Strawberry mineral substance detection-based nutritional ingredient analysis method and system

Abstract

The invention relates to the technical field of food detection and nutrition analysis, in particular to a nutrient component analysis method and system based on strawberry mineral detection, and the method comprises the steps: obtaining a strawberry sample mineral content data set, and carrying out preliminary component distribution detection, key element recognition, synergistic relationship calculation and multi-dimensional feature fusion. And a nutritional ingredient evaluation coefficient is obtained by using a pre-trained nutritional ingredient analysis model. Through precise mineral substance detection and dynamic fertilization regulation and control, the problem of lack of real-time mineral substance detection and dynamic fertilization optimization in the prior art is solved, efficient analysis and evaluation of nutritional ingredients of strawberries are realized, and the method has a relatively high agricultural application value.

Description

Technical Field

[0001] The present invention belongs to the technical field of food detection and nutritional analysis, and specifically relates to a nutritional component analysis method and system based on strawberry mineral detection. Background Art

[0002] With the continuous advancement of strawberry cultivation and processing technologies, the analysis and optimization of its nutritional composition has become a research hotspot. In particular, in the areas of improving strawberry quality, precision fertilization, and deep processing, nutrient analysis methods and systems based on mineral detection can provide an important basis for scientific decision-making. However, existing strawberry-related technical solutions still have certain limitations in mineral detection and nutrient analysis, making it difficult to fully meet the requirements for efficiency and accuracy.

[0003] A search revealed that patent publication number CN103069991B (published on January 29, 2014) proposes a pollution-free strawberry cultivation method that improves yield and quality by managing fertilization and pesticide use according to different growth stages. However, this technical solution primarily focuses on fertilization and pesticide application strategies throughout the strawberry's overall growth process and does not involve specific testing and analysis of the strawberry's mineral content. This fertilization strategy relies on empirical formulas and lacks the ability to dynamically adjust based on mineral testing data, potentially leading to inefficient fertilization or deficiencies in certain key nutrients. Furthermore, patent publication number CN110304948B (published on July 19, 2024) proposes a strawberry cultivation fertilization method that improves the absorption and utilization of trace elements (such as zinc, boron, and molybdenum) by designing the composition and state of various fertilizers. Although this technical solution emphasizes the supplementation of trace elements, its fertilization plan still relies primarily on preset formulas and fails to incorporate real-time mineral testing results for precise regulation. At the same time, this method lacks in-depth analysis of the correlation between strawberry mineral content and nutritional components, which may affect the maximization of fertilization effects.

[0004] The above problems indicate that existing strawberry-related technical solutions still have significant deficiencies in combining mineral detection with nutrient analysis, primarily manifested in the following aspects: First, there is a lack of real-time detection methods for strawberry mineral content, making dynamic monitoring difficult; second, the failure to dynamically combine mineral detection results with fertilization or processing techniques leads to inaccurate fertilization and processing processes; and third, insufficient analysis of the correlation between strawberry minerals and nutrients limits the scientific nature and effectiveness of nutrient optimization. Therefore, there is an urgent need for a nutrient analysis method and system based on strawberry mineral detection. By combining high-precision mineral detection technology with data analysis models, accurate assessment and optimization of strawberry nutrients can be achieved, thereby providing scientific support for strawberry cultivation, processing, and quality improvement. Summary of the Invention

[0005] The present invention provides a nutrient analysis method and system based on strawberry mineral detection, the main purpose of which is to solve the problem of the lack of real-time detection of strawberry mineral content and dynamic fertilization control in the prior art. To achieve the above purpose, the present invention provides a nutrient analysis method based on strawberry mineral detection, comprising: obtaining a mineral content dataset from a strawberry sample, performing a preliminary component distribution test on the mineral content dataset to obtain a preliminary mineral distribution test result; identifying key elements on the preliminary mineral distribution test result to obtain a mineral distribution feature; calculating the synergistic relationship between different minerals in the strawberry sample to obtain a synergistic relationship dataset; extracting key information from the synergistic relationship dataset to obtain a synergistic feature; performing multidimensional feature fusion on the synergistic feature and the mineral distribution feature to obtain a comprehensive nutrient feature; and using a pre-trained nutrient analysis model to perform nutrient analysis on the strawberry sample based on the comprehensive nutrient feature to obtain a nutrient evaluation coefficient.

[0006] Optionally, performing preliminary component distribution detection on the mineral content dataset to obtain preliminary mineral distribution detection results includes: performing abnormal fluctuation detection on the mineral content dataset to obtain abnormal fluctuation detection results; performing missing value filling processing on the mineral content dataset to obtain missing value filling results; and performing data integration on the abnormal fluctuation detection results and the missing value filling results to obtain preliminary mineral distribution detection results.

[0007] Optionally, performing abnormal fluctuation detection on the mineral content data set to obtain an abnormal fluctuation detection result includes: calculating a trend change curve of the mineral content data set; determining a normal fluctuation range based on the trend change curve; and performing abnormal screening on the mineral content data set based on the normal fluctuation range to obtain an abnormal fluctuation detection result.

[0008] Optionally, the key element identification of the preliminary mineral distribution detection results to obtain mineral distribution characteristics includes: segmenting the preliminary mineral distribution detection results according to preset concentration intervals; counting the number of samples in each concentration interval to obtain sample statistical results; constructing a distribution graph according to the sample statistical results with the horizontal axis being the concentration interval and the vertical axis being the sample number; converting the preliminary mineral distribution detection results into a matrix form according to the data form of the distribution graph to obtain a mineral distribution matrix; and performing dimensional optimization processing on the mineral distribution matrix to obtain mineral distribution characteristics.

[0009] Optionally, extracting key information from the synergistic relationship data set to obtain synergistic features includes: analyzing key association points of the synergistic relationship data set; performing weight assignment on the synergistic relationship data set to obtain weight parameters; converting the key association points into vector form according to the weight parameters to obtain a synergistic vector; and performing dimensionality reduction processing on the synergistic vector using principal component analysis to obtain synergistic features.

[0010] Optionally, the synergistic effect characteristics and the mineral distribution characteristics are subjected to multidimensional feature fusion to obtain comprehensive nutritional characteristics, including: converting the synergistic effect characteristics and the mineral distribution characteristics into multidimensional feature space representation to obtain a synergistic effect feature space and a mineral distribution feature space; using a feature mapping algorithm to decompose the synergistic effect feature space and the mineral distribution feature space to obtain a synergistic low-dimensional subspace and a mineral distribution low-dimensional subspace; using a feature weighted fusion algorithm to integrate the synergistic low-dimensional subspace and the mineral distribution low-dimensional subspace to obtain comprehensive nutritional characteristics.

[0011] Optionally, before the nutrient component analysis model is used to perform nutrient component analysis on the strawberry sample according to the comprehensive nutrient component characteristics to obtain the nutrient component evaluation coefficient, the method further comprises: obtaining a large number of public strawberry sample data sets with nutrient component annotations, calculating the synergistic relationship between different minerals in the public strawberry sample data sets, and obtaining a public synergistic relationship data set; performing feature extraction on the public synergistic relationship data set to obtain a public synergistic feature, extracting the mineral distribution feature of the public strawberry sample data set, performing multidimensional feature fusion on the mineral distribution feature and the public synergistic feature, and obtaining a public sample feature; dividing the public sample features into training samples and test samples, performing nutrient component analysis on the training samples using a machine learning model, and obtaining a training sample. training nutrient evaluation coefficient; calculating the error value between the training nutrient evaluation coefficient and the nutrient evaluation coefficient pre-labeled for the training sample; if the error value is greater than or equal to a preset error threshold, adjusting the parameters of the machine learning model and re-executing the above-mentioned step of performing nutrient analysis on the training sample using the machine learning model to obtain the training nutrient evaluation coefficient; if the error value is less than the error threshold, performing a model quality assessment on the machine learning model using the test sample; if the machine learning model fails the model quality assessment, reducing the complexity of the machine learning model and re-executing the above-mentioned step of performing model quality assessment on the machine learning model using the test sample; if the machine learning model passes the model quality assessment, confirming that the model training is complete, and obtaining a nutrient analysis model.

[0012] Optionally, the method of using a pre-trained nutrient analysis model to perform nutrient analysis on a strawberry sample according to the comprehensive nutrient characteristics to obtain a nutrient evaluation coefficient includes: using the nutrient analysis model to perform feature recombination processing on the comprehensive nutrient characteristics to obtain a feature set; using the nutrient analysis model to perform feature enhancement processing on the feature set through a feature enhancement layer to obtain enhanced features; using the nutrient analysis model to perform key feature extraction on the enhanced features through a feature screening layer to obtain screening features; and using the nutrient analysis model to perform nutrient analysis on the strawberry sample according to the screening features through a regression analysis module to obtain a nutrient evaluation coefficient.

[0013] In order to solve the above problems, the present invention also provides a nutritional component analysis system based on strawberry mineral detection, which includes: a component detection module, which obtains a mineral content data set in a strawberry sample, performs a preliminary component distribution detection on the mineral content data set, and obtains a preliminary mineral distribution detection result; an element identification module, which performs key element identification on the preliminary mineral distribution detection result to obtain a mineral distribution characteristic; a synergy analysis module, which calculates the synergy relationship between different minerals in the strawberry sample to obtain a synergy relationship data set; a feature fusion module, which extracts key information from the synergy relationship data set to obtain a synergy characteristic, and performs multidimensional feature fusion on the synergy characteristic and the mineral distribution characteristic to obtain a comprehensive nutritional component characteristic; and a nutritional analysis module, which uses a pre-trained nutritional component analysis model to perform nutritional component analysis on the strawberry sample according to the comprehensive nutritional component characteristic to obtain a nutritional component evaluation coefficient.

[0014] The embodiment of the present invention obtains a mineral content dataset from a strawberry sample, performs a preliminary component distribution test on the mineral content dataset, obtains a preliminary mineral distribution test result, provides basic data for subsequent nutrient analysis, performs key element identification on the preliminary mineral distribution test result, obtains a mineral distribution feature, improves the interpretation ability of the mineral data, calculates the synergistic relationship between different minerals in the strawberry sample, obtains a synergistic relationship dataset, extracts key information from the synergistic relationship dataset, obtains a synergistic feature, performs multidimensional feature fusion on the synergistic feature and the mineral distribution feature, obtains a comprehensive nutrient feature, enhances the feature expression ability, and uses a pre-trained nutrient analysis model to perform nutrient analysis on the strawberry sample based on the comprehensive nutrient feature to obtain a nutrient evaluation coefficient. Therefore, the nutrient analysis method and system based on strawberry mineral detection proposed in the present invention can solve the problem of lack of accurate mineral detection and dynamic fertilization control in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 A schematic flow chart of a method for analyzing nutrient components in strawberries based on mineral detection according to one embodiment of the present invention; Figure 2 A schematic diagram of a process flow for abnormal fluctuation detection and missing value filling provided by one embodiment of the present invention; Figure 3 This is a functional module diagram of a nutritional component analysis system based on strawberry mineral detection provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0016] The present invention provides a method and system for analyzing nutrient components based on strawberry mineral detection. The core of the method is to achieve accurate evaluation of the nutrient components of strawberry samples by acquiring, processing and extracting the mineral content data in strawberry samples, combining synergistic relationship analysis with multidimensional feature fusion technology. Figure 1 To the attached Figure 3 Specific embodiments of the present invention are described in detail.

[0017] In practical applications, the mineral content dataset in strawberry samples is first collected through the component detection module. This dataset comes from the original data obtained by laboratory testing equipment or field sensors, including but not limited to the concentration values ​​of key mineral elements such as potassium, calcium, magnesium, and iron. In order to ensure data quality, the data needs to be detected for abnormal fluctuations and missing values ​​need to be filled in during the preliminary component distribution detection stage. Specifically, for abnormal fluctuation detection, the trend change curve of the mineral content dataset is first calculated. The formula is as follows: Assume that the mineral content data is a time series , then the trend change curve can be calculated by the moving average method, that is

[0018] Where w is the window size. The normal fluctuation range is determined according to the trend change curve, usually Indicates that is the standard deviation. Data points outside this range are marked as abnormal fluctuations, and abnormal fluctuation detection results are generated. Subsequently, the missing values ​​are filled by using the mean filling method or interpolation method, such as the linear interpolation formula: If the time series X If missing, ; After integrating the abnormal fluctuation detection results with the missing value filling results, the preliminary detection results of mineral distribution were obtained.

[0019] Next, the element identification module is used to identify key elements in the preliminary detection results of mineral distribution to obtain the mineral distribution characteristics. The specific steps include: segmenting the preliminary detection results of mineral distribution according to the preset concentration range, for example, dividing each 0.5 mg / L into an interval, and counting the number of samples in each concentration interval to construct the sample statistical results. Then, a distribution graph is drawn according to the sample statistical results, with the horizontal axis representing the concentration interval and the vertical axis representing the number of samples. By observing the data form of the distribution graph, the preliminary detection results of mineral distribution are converted into a matrix form to obtain the mineral distribution matrix. Furthermore, the mineral distribution matrix is ​​dimensionally optimized, for example, by principal component analysis and dimensionality reduction, retaining the top several principal components with higher contribution rates as mineral distribution characteristics. This process improves the interpretability of mineral data and lays the foundation for subsequent analysis.

[0020] The synergistic analysis module is responsible for calculating the synergistic relationship between different minerals in the strawberry sample and obtaining a synergistic relationship dataset. The calculation of the synergistic relationship is based on correlation analysis, such as the Pearson correlation coefficient formula: ; Where x and y represent the concentration values ​​of two minerals, and Their means are respectively. A synergistic relationship matrix is ​​formed by calculating the correlation coefficients of all mineral pairs. Subsequently, key information is extracted from the synergistic relationship dataset to obtain synergistic characteristics. The specific operations include: analyzing the key correlation points of the synergistic relationship dataset, such as correlation coefficients with large absolute values; assigning weights to the dataset, with the weight parameters calculated using the entropy method; converting the key correlation points into vector form to obtain the synergistic vector; and finally, using principal component analysis to reduce the dimensionality of the synergistic vector to obtain the synergistic characteristics.

[0021] The feature fusion module fuses the synergistic feature and the mineral distribution feature into multidimensional features to obtain the comprehensive feature of the nutritional components. The specific steps include: mapping the synergistic feature and the mineral distribution feature into a multidimensional feature space representation to obtain the synergistic feature space and the mineral distribution feature space respectively. The two feature spaces are decomposed using the feature mapping algorithm to obtain the synergistic low-dimensional subspace and the mineral distribution low-dimensional subspace. Subsequently, the feature weighted fusion algorithm is used to integrate the two low-dimensional subspaces. The formula is as follows: Let the synergistic low-dimensional subspace be , the low-dimensional subspace of mineral distribution is , then the comprehensive characteristics of the nutrient components after fusion are ,in and is the weight parameter, which can be determined by cross-validation. This process enhances the expressive power of features and provides high-dimensional feature support for subsequent nutritional analysis.

[0022] Before using a pre-trained nutrient analysis model for nutrient analysis, the model must be trained. Specifically, a large dataset of publicly available strawberry samples with nutrient labeling is obtained. The synergistic relationships between different minerals in the dataset are calculated to obtain a publicly available synergistic relationship dataset. Feature extraction is performed on the publicly available synergistic relationship dataset to obtain public synergistic features. Mineral distribution features are also extracted from the publicly available strawberry sample dataset. Multidimensional feature fusion is performed on the mineral distribution features and the publicly available synergistic features to obtain public sample features. Subsequently, the public sample features are divided into training and test samples. A machine learning model (such as a random forest or support vector machine) is used to perform nutrient analysis on the training samples to obtain training nutrient evaluation coefficients. Model performance is assessed by calculating the error between the training nutrient evaluation coefficients and the pre-labeled nutrient evaluation coefficients of the training samples. If the error is greater than or equal to a preset error threshold, the model parameters are adjusted and retrained. If the error is less than the error threshold, the model is quality assessed using test samples. If the model fails the quality assessment, the model complexity is reduced and reassessed. If the model passes the quality assessment, model training is confirmed complete, resulting in a nutrient analysis model.

[0023] Finally, the nutritional analysis module was used to analyze the strawberry samples' nutritional composition and obtain a nutritional evaluation coefficient. The specific process involved: using the nutritional analysis model to reorganize the nutritional characteristics of the comprehensive features to obtain a feature set; using the feature enhancement layer to enhance the feature set to obtain enhanced features; using the feature screening layer to extract key features from the enhanced features to obtain screening features; and finally, using the regression analysis module to analyze the nutritional composition of the strawberry samples based on the screening features to obtain a nutritional evaluation coefficient.

[0024] The embodiment of the present invention realizes the real-time detection of the mineral content in the strawberry sample and dynamic fertilization control through the above process, which solves the problem of lack of accurate mineral detection and dynamic fertilization control in the prior art. For example, in an actual application scenario, a strawberry planting base used the system provided by the present invention to detect the mineral content of strawberry samples and found that the potassium content was low. At the same time, synergistic analysis showed that there was a significant positive correlation between potassium and magnesium. Based on this result, the grower adjusted the fertilization plan in time and increased the amount of potassium fertilizer applied, thereby effectively improving the nutritional value and yield of strawberries. In addition, the present invention can also be applied to the field of mineral detection and nutrient analysis of other fruit and vegetable crops, and has broad application prospects.

Claims

1. A nutritional component analysis method based on strawberry mineral detection, characterized in that: The method comprises: Obtaining a mineral content dataset from a strawberry sample, performing a preliminary component distribution test on the mineral content dataset, and obtaining a preliminary mineral distribution test result; Identifying key elements on the preliminary mineral distribution test results to obtain mineral distribution characteristics; calculating the synergistic relationship between different minerals in the strawberry sample to obtain a synergistic relationship data set; Extracting key information from the synergy relationship data set to obtain synergy features; Performing multi-dimensional feature fusion on the synergistic effect characteristics and the mineral distribution characteristics to obtain comprehensive nutrient component characteristics; The pre-trained nutrient analysis model is used to perform nutrient analysis on the strawberry sample according to the comprehensive nutrient characteristics to obtain a nutrient evaluation coefficient.

2. The nutrient component analysis method based on strawberry mineral detection according to claim 1, wherein The preliminary component distribution detection is performed on the mineral content data set to obtain a preliminary mineral distribution detection result, including: performing abnormal fluctuation detection on the mineral content data set to obtain an abnormal fluctuation detection result; Performing missing value filling processing on the mineral content data set to obtain a missing value filling result; The abnormal fluctuation detection results and the missing value filling results are integrated to obtain preliminary mineral distribution detection results.

3. The nutrient component analysis method based on strawberry mineral detection according to claim 2, wherein: The abnormal fluctuation detection is performed on the mineral content data set to obtain the abnormal fluctuation detection result, including: Calculating a trend change curve of the mineral content data set; Determine the normal fluctuation range according to the trend change curve; The mineral content data set is screened for abnormalities according to the normal fluctuation range to obtain an abnormal fluctuation detection result.

4. The nutrient component analysis method based on strawberry mineral detection according to claim 1, wherein The key elements of the preliminary mineral distribution detection results are identified to obtain mineral distribution characteristics, including: Segmenting the preliminary mineral distribution detection results according to preset concentration intervals; Count the number of samples in each concentration range to obtain sample statistical results; Construct a distribution graph based on the sample statistical results with the horizontal axis representing the concentration range and the vertical axis representing the sample number; Converting the preliminary mineral distribution detection results into a matrix form according to the data form of the distribution map to obtain a mineral distribution matrix; The mineral distribution matrix is ​​subjected to dimension optimization processing to obtain mineral distribution characteristics.

5. The nutrient component analysis method based on strawberry mineral detection according to claim 1, wherein: The extracting key information from the synergy relationship dataset to obtain synergy features includes: Analyzing key correlation points of the synergistic relationship data set; Performing weight assignment on the synergistic relationship data set to obtain a weight parameter; Converting the key association points into vector form according to the weight parameters to obtain a synergy vector; The synergy vector is subjected to dimension reduction processing by principal component analysis to obtain synergy features.

6. The nutrient component analysis method based on strawberry mineral detection according to claim 1, wherein: The multi-dimensional feature fusion of the synergistic effect feature and the mineral distribution feature is performed to obtain a comprehensive nutrient component feature, including: Converting the synergistic effect characteristics and the mineral distribution characteristics into a multidimensional feature space representation to obtain a synergistic effect feature space and a mineral distribution feature space; Decomposing the synergistic feature space and the mineral distribution feature space by using a feature mapping algorithm to obtain a synergistic low-dimensional subspace and a mineral distribution low-dimensional subspace; The synergistic low-dimensional subspace and the mineral distribution low-dimensional subspace are integrated using a feature weighted fusion algorithm to obtain comprehensive nutritional characteristics.

7. The nutrient component analysis method based on strawberry mineral detection according to claim 1, wherein: Before performing nutrient component analysis on the strawberry sample according to the comprehensive nutrient component characteristics using the pre-trained nutrient component analysis model, the method further includes: Obtaining a large number of publicly available strawberry sample data sets with nutrient component annotations, calculating the synergistic relationship between different minerals in the publicly available strawberry sample data sets to obtain a publicly available synergistic relationship data set; performing feature extraction on the public synergy relationship dataset to obtain public synergy features; Extracting mineral distribution characteristics of the public strawberry sample dataset; Performing multi-dimensional feature fusion on the mineral distribution characteristics and the disclosed synergistic effect characteristics to obtain disclosed sample characteristics; Dividing the public sample characteristics into training samples and test samples, and using a machine learning model to analyze the nutritional components of the training samples to obtain training nutritional component evaluation coefficients; Calculating the error between the training nutrient evaluation coefficient and the nutrient evaluation coefficient pre-labeled for the training sample; If the error value is greater than or equal to a preset error threshold, the parameters of the machine learning model are adjusted and the step of performing nutrient analysis on the training sample using the machine learning model is re-executed; If the error value is less than the error threshold, performing a model quality assessment on the machine learning model using the test sample; If the machine learning model fails the model quality assessment, reducing the complexity of the machine learning model and re-performing the step of performing model quality assessment on the machine learning model using the test sample; If the machine learning model passes the model quality assessment, the model training is confirmed to be complete and a nutrient analysis model is obtained.

8. The nutrient component analysis method based on strawberry mineral detection according to claim 1, wherein: The pre-trained nutrient analysis model is used to perform nutrient analysis on the strawberry sample according to the comprehensive nutrient characteristics to obtain a nutrient evaluation coefficient, including: Using the nutrient component analysis model to perform feature recombination processing on the comprehensive features of the nutrient components to obtain a feature set; Using the nutrient component analysis model to perform feature enhancement processing on the feature set through a feature enhancement layer to obtain enhanced features; Using the nutrient component analysis model to extract key features of the fortification features through a feature screening layer to obtain screening features; The nutrient component analysis model is used to perform nutrient component analysis on the strawberry sample according to the screening characteristics through a regression analysis module to obtain a nutrient component evaluation coefficient.

9. A nutritional component analysis system based on strawberry mineral detection, characterized in that: The system comprises: A component detection module is used to obtain a mineral content dataset in a strawberry sample, perform a preliminary component distribution test on the mineral content dataset, and obtain a preliminary mineral distribution test result; An element identification module is used to identify key elements based on the preliminary detection results of mineral distribution to obtain mineral distribution characteristics; a synergistic analysis module, configured to calculate the synergistic relationship between different minerals in the strawberry sample to obtain a synergistic relationship data set; a feature fusion module for extracting key information from the synergistic relationship dataset to obtain synergistic features, and performing multidimensional feature fusion on the synergistic features and the mineral distribution features to obtain comprehensive nutrient component features; The nutritional analysis module is used to use a pre-trained nutritional analysis model to perform nutritional analysis on the strawberry sample according to the comprehensive nutritional characteristics to obtain a nutritional evaluation coefficient.

10. The nutritional component analysis system based on strawberry mineral detection according to claim 9, characterized in that: The component detection module includes an abnormal fluctuation detection unit and a missing value filling unit. The abnormal fluctuation detection unit is used to calculate the trend change curve of the mineral content data set and determine the normal fluctuation range for abnormal screening. The missing value filling unit is used to perform missing value filling processing on the mineral content data set.

Images