Intelligent data analysis system and method based on aluminum oxide production

Abstract

The invention relates to the technical field of intelligent analysis of aluminum oxide, in particular to an intelligent data analysis system and method based on aluminum oxide production. Acquiring ball mill parameters, raw material parameters and a spherical alumina impurity threshold, and determining a standard threshold according to the ball mill parameters and the raw material parameters; sampling and detecting the processed spherical aluminum oxide to obtain a purity and morphology image set of the spherical aluminum oxide; constructing a morphology image segmentation model to segment the morphology image set to obtain a first image set and a second image set; constructing a first image recognition model to recognize the first image set to obtain a spherical alumina parameter set, and calculating to obtain a first coefficient; constructing a second image recognition model to recognize the second image set to obtain an impurity parameter set, and calculating to obtain a second coefficient; and calculating the quality coefficient according to the first coefficient, the second coefficient, the spherical alumina impurity threshold and the spherical alumina purity. The comprehensive evaluation of the spherical aluminum oxide is realized through the quality coefficient of the spherical aluminum oxide.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent analysis of alumina, and in particular to an intelligent data analysis system and method based on alumina production. Background Art

[0002] Spherical alumina powder is an important raw material in industry and is widely used in ceramic materials, abrasives, polishing agents, catalyst carriers and fillers. The performance of spherical alumina powder is related to the morphology of its particles, and the particle morphology can greatly affect the performance of spherical alumina powder products.

[0003] Ball milling is one of the methods for producing spherical alumina powder. The spherical alumina powder produced by ball milling usually has uniform particle size distribution and high purity, and is suitable for applications in high-end ceramics, electronic materials, magnetic materials, etc. However, during the ball milling production process, due to the wear of ball milling media or other non-alumina materials, pollution in the air environment, etc., impurities will appear in the spherical alumina powder, affecting the comprehensive performance of spherical alumina.

[0004] Therefore, testing and analyzing the quality of spherical alumina powder is of great significance to its performance.

[0005] Existing technologies for detecting microscopic particles mainly observe through dimensions such as particle size and sphericity, without comprehensively considering the morphology indicators of spherical alumina powder and the morphology of impurities contained, resulting in inaccurate overall evaluation of alumina powder.

[0006] Therefore, an intelligent data analysis system and method based on alumina production is proposed. Summary of the invention

[0007] The purpose of the present invention is to provide an intelligent data analysis system and method based on alumina production, to obtain ball mill parameters, raw material parameters and spherical alumina impurity thresholds, and to determine the standard thresholds through the ball mill parameters and raw material parameters; to sample and test the processed spherical alumina to obtain the purity and morphology image set of the spherical alumina; to construct a morphology image segmentation model to segment the morphology image set to obtain a first image set and a second image set; to construct a first image recognition model to recognize the first image set to obtain a spherical alumina parameter set, and to calculate a first coefficient; to construct a second image recognition model to recognize the second image set to obtain an impurity parameter set, and to calculate a second coefficient; to calculate a quality coefficient based on the first coefficient, the second coefficient and the purity of the spherical alumina; and to achieve a comprehensive evaluation of the microscopic morphology of spherical alumina.

[0008] To achieve the above object, the present invention provides the following technical solutions:

[0009] An intelligent data analysis system based on alumina production, comprising:

[0010] A parameter acquisition module is used to acquire ball mill parameters, raw material parameters and spherical alumina impurity thresholds; the ball mill parameters and raw material parameters are identified according to a threshold prediction model to determine a standard threshold;

[0011] The sample acquisition module processes the alumina raw material by a ball mill to obtain spherical alumina; detects the purity of the spherical alumina, and samples the spherical alumina by random sampling to obtain multiple samples; observes the samples by an electron microscope to obtain a morphology image set;

[0012] An image segmentation module, constructing a morphology image segmentation model to segment the morphology image set to obtain a first image set and a second image set; the first image set is the microscopic morphology of spherical aluminum oxide, and the second image set is the microscopic morphology of impurities;

[0013] A first recognition module constructs a first image recognition model to recognize the first image set to obtain a spherical alumina parameter set; the spherical alumina parameter set includes particle size data, sphericity data and a smoothness coefficient of spherical alumina particles; and a first coefficient is calculated based on the spherical alumina parameter set;

[0014] A second recognition module constructs a second image recognition model to recognize the second image set to obtain an impurity parameter set; the impurity parameter set includes impurity particle size, impurity sphericity and impurity smoothness coefficient; and a second coefficient is calculated based on the impurity parameter set;

[0015] A comprehensive evaluation module is provided, wherein the first coefficient, the second coefficient, the impurity threshold of spherical alumina and the purity of spherical alumina are used to calculate the quality coefficient.

[0016] The ball mill parameters include ball milling time, ball milling speed, ball milling medium and ball milling pressure; the raw material parameters include raw material purity; the standard thresholds include particle size threshold, sphericity threshold and smoothness threshold.

[0017] The threshold prediction model includes a first prediction unit, a second prediction unit and a third prediction unit, and the training process is:

[0018] Obtain historical data on the processing of alumina raw materials through a ball mill;

[0019] Processing the historical data to obtain a first data set, a second data set, and a third data set;

[0020] The first data set includes ball mill parameters, raw material parameters and particle size of spherical alumina particles; the second data set includes ball mill parameters, raw material parameters and sphericity of spherical alumina particles; the third data set includes ball mill parameters, raw material parameters and smoothness coefficient of spherical alumina particles;

[0021] The first prediction unit, the second prediction unit and the third prediction unit are trained respectively by using the first data set, the second data set and the third data set, and a threshold prediction model is obtained by combining them.

[0022] The morphological image segmentation model includes an image input layer, an image preprocessing layer, an image segmentation layer, an image recognition layer, an image classification layer and a result output layer;

[0023] The image input layer is used to input the topographic image set into the model;

[0024] The image preprocessing layer performs image preprocessing on the topographic image set to obtain a preprocessed topographic image;

[0025] The image segmentation layer identifies and segments the particle shape region in the preprocessed topographic image according to the pixel data to obtain a particle image subset;

[0026] The image recognition layer recognizes the particle image subset and determines whether the particle is a spherical aluminum oxide particle based on size characteristics, shape characteristics and texture characteristics;

[0027] The image classification layer divides the particle images of spherical aluminum oxide particles into a first image set, and divides the particle images of non-spherical aluminum oxide particles into a second image set;

[0028] The result output layer is used to output the first image set and the second image set.

[0029] The first image recognition model includes a particle size recognition unit, a sphericity recognition unit and a smoothness coefficient recognition unit;

[0030] The particle size recognition unit is used to recognize the first image set and obtain the particle size data of the particles in the image; the sphericity recognition unit is used to recognize the first image set and obtain the sphericity data of the particles in the image; the smoothness coefficient recognition unit is used to recognize the first image set and obtain the smoothness coefficient of the particles in the image.

[0031] The calculation process of the first coefficient is:

[0032] Acquire a spherical alumina parameter set and a standard threshold; the spherical alumina parameter set includes particle size data, sphericity data and smoothness coefficient of spherical alumina particles; the standard threshold includes particle size threshold, sphericity threshold and smoothness threshold;

[0033] The parameter anomalies in the spherical alumina parameter set are identified by using a standard threshold value to obtain the first coefficient.

[0034] The calculation process of the second coefficient is:

[0035] Obtaining an impurity parameter set and a standard threshold; the impurity parameter set includes impurity particle size, impurity sphericity and impurity smoothness coefficient; the standard threshold includes particle size threshold, sphericity threshold and smoothness threshold;

[0036] The parameter anomalies in the spherical alumina parameter set are identified by using a standard threshold value to obtain the second coefficient.

[0037] The calculation formula of the quality coefficient is:

[0038] A particle index is obtained through the purity of spherical alumina and the first coefficient; an impurity index is obtained through the purity of spherical alumina, the second coefficient and the spherical alumina impurity threshold; and a quality coefficient is obtained through the particle index and the impurity index.

[0039] A method for intelligent data analysis based on alumina production, comprising:

[0040] Obtaining ball mill parameters, raw material parameters and spherical alumina impurity thresholds; identifying the ball mill parameters and raw material parameters according to a threshold prediction model to determine a standard threshold;

[0041] Processing the alumina raw material by a ball mill to obtain spherical alumina; detecting the purity of the spherical alumina, and sampling the spherical alumina by random sampling to obtain a plurality of samples; observing the samples by an electron microscope to obtain a morphology image set;

[0042] Constructing a morphology image segmentation model to segment the morphology image set to obtain a first image set and a second image set; the first image set is the microscopic morphology of spherical aluminum oxide, and the second image set is the microscopic morphology of impurities;

[0043] Constructing a first image recognition model to recognize the first image set, and obtaining a spherical alumina parameter set; the spherical alumina parameter set includes particle size data, sphericity data, and a smoothness coefficient of spherical alumina particles; and calculating a first coefficient according to the spherical alumina parameter set;

[0044] Constructing a second image recognition model to recognize the second image set, and obtaining an impurity parameter set; the impurity parameter set includes an impurity particle size, an impurity sphericity, and an impurity smoothness coefficient; and calculating a second coefficient according to the impurity parameter set;

[0045] The quality coefficient is calculated by using the first coefficient, the second coefficient, the impurity threshold of spherical alumina and the purity of spherical alumina.

[0046] The morphological image segmentation model includes an image input layer, an image preprocessing layer, an image segmentation layer, an image recognition layer, an image classification layer and a result output layer;

[0047] The image input layer is used to input the topographic image set into the model;

[0048] The image preprocessing layer performs image preprocessing on the topographic image set to obtain a preprocessed topographic image;

[0049] The image segmentation layer identifies and segments the particle shape region in the preprocessed topographic image according to the pixel data to obtain a particle image subset;

[0050] The image recognition layer recognizes the particle image subset and determines whether the particle is a spherical aluminum oxide particle based on size characteristics, shape characteristics and texture characteristics;

[0051] The image classification layer divides the particle images determined to be spherical aluminum oxide particles into a first image set, and divides the particle images of non-spherical aluminum oxide particles into a second image set;

[0052] The result output layer is used to output the first image set and the second image set.

[0053] Compared with the prior art, the present invention has the following beneficial effects:

[0054] 1. The present invention collects historical data of spherical alumina processing, combines ball mill parameters and raw material parameters with the particle size, sphericity and smoothness coefficient of spherical alumina particles to obtain three data sets; then trains the BP neural network according to the three data sets respectively to obtain three prediction units; obtains a threshold prediction model according to the combination of the three prediction units; and realizes accurate prediction of the standard threshold through the threshold prediction model.

[0055] 2. The present invention constructs a morphology image segmentation model based on a convolutional neural network model, and uses the morphology image segmentation model to identify and segment the morphology image set to obtain a particle image subset; the particle image subset is identified according to size characteristics, shape characteristics and texture characteristics to determine whether it is a spherical aluminum oxide particle, thereby achieving accurate segmentation and identification of the morphology image set, and accurately obtaining spherical aluminum oxide particle images and impurity particle images.

[0056] 3. The present invention is based on a convolutional neural network model, and training and optimization are performed to obtain a first image recognition model; the first image recognition model includes a particle size recognition unit, a sphericity recognition unit, and a smoothness coefficient recognition unit; the particle size, sphericity, and smoothness coefficient of the particles are accurately identified through the pixel distribution data of the particle area in the first image set.

[0057] 4. The present invention takes the degree of deviation between the particle size data of spherical alumina particles and the particle size threshold as the core indicator; the sphericity of the spherical alumina particles is quantified by the sphericity threshold, and the smoothness coefficient of the spherical alumina particles is quantified by the smoothness threshold, and then the sphericity weight and the smoothness weight are weighted to obtain the auxiliary indicator; finally, the core indicator is weighted by the auxiliary indicator to obtain the first coefficient, so as to accurately evaluate the morphology of the spherical alumina particles.

[0058] 5. The present invention weights the first coefficient according to the purity of the spherical alumina, weights the difference between the impurity threshold of the spherical alumina and the second coefficient according to the impurity content, and then performs an exponential transformation on the difference between the two weighted results to obtain the quality coefficient of the spherical alumina, thereby accurately achieving a comprehensive evaluation of the morphology of the spherical alumina particles and the impurity morphology. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 It is a structural schematic diagram of an intelligent data analysis system based on alumina production according to the present invention;

[0060] Figure 2 It is a structural schematic diagram of the topographic image segmentation model of the present invention;

[0061] Figure 3 The present invention is a flow chart of an intelligent data analysis system method based on alumina production. DETAILED DESCRIPTION

[0062] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0063] During the ball milling production process, impurities may appear in the spherical alumina powder due to the wear of the ball milling medium or other non-alumina materials, pollution in the air environment, etc., which will affect the comprehensive properties of the spherical alumina and reduce its thermal stability and catalytic activity.

[0064] Therefore, an intelligent data analysis system and method based on alumina production is proposed.

[0065] Embodiment 1

[0066] Chemical Plant A produces spherical alumina powder by ball milling. In order to quantify the quality of the produced spherical alumina powder, an intelligent data analysis system based on alumina production according to the present invention is used for detection.

[0067] The structure of the intelligent data analysis system based on alumina production is as follows: Figure 1 As shown, including:

[0068] Parameter acquisition module, sample acquisition module, image segmentation module, first recognition module, second recognition module and comprehensive evaluation module;

[0069] The parameter acquisition module acquires the ball mill parameters, raw material parameters and spherical alumina impurity thresholds; identifies the ball mill parameters and raw material parameters according to the threshold prediction model, and determines the standard thresholds.

[0070] The spherical alumina impurity threshold is determined by historical data; the ball mill parameters include ball milling time, ball milling speed, ball milling medium and ball milling pressure; the raw material parameters include raw material purity; the standard threshold includes particle size threshold, sphericity threshold and smoothness threshold.

[0071] Ball mill is a common powder preparation equipment that crushes and mixes raw powders through friction and collision.

[0072] The present invention collects data on production machines and raw materials in the production process of spherical alumina to obtain ball mill parameters, raw material parameters and spherical alumina impurity thresholds; then predicts standard thresholds based on the ball mill parameters and raw material parameters; and provides a data basis for subsequent quality inspection of spherical alumina.

[0073] The standard threshold is obtained by identifying and predicting the ball mill parameters and raw material parameters through a threshold prediction model;

[0074] The threshold prediction model includes a first prediction unit, a second prediction unit and a third prediction unit;

[0075] The first prediction unit is used to predict the particle size of the spherical alumina particles, and the prediction result is used as the particle size threshold; the second prediction unit is used to predict the sphericity of the spherical alumina particles, and the prediction result is used as the sphericity threshold; the third prediction unit is used to predict the smoothness coefficient of the spherical alumina particles, and the prediction result is used as the smoothness threshold. The value range of the smoothness coefficient and the sphericity is processed to be between 0 and 1.

[0076] The standard threshold values ​​predicted by the threshold prediction model of Chemical Plant A are shown in Table 1; the unit of the particle size threshold value is micrometer.

[0077] Table 1 Chemical Plant A Standard Threshold Data Table

[0078] Standard threshold data Particle size threshold 4.527 Sphericity threshold 0.814 Smooth Threshold 0.838

[0079] The training process of the first prediction unit, the second prediction unit and the third prediction unit is:

[0080] Obtain historical data on the processing of alumina raw materials through a ball mill;

[0081] Processing the historical data to obtain a first data set, a second data set, and a third data set;

[0082] The first data set includes ball mill parameters, raw material parameters and the average particle size of spherical alumina particles; the second data set includes ball mill parameters, raw material parameters and the sphericity of spherical alumina particles; the third data set includes ball mill parameters, raw material parameters and the smoothness coefficient of spherical alumina particles;

[0083] The first prediction unit, the second prediction unit and the third prediction unit are trained respectively by using the first data set, the second data set and the third data set, and a threshold prediction model is obtained by combining them.

[0084] The present invention collects historical data of spherical alumina processing, combines ball mill parameters and raw material parameters with the average particle size, sphericity and smoothness coefficient of spherical alumina particles to obtain three data sets; then trains a BP neural network according to the three data sets respectively to obtain three prediction units; obtains a threshold prediction model according to the combination of the three prediction units; and realizes accurate prediction of a standard threshold through the threshold prediction model.

[0085] The sample acquisition module processes the alumina raw material through a ball mill to obtain spherical alumina; detects the purity of the spherical alumina, and samples the spherical alumina by random sampling to obtain multiple samples; observes the samples through an electron microscope to obtain a morphology image set.

[0086] The alumina raw material is processed according to the set ball mill parameters to obtain spherical alumina.

[0087] Commonly used methods for detecting the purity of spherical alumina include: X-ray fluorescence spectrometry, inductively coupled plasma emission spectrometry and inductively coupled plasma mass spectrometry.

[0088] The image segmentation module constructs a morphology image segmentation model to segment the morphology image set to obtain a first image set and a second image set; the first image set is the microscopic morphology of spherical aluminum oxide, and the second image set is the microscopic morphology of impurities.

[0089] Alumina prepared by ball milling has relatively uniform spherical particles, and the particle size is generally small and the distribution is relatively uniform. However, during the preparation process of the ball milling method, the mixing and grinding of the raw alumina powder and the ball milling medium will cause impurities to appear in the prepared spherical alumina, affecting the comprehensive performance of the spherical alumina.

[0090] There are some differences between the impurity particle size and morphology and the spherical alumina particles, as shown in the following:

[0091] Ultrafine alumina particles usually have a narrow particle size distribution, while impurities may cause abnormal particle size distribution or additional small-sized particles. Ultrafine alumina particles usually present a uniform spherical or near-spherical shape, while impurities may form additional layers or dot-like materials on the particle surface, presenting irregular shapes or uneven distribution, thus distinguishing from alumina particles.

[0092] Therefore, the present invention constructs a morphological image segmentation model based on a convolutional neural network to identify particles.

[0093] The structure of the topographic image segmentation model is as follows: Figure 2 As shown, it includes an image input layer, an image preprocessing layer, an image segmentation layer, an image recognition layer, an image classification layer and a result output layer;

[0094] The image input layer is used to input the topographic image set into the model;

[0095] The image preprocessing layer performs image preprocessing on the topographic image set to obtain a preprocessed topographic image;

[0096] The image segmentation layer identifies and segments the particle shape region in the preprocessed topographic image according to the pixel data to obtain a particle image subset;

[0097] The image recognition layer recognizes the particle image subset to obtain particle size characteristics, particle shape characteristics and particle texture characteristics, and determines whether the particle is a spherical aluminum oxide particle based on the above characteristics;

[0098] The image classification layer divides the particle images determined to be spherical aluminum oxide particles into a first image set, and divides the particle images of non-spherical aluminum oxide particles into a second image set;

[0099] The result output layer is used to output the first image set and the second image set.

[0100] The present invention constructs a morphology image segmentation model based on a convolutional neural network model, and uses the morphology image segmentation model to identify and segment a morphology image set to obtain a particle image subset; the particle image subset is identified according to size characteristics, shape characteristics and texture characteristics to determine whether it is a spherical aluminum oxide particle, thereby achieving accurate segmentation and identification of the morphology image set, and accurately obtaining spherical aluminum oxide particle images and impurity particle images.

[0101] A first recognition module constructs a first image recognition model to recognize the first image set to obtain a spherical alumina parameter set; the spherical alumina parameter set includes particle size data, sphericity data and smoothness coefficient of spherical alumina particles; and a first coefficient is calculated based on the spherical alumina parameter set.

[0102] The present invention is based on a convolutional neural network model, and trains and optimizes to obtain a first image recognition model;

[0103] The first image recognition model includes a particle size recognition unit, a sphericity recognition unit and a smoothness coefficient recognition unit;

[0104] The particle size recognition unit is used to recognize the first image set, obtain the pixel coordinates of the particle area in the image, obtain the circumference of the particle in the image according to the edge pixel coordinates; obtain the diameter of a circle equal to the circumference of the particle, and obtain the projected circumference circle equivalent diameter; and use the projected circumference circle equivalent diameter as the particle size of the particle;

[0105]

[0106] Wherein, par represents the particle size of the spherical alumina particles; L represents the circumference of the particle area; and π represents the circumference.

[0107] The sphericity recognition unit is used to recognize the first image set, obtain the perimeter of the particle area according to the pixel points of the particle edge area in the image; obtain the area of ​​the particle area according to the pixel points of the entire particle area in the image; and calculate the circularity of the particle area according to the perimeter and area of ​​the particle area, and use the circularity as the sphericity of the particle.

[0108]

[0109] Among them, c represents the circularity of the particle area; S represents the area of ​​the particle area; L represents the circumference of the particle area; when c is 1, the particle area is circular; the smaller c is, the more irregular the particle area is, and the greater the difference from the circle.

[0110] The smoothness coefficient recognition unit is used to recognize the first image set, obtain the pixel value of the particle area in the image, and calculate the smoothness coefficient of the particles in the image according to the average value and fluctuation degree of the pixel value; for example, it can be determined by the variance of the pixel and the variance threshold.

[0111] The present invention is based on a convolutional neural network model, and training and optimization are used to obtain a first image recognition model; the first image recognition model includes a particle size recognition unit, a sphericity recognition unit and a smoothness coefficient recognition unit; the particle size, sphericity and smoothness coefficient of the particles are accurately recognized through the pixel distribution data of the particle area in the first image set.

[0112] The calculation process of the first coefficient is:

[0113] Acquire a spherical alumina parameter set and a standard threshold; the spherical alumina parameter set includes particle size data, sphericity data and smoothness coefficient of spherical alumina particles; the standard threshold includes particle size threshold, sphericity threshold and smoothness threshold;

[0114] The parameter anomalies in the spherical alumina parameter set are identified by using a standard threshold value to obtain the first coefficient.

[0115] The calculation formula of the first coefficient is:

[0116]

[0117] Among them, ent1 represents the first coefficient; par v represents the average particle size of spherical alumina particles; PT represents the particle size threshold; sph represents the sphericity of spherical alumina particles; ST represents the sphericity threshold; α1 represents the first sphericity weight; smo represents the smoothness coefficient of spherical alumina particles; MT represents the smoothness threshold; β1 represents the first smoothness weight; par i represents the particle size of the i-th spherical aluminum oxide particle; exp represents an exponential function with the natural constant e as the base; n represents the number of spherical aluminum oxide particles identified in the first image set.

[0118] SPH is obtained by the average sphericity of all identified spherical alumina particles; SMO is obtained by the average smoothness coefficient of all identified spherical alumina particles.

[0119] The first sphericity weight and the first smoothness weight are determined according to historical data of spherical alumina particles.

[0120] The present invention takes the degree of deviation between the particle size data of spherical alumina particles and the particle size threshold as the core indicator; the sphericity of the spherical alumina particles is quantified by the sphericity threshold, the smoothness coefficient of the spherical alumina particles is quantified by the smoothness threshold, and then the auxiliary indicator is weighted by the sphericity weight and the smoothness weight; finally, the core indicator is weighted by the auxiliary indicator to obtain the first coefficient, so as to accurately evaluate the morphology of the spherical alumina particles.

[0121] The second recognition module constructs a second image recognition model to recognize the second image set to obtain an impurity parameter set; the impurity parameter set includes impurity particle size, impurity sphericity and impurity smoothness coefficient; and the second coefficient is calculated based on the impurity parameter set.

[0122] The second image recognition model is obtained by training with collected impurity particle images on the basis of the first image recognition model, and can be used to recognize the second image set to obtain the impurity particle size, impurity sphericity and impurity smoothness coefficient.

[0123] The calculation process of the second coefficient is:

[0124] Obtaining an impurity parameter set and a standard threshold; the impurity parameter set includes impurity particle size, impurity sphericity and impurity smoothness coefficient; the standard threshold includes particle size threshold, sphericity threshold and smoothness threshold;

[0125] The parameter anomalies in the spherical alumina parameter set are identified by using a standard threshold value to obtain the second coefficient.

[0126] The calculation formula of the second coefficient is:

[0127]

[0128] Among them, ent2 represents the second coefficient; zar v represents the average particle size of the impurity particles; PT represents the particle size threshold; zph represents the sphericity of the impurity particles; ST represents the sphericity threshold; α2 represents the second sphericity weight; zmo represents the smoothness coefficient of the impurity particles; MT represents the smoothness threshold; β2 represents the second smoothness weight; zar i represents the particle size of the i-th impurity particle; exp represents an exponential function with the natural constant e as the base; m represents the number of impurity particles identified in the second image set.

[0129] zph is obtained by the average value of the sphericity of all impurity particles identified; zmo is obtained by the average value of the smoothness coefficient of all impurity particles identified.

[0130] The second sphericity weight and the second smoothness weight are determined according to historical data of the impurity particles.

[0131] The present invention takes the degree of deviation between the particle size data of impurity particles and the particle size threshold as the core indicator; quantifies the sphericity of the impurity particles through the sphericity threshold, quantifies the smoothness coefficient of the impurity particles through the smoothness threshold, and then weights the sphericity weight and the smoothness weight to obtain an auxiliary indicator; finally, weights the core indicator through the auxiliary indicator to obtain a second coefficient, so as to accurately evaluate the morphology of the impurity particles.

[0132] The comprehensive evaluation module calculates the quality coefficient through the first coefficient, the second coefficient, the impurity threshold of spherical alumina and the purity of spherical alumina.

[0133] The calculation process of the quality coefficient is:

[0134] A particle index is obtained through the purity of spherical alumina and the first coefficient, and an impurity index is obtained through the purity of spherical alumina, the second coefficient and the spherical alumina impurity threshold; and a quality coefficient is obtained through the particle index and the impurity index.

[0135] The calculation formula of the quality coefficient is:

[0136] QF = exp(posit-negat);

[0137] posit = p * ent1;

[0138] negat=(1-p)*(ET-ent2);

[0139] Among them, QF represents the quality coefficient of spherical alumina; posit represents the particle index; negat represents the impurity index; p represents the purity of spherical alumina; ent1 represents the first coefficient; ET represents the impurity threshold of spherical alumina; ent2 represents the second coefficient.

[0140] The present invention weights the first coefficient according to the purity of spherical alumina, weights the difference between the impurity threshold of spherical alumina and the second coefficient according to the impurity content, and then performs exponential transformation on the difference between the two weighted results to obtain the quality coefficient of spherical alumina, thereby accurately achieving a comprehensive evaluation of the morphology of spherical alumina particles and the impurity morphology.

[0141] By means of the intelligent data analysis system based on alumina production described in the present invention, chemical plant A performs quality grading on the spherical alumina powder it produces, which is beneficial to the production technology upgrade and product pricing of chemical plant A.

[0142] The present invention obtains ball mill parameters, raw material parameters and spherical alumina impurity thresholds during the process of processing alumina raw materials by a ball mill; determines a standard threshold through the ball mill parameters and raw material parameters; then samples and detects the processed spherical alumina to obtain a spherical alumina purity and morphology image set; constructs a morphology image segmentation model to segment the morphology image set to obtain a first image set and a second image set; constructs a first image recognition model to recognize the first image set to obtain a spherical alumina parameter set, and calculates a first coefficient based on the spherical alumina parameters; constructs a second image recognition model to recognize the second image set to obtain an impurity parameter set, and calculates a second coefficient based on the impurity parameter set; calculates a quality coefficient based on the first coefficient, the second coefficient and the purity of the spherical alumina; and realizes a comprehensive evaluation of the microscopic morphology of the spherical alumina.

[0143] Embodiment 2

[0144] Chemical Plant B adopts the present invention to propose a method for intelligent data analysis based on alumina production to identify the quality of spherical alumina. The method flow is as follows: Figure 3 As shown, including:

[0145] Obtaining ball mill parameters, raw material parameters and spherical alumina impurity thresholds; identifying the ball mill parameters and raw material parameters according to a threshold prediction model to determine a standard threshold;

[0146] The standard threshold values ​​predicted by Chemical Plant B through the threshold prediction model are shown in Table 2; the unit of the particle size threshold value is micrometer.

[0147] Table 2 Chemical Plant B Standard Threshold Data Table

[0148] Standard threshold data Particle size threshold 3.243 Sphericity threshold 0.784 Smooth Threshold 0.802

[0149] Processing the alumina raw material by a ball mill to obtain spherical alumina; detecting the purity of the spherical alumina, and sampling the spherical alumina by random sampling to obtain a plurality of samples; observing the samples by an electron microscope to obtain a morphology image set;

[0150] Constructing a morphology image segmentation model to segment the morphology image set to obtain a first image set and a second image set; the first image set is the microscopic morphology of spherical aluminum oxide, and the second image set is the microscopic morphology of impurities;

[0151] Constructing a first image recognition model to recognize the first image set, and obtaining a spherical alumina parameter set; the spherical alumina parameter set includes particle size data, sphericity data, and a smoothness coefficient of spherical alumina particles; and calculating a first coefficient according to the spherical alumina parameter set;

[0152] Constructing a second image recognition model to recognize the second image set, and obtaining an impurity parameter set; the impurity parameter set includes an impurity particle size, an impurity sphericity, and an impurity smoothness coefficient; and calculating a second coefficient according to the impurity parameter set;

[0153] The quality coefficient is calculated by using the first coefficient, the second coefficient, the impurity threshold of spherical alumina and the purity of spherical alumina.

[0154] The calculation formula of the first coefficient is:

[0155]

[0156] Among them, ent1 represents the first coefficient; par v represents the average particle size of spherical alumina particles; PT represents the particle size threshold; sph represents the sphericity of spherical alumina particles; ST represents the sphericity threshold; α1 represents the first sphericity weight; smo represents the smoothness coefficient of spherical alumina particles; MT represents the smoothness threshold; β1 represents the first smoothness weight; par irepresents the particle size of the i-th spherical alumina particle; exp represents an exponential function with the natural constant e as the base; n represents the number of identified spherical alumina particles.

[0157] The calculation formula of the second coefficient is:

[0158]

[0159] Among them, ent2 represents the second coefficient; zar v represents the average particle size of the impurity particles; PT represents the particle size threshold; zph represents the sphericity of the impurity particles; ST represents the sphericity threshold; α2 represents the second sphericity weight; zmo represents the smoothness coefficient of the impurity particles; MT represents the smoothness threshold; β2 represents the second smoothness weight; zar i represents the particle size of the i-th impurity particle; exp represents the exponential function with the natural constant e as the base; m represents the number of impurity particles identified.

[0160] The calculation formula of the quality coefficient is:

[0161] QF = exp(posit-negat);

[0162] posit = p * ent1;

[0163] negat=(1-p)*(ET-ent2);

[0164] Among them, QF represents the quality coefficient of spherical alumina; posit represents the particle index; negat represents the impurity index; p represents the purity of spherical alumina; ent1 represents the first coefficient; Et represents the impurity threshold of spherical alumina; ent2 represents the second coefficient.

[0165] The present invention obtains the ball mill parameters, raw material parameters and spherical alumina impurity thresholds during the processing process, determines the standard thresholds through the ball mill parameters and raw material parameters; samples and detects the processed spherical alumina to obtain the purity and morphology image set of the spherical alumina; constructs a morphology image segmentation model to segment the morphology image set to obtain a first image set and a second image set; constructs a first image recognition model to recognize the first image set to obtain a spherical alumina parameter set, and calculates the first coefficient; constructs a second image recognition model to recognize the second image set to obtain an impurity parameter set, and calculates the second coefficient; and calculates the quality coefficient according to the first coefficient, the second coefficient and the purity of the spherical alumina. The present invention realizes a comprehensive evaluation of the microscopic morphology of spherical alumina through the quality coefficient of spherical alumina.

[0166] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent data analysis system based on alumina production, characterized in that: include: Parameter acquisition module, which acquires ball mill parameters, raw material parameters and spherical alumina impurity thresholds; Identify the ball mill parameters and raw material parameters according to the threshold prediction model to determine the standard threshold; The sample acquisition module processes the alumina raw material through a ball mill to obtain spherical alumina; The purity of the spherical alumina is detected, and the spherical alumina is sampled by random sampling to obtain a plurality of samples; the samples are observed by electron microscope to obtain a morphology image set; An image segmentation module, constructing a morphology image segmentation model to segment the morphology image set to obtain a first image set and a second image set; the first image set is the microscopic morphology of spherical aluminum oxide, and the second image set is the microscopic morphology of impurities; A first recognition module constructs a first image recognition model to recognize the first image set to obtain a spherical alumina parameter set; the spherical alumina parameter set includes particle size data, sphericity data and a smoothness coefficient of spherical alumina particles; and a first coefficient is calculated based on the spherical alumina parameter set; A second recognition module constructs a second image recognition model to recognize the second image set to obtain an impurity parameter set; the impurity parameter set includes impurity particle size, impurity sphericity and impurity smoothness coefficient; and a second coefficient is calculated based on the impurity parameter set; The comprehensive evaluation module calculates the quality coefficient through the first coefficient, the second coefficient, the impurity threshold of spherical alumina and the purity of spherical alumina.

2. The intelligent data analysis system based on alumina production according to claim 1 is characterized in that: The ball mill parameters include ball milling time, ball milling speed, ball milling medium and ball milling pressure; the raw material parameters include raw material purity; the standard thresholds include particle size threshold, sphericity threshold and smoothness threshold.

3. The intelligent data analysis system based on alumina production according to claim 2 is characterized in that: The threshold prediction model includes a first prediction unit, a second prediction unit and a third prediction unit, and the training process is: Obtain historical data on the processing of alumina raw materials through a ball mill; Processing the historical data to obtain a first data set, a second data set, and a third data set; The first data set includes ball mill parameters, raw material parameters and particle size of spherical alumina particles; the second data set includes ball mill parameters, raw material parameters and sphericity of spherical alumina particles; the third data set includes ball mill parameters, raw material parameters and smoothness coefficient of spherical alumina particles; The first prediction unit, the second prediction unit and the third prediction unit are trained respectively by using the first data set, the second data set and the third data set, and a threshold prediction model is obtained by combining them.

4. The intelligent data analysis system based on alumina production according to claim 1 is characterized in that: The morphological image segmentation model includes an image input layer, an image preprocessing layer, an image segmentation layer, an image recognition layer, an image classification layer and a result output layer; The image input layer is used to input the topographic image set into the model; The image preprocessing layer performs image preprocessing on the topographic image set to obtain a preprocessed topographic image; The image segmentation layer identifies and segments the particle shape region in the preprocessed topographic image according to the pixel data to obtain a particle image subset; The image recognition layer recognizes the particle image subset and determines whether the particle is a spherical aluminum oxide particle based on size characteristics, shape characteristics and texture characteristics; The image classification layer divides the particle images of spherical aluminum oxide particles into a first image set, and divides the particle images of non-spherical aluminum oxide particles into a second image set; The result output layer is used to output the first image set and the second image set.

5. The intelligent data analysis system based on alumina production according to claim 1 is characterized in that: The first image recognition model includes a particle size recognition unit, a sphericity recognition unit and a smoothness coefficient recognition unit; The particle size recognition unit is used to recognize the first image set and obtain the particle size data of the particles in the image; the sphericity recognition unit is used to recognize the first image set and obtain the sphericity data of the particles in the image; the smoothness coefficient recognition unit is used to recognize the first image set and obtain the smoothness coefficient of the particles in the image.

6. The intelligent data analysis system based on alumina production according to claim 1 is characterized in that: The calculation process of the first coefficient is: Acquire a spherical alumina parameter set and a standard threshold; the spherical alumina parameter set includes particle size data, sphericity data and smoothness coefficient of spherical alumina particles; the standard threshold includes particle size threshold, sphericity threshold and smoothness threshold; The parameter anomalies in the spherical alumina parameter set are identified by using a standard threshold value to obtain the first coefficient.

7. The intelligent data analysis system based on alumina production according to claim 1 is characterized in that: The calculation process of the second coefficient is: Obtaining an impurity parameter set and a standard threshold; the impurity parameter set includes impurity particle size, impurity sphericity and impurity smoothness coefficient; the standard threshold includes particle size threshold, sphericity threshold and smoothness threshold; The parameter anomalies in the spherical alumina parameter set are identified by using a standard threshold value to obtain the second coefficient.

8. The intelligent data analysis system based on alumina production according to claim 1 is characterized in that: The calculation formula of the quality coefficient is: A particle index is obtained through the purity of spherical alumina and the first coefficient; an impurity index is obtained through the purity of spherical alumina, the second coefficient and the spherical alumina impurity threshold; and a quality coefficient is obtained through the particle index and the impurity index.

9. A method for intelligent data analysis based on alumina production, characterized in that: include: Obtain ball mill parameters, raw material parameters and spherical alumina impurity thresholds; Identify the ball mill parameters and raw material parameters according to the threshold prediction model to determine the standard threshold; Processing the alumina raw material by a ball mill to obtain spherical alumina; The purity of the spherical alumina is detected, and the spherical alumina is sampled by random sampling to obtain a plurality of samples; the samples are observed by electron microscope to obtain a morphology image set; Constructing a morphology image segmentation model to segment the morphology image set to obtain a first image set and a second image set; the first image set is the microscopic morphology of spherical aluminum oxide, and the second image set is the microscopic morphology of impurities; Constructing a first image recognition model to recognize the first image set, and obtaining a spherical alumina parameter set; the spherical alumina parameter set includes particle size data, sphericity data, and a smoothness coefficient of spherical alumina particles; and calculating a first coefficient according to the spherical alumina parameter set; Constructing a second image recognition model to recognize the second image set, and obtaining an impurity parameter set; the impurity parameter set includes an impurity particle size, an impurity sphericity, and an impurity smoothness coefficient; and calculating a second coefficient according to the impurity parameter set; The quality coefficient is calculated by using the first coefficient, the second coefficient, the impurity threshold of spherical alumina and the purity of spherical alumina.

10. The method for intelligent data analysis based on alumina production according to claim 9, characterized in that: The morphological image segmentation model includes an image input layer, an image preprocessing layer, an image segmentation layer, an image recognition layer, an image classification layer and a result output layer; The image input layer is used to input the topographic image set into the model; The image preprocessing layer performs image preprocessing on the topographic image set to obtain a preprocessed topographic image; The image segmentation layer identifies and segments the particle shape region in the preprocessed topographic image according to the pixel data to obtain a particle image subset; The image recognition layer recognizes the particle image subset and determines whether the particle is a spherical aluminum oxide particle based on size characteristics, shape characteristics and texture characteristics; The image classification layer divides the particle images determined to be spherical aluminum oxide particles into a first image set, and divides the particle images of non-spherical aluminum oxide particles into a second image set; The result output layer is used to output the first image set and the second image set.

Images