Pit mud quality detection method, device and equipment based on multiple modes and medium

Abstract

The invention relates to a multi-mode-based pit mud quality detection method, device, equipment and medium, and the method comprises the following steps: acquiring near infrared spectrum data NIR and mid-infrared spectrum data MIR of a pit mud sample to obtain original spectrum data of the pit mud sample, and preprocessing the original spectrum data to obtain standard spectrum data; dynamically adjusting a data fusion mode of NIR and MIR by adopting a multivariate statistical analysis algorithm based on the standard spectral data, and further constructing an NIR-MIR fusion model; carrying out spectral analysis on the pit mud sample based on an NIR-MIR fusion model so as to extract key component information in the pit mud sample; and based on the key component information, according to a preset multi-dimensional evaluation strategy, carrying out comprehensive evaluation on the pit mud sample, and generating a quality evaluation result. The pit mud detection device has the effect of improving pit mud detection efficiency.

Description

Technical Field

[0001] The present application relates to the technical field of pit mud quality detection, and in particular, to a multi-modal based pit mud quality detection method, device, equipment and medium. Background Technique

[0002] At present, pit mud plays a crucial role in industries such as wine brewing and fermentation, and its quality directly affects the fermentation process and the quality of the final product. Therefore, accurately and quickly evaluating the quality of pit mud is of great significance for improving product quality and optimizing production processes.

[0003] Existing pit mud quality detection technologies mainly rely on chemical analysis methods or single-spectral detection methods. For example, traditional chemical analysis methods usually require collecting pit mud samples and conducting component detections in a laboratory, such as measuring mineral components, organic matter content, moisture ratio, pH value, and microbial activity in it. Although these methods are highly accurate, the detection process is cumbersome, time-consuming, requires a large amount of chemical reagents, has a high detection cost, and is easily affected by human factors.

[0004] The above-mentioned existing technical solutions have the following defects: The existing pit mud quality detection methods are cumbersome to detect, resulting in low detection efficiency of pit mud, so there is room for improvement. Summary of the Invention

[0005] In order to improve the efficiency of pit mud detection, the present application provides a multi-modal based pit mud quality detection method, device, equipment and medium.

[0006] The first invention object of the present application is achieved through the following technical solutions: A multi-modal based pit mud quality detection method, the multi-modal based pit mud quality detection method includes: By collecting near-infrared spectral data and mid-infrared spectral data of a pit mud sample, obtaining the original spectral data of the pit mud sample, and preprocessing the original spectral data to obtain standard spectral data; Based on the standard spectral data, using a multivariate statistical analysis algorithm to dynamically adjust the data fusion method of the near-infrared spectral data and the mid-infrared spectral data, and then constructing a NIR-MIR fusion model; Based on the NIR-MIR fusion model, performing spectral analysis on the pit mud sample to extract key component information in the pit mud sample; Based on the key component information, according to a preset multi-dimensional evaluation strategy, comprehensively evaluating the pit mud sample to generate a quality evaluation result.

[0007] By adopting the above technical solutions, by collecting the near-infrared spectral data NIR and mid-infrared spectral data MIR of the pit mud samples and preprocessing the original spectral data, it is possible to reduce the spectral signal errors caused by factors such as environmental noise, equipment deviation, and baseline drift, thereby improving the consistency and reliability of the spectral data, and enabling subsequent analysis to be processed based on more accurate data; by adopting a multivariate statistical analysis algorithm to dynamically adjust the data fusion method of NIR and MIR, and constructing an NIR-MIR fusion model, it is possible to adaptively optimize the fusion ratio of NIR and MIR under the condition of different characteristics of pit mud samples, maximize the complementarity of spectral information, and thus improve the overall evaluation accuracy of the quality of pit mud samples; by performing spectral analysis on the pit mud samples based on the NIR-MIR fusion model to extract the key component information in the pit mud samples, it is possible to accurately identify the core components affecting the quality of pit mud in a data-driven manner, reduce the interference of human factors, and thus improve the stability and repeatability of the pit mud quality detection; by based on the key component information, according to the preset multi-dimensional evaluation strategy, comprehensively evaluate the pit mud samples and generate a quality evaluation result, it is possible to quantitatively analyze the pit mud samples from multiple perspectives, and combine different evaluation indicators to obtain a comprehensive quality evaluation result, thereby improving the scientificity of the pit mud quality evaluation and facilitating the further optimization of the production and application of pit mud.

[0008] In one example, the present application can be further configured as follows: based on the standard spectral data, a multivariate statistical analysis algorithm is used to dynamically adjust the data fusion method of the near-infrared spectral data and the mid-infrared spectral data, and then an NIR-MIR fusion model is constructed, specifically including: Using a multivariate statistical analysis algorithm, respectively extract the features of the near-infrared spectral data and the mid-infrared spectral data in the standard spectral data, and screen the key spectral features that affect the quality of the pit mud samples; According to the key spectral features, calculate the contribution degrees of each spectral component, and determine the weighting coefficients of each spectral component according to the contribution degrees, and then adjust the fusion ratio of the near-infrared spectral data and the mid-infrared spectral data through the weighting coefficients; Based on the fusion ratio, adjust the data fusion method of the near-infrared spectral data and the mid-infrared spectral data, and then construct the NIR-MIR fusion model.

[0009] By adopting the above technical solution, by performing feature extraction on the NIR and MIR spectral data in the standard spectral data and screening the key spectral features that affect the quality of the pit mud sample, redundant information in the spectral data can be effectively removed, making the analysis process more targeted, thereby reducing the computational burden and improving the detection accuracy; by calculating the contribution degrees of each spectral component and determining the weighting coefficients of each spectral component according to the contribution degrees, and then adjusting the fusion ratio of NIR and MIR, it is possible to adaptively optimize the data fusion method based on the influence degree of different spectral bands on the quality of the pit mud, thereby ensuring that the fused data is more representative and improving the reliability of spectral analysis; by adjusting the data fusion method of NIR and MIR based on the fusion ratio and then constructing an NIR-MIR fusion model, the spectral model can be dynamically adapted to the changes of different batches of pit mud samples, thereby improving the generalization ability of the model and enabling it to be applicable to a wider range of pit mud quality assessment tasks.

[0010] In one example, the present application can be further configured as follows: The step of respectively performing feature extraction on the near-infrared spectral data and the mid-infrared spectral data in the standard spectral data and screening the key spectral features that affect the quality of the pit mud sample specifically includes: Dividing the wavelength ranges of the NIR and MIR spectral data in the standard spectral data respectively to identify the characteristic regions related to the quality of the pit mud sample; Performing a correlation analysis on the NIR and MIR to obtain the correlations between each band and the quality of the pit mud sample, screening out the bands that exceed the preset correlation threshold, and then obtaining the corresponding relevant spectral features; Using the principal component analysis algorithm to perform dimensionality reduction processing on the relevant spectral features, extracting the corresponding characteristic information, and then selecting the key spectral features according to the characteristic information.

[0011] By adopting the above technical solution, by dividing the wavelength ranges of the NIR and MIR spectral data in the standard spectral data to identify the characteristic regions related to the quality of the pit mud sample, it is possible to screen out the bands that are irrelevant or have little influence on the quality analysis in the data preprocessing stage, thereby improving the analysis efficiency and reducing the complexity of data processing; by performing a correlation analysis on the NIR and MIR and screening out the bands that exceed the preset correlation threshold, and then obtaining the corresponding relevant spectral features, it is possible to accurately select the most representative spectral features based on the statistical analysis method, thereby improving the pertinence of the analysis and enhancing the accuracy of subsequent data modeling; by using the principal component analysis algorithm to perform dimensionality reduction processing on the relevant spectral features and extracting the key spectral features, it is possible to reduce the dimension of the characteristic variables, remove redundant information, thereby optimizing the data calculation efficiency and improving the stability and generalization ability of the model.

[0012] In one example, the present application can be further configured as follows: performing spectral analysis on the pit mud sample based on the NIR-MIR fusion model to extract key component information in the pit mud sample, specifically including: Performing cluster analysis on the key spectral features to identify potential patterns related to the quality of the pit mud sample; Speculating on the potential patterns through a pattern recognition algorithm to obtain the speculated component types of the pit mud sample, and analyzing the speculated component types in combination with known quality data to generate the key component information.

[0013] By adopting the above technical solution, by performing cluster analysis on the key spectral features to identify potential patterns related to the quality of the pit mud sample, it is possible to identify the similarity in spectral features of different types of pit mud samples in a data-driven manner, thereby providing a clearer category division for quality classification and improving the accuracy of detection; by speculating on the potential patterns through a pattern recognition algorithm to obtain the speculated component types of the pit mud sample and analyzing in combination with known quality data to generate the key component information, it is possible to automatically identify and speculate the main components of the pit mud based on the spectral pattern, improve the degree of automation of the analysis, thereby reducing the errors caused by manual intervention and making the quality assessment more objective and efficient.

[0014] In one example, the present application can be further configured as follows: based on the key component information, performing a comprehensive evaluation on the pit mud sample according to a preset multi-dimensional evaluation strategy to generate a quality evaluation result, specifically including: According to the key component information, assigning weight coefficients to each evaluation dimension in the multi-dimensional evaluation strategy, where the weight coefficient is the proportion of the correlation of different evaluation dimensions in the quality evaluation, and the sum of the weight coefficients is 1; Performing standardization processing on each evaluation dimension, and then according to a preset scoring algorithm Obtaining the comprehensive sample score, where S is the comprehensive sample score, n is the total number of evaluation dimensions, ω i is the weight coefficient of the i-th evaluation dimension, and X is the value of the evaluation dimension after standardization; According to the comprehensive sample score, dividing the corresponding quality grade for the pit mud sample, and generating the quality evaluation result according to the contribution degree of each evaluation dimension.

[0015] By adopting the above technical solution, by allocating weight coefficients to each evaluation dimension in the multi-dimensional evaluation strategy according to the key component information, it is possible to quantitatively evaluate the importance of each evaluation dimension for the quality of pit mud based on different evaluation dimensions, thereby ensuring the rationality of each dimension in quality analysis and improving the scientific nature of the evaluation strategy; by standardizing each evaluation dimension and obtaining the comprehensive score of the sample according to the preset scoring algorithm, it is possible to eliminate the dimensional differences of different evaluation dimensions, enabling each index to be calculated on the same scale, thereby ensuring the accuracy and comparability of the scoring results; by dividing the corresponding quality grades for the pit mud samples according to the comprehensive score of the samples and generating a quality evaluation result based on the contribution degree of each evaluation dimension, it is possible to achieve hierarchical classification of the quality of pit mud on the basis of ensuring the evaluation accuracy, thereby providing a decision-making basis for subsequent process optimization and product quality control.

[0016] In one example, the present application can be further configured as follows: The multi-modal pit mud quality detection method further includes: Based on the key component information of the pit mud sample, extract the core parameters affecting the brewing process, and construct an analysis index for the suitability of pit mud quality according to the core parameters; According to the preset brewing process requirements, set the range of key components of the pit mud corresponding to the brewing process requirements, and match the key component information of the pit mud sample with the key component range; Calculate the matching degree between the key component information and the key component range, and generate the corresponding brewing process for the pit mud sample according to the matching degree.

[0017] By adopting the above technical solution, by extracting the core parameters affecting the brewing process based on the key component information of the pit mud sample and constructing an analysis index for the suitability of pit mud quality, it is possible to quantitatively describe the quality attributes of the pit mud sample according to the quality requirements of different brewing processes, thereby ensuring the suitability between the pit mud and the brewing requirements and improving the accuracy of quality control; by setting the range of key components of the pit mud corresponding to the brewing process requirements and matching the key component information of the pit mud sample with the key component range, it is possible to screen the most suitable pit mud for different brewing processes based on the method of data matching, thereby optimizing the stability and quality consistency of brewing production; by calculating the matching degree between the key component information and the key component range and generating the corresponding brewing process for the pit mud sample according to the matching degree, it is possible to intuitively reflect the matching relationship between the pit mud and different brewing processes through data calculation, thereby reducing the time of experimental screening, improving the brewing production efficiency, and at the same time reducing the risk of process defects caused by the mismatch of pit mud quality.

[0018] The above-mentioned second invention object of the present application is achieved by the following technical solutions: A multimodal-based pit mud quality detection device, the multimodal-based pit mud quality detection device includes: A spectral data acquisition module, configured to obtain the original spectral data of the pit mud sample by collecting the near-infrared spectral data and mid-infrared spectral data of the pit mud sample, and preprocess the original spectral data to obtain standard spectral data; A spectral fusion modeling module, configured to dynamically adjust the data fusion method of the near-infrared spectral data and the mid-infrared spectral data based on the standard spectral data by using a multivariate statistical analysis algorithm, and then construct a NIR-MIR fusion model; A spectral analysis module, configured to perform spectral analysis on the pit mud sample based on the NIR-MIR fusion model to extract key component information in the pit mud sample; A quality evaluation module, configured to comprehensively evaluate the pit mud sample based on the key component information according to a preset multi-dimensional evaluation strategy, and generate a quality evaluation result.

[0019] By adopting the above technical solutions, by collecting the near-infrared spectral data NIR and mid-infrared spectral data MIR of the pit mud sample and preprocessing the original spectral data, it is possible to reduce the spectral signal error caused by factors such as environmental noise, equipment deviation, and baseline drift, thereby improving the consistency and reliability of the spectral data, and enabling subsequent analysis to be processed based on more accurate data; by using a multivariate statistical analysis algorithm to dynamically adjust the data fusion method of NIR and MIR and construct a NIR-MIR fusion model, it is possible to adaptively optimize the fusion ratio of NIR and MIR under the condition of different characteristics of pit mud samples, maximize the complementarity of spectral information, and thereby improve the overall evaluation accuracy of the quality of pit mud samples; by performing spectral analysis on the pit mud sample based on the NIR-MIR fusion model to extract the key component information in the pit mud sample, it is possible to accurately identify the core components affecting the quality of the pit mud in a data-driven manner, reduce the interference of human factors, and thereby improve the stability and repeatability of the pit mud quality detection; by comprehensively evaluating the pit mud sample based on the key component information according to a preset multi-dimensional evaluation strategy and generating a quality evaluation result, it is possible to quantitatively analyze the pit mud sample from multiple perspectives, and obtain a comprehensive quality evaluation result by combining different evaluation indicators, thereby improving the scientificity of the pit mud quality evaluation and facilitating the further optimization of the production and application of the pit mud.

[0020] The above object three of the present application is achieved by the following technical solutions: A computer 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 computer program, the steps of the above multimodal-based pit mud quality detection method are implemented.

[0021] The fourth above-mentioned object of the present application is achieved by the following technical solutions: A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above-mentioned multimodal-based pit mud quality detection method are implemented.

[0022] In summary, the present application includes the following beneficial technical effects: 1. By collecting the near-infrared spectral data NIR and mid-infrared spectral data MIR of the pit mud sample and preprocessing the original spectral data, the spectral signal error caused by factors such as environmental noise, equipment deviation, and baseline drift can be reduced, thereby improving the consistency and reliability of the spectral data, enabling subsequent analysis to be processed based on more accurate data; by adopting a multivariate statistical analysis algorithm to dynamically adjust the data fusion method of NIR and MIR and constructing an NIR-MIR fusion model, the fusion ratio of NIR and MIR can be adaptively optimized under the condition of different pit mud sample characteristics, maximizing the complementarity of spectral information, thereby improving the overall evaluation accuracy of the pit mud sample quality; by performing spectral analysis on the pit mud sample based on the NIR-MIR fusion model to extract the key component information in the pit mud sample, the core components affecting the pit mud quality can be accurately identified in a data-driven manner, reducing the interference of human factors, thereby improving the stability and repeatability of the pit mud quality detection; by comprehensively evaluating the pit mud sample based on the key component information according to the preset multi-dimensional evaluation strategy and generating a quality evaluation result, the pit mud sample can be quantitatively analyzed from multiple angles, and a comprehensive quality evaluation result can be obtained by combining different evaluation indicators, thereby improving the scientificity of the pit mud quality evaluation and facilitating the further optimization of the production and application of the pit mud; 2. By extracting the features of the NIR and MIR spectral data in the standard spectral data and screening the key spectral features affecting the pit mud sample quality, the redundant information in the spectral data can be effectively removed, making the analysis process more targeted, thereby reducing the calculation burden and improving the detection accuracy; by calculating the contribution degree of each spectral component and determining the weighting coefficient of each spectral component according to the contribution degree, and then adjusting the fusion ratio of NIR and MIR, the data fusion method can be adaptively optimized based on the influence degree of different spectral bands on the pit mud quality, thereby ensuring that the fused data is more representative and improving the reliability of the spectral analysis; by adjusting the data fusion method of NIR and MIR based on the fusion ratio and then constructing an NIR-MIR fusion model, the spectral model can be dynamically adapted to the changes of different batches of pit mud samples, thereby improving the generalization ability of the model and enabling it to be applicable to a wider range of pit mud quality evaluation tasks; 3. By dividing the wavelength ranges of the NIR and MIR spectral data in the standard spectral data, the characteristic regions related to the quality of the pit mud samples are identified, which can filter out the bands that are irrelevant or have little impact on the quality analysis during the data preprocessing stage, thereby improving the analysis efficiency and reducing the complexity of data processing. By performing correlation analysis on NIR and MIR and screening out the bands exceeding the preset correlation threshold, the corresponding relevant spectral features are obtained, and the most representative spectral features can be accurately selected based on statistical analysis methods, thus improving the pertinence of the analysis and enhancing the accuracy of subsequent data modeling. By using the principal component analysis algorithm to perform dimensionality reduction processing on the relevant spectral features and extract the key spectral features, the dimension of the feature variables can be reduced, redundant information can be removed, thereby optimizing the data calculation efficiency and enhancing the stability and generalization ability of the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 is a flowchart of a multi-modal pit mud quality detection method according to an embodiment of the present application; Figure 2 is an implementation flowchart of step S20 in the multi-modal pit mud quality detection method according to an embodiment of the present application; Figure 3 is an implementation flowchart of step S21 in the multi-modal pit mud quality detection method according to an embodiment of the present application; Figure 4 is an implementation flowchart of step S30 in the multi-modal pit mud quality detection method according to an embodiment of the present application; Figure 5 is an implementation flowchart of step S40 in the multi-modal pit mud quality detection method according to an embodiment of the present application; Figure 6 is another implementation flowchart of the multi-modal pit mud quality detection method according to an embodiment of the present application; Figure 7 is a principle block diagram of a multi-modal pit mud quality detection device according to an embodiment of the present application; Figure 8 is a schematic diagram of a device according to an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0024] The present application will be further described in detail below with reference to the accompanying drawings.

[0025] In one embodiment, as Figure 1 shown, the present application discloses a multi-modal pit mud quality detection method, which specifically includes the following steps: S10: By collecting the near-infrared spectral data NIR and mid-infrared spectral data MIR of the pit mud samples, the original spectral data of the pit mud samples are obtained, and the original spectral data are preprocessed to obtain the standard spectral data.

[0026] Specifically, a near-infrared spectrometer NIR and a mid-infrared spectrometer MIR are used to scan the pit mud samples respectively. NIR is used to collect the molecular vibration information of the samples, and MIR is used to detect the chemical bond characteristic information of the samples. During the collection, a fixed light source intensity is adopted to ensure signal stability. For example, the scanning range is set to 800nm - 2500nm (NIR) and 2500nm - 5000nm (MIR) according to the instrument performance. After the data collection is completed, the original spectral data is preprocessed, including steps such as removing environmental noise, background correction, and spectral normalization. Among them, wavelet transform or SG smoothing filtering method can be used for noise removal, polynomial fitting algorithm is used for background correction to remove the influence of baseline drift on the spectrum, and maximum-minimum normalization method is used for spectral normalization processing to ensure the consistency of data between different batches of samples. The spectral data after preprocessing is the standard spectral data.

[0027] S20: Based on the standard spectral data, a multivariate statistical analysis algorithm is used to dynamically adjust the data fusion method of the near-infrared spectral data and the mid-infrared spectral data, and then an NIR-MIR fusion model is constructed.

[0028] Specifically, based on the standard spectral data, first calculate the correlation between different bands to identify the complementary information between the NIR and MIR spectral data. The principal component analysis or independent component analysis method is used to perform dimensionality reduction on the data to reduce the influence of redundant information, while retaining the main features of the spectral data. On this basis, the multivariate linear regression or partial least squares regression method is used to perform weighted combination on the NIR and MIR data, and the fusion weight is dynamically adjusted according to the contribution degree of different bands, so that the data fusion method of the near-infrared spectral data and the mid-infrared spectral data can adapt to the characteristics of different pit mud samples, and finally an NIR-MIR fusion model is constructed.

[0029] S30: Based on the NIR-MIR fusion model, spectral analysis is performed on the pit mud samples to extract the key component information in the pit mud samples.

[0030] Specifically, based on the fusion model, calculate the spectral feature vector of each sample and input it into the pattern recognition algorithm for classification. The clustering analysis method is used to group the spectral features to identify the sample categories with similar spectral features. After the sample classification is completed, calculate the relationship between different spectral features and the sample component content through multivariate regression analysis, and speculate on the key components. During this process, a regularization-based multivariate regression method can be used to reduce the influence of noise on component prediction. Finally, according to the results of the spectral feature matching analysis, the key component information in the pit mud samples is extracted.

[0031] S40: Based on the key component information, comprehensively evaluate the pit mud sample according to the preset multi-dimensional evaluation strategy to generate a quality evaluation result.

[0032] Specifically, according to the key component information of the pit mud sample, establish a multi-dimensional evaluation model, which includes multiple evaluation dimensions such as mineral composition, moisture content, pH value, and microbial activity. To ensure the fairness of each evaluation dimension, first use the standardization method to normalize the data so that different indicators can be compared on the same scale. Subsequently, set the corresponding weight coefficients for the influence degrees of different evaluation dimensions, calculate the comprehensive score through weighted calculation, and divide the quality grade of the sample based on the score threshold. Finally, output the quality evaluation result of the pit mud sample, which is used to guide subsequent process optimization or quality control.

[0033] In one embodiment, as Figure 2 shown, in step S20, that is, based on the standard spectral data, use the multivariate statistical analysis algorithm to dynamically adjust the data fusion method of the near-infrared spectral data and the mid-infrared spectral data, and then construct the NIR-MIR fusion model, which specifically includes: S21: Use the multivariate statistical analysis algorithm to extract the features of the near-infrared spectral data and the mid-infrared spectral data in the standard spectral data respectively, and screen the key spectral features that affect the quality of the pit mud sample.

[0034] Specifically, divide the wavelength ranges of the near-infrared spectral data and the mid-infrared spectral data respectively, eliminate the band information that has no significant impact on the quality evaluation, use the correlation analysis method to calculate the contribution degree of each band to the quality evaluation, and screen the bands with higher correlation as the key spectral features. Subsequently, perform principal component analysis and dimensionality reduction on the screened spectral features to extract the most representative spectral feature components, so as to reduce redundant information and improve the stability and accuracy of subsequent analysis.

[0035] S22: According to the key spectral features, calculate the contribution degree of each spectral component, and determine the weighting coefficient of each spectral component according to the contribution degree, and then adjust the fusion ratio of the near-infrared spectral data and the mid-infrared spectral data through the weighting coefficient.

[0036] Specifically, calculate the feature importance of different spectral bands, use statistical methods to analyze the influence weight of each band on the quality evaluation of the sample, and determine the fusion ratio of each spectral band based on the optimal weighting combination strategy. During the calculation process, the orthogonal least squares method or the gradient boosting decision tree method can be used to optimize the weight parameters so that the fused spectral data can most reflect the quality characteristics of the sample.

[0037] S23: Based on the fusion ratio, adjust the data fusion method of the near-infrared spectral data and the mid-infrared spectral data, and then construct the NIR-MIR fusion model.

[0038] Specifically, the kernel principal component analysis method is adopted to optimize the non-linear mapping ability of the fusion model and improve the robustness of the fused data. During the fusion process, the fusion coefficient is dynamically adjusted according to the characteristics of different samples, so that the fused spectral data has higher resolution and prediction ability in subsequent analysis.

[0039] In one embodiment, as Figure 3 shown, in step S21, the near-infrared spectral data and mid-infrared spectral data in the standard spectral data are respectively subjected to feature extraction, and the key spectral features affecting the quality of the pit mud samples are screened, specifically including: S211: The wavelength ranges of the near-infrared spectral data and mid-infrared spectral data in the standard spectral data are respectively divided to identify the characteristic regions related to the quality of the pit mud samples.

[0040] Specifically, the near-infrared spectral data and mid-infrared spectral data are preprocessed to eliminate background noise and external interference factors. Subsequently, the method of fixed band division is adopted. The near-infrared spectral data is divided into three intervals such as 800 - 1100 nm, 1100 - 1700 nm, and 1700 - 2500 nm, and the mid-infrared spectral data is divided into three intervals such as 2500 - 3500 nm, 3500 - 4500 nm, and 4500 - 5000 nm. To ensure that the selected bands are related to the quality of the pit mud samples, the spectral response analysis method is used to calibrate the spectral absorption peaks of different bands, calculate the response degrees of each band to different components of the pit mud samples, and verify with the existing experimental data. Finally, the characteristic regions most influential to the quality of the pit mud samples are identified.

[0041] S212: The correlation analysis is carried out on the analyzed near-infrared spectral data and mid-infrared spectral data to obtain the correlations between each band and the quality of the pit mud samples, and the bands exceeding the preset correlation threshold are screened out, thereby obtaining the corresponding relevant spectral features.

[0042] Specifically, the partial least squares regression method is adopted to model the near-infrared spectral data and mid-infrared spectral data, calculate the correlations between the spectral signals of different bands and the quality parameters of the pit mud samples, such as moisture content, mineral content, etc., and evaluate them using the Pearson correlation coefficient. The bands with a correlation coefficient exceeding 0.75 are screened out as candidate characteristic bands. To improve the robustness of the data, the mutual information analysis method is introduced at the same time to further calculate the information entropy between different spectral bands and the quality parameters of the pit mud, and finally, the spectral characteristic bands with the strongest correlation with the quality of the pit mud samples are determined according to the comprehensive score of correlation and information entropy as the input for subsequent data processing.

[0043] S213: Use the principal component analysis algorithm to perform dimensionality reduction on the relevant spectral features, extract the corresponding feature information, and then select the key spectral features based on the feature information.

[0044] Specifically, first standardize the selected relevant spectral feature data to eliminate the scale differences between different samples. Subsequently, use the principal component analysis method to calculate the projection values of each spectral feature in different principal components, and analyze the contribution degrees of each principal component to the pit mud quality parameters. Select the principal components with a contribution degree exceeding 85% as the key spectral features. In this process, determine the optimal number of principal components through the Kaiser criterion, and verify the dimensionality reduction effect through the cumulative variance contribution rate. Finally, extract the most representative spectral features to provide data support for subsequent spectral analysis.

[0045] In one embodiment, as Figure 4 shown, in step S30, that is, perform spectral analysis on the pit mud samples based on the NIR-MIR fusion model to extract the key component information in the pit mud samples, specifically including: S31: Perform clustering analysis on the key spectral features to identify potential patterns related to the quality of the pit mud samples.

[0046] Specifically, use the K-means clustering method to perform clustering analysis on the key spectral features. According to the spectral distribution characteristics of different samples, divide them into multiple categories, and calculate the center points of each category to determine the similarity between different categories. To optimize the clustering effect, use the silhouette coefficient to evaluate the rationality of different clustering results, and optimize the K value to make the clustering results best reflect the differences in the quality of the pit mud samples. Finally, determine the category labels of each sample to identify potential patterns related to the pit mud quality.

[0047] S32: Speculate on the potential patterns through a pattern recognition algorithm to obtain the speculated component types of the pit mud samples, and analyze the speculated component types in combination with the known quality data to generate key component information.

[0048] Specifically, use the support vector machine or random forest classification method to classify the spectral patterns of the samples, and learn the corresponding relationship between the spectral patterns and the pit mud quality components based on the training set data. In the process of speculating on the component types, use the cross-validation method to evaluate the performance of the classification model to ensure its generalization ability. After classification, compare the speculated component types with the known quality data, calculate the classification accuracy rate. If the accuracy rate is lower than the set threshold, optimize it by adjusting the feature selection strategy or increasing the training samples. Finally, generate key component information with high confidence.

[0049] In one embodiment, as Figure 5As shown, in step S40, based on the key component information, a comprehensive evaluation of the pit mud sample is carried out according to a preset multi-dimensional evaluation strategy to generate a quality evaluation result, specifically including: S41: According to the key component information, weight coefficients are assigned to each evaluation dimension in the multi-dimensional evaluation strategy. The weight coefficient is the proportion of the correlation of different evaluation dimensions in the quality evaluation, and the sum of the weight coefficients is 1.

[0050] Specifically, the analytic hierarchy process is used to determine the weights of the evaluation dimensions. First, an impact matrix of the evaluation dimensions on the quality evaluation is established, and the importance weights of each dimension are calculated based on historical data, etc. Subsequently, a consistency test is performed on the weights to ensure the rationality of the weight assignment. If the consistency test passes, the weights of each evaluation dimension are normalized so that the sum of all weight coefficients is equal to 1, and finally, the contribution ratio of different evaluation dimensions in the quality evaluation is determined.

[0051] S42: Standardize each evaluation dimension, and then according to a preset scoring algorithm obtain the comprehensive sample score, where S is the comprehensive sample score, n is the total number of evaluation dimensions, ω i is the weight coefficient of the i-th evaluation dimension, and X is the standardized evaluation dimension value.

[0052] Specifically, the min-max normalization method is used to standardize the values of each evaluation dimension, converting them into values between 0 and 1 for weighted calculation. Subsequently, the weighted average method is used to calculate the comprehensive sample score, that is, the standardized values of each evaluation dimension are multiplied by the corresponding weight coefficients, and all weights are summed to obtain the final comprehensive score. After the score calculation is completed, the score result is verified and compared with historical data to ensure the rationality and stability of the calculation result.

[0053] S43: According to the comprehensive sample score, divide the corresponding quality grade for the pit mud sample and generate a quality evaluation result based on the contribution degree of each evaluation dimension.

[0054] Specifically, based on the comprehensive score, the pit mud samples are divided into multiple quality grades, including excellent, good, qualified, and unqualified categories. The division criteria are set according to historical data with different scoring thresholds. For example, samples with a score higher than 0.85 are judged as high-quality pit mud, samples with a score between 0.7 and 0.85 are judged as good, and samples with a score lower than 0.5 are judged as unqualified. To ensure the scientific nature of the quality evaluation result, weighted analysis can also be combined with the contribution degree of each evaluation dimension, and detailed quality evaluation analysis results are provided in the final evaluation report.

[0055] In one embodiment, as Figure 6 shown, that is, the multi-modal pit mud quality detection method further includes: S50: Extract the core parameters affecting the brewing process based on the key component information of the pit mud sample, and construct an analysis index for the suitability of pit mud quality according to the core parameters.

[0056] Specifically, analyze the influencing factors of the pit mud sample in different brewing processes, extract the core parameters crucial for the fermentation process, including mineral content, moisture content, organic matter content, pH value, etc. According to the specific requirements of different brewing processes for the pit mud, set the applicable range of each parameter, and based on historical production data and actual measurement results, construct an analysis index for the suitability of pit mud quality. In the suitability analysis, a weighted comprehensive scoring method is adopted, weights are assigned according to the importance of each parameter, and the comprehensive suitability score of the pit mud sample is obtained through weighted calculation. This score is used to measure the adaptability of the pit mud sample to different brewing processes, thereby constructing an analysis system for the suitability of pit mud quality and providing data support for subsequent brewing process matching.

[0057] S60: According to the preset brewing process requirements, set the range of key components of the pit mud corresponding to the brewing process requirements, and match the key component information of the pit mud sample with the key component range.

[0058] Specifically, first collect the quality requirements of different brewing processes for the pit mud sample, including the proportion of mineral components, pH value, microbial activity, moisture content, etc. Based on existing industry standards and historical data, set the range of key components for different brewing process requirements, and establish a database of brewing process requirements. Subsequently, input the key component information of the pit mud sample into the database, and use the cosine similarity calculation method to calculate the matching degree between the sample and the brewing process requirements. Through the matching degree calculation, the most suitable brewing process category for the pit mud sample can be identified, thereby providing accurate suggestions for the use of pit mud in the production process.

[0059] S70: The matching degree between the key component information and the key component range, and generate the corresponding brewing process for the pit mud sample according to the matching degree.

[0060] Specifically, in the process of calculating the matching degree, the Euclidean distance or Mahalanobis distance method is used to measure the closeness between the key component information of the pit mud sample and the brewing process requirements. When the matching degree is higher than the set threshold, recommend this pit mud sample as the preferred material for a certain brewing process. If the matching degree is between two thresholds, the decision tree analysis or fuzzy logic analysis method can be further used to comprehensively adjust in combination with other influencing factors, such as fermentation temperature, type of koji, etc. On the basis of calculating the matching degree, generate an adaptability report of the pit mud sample and different brewing processes. This report includes the key component indicators of the pit mud sample, the matching degree score, the optimal matching brewing process suggestions, and further production process adjustment suggestions, and finally provides accurate data support for brewing process optimization.

[0061] It should be understood that the sequence numbers of the steps in the above embodiments do not imply the order of execution. The execution order of each process should be determined according to its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present application.

[0062] In one embodiment, a multi-modal pit mud quality detection device is provided. The multi-modal pit mud quality detection device corresponds one-to-one with the multi-modal pit mud quality detection method in the above embodiment. As Figure 7 shown, the multi-modal pit mud quality detection device includes a spectral data acquisition module, a spectral fusion modeling module, a spectral analysis module, and a quality evaluation module. The detailed description of each functional module is as follows: The spectral data acquisition module is used to obtain the original spectral data of the pit mud sample by collecting the near-infrared spectral data and mid-infrared spectral data of the pit mud sample, and preprocess the original spectral data to obtain the standard spectral data; The spectral fusion modeling module is used to dynamically adjust the data fusion method of the near-infrared spectral data and mid-infrared spectral data based on the standard spectral data by using a multivariate statistical analysis algorithm, and then construct a NIR-MIR fusion model; The spectral analysis module is used to perform spectral analysis on the pit mud sample based on the NIR-MIR fusion model to extract the key component information in the pit mud sample; The quality evaluation module is used to comprehensively evaluate the pit mud sample based on the key component information according to the preset multi-dimensional evaluation strategy, and generate a quality evaluation result.

[0063] Optionally, the spectral fusion modeling module specifically includes: The feature extraction sub-module is used to extract features from the near-infrared spectral data and mid-infrared spectral data in the standard spectral data respectively by using a multivariate statistical analysis algorithm, and screen out the key spectral features that affect the quality of the pit mud sample; The fusion weight calculation sub-module is used to calculate the contribution degree of each spectral component according to the key spectral features, and determine the weighting coefficient of each spectral component according to the contribution degree, and then adjust the fusion ratio of the near-infrared spectral data and mid-infrared spectral data through the weighting coefficient; The fusion model construction sub-module is used to adjust the data fusion method of the near-infrared spectral data and mid-infrared spectral data based on the fusion ratio, and then construct a NIR-MIR fusion model.

[0064] Optionally, the feature extraction sub-module specifically includes: The wavelength screening unit is used to divide the wavelength ranges of the NIR and MIR spectral data in the standard spectral data respectively, and identify the characteristic regions related to the quality of the pit mud sample; A correlation analysis unit is used to perform a correlation analysis on the NIR and MIR, obtain the correlation between each band and the quality of the pit mud sample, screen out the bands that exceed the preset correlation threshold, and then obtain the corresponding relevant spectral features. A dimensionality reduction and optimization unit is used to perform dimensionality reduction processing on the relevant spectral features by using the principal component analysis algorithm, extract the corresponding feature information, and then select the key spectral features according to the feature information.

[0065] Optionally, the spectral analysis module specifically includes: A pattern recognition sub-module is used to perform cluster analysis on the key spectral features to identify potential patterns related to the quality of the pit mud sample; a component speculation sub-module is used to speculate on the potential patterns through a pattern recognition algorithm to obtain the speculated component types of the pit mud sample, and analyze the speculated component types in combination with the known quality data to generate key component information.

[0066] Optionally, the quality assessment module specifically includes: A weight calculation sub-module is used to assign weight coefficients to each evaluation dimension in the multi-dimensional evaluation strategy according to the key component information. The weight coefficient is the proportion of the correlation of different evaluation dimensions in the quality assessment, and the sum of the weight coefficients is 1. A normalization processing sub-module is used to perform normalization processing on each evaluation dimension, and then according to the preset scoring algorithm Obtain the comprehensive sample score, where S is the comprehensive sample score, n is the total number of evaluation dimensions, ω i Is the weight coefficient of the i-th evaluation dimension, and X is the normalized evaluation dimension value. A quality grade determination sub-module is used to divide the corresponding quality grade for the pit mud sample according to the comprehensive sample score, and generate a quality assessment result according to the contribution degree of each evaluation dimension.

[0067] Optionally, the multi-modal based pit mud quality detection method further includes: A core parameter extraction module is used to extract the core parameters affecting the brewing process based on the key component information of the pit mud sample, and construct an analysis index for the adaptability of the pit mud quality according to the core parameters. An adaptability evaluation module is used to set the range of key components of the pit mud corresponding to the brewing process requirements according to the preset brewing process requirements, and match the key component information of the pit mud sample with the range of key components. An adaptability report generation module is used to match the key component information with the range of key components, and generate the corresponding brewing process for the pit mud sample according to the matching degree.

[0068] For the specific limitations of the multimodal-based pit mud quality detection device, reference can be made to the limitations of the multimodal-based pit mud quality detection method described above, which will not be elaborated here. Each module in the above multimodal-based pit mud quality detection device can be implemented in whole or in part by software, hardware, or a combination thereof. The above modules can be embedded in the processor of the computer device in hardware form or be independent of it, or be stored in the memory of the computer device in software form, so that the processor can call and execute the operations corresponding to the above modules.

[0069] In one embodiment, a computer device is provided. The computer device can be a server, and its internal structure diagram can be as Figure 8 shown. The computer device includes a processor, a memory, a network interface, and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with external terminals through a network connection. When the computer program is executed by the processor, it implements a multimodal-based pit mud quality detection method.

[0070] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the following steps are implemented: By collecting the near-infrared spectral data and mid-infrared spectral data of the pit mud sample, the original spectral data of the pit mud sample is obtained, and the original spectral data is preprocessed to obtain standard spectral data; Based on the standard spectral data, a multivariate statistical analysis algorithm is used to dynamically adjust the data fusion method of the near-infrared spectral data and the mid-infrared spectral data, and then a NIR-MIR fusion model is constructed; Based on the NIR-MIR fusion model, spectral analysis is performed on the pit mud sample to extract the key component information in the pit mud sample; Based on the key component information, according to the preset multi-dimensional evaluation strategy, a comprehensive evaluation is performed on the pit mud sample to generate a quality evaluation result.

[0071] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by the processor, the following steps are implemented: By collecting the near-infrared spectral data and mid-infrared spectral data of the pit mud sample, the original spectral data of the pit mud sample is obtained, and the original spectral data is preprocessed to obtain standard spectral data; Based on standard spectral data, a multivariate statistical analysis algorithm is used to dynamically adjust the data fusion method of near-infrared spectral data and mid-infrared spectral data, and then a NIR-MIR fusion model is constructed; Based on the NIR-MIR fusion model, spectral analysis is performed on the pit mud samples to extract the key component information in the pit mud samples; Based on the key component information, according to the preset multi-dimensional evaluation strategy, a comprehensive evaluation is performed on the pit mud samples to generate a quality evaluation result.

[0072] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, storage, database or other medium used in the various embodiments provided in the present application can include non-volatile and / or volatile memories. Non-volatile memories can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memories can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0073] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the above division of each functional unit and module is used for illustration. In actual applications, the above functions can be allocated to different functional units and modules as needed, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above.

[0074] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the various embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A method for detecting the quality of pit mud based on multimodality, characterized in that: The multimodal-based pit mud quality detection method comprises: By collecting near-infrared spectrum data and mid-infrared spectrum data of the pit mud sample, original spectrum data of the pit mud sample is obtained, and the original spectrum data is preprocessed to obtain standard spectrum data; Based on the standard spectral data, a multivariate statistical analysis algorithm is used to dynamically adjust the data fusion mode of the near infrared spectral data and the mid infrared spectral data, thereby constructing a NIR-MIR fusion model; Performing spectral analysis on the cellar mud sample based on the NIR-MIR fusion model to extract key component information in the cellar mud sample; Based on the key component information and in accordance with a preset multi-dimensional evaluation strategy, a comprehensive evaluation is performed on the pit mud samples to generate a quality evaluation result.

2. The multimodal cellar mud quality detection method according to claim 1, characterized in that: The method of dynamically adjusting the data fusion mode of the near infrared spectrum data and the mid infrared spectrum data by using a multivariate statistical analysis algorithm based on the standard spectrum data, and then constructing a NIR-MIR fusion model, specifically includes: Using a multivariate statistical analysis algorithm, feature extraction is performed on the near infrared spectrum data and the mid infrared spectrum data in the standard spectrum data, and key spectrum features that affect the quality of the pit mud sample are screened; According to the key spectral features, the contribution of each spectral component is calculated, and the weighting coefficient of each spectral component is determined according to the contribution, and then the fusion ratio of the near-infrared spectral data and the mid-infrared spectral data is adjusted by the weighting coefficient; The data fusion mode of the near infrared spectral data and the mid infrared spectral data is adjusted based on the fusion ratio, and then the NIR-MIR fusion model is constructed.

3. The multimodal cellar mud quality detection method according to claim 2, characterized in that: The method of extracting features from the near infrared spectrum data and the mid infrared spectrum data in the standard spectrum data respectively and screening key spectrum features that affect the quality of the pit mud sample specifically includes: The NIR and MIR spectral data in the standard spectral data are divided into wavelength ranges respectively, and characteristic regions related to the quality of the pit mud sample are identified; Performing correlation analysis on the NIR and MIR analysis to obtain the correlation between each band and the quality of the pit mud sample, and screening out the band that exceeds a preset correlation threshold, thereby obtaining the corresponding related spectral features; The principal component analysis algorithm is used to perform dimensionality reduction processing on the relevant spectral features, extract corresponding feature information, and then select the key spectral features according to the feature information.

4. The multimodal pit mud quality detection method according to claim 3, characterized in that: The spectral analysis of the cellar mud sample based on the NIR-MIR fusion model to extract key component information in the cellar mud sample specifically includes: Performing cluster analysis on the key spectral features to identify potential patterns associated with the quality of the pit mud sample; The potential pattern is inferred through a pattern recognition algorithm to obtain the inferred component type of the pit mud sample, and the inferred component type is analyzed in combination with known quality data to generate the key component information.

5. The multimodal cellar mud quality detection method according to claim 1, characterized in that: Based on the key component information, according to a preset multi-dimensional evaluation strategy, a comprehensive evaluation is performed on the pit mud sample to generate a quality evaluation result, specifically including: According to the key component information, a weight coefficient is assigned to each evaluation dimension in the multidimensional evaluation strategy, wherein the weight coefficient is the correlation ratio of different evaluation dimensions in the quality evaluation, and the sum of the weight coefficients is 1; Each of the evaluation dimensions is standardized, and then the evaluation is performed according to the preset scoring algorithm. Get the comprehensive score of the sample, where S is the comprehensive score of the sample, n is the total amount of the evaluation dimension, ω i is the weight coefficient of the i-th evaluation dimension, and X is the standardized value of the evaluation dimension; According to the comprehensive score of the samples, the pit mud samples are divided into corresponding quality grades, and the quality assessment results are generated according to the contribution of each assessment dimension.

6. The multimodal cellar mud quality detection method according to claim 1, characterized in that: The multimodal-based pit mud quality detection method further includes: Based on the key component information of the pit mud sample, the core parameters affecting the winemaking process are extracted, and the pit mud quality fitness analysis index is constructed according to the core parameters; According to the preset winemaking process requirements, the key component range of the cellar mud corresponding to the winemaking process requirements is set, and the key component information of the cellar mud sample is matched with the key component range; The matching degree between the key component information and the key component range is determined, and the brewing process corresponding to the cellar mud sample is generated according to the matching degree.

7. A multi-modal cellar mud quality detection device, characterized in that: The multi-modal cellar mud quality detection device comprises: The spectral data acquisition module is used to acquire the near-infrared spectral data and mid-infrared spectral data of the cellar mud sample to obtain the original spectral data of the cellar mud sample, and pre-process the original spectral data to obtain standard spectral data; A spectral fusion modeling module is used to dynamically adjust the data fusion mode of the near-infrared spectral data and the mid-infrared spectral data based on the standard spectral data by using a multivariate statistical analysis algorithm, so as to construct a NIR-MIR fusion model; a spectral analysis module is used to perform spectral analysis on the cellar mud sample based on the NIR-MIR fusion model to extract key component information in the cellar mud sample; The quality assessment module is used to conduct a comprehensive assessment of the pit mud sample based on the key component information and according to a preset multi-dimensional assessment strategy to generate a quality assessment result.

8. The multi-modal cellar mud quality detection device according to claim 7 is characterized in that: The spectral fusion modeling module specifically includes: A feature extraction submodule is used to use a multivariate statistical analysis algorithm to perform feature extraction on the near-infrared spectrum data and the mid-infrared spectrum data in the standard spectrum data, and screen key spectrum features that affect the quality of the pit mud sample; a fusion weight calculation submodule is used to calculate the contribution of each spectrum component according to the key spectrum features, and determine the weighting coefficient of each spectrum component according to the contribution, and then adjust the fusion ratio of the near-infrared spectrum data and the mid-infrared spectrum data by the weighting coefficient; The fusion model construction submodule is used to adjust the data fusion mode of the near-infrared spectral data and the mid-infrared spectral data based on the fusion ratio, so as to construct the NIR-MIR fusion model.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the multimodal cellar mud quality detection method as described in any one of claims 1 to 6 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the multimodal pit mud quality detection method as described in any one of claims 1 to 6 are implemented.

Images