A method for predicting the quality of feed oil raw materials based on physical and chemical indicators

By constructing a nonlinear mapping model and relational mapping model, combined with physical and chemical indicators and time series analysis, the problem that traditional models are difficult to predict the quality of feed oil and fat raw materials in different varieties and processing technologies is solved, and accurate prediction and real-time monitoring of feed oil and fat raw materials are achieved. , improves the scientific nature of food safety and production decisions.

CN118569716BActive Publication Date: 2024-11-22山东创脂生物科技有限公司
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410666213.1
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-05-27
Publication Date
2024-11-22
Estimated Expiration
2044-05-27

AI Technical Summary

Technical Problem

It is difficult for traditional single machine learning models to accurately predict the quality of feed oil and raw materials of different varieties and processing technologies, especially under the influence of storage environment and the quality is highly dynamic.

Method used

Build a nonlinear mapping model, dynamic change model and relational mapping model based on physical and chemical indicators. Combining varieties, processing technology and storage environment data, physical and chemical indicators are obtained through chemical analysis, comprehensive feature vectors and time series models are established, and real-time prediction and analysis are carried out. .

Benefits of technology

It improves the accurate prediction ability of feed oil raw materials, can instantly reflect the impact of storage environment and processing technology on quality, provide a more scientific nutritional value evaluation and early warning mechanism, and ensures the accuracy of production decisions and food safety.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118569716B_ABST
    Figure CN118569716B_ABST
Patent Text Reader

Abstract

The present application discloses a method for predicting the quality of feed oil raw materials based on physical and chemical indicators, which relates to the technical field of feed quality detection. The method collects variety data, processing technology data and storage environment parameters of different feed oil raw material samples, obtains physical and chemical indicators, and uses variety data, processing technology data, storage environment parameters and physical and chemical indicators to obtain a relationship mapping model between physical and chemical indicators and different varieties, processing technologies and storage environments. The method collects variety data, processing technology data and storage environment parameters in real time, predicts real-time physical and chemical indicators, calculates nutritional value reference values, presets nutritional value thresholds, calculates nutritional value deviations and a warning index at the current moment. When the warning index at r consecutive moments is greater than the warning trigger threshold, a warning signal is issued, and the abnormal feed oil raw materials are subjected to secondary processing, so as to better predict the quality of feed oil raw materials of different varieties, processing technologies and storage environments.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the technical field of feed quality detection, and in particular to a method for predicting the quality of feed oil raw materials based on physical and chemical indicators. Background Art

[0002] Animal feed refers to products used to feed livestock, poultry, aquatic products and other animals. Its main purpose is to provide the nutrients required by animals to maintain health, growth and production. The basic components of feed usually include: protein, carbohydrates, fats, vitamins and minerals, and additives. Feed fats play multiple roles in animal feed: fats are high-energy-density feed components that can provide more energy than carbohydrates and proteins; fats help the absorption of fat-soluble vitamins, which are essential for animal growth and reproduction; fats increase the palatability of feed and promote animal food intake.

[0003] Traditional methods of oil quality testing use chemical analysis to determine the free fatty acid content, peroxide content, moisture and impurities in the oil, instrumental analysis to analyze the fatty acid composition of the oil, determine specific compounds in the oil such as vitamins or pigments, detect functional groups in the oil, and evaluate quality. With the advancement of technology, machine learning is becoming an important tool in the field of feed quality testing technology. Machine learning models can learn from large amounts of data and predict the quality of oils.

[0004] However, there are many types of feed oil raw materials, and the physical and chemical properties of oils of different varieties and processing technologies vary greatly. Feed oil raw materials are easily affected by the storage environment, and their quality has a certain dynamic variability. In addition, feed oil raw materials of different varieties and different processing technologies are affected by the storage environment to different degrees. Traditional single machine learning models are difficult to accurately predict the quality of feed oil raw materials of different varieties and processing technologies. Summary of the invention

[0005] The present application aims to solve at least one of the technical problems in the related art to a certain extent. To this end, one purpose of the present application is to propose a method for predicting the quality of feed oil raw materials based on physical and chemical indicators, which can better predict the quality of feed oil raw materials of different varieties, processing technologies and storage environments.

[0006] One aspect of the present application provides a method for predicting the quality of feed oil raw materials based on physical and chemical indicators, comprising:

[0007] Step S100: collecting variety data, processing technology data and storage environment parameters of different feed oil raw material samples, and obtaining physical and chemical indicators of the feed oil raw material samples;

[0008] The specific method for collecting the variety data, processing technology data and storage environment parameters of different feed oil raw material samples and obtaining the physical and chemical indicators of the feed oil raw material samples is:

[0009] Step S110: Collect feed oil raw material samples covering different varieties, different processing technologies, and different storage environments, and obtain variety data, processing technology data, and storage environment parameters corresponding to each feed oil raw material sample;

[0010] Step S120: Analyze the components of the feed oil raw material sample by chemical analysis to obtain the physical and chemical indicators of the feed oil raw material, wherein the physical and chemical indicators include acid value, peroxide value, and water content;

[0011] Step S200: using variety data, processing technology data, storage environment parameters and physical and chemical indicators to obtain a relationship mapping model between the physical and chemical indicators of the feed oil raw material sample and different varieties, processing technologies and storage environments;

[0012] The specific method of using variety data, processing technology data, storage environment parameters and physical and chemical indicators to obtain the relationship mapping model between the physical and chemical indicators of feed oil raw material samples and different varieties, processing technologies and storage environments is:

[0013] Step S210: vectorize the variety data and processing technology data of feed oil raw material samples of different varieties and different processing technologies to obtain the variety feature vector V s and the processing feature vector V p ;

[0014] Step S220: Normalize the storage environment parameters to obtain a normalized storage environment feature vector V e ;

[0015] Step S230: The variety feature vector V s , processing technology feature vector V p and storage environment feature vector V e Splice into a comprehensive feature vector V c , with the variety characteristic vector V s , processing technology feature vector V p and storage environment feature vector V e As the first input, the corresponding comprehensive feature vector V c As the first output, the comprehensive feature vector V c Paired with the corresponding time series of physical and chemical indicators to form training samples and construct variety feature vector V s , processing technology feature vector V p and store the environmental feature vector v e The nonlinear mapping model betweens , v p , v e );

[0016] Step S240: nonlinear mapping model f(v s , v p , v e ) is trained to optimize the model parameters of the nonlinear mapping model by minimizing the first loss function to obtain a weight matrix of the nonlinear mapping model and a bias vector of the nonlinear mapping model;

[0017] The model expression of the nonlinear mapping model is: Among them, W1 is the weight matrix of the nonlinear mapping model, b1 is the bias vector of the nonlinear mapping model, Represents a vector concatenation operation;

[0018] The first loss function is the mean square error, and the calculation formula is: Among them, n is the number of training samples, v c,i is the true comprehensive feature vector of the i-th training sample, f(v c,i ) is the predicted value of the comprehensive feature vector of the i-th training sample by the nonlinear mapping model;

[0019] Step S250: The time series of the physical and chemical indicators at the previous k moments and the storage environment characteristic vector are used as the second input, and the physical and chemical indicators at the current moment t are used as the second output to construct a dynamic change model g(y t-1 ,y t-2 , ..., y t-k , V e ), optimizing the model parameters of the dynamic change model by minimizing the second loss function, and obtaining the coefficient matrix of the dynamic change model, the environmental feature weight matrix and the bias vector of the dynamic change model;

[0020] The model expression of the dynamic change model is: Among them, y t represents the physical and chemical index at time t, y t-1 ,y t-2 , ..., y t-k represents the time series of physical and chemical indicators at the previous k moments, W 2z represents the coefficient matrix of the dynamic change model, W3 represents the environmental characteristic weight matrix, and b2 represents the bias vector of the dynamic change model of physical and chemical indicators;

[0021] The second loss function is the mean square error, and the calculation formula is: Among them, y t,i is the real physical and chemical index of the i-th training sample at time t, g(X t,i , ve,i ) is the input sequence X of the physical and chemical indicators of the dynamic change model for the i-th training sample t,i and store the environmental feature vector v e,i The predicted value of the physical and chemical index input sequence X t,i is the time series of physical and chemical indicators at the previous k moments;

[0022] Step S260: Using the comprehensive feature vector V c , the second output of the dynamic change model and the storage environment feature vector are used as the third input, and the physical and chemical indicators are used as the third output, to construct a relationship mapping model, optimize the model parameters by minimizing the third loss function, and obtain the weight matrix of the relationship mapping model, the bias vector of the relationship mapping model and the trained relationship mapping model;

[0023] The model expression of the relationship mapping model is: , where h(V c ,y t-1 ,y t-2 , ..., y t-k , V e ) represents the final relationship mapping model, W4 represents the weight matrix of the relationship mapping model, and b3 represents the bias vector of the relationship mapping model;

[0024] The third loss function is the mean square error, and the calculation formula is: Among them, h(v c,i , g(X t,i , v e,i ), v e,i ) is the physical and chemical index predicted by the relational mapping model based on the third input;

[0025] Step S300: collecting the variety data, processing technology data, and storage environment parameters of the feed oil raw materials in real time, using the relationship mapping model to predict the real-time physical and chemical indicators of the feed oil raw materials, and calculating the nutritional value reference value of the feed oil raw materials based on the real-time physical and chemical indicators;

[0026] The specific method of collecting the variety data, processing technology data, and storage environment parameters of the feed oil raw materials in real time, predicting the real-time physical and chemical indicators of the feed oil raw materials by using the relationship mapping model, and calculating the reference value of the nutritional value of the feed oil raw materials according to the real-time physical and chemical indicators is as follows:

[0027] Step S310: Real-time collection of feed oil raw material variety data, processing technology data, storage environment parameters, and use of a relationship mapping model to predict real-time physical and chemical indicators of the feed oil raw materials, including but not limited to: real-time acid value Real-time peroxide value Real-time moisture content

[0028] Step S320: Obtain the standard physical and chemical index reference values ​​and standard nutritional values ​​of the corresponding feed oil raw materials from the nutrition database. The standard physical and chemical index reference values ​​include but are not limited to: acid value reference value Y 1 , Peroxide value reference value Y 2 , moisture content reference value Y 3 ;

[0029] Step S330: Preset nutritional value influencing factor α j , using real-time physical and chemical indicators and standard physical and chemical index reference values ​​to calculate the estimated nutritional value of feed oil raw materials;

[0030] The preset nutritional value influencing factor α j The specific method for calculating the estimated value of the nutritional value of feed oil raw materials using real-time physical and chemical indicators and standard physical and chemical index reference values ​​is:

[0031] Step S331: Preset nutritional value influencing factor α j , the influencing factors of nutritional value and real-time physical and chemical indicators are combined to obtain the influencing sequence of physical and chemical indicators in It represents the mth real-time physical and chemical index at time t, where m is the total number of real-time physical and chemical indexes;

[0032] Step S332: Calculate the similarity function based on the real-time physical and chemical indexes and the standard physical and chemical index reference values, as well as the physical and chemical index influence sequence

[0033] The calculation formula of the similarity function is: in, represents the jth real-time physical and chemical index, Y j It represents the reference value of the jth standard physical and chemical index;

[0034] Step S333: Calculate the estimated nutritional value based on the similarity function

[0035] The estimated nutritional value is calculated as follows: Where m is the index number of real-time physical and chemical index or standard physical and chemical index reference value, ω j is the similarity function corresponding to the jth real-time physical and chemical index The weight coefficient of , β is the baseline nutritional value;

[0036] Step S340: Calculate the nutritional value reference value based on the estimated nutritional value, the nutritional value reference value V ref The calculation formula is: Where λ is the proportional coefficient of the estimated nutritional value, (1-λ) is the proportional coefficient of the standard nutritional value, and V std Indicates standard nutritional value, is an estimate of nutritional value;

[0037] Step S400: Preset a nutritional value threshold, calculate the nutritional value deviation according to the nutritional value reference value, and calculate the warning index at the current moment in combination with the warning index at the previous moment. When the warning index at r consecutive moments is greater than the warning trigger threshold, issue a warning signal;

[0038] The preset nutritional value threshold, the nutritional value deviation is calculated according to the nutritional value reference value, and the warning index at the current moment is calculated in combination with the warning index at the previous moment. When the warning index at r consecutive moments is greater than the warning trigger threshold, the specific method of issuing the warning signal is:

[0039] Step S410: Collect the nutritional value reference values ​​of feed oil raw materials with different storage time, different varieties, different storage environments and different processing technologies, and calculate the average of all nutritional value reference values and standard deviation δ, the nutritional value threshold is calculated;

[0040] The nutritional value threshold includes an upper threshold and a lower threshold, and the specific calculation method is:

[0041] The upper threshold of the nutritional value threshold is:

[0042] The lower limit of the nutritional value threshold is:

[0043] Step S420: For a given feed oil raw material, calculate its nutritional value deviation d, and the calculation formula of the nutritional value deviation d is:

[0044] Step S430: Calculate the warning index u t , the calculation formula of the warning index is: t =ε×u t-1 +(1-ε)×|d t |, where u t represents the warning index at the current time t, u t-1 represents the warning index at the previous moment, ε is the time decay factor, d t Indicates the deviation of nutritional value at the current time t;

[0045] Step S440: When the warning index for r consecutive moments is greater than the preset warning trigger threshold U, a warning signal is sent to the feed storage management personnel to remind the feed storage management personnel to take corresponding quality control measures;

[0046] Step S500: separating the feed and fat raw materials that have issued warning signals as abnormal feed and fat raw materials from normal batches, analyzing the causes of the warnings, and performing secondary processing on the abnormal feed and fat raw materials according to the causes of the warnings;

[0047] The specific method of separating the feed oil raw materials that have issued warning signals as abnormal feed oil raw materials from normal batches, analyzing the warning causes, and performing secondary processing on the abnormal feed oil raw materials according to the warning causes is as follows:

[0048] Step S510: Mark the feed oil raw materials that have issued warning signals as abnormal feed oil raw materials, separate them from normal batches, and perform quality inspections on them, including but not limited to: physical and chemical index analysis, pollutant detection and microbial detection;

[0049] Step S520: Analyze the quality inspection results to obtain the causes of the warning signal issued by the abnormal feed oil raw material, including but not limited to: moisture, microbial contamination, impurity contamination;

[0050] Step S530: performing secondary processing on the abnormal feed oil raw material according to the cause of the warning, wherein the secondary processing method includes but is not limited to: re-refining, filtering, adding additives, and changing the storage environment;

[0051] Step S540: for the abnormal feed oil raw material after secondary processing, predict its physical and chemical indicators again, calculate the nutritional value reference value and the early warning index, and determine whether the early warning index at this time is greater than the early warning trigger threshold;

[0052] Step S550: When the warning index at this time is greater than the warning trigger threshold, the abnormal feed and fat raw material is determined to be a batch whose quality cannot be restored, and other qualified feed and fat raw materials are selected to replace it to maintain the operation of the production line. When the warning index at this time is less than or equal to the warning trigger threshold, the storage environment of the abnormal feed and fat raw material after secondary processing is changed.

[0053] One aspect of the present application provides a feed oil raw material quality prediction system based on physical and chemical indicators, comprising:

[0054] The oil and fat raw material data collection module is used to collect the variety data, processing technology data and storage environment parameters of different feed oil and fat raw material samples, and obtain the physical and chemical indicators of the feed oil and fat raw material samples;

[0055] A relationship mapping model building module is used to obtain a relationship mapping model between the physical and chemical indicators of feed oil raw material samples and different varieties, processing technologies and storage environments using variety data, processing technology data, storage environment parameters and physical and chemical indicators;

[0056] The raw material nutritional value calculation module is used to collect the variety data, processing technology data, and storage environment parameters of feed oil raw materials in real time, use the relationship mapping model to predict the real-time physical and chemical indicators of feed oil raw materials, and calculate the nutritional value reference value of feed oil raw materials based on the real-time physical and chemical indicators;

[0057] The early warning calculation and triggering module is used to preset the nutritional value threshold, calculate the nutritional value deviation according to the nutritional value reference value, and calculate the early warning index at the current moment in combination with the early warning index at the previous moment. When the early warning index at r consecutive moments is greater than the early warning triggering threshold, an early warning signal is issued;

[0058] The cause analysis and secondary processing module is used to separate the feed and oil raw materials that have issued warning signals as abnormal feed and oil raw materials from normal batches, analyze their warning causes, and perform secondary processing on the abnormal feed and oil raw materials based on the warning causes.

[0059] One aspect of the present application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps in a method for predicting the quality of feed oil raw materials based on physical and chemical indicators are implemented.

[0060] One aspect of the present application provides a readable storage medium storing a computer program, wherein the computer program is suitable for loading by a processor to execute steps in a method for predicting the quality of feed oil raw materials based on physical and chemical indicators.

[0061] Compared with the prior art, the feed oil raw material quality prediction method based on physical and chemical indicators proposed in this application has the following advantages:

[0062] This application captures the complex connections and nonlinear relationships between varieties, processing technology and storage environment by constructing nonlinear mapping models, dynamic change models and relationship mapping models, tracks the changes of physical and chemical indicators over time, and comprehensively considers the dynamic change trends of varieties, processing technology, storage environment and physical and chemical indicators. The outputs of the three models are both different and related. Through step-by-step modeling and optimization, an accurate prediction of physical and chemical indicators is finally obtained, the accuracy of the prediction is improved, and the impact of the storage environment on feed oil raw materials of different varieties and processing technologies is more accurately analyzed;

[0063] This application can instantly respond to any changes in the state of raw materials by collecting real-time data on the variety, processing technology and storage environment of feed oil raw materials. This is more advanced and practical than traditional detection methods. It uses a relational mapping model to predict real-time physical and chemical indicators and conduct a more accurate and immediate assessment of material quality. The data and analysis provided by the entire method can directly support production decisions;

[0064] This application continuously monitors the nutritional value deviation and early warning index. The system can evaluate the quality status of feed and oil raw materials in real time, discover problems in time and issue early warnings, which is more efficient than traditional periodic inspections. The mean and standard deviation are used to set the upper and lower thresholds of the nutritional value. This method can more scientifically reflect the normal nutritional value range of feed and oil raw materials under different conditions, thereby improving the accuracy of early warnings. The calculation of the early warning index not only considers the nutritional value deviation at the current moment, but also combines the early warning index at the previous moment, and adjusts the influence through the time decay factor. This method enables the early warning system to have a certain memory and adaptability, and can handle data fluctuations more smoothly. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 A method flow chart of a method for predicting the quality of feed oil raw materials based on physical and chemical indicators provided in this application;

[0066] Figure 2 A time sequence logic diagram of a method for predicting the quality of feed oil raw materials based on physical and chemical indicators provided in this application;

[0067] Figure 3 A functional module diagram of a feed oil raw material quality prediction system based on physical and chemical indicators provided in this application;

[0068] Figure 4 A schematic diagram of the structure of an electronic device provided in this application;

[0069] Figure 5 It is a schematic diagram of the structure of a readable storage medium provided by this application. DETAILED DESCRIPTION

[0070] In order to better understand the present application, various aspects of the present application will be described in more detail with reference to the accompanying drawings. It should be understood that these detailed descriptions are only descriptions of exemplary embodiments of the present application and do not limit the scope of the present application in any way. Throughout the specification, the same reference numerals refer to the same elements. The expression "and / or" includes any and all combinations of one or more of the associated listed items.

[0071] In the accompanying drawings, the size, dimensions and shapes of the elements have been slightly adjusted for ease of illustration. The drawings are for illustration only and are not strictly drawn to scale. As used herein, the terms "substantially", "approximately" and similar terms are used as terms of approximation, not as terms of degree, and are intended to illustrate the inherent deviations in measurements or calculations that will be recognized by those of ordinary skill in the art. In addition, in this application, the order in which the steps are described does not necessarily represent the order in which these processes occur in actual operation, unless otherwise specified or can be derived from the context.

[0072] It should also be understood that expressions such as "including", "comprising", "having", "containing" and / or "comprising" are open rather than closed expressions in this specification, which indicate the presence of the stated features, elements and / or components, but do not exclude the presence of one or more other features, elements, components and / or combinations thereof. In addition, when expressions such as "at least one of..." appear after a list of listed features, they modify the entire list of features rather than just the individual elements in the list. In addition, when describing embodiments of the present application, "may" is used to mean "one or more embodiments of the present application". And, the term "exemplary" is intended to refer to an example or illustration.

[0073] Unless otherwise specified, all words (including engineering terms and scientific and technological terms) used in this article have the same meaning as those commonly understood by ordinary technicians in the field to which this application belongs. It should also be understood that, unless clearly stated in this application, words defined in common dictionaries should be interpreted as having the same meaning as their meaning in the context of the relevant technology, and should not be interpreted in an idealized or overly formal sense.

[0074] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0075] Example 1

[0076] like Figure 1-Figure 2 As shown, a method for predicting the quality of feed oil raw materials based on physical and chemical indicators provided by this application includes:

[0077] Step S100: collecting variety data, processing technology data and storage environment parameters of different feed oil raw material samples, and obtaining physical and chemical indicators of the feed oil raw material samples;

[0078] The specific method for collecting the variety data, processing technology data and storage environment parameters of different feed oil raw material samples and obtaining the physical and chemical indicators of the feed oil raw material samples is:

[0079] Step S110: Collect feed oil raw material samples covering different varieties, different processing technologies, and different storage environments, and obtain variety data, processing technology data, and storage environment parameters corresponding to each feed oil raw material sample;

[0080] Step S120: Analyze the components of the feed oil raw material sample by chemical analysis to obtain the physical and chemical indicators of the feed oil raw material, wherein the physical and chemical indicators include acid value, peroxide value, and water content;

[0081] This step uses chemical analysis to analyze the components of feed oil raw material samples, obtain physical and chemical indicators, and collect various relevant data of feed oil raw material samples, providing basic data for quality assessment of feed oil raw materials;

[0082] Step S200: using variety data, processing technology data, storage environment parameters and physical and chemical indicators to obtain a relationship mapping model between the physical and chemical indicators of the feed oil raw material sample and different varieties, processing technologies and storage environments;

[0083] The specific method of using variety data, processing technology data, storage environment parameters and physical and chemical indicators to obtain the relationship mapping model between the physical and chemical indicators of feed oil raw material samples and different varieties, processing technologies and storage environments is:

[0084] Step S210: vectorize the variety data and processing technology data of feed oil raw material samples of different varieties and different processing technologies to obtain the variety feature vector V s and the processing feature vector V p ;

[0085] Step S220: Normalize the storage environment parameters to obtain a normalized storage environment feature vector V e ;

[0086] Step S230: The variety feature vector V s , processing technology feature vector V p and storage environment feature vector V e Splice into a comprehensive feature vector V c , with the variety feature vector V s , processing technology feature vector V p and storage environment feature vector V e As the first input, the corresponding comprehensive feature vector V c As the first output, the comprehensive feature vector V c Paired with the corresponding time series of physical and chemical indicators to form training samples and construct variety feature vector V s , processing technology feature vector V p and storage environment feature vector V e The nonlinear mapping model between s , v p , v e );

[0087] Step S240: nonlinear mapping model f(v s , v p , v e) is trained to optimize the model parameters of the nonlinear mapping model by minimizing the first loss function to obtain a weight matrix of the nonlinear mapping model and a bias vector of the nonlinear mapping model;

[0088] The nonlinear mapping model is used to calculate the variety feature vector V s , processing technology feature vector V p and storage environment feature vector V e Predict the comprehensive feature vector V c The machine learning model reflects the comprehensive impact of varieties, processing technology and storage environment on physical and chemical indicators;

[0089] The model expression of the nonlinear mapping model is: Among them, W1 is the weight matrix of the nonlinear mapping model, b1 is the bias vector of the nonlinear mapping model, Represents a vector concatenation operation;

[0090] Preferably, the initial model of the nonlinear mapping model is a neural network model;

[0091] The first loss function is the mean square error, and the calculation formula is: Among them, n is the number of training samples, v c,i is the true comprehensive feature vector of the i-th training sample, f(v c,i ) is the predicted value of the comprehensive feature vector of the i-th training sample by the nonlinear mapping model;

[0092] This step creates a comprehensive feature vector containing all necessary features, which helps capture the interactions between different input features;

[0093] Step S250: The time series of the physical and chemical indicators at the previous k moments and the storage environment characteristic vector are used as the second input, and the physical and chemical indicators at the current moment t are used as the second output to construct a dynamic change model g(y t-1 ,y t-2 , ..., y t-k , V e ), optimizing the model parameters of the dynamic change model by minimizing the second loss function, and obtaining the coefficient matrix of the dynamic change model, the environmental feature weight matrix and the bias vector of the dynamic change model;

[0094] The dynamic change model g(y t-1 ,y t-2 , ..., y t-k , V e ) is used to predict the physical and chemical index y at time t based on the physical and chemical indexes at the k moments before time t and the storage environment characteristic vector t The machine learning model reflects the dynamic trend of physical and chemical indicators;

[0095] The model expression of the dynamic change model is: Among them, y t represents the physical and chemical index at time t, y t-1 ,y t-2 , ..., y t-k represents the time series of physical and chemical indicators at the previous k moments, W 2z represents the coefficient matrix of the dynamic change model, W3 represents the environmental characteristic weight matrix, and b2 represents the bias vector of the dynamic change model of physical and chemical indicators;

[0096] Preferably, the initial model of the dynamic change model selects an LSTM model;

[0097] The second loss function is the mean square error, and the calculation formula is: Among them, y t,i is the real physical and chemical index of the i-th training sample at time t, g(X t,i , v e,i ) is the input sequence X of the physical and chemical indicators of the dynamic change model for the i-th training sample t,i and store the environmental feature vector v e,i The predicted value of the physical and chemical index input sequence X t,i is the time series of physical and chemical indicators at the previous k moments;

[0098] This step further understands the influence of time factors on physical and chemical indicators by simulating the changes of physical and chemical indicators over time;

[0099] Step S260: Using the comprehensive feature vector V c , the second output of the dynamic change model and the storage environment feature vector are used as the third input, and the physical and chemical indicators are used as the third output, to construct a relationship mapping model, optimize the model parameters by minimizing the third loss function, and obtain the weight matrix of the relationship mapping model, the bias vector of the relationship mapping model and the trained relationship mapping model;

[0100] The relationship mapping model is based on the comprehensive feature vector V c , the second output of the dynamic change model and the machine learning model for predicting physical and chemical indicators based on the storage environment characteristic vector, which comprehensively considers the impact of varieties, processing technology, storage environment and the dynamic change trend of physical and chemical indicators on physical and chemical indicators, and is the final prediction of physical and chemical indicators;

[0101] The model expression of the relationship mapping model is: , where h(v c ,y t-1 ,y t-2 , ..., y t-k , V e) represents the final relationship mapping model, W4 represents the weight matrix of the relationship mapping model, and b3 represents the bias vector of the relationship mapping model;

[0102] Preferably, the initial model of the relationship mapping model is a neural network model;

[0103] The third loss function is the mean square error, and the calculation formula is: Among them, h(v c,i , g(X t,i , v e,i ), v e,i ) is the physical and chemical index predicted by the relational mapping model based on the third input;

[0104] This step takes into account all relevant factors, including variety characteristics, processing technology, storage conditions and their dynamic changes, by providing a comprehensive tool to predict physicochemical indicators;

[0105] The above steps gradually build and refine the model, so that the final model can accurately predict and understand how various factors jointly affect the physical and chemical indicators of feed oil raw materials;

[0106] Step S300: collecting the variety data, processing technology data, and storage environment parameters of the feed oil raw materials in real time, using the relationship mapping model to predict the real-time physical and chemical indicators of the feed oil raw materials, and calculating the nutritional value reference value of the feed oil raw materials based on the real-time physical and chemical indicators;

[0107] The specific method of collecting the variety data, processing technology data, and storage environment parameters of the feed oil raw materials in real time, predicting the real-time physical and chemical indicators of the feed oil raw materials by using the relationship mapping model, and calculating the reference value of the nutritional value of the feed oil raw materials according to the real-time physical and chemical indicators is as follows:

[0108] Step S310: Real-time collection of feed oil raw material variety data, processing technology data, storage environment parameters, and use of a relationship mapping model to predict real-time physical and chemical indicators of the feed oil raw materials, including but not limited to: real-time acid value Real-time peroxide value Real-time moisture content

[0109] Step S320: Obtain the standard physical and chemical index reference values ​​and standard nutritional values ​​of the corresponding feed oil raw materials from the nutrition database. The standard physical and chemical index reference values ​​include but are not limited to: acid value reference value Y 1 , Peroxide value reference value Y 2 , moisture content reference value Y 3 ;

[0110] Step S330: Preset nutritional value influencing factor α j, using real-time physical and chemical indicators and standard physical and chemical index reference values ​​to calculate the estimated nutritional value of feed oil raw materials;

[0111] The preset nutritional value influencing factor α j The specific method for calculating the estimated value of the nutritional value of feed oil raw materials using real-time physical and chemical indicators and standard physical and chemical index reference values ​​is:

[0112] Step S331: Preset nutritional value influencing factor α j , the influencing factors of nutritional value and real-time physical and chemical indicators are combined to obtain the influencing sequence of physical and chemical indicators in It represents the mth real-time physical and chemical index at time t, where m is the total number of real-time physical and chemical indexes;

[0113] The nutritional value influencing factors refer to the factors affecting the nutritional value by physical and chemical indicators;

[0114] Step S332: Calculate the similarity function based on the real-time physical and chemical indexes and the standard physical and chemical index reference values, as well as the physical and chemical index influence sequence

[0115] The calculation formula of the similarity function is: in, represents the jth real-time physical and chemical index, Y j It represents the reference value of the jth standard physical and chemical index;

[0116] Step S333: Calculate the estimated nutritional value based on the similarity function

[0117] The estimated nutritional value is calculated as follows: Where m is the index number of real-time physical and chemical index or standard physical and chemical index reference value, ω j is the similarity function corresponding to the jth real-time physical and chemical index The weight coefficient of , β is the baseline nutritional value;

[0118] The baseline nutritional value refers to the benchmark value of nutritional value;

[0119] Preferably, m is equal to 3;

[0120] Step S340: Calculate the nutritional value reference value based on the estimated nutritional value, the nutritional value reference value V ref The calculation formula is: Where λ is the proportional coefficient of the estimated nutritional value, (1-λ) is the proportional coefficient of the standard nutritional value, and V std Indicates standard nutritional value, is an estimate of nutritional value;

[0121] The size of the proportionality coefficient λ of the estimated value of the nutritional value is set according to expert experience or experiments;

[0122] This step combines real-time data collection, predictive model application, nutrition database query and nutritional value calculation to provide key information during production and processing. By providing real-time monitoring and prediction, it helps to adjust the production process immediately to ensure product quality, reduce production losses and improve efficiency. By providing a quantitative and comparable nutritional value indicator, it helps consumers understand the nutritional value of the product and provides support for marketing.

[0123] Step S400: Preset a nutritional value threshold, calculate the nutritional value deviation according to the nutritional value reference value, and calculate the warning index at the current moment in combination with the warning index at the previous moment. When the warning index at r consecutive moments is greater than the warning trigger threshold, issue a warning signal;

[0124] The preset nutritional value threshold, the nutritional value deviation is calculated according to the nutritional value reference value, and the warning index at the current moment is calculated in combination with the warning index at the previous moment. When the warning index at r consecutive moments is greater than the warning trigger threshold, the specific method of issuing the warning signal is:

[0125] Step S410: Collect the nutritional value reference values ​​of feed oil raw materials with different storage time, different varieties, different storage environments and different processing technologies, and calculate the average of all nutritional value reference values and standard deviation δ, the nutritional value threshold is calculated;

[0126] The nutritional value threshold includes an upper threshold and a lower threshold, and the specific calculation method is:

[0127] The upper threshold of the nutritional value threshold is:

[0128] The lower limit of the nutritional value threshold is:

[0129] Step S420: For a given feed oil raw material, calculate its nutritional value deviation d, and the calculation formula of the nutritional value deviation d is:

[0130] The nutritional value deviation d represents the degree of deviation between the nutritional value reference value of a given feed oil raw material and the nutritional value threshold value;

[0131] Step S430: Calculate the warning index u t , the calculation formula of the warning index is: t =ε×u t-1+(1-ε)×|d t |, where u t represents the warning index at the current time t, u t-1 represents the warning index at the previous moment, ε is the time decay factor, d t Indicates the deviation of nutritional value at the current time t;

[0132] The time decay factor ε∈(0, 1] represents the speed at which the warning index of historical time decays over time;

[0133] Step S440: When the warning index for r consecutive moments is greater than the preset warning trigger threshold U, a warning signal is sent to the feed storage management personnel to remind the feed storage management personnel to take corresponding quality control measures;

[0134] The warning triggering threshold U is determined by those skilled in the art according to specific application requirements;

[0135] The above steps ensure the quality control and risk management of feed, which is an important means to maintain product quality and safety. Establishing thresholds allows the system to identify and warn of situations that deviate from the normal nutritional value range, thereby preventing unqualified products from entering the market and ensuring animal health and food safety. The calculation of the degree of deviation provides a direct basis for subsequent warnings, which helps to timely discover product quality problems, reduce economic losses and potential health risks, and timely warning signals can prompt warehouse management personnel to respond quickly, thereby maintaining high product quality standards;

[0136] Step S500: separating the feed and fat raw materials that have issued warning signals as abnormal feed and fat raw materials from normal batches, analyzing the causes of the warnings, and performing secondary processing on the abnormal feed and fat raw materials according to the causes of the warnings;

[0137] The specific method of separating the feed oil raw materials that have issued warning signals as abnormal feed oil raw materials from normal batches, analyzing the warning causes, and performing secondary processing on the abnormal feed oil raw materials according to the warning causes is as follows:

[0138] Step S510: Mark the feed oil raw materials that have issued warning signals as abnormal feed oil raw materials, separate them from normal batches, and perform quality inspections on them, including but not limited to: physical and chemical index analysis, pollutant detection and microbial detection;

[0139] Step S520: Analyze the quality inspection results to obtain the causes of the warning signal issued by the abnormal feed oil raw materials, including but not limited to: moisture, microbial contamination, impurity contamination;

[0140] Step S530: performing secondary processing on the abnormal feed oil raw material according to the cause of the warning, wherein the secondary processing method includes but is not limited to: re-refining, filtering, adding additives, and changing the storage environment;

[0141] Step S540: for the abnormal feed oil raw material after secondary processing, predict its physical and chemical indicators again, calculate the nutritional value reference value and the early warning index, and determine whether the early warning index at this time is greater than the early warning trigger threshold;

[0142] Step S550: when the warning index at this time is greater than the warning trigger threshold, the abnormal feed and fat raw material is determined to be a batch whose quality cannot be restored, and other qualified feed and fat raw materials are selected to replace it to maintain the operation of the production line. When the warning index at this time is less than or equal to the warning trigger threshold, the storage environment of the abnormal feed and fat raw material after secondary processing is changed;

[0143] Although it is impossible to completely restore the feed oil raw materials that have experienced quality decline to their original state, there are still some ways to try to improve their quality. By first conducting a comprehensive quality test on the feed oil raw materials, the appropriate treatment method can be determined;

[0144] The above steps construct a comprehensive processing flow for processing and improving feed oil raw materials that have issued warning signals. It not only helps to restore or improve the quality of abnormal raw materials, but also ensures the stable operation of the production line and the quality and safety of the products.

[0145] Example 2

[0146] like Figure 3 As shown, a feed oil raw material quality prediction system based on physical and chemical indicators provided by this application includes:

[0147] The oil and fat raw material data collection module is used to collect the variety data, processing technology data and storage environment parameters of different feed oil and fat raw material samples, and obtain the physical and chemical indicators of the feed oil and fat raw material samples;

[0148] A relationship mapping model building module is used to obtain a relationship mapping model between the physical and chemical indicators of feed oil raw material samples and different varieties, processing technologies and storage environments using variety data, processing technology data, storage environment parameters and physical and chemical indicators;

[0149] The raw material nutritional value calculation module is used to collect the variety data, processing technology data, and storage environment parameters of feed oil raw materials in real time, use the relationship mapping model to predict the real-time physical and chemical indicators of feed oil raw materials, and calculate the nutritional value reference value of feed oil raw materials based on the real-time physical and chemical indicators;

[0150] The early warning calculation and triggering module is used to preset the nutritional value threshold, calculate the nutritional value deviation according to the nutritional value reference value, and calculate the early warning index at the current moment in combination with the early warning index at the previous moment. When the early warning index at r consecutive moments is greater than the early warning triggering threshold, an early warning signal is issued;

[0151] The cause analysis and secondary processing module is used to separate the feed and oil raw materials that have issued warning signals as abnormal feed and oil raw materials from normal batches, analyze their warning causes, and perform secondary processing on the abnormal feed and oil raw materials based on the warning causes.

[0152] Example 3

[0153] Figure 4 1 is a schematic diagram of the structure of an electronic device provided by an embodiment of the present application. Figure 4 As shown, according to another aspect of the present application, an electronic device is also provided. The electronic device may include one or more processors and one or more memories. The memories store computer readable codes, and when the computer readable codes are executed by one or more processors, the feed oil raw material quality prediction method based on physical and chemical indicators as described above can be executed.

[0154] The method or system according to the embodiment of the present application can also be Figure 4 The electronic device architecture shown in FIG. Figure 4As shown, the electronic device may include a bus, one or more CPUs, a read-only memory (ROM), a random access memory (RAM), a communication port connected to a network, an input / output component, a hard disk, etc. A storage device in the electronic device, such as a ROM or a hard disk, may store a method for predicting the quality of feed oil raw materials based on physical and chemical indicators provided in the present application. A method for predicting the quality of feed and oil raw materials based on physical and chemical indicators may, for example, include: collecting variety data, processing technology data and storage environment parameters of different feed and oil raw material samples, and obtaining the physical and chemical indicators of the feed and oil raw material samples; using variety data, processing technology data, storage environment parameters and physical and chemical indicators to obtain a relationship mapping model between the physical and chemical indicators of the feed and oil raw material samples and different varieties, processing technologies and storage environments; collecting variety data, processing technology data and storage environment parameters of feed and oil raw materials in real time, using the relationship mapping model to predict the real-time physical and chemical indicators of the feed and oil raw materials, and calculating the nutritional value reference value of the feed and oil raw materials based on the real-time physical and chemical indicators; presetting a nutritional value threshold, calculating the nutritional value deviation based on the nutritional value reference value, and calculating the warning index at the current moment in combination with the warning index at the previous moment, and issuing a warning signal when the warning index for r consecutive moments is greater than the warning trigger threshold; separating the feed and oil raw materials that have issued warning signals as abnormal feed and oil raw materials from normal batches, analyzing their warning causes, and performing secondary processing on the abnormal feed and oil raw materials based on the warning causes. Furthermore, the electronic device may also include a user interface. Of course, Figure 4 The architecture shown is only exemplary and can be omitted according to actual needs when implementing different devices. Figure 4 One or more components of an electronic device are shown.

[0155] Example 4

[0156] Figure 5 Schematic diagram of a readable storage medium structure provided by an embodiment of the present application. Figure 5 As shown, it is a readable storage medium according to an embodiment of the present application. Computer-readable instructions are stored on the computer-readable storage medium. When the computer-readable instructions are executed by the processor, a method for predicting the quality of feed oil raw materials based on physical and chemical indicators according to an embodiment of the present application described with reference to the above figures can be executed. The storage medium includes, but is not limited to, for example, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and cache memory (cache), etc. Non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc.

[0157] In addition, according to the implementation mode of the present application, the process described in the above reference flowchart can be implemented as a computer software program. For example, the present application provides a non-transitory machine-readable storage medium, the non-transitory machine-readable storage medium stores machine-readable instructions, the machine-readable instructions can be executed by a processor to execute instructions corresponding to the method steps provided in the present application, for example: collecting variety data, processing technology data and storage environment parameters of different feed and oil raw material samples, and obtaining the physical and chemical indicators of the feed and oil raw material samples; using the variety data, processing technology data, storage environment parameters and physical and chemical indicators to obtain a relationship mapping model between the physical and chemical indicators of the feed and oil raw material samples and different varieties, processing technologies and storage environments; collecting the feed and oil raw materials in real time. Variety data, processing technology data, storage environment parameters, use the relationship mapping model to predict the real-time physical and chemical indicators of feed oil raw materials, calculate the nutritional value reference value of feed oil raw materials according to the real-time physical and chemical indicators; preset the nutritional value threshold, calculate the nutritional value deviation according to the nutritional value reference value, combine the warning index of the previous moment, calculate the warning index of the current moment, and when the warning index of r consecutive moments is greater than the warning trigger threshold, issue a warning signal; separate the feed oil raw materials that have issued warning signals as abnormal feed oil raw materials from normal batches, analyze their warning causes, and perform secondary processing on the abnormal feed oil raw materials according to the warning causes. When the computer program is executed by the central processing unit (CPU), the above functions defined in the method of this application are executed.

[0158] The methods, apparatuses, and devices of the present application may be implemented in many ways. For example, the methods, apparatuses, and devices of the present application may be implemented by software, hardware, firmware, or any combination of software, hardware, and firmware. The above order of steps for the method is for illustration only, and the steps of the method of the present application are not limited to the order specifically described above unless otherwise specified. In addition, in some embodiments, the present application may also be implemented as programs recorded in a recording medium, which include machine-readable instructions for implementing the method according to the present application. Therefore, the present application also covers recording media storing programs for executing the method according to the present application.

[0159] In addition, the parts of the above-mentioned technical solutions provided in the embodiments of the present application that are consistent with the implementation principles of the corresponding technical solutions in the prior art are not described in detail to avoid excessive redundancy.

[0160] The specific implementation modes as described above further describe the purpose, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation mode of the present invention and is not intended to limit the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for predicting the quality of feed oil raw materials based on physical and chemical indicators, characterized in that: include: Collect the variety data, processing technology data and storage environment parameters of different feed oil raw material samples, and obtain the physical and chemical indicators of feed oil raw material samples; Using variety data, processing technology data, storage environment parameters and physical and chemical indicators, a relationship mapping model between the physical and chemical indicators of feed oil raw material samples and different varieties, processing technologies and storage environments is obtained; Collect the variety data, processing technology data, and storage environment parameters of feed oil raw materials in real time, use the relationship mapping model to predict the real-time physical and chemical indicators of feed oil raw materials, and calculate the reference value of the nutritional value of feed oil raw materials based on the real-time physical and chemical indicators; Preset the nutritional value threshold, calculate the nutritional value deviation according to the nutritional value reference value, and calculate the warning index at the current moment in combination with the warning index at the previous moment. When the warning index at r consecutive moments is greater than the warning trigger threshold, issue a warning signal; Separate the feed and oil raw materials that have issued warning signals as abnormal feed and oil raw materials from normal batches, analyze the causes of the warnings, and perform secondary processing on the abnormal feed and oil raw materials based on the causes of the warnings; The specific method of using variety data, processing technology data, storage environment parameters and physical and chemical indicators to obtain the relationship mapping model between the physical and chemical indicators of feed oil raw material samples and different varieties, processing technologies and storage environments is: The variety data and processing technology data of feed oil raw material samples of different varieties and different processing technologies are vectorized to obtain the variety feature vector v s and the processing feature vector v p ; Normalize the storage environment parameters to obtain the normalized storage environment feature vector v e ; The variety feature vector v s , processing technology feature vector v p and store the environmental feature vector v e Splice into a comprehensive feature vector v c , with the variety feature vector v s , processing technology feature vector v p and store the environmental feature vector v e As the first input, the corresponding comprehensive feature vector v c As the first output, the comprehensive feature vector v c Paired with the corresponding time series of physical and chemical indicators to form training samples and construct variety feature vector v s , processing technology feature vector v p and store the environmental feature vector v e The nonlinear mapping model between s ,v p ,v e ); For the nonlinear mapping model f(v s ,v p ,v e ) is trained to optimize the model parameters of the nonlinear mapping model by minimizing the first loss function to obtain a weight matrix of the nonlinear mapping model and a bias vector of the nonlinear mapping model; The model expression of the nonlinear mapping model is: Among them, W1 is the weight matrix of the nonlinear mapping model, b1 is the bias vector of the nonlinear mapping model, Represents a vector concatenation operation; The first loss function is the mean square error, and the calculation formula is: Among them, n is the number of training samples, v c,i is the true comprehensive feature vector of the i-th training sample, f(v c,i ) is the predicted value of the comprehensive feature vector of the i-th training sample by the nonlinear mapping model; The time series of the physical and chemical indicators at the previous k moments and the storage environment characteristic vector are used as the second input, and the physical and chemical indicators at the current moment t are used as the second output to construct a dynamic change model g(y t-1 ,y t-2 ,...,y t-k ,v e ), optimizing the model parameters of the dynamic change model by minimizing the second loss function, and obtaining the coefficient matrix of the dynamic change model, the environmental feature weight matrix and the bias vector of the dynamic change model; The model expression of the dynamic change model is: Among them, y t represents the physical and chemical index at time t, y t-1 ,y t-2 ,...,y t-k represents the time series of physical and chemical indicators at the previous k moments, W 2z represents the coefficient matrix of the dynamic change model, W3 represents the environmental characteristic weight matrix, and b2 represents the bias vector of the dynamic change model of physical and chemical indicators; The second loss function is the mean square error, and the calculation formula is: Among them, y t,i is the real physical and chemical index of the i-th training sample at time t, g(X t,i ,v e,i ) is the input sequence X of the physical and chemical indicators of the dynamic change model for the i-th training sample t,i and store the environmental feature vector v e,i The predicted value of the physical and chemical index input sequence X t,i is the time series of physical and chemical indicators at the previous k moments; Take the comprehensive feature vector v c , the second output of the dynamic change model and the storage environment feature vector are used as the third input, and the physical and chemical indicators are used as the third output, to construct a relationship mapping model, optimize the model parameters by minimizing the third loss function, and obtain the weight matrix of the relationship mapping model, the bias vector of the relationship mapping model and the trained relationship mapping model; The model expression of the relationship mapping model is: Among them, h(vc,yt-1,yt-2,...,yt-k,ve) represents the final relationship mapping model, W4 represents the weight matrix of the relationship mapping model, and b3 represents the bias vector of the relationship mapping model; The third loss function is the mean square error, and the calculation formula is: Among them, h(v c,i ,g(X t,i ,v e,i ),v e,i ) is the physical and chemical index predicted by the relational mapping model based on the third input; The specific method of collecting the variety data, processing technology data, and storage environment parameters of the feed oil raw materials in real time, predicting the real-time physical and chemical indicators of the feed oil raw materials by using the relationship mapping model, and calculating the reference value of the nutritional value of the feed oil raw materials according to the real-time physical and chemical indicators is as follows: Real-time collection of feed oil raw material variety data, processing technology data, storage environment parameters, and use of relationship mapping models to predict the real-time physical and chemical indicators of feed oil raw materials. The real-time physical and chemical indicators include: real-time acid value Real-time peroxide value Real-time moisture content The standard physical and chemical index reference values ​​and standard nutritional values ​​of the corresponding feed oil raw materials are obtained from the nutrition database. The standard physical and chemical index reference values ​​include: acid value reference value Y 1 , Peroxide value reference value Y 2 , moisture content reference value Y 3 ; Preset nutritional value influencing factor α j , using real-time physical and chemical indicators and standard physical and chemical index reference values ​​to calculate the estimated nutritional value of feed oil raw materials; The nutritional value reference value is calculated based on the estimated nutritional value, and the nutritional value reference value V ref The calculation formula is: Where λ is the proportional coefficient of the estimated nutritional value, (1-λ) is the proportional coefficient of the standard nutritional value, and V std Indicates standard nutritional value, The nutritional value is an estimate.

2. A method for predicting the quality of feed oil raw materials based on physical and chemical indicators as claimed in claim 1, characterized in that: The specific method for collecting the variety data, processing technology data and storage environment parameters of different feed oil raw material samples and obtaining the physical and chemical indicators of the feed oil raw material samples is: Collect feed oil raw material samples covering different varieties, different processing technologies, and different storage environments, and obtain variety data, processing technology data, and storage environment parameters corresponding to each feed oil raw material sample; The components of the feed oil raw material samples are analyzed by chemical analysis to obtain the physical and chemical indicators of the feed oil raw materials, which include acid value, peroxide value and water content.

3. A method for predicting the quality of feed oil raw materials based on physical and chemical indicators as claimed in claim 2, characterized in that: The preset nutritional value influencing factor α j The specific method for calculating the estimated value of the nutritional value of feed oil raw materials using real-time physical and chemical indicators and standard physical and chemical index reference values ​​is: Preset nutritional value influencing factor α j , the influencing factors of nutritional value and real-time physical and chemical indicators are combined to obtain the influencing sequence of physical and chemical indicators in It represents the mth real-time physical and chemical index at time t, where m is the total number of real-time physical and chemical indexes; Calculate the similarity function based on real-time physical and chemical indicators, standard physical and chemical indicator reference values, and physical and chemical indicator influence sequence The calculation formula of the similarity function is: in, represents the jth real-time physical and chemical index, Y j It represents the reference value of the physical and chemical index of the jth standard; The estimated nutritional value is calculated based on the similarity function The estimated nutritional value is calculated as follows: Where m is the index number of real-time physical and chemical index or standard physical and chemical index reference value, ω j is the similarity function corresponding to the jth real-time physical and chemical index The weight coefficient is , and β is the baseline nutritional value.

4. A method for predicting the quality of feed oil raw materials based on physical and chemical indicators as claimed in claim 3, characterized in that: The preset nutritional value threshold, the nutritional value deviation is calculated according to the nutritional value reference value, and the warning index at the current moment is calculated in combination with the warning index at the previous moment. When the warning index at r consecutive moments is greater than the warning trigger threshold, the specific method of issuing the warning signal is: Collect the nutritional value reference values ​​of feed oil raw materials with different storage time, different varieties, different storage environments and different processing technologies, and calculate the mean of all nutritional value reference values and standard deviation δ, the nutritional value threshold is calculated; The nutritional value threshold includes an upper threshold and a lower threshold, and the specific calculation method is: The upper threshold of the nutritional value threshold is: The lower limit of the nutritional value threshold is: For a given feed oil raw material, calculate its nutritional value deviation d, the calculation formula of the nutritional value deviation d is: Calculate the early warning index u t , the calculation formula of the warning index is: t =ε×u t-1 +(1-ε)×|d t |, where u t represents the warning index at the current time t, u t-1 represents the warning index at the previous moment, ε is the time decay factor, d t Indicates the deviation of nutritional value at the current time t; When the warning index for r consecutive moments is greater than the preset warning trigger threshold U, a warning signal is sent to the feed storage management personnel to remind them to take corresponding quality control measures.

5. A method for predicting the quality of feed oil raw materials based on physical and chemical indicators as claimed in claim 4, characterized in that: The specific method of separating the feed oil raw materials that have issued warning signals as abnormal feed oil raw materials from normal batches, analyzing the warning causes, and performing secondary processing on the abnormal feed oil raw materials according to the warning causes is as follows: Feed oil and fat raw materials that have issued warning signals will be marked as abnormal feed oil and fat raw materials, separated from normal batches, and subjected to quality testing, including: physical and chemical index analysis, contaminant detection and microbial detection; The results of quality inspection are analyzed to obtain the causes of early warning signals of abnormal feed oil raw materials, including: moisture, microbial contamination, impurity contamination; Perform secondary processing on abnormal feed oil raw materials according to the cause of the warning, wherein the secondary processing method includes: re-refining, filtering, adding additives, and changing the storage environment; For abnormal feed oil raw materials after secondary processing, their physical and chemical indicators are predicted again, and the nutritional value reference value and early warning index are calculated to determine whether the early warning index at this time is greater than the early warning trigger threshold; When the warning index at this time is greater than the warning trigger threshold, the abnormal feed and fat raw material is determined to be a batch whose quality cannot be restored, and other qualified feed and fat raw materials are selected to replace it to maintain the operation of the production line. When the warning index at this time is less than or equal to the warning trigger threshold, the storage environment of the abnormal feed and fat raw material after secondary processing is changed.

6. A feed oil raw material quality prediction system based on physical and chemical indicators, characterized in that: include: The oil and fat raw material data collection module is used to collect the variety data, processing technology data and storage environment parameters of different feed oil and fat raw material samples, and obtain the physical and chemical indicators of the feed oil and fat raw material samples; A relationship mapping model building module is used to obtain a relationship mapping model between the physical and chemical indicators of feed oil raw material samples and different varieties, processing technologies and storage environments by using variety data, processing technology data, storage environment parameters and physical and chemical indicators; The raw material nutritional value calculation module is used to collect the variety data, processing technology data, and storage environment parameters of feed oil raw materials in real time, use the relationship mapping model to predict the real-time physical and chemical indicators of feed oil raw materials, and calculate the nutritional value reference value of feed oil raw materials based on the real-time physical and chemical indicators; The early warning calculation and triggering module is used to preset the nutritional value threshold, calculate the nutritional value deviation according to the nutritional value reference value, and calculate the early warning index at the current moment in combination with the early warning index at the previous moment. When the early warning index at r consecutive moments is greater than the early warning triggering threshold, an early warning signal is issued; Cause analysis and secondary processing module, used to separate feed oil raw materials that have issued warning signals as abnormal feed oil raw materials from normal batches, analyze their warning causes, and perform secondary processing on the abnormal feed oil raw materials according to the warning causes; The specific method of using variety data, processing technology data, storage environment parameters and physical and chemical indicators to obtain the relationship mapping model between the physical and chemical indicators of feed oil raw material samples and different varieties, processing technologies and storage environments is: The variety data and processing technology data of feed oil raw material samples of different varieties and different processing technologies are vectorized to obtain the variety feature vector v s and the processing feature vector v p ; Normalize the storage environment parameters to obtain the normalized storage environment feature vector v e ; The variety feature vector v s , processing technology feature vector v p and store the environmental feature vector v e Splice into a comprehensive feature vector v c , with the variety feature vector v s , processing technology feature vector v p and store the environmental feature vector v e As the first input, the corresponding comprehensive feature vector v c As the first output, the comprehensive feature vector v c Paired with the corresponding time series of physical and chemical indicators to form training samples and construct variety feature vector v s , processing technology feature vector v p and store the environmental feature vector v e The nonlinear mapping model between s ,v p ,v e ); For the nonlinear mapping model f(v s ,v p ,v e ) is trained to optimize the model parameters of the nonlinear mapping model by minimizing the first loss function to obtain a weight matrix of the nonlinear mapping model and a bias vector of the nonlinear mapping model; The model expression of the nonlinear mapping model is: Among them, W1 is the weight matrix of the nonlinear mapping model, b1 is the bias vector of the nonlinear mapping model, Represents a vector concatenation operation; The first loss function is the mean square error, and the calculation formula is: Among them, n is the number of training samples, v c,i is the true comprehensive feature vector of the i-th training sample, f(v c,i ) is the predicted value of the comprehensive feature vector of the i-th training sample by the nonlinear mapping model; The time series of the physical and chemical indicators at the previous k moments and the storage environment characteristic vector are used as the second input, and the physical and chemical indicators at the current moment t are used as the second output to construct a dynamic change model g(y t-1 ,y t-2 ,...,y t-k ,v e ), optimizing the model parameters of the dynamic change model by minimizing the second loss function, and obtaining the coefficient matrix of the dynamic change model, the environmental feature weight matrix and the bias vector of the dynamic change model; The model expression of the dynamic change model is: Among them, y t represents the physical and chemical index at time t, y t-1 ,y t-2 ,...,y t-k represents the time series of physical and chemical indicators at the previous k moments, W 2z represents the coefficient matrix of the dynamic change model, W3 represents the environmental characteristic weight matrix, and b2 represents the bias vector of the dynamic change model of physical and chemical indicators; The second loss function is the mean square error, and the calculation formula is: Among them, y t,i is the real physical and chemical index of the i-th training sample at time t, g(X t,i ,v e,i ) is the input sequence X of the physical and chemical indicators of the dynamic change model for the i-th training sample t,i and store the environmental feature vector v e,i The predicted value of the physical and chemical index input sequence X t,i is the time series of physical and chemical indicators at the previous k moments; Take the comprehensive feature vector v c , the second output of the dynamic change model and the storage environment feature vector are used as the third input, and the physical and chemical indicators are used as the third output, to construct a relationship mapping model, optimize the model parameters by minimizing the third loss function, and obtain the weight matrix of the relationship mapping model, the bias vector of the relationship mapping model and the trained relationship mapping model; The model expression of the relationship mapping model is: Among them, h(vc,yt-1,yt-2,...,yt-k,ve) represents the final relationship mapping model, W4 represents the weight matrix of the relationship mapping model, and b3 represents the bias vector of the relationship mapping model; The third loss function is the mean square error, and the calculation formula is: Among them, h(v c,i ,g(X t,i ,v e,i ),v e,i ) is the physical and chemical index predicted by the relational mapping model based on the third input; The specific method of collecting the variety data, processing technology data, and storage environment parameters of the feed oil raw materials in real time, predicting the real-time physical and chemical indicators of the feed oil raw materials by using the relationship mapping model, and calculating the reference value of the nutritional value of the feed oil raw materials according to the real-time physical and chemical indicators is as follows: Real-time collection of feed oil raw material variety data, processing technology data, storage environment parameters, and use of relationship mapping models to predict the real-time physical and chemical indicators of feed oil raw materials. The real-time physical and chemical indicators include: real-time acid value Real-time peroxide value Real-time moisture content The standard physical and chemical index reference values ​​and standard nutritional values ​​of the corresponding feed oil raw materials are obtained from the nutrition database. The standard physical and chemical index reference values ​​include: acid value reference value Y 1 , Peroxide value reference value Y 2 , moisture content reference value Y 3 ; Preset nutritional value influencing factor α j , using real-time physical and chemical indicators and standard physical and chemical index reference values ​​to calculate the estimated nutritional value of feed oil raw materials; The nutritional value reference value is calculated based on the estimated nutritional value, and the nutritional value reference value V ref The calculation formula is: Where λ is the proportional coefficient of the estimated nutritional value, (1-λ) is the proportional coefficient of the standard nutritional value, and V std Indicates standard nutritional value, The nutritional value is an estimate.

7. An electronic device, characterized in that: The invention comprises a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the program, the steps in the method for predicting the quality of feed oil raw materials based on physical and chemical indicators as described in any one of claims 1 to 5 are implemented.

8. A readable storage medium, characterized in that: The readable storage medium stores a computer program, and the computer program is suitable for being loaded by a processor to execute the steps in the method for predicting the quality of feed oil raw materials based on physical and chemical indicators as described in any one of claims 1 to 5.