proper value of C is used in calculating the values of B, it is advantageous to measure it in real time. This can be achieved by directly measuring the pulses emitted by the laser diode, which are used as references to normalize the signals reflected by the FBG's. Three possible ways of performing that measurement are illustrated in FIGS. 6A-C.

(48) In a first embodiment, a photodiode 120 can be inserted at the output of the second branch of the optical fiber coupler 15 (FIG. 6A). The second embodiment (FIG. 6B) uses the internal photodiode 120 in the laser diode package. A third embodiment (FIG. 6C) uses the reflection from the end face of the second branch of the optical fiber coupler 15, or any reflective device inserted there, which is then detected by the same photodetector 20 that detects the echoes from the sensor array 55. In the latter case, the length of fiber between the fiber coupler and the reflective surface must be adjusted such that the reflected pulse arrives on the photodiode at a different time than all the other echoes from the sensor array. Those reference pulses, after detection and amplification, are similarly amplified and digitized to provide either a peak value, or an integrated value, similar to the echo pulses. Other equivalent embodiments can provide similar results.

(49) After digitization of a sequence of N pulses at wavelengths λ.sub.+, the integrated energy E.sub.+i can be calculated for each i.sup.th pulse in the sequence by integrating the digitized values over the temporal window of each pulse. This is done for the sequence of pulses at wavelengths λ.sub.− give E.sub.−i. The same procedure is applied to the reference pulses, which gives values E.sub.ref+ and E.sub.ref− If bursts of pulses at λ.sub.+ are followed by bursts of pulses at then the integrated energies for each individual pulses are averaged over the number of pulses in the bursts. The N values of B.sub.i for each of the i.sup.th pulses are then extracted from those values by calculating:
B.sub.i=((E.sub.+i/E.sub.ref+)−(E.sub.−i/E.sub.ref−))/((E.sub.+i/E.sub.ref+)+(E.sub.−i/E.sub.ref−))  (8)

(50) This is illustrated in the graphs of FIGS. 3C and 4C, where the shaded areas correspond to the integrated energy of the first pulses in the sequence. As described above, the values B.sub.i can be directly related to the value of the measurand for the i.sup.th FBG sensor in the array.

(51) The module 10 described above can be used as a standalone apparatus, to measure a sensor array 55 of N FBG sensors 50. However, multiple modules 10 having light sources 11 at different wavelengths λ.sub.av can be used to independently interrogate multiple sensor arrays 55 of FBG sensors 50, accessed via a single optical fiber 30. The spacing between the wavelengths of the various sources must be larger than the spectral window. This is accomplished by multiplexing the outputs of the multiple modules 10 into a single optical fiber 30, with a wavelength division multiplexer 60. For example, and according to an embodiment, the light source 11 comprises laser diodes which can be of the type used for so-called Coarse Wavelength Division Multiplexing systems (CWDM), which is an established standard for optical telecommunications systems. Such diodes have wavelengths separated by 20 nm, between 1270 and 1610 nm, providing 18 different channels. Off-the-shelf multiplexers are commercially available to combine up to 16 of those channels into a single fiber. The multiplexing configuration (60, 65) is illustrated in FIG. 2A. A 20 nm spacing is large enough to accommodate most practical spectral windows. Therefore the maximum capability of the apparatus 1 does not depend on the value of the spectral window. Furthermore, the standardized wavelengths mean that the FBG sensors 50 can also have a standardized design, which lowers manufacturing costs.

(52) One advantage of having all the light sources 11 combined into a single optical fiber 30 is that all the FBG sensors 50 can be inscribed along a single fiber. Thus the maximum number of FBG sensors 50 along that optical fiber 30 is the product of the maximum number of sensors interrogated by each module 10, times the number of modules 10.

(53) However, one other advantage is that all the sensors can be addressed via a single entry point for the optical fiber 30, passed which the light from different modules 10 can be redirected into side branches. For example, if one wants to measure the temperature or strain at multiple locations in a rotating equipment, such as the rotor of a generator, then a fiber optic rotary joint 70 located on the axis of the rotor can be used between the interrogator 12 and the sensors, as illustrated in FIG. 2A. Passed the fiber optic rotary joint 70, different wavelengths can be sent onto separate branches from the main fiber, with the use of add/drop multiplexers 65. Such devices can either extract one channel (at one wavelength) and launch it into another fiber, or extract a group of channels. For example, 8 out of 16 channels can be sent into a separate branch, or 4 out of 8 channels, as illustrated in FIG. 2A. According to another embodiment shown in FIG. 2B, a wavelength independent fiber splitter can be used to distribute all the wavelengths from the different modules more or less equally among multiple optical fibres that comprise multiple parallel arrays of sensors, each interrogated by a different module. This has the advantage of providing redundancy in case of failure or breakage of one of the branches. Having all interrogating signals in the same fiber is also advantageous if the interrogator 12 is located at a large distance from the FBG sensors 50.

(54) Thus the modular architecture of the apparatus 1 offers a lot of flexibility. Various configurations for locating the FBG sensors 50 are shown in FIGS. 7A-7C. The modules 10 operate independently, and the associated sensor arrays 55 can all have different characteristics, such as different distance between sensors, or different spectral width of the FBG sensors 50. The latter determines the measurand range, but also its resolution. Thus sensors with large span and low resolution can be combined with sensors of small span but high resolution. The sensor arrays 55 can be interleaved with regular spacing (FIG. 7A), or bunches of sensors from different arrays bundled with close spacing, at regular intervals, which can provide higher spatial resolution (FIG. 7B). Or else the sensor arrays 55 can be concatenated one after the other (FIG. 7C). The modular nature of the apparatus 1 also means that the cost is incremental with the number of modules used. FIG. 8 shows how add/drop multiplexers 65 can be used to launch signals from one or more modules 10 into separate branches.

(55) DFB laser diodes can be made to generate pulses of 1 nanosecond or even less. Thus the minimum spacing between sensors for one array can be as small as 10 cm. The speed of the electronic processing ultimately limits that spacing. Because signal is lost on each reflection, a limited number of sensors can be used in one array, typically about 15. Since as many as 16 channels can be combined with off-the-shelf multiplexers, the maximum capability of the apparatus 1 can be as high as 240 sensors.

(56) Since the wavelength emitted by the laser diode is a function of the temperature, a good temperature control 13 of the diode is required. This can be achieved with a thermoelectric cooling element, using the signal from a temperature sensor located close to the diode as an error signal, in a manner that is well known to the art. For long term stability, it is important that the central wavelength of the diode λ.sub.av, as well as the difference in wavelength δλ between the two types of pulses remain constant in time or, if they are not, that their value be monitored and used to correct the calibration of the instrument. Laser diodes are known to age over time, therefore these two values are likely to drift slowly over time. To account for long term drift of the central wavelength, and of the value δλ, it is beneficial to use a reference sensor that is well calibrated and known to be stable in time. The reference sensor can itself be a fiber Bragg grating of similar design to the other sensors, but whose temperature and strain are known with good accuracy. This can be achieved by measuring the FBG temperature with a another precision temperature sensor such as a thermistance, or else by maintaining the FBG at a stable temperature and strain, in manners known in the art.

(57) The nature of TDM interrogation allows the interrogation of such a reference sensor simultaneously with the other sensors in the array. For this, the sensor has to be located at a distance from the interrogation unit that is such that the echo from the reference grating uses one of the available temporal windows. This is done by adjusting the distance from the source to the reference sensor. The reference sensor can be located in the second branch of the optical fiber coupler 15, as long as its position is such that the echo falls within one of the temporal windows.

(58) From the embodiments described above, it will be understood that the each FBG sensor 50 has a known reflection spectrum, also known as a reference reflection spectrum, which can shift under given environmental conditions applied to each one of the FBG sensors 50, thereby defining a shift for each one of the FBG sensors 50. Each one of the FBG sensors 50 is located at a different distance of the light source 11 and the receiver (there is spacing between each one of them among a given array 55), thereby allowing the receiver to temporally discriminate each reflected pulse received from the FBG sensors 50 from a given array.

(59) When more than one array 55 is used, FBG sensors 55 of different arrays 55 can be at the same distance from the light source 11 and from the receiver than another FBG sensor 50 of another array 55. Therefore, there is a need for a spectral separation between each one of the arrays 55, i.e., the FBG sensors 50 of a given array 55 all respond inside a given spectral window. Spectral windows characterizing different arrays 55 should not overlap. Therefore, the receiver, when receiving a reflected pulse, is able to unambiguously identify the FBG sensor 50 on which the pulse was reflected based on the time at which it was received and on the spectral window characterizing the reflected pulse.

(60) Since each array 55 is made to reflect in a given spectral window, there must be an interrogator adapted to emit at a wavelength corresponding to the given spectral window, and avoiding using broadband or tunable sources, as mentioned above in the background section. Therefore, each one of the arrays 55 has a corresponding light source 11. The light source 11 corresponding to an array 55 is adapted to emit within the spectral window of its corresponding array. Therefore, each one of the arrays 55 and its corresponding light source 11 operates in their respective spectral window.

(61) As described above, the light source 11 does not need to emit on a continuous range of wavelengths to get a “picture” of the reflected pulse, as performed in the prior art. Since the shape of the reference reflection spectrum is known for each FBG sensor 50, and since the shift is known to be relatively small (smaller than the spectral window of the FBG sensor 50), only a discrete number of points on the actual (shifted) reflection spectrum need to be interrogated. In the embodiments described above, the shape was known to be (at least approximately) a gaussian shape, for which only two points in the actual (shifted) reflection spectrum needed to be determined in order to calculate the shift of the actual reflection spectrum with respect to the reference (known) reflection spectrum. More complex reflection spectra could require a different number of wavelengths for unambiguous determination of the shift.

(62) For this reason, each light source 11 does not emit on a whole spectral window but rather emits a discrete number of wavelengths (such as λ.sub.+ and λ.sub.− above) within the respective spectral window of the light source 11. Since both λ.sub.+ and λ.sub.− are within this spectral window, one can formalize the situation by defining an average wavelength λ.sub.av about which λ.sub.+ and λ.sub.− (or other wavelengths if there are more than two) are emitted. Therefore, each light source 11 is characterized by its respective average wavelength within the respective spectral window of the corresponding one of the arrays 55, about which a given number of discrete wavelengths (e.g., 2) are emitted.

(63) Pulses reflected by each FBG sensor 50 are received at a given time and wavelength which are used to unambiguously identify the FBG sensor 50 which reflected that pulse. The reflected pulse is also indicative of how much the actual reflection spectrum of a given FBG sensor 50 shifted from its reference reflection spectrum (i.e., the shift of each FBG sensor 50). It is therefore also indicative of the conditions prevailing for this FBG sensor 50. The shift for each FBG sensor 50 can be calculated as explained above. The conditions prevailing at a given FBG sensor 50 can also be retrieved via the shift using a calibration table.

(64) While embodiments have been described above and illustrated in the accompanying drawings, it will be evident to those skilled in the art that modifications may be made without departing from this disclosure. Such modifications are considered as possible variants comprised in the scope of the disclosure.