Method for ensuring uniformity of grain and oil matrix standard substance based on characteristic value model

Abstract

The application discloses a kind of based on characteristic value model's grain oil matrix standard substance preparation uniformity guarantee method, the method includes the following steps: obtaining the characteristic signal intensity data of pollutant in different depth and area of grain seed;Processing characteristic signal intensity data, obtain the concentration distribution diagram of pollutant;Based on the concentration distribution data obtained, construct the probability density function of pollutant distribution;Intelligent optimization model of comminution process is constructed and run, and the optimal comminution process parameters are output;Based on the optimal process parameters output, execute comminution, and based on the probability density function, establish the comminution coefficient model;Based on the constructed comminution coefficient model, calculate the theoretical minimum sample weight;Based on the theoretical minimum sample weight, develop dynamic sample weighing algorithm;Dynamic sample weighing algorithm is used to execute sample weighing, and the uniformity of sample weighing is verified;Complete uniformity weighing sample.The method can guarantee that the obtained grain oil matrix standard substance preparation is uniform, improve the accuracy of data measurement.

Description

Technical Field

[0001] This invention relates to the field of standard substance preparation technology, and in particular to a method for ensuring the uniformity of grain and oil matrix standard substance preparation based on characteristic value model. Background Technology

[0002] Matrix reference materials refer to standard materials that possess the characteristics of actual samples. They are intended for quality control, rapid detection, product traceability, consistency assurance, and related research and development in the analysis of actual samples with the same or similar matrices. Grain and oil matrix reference materials are a class of standard materials prepared from raw materials such as rice, wheat, corn, and vegetable oil. They represent the characteristics of actual grain and oil samples and possess features such as stability, homogeneity, accuracy, and traceability. They are a key benchmark for ensuring the quality of grain quality and safety testing data. Grain and oil quality and safety testing is an effective technical means to ensure the quality and safety of grain and oil products and prevent substandard or even contaminated grain and oil products from entering the market.

[0003] Existing methods for screening the uniformity of grain and oil matrix standard reference materials have the following drawbacks:

[0004] (1) The problem of ensuring the homogeneity of standard materials lacks systematic theoretical support;

[0005] (2) The distribution of pollutants in grain grains is uneven. The main components of grains are different, their physicochemical properties are different, and the crushing efficiency of each component is different. Traditional methods of ensuring uniformity are difficult to guarantee the uniformity and representativeness of the samples.

[0006] (3) The correlation between sample quantity and pollutant detection results has not been sufficiently studied;

[0007] (4) The problem of synergistic distribution-crushing-weighing of target substances in the matrix of pollutant analysis has not yet been solved. Summary of the Invention

[0008] The technical problem to be solved by the present invention is how to provide a method for accurately obtaining the uniformity of the preparation of grain and oil matrix standard materials.

[0009] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: a method for screening the homogeneity of grain and oil matrix standard materials based on characteristic value models, comprising the following steps:

[0010] S1, collect in-situ distribution data of pollutants in grain grains, and obtain characteristic signal intensity data of pollutants at different depths and regions of grain grains;

[0011] S2, process the characteristic signal intensity data to obtain the pollutant concentration distribution map;

[0012] S3, Based on the acquired concentration distribution data, construct the pollutant distribution probability density function;

[0013] S4: Build and run the intelligent optimization model of the pulverizing process, and output the optimal pulverizing process parameters;

[0014] S5 executes pulverization based on the output optimal process parameters and establishes a pulverization coefficient model based on the probability density function;

[0015] S6, based on the constructed crushing coefficient model, calculate the theoretical minimum sample weight;

[0016] S7, based on the theoretical minimum sample quantity, developed a dynamic sample weighing algorithm;

[0017] S8 uses a dynamic weighing algorithm to perform weighing and verifies the uniformity of the weighing.

[0018] S9. Complete the homogeneity weighing. If the homogeneity meets the requirements, the standard substance is judged to be homogeneous. Otherwise, it is fed back to step S4 to adjust the pulverization process or step S5 to update the model parameters and rebuild the pulverization coefficient model until the homogeneity meets the requirements.

[0019] The beneficial effects of adopting the above technical solution are as follows: The method described in this invention solves the black box problem of uniformity assurance by establishing a quantitative relationship model between grain contaminant distribution, crushing parameters, and sample weight. It combines in-situ detection, dynamic algorithms, and process control to achieve a closed loop from theory to application. A mapping model between crushing process parameters and particle size results is established using machine learning algorithms, and optimized process parameters are output, enabling autonomous decision-making and real-time control of process parameters. Through systematic control, the uniformity qualification rate of standard materials is increased to over 95%. Attached Figure Description

[0020] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0021] Figure 1 This is the main flowchart of the method described in the embodiments of the present invention;

[0022] Figure 2 This is a schematic cross-sectional view of rice obtained by LA-ICP-MS in an application example of the present invention. Detailed Implementation

[0023] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0024] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0025] like Figure 1 As shown in the figure, this invention discloses a method for screening the homogeneity of grain and oil matrix standard materials based on a characteristic value model, comprising the following steps:

[0026] S1, collect in-situ distribution data of pollutants in grain grains, and obtain characteristic signal intensity data of pollutants at different depths and regions of grain grains;

[0027] S2, process the characteristic signal intensity data to obtain the pollutant concentration distribution map;

[0028] S3, Based on the acquired concentration distribution data, construct the pollutant distribution probability density function;

[0029] S4: Build and run the intelligent optimization model of the pulverizing process, and output the optimal pulverizing process parameters;

[0030] S5 executes pulverization based on the output optimal process parameters and establishes a pulverization coefficient model based on the probability density function;

[0031] S6, based on the constructed crushing coefficient model, calculate the theoretical minimum sample weight;

[0032] S7, based on the theoretical minimum sample quantity, developed a dynamic sample weighing algorithm;

[0033] S8 uses a dynamic weighing algorithm to perform weighing and verifies the uniformity of the weighing.

[0034] S9. Complete the homogeneity weighing. If the homogeneity meets the requirements, the standard substance is judged to be homogeneous. Otherwise, it is fed back to step S4 to adjust the pulverization process or step S5 to update the model parameters and rebuild the pulverization coefficient model until the homogeneity meets the requirements.

[0035] The above steps will be explained in detail below with specific methods:

[0036] Step S1: Collect in-situ distribution data of pollutants in grain grains:

[0037] Laser ablation-inductively coupled plasma mass spectrometry (LA-ICP-MS), high-resolution mass spectrometry, or non-targeted screening mass spectrometry, combined with flexible milling and layering processing technology, are used to perform layer-by-layer scanning and component analysis of grain grains to obtain characteristic signal intensity data of pollutants at different depths and regions of the grains (the input for this step is: the original grain sample; the output is: a three-dimensional signal intensity matrix of pollutants).

[0038] Step S2: Process the signal intensity data from step S1 to obtain the pollutant concentration distribution map:

[0039] The signal intensity matrix obtained in step S1 is quantitatively converted using the standard curve method. The calculation formula is as follows: ,in For concentration, To measure signal strength, The blank signal strength, The sensitivity coefficient is used to obtain the three-dimensional concentration distribution data of pollutants in the grains (the input for this step is: signal intensity matrix; the output is: three-dimensional concentration distribution map of pollutants).

[0040] Step S3: Based on the concentration distribution data in Step S2, construct the pollutant distribution probability density function:

[0041] Spatial statistics and kernel density estimation (KDE) are performed on the three-dimensional concentration data obtained in step S2, and a probability density function (PDF) for the pollutant distribution is generated using a Gaussian kernel function. (The input for this step is: three-dimensional concentration distribution data; the output is: pollutant distribution probability density function). In this application, the probability density function is:

[0042] ;

[0043] in: For the spatial location of pollutants The probability density value at that location; The three-dimensional coordinates of any point within the grain; n Let be the number of sample points; i is the i-th sample point; Let i be the spatial coordinates of the i-th sample point; They are respectively The bandwidth parameter in the direction is used to control the smoothness of the kernel function; K Let Gaussian kernel function be defined as follows:

[0044] ;

[0045] Where u, v, and w are standardized relative displacement variables, and their expressions are:

[0046] .

[0047] Step S4: Build and run the intelligent optimization system for the pulverizing process, and output the optimal pulverizing process parameters.

[0048] Step S4-1 (Data Acquisition): Real-time acquisition of multi-source data, including: distribution characteristic values ​​represented by the probability density function of pollutant distribution based on step S3, real-time operating parameters of the crushing equipment (including but not limited to spindle speed, feed rate, grinding media ratio and crushing time), and particle size distribution data monitored in real time by an online laser particle size analyzer (output: multi-source real-time dataset).

[0049] Step S4-2 (AI Optimization Engine): Using the multi-source real-time dataset collected in Step S4-1 as input, a machine learning algorithm (such as random forest regression or neural network) is employed to establish a nonlinear mapping model between the crushing process parameters and the target particle size distribution. This model uses equipment operating parameters as the core input features and particle size distribution D as the input. 50 D 90 With the distribution width as the prediction target, the model weights are dynamically optimized by continuously inputting new production data and performing incremental learning, and finally outputting a set of recommended values ​​for optimal crushing process parameters (the input of this step is: multi-source real-time dataset; the output is: optimal combination of crushing process parameters).

[0050] Step S5: Perform pulverization based on the optimal process parameters output in Step S4, and establish a pulverization coefficient model based on the probability density function PDF in Step S3:

[0051] The optimal parameters obtained in step S4-2 were used to control the pulverizing equipment for preparation, and the particle size distribution after pulverization was measured. The optimized particle size distribution data was coupled with the probability density function (PDF) from step S3, and a precise pulverization coefficient model was established through multiple regression.

[0052] ;

[0053] in The crushing coefficient is... The median particle size, For particle size standard deviation, , , The regression coefficients are the inputs for this step: probability density function and optimized particle size distribution data; the outputs are: high-precision crushing coefficient model and key parameters.

[0054] Step S6: Based on the crushing coefficient model in Step S5, calculate the theoretical minimum sample weight.

[0055] According to statistical principles, given a confidence level (e.g., 95%) and an allowable uncertainty (U), the formula is used... Calculate the theoretical minimum sample size required to ensure representativeness, where for Distribution critical value, The standard deviation of the sample is 1. The crushing coefficient is... To allow for uncertainty (input: crushing coefficient model and statistical parameters; output: minimum sample weight value).

[0056] Step S7: Based on the minimum sample quantity in Step 6, develop a dynamic sample weighing algorithm.

[0057] Step S7-1: Embed the minimum sample weight into the automatic sample weighing control system and design a dynamic sample weighing algorithm: monitor particle size distribution in real time (via an online particle size analyzer) and adjust the sample weight accordingly.

[0058] ;

[0059] in For the current particle size, The reference particle size is k, and the correction index is k (the input for this step is: minimum sample weight and real-time particle size data; the output is: dynamic sample weighing command signal).

[0060] Step S7-2 Dynamic Decision Logic: The control program calculates and outputs the current required sample quantity m_actual in real time based on the following logic tree:

[0061] If the real-time particle size D 50 _current ∈ [D 50 _low, D 50 _high], then:

[0062] m_actual=m_min_theoretical (particle size is within the ideal range, theoretical value is used);

[0063] Otherwise if D 50 _current <D 50 _low, then:

[0064] m_actual=m_min_theoretical k1 (Fineer particles, better uniformity, reduce sample weight; k1 is the reduction coefficient, 0) <k1<1);

[0065] Otherwise if D 50 _current>D 50 _high, then:

[0066] m_actual = m_min_theoretical k2 (the particles are coarser, with a high risk of non-uniformity. Increase the sample weighing amount. k2 is the coefficient for increase, k2 > 1);

[0067] Step S8: Perform sample weighing using the dynamic sample weighing algorithm in Step S7 and verify the uniformity.

[0068] Use an automatic sample weighing device to perform sample weighing according to the dynamic algorithm, prepare a reference material sample, compare the within-group and between-group differences using the analysis of variance (ANOVA), and calculate the F value:

[0069] F = MS_between / MS_within, and establish a Shewhart control chart to monitor the uniformity (input: dynamic sample weighing instructions and samples; output: uniformity evaluation results and control chart).

[0070] Step S9: Complete the uniformity guarantee:

[0071] Based on the verification results in Step S8, if the uniformity meets the requirements (e.g., F < F_critical and the control chart is in control), it is determined that the uniformity of the reference material meets the standard; otherwise, feedback to Step S4 to adjust the comminution process or Step S5 to update the model parameters (the input for this step is: uniformity evaluation results; the output is: uniformity guarantee conclusion and feedback adjustment instructions).

[0072] This technical solution includes three modules: namely, Module 1 - Pollutant Distribution Research, Module 2 - Intelligent Comminution Optimization System, and Module 3 - Dynamic Sample Weighing and Verification. The three modules are an organic whole with close cooperation and closed-loop data flow.

[0073] Module 1: The constructed probability density function (PDF) of pollutant distribution is the theoretical basis and starting point of the entire technology. This PDF is provided to both Module 2 and Module 3 simultaneously, providing key feature inputs for the AI optimization engine and calculation basis for the comminution coefficient model (η).

[0074] Module 2: Receive the PDF data from Module 1 and guide the optimization of the AI model. The output optimal comminution process parameters and optimized particle size distribution data are the prerequisites for ensuring that the sample reaches the expected uniformity level. Precisely transfer the processing results (particle size data) to Module 3 as the basis for sample weighing decision-making.

[0075] Module 3: Receive the PDF from Module 1 and the particle size data from Module 2, and calculate the theoretical sample weighing amount. Its core - the dynamic sample weighing control program, through real-time linkage with the on-line detection device, forms a closed-loop feedback system:

[0076] Feedforward control: Dynamically adjust the sample weighing amount according to the real-time particle size (from the particle size analyzer).

[0077] Feedback control: Based on real-time detection results (from rapid detection equipment), it determines whether the uniformity deviates from expectations. If a deviation occurs, a signal is immediately generated and fed back to the AI ​​optimization engine in module two, triggering further optimization of the pulverizing process, thus forming a powerful self-correcting mechanism.

[0078] The final uniformity verification results will also be fed back to Module 2 for continuous optimization of the AI ​​model.

[0079] Compared to traditional methods that lack theoretical support (such as the quantitative relationship between crushing parameters and contaminant distribution) and rely mainly on empirical operations, the method described in this application achieves standardized control through mathematical models and algorithms. Furthermore, while existing technologies focus on single steps (such as crushing or weighing), this method integrates the entire chain of "distribution patterns - crushing - weighing," forming a systematic solution. Through systematic theoretical support, this application increases the uniformity qualification rate of standard materials from 78% using traditional methods to 95%, and the related technology has been successfully applied to the national grain standard material development project.

[0080] Application Example 1:

[0081] Homogeneity control of rice cadmium analysis standard reference material:

[0082] Commissioned by a company, we prepared a batch of natural cadmium-contaminated rice matrix standard materials for proficiency testing of grain heavy metal rapid detection equipment. Traditional methods, due to the uneven distribution of cadmium in rice grains, typically rely on examining the impact of single factors such as particle size, time, rotation speed, and mixing time on the repeatability of initial test results. This requires multiple samplings and repeated mixing to achieve the desired uniformity. Furthermore, the minimum sample size is usually determined based on experience or standard requirements, and is thus loosely controlled, leading to repetitive, time-consuming processes that negatively impact the consistency of test results.

[0083] Pollutant distribution modeling:

[0084] LA-ICP-MS was used to perform layer-by-layer scanning of typical cadmium-contaminated rice to obtain signal intensity data of cadmium at different depths of the grain (cortex, endosperm, germ), and a three-dimensional concentration distribution map was constructed. The spatial distribution probability density function (PDF) of cadmium was established by kernel density estimation, and it was found that cadmium was mainly concentrated in the cortex and germ, with a concentration difference of up to 3.2 times, as shown in Table 1.

[0085] Table 1: Cadmium signal intensity (raw data from LA-ICP-MS scan)

[0086]

[0087] The signal strength was converted using the standard curve method, with a sensitivity coefficient k = 0.000050 mg / kg·cps and a blank signal of 210 cps.

[0088] The PDF shows that Cd is enriched in the cortex and germ (peak value 1.05 mg / kg), while the endosperm region has only 0.24 mg / kg, with a concentration gradient of 4.4 times. The uneven distribution of pollutants provides weight input for subsequent pulverization-weighing optimization.

[0089] Grinding optimization:

[0090] A grinding process AI optimization model is constructed, with input parameters including PDF feature values, equipment rotation speed, feed rate, and grinding media ratio, and the objective is to control the particle size D. 50 At 80±5 μm, D 90 <150 μm. Recommended parameters for model output: spindle speed 3200 rpm, feed rate 1.2 kg / h, zirconia ball ratio 70%. After crushing with these parameters, the particle size distribution is uniform. Specific data are shown in Table 2.

[0091] Table 2 Particle size distribution (detected online by laser particle size analyzer)

[0092]

[0093] D in Table 2 50 =79.4 µm, σ=25.1 µm and substitute in equation (7) η=a+b·D 50 From +c·σ, we get η=0.87 (R 2 =0.96), based on obtaining a high pulverization coefficient, the minimum sample weight required to ensure uniformity can be calculated according to step S6 of this application. Among them, a, b, and c are obtained by fitting the experimental data of this batch and are not general constants; if the raw materials or equipment are changed, the experiment and regression must be repeated according to step S4 → step S5.

[0094] Minimum sample quantity calculation and dynamic sample weighing:

[0095] At a 95% confidence level (t = 1.96) and an allowable uncertainty of 5%, with a sample standard deviation s = 0.18 mg / kg and a pulverization coefficient η = 0.87, the theoretical minimum sample weight was calculated to be 0.38 g. The dynamic weighing algorithm automatically adjusts the sample weight based on real-time particle size feedback. When the particle size is too fine, the weight is lowered to 0.32 g (k1 = 0.85), and when it is too coarse, it is raised to 0.45 g (k2 = 1.18). k1 and k2 are incrementally learned from the first 20 sets of online data, conforming to the logic tree in step S7-2 of this application, ensuring the representativeness of each sample. After weighing according to the dynamic algorithm, ANOVA verification was performed according to step S8 of this application, and the results are as follows.

[0096] Homogeneity verification:

[0097] Randomly select 10 groups, with 3 bottles in each group, for a total of 30 independent samples (df between groups = 10 - 1 = 9, df within groups = 30 - 10 = 20). Use ANOVA to analyze the cadmium content. The F value is 1.34 (critical value F 0.05 = 1.87). The Shewhart control chart shows that all points are within the control limits, and the homogeneity is qualified. Compared with the traditional method (F = 2.12, unqualified), the homogeneity is significantly improved, as shown in Table 3.

[0098] Table 3 Homogeneity verification (ANOVA)

[0099]

[0100] F = MS_between / MS_within = 0.0021 / 0.0016 = 1.31; F_critical (0.05, 9, 20) = 2.39; Conclusion: F < F_critical, the homogeneity is qualified;

[0101] After the above control, based on three consecutive preparation batches (N = 90), F < F_critical and |Cd - certified value| ≤ 3% (k = 2). The homogeneity qualification rate has increased from 78% in the traditional method to 96%. In addition, this method makes the sample weighing amount more clear, and the preparation cycle is shortened by 30%, achieving efficient and standardized production. While improving the development level of reference materials, it also provides scientific data support for guiding the use of reference materials.

[0102] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use, and processing of relevant data need to comply with relevant regulations.

[0103] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0104] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic units, data processing logic units, etc., and are not limited to these.

[0105] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0106] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for ensuring the homogeneity of grain and oil matrix standard material preparation based on a characteristic value model, characterized in that... Includes the following steps: S1, collect in-situ distribution data of pollutants in grain grains, and obtain characteristic signal intensity data of pollutants at different depths and regions of grain grains; S2, process the characteristic signal intensity data to obtain the pollutant concentration distribution map; S3, Based on the acquired concentration distribution data, construct the pollutant distribution probability density function; S4: Build and run the intelligent optimization model of the pulverizing process, and output the optimal pulverizing process parameters; S5 executes pulverization based on the output optimal process parameters and establishes a pulverization coefficient model based on the probability density function; S6, based on the constructed crushing coefficient model, calculate the theoretical minimum sample weight; S7, based on the theoretical minimum sample size, developed a dynamic sample weighing algorithm: S7-1: Embed the minimum sample weight into the automatic sample weighing control system and design a dynamic sample weighing algorithm: Real-time monitoring of particle size distribution and adjustment of sample weight: ； in For the current particle size, The reference particle size is given by k, which is the correction index. S7-2: Calculate and output the current required sample quantity m_actual in real time based on the following logic tree: If the real-time particle size D 50 If _current is within the ideal particle size range, then the theoretical value is used. The current required sample size is m_actual; If the particles are finer and more uniform, then reduce the sample weight: m_actual = m_min_theoretical k1, where k1 is the down adjustment coefficient, 0 <k1<1; If the particles are coarser and the risk of unevenness is higher, then increase the sample weight: m_actual = m_min_theoretical k2, where k2 is the upward adjustment coefficient, and k2>1; S8 uses a dynamic weighing algorithm to perform weighing and verifies the uniformity of the weighing. S9. Complete the homogeneity weighing. If the homogeneity meets the requirements, the standard substance is judged to be homogeneous. Otherwise, it is fed back to step S4 to adjust the pulverization process or step S5 to update the model parameters and rebuild the pulverization coefficient model until the homogeneity meets the requirements.

2. The method for ensuring uniformity in the preparation of grain and oil matrix standard materials based on a characteristic value model as described in claim 1, characterized in that, S1 specifically includes the following steps: Laser ablation-inductively coupled plasma mass spectrometry, high-resolution mass spectrometry, or non-targeted screening mass spectrometry are used in conjunction with flexible milling and layered processing to perform layer-by-layer scanning and component analysis of grain grains, thereby obtaining characteristic signal intensity data of pollutants at different depths and regions of the grains.

3. The method for ensuring uniformity in the preparation of grain and oil matrix standard materials based on a characteristic value model as described in claim 1, characterized in that, S2 specifically includes the following steps: The acquired signal strength matrix is ​​quantitatively transformed using the standard curve method. The calculation formula is as follows: ； in For concentration, To measure signal strength, The blank signal strength, The sensitivity coefficient is used to obtain the three-dimensional concentration distribution data of pollutants in the grains.

4. The method for ensuring uniformity in the preparation of grain and oil matrix standard materials based on a characteristic value model as described in claim 1, characterized in that, S3 specifically includes the following steps: Spatial statistics and kernel density estimation were performed on the obtained three-dimensional concentration data, and the probability density function PDF of pollutant distribution was generated using the Gaussian kernel function.

5. The method for ensuring uniformity in the preparation of grain and oil matrix standard materials based on a characteristic value model as described in claim 1, characterized in that, S4 specifically includes the following steps: S4-1: Real-time acquisition of multi-source data, including: distribution characteristic values ​​based on the probability density function of pollutant distribution, real-time operating parameters of the crushing equipment, and particle size distribution data monitored in real time by an online laser particle size analyzer; S4-2: Using the collected multi-source real-time datasets as input, a machine learning algorithm is employed to establish a nonlinear mapping model between crushing process parameters and the target particle size distribution. This model uses equipment operating parameters as the core input features and particle size distribution D as the input. 50 D 90 Using the distribution width as the prediction target, the model weights are dynamically optimized by continuously inputting new production data and performing incremental learning, and finally outputting a set of recommended values ​​for optimal crushing process parameters.

6. The method for ensuring uniformity in the preparation of grain and oil matrix standard materials based on a characteristic value model as described in claim 1, characterized in that, S5 specifically includes the following steps: The grinding equipment was operated using the recommended optimal grinding process parameters for preparation, and the particle size distribution after grinding was measured. The optimized particle size distribution data was coupled with the probability density function (PDF) to establish an accurate grinding coefficient model through multiple regression. ； in The crushing coefficient is... The median particle size, For particle size standard deviation, , , is the regression coefficient.

7. The method for ensuring uniformity in the preparation of grain and oil matrix standard materials based on a characteristic value model as described in claim 1, characterized in that, S6 specifically includes the following steps: According to statistical principles, given a confidence level and allowable uncertainty, the formula is used: Calculate the theoretical minimum sample size required to ensure representativeness; in for Distribution critical value, The standard deviation of the sample is 1. The crushing coefficient is... To allow for uncertainty.

8. The method for ensuring uniformity in the preparation of grain and oil matrix standard materials based on a characteristic value model as described in claim 1, characterized in that, S8 specifically includes the following steps: Using automated weighing equipment and a dynamic algorithm, standard substance samples were prepared. ANOVA (Analysis of Variance) was used to compare within-group and between-group differences, and the homogeneity F-value was calculated. F = MS_between / MS_within, and a Shewhart control chart is established to monitor uniformity; MS_between represents the variance between groups, and MS_within represents the variance within groups.

9. The method for ensuring uniformity in the preparation of grain and oil matrix standard materials based on a characteristic value model as described in claim 4, characterized in that, The pollutant probability density function PDF is: ； in: For the spatial location of pollutants The probability density value at that location; The three-dimensional coordinates of any point within the grain; n Let be the number of sample points; i is the i-th sample point; Let i be the spatial coordinates of the i-th sample point; They are respectively The bandwidth parameter in the direction is used to control the smoothness of the kernel function; K Let Gaussian kernel function be defined as follows: ； Where u, v, and w are standardized relative displacement variables, and their expressions are: 。

Images