WO2022213958A1

WO2022213958A1 - Soft measurement method for purity of phosphorus recovery product

Info

Publication number: WO2022213958A1
Application number: PCT/CN2022/085201
Authority: WO
Inventors: 李兵; 黄跃飞; 韩京成; 武晓峰
Original assignee: 清华大学深圳国际研究生院
Priority date: 2021-04-06
Filing date: 2022-04-03
Publication date: 2022-10-13
Also published as: CN113096087A

Abstract

A soft measurement method for the purity of a phosphorus recovery product, comprising the following steps: (1) photographing a collected phosphorus recovery product with an optical microscope, and recording a microstructure of the phosphorus recovery product on a two-dimensional plane; (2) processing the picture taken in step (1); (3) measuring and calculating: the distance nF_max between the farthest two points at the periphery of a particle, the distance nF_min between the closest two points at the periphery of the particle, the elongation n_El1, the length-to-diameter ratio n_El2, the total area of particles n_A, the particle perimeter n_P, the particle perimeter n_Ci, the equivalent diameter n_Ed, the solidity n_s, and the convexity-concavity n_Co; and (4) establishing a soft measurement model of the purity of the phosphorus recovery product by using a non-linear partial least squares (NLPLS) method. For the new phosphorus recovery product, predictive analysis is performed on the purity of the product by using the NLPLS model, and the model is updated by using the new predictive analysis data. The present invention provides a fast, simple and accurate method for purity determination of phosphorus recovery products.

Description

A soft-sensor method for the purity of phosphorus recovery products

technical field

The present invention is in the technical field of environmental protection, in particular to a soft measurement method for the purity of a phosphorus recovery product.

Background technique

Phosphorus is an important nutrient element, which is of great significance to human beings and various organisms in nature. Due to the massive exploitation of human beings and the unidirectional mobility of phosphorus in nature, the phosphate rock will be exhausted in the next 50 to 100 years, which will lead to a serious crisis of phosphorus resources. In order to alleviate this phenomenon, the recovery and reuse of phosphorus resources has received extensive attention. In the past ten years, the struvite crystallization method has been regarded as a very promising method to solve the crisis of phosphorus resources. It can not only precipitate phosphate in wastewater, but also the product MgNH4PO3·H2O (struvite) is a good Slow-release fertilizers have therefore received widespread attention. The crystallization process is affected by the combination of liquid-solid equilibrium thermodynamics and mass transfer between solid and liquid phases. Other process factors, such as pH, degree of supersaturation, mixing energy, temperature and the presence of competing ions, can affect the purity of the product struvite. For struvite to be an economically viable commercial fertilizer, high strength, low dissolution rate, slow release properties, and high purity are the primary characteristics.

Usually, in the process of struvite crystallization, many impurity ions will precipitate with magnesium and phosphate, resulting in impure crystals, which will reduce the application effect of the product as a fertilizer. Therefore, it is necessary to determine the guano in the precipitated product. The purity of the stone, that is, the proportion of struvite in all precipitation products. Commonly used methods for determining the purity of struvite are: 1) qualitative determination, using Fourier transform infrared spectroscopy (FTIR) or X-ray diffraction (XRD) spectroscopy to compare the morphology of solid precipitates and pure struvite; The theoretical value of struvite is used as a reference, and the percentage of each element in the solid is compared; 3) The nitrogen content is measured by an elemental analyzer and used as a reference value; 4) Quantitative X-ray diffraction method. However, due to the additional time required for sample preparation and analysis, these purity determination methods limit the overall production process, as the production line needs to be paused during the assay period until suitable product purity is achieved. Therefore, the currently used methods for measuring the purity of phosphorus recovery products all have defects such as complicated operation and long time, which greatly reduces the efficiency of the entire production process.

It should be noted that the information disclosed in the above Background section is only for understanding of the background of the application, and therefore may include information that does not form the prior art known to a person of ordinary skill in the art.

SUMMARY OF THE INVENTION

The main purpose of the present invention is to overcome the defects of the above-mentioned background technology, and to provide a soft measurement method for the purity of phosphorus recovery products.

To achieve the above object, the present invention adopts the following technical solutions:

A soft measurement method for the purity of a phosphorus recovery product, comprising the steps:

(1) Use an optical microscope to photograph the collected phosphorus recovery products, and record their microscopic topography on a two-dimensional plane;

(2) processing the pictures taken in step (1) so as to measure the physical parameters of the particles in the pictures;

(3) measure and calculate the following physical parameters of the particulate matter in the picture processed by step (2): the distance nF _max of the two farthest points around the particle; the distance nF _min of the two nearest points around the particle; elongation n _E11 ; aspect ratio n _El2 ; particle total area n _A ; particle perimeter n _P ; particle perimeter n _Ci ; equivalent diameter n _Ed ; _solidity _ns ;

(4) Based on the physical parameters of the product measured and calculated in step (3), a non-linear partial least squares NLPLS method is used to establish a soft sensor model for the purity of the phosphorus recovery product; for new phosphorus recovery products, the established NLPLS method is used. The model predicts and analyzes the product purity, and uses the new predictive analysis data combined with the original NLPLS model to update the model with a recursive algorithm.

further:

The processing in step (2) includes enhancing chromatic aberration and contrast, and performing stroke processing on the two-dimensional plane shape of the phosphorus recovery product in the picture.

In step (4), the soft-sensor model of establishing phosphorus recovery product purity specifically includes the following steps:

Ⅰ. Take the measured and calculated sample topography parameters as a data-driven sample set, expressed as {x(i), y(i)}, where x(i) represents the i-th group of input data, that is, step (3) The parameters obtained in , y(i) represents the output data of the ith group, which is the purity of the product, and the input data is formed into a matrix X, and the output data is formed into an output matrix Y;

Ⅱ. Build an NLPLS model based on input and output data, including:

Normalize the matrices X and Y so that the mean value is 0 and the variance is 1, and then the input matrix is column-expanded, and the expansion term is the hidden node output matrix G of the radial basis function (RBF) neural network and the elements are full. is a column vector 1 of 1, where each row of G corresponds to the output of a hidden node under the action of an input vector, and the coefficient of the bias term of the hidden node is 1; Partial least squares (PLS) is performed on the following augmented input matrix and output matrix )Recycle:

{[1 X G], Y}, the resulting NLPLS model is expressed as:

where XE is the augmented input matrix, A and H are the weight coefficient matrix corresponding to the original input vector and the corresponding output vector of the hidden node of the RBF network, b is the input bias vector, T is the transpose, and c is the hidden node center vector , σ is the corresponding width vector, A and H are the weight coefficient matrix.

In step II, the hidden node center vector c, the corresponding width vector σ, the weight coefficient matrices A and H, and the input bias vector b are determined as follows:

①Use the k-means clustering algorithm to cluster the input data to obtain the hidden node center c;

②Use the p-nearest neighbor rule to calculate the hidden node width:

where N is the number of hidden node centers, and ci is the p hidden node centers closest to the jth hidden node center;

③ Use PLS regression to determine the weight coefficient matrix A, H and bias vector b:

Calculate the hidden node output matrix G according to the obtained hidden node center and width, and then expand the input matrix to obtain the augmented input matrix [1 X G], perform PLS regression on the data {[1 X G], Y}, and obtain PLS The model parameter matrix {T,W,P,B,Q}, the number of extracted feature variables a is determined by the cross-check method, the obtained parameter matrix is recorded as {Ta,Wa,Pa,Ba,Qa}, and the PLS is further calculated. Regression coefficient matrix β, resulting in A, H and b.

In step (4), the new prediction analysis data is combined with the original NLPLS model, and the recursive algorithm is used to update the model, which specifically includes the following steps:

a. Record the input and output data of the new phosphorus recovery product batch as X1 and Y1 respectively, and do not contain abnormal points;

b. Determine whether to add a new hidden node. If the distance between the new data X1 and the existing RBF network hidden node center is greater than the set value, add a new hidden node; if the distance between X1 and the existing RBF network hidden node center If it is less than or equal to the set value, there is no need to add hidden nodes;

c. Expand X1, perform PLS regression, and obtain an updated NLPLS model; then calculate the weight coefficient matrix A, H and bias vector b according to the method of step ③.

In step b, if the distance between the new data X1 and the existing hidden node center of the RBF network is greater than the set value, record the new hidden node center matrix as C _gnew , and the corresponding width vector as σ _gnew . The matrix C _g , the corresponding width vector σ _g and the loading matrix P are expanded as follows:

If the distance between X1 and the existing RBF network hidden node center is less than or equal to the set value, C _g , σ _g and P remain unchanged.

Extend X1 to XE1=[1 X1 G1], where G1 is the output matrix of the hidden node for X1, let

Perform PLS regression on the data pair {X _E , Y} to get the updated NLPLS model.

A computer-readable storage medium, storing a computer program, when the computer program is executed by a processor, realizes the steps (2) to (4) of the soft measurement method of the described phosphorus recovery product purity, specifically comprising:

(2) processing the picture to measure the physical parameters of the particulate matter in the picture; wherein, the picture is obtained by using an optical microscope to photograph the collected phosphorus recovery product, and the picture records the phosphorus recovery product in a two-dimensional plane The microstructure on the top;

The present invention has the following beneficial effects:

The invention proposes a soft measurement method for the purity of phosphorus recovery products, which quantitatively describes the two-dimensional topography and structure of the product, and then adopts a nonlinear least squares (NLPLS) method to establish a soft measurement model between the shape factor and the product purity. It overcomes the defects of complicated operation and long time of the existing purity determination method, and greatly improves the efficiency of the whole production process. The method of the invention is based on the microscope photographing technology, the shape factor of the product is represented by a physical parameter, and then a non-linear least squares method is used to establish a soft measurement model between the shape factor and the product purity, and the model is used to predict the product purity, which is the best way for production Control and optimization provide operational guidance. The method of the invention effectively avoids the problems of long time consumption and signal distortion in the traditional purity measurement process, and provides a fast, simple and accurate method for the determination of the purity of the phosphorus recovery product.

Description of drawings

FIG. 1 is a schematic flowchart of a soft sensing method for a phosphorus recovery product according to an embodiment of the present invention.

Detailed ways

Embodiments of the present invention will be described in detail below. It should be emphasized that the following description is exemplary only, and is not intended to limit the scope of the invention and its application.

It should be noted that when an element is referred to as being "fixed to" or "disposed on" another element, it can be directly on the other element or indirectly on the other element. When an element is referred to as being "connected to" another element, it can be directly connected to the other element or indirectly connected to the other element. In addition, the connection can be used for both the fixing function and the coupling or communication function.

It is to be understood that the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top" , "bottom", "inside", "outside", etc. indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, which are only for the convenience of describing the embodiments of the present invention and simplifying the description, rather than indicating or implying that The device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention.

In addition, the terms "first" and "second" are only used for descriptive purposes, and should not be construed as indicating or implying relative importance or implying the number of indicated technical features. Thus, a feature defined as "first", "second" may expressly or implicitly include one or more of that feature. In the description of the embodiments of the present invention, "plurality" means two or more, unless otherwise expressly and specifically defined.

Referring to Fig. 1, a soft measurement method for the purity of a phosphorus recovery product provided by the embodiment of the present invention comprises the following steps:

When the ratio of each ion concentration Mg ²⁺ :NH ₄ ⁺ :PO ₄ ^3- :Ca ²⁺ in the wastewater is 2:4:1:0.4, white solid precipitates such as calcium phosphate, magnesium phosphate and struvite will be formed. These solid precipitated products were collected, and the phosphorus recovery products were photographed with an optical microscope to record their microscopic topography on a two-dimensional plane.

The captured pictures are processed, the background of the picture is adjusted to white, the particles are adjusted to black, the color difference and contrast are enhanced, and the two-dimensional plane shape of the phosphorus recovery product on the picture is stroked.

The following physical parameters of the visible particles in the picture were measured:

The distance between the two farthest points around the particle (Max Feret Diameter): nF _max

The distance between the two nearest points around the particle (Min Feret Diameter): nF _min

Total particle area (Area): n _A

Particle perimeter (Perimeter): n _P

The following physical quantities are calculated using the measured parameters:

Elongation:

Aspect ratio:

Circularity:

Circle equivalent Diameter:

Solidity:

Convexity:

The model establishment and purity prediction process are as follows:

Step 1: Based on the physical parameters of the product obtained by the above measurement and calculation, a non-linear partial least squares (NLPLS) method is used to establish a soft sensor model of the purity of the phosphorus recovery product. The specific method is:

Ⅰ. Take the measured and calculated sample topography parameters as a data-driven sample set, expressed as {x(i), y(i)}, where x(i) represents the i-th group of input data, which is the The parameter of purity is the parameter obtained in (3) above. y(i) represents the output data of the i-th group, which is the purity of the product. The input data is formed into a matrix X, and the output data is formed into an output matrix Y;

Ⅱ. Build NLPLS model based on input and output data. Proceed as follows:

Normalize the matrices X and Y so that they have a mean of 0 and a variance of 1. Then the input matrix is expanded by columns, and the expanded items are the hidden node output matrix G of the radial basis function (RBF) neural network and the column vector 1 whose elements are all 1, where each row of G corresponds to a hidden node under the action of the input vector. The output of , the bias term coefficient of the hidden node is 1; the partial least squares (PLS) recovery is performed on the following augmented input matrix and output matrix:

{[1 X G], Y}, the resulting NLPLS model is expressed as:

where XE is the augmented input matrix, A and H are the weight coefficient matrix corresponding to the original input vector and the corresponding output vector of the hidden node of the RBF network, b is the input bias vector, and T is the transpose.

The unknown parameters in the NLPLS model are the hidden node center vector c, the corresponding width vector σ, the weight coefficient matrices A and H, and the model bias vector b. These parameters are determined as follows:

①Cluster the input data with the k-means clustering algorithm to obtain the hidden node center c; the algorithm can determine the optimal number of clustering centers, and at the same time, the clustering centers can be reasonably distributed in the data space;

②Use the p-nearest neighbor rule to calculate the hidden node width:

where N is the number of hidden node centers, and ci is the p hidden node centers closest to the jth hidden node center.

Calculate the hidden node output matrix G according to the obtained hidden node center and width, and then expand the input matrix to obtain the augmented input matrix [1 X G]. Perform PLS regression on the data {[1 X G], Y} to obtain the PLS model parameter matrix {T,W,P,B,Q}. In order to retain all the information in the subsequent model update, the number of extracted feature variables a is determined by the cross-check method, and the obtained parameter matrix is recorded as {Ta, Wa, Pa, Ba, Qa}, and the PLS regression coefficients are calculated from them. matrix β, resulting in A, H and b.

Step 2: For a new batch, use the established NLPLS model to predict and analyze the product purity: transfer the product shape parameters of the new batch to the NLPLS model to estimate the product purity of the batch.

Step 3: When the new batch ends, use the new data combined with the original NLPLS model, and use the recursive algorithm to update the model, so that the model can continuously adopt new information and adapt to changes in the process. Specific steps are as follows:

a. Record the input and output data of the newly obtained batch as X1 and Y1 respectively (it can be the data of one batch or the data of multiple batches), and it does not contain abnormal points. First, perform data preprocessing on the new data using the same method as in step 1.

b. Determine whether to add new hidden nodes:

If the distance between the new data X1 and the existing hidden node center of the RBF network is greater than the set value, a new hidden node is added; the new hidden node center matrix is C _gnew , and the corresponding width vector is σ _gnew . The hidden node center matrix C _g , the corresponding width vector σ _g and the load matrix P are extended as follows:

If the distance between X1 and the center of the hidden nodes of the existing RBF network is less than or equal to the set value, there is no need to add hidden nodes, and C _g , σ _g and P remain unchanged;

c. Extend X1 to XE1=[1 X1 G1], where G1 is the output matrix of the hidden node for X1, let

Perform PLS regression on the data pair {X _E , Y} to get a new NLPLS model:

Then calculate the weight coefficient matrix A, H and the bias vector b according to the method of step 3 in step 1;

d. After getting the new model, go back to step 2 and apply it to the newly obtained batch of data.

The Background of the Invention section may contain background information about the problem or environment of the invention and is not necessarily a description of the prior art. Therefore, what is contained in the Background section is not an admission of prior art by the applicant.

The above content is a further detailed description of the present invention in conjunction with specific/preferred embodiments, and it cannot be considered that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art to which the present invention pertains, without departing from the concept of the present invention, they can also make several substitutions or modifications to the described embodiments, and these substitutions or modifications should be regarded as It belongs to the protection scope of the present invention. In the description of this specification, reference to the terms "one embodiment," "some embodiments," "preferred embodiment," "example," "specific example," or "some examples" or the like is meant to be used in conjunction with the description. A particular feature, structure, material, or characteristic described by an example or example is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms are not necessarily directed to the same embodiment or example. Furthermore, the particular features, structures, materials or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Those skilled in the art may combine and combine the different embodiments or examples described in this specification, as well as the features of the different embodiments or examples, without conflicting each other. Although the embodiments of the present invention and their advantages have been described in detail, it should be understood that various changes, substitutions and alterations can be made herein without departing from the scope of the patent application.

Claims

A soft measurement method of phosphorus recovery product purity, is characterized in that, comprises the steps:

(1) Use an optical microscope to photograph the collected phosphorus recovery products, and record their microscopic topography on a two-dimensional plane;

(2) processing the pictures taken in step (1) so as to measure the physical parameters of the particles in the pictures;

(3) measure and calculate the following physical parameters of the particulate matter in the picture processed by step (2): the distance nF max of the two farthest points around the particle; the distance nF min of the two nearest points around the particle; elongation n E11 ; aspect ratio n El2 ; particle total area n A ; particle perimeter n P ; particle perimeter n Ci ; equivalent diameter n Ed ; solidity ns ;

(4) Based on the physical parameters of the product measured and calculated in step (3), a non-linear partial least squares NLPLS method is used to establish a soft sensor model for the purity of the phosphorus recovery product; for new phosphorus recovery products, the established NLPLS method is used. The model predicts and analyzes the product purity, and uses the new predictive analysis data combined with the original NLPLS model to update the model with a recursive algorithm.
The soft measurement method for the purity of phosphorus recovery products according to claim 1, wherein the processing in step (2) includes enhancing color difference and contrast, and performing stroke processing on the two-dimensional plane shape of the phosphorus recovery products in the picture.
The soft-sensor method of phosphorus recovery product purity as claimed in claim 1, is characterized in that, in step (4), the soft-sensor model of establishing phosphorus recovery product purity specifically comprises the steps:

Ⅰ. Take the measured and calculated sample topography parameters as a data-driven sample set, expressed as {x(i), y(i)}, where x(i) represents the i-th group of input data, that is, step (3) The parameters obtained in , y(i) represents the output data of the ith group, which is the purity of the product, and the input data is formed into a matrix X, and the output data is formed into an output matrix Y;

Ⅱ. Build an NLPLS model based on input and output data, including:

Normalize the matrices X and Y so that the mean value is 0 and the variance is 1, and then the input matrix is column-expanded, and the expansion term is the hidden node output matrix G of the radial basis function (RBF) neural network and the elements are full. is a column vector 1 of 1, where each row of G corresponds to the output of a hidden node under the action of an input vector, and the coefficient of the bias term of the hidden node is 1; Partial least squares (PLS) is performed on the following augmented input matrix and output matrix )Recycle:

{[1 X G], Y}, the resulting NLPLS model is expressed as:

where XE is the augmented input matrix, A and H are the weight coefficient matrix corresponding to the original input vector and the corresponding output vector of the hidden node of the RBF network, b is the input bias vector, T is the transpose, and c is the hidden node center vector , σ is the corresponding width vector, A and H are the weight coefficient matrix.
The soft measurement method for the purity of phosphorus recovery products as claimed in claim 3, wherein in step II, the hidden node center vector c, the corresponding width vector σ, the weight coefficient matrices A and H, and the input bias vector b are as follows Steps to determine:

①Use the k-means clustering algorithm to cluster the input data to obtain the hidden node center c;

②Use the p-nearest neighbor rule to calculate the hidden node width:

where N is the number of hidden node centers, and ci is the p hidden node centers closest to the jth hidden node center;

③ Use PLS regression to determine the weight coefficient matrix A, H and bias vector b:

Calculate the hidden node output matrix G according to the obtained hidden node center and width, and then expand the input matrix to obtain the augmented input matrix [1 X G], perform PLS regression on the data {[1 X G], Y}, and obtain PLS The model parameter matrix {T,W,P,B,Q}, the number of extracted feature variables a is determined by the cross-check method, the obtained parameter matrix is recorded as {Ta,Wa,Pa,Ba,Qa}, and the PLS is further calculated. Regression coefficient matrix β, resulting in A, H and b.
The soft measurement method of phosphorus recovery product purity as claimed in claim 3 or 4, is characterized in that, in step (4), utilizes new pre-estimation analysis data in conjunction with original NLPLS model, adopts recursive algorithm to update the model, Specifically include the following steps:

a. Record the input and output data of the new phosphorus recovery product batch as X1 and Y1 respectively, and do not contain abnormal points;

b. Determine whether to add a new hidden node. If the distance between the new data X1 and the existing RBF network hidden node center is greater than the set value, add a new hidden node; if the distance between X1 and the existing RBF network hidden node center If it is less than or equal to the set value, there is no need to add hidden nodes;

c. Expand X1, perform PLS regression, and obtain the updated NLPLS model; then calculate the weight coefficient matrix A, H and the bias vector b according to the method of step ③.
The soft measurement method of phosphorus recovery product purity as claimed in claim 5, characterized in that, in step b, if the distance between the new data X1 and the existing RBF network hidden node center is greater than a set value, record the new hidden node center The matrix is C gnew , and the corresponding width vector is σ gnew . The original hidden node center matrix C g , the corresponding width vector σ g and the load matrix P are expanded as follows:

If the distance between X1 and the existing RBF network hidden node center is less than or equal to the set value, C g , σ g and P remain unchanged.
The soft measurement method of phosphorus recovery product purity as claimed in claim 6, is characterized in that,

Extend X1 to XE1=[1 X1 G1], where G1 is the output matrix of the hidden node for X1, let
Perform PLS regression on the data pair {X E , Y} to get the updated NLPLS model.
A computer-readable storage medium storing a computer program, characterized in that, when the computer program is executed by a processor, the steps of the soft-measurement method for the purity of the phosphorus recovery product according to any one of claims 1 to 7 are realized. (2) to step (4), specifically including:

(2) processing the picture to measure the physical parameters of the particulate matter in the picture; wherein, the picture is obtained by using an optical microscope to photograph the collected phosphorus recovery product, and the picture records the phosphorus recovery product in a two-dimensional plane on the microstructure of the topography;

(3) measure and calculate the following physical parameters of the particulate matter in the picture processed by step (2): the distance nF max of the two farthest points around the particle; the distance nF min of the two nearest points around the particle; elongation n E11 ; aspect ratio n El2 ; particle total area n A ; particle perimeter n P ; particle perimeter n Ci ; equivalent diameter n Ed ; solidity ns ;

(4) Based on the physical parameters of the product measured and calculated in step (3), a non-linear partial least squares NLPLS method is used to establish a soft sensor model for the purity of the phosphorus recovery product; for new phosphorus recovery products, the established NLPLS method is used. The model predicts and analyzes the product purity, and uses the new predictive analysis data combined with the original NLPLS model to update the model with a recursive algorithm.