X-ray-based online monitoring method and system for chlorine content of crude oil

Abstract

The application discloses an X-ray-based online monitoring method and system for chlorine content of crude oil, relates to the technical field of online analysis of crude oil components, and solves the technical problem that existing XRF online analysis only outputs total chlorine concentration and is difficult to separately quantify organic chlorine and inorganic chlorine; the method comprises the following steps: arranging low-energy X-ray sources and high-energy X-ray sources along the flow direction of a pipeline at intervals, collecting low-energy excitation energy spectrum and high-energy excitation energy spectrum respectively; online acquiring the water content and salinity parameters of crude oil as constraint conditions; calculating the Fisher information ratio of Compton scattering continuous spectrum to extract a scattering fingerprint characteristic parameter; taking the scattering fingerprint characteristic parameter, the water content and the salinity as constraints, establishing a response matrix dynamic updating mechanism, constructing a total chlorine signal deconvolution mathematical model and solving the model by regularization, and decomposing the total chlorine signal into organic chlorine and inorganic chlorine components and outputting the concentration values of the components respectively. The application realizes the form differentiation and online synchronous quantification of organic chlorine and inorganic chlorine in crude oil.

Description

Technical Field

[0001] This invention belongs to the field of online crude oil composition analysis technology, specifically relating to an online monitoring method and system for crude oil chlorine content based on X-rays. Background Technology

[0002] Chlorides in crude oil are mainly divided into two categories: inorganic chlorine and organic chlorine. Inorganic chlorine is dissolved in trace amounts of aqueous phase in crude oil in the form of salts such as sodium chloride, calcium chloride, and magnesium chloride, and dispersed in the oil phase as salt droplets. It can be removed by electro-desalting. Organic chlorine, on the other hand, is bound to organic molecules in crude oil in the form of covalent bonds and is uniformly dissolved in the oil phase. It cannot be removed by electro-desalting and is prone to hydrolysis to generate hydrogen chloride during downstream processing, causing serious equipment corrosion and catalyst poisoning in atmospheric and vacuum distillation, catalytic reforming and other equipment.

[0003] Current online monitoring technologies for crude oil chlorine content mainly employ monochromatic energy-dispersive X-ray fluorescence spectroscopy, such as the Clora online analyzer. These instruments use a single X-ray excitation source to measure total chlorine content, but they face the following technical challenges: First, in actual refinery operation, electro-desalting units can only remove inorganic chlorine and are ineffective against organic chlorine. The total chlorine concentration alone cannot determine whether the desalting process has achieved the expected results, nor can it provide early warning of the risk of excessive organic chlorine levels. Second, inorganic chlorine is dispersed in the oil phase as brine droplets, and the particle size and spatial distribution of these droplets vary depending on factors such as crude oil properties, water content, and flow rate. The dynamic changes in sulfur content, the limited X-ray irradiation area of ​​conventional XRF measurements, and the uneven distribution of droplets directly lead to poor measurement repeatability. Furthermore, in existing XRF analyses, sulfur interferes with the characteristic peaks of chlorine due to spectral interference and matrix absorption effects. Although this can be compensated for by mathematical correction formulas, the correction coefficients are usually fixed values, which cannot adapt to the large fluctuations in sulfur content in crude oils from different sources. Finally, the size and distribution of brine droplets in crude oil affect the intensity ratio and peak shape characteristics of Compton and Rayleigh scattering, but current techniques do not correlate these scattering characteristics with the quantitative analysis of chlorine speciation.

[0004] Therefore, this invention provides an online monitoring method and system for crude oil chlorine content based on X-rays. By synchronously exciting dual-energy X-rays, combining statistical feature extraction of the scattering continuous spectrum, and online correction of water content and salinity, the method can achieve separate quantification of organic and inorganic chlorine. Summary of the Invention

[0005] The present invention aims to solve at least one of the technical problems existing in the prior art; to this end, the present invention proposes an online monitoring method and system for crude oil chlorine content based on X-rays, which is used to solve the technical problem that the existing online XRF analysis of crude oil chlorine content only outputs the total chlorine concentration and cannot distinguish the occurrence forms and contents of organic chlorine and inorganic chlorine.

[0006] To address the above problems, a first aspect of the present invention provides a method for online monitoring of crude oil chlorine content based on X-rays, comprising the following steps: S1: A dual-level excitation is simultaneously applied to the crude oil flowing in the pipeline using a low-energy X-ray source and a high-energy X-ray source, and the low-energy excitation energy spectrum and the high-energy excitation energy spectrum are collected respectively. The low-energy excitation spectrum contains the Cl-Kα characteristic peak of chlorine, and the high-energy excitation spectrum contains the Cl-Kα characteristic peak of chlorine and the S-Kα characteristic peak of sulfur. S2: Obtain the current water content and salinity parameters of crude oil online and input them as constraints into the deconvolution model; S3: Perform statistical feature analysis on the Compton scattering continuous spectrum in the high-energy excitation spectrum to extract scattering fingerprint feature parameters; S4: Using the scattering fingerprint feature parameters and the water content and salinity parameters as known constraints, construct a mathematical model for the deconvolution of the total chlorine signal, and iteratively solve the low-energy excitation spectrum and the high-energy excitation spectrum to decompose the total chlorine signal into organic chlorine signal components and inorganic chlorine signal components, and output the organic chlorine concentration value and the inorganic chlorine concentration value respectively.

[0007] Preferably, in step S1, the low-energy X-ray source and the high-energy X-ray source are arranged at intervals along the crude oil flow direction in the pipeline. The low-energy excitation beam forms a first excitation spot in the crude oil measurement area, and the high-energy excitation beam forms a second excitation spot in the crude oil measurement area. The distance between the first excitation spot and the second excitation spot in the flow direction is L, where L≥d, and d is the width of the first excitation spot in the flow direction. It also includes a flow velocity measurement unit, which is used to measure the flow velocity v of crude oil in the pipeline in real time, and to measure and output the flow velocity v to the trigger control terminal of the second X-ray detector in real time at a preset sampling frequency. L = v × Δt, where Δt is a preset time delay and is dynamically updated according to the flow velocity v; The low-energy excitation spectrum is acquired by a first X-ray detector, and the high-energy excitation spectrum is acquired by a second X-ray detector. The second X-ray detector starts acquisition after a delay of Δt. During acquisition, the bypass quick-opening solenoid valves at both ends of the measurement area of ​​the control pipeline are switched to a stop-flow closed state, and the energy spectrum is integrated synchronously on the static crude oil section. Alternatively, a high-speed multichannel pulse analyzer is used to perform multi-cycle cyclic acquisition with a short integration window width on the order of milliseconds, and sliding matching and spectral superposition of each cycle short window sequence are performed on the digital end according to the delay Δt. The low-energy excitation beam and the high-energy excitation beam are respectively incident on the crude oil measurement area inside the pipeline in a manner perpendicular to the flow direction.

[0008] Preferably, step S2 includes the following steps: A single microwave resonant cavity dual-frequency multiplexing structure is adopted to transmit microwave signals of the first and second frequency bands to crude oil flowing through the same measurement section, and to collect the first resonant frequency and the first quality factor, as well as the second resonant frequency and the second quality factor. The high and low frequency signal responses are extracted for differential calculation. The volumetric water content and salinity parameters of the current crude oil are calculated by inversion based on the frequency difference signal and the quality factor difference signal. A water content-salinity joint constraint matrix is ​​constructed using volumetric water content and salinity parameters, and the joint constraint matrix is ​​used as the input of the regularization constraint term of the deconvolution model.

[0009] Preferably, step S3 includes the following steps: The crude oil measurement area was irradiated with low-energy and high-energy excitation beams, respectively, and the low-energy Compton scattering continuous spectrum was collected. low (s) and high-energy Compton scattering continuous spectrum l high (s); Using the preset energy resolution to l low (s) and l high (s) are discretized into N energy channels, and the probability of photon count in each energy channel relative to the total count is calculated to obtain the probability distribution p of the low-energy spectrum. low i The probability distribution p of the high energy spectrum high i Where i = 1, 2, ..., N, and N is the total number of energy channels; calculate the fluctuation characteristic parameters of the low-energy scattering spectrum respectively. Fluctuation characteristic parameters of high-energy scattering spectrum ; where p low i and p low i+1 p represents the probability values ​​of the i-th and (i+1)-th energy channels in the low-energy scattering spectrum, respectively. high i and p high i+1 These are the probability values ​​of the i-th and (i+1)-th energy channels in the high-energy scattering spectrum, respectively. Calculate the Fisher information content F of the low-energy scattering spectrum respectively. low Fisher information content F of high-energy scattering spectrum high ; Calculate Fisher information ratio .

[0010] Preferably, in step S3, l is set to a preset energy resolution. low (s) and l high (s) is discretized into N energy channels, and the following steps are also included: High-energy Compton scattering continuous spectrum lhigh The photon counting probability p of each energy channel of (s) high i Calculate the fluctuation characteristic parameter σ high i To obtain the fluctuation feature sequence {σ high 1,σ high 2,…,σ high N}; Using the sliding window length M to represent the fluctuation feature sequence {σ high i} Perform local standard deviation calculation to obtain the local standard deviation sequence {τ} j}, where j = N - M + 1, τ j =std(σ high j ,σ high j+1 ,…,σ high j+M-1 ), where std is the standard deviation; The length M of the sliding window is dynamically determined based on the excitation energy of the high-energy excitation beam. Extract the local standard deviation sequence {τ j The peak position j peak ; The peak position j peak Together with the Fisher information ratio R, a scattering fingerprint feature vector F=[j peak ,R], is used as a scattering fingerprint feature parameter.

[0011] Preferably, in step S4, constructing the deconvolution mathematical model of the total chlorine signal includes the following steps: The net peak area A of the Cl-Kα characteristic peak in the low-energy excitation spectrum Cl,low The net peak area A of the Cl-Kα characteristic peak in the high-energy excitation spectrum Cl,high Construct the measured total chlorine signal vector b=[A Cl,low A Cl,high ] T ; The unit concentration response coefficient k of organochlorine under low-energy excitation org,low The unit concentration response coefficient k under high-energy excitation org,high And the unit concentration response coefficient k of inorganic chlorine under low-energy excitation. inorg,low The unit concentration response coefficient k under high-energy excitation inorg,high Construct the response matrix K=[[k org,low ,k inorg,low ],[k org,high ,k inorg,high ]]; Where, k inorg,low With k inorg,high Based on Fisher's information ratio R, the updates are dynamic, specifically: k inorg,low =k 0 inorg,low ·(1+α low ·R); k inorg,high =k 0 inorg,high ·(1+α high ·R); where k 0 inorg,low k is the reference response coefficient of inorganic chlorine under low-energy excitation. 0 inorg,high α is the baseline response coefficient of inorganic chlorine under high-energy excitation. low α is the low-energy adjustment factor. high This is a high-energy adjustment factor; Construct the theoretical prediction vector K·[C org C inorg ] T The residual sum of squares between the measured total chlorine signal vector b and the measured total chlorine signal vector b, and using the water content-salinity joint constraint matrix constructed with water content and salinity parameters as regularization constraints, the organochlorine concentration C is solved using the Tikhonov regularization method. org and inorganic chlorine concentration C inorg .

[0012] Preferably, the step of using the Tikhonov regularization method to solve the regularization objective function of the matrix equation specifically involves: min{||K·[C org C inorg ] T -b|| 2 +λ·||Γ·[C org C inorg ] T || 2}, Where λ is the regularization parameter and Γ is the penalty matrix constructed based on the water content-salinity joint constraint matrix; The penalty matrix Γ is a 2×2 diagonal matrix, specifically: Γ=diag(γ org ,γ inorg ), γ inorg =η·w·s,γ org =ε·γ inorg Where, diag is a diagonal matrix, γ org γ represents the penalty weight corresponding to the organochlorine concentration. inorg η is the penalty weight corresponding to the inorganic chlorine concentration, w is the obtained volumetric water content, s is the obtained salinity, η is the first proportionality coefficient, and ε is the second proportionality coefficient.

[0013] A second aspect of the present invention provides an X-ray-based online monitoring system for crude oil chlorine content, comprising the following modules: Dual-energy synchronous excitation acquisition module: A low-energy X-ray source and a high-energy X-ray source are used to simultaneously apply dual-level excitation to the crude oil flowing in the pipeline, and the low-energy excitation energy spectrum and the high-energy excitation energy spectrum are acquired respectively; The low-energy excitation spectrum contains the Cl-Kα characteristic peak of chlorine, and the high-energy excitation spectrum contains the Cl-Kα characteristic peak of chlorine and the S-Kα characteristic peak of sulfur. Online process parameter acquisition module: Acquires the current water content and salinity parameters of crude oil online, and inputs them as constraints into the deconvolution model; Scattering fingerprint feature extraction module: Performs statistical feature analysis on the Compton scattering continuous spectrum in the high-energy excitation energy spectrum to extract scattering fingerprint feature parameters; Total chlorine signal deconvolution solution module: Using the scattering fingerprint feature parameters and the water content and salinity parameters as known constraints, a mathematical model for deconvolution of the total chlorine signal is constructed. The low-energy excitation energy spectrum and the high-energy excitation energy spectrum are combined for iterative solution to decompose the total chlorine signal into organic chlorine signal components and inorganic chlorine signal components, and the organic chlorine concentration value and inorganic chlorine concentration value are output respectively.

[0014] The beneficial effects of this invention are: This invention employs simultaneous excitation from low-energy and high-energy dual X-ray sources and utilizes the Fisher information ratio of the Compton scattering continuous spectrum as a fingerprint parameter for the morphology of brine droplets. This fingerprint parameter is embedded in the update function of the response matrix, enabling the response coefficient of inorganic chlorine to adaptively adjust with the size distribution of brine droplets. This achieves online morphological differentiation between organic and inorganic chlorine, allowing refineries to obtain real-time information on organic chlorine concentration and providing a direct basis for adjusting the electro-desalination process. This invention arranges low-energy and high-energy excitation sources at intervals along the pipeline flow direction, and uses flow velocity measurement and delayed trigger control to make the two sets of energy spectra correspond to the same measurement volume of the same flowing oil slick, effectively eliminating the influence of heterogeneous distribution on the measurement results. This invention employs a dual microwave resonant cavity differential structure to eliminate common-mode interference from environmental factors such as temperature and pipe wall scaling on measurements. The corrected water content and salinity are then used as regularization constraint terms input into the deconvolution model, ensuring that the quantification of inorganic chlorine is unaffected by fluctuations in water content and salinity. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the method flow of the present invention; Figure 2 This is a schematic diagram of the module flow of the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Please see Figure 1 As shown, this invention is an online monitoring method for the chlorine content of crude oil based on X-rays, comprising the following steps: S1: A dual-level excitation is simultaneously applied to the crude oil flowing in the pipeline using a low-energy X-ray source and a high-energy X-ray source, and the low-energy excitation energy spectrum and the high-energy excitation energy spectrum are collected respectively. The low-energy excitation spectrum contains the Cl-Kα characteristic peak of chlorine, and the high-energy excitation spectrum contains the Cl-Kα characteristic peak of chlorine and the S-Kα characteristic peak of sulfur. S2: Obtain the current water content and salinity parameters of crude oil online and input them as constraints into the deconvolution model; S3: Perform statistical feature analysis on the Compton scattering continuous spectrum in the high-energy excitation spectrum to extract scattering fingerprint feature parameters; S4: Using the scattering fingerprint feature parameters and the water content and salinity parameters as known constraints, construct a mathematical model for the deconvolution of the total chlorine signal, and iteratively solve the low-energy excitation spectrum and the high-energy excitation spectrum to decompose the total chlorine signal into organic chlorine signal components and inorganic chlorine signal components, and output the organic chlorine concentration value and the inorganic chlorine concentration value respectively.

[0018] Specifically, scattering fingerprint feature parameters refer to one or more statistical quantities extracted from the Compton scattering continuum that can characterize the morphological features (including average particle size and particle size distribution width) of brine droplets in crude oil.

[0019] In one embodiment of the present invention, in step S1, the low-energy X-ray source and the high-energy X-ray source are arranged at intervals along the crude oil flow direction in the pipeline. The low-energy excitation beam forms a first excitation spot in the crude oil measurement area, and the high-energy excitation beam forms a second excitation spot in the crude oil measurement area. The distance between the first excitation spot and the second excitation spot in the flow direction is L, where L≥d, and d is the width of the first excitation spot in the flow direction. It also includes a flow velocity measurement unit, which is used to measure the flow velocity v of crude oil in the pipeline in real time, and to measure and output the flow velocity v to the trigger control terminal of the second X-ray detector in real time at a preset sampling frequency. L = v × Δt, where Δt is a preset time delay and is dynamically updated according to the flow velocity v; The low-energy excitation spectrum is acquired by a first X-ray detector, and the high-energy excitation spectrum is acquired by a second X-ray detector. The second X-ray detector starts acquisition after a delay of Δt. During acquisition, the bypass quick-opening solenoid valves at both ends of the measurement area of ​​the control pipeline are switched to a stop-flow closed state, and the energy spectrum is integrated synchronously on the static crude oil section. Alternatively, a high-speed multichannel pulse analyzer is used to perform multi-cycle cyclic acquisition with a short integration window width on the order of milliseconds, and sliding matching and spectral superposition of each cycle short window sequence are performed on the digital end according to the delay Δt. The low-energy excitation beam and the high-energy excitation beam are respectively incident on the crude oil measurement area inside the pipeline in a manner perpendicular to the flow direction.

[0020] Specifically, a low-energy X-ray source and a high-energy X-ray source are respectively installed on the upstream and downstream sides of the crude oil pipeline measurement area. They are arranged at intervals along the crude oil flow direction within the pipeline. The low-energy X-ray source emits a low-energy excitation beam that forms a first excitation spot in the crude oil measurement area, while the high-energy X-ray source emits a high-energy excitation beam that forms a second excitation spot. The distance between the first and second excitation spots in the flow direction is denoted as L, where L is greater than or equal to the width d of the first excitation spot in the flow direction, ensuring that the two excitation spots are completely separated in space and have no overlapping areas. A flow velocity measurement unit is also installed on the pipeline to measure the flow velocity v of the crude oil in real time. The flow velocity measurement unit continuously measures the flow velocity v at a sampling frequency of not less than 1 Hz and outputs the measurement results to the trigger control terminal of the second X-ray detector in real time. The distance L between the first and second excitation spots satisfies the relationship L = v × Δt with respect to the flow velocity v, where Δt is a preset time delay. The preset time delay is dynamically updated according to the flow velocity v, and the update method is as follows: That is, when the flow velocity changes, the time delay is adaptively adjusted accordingly; specifically, during the energy spectrum acquisition cycle of the detector, if the flow velocity measurement unit detects that the fluctuation amplitude of the instantaneous flow velocity v is greater than the preset amplitude threshold (e.g., ±5%), the system triggers an acquisition interruption, discards the energy spectrum data of the current acquisition cycle; after the flow velocity stabilizes again, Δt is recalculated and updated based on the current v, and the dual-energy synchronous acquisition process is restarted. In the actual acquisition process, the low-energy excitation spectrum is continuously acquired by the first X-ray detector located at the corresponding position of the low-energy X-ray source; the high-energy excitation spectrum is acquired by the second X-ray detector located at the corresponding position of the high-energy X-ray source. However, the second X-ray detector is not started synchronously with the first X-ray detector, but rather after a delay of Δt. This is because the time required for crude oil to flow from the first excitation spot to the second excitation spot is exactly equal to Δt. Therefore, the spectrum acquired by the second X-ray detector corresponds to the same oil slick flowing from the first excitation spot. Thus, although the low-energy and high-energy excitation spectra originate from different spatial locations within the pipeline, they correspond to the same measured oil slick. The low-energy and high-energy excitation beams are incident on the crude oil measurement area perpendicular to the flow direction of the crude oil within the pipeline, ensuring that the excitation beams are orthogonal to the flow direction. This guarantees the stability of the cross-sectional shape of the excitation spot in the flow direction, facilitating precise control of the spacing L and the delay time Δt.

[0021] In one embodiment of the present invention, step S2 includes the following steps: A single microwave resonant cavity dual-frequency multiplexing structure is adopted to transmit microwave signals of the first and second frequency bands to crude oil flowing through the same measurement section, and to collect the first resonant frequency and the first quality factor, as well as the second resonant frequency and the second quality factor. The high and low frequency signal responses are extracted for differential calculation. The volumetric water content and salinity parameters of the current crude oil are calculated by inversion based on the frequency difference signal and the quality factor difference signal. A water content-salinity joint constraint matrix is ​​constructed using volumetric water content and salinity parameters, and the joint constraint matrix is ​​used as the input of the regularization constraint term of the deconvolution model.

[0022] Specifically, the microwave resonant cavity sensor utilizes the physical property that oil and water have significantly different dielectric constants—the dielectric constant of crude oil at room temperature is about 2.2, while the dielectric constant of water is about 80. Therefore, the dielectric constant of an oil-water mixture is mainly determined by its water content. When microwaves pass through water-containing crude oil, the resonant frequency and quality factor of the resonant cavity change in a regular manner. The first resonant frequency f1 and first quality factor Q1 of the first microwave resonant cavity sensor, as well as the second resonant frequency f2 and second quality factor Q2, are collected respectively. The high and low frequency signal responses are extracted and differentially calculated. The difference between the first and second resonant frequencies is taken as the frequency difference signal Δf = f1 - f2, and the difference between the first and second quality factors is taken as the quality factor difference signal ΔQ = Q1 - Q2. When the pipeline is filled with crude oil, the resonant frequency of the resonant cavity shifts. The shift Δf is related to the relative permittivity ε of the measured crude oil. r The relationship between them satisfies a perturbation formula, specifically: Where f0 is the cavity resonant frequency, V sample V is the volume of the sample being tested.cavity Let be the volume of the resonant cavity; simultaneously, the relationship between the change in quality factor ΔQ and the loss tangent tanδ of the dielectric satisfies: Where Q0 is the cavity quality factor, and C is a constant related to the cavity geometry. The constant C is obtained through the initial calibration before the system leaves the factory or is put into operation. Specifically, dry air and standard pure water with known conductivity are sequentially introduced into the pipeline. The resonant frequency and quality factor parameters in the cavity and pure water full states are recorded respectively. The inherent constant C is obtained by substituting it into the above perturbation formula and then solidified into the non-volatile memory of the system. Based on the frequency difference signal Δf and the quality factor difference signal ΔQ, combined with the above perturbation formula and the equivalent dielectric constant model of the crude oil mixture (e.g., a series-parallel hybrid model), the current volumetric water content w and salinity parameter s of the crude oil are calculated by inversion. Specifically, the salinity parameter s is obtained as follows: salinity changes simultaneously affect the resonant frequency shift and the quality factor change. The former reflects the energy storage characteristics of the medium (real part of the dielectric constant), and the latter reflects the energy dissipation characteristics of the medium (imaginary part of the dielectric constant). The two difference signals are combined to establish a system of equations including salinity s as an unknown: Δf = f(εQ / ΔQ). r (w,s)); ΔQ=g(ε r (w,s),tanδ(w,s)); By iteratively solving the above system of equations, the volumetric water content w and the salinity parameter s can be obtained simultaneously; Construct a water content-salinity joint constraint matrix using volumetric water content w and salinity parameter s: M constraint =[α·w,β·s], where α is the water content normalization coefficient and β is the salinity normalization coefficient. Both are pre-calibrated according to the measurement range. In this embodiment, the value of α is 0.6 and the value of β is 0.4. The joint constraint matrix M is... constraint As input to the regularization constraint term of the deconvolution model, it is used to limit the search range of the inorganic chlorine signal component in the subsequent step S4.

[0023] In one embodiment of the present invention, step S3 includes the following steps: The crude oil measurement area was irradiated with low-energy and high-energy excitation beams, respectively, and the low-energy Compton scattering continuous spectrum was collected. low (s) and high-energy Compton scattering continuous spectrum l high (s); Using the preset energy resolution to l low (s) and l high (s) are discretized into N energy channels, and the probability of photon count in each energy channel relative to the total count is calculated to obtain the probability distribution p of the low-energy spectrum. low i The probability distribution p of the high energy spectrum highi Where i = 1, 2, ..., N, and N is the total number of energy channels; Calculate the fluctuation characteristic parameters of the low-energy scattering spectrum respectively. Fluctuation characteristic parameters of high-energy scattering spectrum ; where p low i and p low i+1 p represents the probability values ​​of the i-th and (i+1)-th energy channels in the low-energy scattering spectrum, respectively. high i and p high i+1 These are the probability values ​​of the i-th and (i+1)-th energy channels in the high-energy scattering spectrum, respectively. Calculate the Fisher information content F of the low-energy scattering spectrum respectively. low Fisher information content F of high-energy scattering spectrum high ; Calculate Fisher information ratio .

[0024] Specifically, after completing the acquisition of low-energy and high-energy excitation spectra, the crude oil measurement area was irradiated with low-energy and high-energy excitation beams respectively, and the low-energy Compton scattering continuous spectrum was acquired. low (s) and high-energy Compton scattering continuous spectrum l high (s), where the Compton scattering continuous spectrum refers to the scattering signal with a continuous energy distribution generated after X-ray photons collide with electrons in crude oil in an inelastic manner. Its intensity varies with the scattering angle and energy, reflecting the electron density distribution information of the scatterer. Using the preset energy resolution to l low (s) and l high (s) are discretized into N energy channels, where the total number of energy channels N is determined by the energy spectrum measurement range and the preset energy resolution, specifically as follows: E max With E min δE represents the upper and lower energy limits of the energy spectrum, respectively, and δE is the preset energy resolution. The probability distribution p of the low-energy spectrum is obtained by statistically analyzing the photon count in each energy channel relative to the total count. low i The probability distribution p of the high energy spectrum high i Where i = 1, 2, ..., N; the probability distribution is calculated as follows: , where C i Count the number of photons in the i-th energy channel; Among them, fluctuation characteristic parameters This reflects the drastic change in photon counting probability between adjacent energy channels; when the scattering spectrum contains structural information in a certain energy region (e.g., superimposed with the scattering characteristics of salt water droplets), the probability change rate between adjacent channels in that region increases, and the fluctuation characteristic parameter increases accordingly; where, when p i +p i+1 When =0, let σ i =0; for i=N, define p N+1 =p N ; Among them, the Fisher information content F of the low-energy scattering spectrum low The specific calculation method is as follows: Fisher information content F of high-energy scattering spectrum high The specific calculation method is as follows: Fisher information is a statistical measure in information theory used to measure the amount of information carried by the probability distribution of a random variable. In physical measurements, Fisher information reflects the cumulative effect of the "steepness" or "rate of change" of the probability distribution curve. For X-ray scattering spectra, when the scattering signal contains structural information of brine droplets, the probability distribution shows concentration or violent fluctuations in some energy channels, and the Fisher information increases accordingly. Therefore, Fisher information can be used as a quantitative indicator to describe the morphological characteristics of brine droplets implied in the scattering spectrum.

[0025] In one embodiment of the present invention, in step S3, l is set to a preset energy resolution. low (s) and l high (s) is discretized into N energy channels, and the following steps are also included: High-energy Compton scattering continuous spectrum l high The photon counting probability p of each energy channel of (s) high i, calculate the fluctuation characteristic parameter σ high i To obtain the fluctuation feature sequence {σ high 1,σ high 2,…,σ high N}; The fluctuation feature sequence σ is subjected to a sliding window length M. high i Perform local standard deviation calculation to obtain the local standard deviation sequence {τ}. j}, where j = N - M + 1, τ j =std(σ high j ,σ high j+1 ,…,σ high j+M-1 ), where std is the standard deviation; The length M of the sliding window is dynamically determined based on the excitation energy of the high-energy excitation beam. Extract the local standard deviation sequence {τ j The peak position j peak ; The peak position j peak Together with the Fisher information ratio R, a scattering fingerprint feature vector F=[j peak ,R], is used as a scattering fingerprint feature parameter.

[0026] Specifically, a sliding window refers to starting from the first element of the sequence, taking M consecutive data points as a window each time, calculating the standard deviation of the M data points within that window, then moving the window forward by one data point, and repeating the above calculation until the entire sequence is covered; where the local standard deviation sequence {τ} is the standard deviation of the sequence. j The calculation method for} is as follows: For the j-th window, j=1,2,…,N-M+1, this window contains the fluctuation characteristic parameter σ. high j ,σ high j+1 ,…,σ high j+M-1 There are a total of M data points; first, calculate the average of the M data points within this window. Then calculate the standard deviation of the M data points within that window. Where sqrt is the square root operation; calculated sequentially from j=1 to j=N-M+1 in the above manner, the local standard deviation sequence {τ1,τ2,…,τ} is obtained. N-M+1}; where the value of the sliding window length M is determined by the excitation energy E of the high-energy excitation beam. high Dynamically determined, specifically: E window The local capability window width for feature extraction is preset, for example, 2keV~5keV near the main scattering peak, and δE is the preset energy resolution. This setting method makes the window length match the actual resolution of the energy spectrum, ensuring that the number of data points in each window is sufficient for statistically meaningful calculations. Among them, the fluctuation characteristic sequence {σ high i The local standard deviation sequence {τ} reflects the degree of drastic change in the photon count probability between adjacent channels. j Based on this, it further describes "in which regions the amplitude of change fluctuates the most"; when the fluctuation characteristics show drastic changes in a certain energy region (with multiple large σ values ​​appearing in clusters), the local standard deviation τ of that region... j The corresponding increase; therefore, τ j Peak position j of the sequence peakThe corresponding energy region is precisely the area where the superimposed structural information is most concentrated in the Compton scattering continuum spectrum. This location is positively correlated with the main peak position of the droplet size distribution in brine microparticles in crude oil, specifically: j peak A rightward shift indicates that structural information is concentrated at the high-energy end, corresponding to smaller droplets; j peak A leftward shift indicates that the structural information is concentrated at the low-energy end, corresponding to a larger-sized droplet.

[0027] In one embodiment of the present invention, step S4, constructing the deconvolution mathematical model of the total chlorine signal, includes the following steps: The net peak area A of the Cl-Kα characteristic peak in the low-energy excitation spectrum Cl,low The net peak area A of the Cl-Kα characteristic peak in the high-energy excitation spectrum Cl,high Construct the measured total chlorine signal vector b=[A Cl,low, A Cl,high ] T ; The unit concentration response coefficient k of organochlorine under low-energy excitation org,low The unit concentration response coefficient k under high-energy excitation org,high And the unit concentration response coefficient k of inorganic chlorine under low-energy excitation. inorg,low The unit concentration response coefficient k under high-energy excitation inorg,high Construct the response matrix K=[[k org,low ,k inorg,low ],[k org,high ,k inorg,high ]]; Where, k inorg,low With k inorg,high Based on Fisher's information ratio R, the updates are dynamic, specifically: k inorg,low =k 0 inorg,low ·(1+α low ·R); k inorg,high =k 0 inorg,high ·(1+α high ·R); where k 0 inorg,low k is the reference response coefficient of inorganic chlorine under low-energy excitation. 0 inorg,high α is the baseline response coefficient of inorganic chlorine under high-energy excitation. low α is the low-energy adjustment factor. high This is a high-energy adjustment factor; Construct the theoretical prediction vector K·[C org C inorg ] TThe residual sum of squares between the measured total chlorine signal vector b and the measured total chlorine signal vector b, and using the water content-salinity joint constraint matrix constructed with water content and salinity parameters as regularization constraints, the organochlorine concentration C is solved using the Tikhonov regularization method. org and inorganic chlorine concentration C inorg .

[0028] Specifically, based on completing steps S1 to S3 above, the mathematical relationship between the total chlorine signal and the concentrations of organic and inorganic chlorine is established using the basic parameter method in X-ray fluorescence quantitative analysis. Specifically, the energy of the acquired low-energy excitation spectra and high-energy excitation spectra is calibrated, background subtraction is performed, and characteristic peak areas are extracted. A digital filtering method (e.g., the SNIP algorithm) is used to subtract the continuous background. A Gaussian function or Voigt function is used to fit the Cl-Kα characteristic peaks to obtain the net peak area A of the Cl-Kα characteristic peaks in the low-energy excitation spectra. Cl,low The net peak area A of the Cl-Kα characteristic peak in the high-energy excitation spectrum Cl,high ; to A Cl,low With A Cl,high Construct the measured total chlorine signal vector b=[A Cl,low, A Cl,high ] T Where T is the transpose operator; For chlorine-containing components in crude oil, the characteristic X-ray intensity (based on net peak area) produced under X-ray excitation exhibits a linear relationship with the component concentration: I = k·C, where I is the net peak area of ​​the characteristic peak, k is the response coefficient per unit concentration, and C is the component concentration. Based on this principle, a response matrix K = [[k org,low ,k inorg,low ],[k org,high ,k inorg,high Since organochlorine compounds are uniformly dissolved in the oil phase in the form of covalent bonds, their form does not change with changes in the aqueous phase of crude oil. Therefore, k org,low With k org,high Under given measurement conditions, the value is fixed and obtained through standard sample determination; however, inorganic chlorine exists in saline droplets, and its X-ray excitation efficiency is affected by the droplet size and distribution. When the morphology of the saline droplets changes, the response coefficient of inorganic chlorine changes accordingly. Therefore, k... inorg,low With k inorg,high Set to dynamically update based on the Fisher information ratio R, specifically: k inorg,low =k 0 inorg,low ·(1+α low ·R); k inorg,high =k 0 inorg,high ·(1+α high ·R); where k 0inorg,low k represents the baseline response coefficient of inorganic chlorine under low-energy excitation (i.e., the response coefficient when R=0). 0 inorg,high The reference response coefficients for inorganic chlorine under high-energy excitation are α and β, which are obtained by measuring standard samples. high These are high-energy adjustment coefficients, all dimensionless positive numbers. Their values ​​are determined through least-squares fitting. Specifically, a set of standard oil samples containing inorganic chlorine is prepared, and the R-value and characteristic peak area of ​​inorganic chlorine for each sample are recorded. k... inorg =k 0 inorg ·(1+α·R) represents the fitting model, and α is determined using the least squares method. low and α high The values ​​of , where α high >α low >0.

[0029] In one embodiment of the present invention, the method of using the Tikhonov regularization method to solve the regularization objective function of the matrix equation specifically includes: min{||K·[C org C inorg ] T -b|| 2 +λ·||Γ·[C org C inorg ] T || 2}, Where λ is the regularization parameter and Γ is the penalty matrix constructed based on the water content-salinity joint constraint matrix; The penalty matrix Γ is a 2×2 diagonal matrix, specifically: Γ=diag(γ org ,γ inorg ), γ inorg =η·w·s,γ org =ε·γ inorg Where, diag is a diagonal matrix, γ org γ represents the penalty weight corresponding to the organochlorine concentration. inorg η is the penalty weight corresponding to the inorganic chlorine concentration, w is the obtained volumetric water content, s is the obtained salinity, η is the first proportionality coefficient, and ε is the second proportionality coefficient.

[0030] Specifically, given the measured total chlorine signal vector b and the organic chlorine concentration C to be determined... org Inorganic chlorine concentration C inorg The response matrix equation b=K·[C] is satisfied between them. org C inorg ] THowever, since the measured total chlorine signal vector b contains measurement noise and the condition number of the response matrix K may be large, directly solving the matrix equation will lead to instability of the solution—small noise causes large fluctuations in the concentration solution; in order to obtain a stable concentration solution, the Tikhonov regularization method is used to transform the original problem into an optimization problem with a penalty term. Where λ is the regularization parameter, used to balance the weight between fitting accuracy and solution stability; the value of λ is determined by using the L-curve method, within a range of 10... -4 ~10 4 Fifty candidate λ values ​​are selected from the range, and each is substituted into the objective function to solve the problem. The sum of squared residuals and the logarithm of the regularization term corresponding to each λ value are recorded. An L-curve is plotted with these two values ​​as the x and y axes. The curvature of each point on the curve is calculated, and the λ value corresponding to the point with the maximum curvature is taken as the optimal regularization parameter. The formula for calculating curvature κ(λ) is as follows: Where ρ = log||K·Cb|| 2 θ=log||Γ·C|| 2 , and Let be the first and second derivatives of ρ with respect to logλ, respectively. and The first and second derivatives of θ with respect to logλ are given, respectively. The objective function for each value of λ is solved using the conjugate gradient method, and the iterative convergence condition is set to a residual descent rate of less than 10. -6 The iteration terminates at the specified time. Wherein, η is the first proportionality coefficient and ε is the second proportionality coefficient. η and ε are determined by fitting standard samples. The fitting method for η is: prepare 5 to 10 standard samples with known w, s and inorganic chlorine concentrations, and fit η with the goal of minimizing the inversion error of all standard samples; the fitting method for ε is: prepare standard samples containing organic chlorine, and fit ε with the goal of minimizing the inversion error of organic chlorine concentration. Substituting the determined regularization parameter λ and penalty matrix Γ into the objective function, which is a convex function, we can iteratively solve it using gradient descent or conjugate gradient methods to obtain the value C that minimizes the objective function. org and C inorg The values ​​represent the final output concentrations of organic and inorganic chlorine.

[0031] Please see Figure 2 As shown, this invention is an online monitoring system for crude oil chlorine content based on X-rays, comprising the following modules: Dual-energy synchronous excitation acquisition module: A low-energy X-ray source and a high-energy X-ray source are used to simultaneously apply dual-level excitation to the crude oil flowing in the pipeline, and the low-energy excitation energy spectrum and the high-energy excitation energy spectrum are acquired respectively; The low-energy excitation spectrum contains the Cl-Kα characteristic peak of chlorine, and the high-energy excitation spectrum contains the Cl-Kα characteristic peak of chlorine and the S-Kα characteristic peak of sulfur. Online process parameter acquisition module: Acquires the current water content and salinity parameters of crude oil online, and inputs them as constraints into the deconvolution model; Scattering fingerprint feature extraction module: Performs statistical feature analysis on the Compton scattering continuous spectrum in the high-energy excitation energy spectrum to extract scattering fingerprint feature parameters; Total chlorine signal deconvolution solution module: Using the scattering fingerprint feature parameters and the water content and salinity parameters as known constraints, a mathematical model for deconvolution of the total chlorine signal is constructed. The low-energy excitation energy spectrum and the high-energy excitation energy spectrum are combined for iterative solution to decompose the total chlorine signal into organic chlorine signal components and inorganic chlorine signal components, and the organic chlorine concentration value and inorganic chlorine concentration value are output respectively.

[0032] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A method for online monitoring of crude oil chlorine content based on X-rays, characterized in that, Includes the following steps: S1: A dual-level excitation is simultaneously applied to the crude oil flowing in the pipeline using a low-energy X-ray source and a high-energy X-ray source, and the low-energy excitation energy spectrum and the high-energy excitation energy spectrum are collected respectively. The low-energy excitation spectrum contains the Cl-Kα characteristic peak of chlorine, and the high-energy excitation spectrum contains the Cl-Kα characteristic peak of chlorine and the S-Kα characteristic peak of sulfur. S2: Obtain the current water content and salinity parameters of crude oil online and input them as constraints into the deconvolution model; S3: Perform statistical feature analysis on the Compton scattering continuous spectrum in the high-energy excitation spectrum to extract scattering fingerprint feature parameters; S4: Using the scattering fingerprint feature parameters and the water content and salinity parameters as known constraints, construct a mathematical model for the deconvolution of the total chlorine signal, and iteratively solve the low-energy excitation spectrum and the high-energy excitation spectrum to decompose the total chlorine signal into organic chlorine signal components and inorganic chlorine signal components, and output the organic chlorine concentration value and the inorganic chlorine concentration value respectively.

2. The method for online monitoring of crude oil chlorine content based on X-rays according to claim 1, characterized in that, In step S1, the low-energy X-ray source and the high-energy X-ray source are arranged at intervals along the crude oil flow direction in the pipeline. The low-energy excitation beam forms a first excitation spot in the crude oil measurement area, and the high-energy excitation beam forms a second excitation spot in the crude oil measurement area. The distance between the first excitation spot and the second excitation spot in the flow direction is L, where L≥d, and d is the width of the first excitation spot in the flow direction. It also includes a flow velocity measurement unit, which is used to measure the flow velocity v of crude oil in the pipeline in real time, and to measure and output the flow velocity v to the trigger control terminal of the second X-ray detector in real time at a preset sampling frequency. L = v × Δt, where Δt is a preset time delay and is dynamically updated according to the flow velocity v; The low-energy excitation spectrum is acquired by a first X-ray detector, and the high-energy excitation spectrum is acquired by a second X-ray detector. The second X-ray detector starts acquisition after a delay of Δt. During acquisition, the bypass quick-opening solenoid valves at both ends of the measurement area of ​​the control pipeline are switched to a stop-flow closed state, and the energy spectrum is integrated synchronously on the static crude oil section. Alternatively, a high-speed multichannel pulse analyzer is used to perform multi-cycle cyclic acquisition with a short integration window width on the order of milliseconds, and sliding matching and spectral superposition of each cycle short window sequence are performed on the digital end according to the delay Δt. The low-energy excitation beam and the high-energy excitation beam are respectively incident on the crude oil measurement area inside the pipeline in a manner perpendicular to the flow direction.

3. The method for online monitoring of crude oil chlorine content based on X-rays according to claim 1, characterized in that, Step S2 includes the following steps: A single microwave resonant cavity dual-frequency multiplexing structure is adopted to transmit microwave signals of the first and second frequency bands to crude oil flowing through the same measurement section, and to collect the first resonant frequency and the first quality factor, as well as the second resonant frequency and the second quality factor. The high and low frequency signal responses are extracted for differential calculation. The volumetric water content and salinity parameters of the current crude oil are calculated by inversion based on the frequency difference signal and the quality factor difference signal. A water content-salinity joint constraint matrix is ​​constructed using volumetric water content and salinity parameters, and this joint constraint matrix is ​​used as the input of the regularization constraint term for the deconvolution model.

4. The method for online monitoring of crude oil chlorine content based on X-rays according to claim 1, characterized in that, Step S3 includes the following steps: The crude oil measurement area was irradiated with low-energy and high-energy excitation beams, respectively, and the low-energy Compton scattering continuous spectrum was collected. low (s) and high-energy Compton scattering continuous spectrum l high (s); Using the preset energy resolution to l low (s) and l high (s) are discretized into N energy channels, and the probability of photon count in each energy channel relative to the total count is calculated to obtain the probability distribution p of the low-energy spectrum. low i The probability distribution p of the high energy spectrum high i Where i = 1, 2, ..., N, and N is the total number of energy channels; Calculate the fluctuation characteristic parameters of the low-energy scattering spectrum respectively. Fluctuation characteristic parameters of high-energy scattering spectrum ; where p low i and p lo00w i+1 p represents the probability values ​​of the i-th and (i+1)-th energy channels in the low-energy scattering spectrum, respectively. high i and p high i+1 These are the probability values ​​of the i-th and (i+1)-th energy channels in the high-energy scattering spectrum, respectively. Calculate the Fisher information content F of the low-energy scattering spectrum respectively. low Fisher information content F of high-energy scattering spectrum high ; Calculate Fisher information ratio .

5. The method for online monitoring of crude oil chlorine content based on X-rays according to claim 4, characterized in that, In step S3, l is set to a preset energy resolution. low (s) and l high (s) is discretized into N energy channels, and the following steps are also included: High-energy Compton scattering continuous spectrum l high The photon counting probability p of each energy channel of (s) high i Calculate the fluctuation characteristic parameter σ high i To obtain the fluctuation feature sequence {σ high 1,σ high 2,…,σ high N }; Using the sliding window length M to represent the fluctuation feature sequence {σ high i } Perform local standard deviation calculation to obtain the local standard deviation sequence {τ} j }, where j = N - M + 1, τ j =std(σ high j ,σ high j+1 ,…,σ high j+M-1 ), where std is the standard deviation; The length M of the sliding window is dynamically determined based on the excitation energy of the high-energy excitation beam. Extract the local standard deviation sequence {τ j The peak position j peak ; The peak position j peak Together with the Fisher information ratio R, a scattering fingerprint feature vector F=[j peak ,R], is used as a scattering fingerprint feature parameter.

6. The method for online monitoring of crude oil chlorine content based on X-rays according to claim 1, characterized in that, In step S4, constructing the deconvolution mathematical model of the total chlorine signal includes the following steps: The net peak area A of the Cl-Kα characteristic peak in the low-energy excitation spectrum Cl,low The net peak area A of the Cl-Kα characteristic peak in the high-energy excitation spectrum Cl,high Construct the measured total chlorine signal vector b=[A Cl,low A Cl,high ] T ; The unit concentration response coefficient k of organochlorine under low-energy excitation org,low The unit concentration response coefficient k under high-energy excitation org,high And the unit concentration response coefficient k of inorganic chlorine under low-energy excitation. inorg,low The unit concentration response coefficient k under high-energy excitation inorg,high Construct the response matrix K=[[k org,low ,k inorg,low ],[k org,high ,k inorg,high ]]; Where, k inorg,low With k inorg,high Based on Fisher's information ratio R, the updates are dynamic, specifically: k inorg,low =k 0 inorg,low ·(1+α low ·R); k inorg,high =k 0 inorg,high ·(1+α high ·R); where k 0 inorg,low k is the reference response coefficient of inorganic chlorine under low-energy excitation. 0 inorg,high α is the baseline response coefficient of inorganic chlorine under high-energy excitation. low α is the low-energy adjustment factor. high This is a high-energy adjustment factor; Construct the theoretical prediction vector K·[C org C inorg ] T The residual sum of squares between the measured total chlorine signal vector b and the measured total chlorine signal vector b, and using the water content-salinity joint constraint matrix constructed with water content and salinity parameters as regularization constraints, the organochlorine concentration C is solved using the Tikhonov regularization method. org and inorganic chlorine concentration C inorg .

7. The method for online monitoring of crude oil chlorine content based on X-rays according to claim 6, characterized in that, The regularization objective function for solving the matrix equation using the Tikhonov regularization method is as follows: min{||K·[C org ,C inorg ] T -b|| 2 +λ·||Γ·[C org ,C inorg ] T || 2 }, Where λ is the regularization parameter and Γ is the penalty matrix constructed based on the water content-salinity joint constraint matrix; The penalty matrix Γ is a 2×2 diagonal matrix, specifically: Γ=diag(γ org ,γ inorg ), γ inorg =η·w·s,γ org =ε·γ inorg Where, diag is a diagonal matrix, γ org γ represents the penalty weight corresponding to the organochlorine concentration. inorg η is the penalty weight corresponding to the inorganic chlorine concentration, w is the obtained volumetric water content, s is the obtained salinity, η is the first proportionality coefficient, and ε is the second proportionality coefficient.

8. An X-ray-based online monitoring system for crude oil chlorine content, used to implement the X-ray-based online monitoring method for crude oil chlorine content as described in any one of claims 1-7, characterized in that, Includes the following modules: Dual-energy synchronous excitation acquisition module: A low-energy X-ray source and a high-energy X-ray source are used to simultaneously apply dual-energy level excitation to the crude oil flowing in the pipeline, and the low-energy excitation energy spectrum and the high-energy excitation energy spectrum are acquired respectively; The low-energy excitation spectrum contains the Cl-Kα characteristic peak of chlorine, and the high-energy excitation spectrum contains the Cl-Kα characteristic peak of chlorine and the S-Kα characteristic peak of sulfur. Online process parameter acquisition module: Acquires the current water content and salinity parameters of crude oil online, and inputs them as constraints into the deconvolution model; Scattering fingerprint feature extraction module: Performs statistical feature analysis on the Compton scattering continuous spectrum in the high-energy excitation energy spectrum to extract scattering fingerprint feature parameters; Total chlorine signal deconvolution solution module: Using the scattering fingerprint feature parameters and the water content and salinity parameters as known constraints, a mathematical model for deconvolution of the total chlorine signal is constructed. The low-energy excitation energy spectrum and the high-energy excitation energy spectrum are combined for iterative solution to decompose the total chlorine signal into organic chlorine signal components and inorganic chlorine signal components, and the organic chlorine concentration value and inorganic chlorine concentration value are output respectively.

Images