System for detecting defects in a fluid flowing through a circuit containing a fluidic device and for preventing failure of the device as a result of the defect

Abstract

The present invention essentially consists of a system for detecting one or more defects that may cause a malfunction or even a breakdown of a fluidic device installed in a fluid circuit. The system includes electrical sensors with electrodes, preferably regularly distributed around the upstream part of the device, with their ends flush with the inner wall of the part. EIT measurements (EIT stands for electrical impedance tomography) are performed using signals, either triangular or applied via pairs of electrodes, either simultaneously to excite all electrodes or sequentially to a subset of electrodes. This measurement allows a continuous, real-time view of the interior of the part and thus the fluid flowing therein, by measuring electrical properties (current and potential), non-invasively and non-destructively.

Description

[Technical Field]

[0001] The present invention relates to the field of monitoring fluidic devices intended to be used in fluid circuits and in particular in industrial plants.

[0002] The invention more particularly relates to the detection of any kind of defect that may be present in a fluid flowing through a fluid circuit and that may cause failure or even damage to a fluidic device in said circuit.

[0003] Although described with reference to its application to pump monitoring, the invention can be applied to the monitoring of any kind of fluid device that may be present in fluid circuits, and particularly in industrial installations.

[0004] The term "fluidic device" is understood here and in the context of the present invention to mean any component or set of mechanical or electromechanical components that can be used in a fluid circuit. This can be a matter of a device (pump) or a device using fluid forces (turbine) whose at least one rotating component continuously and / or discontinuously displaces a volume of fluid. This can be a matter of a device with fixed or moving components that modify the flow of a fluid, such as a valve, check valve, piston, elbow, or gooseneck. This can also be a matter of a duct in a fluid circuit with at least one undesirable leak. This can be a matter of a fluid duct that is at risk of clogging, blockage, or fouling. This can also be a matter of an intrusive or non-intrusive fluid sensor for measuring at least one characteristic of a fluid.

[0005] By "defect" what is meant here and in the context of the present invention is any abnormality present in the fluid flowing through a fluid circuit that changes the expected properties of the fluid (e.g. its temperature, pressure, flow rate, viscosity, or its single-phase behavior), i.e. the properties it was originally intended to have. This can be a problem of foreign objects suspended in the flowing fluid, such as metal bodies (one or more residues of welding, bolts, etc.). It can also be a problem of one or more agglomerates of solid material, of wall fouling, of blockages by solid particles, of clogging, or of the formation of one or more plugs. It can also be a problem of gas bubbles, either due to one or more leaks or cavitation. [Background technology]

[0006] Pumps are essential components of industrial plants and are currently the second most manufactured device in the world after electric motors.

[0007] Pumps transport fluids necessary for human activity, especially industrial activity, such as most raw materials, water, fluids in food production and processing, in the chemical industry, pharmaceutical industry, construction industry, paper industry, electronics industry, mining industry, petroleum industry, etc.

[0008] A failure, including a broken pump, can quickly lead to a shutdown of the entire factory, and the costs associated with the production outage can be very high. The inventor has been able to quantify that among manufacturers in one sector, approximately 5% of factories installing new pumps were forced to shut down upon initial start-up due to the presence of metallic foreign objects (e.g., welding residue, bolts, etc.) remaining in the pump and damaging its mechanism.

[0009] Furthermore, the causes of pump damage are, first of all, the so-called cavitation effect, i.e. the presence of vapor bubbles caused by too high a suction force or the absence of fluid at the suction end of the pump, The presence of foreign bodies, Blockage due to solid matter is.

[0010] To solve this problem, the pump can be disassembled and parts replaced or cleaned as part of routine maintenance, however this can result in unnecessary costs as the pump must be inspected more frequently than is actually necessary for safety reasons.

[0011] There are numerous systems for monitoring fluid circuits that incorporate one or more pumps.

[0012] These are mainly based on pressure differences, vibration measurements, or infrared or ultrasonic measurements, as described in Patent Documents 1 to 3, respectively.

[0013] These existing systems do not actually predict pump failure and / or do not accurately identify all the anomalies that cause failure.

[0014] In particular, implemented commercial solutions allow the detection of high degrees of cavitation, but do not measure the risk of foreign objects, blockages or clogging in the fluid.

[0015] There is therefore a need for a system for monitoring a pump in a fluid circuit that overcomes the drawbacks of known systems, in particular to make it possible to detect any kind of defect other than cavitation that is likely to cause the failure or even destruction of the pump.

[0016] More generally, there is a need for a system that is able to monitor fluidic devices used in fluidic circuits and detect any kind of defect that could cause failure or even damage to at least part of the device, and thus prevent said failure. [Prior art documents] [Patent documents]

[0017] [Patent Document 1] French Patent Application Publication No. 3026443 [Patent Document 2] Chinese Patent Application Publication No. 111197581 [Patent Document 3] European Patent No. 1297331 Summary of the Invention [Problem to be solved by the invention]

[0018] One object of the present invention is to at least partially meet these needs. [Means for solving the problem]

[0019] Thus, one subject of the present invention, according to one of its aspects, is a system for detecting at least one defect that may be present in a fluid flowing through a fluid circuit comprising at least one fluidic device and, where appropriate, at least one fluid branch upstream of the fluidic device, and for preventing failures of the fluidic device specific to the defect or defects, at least one electrical sensor including a plurality of electrodes distributed around a portion of the fluid circuit upstream of the fluidic device, the electrodes having ends flush with an inner surface of the portion; an electronic circuit for controlling the electrodes and for performing electrical impedance tomography measurements, the electronic circuit being configured to simultaneously excite each electrode and to measure the electrical impedance matrix of a fluid flowing through the portion; a signal processing unit configured to identify one or more defects and / or reconstruct an image of a fluid flowing through the portion based on measurements of the impedance matrix of the electronic circuit; an electronic unit for automatically controlling a controller of the fluidic device or the fluidic device itself or, where appropriate, a fluid branch, configured to change the operating frequency of the fluidic device so as to prevent a failure of the fluidic device, or to control the operation of the fluidic device, or to interrupt the operation of the fluidic device, or to physically remove one or more defects upstream of the fluidic device, respectively, in response to signals of the flowing fluid indicative of one or more defects; Includes.

[0020] According to one variant of embodiment, the fluidic device is a pump or a turbine and the controller is a variable frequency drive.

[0021] Advantageously, the operation of the electronic unit for automatically controlling the fluidic device is subordinated to the operation of the signal processing unit.

[0022] According to one advantageous construction mode, the system comprises a module to be placed in the fluid circuit, upstream of the fluidic device, said module comprising: a duct defining a fluid portion; a casing fixed around the duct, containing an electric sensor with electrodes individually housed and fixed in a sealed manner to the through-hole of the duct; Includes.

[0023] Preferably, the casing contains a port for supplying power to the electrodes.

[0024] Also preferably, the duct includes a flange at at least one of its ends for fastening to a duct of the fluid circuit.

[0025] Such a module can therefore be easily integrated directly into existing or new fluid circuits, and can therefore be used as a retrofit module for existing circuits.

[0026] The module can also be directly integrated into fluidic devices (pump bodies, valve bodies, etc.).

[0027] According to an advantageous variant of embodiment, the signal processing unit is configured to implement an image reconstruction neural network.

[0028] According to one advantageous embodiment, the control circuit is configured to carry out the following steps:

[0029] a / energizing the electrodes, each electrode being of the form [Formula 2]

number

number

number

number

number

[0030] b / electrical characteristics of fluid flow at electrodes V n meas A step of measuring.

[0031] c / Processing the data generated in the measurement step b / , which comprises the following sub-steps:

[0032] c1 / Each electrode E n Regarding [Formula 10]

number

[0033] If the c2 / excitation format [Equation 2] is used, Eq. [Formula 13]

number

number

[0034] c3 / forming a signed data matrix, the elements of which are the phase shift Φ between the excitation potential of electrode l and the measured current of electrode n n,l When (k) is less than π / 2, the following equation [Formula 20]

number

number

[0035] Preferably, the control circuit controls the potential V n exc All of the above are conditions [Formula 4]

number

[0036] According to one advantageous configuration, the electrodes of the electrical sensor are regularly angularly distributed around the part of the fluid circuit upstream of the pump.

[0037] Another subject of the present invention is to use a detection and prevention system as described above to measure foreign objects and / or blockages, clogging and / or fouling induced by foreign objects, cavitation, bubbles, lack of fluid and the presence of gas.

[0038] Another subject of the invention is an industrial plant comprising at least a fluid circuit and a detection and prevention system such as those described above.

[0039] In general, the invention can be implemented in any fluid circuit, in particular in factories or industrial production sites, especially in the fields of food processing and manufacturing, pharmaceuticals and cosmetics, chemical and petrochemical, and water distribution and treatment.

[0040] The invention therefore essentially consists of a system for detecting one or more defects that may cause the failure, or even destruction, of a fluidic device installed in a fluidic circuit.

[0041] The system comprises electrical sensors having electrodes distributed, preferably regularly, around the upstream portion of the device, with ends flush with the inner wall of said portion.

[0042] EIT measurements (EIT stands for Electrical Impedance Tomography) are performed using signals, which are either triangular or applied via pairs of electrodes, either simultaneously to excite all electrodes or sequentially to subsets of electrodes.

[0043] This measurement allows a continuous, real-time view of the interior of the part and therefore the fluid flowing therein, non-invasively and non-destructively, by measuring electrical properties (current and potential).

[0044] The flow through the pipe is analyzed either directly via the measured signals or by solving an inverse problem to reconstruct a map of the electrical impedance inside the section.

[0045] The speed of acquisition of images reconstructed by the processing unit can be very high, typically up to 31,250 images / second, which makes it possible to observe fluids flowing at speeds of up to 300 meters / second.

[0046] In response to signals in the flowing fluid indicative of one or more defects, the electronic unit will automatically control either the controller of the fluidic device or the fluidic device itself or, if appropriate, an upstream fluid branch.

[0047] This unit therefore makes it possible to change the operating frequency of the fluidic device to prevent failure of the fluidic device, to control or interrupt the operation of the fluidic device, or to physically remove one or more defects upstream of the fluidic device.

[0048] The invention described here has many advantages, among which are: Non-invasive measurement of fluids without pressure drop or dirt buildup; direct measurement of defects that cause failure or even destruction of fluidic devices; The ability to take measurements / images at extremely high speeds of up to 31,250 images / second, allowing for fast detection and preventative action; A robust system that can be relied upon regardless of the fluid used, even at very high pressures (typically above 300 bar) and / or very high temperatures (typically above 600°C), the ability to detect all types of defects, such as foreign objects, blockages, dirt, clogs, leaks or cavitation bubbles, with the same system, unlike prior art systems that are configured to detect only one type of defect; Examples include: [Brief explanation of the drawings]

[0049] [Figure 1] 1 shows a fluid circuit including a pump in which a flowing fluid that may have a defect can be detected by a system for detecting and preventing pump failures according to the present invention, including an electrical sensor with electrodes for electrical impedance tomography measurements. [Figure 2] 1 shows a schematic representation of the hardware and software means of a system for detecting and preventing pump failures according to the present invention; [Figure 3] 1 shows a module incorporating an electrical sensor of a system according to the invention that can be installed directly into a new or existing fluid circuit. [Figure 3A] 4 is a cross-sectional view of the module according to FIG. 3, the cross-section being taken through the electrodes of the electrical sensor. [Figure 4] 1 shows a spatial cosine pattern. [Figure 5] 1 shows part of the electronic circuitry that makes it possible to energize the electrodes. [Figure 6] 3 shows a method for generating signals for exciting electrodes of an electrical sensor according to a first variant of the invention; [Figure 7] 1 illustrates a method for measuring a signal generated by an electrode. [Figure 8] An excitation signal of the form [Equation 2] and some of its characteristics are shown in graphical form. [Figure 9] The sign matrix for a 16-electrode electrical sensor associated with excitation of the form [Equation 2] is shown. [Figure 10] The sign matrix for a 32-electrode electrical sensor associated with excitation of the form [Equation 2] is shown. [Figure 11] 1A and 1B show images reconstructed according to the prior art and experimentally, respectively, and simulated by a neural network in a system according to the invention. DETAILED DESCRIPTION OF THE INVENTION

[0050] 1 shows an example of an implementation of a system 1 for detecting defects in a fluid flowing in a circuit 2 including a pump 20. This system further makes it possible to prevent failure of the pump 20 as a result of the defect according to the invention.

[0051] The defects may be of any type, such as foreign bodies (especially metallic foreign bodies), clogs, blockages, or bubbles due to cavitation or leaks.

[0052] The system 1 firstly comprises an electrical sensor 10 placed in a portion of the circuit 2 upstream of a pump 20. This electrical sensor 10 makes it possible to carry out electrical impedance tomography measurements on the fluid flowing through this portion.

[0053] Unit 11 enables operational control of the pump's variable frequency drive according to real-time fault measurements in response to flowing fluid signals indicative of one or more faults, as described in more detail below, to prevent failure of pump 20.

[0054] More precisely, the electrical sensor 10 comprises electrodes 100 angularly distributed, regularly or irregularly, around the portion 21 so that their ends are flush with the inner surface of the portion 21 .

[0055] A block diagram of the hardware and software means of the system 1 is shown in FIG.

[0056] The electrodes 100 are connected to a printed circuit board 12, which is an electronic circuit for controlling the electrodes 100 and for performing electrical impedance tomography measurements, the circuit being configured to simultaneously excite each electrode and measure the electrical impedance matrix of the fluid flowing through portion 21.

[0057] The signal processing unit 13 makes it possible to identify one or more defects and to reconstruct an image of the fluid flowing through the portion 21 based on measurements of the impedance matrix of the electronic circuit 12 .

[0058] The automatic electronic control unit 14 makes it possible to control, preferably via digital communication, the variable frequency drive of the pump 20 based on the presence or absence of risks 13 identified for the pump 20, and more generally to supervise the industrial plant in which the fluid circuit 2 is installed.

[0059] 3 and 3A show one advantageous mode of incorporating the electrical sensor 10, in the form of a module M, into a fluid circuit.

[0060] The module M first includes a duct 22 which defines a fluid portion 21 .

[0061] A casing 23 fixed around the duct 22 houses the electric sensor 10, with the electrodes 100 individually housed and fixed in a sealed state in the through holes 24 of the duct.

[0062] The casing 23 may house a port 25 for supplying power to the electrodes.

[0063] For easy and quick assembly and fixing of the module M in the fluid circuit, the duct 22 includes flanges 26 at both ends for fixing to the ducts of the fluid circuit.

[0064] The control circuitry 12 can be connected to a data acquisition system. A screen 11 can make it possible to view the data and images generated from these data. The data acquisition system includes a Linux operating system (HOST) that controls an FPGA that is also included in the data acquisition system. The FPGA can be replaced by a microcontroller.

[0065] The acquisition system makes it possible to generate an analog excitation signal and to measure an analog measurement signal delivered by the electrode 100 .

[0066] The acquisition system may, for example, incorporate a cRIO-9039 controller from the manufacturer National Instruments, including an FPGA, an NI-9262 module from the manufacturer National Instruments for generating an analog excitation signal, and an NI-9223 module from the manufacturer National Instruments for measuring the analog signal delivered by the electrode 100.

[0067] The method of measuring and processing signals implemented by the system for detecting pump defects and preventing their failure according to the invention will now be described.

[0068] It will be noted that the electrical sensor 10 is pre-installed in the fluid circuit with the electrodes 100 positioned as shown above.

[0069] Exciting the electrodes The first step consists in exciting all the electrodes 100 simultaneously with a potential having an appropriately chosen form.

[0070] Electrodes 100 form a set of linearly independent electrodes that are used to generate electrical excitation on the surface of a body and measure its electrical properties.

[0071] The excitation by the trigonometric template transformation [Equation 2] is: The number of electrodes ne About (n e -1) linearly independent excitation patterns. [Formula 1]

number

[1999] , Fourier basis functions are a natural choice to describe these linearly independent patterns.

[0072] According to a first alternative, all of the electrodes are excited simultaneously using a triangular type of excitation.

[0073] This set of simultaneous excitations is decomposed into spatial and temporal oscillations in a Fourier basis.

[0074] The different frequencies are used to establish distinction between the different frequency multiplexed triangular signals.

[0075] In each triangular excitation pattern, each electrode E n is one static voltage V n sta is related to.

[0076] n e Static voltage V n sta The set of m forms a set of sine and cosine functions with various spatial frequencies m.

[0077] Figure 4 shows the cosine spatial patterns for m varying from 1 to 5. The sine patterns are not shown.

[0078] n e The electrodes are positioned on the perimeter ∂Ω of the body, which is represented by the dotted line in Figure 4. The solid lines represent the excitation potentials applied to the electrodes.

[0079] Each electrode E n The electrostatic potential V related to n,m sta is 1 and n e The n between is defined by the following formula [Formula 2]. [Formula 2]

number

[0080] where A is the signal amplitude, m∈1,...,(n_e-1) represents the spatial frequency, and n e is the number of electrodes, and θ n is the angular position of electrode n, and f m =m*f0 is the vibration frequency, and f0 is f m is the fundamental frequency chosen to be below the Nyquist frequency of the system for any m,

number

number

[0081] Given n e For a set of electrodes, all independent excitation patterns are (n e -1) distinct spatial frequencies. n =2πn / n e In certain cases, the electrodes are regularly distributed around the body.

[0082] Each spatial frequency m corresponds to a temporal frequency f m and are applied simultaneously to each of the electrodes.

[0083] Therefore, each simultaneous excitation potential V n exc is (n e -1) trigonometric functions, each function at a particular frequency f m It vibrates.

[0084] The excitation signal for electrode n is given by the following equation: [Formula 3]

number

[0085] Each potential V defined in this way n exc (t) is the value of each electrode E n Regardless of the time t, n e Care must be taken to ensure that the sum of the excitation voltages on the electrodes is zero. This is achieved under the condition [Formula 4]

number

[0086] According to a second alternative of the invention, the excitation form is [Formula 5]

number

[0087] Here, by convention, if the value of the starting index is greater than the value of the ending index, the sum is also zero, where: [Formula 6] Ψ i (t) = A sin(2πf i t), [Formula 7]

number

number

[0088] where [V j (t)] n-1 is the voltage V j (t) refers to the (n-1)th element of the identity that defines (t), and in such an identity, the term Ψ iWe follow the convention that the are always arranged in increasing order of index i. Excitation can be performed on all electrodes simultaneously, or on one or more subsets of electrodes sequentially.

[0089] Measuring electrical characteristics The electrical properties of the fluid flowing through portion 21 are then measured using electrodes 100 .

[0090] Unit 12 controls the FPGA of the acquisition system to generate 16 excitation signals with the aforementioned characteristics. These 16 digital signals are converted to analog signals by the NI-9262 module and transmitted to the electrode 100 by a coaxial cable.

[0091] The printed circuit board 12 includes one excitation circuit for each electrode 100, and each of these circuits includes a resistor R, as shown in Figure 5. As can be seen in this figure, the potential V n exc is applied to one side of resistor R, and the other side is connected to electrode E n is connected to.

[0092] Electrode E n The Neumann boundary condition at n This current flows through the excitation circuit. This current generates a voltage V across the resistor. n meas =RI n It is obtained by measuring V n exc With respect to this, the signal is a sum of trigonometric functions.

[0093] Process the data In a fourth step, the data measured in the second step is processed to either feed the information directly into a processing algorithm or to obtain a signed data matrix representing the image.

[0094] Data point M n Generate In the first sub-step of the fourth step of the measurement method, data point Mn is calculated for each electrode n.

[0095] The measured signal V n meas The Fourier transform of the P-point current measurement sequence I n (p), where p is the discrete time, and 0≦p≦P is observed, i.e., [Formula 9]

number

[0096] where

number

number

number

[0097] The Fourier transform may be calculated at frequency f1, which corresponds to the frequency at which the P Fourier coefficients are calculated.

[0098] Voltage V n exc frequency f m is chosen to be a harmonic of f1, which allows us to distinguish between the measured signals. Thus, each coefficient k is used to measure one specific frequency f m is related to.

[0099] Data is thus generated at frequency f1, and the resolution in Fourier space is Δf=f m+1 -f m = f1. The highest frequency is the Nyquist frequency of the system, f Nyq It should be noted that the Δp is chosen to be lower than Δp = ½Δp, where Δp is the sampling time.

[0100] Data point M n (k) is the coefficient of each Fourier coefficient k for each electrode n. [Formula 10]

number

[0101] Each data point defines a given triangular pattern of current at a given electrode. The set M of data points for all n and all k n (k) forms the measurement data.

[0102] The excitation frequency is determined by the sampling frequency f of the data acquisition system. DAQ is the voltage V n exec The maximum frequency f m and limit the Nyquist frequency f Nyq =f DAQ The upper limit is / 2.

[0103] f DAQ For a data acquisition system where = 1 MS / s, the Nyquist frequency is equal to 500 kHz.

[0104] A continuous signal must be delivered to avoid small residual voltage errors resulting from energy stored in the electrode-electrolyte contact impedance. To generate continuous signals at various frequencies, the signals are chosen to be harmonics of the lowest frequency to be generated, f1.

[0105] When the application of the potential to the electrode is stopped, some of the electrical energy accumulates at the interface between the electrode and the medium for a few tens of microseconds. This contact impedance effect leads to errors in the measurement and implies the need to introduce a dead time between two successive excitations to allow this energy to dissipate. Generating a continuous signal has the advantage that the applied voltage is never stopped and therefore avoids the need to introduce errors and dead time associated with contact impedance.

[0106] With 16 electrodes, a set of excitation signals at 15 different frequencies (according to Equation 2) or 120 different frequencies (according to Equation 5) is generated. Considering a sampling rate of the acquisition system, e.g., 1 MS / s, the frequencies are f i =i*f0, where f0 is the fundamental frequency and i is between 1 and 15 or 1 and 120, respectively.

[0107] Furthermore, if there are 15 frequencies, the discrete Fourier transform can be performed with P = 32 points, since only the positive results are considered. This results in an image data acquisition rate of 1*10 per second. 6 / 32 = 31,250 images. This choice is made because the lowest sinusoidal signal frequency f1 is equal to the frequency of the discrete Fourier transform calculation, and the highest frequency f 15 = 15 * f0 = 468.875 kHz, which means that it is below the Nyquist limit of 500 kHz for the system in question.

[0108] The excitation amplitude is determined as follows:

[0109] The voltage generation and acquisition module operates in ±10 V intervals. Voltage V n exc Considering this, the amplitude A of the sine waves will necessarily be much lower than the sum of the sine waves produced as a result of constructive interference. However, the amplitude A of the signal should be as large as possible to minimize the signal-to-noise ratio.

[0110] Another limitation to consider is the maximum permissible variation between two successively generated potentials. Real-time control in the acquisition system allows us to select an appropriate value of A = 0.15 V, which gives a resonant peak of ±2.25 V. The rapid transition between positive and negative values ​​of the signal prevents the appearance of electrolysis effects. For example, when DC voltages greater than 1.2 V are applied, electrolysis-related effects appear in water. This effect does not appear with AC voltages greater than 1.2 V if these voltages are changed quickly enough.

[0111] Obtain Data Matrix D In the second substep, data point M n The data matrix D is obtained from

[0112] The 16-electrode system in question, using the NI-9223 module, can use a fixed-point data format with 20 numeric bits, including 5 bits for precision. The electrode index n, which is between 1 and 16, and the Fourier coefficient k, which is between 1 and 15, can be described by a binary number coded on 4 bits.

[0113] In a 32-electrode system, each data point coefficient M n (k) is of the form [Formula 11]

number

[0114] Here, the fixed-point format<s,b,p> is used, where s is signed / unsigned, b is the number of allocated bits, and p is the number of precision bits. n (k) is encoded in U32 format.

[0115] For a given Fourier coefficient k, i.e., for a given frequency f m About e n measured at electrodes e These data points give the data vector [Formula 12]

number

[0116] If the excitation format [Equation 2] is used, the data matrix D [Formula 13]

number

number

[0117] The size of the data is then S=n e (n e -1).

[0118] Only the coefficients of the Fourier transform form part of the data.

[0119] For an image, the data size is S * 32 bits = 4 kB. In comparison, the methods described in [2], [3] and [4] yield 127 kB of data per image without any additional information about boundary conditions.

[0120] Therefore, the measurement method implemented by the system according to the invention allows a higher image acquisition rate and reduces the size of the data by n with respect to the methods described in [2], [3] and [4]. e It is also possible to reduce it by / 2.

[0121] Obtain the signed data matrix In a third sub-step, a signed data matrix representing the image is obtained.

[0122] The Fourier transform yields the coefficients and phase, and the sign of each data point is inferred from the phase.

[0123] Therefore, the format [Formula 15]

number

number

[0124] frequency f m Electrode E l The current I measured at l meas The phase of (t) is [Formula 17]

number

[0125] Assuming synchronous sampling of the analog signal AI input and the sampled signal AO output, the phase shift between the excitation potential and the measured current is [Formula 18]

number

[0126] The phase shift depends on the design of the EIT sensor and the nature of the flow within the body. If the phase shift is large, wraparound effects may make it impossible to reconstruct the sign of the data. Specifically, the phase is symmetric under a transformation of 2πN, where N is an integer. If the phase is greater than 2π, it will wrap around on itself.

[0127] The following two cases have been identified: [Formula 19]

number

[0128] In the first case, the sign of the data point is calculated using the formula above. The element of the signed data matrix corresponding to the k-th Fourier coefficient and to the n-th electrode is then [Formula 20]

number

[0129] In the second case, the wraparound effect causes D n k A code matrix Σ is then introduced to assign arbitrary codes to the data.

[0130] The sign matrix Σ is estimated by introducing a sign function based on the sign of the excitation signal at t=0 for a given harmonic at a given electrode. [Formula 21]

number

[0131] The sign matrix Σ is more specifically defined so that its rows represent alternating signs of cosine functions and sine functions with an integer number of periods in each row: the first two rows have a single period, and the number of periods increases by one for each pair of subsequent rows.

[0132] In other words, if the excitation format [Equation 2] is used, the rows of the code matrix Σ are such that the i-th element of the j-th row is cosine ([2π / ([j+1] / 2)]*(I-1) / n e ), sign([2π / (j / 2)]*(i-1) / n for even j e) is defined to be the sign of

[0133] For example, the first element in the first row of the sign matrix is ​​the sign of cosine (0), i.e., +. The number of electrodes, n e is equal to 16, the sign of the fifth element in the first row is the sign of cosine (2π*[4 / 16]), i.e., 0.

[0134] Thus, the first row of the code matrix contains the code for one period of the cosine function, i.e., the cosine for the i-th element of the row (2π*(i-1) / n e ) The second line represents the sign of one period of the sine function, i.e., the sine (2π*(i-1) / n e ) The third line represents the sign of two periods of the cosine function, i.e., cosine ([2π / 2]*(i-1) / n e ) The fourth line represents the sign of two periods of the sine function, i.e., the sine ([2π / 2]*(i-1) / n e ) The fifth line represents the sign of the three periods of the cosine function, i.e., cosine ([2π / 3]*(i-1) / n e ) and (n e This continues up to line -1).

[0135] For example, for a device with 16 electrodes, the code matrix takes the form shown in Figure 9. For a device with 32 electrodes, the code matrix takes the form shown in Figure 10.

[0136] If the excitation format [Equation 5] is used, the closest excitation electrode is assigned the opposite sign, especially for measurement electrodes placed equidistant between two excitation electrodes. The amplitude has been experimentally measured to be on the order of O(10-7) A, making its contribution to the data negligible.

[0137] The use of such a sign matrix allows for the optimization of the processing of the data to form an image, allowing the sign of each data point to be estimated and the image to be reconstructed.

[0138] For large phase shifts, i.e., greater than π / 2, n e n of electrodes e The arbitrarily signed amplitude of one excitation pattern, or in other words the element of the signed data matrix corresponding to the k-th Fourier coefficient and to the n-th electrode, is given by: [Formula 22]

number

[0139] Algorithm implementation The HOST part of the acquisition system continuously sends frequency and amplitude parameters to the FPGA of the acquisition system.

[0140] FIG. 6 shows the algorithm for generating excitation signals for the 16 electrodes.

[0141] In the system in question, the FPGA receives 16x16 data points in a loop clocked at 1MS / s to create 16 analog signals.

[0142] In the first step, the system is initialized.

[0143] Initially, the FPGA is empty. The HOST loads the FPGA, then the NI-9262 module is reset.

[0144] In the second step, interrupt requests are sent and received.

[0145] When the FPGA is ready to start data acquisition, it notifies the HOST using a hardware interrupt. The FPGA waits for HOST verification and starts acquisition.

[0146] In the third step, the sampling is checked.

[0147] The sample pulse generator function is called to begin generating data points. The frequency with which the function is called determines the sampling rate used to generate data points. In parallel, the I / O status write function is called with the same frequency to check the status of each sample generated.

[0148] In the fourth step, a digital excitation signal function is generated.

[0149] The HOST instructs the FPGA to begin generating the excitation signal function.

[0150] In the fifth step, analog excitation signals are generated: 16 excitation signals are sent to the electrodes.

[0151] In the sixth step, the HOST checks the generation of the signal and reports any errors at the HOST or FPGA level.

[0152] Steps 1 and 2 are executed once when the algorithm is launched. Steps 3 through 6 are repeated for each output point at the sampling frequency.

[0153] The sampling frequency may be comprised between 10 kHz and 500 MHz, preferably between 500 kHz and 50 MHz.

[0154] FIG. 7 shows an algorithm for processing the data received from the electrode 100.

[0155] Measuring the voltage across the terminals of resistor R makes it possible to deduce the Neumann boundary conditions used to implement the image reconstruction algorithm. An acquisition rate of 1 MS / s on 16 channels corresponds to a data transfer rate of 320 MB / s.

[0156] Using a Fast Fourier Transform, considering only the Fourier coefficients relevant to the generated signal, it is possible to reduce its size without affecting its quality. This also acts as an effective bandpass filter. However, the real-time calculation of 16 Fast Fourier Transforms requires high computational power. FPGAs are suitable tools to perform this task, as they allow the real-time parallel conversion of a signal into its Fourier components on multiple channels.

[0157] In the first step A, the system is initialized.

[0158] The FPGA resets the NI-9223 analog signal acquisition module.

[0159] In a second step B, the memory is configured.

[0160] The HOST configures and initiates direct access to the FPGA's memory. The FPGA configures and initiates 16 first-in, first-out queues to ensure communication between the 16 measurement channels and their fast Fourier transform calculations.

[0161] In a third step C, interrupt requests are sent and received.

[0162] Hardware interrupts make it possible to guarantee that queues and direct memory accesses are ready.

[0163] In a fourth step D, the sampling is checked.

[0164] It calls the sample pulse generation function to control the sampling frequency, and calls the I / O status reading function at the same frequency to check the status of each sample and report any errors to the HOST.

[0165] In a fifth step E, an analog measurement signal is acquired.

[0166] The I / O Read function is configured to read a single sample from each channel of each NI-9223 module. This function is called at 1MHz and is timed by the Generate Sample Pulse function.

[0167] In the sixth step F, the fast Fourier transform of each channel is calculated in a 1 MHz loop. The calculation time depends on the number of points P considered for the Fourier transform. After P measured data points have been transferred for the Fourier transform calculation, the function returns P Fourier coefficients, one at each iteration of the Fourier transform loop. The amplitude of the Fourier coefficients is then calculated at a frequency of 1 MHz.

[0168] In a seventh step G, the data is addressed.

[0169] The amplitude data points in U32 format, together with the corresponding Fourier coefficients (in U16 format) and the corresponding channels (in U16 format), form data elements in U64 format. At each iteration of the Fourier transform loop, 16 elements are written to direct access memory for the 16 channels for transmission to the HOST.

[0170] In an eighth step H, a data matrix is ​​constructed.

[0171] The HOST waits for the direct access memory to collect at least 240 elements representing the complete data image. The Fourier coefficients associated with the address and amplitude of the nth electrode are used to form a data matrix D.

[0172] In a ninth step I, the data matrix is ​​stored.

[0173] The data is then used to perform real-time image reconstruction or is stored. The images are based on a one-step iterative least-squares reconstruction algorithm, such as that described in [5].

[0174] Real-time image reconstruction can generate on the order of 100 images per second.

[0175] In a tenth step J, signal acquisition is verified and error checking is performed.

[0176] The synchronization of the analog signal generation and measurement modules is checked and any errors are reported.

[0177] Steps A through C are executed once when the algorithm is launched.

[0178] Steps D through G are repeated for each output point at the sampling frequency.

[0179] n e *(n e Once a complete data matrix containing −1) data points has been acquired, the algorithm proceeds to step H. Steps H through J are repeated at the image acquisition frequency.

[0180] The sampling frequency may be comprised between 10 kHz and 500 MHz, preferably between 500 kHz and 50 MHz.

[0181] Figure 8 shows, in the left-hand graph, six excitation signals for 16 electrodes, each signal consisting of a sum of 15 sinusoidal functions.

[0182] The solid line in the right-hand graph of Figure 8 is the voltage V measured across the resistor R shown in Figure 2, with R=200 Ω, in equivalent Fourier space. meas The dashed-dotted line represents the Fourier transform of the voltage of the generated signal.

[0183] 11 shows experimental images obtained using a neural network to reconstruct a water flow containing plastic foreign objects based on the measurement matrix described above, in accordance with the present invention, and, for comparison, experimental images obtained using a prior art technique. It can be seen that the detection by the present invention is much more accurate than that by the prior art, and corresponds almost perfectly to the experimental situation.

[0184] In response to signals received from the flowing fluid indicative of one or more defects, the electronic automatic control unit 14 modifies, does not modify, or even stops the operating frequency of the pump 20 so as to prevent any malfunction or damage to the pump 20.

[0185] The invention is not limited to the examples described here, and in particular it is possible to combine features of the examples shown with one another in variants not shown.

[0186] However, other variations and modifications can be envisaged without departing from the scope of the invention, and in particular the method implemented by the system according to the invention can also use acquisition systems other than those described.

[0187] Although the fault has been described as a foreign object suspended in the fluid, the system according to the invention can detect any type of fault, in particular the risk of pump failure as a result of cavitation due to a lack of fluid.

[0188] In the examples shown, the number of sensor electrodes was 16 or 32, but a smaller number, in particular 4 or 12 electrodes, or even a number greater than 32 can also be envisaged. (References) [1]Teague, G. (2002). Mass flow measurement of multi-phase mixtures by means of tomographic techniques. University of Cape Town, Faculty of Engineering, Department of Electrical Engineering. [2]Dupre, A., Mylvaganam, S. (2017). Simultaneous and Continuous Excitation Strategy for High Speed EIT: the ONE-SHOT method. In Proceedings of the 9th World Congress on Industrial Process Tomography, pages 667 - 674. [3]Darnajou, M., Dupre, A., Dang, C., Ricciardi, G., Bourennane, S., Bellis, C. (in 2019). On the implementation of simultaneous multi-frequency excitations and measurements for electrical impedance tomography. Sensors, 19(17). [4]Darnajou, M., Dupre, A., Dang, C., Ricciardi, G., Bourennane, S., Bellis, C., Mylvaganam, S. (2020). High Speed EIT with Multifrequency Excitation using FPGA and Response Analysis using FDM. IEEE Sensors. [5]https: / / www.math.colostate.edu / ~siamcsu / files / NOSER.pdf

Explanation of Symbols

[0189] 1 System 2 circuits 10 Electrical Sensors 11 screens 12 Printed Circuit Board 13 Signal Processing Unit 14 Automatic Electronic Control Unit 20 Pump 21 parts 22 Duct 23 Casing 24 through holes 25 ports 26 flange 100 electrodes

Claims

1. 1. A system for detecting at least one defect that may be present in a fluid flowing through a fluid circuit comprising at least one fluid device and, where appropriate, at least one fluid branch upstream of said fluid device, and for preventing failures of said fluid device specific to said defect or defects, comprising: at least one electrical sensor including a plurality of electrodes distributed around a portion of the fluid circuit upstream of the fluidic device, the electrodes having ends flush with an inner surface of the portion; an electronic circuit for controlling the electrodes and for performing electrical impedance tomography measurements, the electronic circuit being configured to simultaneously excite each electrode and measure the electrical impedance matrix of the fluid flowing through the portion; a signal processing unit configured to identify the one or more defects and / or to reconstruct an image of the fluid flowing through the portion based on measurements of an impedance matrix of the electronic circuit; an electronic unit for automatically controlling a controller of the fluidic device or the fluidic device itself or, where appropriate, the fluid branch, configured to change the operating frequency of the fluidic device so as to prevent failure of the fluidic device, or to control the operation of the fluidic device, or to interrupt the operation of the fluidic device, or to physically remove one or more defects upstream of the fluidic device, in response to signals of the flowing fluid indicative of one or more defects; Including, the system.

2. The detection and prevention system of claim 1 , wherein the fluid device is a pump or a turbine and the controller is a variable frequency drive.

3. The detection and prevention system according to claim 1 or 2, wherein the operation of the electronic unit for automatically controlling the fluidic device is subordinate to the operation of the signal processing unit.

4. a module to be disposed in the fluid circuit upstream of the fluidic device, the module comprising: a duct defining the fluid portion; a casing fixed around the duct, the casing housing the electric sensor, the electrodes of which are individually housed and fixed in a sealed state in a through-hole of the duct; 3. The detection and prevention system of claim 1, comprising:

5. The detection and prevention system of claim 4 , wherein the casing houses a port for supplying power to the electrode.

6. 6. The detection and prevention system according to claim 4 or 5, wherein the duct includes a flange at at least one of its ends for fastening to a duct of the fluid circuit.

7. The detection and prevention system of claim 1 , wherein the signal processing unit is configured to implement an image reconstruction neural network.

8. The control circuit a / energizing the electrodes, each electrode being of the form [Formula 2] [Equation 1] where A is the signal amplitude and m∈1,...,(n ｅ −1) represents the spatial frequency, and n ｅ is the number of electrodes, and θ ｎ is the angular position of electrode n, and f ｍ = m * f ０ is the vibration frequency, and f ０ is f ｍ is the fundamental frequency chosen to be below the Nyquist frequency of the system for any m, [Equation 2] is the set of all odd natural numbers, [Equation 3] is the set of all even natural numbers, of the form [Equation 2], or format [Formula 5] [Equation 4] and by convention, if the value of the starting index is greater than the value of the ending index, then the sum is also zero, of the form [Equation 5] potential V ｎ ｅｘｃ is excited by where: [Formula 6] P ｉ (t) = A sin(2πf ｉ t) [Formula 7] [Equation 5] and [Formula 8] [Equation 6] Here, [V ｊ (t)] ｎ－１ is the voltage V ｊ (t) refers to the (n-1)th element of the identity that defines (t), and in such an identity, the term Ψ ｉ follows the convention that the electrodes are always arranged in increasing order of index i, and the excitation is performed either simultaneously on all the electrodes or sequentially on one or more subsets of electrodes according to [Equation 5], Steps and b / the electrical characteristic V of the fluid flow at the electrode ｎ ｍｅａｓ measuring c / a step of processing the data generated in the measurement step b / , which includes: c1 / Each electrode E ｎ Regarding [Formula 10] [Equation 7] A data point M defined by ｎ a sub-step of calculating where R is V ｎ ｍｅａｓ is the resistance of a resistor used to measure V across the terminals of said resistor ｎ ｍｅａｓ =RI ｎ and P is the current I ｎ is the number of points in the discrete sequence of measurements of ｅ -1), and β ｐ = (2πp / P), Sub-steps and If the c2 / excitation format [Equation 2] is used, then Eq. [Formula 13] [Equation 8] or if the excitation form [Equation 5] is used, [Formula 14] [Equation 9] for all n and all k, ｎ forming a data matrix D from (k); c3 / forming a signed data matrix, whose elements are the phase shift Φ between the excitation potential of electrode l and the measured current of electrode n ｎ，ｌ When (k) is less than π / 2, the following equation [Formula 20] [Equation 10] is defined by The element is the phase shift Φ between the excitation potential of electrode l and the measured current of electrode n. ｎ，ｌ When (k) is π / 2 or more, the following formula [Formula 22] [0011] is defined by Here, Σ is If the excitation format [Equation 2] is used, the i-th element of the j-th row of Σ is cos([2π / ([j+1] / 2)]*(i-1) / n for odd j. ｅ ), for even j, sine([2π / (j / 2)]*(i-1) / n ｅ ) is the sign of If the excitation format [Equation 5] is used, the closest excitation electrodes are assigned opposite signs, and in particular, with a measurement electrode placed equidistant between the two excitation electrodes, the amplitude has been experimentally measured to be on the order of O(10-7) A, making its contribution to the data negligible. is the code matrix defined as Sub-steps and and 8. The detection and prevention system of claim 1, configured to perform the following:

9. The control circuit controls the potential V ｎ ｅｘｃ All of the above are conditions [Formula 4] [0012] 9. The detection and prevention system of claim 8, configured to perform step a / by satisfying:

10. 10. The detection and prevention system of claim 1, wherein the electrodes of the electrical sensor are regularly angularly distributed around the portion of the fluid circuit upstream of the pump.

11. Use of the detection and prevention system according to any one of claims 1 to 10 for measuring foreign objects and / or for measuring blockages, clogging and / or fouling induced by said foreign objects, cavitation, bubbles, lack of fluid and the presence of gas.

12. An industrial plant comprising at least a fluid circuit and a detection and prevention system according to any one of claims 1 to 10.

Images