Quantitative detection method and system for geometric characteristics of dicumyl peroxide particles

Abstract

The present invention relates to the field of image recognition and detection, and particularly to a method and system for quantitatively detecting the geometric characteristics of dicumyl peroxide particles. The method includes: collecting pictures of dicumyl peroxide particle samples and standardizing the sample pictures; identifying and correcting the particle regions in the sample pictures; obtaining the measurement parameters of the particles in the sample pictures; obtaining particle area distribution data based on the measurement parameters; obtaining particle aggregate data by principal component analysis based on the measurement parameters; and obtaining the quantitative detection results of the particles based on the particle area distribution data and the particle aggregate data. The present invention improves the detection accuracy and objectivity.

Description

Technical Field

[0001] The present invention relates to the field of image recognition and detection, and particularly to a method and system for quantitatively detecting the geometric characteristics of dicumyl peroxide particles. Background Art

[0002] Dicumyl peroxide (DCP) is a commonly used symmetric di-tert-alkyl peroxide, usually in the form of transparent rhombic particles. DCP is an important additive in polymer materials, serving as a vulcanizing agent in natural rubber and synthetic rubber, and as an initiator or crosslinking agent in polymerization reactions.

[0003] In the industrial production of DCP, crystallization is often used to form DCP into granular solids, and particles with a certain mesh number or above are packaged and sold as products. After crystallization, the DCP particles have problems such as non-uniform particle size distribution, internal encapsulation, and coalescence growth of multiple particles. When the particle distribution width is large, it is easy to cause caking of the packaged DCP, and it is difficult to uniformly disperse it in rubber during application. In actual operation, it is necessary to extend the mixing time and increase the mixing temperature to uniformly disperse the caked DCP in rubber, resulting in energy loss. The presence of internal encapsulation and aggregates in the particles will reduce the purity of DCP and affect the total volatility of the product. In addition, it also has a great negative impact on the appearance of the product.

[0004] Currently, the common methods for measuring particle size distribution are manual screening with a sieve and automatic calculation with a particle size analyzer. The method of manually screening with a sieve is cumbersome and inaccurate, while the particle size analyzer has high requirements for equipment, and the equipment is expensive and the measurement cost is high. For the problems of encapsulation and aggregates in particles, subjective observation is currently relied on, lacking objectivity. Summary of the Invention

[0005] The purpose of the present invention is to provide a method and system for quantitatively detecting the geometric characteristics of dicumyl peroxide particles, improving the detection accuracy and objectivity.

[0006] To solve the above technical problems, the technical solution of the present invention is as follows:

[0007] In the first aspect, a method for quantitatively detecting the geometric characteristics of dicumyl peroxide particles is provided, characterized in that: it includes S100: collecting a sample image of dicumyl peroxide particles and standardizing the sample image;

[0008] S200: identifying and correcting the particle area in the sample image;

[0009] S300: obtaining the measurement parameters of the particles in the sample image;

[0010] S400: obtaining particle area distribution data based on the measurement parameters;

[0011] S500: Obtain the particle aggregate data through the principal component analysis method based on the measurement parameters;

[0012] S600: Obtain the quantitative detection results of the particles based on the particle area distribution data and the particle aggregate data.

[0013] Furthermore, the sample picture standardization method in step S100 is: After placing a scale near the sample, collect the scale and the sample together into a sample picture; calibrate the scale in the picture and perform the conversion between pixel points and corresponding units.

[0014] Furthermore, the specific method of step S200 is: Identify the high-light part of the particle edge multiple times until all particle pictures in the figure are covered by the mask; zoom in on the picture to correct the irregular areas or the adhered areas of the particle edge.

[0015] Furthermore, before step S200, the following steps are also included: Select the effective working area where the particles are located.

[0016] Furthermore, in S300, the measurement parameters of the particles include: area, aspect ratio, granularity, average optical density, optical density deviation, and edge value.

[0017] Furthermore, the specific method of step S400 is: Calculate the mean value, standard deviation, kurtosis, and skewness according to the area of the particles, and obtain the particle size distribution histogram to obtain the particle area distribution data.

[0018] Furthermore, step S500 includes:

[0019] S510: Standardize the five variables of the aspect ratio, granularity, average optical density, optical density deviation, and edge value of the particles to obtain the standard values;

[0020] S520: Reduce the dimension of the five standardized variables;

[0021] S530: Perform factor analysis based on the variables after dimension reduction; If the KMO value and the significance meet the preset conditions, execute S540; Otherwise, the data is unavailable, and collect the sample pictures again;

[0022] S540: Extract the principal component factors according to the preset cumulative contribution rate;

[0023] S550: Convert the initial factor loading matrix into the principal component factor loading matrix;

[0024] S560: Multiply the principal component factor loading matrix by the standard values to obtain the respective principal component values; Use the ratio of the eigenvalue corresponding to each principal component to the sum of the total eigenvalues of the extracted principal components as the weight, and calculate the comprehensive principal component value based on the respective principal component values;

[0025] S570: Take the particles with the comprehensive principal component value less than the preset score threshold as aggregates, so as to obtain the proportion of the number of aggregate particles to the total sampling amount, and obtain the aggregate proportion as the particle aggregate data.

[0026] Further, step S600 includes: preset a discrimination threshold, compare the particle area distribution data and the particle aggregate data with the preset discrimination threshold, and obtain a detection result.

[0027] In a second aspect, a quantitative detection system for the geometric characteristics of diisopropylbenzene peroxide particles is provided, including:

[0028] A memory, a processor, and a computer program stored on the memory and executable on the processor;

[0029] Wherein, when the processor executes the program, it implements the above-mentioned quantitative detection method for the geometric characteristics of diisopropylbenzene peroxide particles.

[0030] In a third aspect, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, it implements the above-mentioned quantitative detection method for the geometric characteristics of diisopropylbenzene peroxide particles.

[0031] The present invention has the following beneficial effects: The present invention collects sample pictures of diisopropylbenzene peroxide particles, standardizes the sample pictures, and converts the actual scales in different images into standard uniform sizes. Then, single particles are identified to extract measurement parameters; particle area distribution data is obtained according to the measurement parameters, and particle aggregate data is obtained by principal component analysis according to the measurement parameters; finally, a quantitative detection result of the geometric characteristics of diisopropylbenzene peroxide particles is obtained. In the determination of the particle size distribution of DCP particles in the present invention, automatic detection and identification are carried out, which is easy to operate, the process is stable, and the data is accurate; in the problems of inclusion and agglomerates in DCP particles, automatic identification is carried out, subjective observation is avoided, objectivity is strong, and the detection accuracy is improved. The whole operation process of the present invention is simple, no additional equipment is required, and the detection accuracy is high. Description of the Drawings

[0032] Figure 1 It is the overall flowchart of the detection method of the present invention;

[0033] Figure 2 It is the flowchart of obtaining particle area distribution data in the detection method of the present invention;

[0034] Figure 3 It is the flowchart of obtaining particle aggregate data in the detection method of the present invention;

[0035] Figure 4 It is a schematic diagram of calibrating the scale in step S100 of Example 1;

[0036] Figure 5 Schematic diagram of the particle mask in step S200 of Example 1;

[0037] Figure 6 Schematic diagram of the identification of a single particle region in step S200 of Example 1;

[0038] Figure 7 Schematic diagram of obtaining measurement parameters in step S300 of Example 1;

[0039] Figure 8 Data graph of the area analysis result in step S400 of Example 1;

[0040] Figure 9 Particle size distribution histogram in step S400 of Example 1;

[0041] Figure 10 Schematic diagram of data standardization in step S510 of Example 1;

[0042] Figure 11 Schematic diagram of the cumulative contribution rate in step S540 of Example 1;

[0043] Figure 12 Schematic diagram of the conversion of the initial factor loading matrix into the principal component loading matrix in step S550 of Example 1;

[0044] Figure 13 Schematic diagram of the calculation of each principal component value in step S560 of Example 1;

[0045] Figure 14 Schematic diagram of the calculation of the comprehensive principal component value in step S560 of Example 1;

[0046] Figure 15 Data graph of the area analysis result in step S400 of Example 2;

[0047] Figure 16 Particle size distribution histogram of batch 1010 in step S400 of Example 2;

[0048] Figure 17 Particle size distribution histogram of batch 0929 in step S400 of Example 2;

[0049] Figure 18 Principal component analysis graph of batches 1010 and 0929 in step S500 of Example 2. Detailed implementation manners

[0050] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0051] Example 1:

[0052] The DCP particles produced by single - batch crystallization are detected.

[0053] Please refer to Figure 1 and Figure 3 , the present invention is a method for quantitatively detecting the geometric characteristics of dicumyl peroxide particles, which includes:

[0054] S100: Collect sample pictures of dicumyl peroxide particles and standardize the sample pictures;

[0055] S200: Identify and correct the particle regions in the sample pictures;

[0056] S300: Obtain the measurement parameters of the particles in the sample pictures;

[0057] S400: Obtain particle area distribution data based on the measurement parameters;

[0058] S500: Obtain particle aggregate data based on the measurement parameters by principal component analysis;

[0059] S600: Obtain the quantitative detection results of the particles based on the particle area distribution data and the particle aggregate data.

[0060] The present invention collects pictures, standardizes the sample pictures, and converts the actual scales in different images into standard uniform sizes. Then, single particles are identified to extract measurement parameters; particle area distribution data is obtained based on the measurement parameters, and particle aggregate data is obtained by principal component analysis based on the measurement parameters; finally, the quantitative detection results of the geometric characteristics of dicumyl peroxide particles are obtained. In the determination of the particle size distribution of DCP particles in the present invention, automatic detection and identification are carried out, which is easy to operate, has a stable process, and accurate data; in the problems of inclusion and coalescence in DCP particles, automatic identification is carried out, subjective observation is avoided, objectivity is strong, and the detection accuracy is improved.

[0061] The following will specifically describe Figure 1 each step in.

[0062] In step S100, the method for collecting sample pictures of dicumyl peroxide particles and standardizing the sample pictures is: after placing a scale near the sample, the scale and the sample are collected together into a sample picture; the scale in the picture is calibrated to perform the conversion between pixel points and corresponding units.

[0063] Specifically, randomly scatter 200 - 400 DCP particles so that there is no adhesion between each particle and there is a spacing of approximately 2 mm to ensure that no adhesion is formed between the particles during automatic image recognition and prevent misjudgment of the area size and other parameters. Place a scale near the sample particles, take a picture, and collect the DCP particles and the scale together to obtain the sample picture. Due to problems with the shooting angle or focal length, the same scale presents different views in the image. Therefore, it is necessary to convert the actual scale in different images into a standard uniform virtual size; in this embodiment, import the sample picture into the Image-Pro Plus software. First, calibrate the picture scale. In Measure>calibration>spatial, click new to create a new standard, and correspond the scale provided by the software with the scale in the picture. For example Figure 4 , after selecting the corresponding unit length, the software will automatically perform the conversion between pixel points and the corresponding unit.

[0064] In step S200, the method for obtaining the particle measurement parameters in the sample picture is as follows: repeatedly identify the highlight part of the particle edge until all the particle pictures in the figure are covered by the mask; magnify the picture and correct the irregular or adhered areas of the particle edge.

[0065] Specifically, in Process>segmentation, use the pipette tool to repeatedly identify the highlight part of the particle edge until all the particle pictures in the figure are covered by the mask, and then close the segmentation tab. Refer to Figure 5 . Open Measure>count / size, click count, and the software will start to automatically identify the area; magnify the picture, select a suitable tool in Measure>count / size>edit to correct the irregular or adhered areas of the particle edge, and ensure that there is no adhesion in the identified particle area in the particle group to complete. At the same time, the particle count is also marked on each particle area, such as Figure 6 .

[0066] Furthermore, before step S200, the following steps are also included: use the Rectangular AOI tool to select the effective working area where the particles are located to reduce the interference of irrelevant areas in the picture. Refer to the square around the particles in Figure 6 . The area inside the square is the effective working area.

[0067] In step S300, obtain the measurement parameters of the particles in the sample picture.

[0068] Specifically, the measurement parameters of the particles include: Area, Aspect, Clumpiness, Density (mean), Density (std.dev.), and Margination. Select the above parameters in Measure>count / size>measure>select measurements, such as Figure 7 . Among them,

[0069] Aspect ratio Ap: The ratio of the longest axis to the shortest axis inside the selected area. When the particles appear as aggregates, the aspect ratio is usually smaller than that of rhombic single crystals.

[0070] Clumpiness Cp: The pixel ratio of the pixels in the selected area that deviate from the average value after magnification, used to reflect the change of texture. The smaller the Cp, the smaller the internal change.

[0071] Mean density Dm: The average brightness density inside the object. The smaller the Dm, the lower the average brightness inside the object.

[0072] Density standard deviation Dsd: The standard deviation of the brightness inside the object. The smaller the Dsd, the more uniform the brightness distribution inside the object.

[0073] Margination Mg: The relative distribution of the brightness between the center and the edge of the object. The larger the Mg value, the higher the center brightness.

[0074] In step S400, based on the measurement parameters, obtain the particle area distribution data. Refer to Figure 2 , the method of S400 is: calculate the mean value, standard deviation, kurtosis, and skewness according to the area of the particles, and obtain the particle size distribution histogram to obtain the particle area distribution data.

[0075] Specifically, the data obtained in step S300 is imported into SPSS Statistics; first, process the area parameter and detect the particle size distribution, calculate the mean value, standard deviation, kurtosis, and skewness, and compare with the drawn histogram.

[0076] The formula for calculating the mean value A is:

[0077] where x is the sample area and n is the number of samples.

[0078] The formula for calculating the standard deviation S is:

[0079] where is the sample mean value.

[0080] The formula for calculating the skewness S k is:

[0081] where s is the standard deviation of the sample.

[0082] The formula for calculating the kurtosis coefficient K is:

[0083]

[0084] Refer to Figure 8 and Figure 9 , it can be seen that a total of 384 particles are selected, and the average particle size without special process parameter adjustment is 2 - 3 mm 2 , and the average area of this batch of particles is 1.58 mm 2 , which are relatively small particles in daily experiments and meet the expected results of process adjustment during the crystallization process. Generally speaking, a standard deviation value within 0.5 indicates that the particle size variation is very small, and a value above 1.5 indicates that the particle size variation is very large. The measured standard deviation of this batch of particles is 0.53, indicating that the particle size distribution variation is relatively small. And considering the kurtosis value K of 2.28, K > 0, which is a leptokurtic distribution, indicating that the data distribution is more concentrated. It can also be directly seen from the histogram Figure 9 that most particles are in the range of 1 - 2 mm 2 . In addition, it can be seen that the skewness S k is 1.36. When S k > 0, it is called a positively skewed distribution. A positively skewed distribution means that the mode appears to the left of the median, that is Figure 9 the situation where the highest peak in

[0085] In step S500, based on the measurement parameters, particle aggregate data is obtained through principal component analysis to detect the proportion of aggregates in this batch of DCP products. Refer to Figure 3 , the S500 method includes:

[0086] S510: Standardize the data of the five variables of the aspect ratio, block size, mean optical density, optical density deviation, and edge value of the particles to obtain standard values; specifically include: Export the data obtained in step S300 to SPSS Statistics to analyze the relevant data of aggregate particles; in Analyze > Descriptive > Descriptive Statistics, check Save standardized scores as variables to standardize the data. Refer to Figure 10 ;

[0087] Suppose there are m evaluation objects for principal component analysis, X1, X2, …, X m , and there are a total of 5 variables. The sample standardization formula is: where i = 1, 2, …, 5; j = 1, 2, …, m; X ij is the j - th index value of the i - th variable, is the sample mean of the i-th variable, S i is the sample standard deviation of the i-th variable;

[0088] S520: Perform dimensionality reduction on the five standardized variables;

[0089] S530: Perform factor analysis based on the variables after dimensionality reduction; if the KMO value and significance meet the preset conditions, execute S540; otherwise, the data is unavailable and sample images need to be recollected; in this embodiment, the specific method is:

[0090] After checking whether the Kaiser-Meyer-Olkin measure of sampling adequacy, i.e., the KMO value, is greater than 0.6 and the significance is less than 0.05 to determine whether the sample data is suitable for principal component analysis, and then ensuring that the cumulative contribution rate of the principal components ≥ 60% can meet the analysis requirements;

[0091] S540: Extract the principal component factors according to the preset cumulative contribution rate;

[0092] Refer to Figure 11 , in this embodiment, the calculated KMO value is 0.804 and the significance is 0, indicating that the sample data is suitable for principal component analysis. In addition, the cumulative contribution rate after extracting three factors is 85.15%, indicating that the explanatory rate of these three factors for the overall is 85.15%, so the first 3 factors can be extracted.

[0093] S550: Convert the initial factor loading matrix into a principal component factor loading matrix;

[0094] Refer to Figure 12 , in the newly created dataset, convert the loading matrix A of the initial factors into the principal component loading matrix U, and the calculation formula is: where λ k is the eigenvalue of each factor, k = 1, 2,..., t; in this embodiment, the number of factors is 3, so k = 1, 2, 3.

[0095] S560: Multiply the principal component factor loading matrix by the standard values to obtain the values of each principal component; use the proportion of the eigenvalue corresponding to each principal component in the sum of the total eigenvalues of the extracted principal components as the weight, and calculate the comprehensive principal component value based on the values of each principal component;

[0096] Specifically, multiply U k by the standard value Z Xij to obtain the values of t principal components Y k , and the calculation formula is: where k = 1, 2,..., t; j = 1, 2,..., m;

[0097] Take the ratio of the eigenvalue corresponding to each principal component to the sum of the eigenvalues of all the extracted principal components as the weight, and calculate the principal component comprehensive model; according to the principal component comprehensive model, the comprehensive principal component value Y can be calculated. j , and the calculation formula is: where j = 1, 2, …, m.

[0098] In this embodiment, referring to Figure 13 and 14 , in the transformation > calculation variable, multiply U k by the standard value Z Xij to obtain the values of 3 principal components Y k . The calculation formula is: where k = 1, 2, 3; j = 1, 2, …, 384;

[0099] Finally, take the ratio of the eigenvalue corresponding to each principal component to the sum of the eigenvalues of all the extracted principal components as the weight, and calculate the principal component comprehensive model; calculate the comprehensive principal component value Y according to the principal component comprehensive model, and the calculation formula is: where j = 1, 2, …, 384.

[0100] S570: Take the particles with the comprehensive principal component value less than the preset score threshold as aggregates, so as to obtain the proportion of the number of aggregate particles to the total sampling amount, and get the aggregate proportion, which is used as the particle aggregate data. For the comprehensive principal component value Y j particles less than -0.6 are classified as aggregate particles, and the proportion of aggregate particles in this batch can be estimated. In this embodiment, there are 132 particles less than -0.6 in the comprehensive principal component, that is, the number of aggregate particles in this batch of DCP product particles, accounting for 34.4% of the total sampling amount; therefore, the aggregate proportion is 34.4%.

[0101] In step S600, preset a discrimination threshold, and compare the particle area distribution data and the particle aggregate data with the preset discrimination threshold to obtain the detection result. The preset discrimination threshold includes the particle area distribution threshold and the aggregate proportion threshold. The staff can set the discrimination thresholds of the mean value, standard deviation, kurtosis and skewness as the particle area distribution threshold. When the detected parameters do not meet the discrimination threshold, the geometric characteristics of the particles are unqualified.

[0102] Comparative Example 1:

[0103] The DCP products in Example 1 were screened using a sieve according to the existing conventional method, and a subjective judgment of agglomerates was made on 384 particles in the photo of Example 1. The sieve mesh sizes selected were 10 mesh, 12 mesh, 14 mesh, 16 mesh, and 18 mesh, corresponding to 1700 μm, 1400 μm, 1180 μm, 1000 μm, and 880 μm respectively. After weighing 100 g of DCP particles, they were sieved, and the particle distribution was obtained as follows: <880 μm: 0.5 g, 880 - 1000 μm: 3.2 g, 1000 - 1180 μm: 13.1 g, 1180 - 1400 μm: 21.4 g, 1400 - 1700 μm: 32.1 g, >1700 μm: 25.2 g. According to the particle distribution obtained by sieving, a relationship line between the cumulative weight percentage on the sieve and the sieve pore size was plotted, and the coefficient of variation (C.V. value) was calculated. Where PD is the sieve pore size value, PD 84％ means that in the sieved crystal sample, the total mass of particles with a particle size smaller than PD 84％ accounts for 84%. PD 16％ and PD 50％ The meaning is deduced by analogy. The C.V. of this batch of particles was calculated to be 22.41%.

[0104] The subjective judgment of the DCP particles in the photo showed a total of 125 agglomerates, accounting for 32.6% of the total sampling amount, which is close to 34.4% in Example 1, indicating that the proportion of agglomerates detected using the present invention has a certain objective credibility.

[0105] Example 2

[0106] Two batches of DCP particles produced by different crystallization processes were detected.

[0107] Two batches of DCP particles produced by different crystallization processes were taken respectively, and 200 - 400 particles were randomly and evenly scattered, photographed and collected, and then imported into the Image - Pro Plus software for analysis according to the steps in Example 1. The area distribution analysis data is referred to Figures 15 - 17 .

[0108] The average area of DCP particles in batch 1010 was 1.71 mm 2 , the standard deviation was 0.67, and the minimum and maximum values were 0.792 mm 2 and 4.541 mm 2 respectively; the average area of DCP particles in batch 0929 was 2.45 mm 2 , the standard deviation was 1.66, and the minimum and maximum values were 0.612 mm 2 and 12.175 mm 2It shows that the particle size distribution of batch 0929 is wider than that of batch 1010, and the particle homogeneity is worse. From the skewness, both batches belong to the positive skewness distribution, and the mode is less than the mean value. The same conclusion can also be obtained by comparing the histograms.

[0109] The variables obtained for the two batches of particles were subjected to principal component analysis in SPSS Statistics software according to the steps in Example 1, and the comprehensive principal component value Y was obtained. The number of particles with Y less than -0.6 in batch 1010 was 105, accounting for 32.8% of the total; the number of particles with Y less than -0.6 in batch 0929 was 148, accounting for 39.6% of the total. That is, the proportion of aggregates in batch 0929 is more than that of aggregate particles in batch 1010. From Figure 18 the PCA diagram, it can be seen that there are great differences between the two batches of particles, and the particle morphology of batch 1010 is better than that of batch 0929.

[0110] In another embodiment, a quantitative detection system for the geometric characteristics of diisopropylbenzene peroxide particles is provided, including a memory, a processor, and a computer program stored on the memory and executable on the processor; wherein, when the processor executes the program, it implements the above-mentioned quantitative detection method for the geometric characteristics of diisopropylbenzene peroxide particles.

[0111] In another embodiment, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, it implements the above-mentioned quantitative detection method for the geometric characteristics of diisopropylbenzene peroxide particles.

[0112] Parts not involved in the present invention are the same as or implemented by the prior art.

[0113] The above content is a further detailed description of the present invention in combination with specific implementation manners. It cannot be determined that the specific implementation of the present invention is only limited to these descriptions. For those of ordinary skill in the technical field to which the present invention belongs, without departing from the concept of the present invention, several simple deductions or substitutions can be made, which should all be regarded as belonging to the protection scope of the present invention.

Claims

1. A method for quantitatively detecting the geometric characteristics of dicumyl peroxide particles, characterized in that: including S100: Collect pictures of dicumyl peroxide particle samples and standardize the sample pictures; S200: Identify and correct the particle regions in the sample pictures; S300: Obtain the measurement parameters of the particles in the sample pictures; S400: Obtain the particle area distribution data based on the measurement parameters; S500: Obtain the particle agglomerate data by principal component analysis based on the measurement parameters; S600: Obtain the quantitative detection results of the particles based on the particle area distribution data and the particle agglomerate data; In step S300, the measurement parameters of the particles include: area, aspect ratio, block size, mean optical density, optical density deviation, and edge value; Step S500 includes: S510: Standardize the data of the five variables of the aspect ratio, block size, mean optical density, optical density deviation, and edge value of the particles to obtain standard values; S520: Reduce the dimension of the five standardized variables; S530: Perform factor analysis based on the variables after dimension reduction; if the KMO value and significance meet the preset conditions, execute S540; otherwise, the data is unavailable and the sample pictures need to be collected again; S540: Extract the principal component factors according to the preset cumulative contribution rate; S550: Convert the initial factor loading matrix into a principal component factor loading matrix; S560: Multiply the principal component factor loading matrix by the standard values to obtain the respective principal component values; use the ratio of the eigenvalue corresponding to each principal component to the sum of the total eigenvalues of the extracted principal components as the weight, and calculate the comprehensive principal component value based on the respective principal component values; S570: Take the particles with the comprehensive principal component value less than the preset score threshold as agglomerates, so as to obtain the proportion of the number of agglomerate particles to the total sampling amount, and obtain the agglomerate proportion as the particle agglomerate data.

2. The quantitative detection method for the geometric characteristics of cumene hydroperoxide particles according to claim 1, characterized in that: The method for standardizing the sample pictures in step S100 is: After placing a scale near the sample, collect the scale and the sample together into a sample picture; calibrate the scale in the picture and perform the conversion between pixel points and corresponding units.

3. The quantitative detection method for the geometric characteristics of cumene hydroperoxide particles according to claim 1, characterized in that: The specific method of step S200 is: Identify the high-light part of the particle edge multiple times until all particle pictures in the figure are covered by the mask; zoom in on the picture to correct the irregular or adhered areas of the particle edge.

4. The method for quantitatively detecting the geometric characteristics of cumene hydroperoxide particles according to claim 3, wherein: Before step S200, the following steps are also included: Select the effective working area where the particles are located.

5. The method for quantitatively detecting the geometric characteristics of cumene hydroperoxide particles according to claim 1, wherein: The specific method of step S400 is: Calculate the mean value, standard deviation, kurtosis, and skewness according to the area of the particles, and obtain the particle size distribution histogram to obtain the particle area distribution data.

6. The method for quantitatively detecting the geometric characteristics of cumene hydroperoxide particles according to claim 1, characterized in that: Step S600 includes: Preset a discrimination threshold, and compare the particle area distribution data and the particle agglomerate data with the preset discrimination threshold to obtain the detection result.

7. A quantification detection system for the geometric characteristics of dicumyl peroxide particles, characterized in that, including: A memory, a processor, and a computer program stored on the memory and executable on the processor; Wherein, when the processor executes the program, it implements the method for quantitatively detecting the geometric characteristics of dicumyl peroxide particles as described in any one of claims 1-6.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the method for quantitatively detecting the geometric characteristics of dicumyl peroxide particles as described in any one of claims 1-6.

Images