СПОСОБ И СИСТЕМА НЕПРЕРЫВНОГО МОНИТОРИНГА КОНСТРУКЦИИ

Abstract

Способ и система непрерывного мониторинга конструкции (10), которая содержит текучую среду, для локализации утечки или источника (E) коррозии, который испускает акустические излучения, причем упомянутые акустические излучения обнаруживаются посредством трех устройств мониторинга (M1-M2-M3), каждое из которых имеет по меньшей мере одну группу обнаружения (G1-G2-G3), которая связана с внешней поверхностью конструкции (10), причем каждая группа обнаружения содержит три сенсорных блока (S0-Sa-Sb), расположенных выровненными друг с другом вдоль линии, параллельной оси (A-A) конструкции (10). Способ включает в себя непрерывный прием от каждой группы обнаружения и параллельно от каждого сенсорного блока (S0-Sa-Sb) соответствующих последовательностей аналоговых сигналов (si(t)) и обработку последовательностей аналоговых сигналов (si(t)) путем генерации последовательностей синхронизированных цифровых сигналов (si(nT)), которые обрабатываются на основе информационного критерия Акаике для идентификации значимых цифровых выборок (si(K'T)). Способ включает в себя предварительную локализацию упомянутого источника (E) для каждого устройства мониторинга (M1-M2-M3) путем определения радиального расстояния упомянутого источника (E) от упомянутого центрального сенсорного блока (S0) на основе разностей времени, определенных посредством временного анализа соответствующих значимых цифровых выборок (si(K'T)), и локализацию упомянутого источника (E) путем обработки радиальных расстояний упомянутых трех устройств мониторинга (M1-M2-M3) с использованием программного обеспечения для обработки.

Description

[0001] "Method and system of continuous monitoring of a structure" ** **

[0002] The present invention relates to a method and system for continuous monitoring of a structure, which comprises a fluid, to detect acoustic emissions for the localization of a leak or a corrosion phenomenon in said structure.

[0003] The structure is preferably large and can be a storage tank made of metal / composite or a civil or industrial construction also made of concrete, comprising fluids that can be liquids such as refined products, water, oil, etc., gaseous and / or multi-phase fluids such as oil, water and natural gas and / or multi-state components such as carbon dioxide, hydrogen mixtures, etc.

[0004] Prior art

[0005] Several methods and systems for monitoring structures and / or constructions or pipelines in order to identify damage, anomalies and / or fluid leaks through the detection of acoustic emissions are known.

[0006] Structural damage and fluid leaks may be due to natural deformation and mechanical fatigue effects, corrosion of the material, accidental break due to undetectable defects in the construction material, or as a consequence of third- party activities, e.g. as a result of mechanical or manual excavation near the structure.

[0007] It is known to place a plurality of sensors arranged in concentric loops circumferentially to circular tanks or vessels. The signals emitted by damages, anomalies and / or fluid leaks are detected by sensors arranged on a same loop and processed by selecting and comparing the detected values with minimum threshold values, L=low, and useful references for maximum values, H=high.

[0008] These known processes and systems, although satisfactory in several aspects and for several applications, are not free from drawbacks such as false alarms when the acoustic signals detected by the sensors are generated by sources outside the tank.

[0009] Indeed, in order to reliably determine breaks or leaks, it is necessary to process homogenous signals, i.e. generated by a same source.

[0010] Furthermore, an error in detecting the signal or an error due to the transmission of the signals sent to the processing unit by the sensors in the same loop may compromise or limit the entire process by reducing the useful references, H=high or L=low. In some cases, a lack of correlation of amplitudes can effectively nullify all the measurements detected by circumferential sensors.

[0011] Other known systems provide periodic check and testing activities that require to stop or shut down production of a portion or of the entire plant. These shutdowns are quite invasive and require the off-line or rest condition of the tank and of all the industrial plant sources related to the instrumentation, including: pumps, motors, rotating equipment, and further external environmental sources that may create acoustic noise and thus interfere with the signals detected by the sensors. Therefore, periodic check and testing activities are particularly time-consuming and expensive.

[0012] Furthermore, these periodic check and testing activities, while advantageous for identifying fluid leaks, are not very reliable for detecting deterioration within the structure and corrosion phenomena that can produce leaks in the time range between two successive check activities.

[0013] It is known that in the presence of leaks due to corrosion phenomena, including chemical corrosion, acoustic emissions and / or vibrations can increase irregularly over time, making it difficult to detect them by means of periodic surveys or they may be detected with considerable delay. The damage to be considered is quantitative as well as qualitative, since environmental damage must also be taken into account.

[0014] Furthermore, external environmental noise can significantly alter the value of the signals detected by the sensors placed on the loops, compromising processing.

[0015] The main task of the present invention is to devise and make available a simple, non-invasive method and system that allow to monitor structures efficiently and effectively for a continuous short- or long-term monitoring, which can also be used in existing structures having structural and functional characteristics that solve the technical problems highlighted, eliminating the drawbacks mentioned with reference to the prior art.

[0016] Brief summary of the invention

[0017] The solution idea underlying the present invention is to continuously monitor the axial sections of the lateral surface of the structure, and to process the detected signals with temporal analysis in order to identify the acoustic emissions emitted by a same source and to localize the source itself, while maintaining overall simplicity and versatility characteristics.

[0018] Based on this solution idea, the present invention relates to a method according to claim 1.

[0019] A system according to claim 6 is also an object of the invention.

[0020] The characteristics and advantages of the method and system according to the invention will result from the following description of a preferred embodiment given by way of indication and not limitation with reference to the accompanying drawings.

[0021] Brief description of the drawings In such drawings:

[0022] -Figure 1 schematically shows a monitoring system according to the present invention;

[0023] -Figure 2 shows a block diagram of a part of the monitoring system in Figure 1;

[0024] -Figure 3 shows a sensor node in a three-quarter view with a partially represented casing;

[0025] -Figures 4-6 show, schematically and respectively in a front view, a three-quarter view and a top view, some steps of the method according to the present invention;

[0026] -Figure 7 shows a circuit diagram of a self-calibration module for the monitoring system according to the present invention;

[0027] -Figures 8a-8e schematically show a simulated propagation and estimated times of arrival of acoustic emissions detected on a portion of the side surface of a tank (Fig. 8a) and the respective signals detected and processed according to the present invention compared with the actual simulation signals;

[0028] -Figures 9a-9d schematically show the results of a test detected by each sensor unit in a detection group;

[0029] -Figures 10, 11 and 12 show the results of tests performed using the monitoring method according to the present invention;

[0030] -Figures 13 show the results of a test performed on the same system used for the tests in Figures 11-12 with the addition of microphones;

[0031] -Figures 14a-14c show the results of the source localization obtained by applying the method according to the present invention in the same plant of Figure 13 at different time periods;

[0032] -Figures 15a-15c show the results of the source localization obtained by applying the method according to the present invention in the same plant of Figure 11.

[0033] Detailed description

[0034] With reference to the attached figures, a method and system of continuous monitoring of a large size structure 10, such as a tank containing a fluid, for locating a leak or a source E of corrosion that emits acoustic emissions, is described.

[0035] Following a fluid leak and / or during the corrosion process, the source or target E produces acoustic emissions, i.e. mechanical vibrational energy such as a signal or wave that is transmitted into the fluid of the tank 10.

[0036] The storage tank 10, which reference will be made, may be a civil or industrial storage construction with a metal or even concrete side surface adapted to contain a fluid that can be liquid, gaseous and / or multi-phase such as oil or refined products.

[0037] The tank 10 may be provided with pumps, motors, rotating equipment, and other instrumentation not shown in the Figures, which create an acoustic noise during operation that also propagates inside the tank 10 itself. Furthermore, in a working context, the tank 10 is subject to an external acoustic noise, such as ambient noise, which can be transmitted inside the tank and propagated through the fluid.

[0038] According to an embodiment, two or more monitoring devices, M1-M2-M3, are arranged on the outer surface of the tank 10, each equipped with at least one detection group, G1-G2-G3, associated and in communication with a respective sensor node, SN1-SN2-SN3.

[0039] A microprocessor-based central processing unit CU is arranged remotely from the tank 10 and is connected to the sensor node, SN1-SN2-SN3, of each monitoring device, M1-M2- M3, via a single port or gate SG. The gate SG is in communication with the central processing unit CU via a connection that can be either wired with a communication bus 20 or with a wi-fi communication.

[0040] In the embodiment shown in Figure 1, the tank 10 has a circular base 12 which is arranged in a plane P and has an axis A-A. The monitoring devices, M1-M2-M3, are three in number and each comprises a single detection group, G1-G2- G3, which is associated with a respective sensor node, SNi- SN2-SN3.

[0041] Each detection group, G1-G2-G3, is equipped with three sensor units, S0-Sa-Sb, each of which comprises at least one detection sensor, e.g. a piezoelectric sensor PZT, to detect the mechanical vibrational energy transmitted through the fluid.

[0042] The three sensor units, S0-Sa-Sb, are associated with the outer lateral-surface of the tank 10 and are arranged at equal distances q from each other. The three sensor units, So-Sa-Sb, are aligned on their respective directions, D1-D1, D2-D2, D3-D3, which are parallel to the axis A-A and are also arranged at equal circumferential distances from each other.

[0043] Each detection group, G1-G2-G3, defines a central sensor unit So, an upper sensor unit Saand a lower sensor unit Sb-

[0044] Each monitoring device, M1-M2-M3, comprises the sensor node, SN1-SN2-SN3, which can be arranged on the same direction as the sensor units S0-Sa-Sb, or near it or can be remotely arranged.

[0045] A separate input channel, SE0-SEa-SEb, of a communication line or bus connects each sensor unit, Sa-So- Sb, of each detection group, G1-G2-G3, to the respective sensor node, SN1-SN2-SN3.

[0046] Each sensor unit, S0-Sa-Sb, comprises at least the piezoelectric sensor PZT which is a transducer adapted to transform the mechanical vibrational energy or detected waves into electrical energy. Additionally, each sensor unit, So-Sa-Sb, can comprise an amplifier and / or filter for preliminary processing of the generated signals. Based on its transfer function Fs, each sensor unit, S0-Sa-Sb, generates packets or time series of analog signals si(t) at a predefined frequency.

[0047] Each sensor node, SN1-SN2-SN3, is configured to receive continuously and in parallel from each of the three sensor units, S0-Sa-Sb, the time series of analog signals si(t) via the separate input channels, SE0-SEa-SEb. In addition, each sensor node, SNi-SN2-SN3fis configured to process the series of analog signals si(t).

[0048] According to an embodiment shown in Figure 2, each sensor node, SN1-SN2-SN3, comprises a micro-controller processing unit 15 with at least one RAM memory 16. In addition, a transceiver XCVR 17 is interposed between the processing unit 15 and a communication bus 19i, and an LDO voltage regulator 18 allows to suitably supply each component of the sensor node, SN1-SN2-SN3.

[0049] Thus, the processing unit 15 receives continuously, in parallel and from each sensor unit, So-Sa-Sb, the series of analog signals s±(t). Furthermore, the processing unit 15 is configured to set time, with the same clock signal, and to sample, at a predefined sampling frequency fs, the series of analog signals s±(t) in order to generate respective series of timed digital signals s±(nT).

[0050] Each processing unit 15 thus comprises a timer that assigns a Time of arrival Timestamp to each analog signal and also comprises an analog-to-digital A / D converter that operates at a predefined sampling frequency fsto generate respective series of timed digital signals s±(nT).

[0051] In addition, the processing unit 15 comprises a processing module that is configured to process, for each sensor unit, So-Sa-Sb, each series of timed digital signals Si(nT) on the basis of an Akaike Information Criterion to identify significant samples Sk(nT) of each time series of analog signals s±(t).

[0052] The Akaike Information Criterion is applied to all the samples of the timed digital signals s±(nT) and allows to generate, for each sample, an index K' which is related to the time of arrival ToA of each digital sample and is defined as the value of k that maximises the following formula: AIC(k)=k*log(var(y(1:k)))+(nsamp-k- 1)*(log(var(y(k+1:nsamp))) wherein: y is the value of the sample amplitude of the timed digital signal s±(nT); nsamp is the number of samples of the timed digital signal s±(nT); k is an index in the range [1; (nsamp-1)]; var is the variance of the digital samples calculated in a first time range [1; k] and in a second time range [ (k+1); nsamp].

[0053] The index K' defines the significant digital sample Si(K'T) of each analog signal s±(t) of the detected series.

[0054] Additionally, the processing module of the processing unit 15 is also configured to pre-localize the source E by determining, by means of a temporal analysis, a radial distance, DiE-D2E-D3E, with respect to the central sensor unit So.

[0055] The radial distance, DIE-D2E_D3E, is calculated on the basis of the propagation velocity v of said acoustic signals in said fluid, and on the basis of the time difference of corresponding significant sample Si(K'T) of the series of acoustic signals received by the three sensor units S0-Sa- Sb.

[0056] In particular, each processing unit 15 is configured to perform a temporal analysis of the series of timed digital signals s±(nT), in the same time slot, to determine a difference of the times of arrival, DToAbo and DToA0a, of the significant samples s±(K'T) of the signals received from the lower sensor unit Sbwith respect to the central sensor unit So and from the upper sensor unit Sawith respect to the central sensor unit So.

[0057] Furthermore, knowing the distance q between the sensor units, S0-Sa-Sb, for each detection group, G1-G2-G3, it is possible to calculate the radial distance DIE-D2E_D3Eof the source E from the central sensor unit So using the formula:

[0058] Wherein: q is the distance, in module, between two of said sensor units in sequence;

[0059] Aa and Ab are the differences in radial distances or simply radial differences between the source E and the sensor units, S0-Sa-Sb calculated as a function of the difference in the number of significant digital samples sa / b(K'T) between the upper sensor Saand the central sensor So and between the central sensor So and the lower sensor Sb-

[0060] The method involves calculating the radial distances, DiE-D2E-D3E, and transmitting these radial distances and the co-ordinates of the respective central sensor unit So, from each processing unit 15 to the central processing unit CU via the detection gate SG.

[0061] The central processing unit CU is configured to process the received data and process it using a software that comprises a multilateration algorithm to determine the position of the source E in one main Cartesian reference system (X*, Y*, Z*).

[0062] The central processing unit CU transforms, by means of a transformation law, the coordinates of the central sensor unit So of each monitoring device, M1-M2-M3, and the radial distances, DiE-D2E-D3E, to the main Cartesian reference system (X*, Y*, Z*). Furthermore, applying the multilateration algorithm on the obtained data, it determines the position of the source E.

[0063] In one embodiment, the main Cartesian reference system (X*, Y*, Z*) has the y-axis Y* as the axis of the base 12, the x-axis X* and the axis Z* perpendicular to each other and comprised in the plane P of the base 12.

[0064] The multilateration algorithm is substantially an algorithm which makes it possible to estimate the position (x', y') of the source E as an unknown point by knowing the positions (x±, y±) of N detection devices, with N>3, known the radial distances df from each detection device to the unknown point. As an illustration and not as a limitation, the multilateration algorithm is described in the article by Meyer, Thomas H., and Ahmed F. Elaksher. "Solving the Multilateration Problem without Iteration", Geomatics 1.3, pages 324-334.

[0065] According to a further aspect of the present invention, the method and the system allow to determine the position of source E even in the special case where, through prelocalization, a radial distance, DiE-D2E-D3E, of the three monitoring devices, M1-M2-M3, is null. In such a case, the source E is positioned on the direction of the corresponding monitoring device and the differences of the calculated radial distances with respect to the central unit So are equal in module to each other:

[0066] Therefore, instead of applying the software with the multilateration algorithm, the central processing unit CU processes the received data using a software comprising trigonometric processing. As shown as illustration and not limitation in Figure 6, the source E is positioned in the direction of the third monitoring device,M3, and the radial distance D3Eis null.

[0067] Using trigonometric processing, the length of the chord Ci,2 between the two central sensor units, SOiand S02, is calculated using the formula:

[0068] Furthermore, considering that the area of the triangle defined by the pair of central sensor unit S01 e S02, and by the source E, of co-ordinates (x'E, y'E), can be calculated using the formula: wherein p is the semi-perimeter of the triangle and considering the following auxiliary quantities: the coordinates of the source E in the main Cartesian reference system (X*, Y*, Z*) are obtained by means of the following formulas: ,o,

[0069] (o) wherein h is the distance of the central unit So from the base 12.

[0070] Determination of the radial distance DIE

[0071] Figure 6 shows, as illustration and not limitation, the source E arranged in the base 12 of the tank 10 with two monitoring devices Mi and M2.

[0072] Considering a first monitoring device Mi, among all the planes eg, eg,eg, etc. of a bundle a of own planes having the same y-axis of ordinates Yi, we consider a first virtual Cartesian reference plane eg (Xi, Yi) with the centre 01 at the central unit So and with the x-axis Xi defined by the perpendicular to the y-axis passing through the central sensor unit So.

[0073] Furthermore, the source E is considered to be comprised within the first virtual Cartesian reference plane eg (Xi, Yi). Therefore, the virtual co-ordinates (xiE, yiE) of the source E can be calculated as a function of the distance q between the sensor units, Sa-So-Sb, and the differences in radial distances or simply radial differences, Aa and Ab, between the source E and the sensor units, Sa-S0-Sb, calculated using the formulas: wherein:

[0074] Nais the difference in the number of digital samples between the significant sample sa(K'T) of the upper sensor saand the corresponding significant sample So(K'T) of the central sensor So;

[0075] Nb is the difference in the number of digital samples between the significant sample So(K'T) of the central sensor unit So and the significant sample Sb(K'T) of the lower sensor unit Sb,- v is the propagation velocity v of the acoustic wave knowing the type of fluid in the tank 1; fsis the sampling frequency.

[0076] The virtual ordinate yiEof the source E is calculated using the following trigonometric formula: wherein q is the distance, in module, between the sensor units in sequence.

[0077] Therefore, the radial distance DiEof the source E from the central sensor unit So is calculated using the formula (1) below:

[0078] The virtual abscissa xiEof the source E is calculated using the formula:

[0079] With pre-localization, the radial distance DiEof the source E from the central sensor unit So is determined, however each plane eg, eg, eg, comprised in the bundle a of planes comprises a point, which belongs to the circumference with radius equal to the radial distance, which satisfies these conditions. Therefore, the multilateration algorithm or trigonometric processing makes it possible to localize the source E.

[0080] Some tests performed on tanks using the described method and system are shown in Figures 8-9 and 11-12, while some representations of localizations of source E are shown in Figures 10 and 15a-15c.

[0081] The method and system described above have some variants, which are described below. In the description that follows, details and co-operating parts having the same structure and function as those described above will be indicated by the same reference numbers and abbreviations.

[0082] In one variant, each processing unit 15 comprises a filter with a threshold module that pre-processes the series of timed digital signals s±(nT) prior to processing on the basis of Akaike Information Criterion.

[0083] This pre-processing allows to maintain the significant components of the timed digital signals s±(nT) with the reduction or cancellation of weak components, thus making it possible to reduce noise.

[0084] In one embodiment, the threshold function is applied to the timed digital samples s±(nT) of each signal by bringing to zero the amplitudes of the samples that are lower than a first threshold X and reducing the amplitudes of the remaining digital samples by a value equal to the first threshold X, i.e: if (|Si(nT)| > X) then yA(nT) = sign(s±(nT)) * (|s±(nT)| - X) (16) otherwise yA(nT) = 0

[0085] With the filter equipped with the threshold function, each timed digital signal s±(nT) is associated with a signature, i.e. an array comprising successively null values and reduced digital samples yA(nT). The signature requires less memory to be processed as all weak components are cancelled. This is advantageous both for the processing with the Akaike criterion and for subsequent pre-processing and processing.

[0086] Of course, all the values and therefore also null values in the signature will be taken into account when calculating the position of source E.

[0087] In a further variant, each monitoring device, M1-M2-M3, can comprise a first detection group, G1-G2-G3, equipped with a sensor unit with piezoelectric sensors PZT configured to detect low-frequency acoustic emissions and a second detection group, GI,-G2,-G3', equipped with sensors configured to detect high-frequency acoustic emissions. The three sensor units of the first and second detection groups can be arranged in the same direction, in succession or alternating with each other. As is well known to those skilled in the art, the sources E generated by corrosion phenomena are high-frequency and observed in the frequency range 50-300 KHz with amplitudes of 80dB. While low- frequency sources E are observed in the lOHz-lOKHz range with amplitudes of 70dB.

[0088] In a further variant, a microphone can be added to said sensor nodes, SNI-SN2-SN3, to detect any acoustic signals due to instrumentation associated with the tank 10, such as: pumps, motors, rotating equipment, or any acoustic signals due to external-environmental noise.

[0089] In this case, the peaks of the acoustic signals detected by the respective microphone are compared with the significant digital samples Si(K'T) of each sensor unit, So- Sa-Sb, of each detection group, GI-G2-G3. This comparison makes it possible to highlight any false negatives and to cancel significant samples corresponding to said peaks, as schematically shown in Figure 13.

[0090] In one embodiment, microphones can be of the MEMS type.

[0091] In a further embodiment, one or more of the sensor nodes, SNI-SN2-SN3, may comprise thermistor sensors with a negative temperature coefficient and possible circuitry to monitor the temperature of the tank and / or contained fluid and / or the ambient temperature.

[0092] Furthermore, the series of timed digital signals s±(nT) can be further processed in real time by each sensor node, SNI-SN2-SN3, to derive additional parameters or characteristic data. Parameters may comprise: peak amplitude, ramping time, overall signal and / or peak duration, high energy components (which could be associated with the opening or closing of a plant valve), number of pulses, average frequency of analog signals s±(t) in each series.

[0093] These additional data / parameters can be sent and processed by the central processing unit CU to maintain or eliminate any localizations of sources E.

[0094] In a further aspect of the present invention, the continuous structure monitoring system provides a single clock device for the monitoring devices, M1-M2-M3, in order to synchronise and standardise the significant samples generated and make the data obtained reliable. According to one embodiment, the timestamp is unique with a 32-bit highspeed hardware counter, configured to generate a 64Mhz clock, and a low-speed 32-bit software counter updated every hour. The clock device provides a synchronisation command for both the hardware and software counters, which can be generated by the gate or the central processing unit CU for all the sensor units, S0-Sa-Sb, of each detection group, Gi- G2-G3.

[0095] According to a further aspect of the present invention, each monitoring device, M1-M2-M3, may comprise at least one self-calibration and self-diagnosis module 40. The module 40 is schematically shown in Figure 7. The self-calibration and self-diagnosis module 40 comprises a conditioning unit 42 that is equipped with an operational amplifier, in a noninverting mode, with a bias block and, in the embodiment shown, is equipped with a digital-to-analog converter DAC. The module 40 also comprises an analog-to-digital converter ADC 43, which is adapted to convert the output voltage of the operational amplifier to provide, as an output, a suitable voltage value to bias the respective sensor unit SNi of the monitoring device, M1-M2-M3.

[0096] The module 40, is associated with each sensor unit, SQ- Sa-Sb, and its sensor node, SNI-SN2-SN3, and allows to automatically perform periodic, predefined, real-time selfassessment tests, determining any damage to each monitoring device, MI-M2-M3, with no need for additional external devices.

[0097] In one embodiment, the self-calibration step involves injecting the following voltages at a predefined time to:

[0098] 117) wherein: VOLD represents high voltage and

[0099] VNEw represents low voltage.

[0100] The piezoelectric sensor PZT of each sensor unit, Sa- So-Sb, can be modelled with a sensor module 41 known as a Van Dyke model, and configured as a capacitor Co with a static capacitance modelling the equivalent electrical capacitance between two upper and lower electrodes. Thus, the voltage V0UT of the PZT output can be represented with the following function: wherein: multiplicative factors or characteristic values that depend on the circuit topology and are estimated a priori;

[0101] T=RICO is a time constant of the piezoelectric sensor PZT, which is calculated as follows: where t* is the time instant at which voltage Vois equal to «VOLD for t^t0and i=t*-t0

[0102] For example, considering the circuit schematised in Figure 6, the parameters or characteristic values ac,Pc and TCare determined by means of the following formulas:

[0103] Such characteristic parameters in the real scenario are subject to drifts that may affect the location of source E.

[0104] In one embodiment, a self-calibration procedure is to be activated to determine the actual characteristic parameters to compensate for a randomness generated by the signal gain.

[0105] The self-calibration procedure involves activating the self-calibration and self-diagnosis module 40 by activating the analog-to-digital converter DAC with the decreasing voltage value VDAC, as indicated in the formula (18), at the input terminal IN. The piezoelectric sensor VPZTis activated and an output voltage VOui is generated.

[0106] The self-calibration procedure then involves detecting the output voltage VOui and calculating, via the sensor unit, SN1-SN2-SN3, the actual parameters using the following formulas: wherein:

[0107] V?eakandt,Peakare the values or the peak voltage output vo from the ADC converter; tv*is the instant of time when Vo= aAV0LDwith t> t0

[0108] Then, the actual calculated parameters are stored in a memory of the sensor unit, So-Sa-Sb, and compared with the respective threshold values a'A, P'Aand T'A.

[0109] In the event that:

[0110] 1. the calculated actual values aA, pA, and are higher than the respective threshold values a'A, P'Aand T'Aa malfunction alarm signal is generated and also sent to the central processing unit CU. One or more mitigation actions are actuated, such as a temporary exclusion of the piezoelectric sensor PZT of the respective sensor unit, So- Sa-Sb, from the continuous monitoring process;

[0111] 2. the calculated actual values aA, pA, and are lower than the respective threshold values a'A, P'Aand T'Athe output voltage V0Uiis pre-processed by the sensor node So-Sa-Sb, and knowing the characteristic values for the circuit, the compensation factors can be calculated:

[0112] Wherein: ac, pcare the characteristic values indicated above.

[0113] According to an aspect of the present invention, the threshold values a'A, P'Aand T'Acan be estimated at a preliminary step by extracting an initial series of actual values of parameters at an early or preparatory step of the system activation. A statistical characterisation of the initial series of actual values, with the determination of the standard and expected deviation, makes it possible to determine the actual values. The threshold can then be estimated for each parameter using the formula: Wherein: are the expected value and standard deviation of the respective parameter; k is a multiplication factor that can be set according to a specific application 3<J,6<Jor9<J, depending on safety.

[0114] The self-calibration module 40 and the respective selfcalibration step allow the system and method to automatically perform periodic, predefined self-assessment tests in real time and without additional external devices. This makes it possible to autonomously determine damage to piezoelectric sensors and their sensor units in real-world scenarios due to operational degradation or damage, such as the detachment of sensor units S0-S3 from the wall of the tank side surface 10.

[0115] The described method and system perform a continuous non-destructive inspection of the structure adapted to easily and independently detect either corrosion processes or a leak, accidental or deliberate.

[0116] The monitoring of the health condition of the structure, tank, as described is minimally invasive relative to the structure and can be used separately or in combination with existing methods.

[0117] In fact, the system and method can be easily used even in existing plants, allowing a variety of uses in different scenarios both on-shore and off-shore, with a limited weight, limited volume and limited wiring.

[0118] The method and system can also be used during the operational step of the structure in the presence of noise or vibrational interference inside or outside the structure.

[0119] The method and system are versatile and can be used in structures independently of the fluid included in the tank, even if the fluid is particularly corrosive or harmful to health.

Claims

CLAIMS1. A method of continuous monitoring of a structure (10), which comprises a fluid, to localize a leak or a source (E) of corrosion that emits acoustic emissions, said acoustic emissions being detected by means of detection sensors associated with the external surface of the structure (10); the method is characterized by:- providing three monitoring devices (M1-M2-M3) each equipped with at least one detection group (G1-G2-G3) with three sensor units (S0-Sa-Sb), each sensor unit (S0-Sa-Sb) comprising at least one of said detection sensors;- arranging, for each monitoring device (M1-M2-M3), said three sensor units (So-Sa-Sb) aligned with each other along a direction parallel to an axis (A-A) of the structure (10); receiving continuously and independently from each detection group (G1-G2-G3) and in parallel from each sensor unit (S0-Sa-Sb) respective series of analog signals (s±(t)),- processing said series of analog signals (s±(t)) providing for:- timing and sampling at a predefined sampling frequency (fs) said series of said analog signals (s±(t)) generating a corresponding series of timed digital signals (s±(nT)); processing each series of timed digital signals (si(nT)) on the basis of an Akaike Information Criterion to identify significant digital samples (si(K'T));- pre-localizing said source (E) for each monitoring device (MI-M2-M3) to determine a radial distance (DIE-D2E_D3E) of said source (E) from said central sensor unit (So) on the basis of time differences (DToAbo and DToAoa) determined through a temporal analysis of corresponding significant digital samples (si(K'T)) relating to the upper sensor unit (Sa) and to the lower sensor unit (Sb) with respect to the central sensor unit (So);- processing the radial distances (DIE-D2E_D3E) of said three monitoring devices (M1-M2-M3), by means of a processing software, to localize said source (E).

2. Method according to claim 1, characterized in that:-if the radial distances (DiE-D2E-D3E) of said three monitoring devices (MI-M2-M3) are all different from zero, said processing software comprises a multilateration algorithm based on said radial distances (DIE-D2E_D3E) and on the coordinates of said central sensor units (So);-if a radial distance (DiE-D2E-D3E) of said three monitoring devices (MI-M2-M3) has a value equal to zero, said processing software, of the remaining two radial distances (DIE-D2E_D3E), comprises a trigonometric elaboration based on the length of a chord (01,2) between the two central sensor units (So) and on the coordinates of said two central sensor units (So).

3. Method according to claim 1, characterized by: -preprocessing said series of timed digital signals (s±(nT)) by means of a filter with a threshold function that is applied to said timed digital samples (s±(nT)) of each sensor unit (S0-Sa-Sb), said threshold function associating a null value to the amplitudes of said timed digital samples (s±(nT)) which are lower than a first threshold (X) and reducing the amplitudes of the remaining timed digital samples (s±(nT)) by a value equal to the first threshold (X).

4. Method according to claim 1, characterized by: equipping said sensor units (S0-Sa-Sb) with detection sensors (PZT) configured to detect low-frequency acoustic emissions or configured to detect high-frequency acoustic emissions, and / or by the fact that said pre-localization of said source (E) is preceded by a comparison of said significant digital samples (si(K'T)) with peaks of acoustic signals detected by at least one microphone.

5. Method according to claim 1, characterized by comprising a self-calibration procedure for each sensor unit (S0-Sa-Sb) of each detection group (G1-G2-G3), said selfcalibration procedure involves polarizing each sensor unit (S0-Sa-Sb) by means of an input voltage (VDAC) for a predefined time and analysing an output voltage to obtain characteristic values (acvPc,Tc) of said sensor unit (So-Sa- Sb), and comparing said characteristic values (ac,Pc,Tc) with respective threshold values to determine compensation factors of said characteristic values or to generate a warning signal.

6. System of continuous monitoring of a structure (10), which comprises a fluid, to localize a leak or a source (E) of corrosion that emits acoustic emissions, said acoustic emissions being detected by means of detection sensors associated with the external surface of the structure (10), characterized in that it comprises three monitoring devices (M1-M2-M3) associated with a microprocessor central processing unit CU, each monitoring device (M1-M2-M3) comprises:- at least one detection group (G1-G2-G3) with three sensor units (S0-Sa-Sb), each sensor unit (S0-Sa-Sb) being equipped with at least one of said detection sensors, said three sensor units (So-Sa-Sb) being arranged aligned with each other along a direction parallel to an axis (A-A) of the structure (10); a sensor node (SNI-SN2-SN3) associated and in communication with each sensor unit (So-Sa-Sb) of said at least one detection group (G1-G2-G3), each sensor node (SNi- SN2-SN3) comprises a microprocessor processing unit (15) which is configured to receive continuously and in parallel from each sensor unit (S0-Sa-Sb) respective series of analog signals (s±(t)) and generate corresponding series of timeddigital signals (s±(nT)), the processing unit (15) comprises a processing module which is configured to process each serie of timed digital signals (s±(nT)) on the basis of an Akaike Information Criterion to identify significant digital samples (si(K'T)) for each sensor unit (S0-Sa-Sb), said processing module is also configured to pre-localize said source (E) determining a radial distance (DiE-D2E-D3E) of said source (E) from said central sensor unit (So), said radial distance (DiE-D2E-D3E) being calculated on the basis of time differences (DToAbo and DToA0a) of corresponding significant digital samples (si(K'T)) of said timed digital signals (si(nT)) relating to the upper sensor unit (Sa) and the lower sensor unit (Sb) compared to the central sensor unit (So); said central processing unit (CU) being configured to receive from said three monitoring devices (MI-M2-M3) the respective radial distances (DiE-D2E-D3E) and is configured to process, by means of a processing software, said radial distances (DiE-D2E-D3E) to locate said source (E).

7. System according to claim 8, characterized in that said central processing unit (CU) is configured to process the radial distances (DiE-D2E-D3E) of said three monitoring devices (MI-M2-M3) with said processing software (15), which comprises a multilateration algorithm if said radial distances (DiE-D2E-D3E) are different from zero, or if a radial distance (DiE-D2E-D3E) of said three monitoring devices (MI-M2-M3) has a value equal to zero, said processing software for processing the remaining two radial distances (DiE-D2E-D3E) comprises a trigonometric processing based on the length of a chord (ci,2) between the two central sensor units (So) and on the coordinates of said two central sensor units (So).

8. System according to claim 6, characterized in that each processing unit (15) comprises a filter with a thresholdmodule configured to pre-process said series of timed digital signals (s±(nT)), said threshold module comprising a threshold function that associates a null value to sample amplitudes lower than a first threshold (X) and reduces the amplitudes of the remaining timed digital samples (s±(nT)) by a value equal to the first threshold (X).

9. System according to claim 6, characterized in that said sensor units (S0-Sa-Sb) comprise detection sensors (PZT) configured to detect low-frequency acoustic emissions or detection sensors (PZT) configured to detect high-frequency acoustic emissions and / or in that said processing unit (15) comprises at least one microphone configured to detect external acoustic signals, said processing module being configured to compare, before pre-localizing said source (E), said significant digital samples (si(K'T)) with peaks of said acoustic signals detected by said at least one microphone.

10. System according to claim 6, characterized in that each monitoring device (M1-M2-M3) comprises at least one self-calibration module (40) interposed between each processing unit (15) and each sensor unit (So-Sa-Sb) which is configured to perform a self-calibration procedure which involves polarizing each sensor unit (So-Sa-Sb) by means of an input voltage (VDAC) for a predefined time and analysing an output voltage to obtain characteristic values (ac,Pc,Tc) of said sensor unit (S0-Sa-Sb), and to compare said characteristic values (ac,Pc,ic) with respective threshold values to determine compensation factors of said characteristic values or to generate a warning signal.

Images