WO1991017414A2

WO1991017414A2 - Non-destructive structural evaluation

Info

Publication number: WO1991017414A2
Application number: PCT/GB1991/000722
Authority: WO
Inventors: Alexander Keith Mackenzie; William Murphy Cummings
Original assignee: The Secretary Of State For Trade And Industry In Her Britannic Majesty's Government Of The United Kingdom Of Great Britain And Northern Ireland
Priority date: 1990-05-09
Filing date: 1991-05-07
Publication date: 1991-11-14
Also published as: WO1991017414A3; AU7799591A; GB9010444D0

Abstract

Excitation is applied to a structure (2) and infrared radiation (5) emitted by the structure (2) passes through a telescope (3), scanning system (30) and focussing optics (40). The signal is then detected by an IR detector (24). The output from the detector (24) is passed to a computer system (17). A reference signal (22) is derived from a strain gauge (4) and is passed to the computer (17). The reference signal has frequency, amplitude and phase content which is characteristic of the excitation and so provides a basis for noise rejection by correlation.

Description

NON-DESTRUCTIVE STRUCTURAL EVALUATION

This invention relates to non-destructive structural evaluation and in particular to such evaluation of structures using scanning detector systems.

For the purpose of this specification a structure means any object or system capable of converting excitation energy into heat or capable of propagating thermal energy or any part of such an object or system.

In particular such structures will be capable of bearing mechanical load but this may not be their primary purpose. Any coating applied to. an object or system is included within this definition of structure. Further, structural evaluation means the assessment by experimental means of: energy coversion processes within the structure or thermal energy propagation through the structure either in their own right or as a means of assessing the integrity of, discontinuities within, the material properties of, or the variation in material properties of or the environmental or service conditions applied to a structure. Evaluation of structures non-destructively has in the past been achieved by analysing the infra-red (or thermal) response of a target structure to mechanical, electromagnetic or other excitation which may be at a single frequency, may be cyclic with content at a number of frequencies may be random with energy content over a range of frequencies or may be transient with energy content over a range of frequencies. Examples of such techniques are Thermoelastic Stress Analysis (TSA), Pulse Video Thermography (PVT), Thermal Evaluation for Residual Stress Analysis (TERSA), Photoacoustic Microscopy and Vibrothermography.

There are drawbacks with each of these techniques. For example, TSA, which is now an established technique, is implemented by means of an infra-red detector element focussed on a small area of the structure. Data is collected from this point and a variety of conventional signal process¬ ing techniques are applied to obtain correlation between the response signal from the detector and an applied load so achieving a high degree of noise rejection. A temperature resolution of 0.001 K is claimed with this technique.

A major disadvantage of this system is that data must be collected on a point by point basis using a single staring detector looking at each successive point on an area long enough to achieve adequate noise rejection and con¬ sequently leading to very long test times for full field test measurements with potential for unnecessary damage to structures. In PVT, where the structure is excited by a transient full-field optical or thermal flux, data capture is by means of a TV rate ie 25 or 30 frames per second imaging system.

TERSA which is the subject of UK Patent No 1601890 again uses a staring detector. In this case localised optical excitation is used. Again long testing times result.

With the above techniques then there is a major disadvantage that, where individual element IR detectors are used in a staring mode the time taken to scan and carry out evaluation of structures is very long. Where fast scanning detector systems are used the resulting temperature resolution is relatively poor.

There is therefore a need for the non-destructive testing of structures by thermal evaluation which can be carried out quickly and easily. Additionally a system is needed which can resolve very small differences in temperature.

By use of the present invention a reference signal will be derived giving a basis for noise rejection and so improved temperature resolution. Also much higher sampling rates will be able to be used thus permitting analysis of materials with high thermal diffusivity or very thin structures eg. coatings.

According to one aspect of the present invention there is provided apparatus for non-destructive structural evaluation comprising: means to apply dynamic excitation energy to a structure; at least one detector responsive to energy radiated from the structure in consequence of interaction of the excitation energy and the structure: scanning means to cyclically scan in a continuous manner the field of view of the detector over an area of the structure; means to record a time series of detector measurements, corresponding to respective discrete points in the area of the structure; and computing means to evaluate the area under inspection; characterised in that the apparatus further comprises means to derive one or more reference signals representative of the excitation energy or of the response of the structure to the excitation energy, and the computing means correlates the time series, of detector measurements corresponding to each discrete point with a corresponding time series of values for the one or more reference signals or with an equivalent numerical representation of such a time series of values.

The means for applying excitation energy to the structure may be a flash tube, a laser or other optical source, a source of mechanical energy or a source of heat. Where suitable excitation is applied to a structure as a result of environmental or service conditions then, for the purpose of this specification, this is considered to constitute a means of applying excitation energy to the structure.

The reference signal is to some degree representative of the applied excitation signal and/or the response of the structure. "Representative" here means correlating with the excitation /response, which means having a fixed relationship between respective frequency, phase, and/or amplitude content. For example, using a laser excitation chopped to give a square profile with an even mark space ratio, a reference signal could be a sine wave at the fundamental frequency and in phase with the chopper or sine waves at odd harmonics of the fundamental frequency again with phase fixed with respect to the chopper.

The reference signal may be derived in many different ways. For example, the reference signal may be generated by monitoring a drive signal to a loading system or from a load cell in the load chain or from a displacement transducer or from a transducer attached to the structure under test. The reference signal may also be derived by non-contacting methods. These include a reference signal derived from a laser or other optical excitation system and also use of the measured infra-red response averaged over part of the structure with high response to the excitation.

A reference signal may also be derived by measuring the level of radiation emitted by the source of exciting energy. In the case of a collimated beam this will typically involve the use of a beam splitter after any chopping or other modulation of the beam. A reference signal could also be the pulsewaveform of a single pulse where the pulse shape determines the frequency content of the applied signal.

Where a chopped beam is used the pulse shape can be calculated from the chopper shape, beam diameter and energy distribution. Preferably chopped laser beam is used. The reference signal(s) may be measured by transducers applied to the structure. Preferably the reference signal(s) are derived during the cyclic scanning of the detector over the structure. Alternatively, the reference signal(s) may be derived using a strain gauge. In a further arrange¬ ment, the reference signal(s) may be synthesised from knowledge or the excitation system and means to determine the relative timing of the excitation and the response signals. The reference time series may also be stored as an equivalent frequency domain representation arrived at by application of a fourier transform or other mathematical transformation ('equivalent' here means that sufficient information is retained to permit correlation by means of a suitable algorithm).

The energy radiated from the material may be detected by a single infra-red detector or by an array of detectors so as to cut down the number of line scans required per frame scan and hence to increase the frame acquisition rate and in consequence the rate of sampling of surface area elements. Preferably the detectors used are SPRITE detectors which are capable of very high rates of data capture and give low noise. This permits time resolved infra-red/temperature response to be logged over line response profiles or area response maps.

The scanning means is preferably a polygon motor for line scanning and an indexed mirror for frame scanning.

The means to record the time series of measurements is preferably an analogue to digital convertor for each detector sampling at an appropriate rate and is preferably synchronised to the frame scan. The appropriate rate is determined by the spatial resolution required or achievable by the detector and its associated optics.

The computing means used preferably employs parallel processing for speed of handling data throughout and correlation with the reference signal(s) where appropriate can be used for noise rejection. This can be used in conjunction with various other signal processing techniques.

In the discussion above one or more reference signals are derived primarily to provide a basis on which information carried by desired signals may be distinguished from noise. In particular the measured infra-red response signal will typically be noisy and information will be extracted by correlation with the reference signal(s).

The more representative the reference signal is of the desired signal then the better the level of noise rejection that can be achieved. For example if the desired signal is sinusoidal then a knowledge of frequency may be used as a reference to distinguish this signal from noise. This is equivalent to filtering of the signal. However if the reference is known to have a fixed phase relationship to the desired signal as well as the same frequency then much higher levels of noise rejection are feasible. If the amplitude of the desired signal is known to hold a fixed ratio to the amplitude of the reference then this may also be used as a basis for noise rejection. This description may be extended to signals and references with more general cyclic or transient or random waveforms by means of a suitable transformation to the frequency domain where relative amplitudes and/or phases may be considered as a function of frequency. The more closely a reference signal correlates with the desired signal then the better basis for noise rejection.

In some cases the physical behaviour of the structure or test system may map energy from one frequency to another. In such cases harmonic analysis will generally be used with the response signal being sought at the frequency of the excitation and at integer multiples of that frequency. A reference signal will also carry information, measured or calculated, relating to the physical behaviour of the structure, the test system, the applied excitation or the environment. This information may be used to permit or improve interpretation of the measured results in terms of physical models. In some cases the information carried by the measured infra-red responses may provide sufficient basis for such interpretation without additional information. In general additional transducers will be applied to the structure and/or the experimental system to give a basis for improved interpretation of results and/or improved control of the test system eg the use of a strain gauge to achieve desired strain levels in a structure through conventional closed loop control or through remote parameter control methods.

As a matter of expediency and efficiency software to assist in inter¬ pretation in terms of physical models or to give control of the test system will be implemented in the same computing system as the time series analysis software used for noise rejection.

Aliasing occurs when a signal is sampled, normally associated with digitisation. ^Normally analogue filters are used to limit the bandwidth before sampling and digitisation. This will allow baseband or zoom processing. In terms of the present invention analogue filters may be used in this way to condition analogue reference signals derived from transducers.

When using a continuously scanning detector system, as in the present invention, sampling is inherent in the scanning process and the normal option of using analogue filters is unavailable. In order to reduce problems associated with aliasing then, the bandwidth of the excitation may be limited or a suitable coating may be applied to the structure to act as a 'thermal filter¹ which attenuates higher frequencies and thus acts as a low pass filter. The bandwidth of the responses for a particular point on the structure which can be described from a single measurement time history can be optimised by selection of the scan area and/or scanning rate of the con¬ tinuously scanning detector system so optimising the resulting sampling rate at each point on the structure. According to another aspect of the present invention there is provided a method of non-destructive structural evaluation comprising the following steps: applying a dynamic excitation energy to a structure; detecting energy radiated from the structure in consequency of interaction of the excitation energy and the structure; cyclically scanning in a continuous manner the field of view of the detector over an area of the structure; recording a time series of detector measurements corresponding to respective discrete points in the area of the structure; evaluating the area of structure under inspection; characterised in that the method further comprises the step of deriving one or more reference signals representative of the excitation energy or of the response of the structure to the excitation energy and in that the evaluation of the structure is by computing means being used to correlate the time series or detector measurements corresponding to each discrete point with a corresponding time series of values for the one or more reference signals, or with an equivalent numerical representation of such a time series of values.

A system and method for non-destructive testing of structures is thus provided which gives full field or line scan analysis of steady state cyclic or steady state random or transient events which is up to 2 orders of magnitude faster than established methods using currently available detectors. This leads to several performance advantages when applied to structural analysis: fewer load cycles leads to less damage to structures, interactive operation in diagnostic or in the field operations, high throughput rendering many quality assurance tests, viable and better noise rejection. A wide range of excitation regimes may be catered for giving flexibility in the design of test procedures. In particular there are advantages in the combination of localised electromagnetic (typically optical) excitation in combination with full field infra-red measurement ie where one or more localised areas of the structure are excited with the infra-red/thermal response measured over an area larger than the areas excited or over areas not including the areas excited. These advantages include more sensitive measurement of thermal diffusion effects, particularly diffusion parallel to the surface of the itructure and reduced interference with the measurement due to radiation reflected by the structure from the excitation source.

The system is also robust and portable and can be used in 'hostile' or inaccessible environments.

The present invention will now be described, by way of example, with reference to the accompanying drawing which is a diagramatic represent- ation of one embodiment of the invention.

The drawing shows a structural evaluation system according to the invention as applied to a controlled mechanical excitation of a structure 2. Mechanical excitation is applied to the structure 2 by an electromagnetic or servohydraulic actuator 1 with control system. Radiation 5 from the structure passes through a telescope 3, scanning system 30 and focussing optics 40. The scanning system 30 comprises a line scanning polygon mirror drum 6, a polygon motor 7, frame scanning mirror 12, a frame motor 13 and mirrors 14, 15, 16. Focussing optics 40 comprises a cold shield 8, a lens 9, a filter 10 and a stop 11. The signal is then detected by a multi- element SPRITE infra-red detector 24, each element of which is connected to conditioning amplifiers and a detector bias system 21. The conditioning amplifiers in turn are connected to fast A/D converters 20. Output

from the A/D convertors 20 is connected to the input of a computer system 17. Signal 23 from the scanning system 30 is also connected to the computer 17 via a digital I/O 19. A strain gauge 4 which may include one or more transducers is attached to the structure and Signal 22 from the strain gauge 4 is connected to the computer via A/D convertors 18. Outputs 26 and 27 from the computer 17 are connected via D/A converters 25 to the control system of the frame motor 13 and the control system of actuator 1 respectively. In use excitation energy is applied to the structure 2 by the actuator 1. A reference signal 22 is generated by monitoring the response from the structure 2 which passes through the A/D convertor 18 to the computer 17. The reference signal has frequency, amplitude and phase content which is characteristic of the excitation and so provides a basis for noise rejection by correlation.

The structure 2, when subjected to the excitation applied to the actuator 1, undergoes temperature changes by thermo-elastic interactions. These are detected by an infra-red (IR) detector system shown generally by 50. Infra-red radiation 5 which is emitted by the structure 2 is collected by telescope 3 onto the line scanning mirror drum 6 which provides continuous scanning across the structure 2 under control of the polygon motor 7. The radiation 5 is directed by various mirrors 14, 15, 16 to the frame scanning mirror 12 towards the focussing optics 40, which direct the radiation 5 towards the multi-element SPRITE detector 24.

The output signals from each element of the SPRITE detector 24 are passed into conditioning amplifiers and a detector bias system 21 where the bias system matches the rate of charge carrier drift in the SPRITE detector elements 24 to the rate of scan of infra-red radiation across the detector surface. The conditioning amplifier outputs are then passed through fast A/D convertors which digitise the signals from the SPRITE detector elements 24 which then forms the input to the computer 17. The computer 17 then runs time series analysis software using correlation based techniques to achieve noise rejection. In this example the infra-red response from the structure 2 may be correlated with the drive signal 27 generated by the computer 17 and with the structural response 22 as monitored by the strain gauge 4. The computer 17 also runs data reduction and display software. A full field display of the structure 2 is thus reconstructed by the computer 17.

The computer system 17 may take many different forms. Two examples are described below:

The fast A/D converters 20 map the digitised response information from one or more of the conditioned detector channels being output from the conditioning amplifiers 21 directly into a transputer's address space. Transputers at this level then partition the information and pass it to other transputers in the network. These transputers carrying out the signal processing tasks are supported by (cascadeable) Digital Signal Processing (DSP) chips residing in their memory-map. A 'display task' is implemented to assemble the processed data into an array representing the spatial distribution of the measured responses. The 'display task¹ is implemented on a separate transputer on the network or using spare capacity. Further data reduction may be necessary. An advantage of this approach is that it takes advantage of the open- ended nature of transputer networks. With sufficient hardware the processing power will handle the data acquisition and signal process¬ ing tasks in real time.

Alternatively data from each conditioned detector channel is passed directly from the A/D convertors 20 to the data lines of cascadeable DSP chips. The DSP chips are programmed by (to achieve signal processing) and return data to a powerful conventional CPU. For each of the above options the A/D convertors 20 carry out block conversions, each block being triggered by the synchronisa¬ tion signal 23 from the polygon scanning system 7. Provision is made for a separate data path giving the signal processing task .informa- tion about the reference signals. Correlation techniques, based on time series analysis, can be used to derive frequency response functions ie the amplitude and phase of the response as a function of frequency. This is expensive in terms of processing and is more than is required in many experimental situations. A less demanding procedure has been explored and has been successfully applied to Thermoelastic Stress Analysis (TSA) for the determination of quasi-static stress distributions, ie where the frequency content of the excitation is kept sufficiently low to avoid resonant behaviour in the structure. A description of the method is included here as an example of a correlation based noise rejection technique.

The 'linear approximation model' is based on two assumptions/ approximations:

(1) The response, at the point on the target under consideration, is a linear function of the excitation (NB strictly the response need only be a linear function of the reference signal(s) but for simplicity this is assumed to be a measure of the excitation)

(2) The net systematic error introduced by drift in the tempera¬ ture (or any other relevant properties) of the target and drift in the signal conditioning system is a linear function of time.

This need only be approximately valid and only over the period required for a single 'measurement set¹:

R = at + ke + c where R = measured response; t = time, measured or known; e -= excitation level, measured or known; a,k,c are constants.

If we consider a 'measurement set' comprising a sequence of three or more measurements we have: ^R1'^R2'^R3 w ^

and in general can solve for a,k,c where: c is a net measure of temperature and other offsets at t=0 k is a measure of the response of the target to the excitation, a is a net measure of the rate of drift of the^'mean termpera- ture and offset due to signal conditioning. Clearly some measurement sets wil-1 not^'give a unique solution for a, k, c or will give rise to ill conditioning when using numerical methods to derive the solution. Such sets should be discarded. Conversely, careful control of the excitation can simplify the equations reducing the number of arithmetic opera¬ tions required to determine the relevant coefficients) .

Given sufficiently large measurement setsthen higher order (quadratic, cubic, et) terms could be solved for.

Given that this approach gives an adequate treatment of systematic error contributions and that the real constants extracted give meaningful information about the response of the structure then we may deal with random error contributions by averaging. In the context of ill conditioning a refinement would be to weight estimates according to their uncertainty to optimise the averaging process.

Where a known gain, attenuation and/or phase shift is imposed on the measured response, say by thermal diffusion effects or signal conditioning components, then this may be corrected for before the linear equations are solved. Alternatively the component of the measured response in phase with the excitation will be derived and, provided this component is sufficiently large, may give an adequate description of the behaviour of the target structure.

In the above invention advantage may be taken of aliasing and so frequencies above the Nyquist frequency can be processed — this is known as zoom processing.

The apparatus and method described above is either directly applicable to or applicable to derivatives of the applications listed in the preamble ie TSA, PVT, TERSA, photoacoustic microscopy and vibrothermography. Although the methods of excitation and of deriving a reference signal will be different for each application, advantages are that results of structural evaluation are obtained at orders of magnitude faster than with previous apparatus and methods, and because very high rates of data capture are feasible this permits time resolved IR/temperature response to be logged over line response profiles or area response maps. Furthermore, A wide range of excitation regimes may be catered for. This gives flexibility in designing test procedures, many service excitations can therefore be handled successfully.

Claims

1. Apparatus for non-destructive structural evaluation comprising: means to apply dynamic excitation energy (1) to a structure (2): at least one detector (24) responsive to energy (5) radiated from the structure in consequency of interaction of the excitation energy and the structure; scanning means (30) to cyclically scan in a continuous manner the field of view of the detector over an area of the structure; means (21) to record a time series of detector measurements corre¬ sponding to respective discrete points in the area of the structure; and computing means (17) to evaluate the area under inspection; characterised in that the apparatus further comprises means(4) to derive one or more reference signals (22) representative of the excitation energy or of the response of the structure to the excitation energy and the computing means correlates the time series of detector measurements corresponding to each discrete point with a corresponding time series of values for the one or more reference signals or with an equivalent numerical represen¬ tation of such a time series of values.

2. Apparatus according to claim 1 characterised in that the excitation energy is cyclic, random or transient.

3. Apparatus according to claim 1 characterised in that the correlation of the time series of radiated energy measurements with the one or more reference signals is carried out to achieve noise rejection.

4. Apparatus according to claim 1 characterised in that the correlation of the time series of radiated energy measurements with the one or more reference signals is used to permit or improve interpretation of the measured results in terms of physical models.

5. Apparatus according to claim 1 characterised in that at least one of the one or more reference signals is derived from transducers applied to the structure.

6. Apparatus according to claim 5 characterised in that the reference signal(s) are derived from the tranducers during the cyclic scanning of the detector over the structure.

7. Apparatus according to claim 1 characterised in that at least one of the one or more reference signals is derived using a strain gauge (4).

8. Apparatus according to claim 1 characterised in that the one or more reference signals are synthesised from knowledge of the excitation system and there is provided means to determine the relative timing of the excitation and the response signals.

9. Apparatus according to claim 1 characterised in that the time series of values for the one or more reference signals may be stored as an equivalent frequency demain representation.

10. Apparatus according to claim 6 or claim 7 characterised in that the apparatus further includes at least one analogue filter

11. Apparatus according to claim 1 characterised in that the detector

(24) comprises a single infra-red detector or an array of infra-red detectors.

12. Apparatus according to claim 1 characterised in that the scanning means (30) include a polygon motor (7) and an indexed mirror (12).

13. Apparatus according to claim 1 characterised in that the means to record detector measurements includes at least one analogue to digital converter (20).

14. Apparatus according to claim 1 characterised in that the structure includes a coating at its surface to act as an analogue low pass filter for infra-red and thermal signals and to filter out high frequency thermal responses.

15. Apparatus according to claim 1 characterised in that the computing means employ parallel processing techniques.

16. A method of non-destructive structural evaluation comprising the following steps: applying dynamic excitation energy to the structure; detecting energy radiated from the structure in consequence of inter¬ action of the excitation energy and the structure; recording a time series of detector measurements corresponding to respective discrete points in the area of the structure; and evaluating the area of the structure under inspection; characterised in that the method further comprises the step of deriving one or more reference signals representative of the excitation energy or of the response of the structure to the excitation energy and in that the evaluation of the structure is by computing means being used to correlate the time series of detector measurements corresponding to each discrete point with a corresponding time series of values for the one or more reference signals or with an equivalent numerical representation of such a time series of values.

17. A method according to claim 14 characterised in that an area of excitation and an area which is scanned may or may not be cotermin»«s«