Optimal matching method of transmitting and receiving structure of optical tomography sensor

Through simulation measurement model and simulation optimization, combined with image reconstruction algorithm, the optimal emission-receiving structure of optical tomography sensors was determined, which solved the problems of high cost and reduced marginal benefits of image quality improvement in the prior art, and achieved different Efficient detection of optical tomography sensors in the environment.

CN113312764BActive Publication Date: 2024-11-12HANGZHOU DIANZI UNIV
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
CN202110545856.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-05-19
Publication Date
2024-11-12
Estimated Expiration
2041-05-19

AI Technical Summary

Technical Problem

The existing optical tomography sensors lack the system's emission-receive structure optimization in multiphase flow detection, resulting in high costs and reduced marginal benefits of image reconstruction quality, making it difficult to determine the optimal proportional structure under cost constraints.

Method used

By designing a simulation measurement model, using Comsol Multiphysics for geometric modeling and simulation, combining Matlab for image reconstruction algorithm optimization, building the optimal emission-receiving structure of the ring optical array sensor, using a fan-shaped laser diode and photodiode to optimize the laser diode and photoelectricity The number of diodes is used to reconstruct the image using linear backprojection and Landweber algorithm, and the optimal ratio is determined through cost function optimization.

Benefits of technology

The optimal proportional structure of optical tomography sensors under different cost models is realized, which improves image reconstruction quality, reduces costs, and provides a new idea for multi-phase flow detection.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN113312764B_ABST
    Figure CN113312764B_ABST
Patent Text Reader

Abstract

The present invention relates to an optimal matching method for the transmitting and receiving structure of an optical tomography sensor. The present invention designs a simulation measurement model and performs geometric modeling through Comsol Multiphysics; imports the output sensitive field, simulated projection signal and image reconstruction algorithm into Matlab to perform image reconstruction; uses a linear back-projection algorithm and a new Landweber algorithm to perform image reconstruction; performs error analysis on multiple groups of reconstructed image data on the impact of the accuracy of the reconstructed image, and simultaneously models the sensor cost function for optimization research and produces corresponding images; determines the optimal transmitting and receiving structure of optical tomography. The present invention overcomes the problem that the specific structure of the annular optical array sensor is difficult to determine directly under different parameter environments. It can also be applied to other practical measurement fields, and provides new ideas for the optimization of specific measurement structures.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field of optical tomography, and relates to an optimal matching method of a transmitting-receiving structure of an optical tomography sensor. Background Art

[0002] Multiphase flow systems are widely present in many fields such as industry and agriculture, chemical industry, energy industry and environmental engineering. In industrial production activities, the uncertainty of multiphase flow will have an important impact on the overall production process, which is directly related to the quality and safety of the products produced. Therefore, the detection and control of the multiphase flow process is extremely necessary. In recent years, with the development of national science and technology and the needs of economic construction, people have paid more and more attention to the detection of multiphase flow parameters. However, in the process of multiphase flow detection, the complex and changeable flow characteristics of multiphase flow, such as non-uniform mixing, unstable flow, and interaction between phases, bring great difficulty to the detection. Among the existing detection technologies, tomography technology can obtain cross-sectional information and reconstruct cross-sectional flow images, and it has been widely valued in the detection of multiphase flow parameters. Among them, compared with ultrasonic tomography, nuclear magnetic resonance tomography, electrical tomography and other technologies, the measurement sensitive field of optical tomography technology is a hard field, which has little interference with the flow field, is sensitive and not easily affected, and has fast and simple data acquisition and high temporal and spatial resolution. It has its unique advantages, especially in the case of low solid and liquid concentration distribution, and is easier to achieve real-time and long-distance continuous measurement of multiphase flow. Therefore, domestic and foreign researchers have paid much attention to optical PT technology.

[0003] In recent years, domestic and foreign scholars have proposed a variety of improved structures for the sensor structure of optical tomography technology and achieved good results. In 2003, Yan Chunsheng of Zhejiang University designed an optical fiber process tomography (OFPT) structure with eight OFS units, which solved the problems of low resolution and large volume; in 2007, E Schleicher of Germany developed and invented a circular fan beam space array sensor with 256 (fan beam) laser transmitters and 32 receivers for the measurement of single-phase flow and gas-liquid two-phase flow; in 2013, Mohd Fadzil et al. of the University of Technology of Malaysia developed a bubble column laser optical tomography sensor with 16 (parallel beam) laser diodes and 16 photodiodes for the detection of gas-liquid two-phase flow.

[0004] However, in order to obtain rich optical signals and accurate parameter models, most researchers have used a large number of transmitters and receivers, but have not conducted systematic optimization research on the structure of the transmitter-receiver sensor. Studies have shown that the marginal benefits of increasing the number of transmitters and receivers to improve image reconstruction quality decrease with the increase in the number, and the marginal benefits of receivers are higher than those of transmitters. Considering the cost, how to find the optimal ratio structure of the annular optical array transmitter-receiver sensor under certain cost constraints still lacks systematic research and theoretical analysis. Summary of the invention

[0005] The purpose of the present invention is to provide a method for optimally matching the transmitting and receiving structures of an optical tomography sensor, and to construct a sensor structure with the best matching ratio for different cost models in different scenarios.

[0006] The present invention specifically comprises the following steps:

[0007] Step 1: Design a simulation measurement model; the inner diameter of the annular optical array sensor is d i , the outer diameter is d o ; The transmitter and receiver are laser diodes and photodiodes respectively, the transmitter and receiver are evenly and equidistantly distributed on the optical array sensor, and the laser type is fan-shaped laser;

[0008] Step 2: Preliminarily set the number of laser diodes and photodiodes to M×N;

[0009] Step 3: Build a physical model, as follows:

[0010] Step (1), firstly, geometric modeling is performed by Comsol Multiphysics, where there are M laser diodes and N photodiodes on the pipeline;

[0011] Step (2), calculate the sensitive field of tomography. The sensitive field is divided into square grids with a side length of b and a number of pixels in the target area of ​​p. n Using Comsol ray optics module, regional pixel points are set in sequence and single laser is activated to obtain photodiode measurement signals. The complete sensitive field matrix dimension is M×N×p n ;

[0012] Step (3), place a diameter d in the pipeline through Comsol simulation. c The black circular occluder is used to construct the geometric model in this case;

[0013] Step (4), using the Comsol ray optics module to obtain a projection signal, activating M laser tubes in sequence, and using N photodiodes to collect simulated projection signals, the simulated projection signal matrix dimension obtained by a single measurement is M×N;

[0014] Step 4: Import the sensitive field, simulated projection signal and image reconstruction algorithm output by Comsol Multiphysics into Matlab for image reconstruction; use the linear back-projection algorithm and a new Landweber algorithm for image reconstruction. The steps are as follows:

[0015] Step 1: Use the linear back-projection algorithm to find the initial solution for image reconstruction:

[0016] P=D T T;

[0017] In the formula, P represents the parameter matrix to be measured, T represents the projection data matrix, D represents the sensitive field matrix information, and D T It can be regarded as an adjoint mapping from the projection information to the required area information;

[0018] Step 2: Determine the minimum mean square error of the noise data as the objective function: min{f(P)}; in the formula

[0019]

[0020] Step 3: Set the descent gradient to:

[0021] Step 4: Substitute the steepest descent formula to obtain:

[0022]

[0023] In the formula, μ represents the iteration step, k represents the number of iterations, k = 1, 2, 3…

[0024] Step 5: Use the convex projection function Q(·) to transform P k Projecting into the specified value range [0,1] accelerates the convergence of the algorithm:

[0025]

[0026] Step 6: Set the loop termination condition. If the result meets the termination condition, exit the iteration. Otherwise, repeat steps 4-5.

[0027] Step 5: Change the values ​​of M and N in step 2 to study the effect of the number of laser diodes and photodiodes on the accuracy of the reconstructed image. Perform error analysis on multiple sets of reconstructed image data and model the sensor cost function for optimization and generate corresponding images. The error analysis evaluation index uses relative image error to evaluate the quality of the reconstructed image, which is defined as follows:

[0028]

[0029] In the formula, γ represents the relative image error, represents the algorithm reconstructed image, and x represents the target image;

[0030] The sensor cost function is: Y = K1X1 + K2X2 + K8 (X1X2) 2 ; In the formula, Y represents the total price; X1 and X2 represent the number of laser diodes and photodiodes respectively; K1, K2, K8 represent the corresponding price coefficients;

[0031] Step 6: Determine the optimal transmitting-receiving structure for optical tomography through the error analysis and sensor cost function in step 5.

[0032] The present invention provides a research method for studying the optimal ratio structure of optical tomography transmitter-receiver sensors, which overcomes the difficulty of directly determining the specific structure of annular optical array sensors under different parameter environments. It can also be applied to other practical measurement fields, providing new ideas for the optimization of specific measurement structures. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 is a flow chart of the present invention;

[0034] Figure 2 It is the COMSOL simulation model structure;

[0035] FIG3(a) is a table of image reconstruction data when the number of photodiodes is changed;

[0036] FIG3( b ) is a graph showing the image reconstruction error when the number of photodiodes is changed;

[0037] FIG4(a) is a table showing image reconstruction data when the number of laser diodes is changed;

[0038] FIG4( b ) is a graph showing the image reconstruction error when the number of laser diodes is changed;

[0039] FIG5( a ) is a cost curve diagram showing the cost of adding a cost function and changing the number of photodiodes;

[0040] FIG5( b ) is a cost curve diagram showing the cost function added and the number of laser diodes changed. DETAILED DESCRIPTION

[0041] In order to overcome the problem that the specific structure of optical tomography sensors is difficult to determine directly under different environments, the present invention carries out optimization research on factors such as transmission-reception structure, cost, and algorithm, and constructs a more accurate optical array receiving-transmitting sensor ratio under certain constraints, providing new ideas for the practical application of optical tomography sensors.

[0042] The following embodiments are only for illustrating the technical concept and features of the present invention, and their purpose is to enable people familiar with the technology to understand the content of the present invention and implement it accordingly, and they cannot be used to limit the protection scope of the present invention. Any equivalent changes or modifications made according to the spirit of the present invention should be included in the protection scope of the present invention.

[0043] like Figure 1 As shown, a method for optimizing the ratio of the transmitting and receiving structures of an optical tomography sensor is described, and the specific implementation process is as follows:

[0044] Step 1: Design a simulation measurement model based on the actual application. The inner diameter of the annular optical array sensor is d i , the outer diameter is d o The transmitter and receiver are laser diodes and photodiodes respectively, which are evenly and equidistantly distributed on the optical array sensor, and the laser type is fan-shaped laser.

[0045] Step 2: Preliminarily set the number of laser diodes and photodiodes to M×N.

[0046] Step 3: Use Comsol Multiphysics to build a physical model for this scenario.

[0047] (COMSOL Multiphysics is a simulation platform that implements all steps involved in the modeling workflow: from geometric modeling, defining material properties, setting up the physics fields to describe the physical phenomena, to solving the model, and post-processing the model to provide accurate and reliable results)

[0048] The details are as follows:

[0049] Step (1), first use Comsol Multiphysics to perform Figure 2 In the geometric modeling shown, there are M laser diodes and N photodiodes on the pipeline.

[0050] Step (2), calculate the sensitive field of tomography. The sensitive field is divided into square grids with a side length of b and a number of pixels in the target area of ​​p. nUsing the Comsol ray optics module, regional pixel points are set in sequence and a single laser is activated to obtain the photodiode measurement signal. The complete sensitive field matrix dimension is M×N×p n .

[0051] Step (3), place a diameter d in the pipeline through Comsol simulation. c The black circular occluder is used to construct the geometric model in this case.

[0052] Step (4), using the Comsol ray optics module to obtain the projection signal, activating M laser tubes in sequence, and using N photodiodes to collect the simulated projection signal. The simulated projection signal matrix dimension obtained by a single measurement is M×N.

[0053] Step 4: Import the sensitive field, simulated projection signal and image reconstruction algorithm output by Comsol Multiphysics into Matlab for image reconstruction. The linear back-projection algorithm and a new Landweber algorithm are used for image reconstruction. The steps are as follows:

[0054] Step 1: Use the linear back projection algorithm (LBP) to find the initial solution for image reconstruction:

[0055] P=D T T;

[0056] In the formula, P represents the parameter matrix to be measured, T represents the projection data matrix, D represents the sensitive field matrix information, and D T It can be regarded as an accompanying mapping from projection information to the required area information.

[0057] Step 2: Determine the minimum mean square error of the noise data as the objective function: min{f(P)}; in the formula

[0058]

[0059] Step 3: Set the descent gradient to:

[0060] Step 4: Substitute the steepest descent formula to obtain:

[0061]

[0062] In the formula, μ represents the iteration step, k represents the number of iterations, k = 1, 2, 3…

[0063] Step 5: Use the convex projection function Q(·) to transform P k Projecting into the specified value range [0,1] accelerates the convergence of the algorithm:

[0064]

[0065] Step 6, set the loop termination condition. If the result meets the stop condition, exit the iteration. Otherwise, repeat steps 4-5. In this example, the loop termination condition is set as follows: when the number of iterations exceeds 2000, stop the iteration; or when the difference between the result of the iteration and the result of the previous iteration is less than or equal to 0.01 after a certain number of iterations, stop the iteration. Among them, the number of iterations 2000 and the difference of 0.01 between adjacent iteration results are not optimal values. From this example, it can be seen that the number of iterations is 1500-2000, and the difference threshold of adjacent iteration results is 0.01-0.1, which is a better result.

[0066] Step 5: Change the values ​​of M and N in step 2 to study the impact of the number of laser diodes and photodiodes on the accuracy of the reconstructed image. Perform error analysis on multiple sets of reconstructed image data and model the sensor cost function for optimization and generate corresponding images. The error analysis evaluation index uses relative image error to evaluate the quality of the reconstructed image, which is defined as follows:

[0067]

[0068] In the formula, γ represents the relative image error, represents the algorithm reconstructed image, and x represents the target image.

[0069] The sensor cost function is: Y = K1X1 + K2X2 + K8 (X1X2) 2 ; In the formula, Y represents the total cost; X1 and X2 represent the number of laser diodes and photodiodes respectively; K1, K2, K8 represent the corresponding price coefficients.

[0070] Step 6: Determine the optimal transmitting-receiving structure for optical tomography through the error analysis and sensor cost function in step 5.

[0071] In this embodiment, the diameter d i The optical tomography sensor structure under the 30cm pipeline is optimized and designed, where d o 30.5cm,d c is 5cm, set the simulation parameter b to 8mm, p n is 1017. The price coefficients K1, K2, and K8 are 1, 10, and 0.01 respectively. The image reconstruction effect of the number of photodiodes and the number of laser tubes M and N at 5, 10, 15, 30, and 60 is analyzed, and the optimal sensor transmission-receiving structure is determined. Figure 3(a) , 3(b)As shown in Figure 4(a) and Figure 4(b), increasing the number of photodiodes or laser diodes can improve the quality of reconstructed images and reduce image errors; the marginal benefits of increasing the number of laser diodes and photodiodes on image quality decrease as the number increases. Figure 5(a) and 5(b) As shown in Figure 1, the cost factors such as sensor structure complexity and cost increase significantly with the increase in the number of laser diodes or photodiodes. Since all curves are concave functions, the optimal transmission-reception structure of the optical array sensor is set to 30 for both the number of laser diodes and the number of photodiodes.

[0072] Matters not covered by the present invention are known technologies.

Claims

1. An optimal matching method for the transmitting-receiving structure of an optical tomography sensor, characterized in that: The specific steps include: Step 1: Design a simulation measurement model; the inner diameter of the annular optical array sensor is d i , the outer diameter is d o ; The transmitter and receiver are laser diodes and photodiodes respectively, the transmitter and receiver are evenly and equidistantly distributed on the optical array sensor, and the laser type is fan-shaped laser; Step 2: Preliminarily set the number of laser diodes and photodiodes to M×N; Step 3: Build a physical model, as follows: Step (1), firstly, geometric modeling is performed by Comsol Multiphysics, where there are M laser diodes and N photodiodes on the pipeline; Step (2), calculate the sensitive field of tomography. The sensitive field is divided into square grids with a side length of b and a number of pixels in the target area of ​​p. n Using Comsol ray optics module, regional pixel points are set in sequence and single laser is activated to obtain photodiode measurement signals. The complete sensitive field matrix dimension is M×N×p n ; Step (3), place a diameter d in the pipeline through Comsol simulation. c A black circular occluder is used to construct a geometric model; Step (4), using the Comsol ray optics module to obtain a projection signal, activating M laser tubes in sequence, and using N photodiodes to collect simulated projection signals, the simulated projection signal matrix dimension obtained by a single measurement is M×N; Step 4: Import the sensitive field, simulated projection signal and image reconstruction algorithm output by Comsol Multiphysics into Matlab for image reconstruction; use the linear back-projection algorithm and a new Landweber algorithm for image reconstruction. The steps are as follows: Step 1: Use the linear back-projection algorithm to find the initial solution for image reconstruction: P=D T T; In the formula, P represents the parameter matrix to be measured, T represents the projection data matrix, D represents the sensitive field matrix information, and D T It can be regarded as an adjoint mapping from the projection information to the required area information; Step 2: Determine the minimum mean square error of the noise data as the objective function: min{f(P)}; in the formula Step 3: Set the descent gradient to: Step 4: Substitute the steepest descent formula to obtain: In the formula, μ represents the iteration step, k represents the number of iterations, k = 1, 2, 3… Step 5: Use the convex projection function Q(·) to transform P k Projecting into the specified value range [0,1] accelerates the convergence of the algorithm: Step 6: Set the loop termination condition. If the result meets the termination condition, exit the iteration. Otherwise, repeat steps 4-5. Step 5: Change the values ​​of M and N in step 2 to study the effect of the number of laser diodes and photodiodes on the accuracy of the reconstructed image. Perform error analysis on multiple sets of reconstructed image data and model the sensor cost function for optimization and generate corresponding images. The error analysis evaluation index uses relative image error to evaluate the quality of the reconstructed image, which is defined as follows: In the formula, γ represents the relative image error, represents the algorithm reconstructed image, and x represents the target image; The sensor cost function is: Y = K1X1 + K2X2 + K3(X1X2) 2 ; In the formula, Y represents the total price; X1 and X2 represent the number of laser diodes and photodiodes respectively; K1, K2, K3 represent the corresponding price coefficients; Step 6: Determine the optimal transmitting-receiving structure for optical tomography through the error analysis and sensor cost function in step 5.

Citation Information

Patent Citations

  • A high-resolution electrical tomography conductivity distribution reconstruction method

    CN109919844A

  • Optical tomography reconstruction method and system using sparse projection-based Nesterov homotopy perturbation iteration

    CN111458300A

  • Ring-shape fan-beam optical chromatography imaging laser space array sensor

    CN2548150Y