Diester oil quality detection method and system based on multi-source data

Abstract

The invention relates to the technical field of food quality detection, in particular to a diester oil quality detection method and system based on multi-source data, and the method comprises the steps: collecting multi-source heterogeneous data, carrying out feature extraction and dimension reduction, and generating a quality feature vector; multi-source data fusion is realized through a dynamic weight distribution model, and time sequence quality evaluation data is generated; performing error compensation on the low-confidence data by adopting a self-adaptive correction algorithm, and performing chain verification on the high-confidence data; and through bidirectional consistency verification, optimizing path planning, generating optimization feedback data and finishing final quality confirmation. Through multi-source data fusion, real-time analysis and automatic verification, the accuracy and efficiency of diester oil quality detection are remarkably improved, and the modern industrial intelligent requirement is met.

Description

Technical Field

[0001] The present invention belongs to the technical field of food quality detection, and in particular relates to a diester oil quality detection method and system based on multi-source data. Background Art

[0002] With the continuous development of the food industry and oil processing technology, diester oils have gradually become a hot topic of research and application due to their unique functional properties and health benefits. Rapid and accurate quality testing of diester oils is crucial for ensuring product quality stability and production efficiency during their production and application. However, existing diester oil quality testing methods and systems still have significant deficiencies in multi-source data fusion, real-time performance, accuracy, and automation, making them unable to meet the modern industry's demand for efficient and intelligent quality testing.

[0003] After searching, it was found that the prior art discloses a data monitoring method and system for cigarette filter rod production (publication number CN111184254B, publication date October 15, 2021). This patent obtains multiple key parameters in the filter rod production process (such as bundle moisture content, bundle density and triacetin content) and combines the process capability index Cpk for quality control, thereby improving the stability of product quality and the level of control over the production process. However, this technical solution is mainly aimed at the specific scenario of cigarette filter rod production, and its data collection and analysis methods are relatively simple, and it fails to make full use of multi-source heterogeneous data (such as spectra, images, chemical composition, etc.) for comprehensive analysis. In addition, the solution has weak adaptability to the dynamically changing production environment, and it is difficult to achieve rapid response and real-time adjustment. Therefore, it cannot be directly applied to the quality detection scenario of diester oil.

[0004] Another related technology is a method for preparing high-purity phlorizin and testing its quality (publication number CN110540558B, publication date April 25, 2023). This patent achieves precise control of phlorizin purity by adopting multiple separation and purification techniques and high-performance liquid chromatography (HPLC) detection methods, thereby ensuring the stability of product quality. However, this technical solution mainly relies on laboratory-level separation and detection equipment and lacks the ability to integrate and analyze multi-source data in large-scale industrial production. At the same time, its detection process is relatively complex, involving multiple steps (such as extraction, column chromatography, recrystallization, etc.), resulting in a long detection cycle, which makes it difficult to meet the demand for rapid, online quality detection in the diester oil production process.

[0005] The above issues demonstrate that existing quality detection methods and systems still have shortcomings in terms of multi-source data integration, real-time performance, detection efficiency, and adaptability. In particular, facing complex production environments and diverse testing requirements, existing technologies struggle to achieve comprehensive, rapid, and accurate detection of diester oil quality. To address this, the present invention provides a diester oil quality detection method and system based on multi-source data. Summary of the Invention

[0006] In order to make up for the deficiencies of the prior art, at least one technical problem raised in the background technology is solved.

[0007] The technical solution adopted by the present invention to solve the technical problem is: a diester oil quality detection method based on multi-source data according to the present invention comprises the following steps: Step S1: collecting a multi-source heterogeneous data set, wherein the multi-source heterogeneous data set includes spectral data, chemical composition data, and process parameter data; performing feature extraction and dimensionality reduction processing on the multi-source heterogeneous data set to generate a diester oil quality feature vector; constructing a dynamic weight allocation model based on the diester oil quality feature vector to generate a multi-source data fusion weight matrix; Step S2: Performing real-time analysis on diester oil quality feature vectors based on a multi-source data fusion weight matrix to generate quality assessment time series data; classifying diester oil samples based on the quality assessment time series data to generate high-confidence quality data and low-confidence quality data; applying an adaptive correction algorithm to error-compensate the low-confidence quality data to generate corrected low-confidence quality data; and performing rapid chain verification on the high-confidence quality data to generate a high-confidence chain verification result. Step S3: Perform a bidirectional consistency check on the corrected low-confidence quality data and the high-confidence chain verification result to generate consistency verification data; perform optimized path planning on the corrected low-confidence quality data and the high-confidence chain verification result based on the consistency verification data to generate quality inspection optimization feedback data; Step S4: Perform final quality confirmation on the quality inspection optimization feedback data to obtain a diester oil quality inspection report; upload the diester oil quality inspection report to the cloud storage platform, and display the inspection results through a visual interface to generate a diester oil quality inspection completion report.

[0008] Preferably, according to step S1, the dynamic weight allocation model uses the following formula to calculate the weight of the i-th data source:

[0009] in, Indicates the The weight of each data source, Indicates the The error value of each data source, To adjust the parameters, The total number of data sources.

[0010] Preferably, according to step S2, the adaptive correction algorithm uses the following formula to perform error compensation on low confidence quality data:

[0011] in, represents the corrected quality data, represents the original quality data, Indicates the reference standard value, is the correction factor, and its value range is 0.1 to 0.5.

[0012] Preferably, according to step S1, the feature extraction and dimensionality reduction process uses principal component analysis to process the multi-source heterogeneous data set, and retains several principal components with the largest variance to generate diester oil quality feature vectors.

[0013] Preferably, according to step S2, the real-time analysis uses a sliding time window technology to continuously monitor the diester oil quality feature vector, with the window length set to 10 seconds to 30 seconds and the step length being 1-5 seconds.

[0014] Preferably, according to step S2, the classification process uses a support vector machine algorithm to classify the diester oil samples, the kernel function selects a radial basis function and the hyperparameters are optimized through cross-validation.

[0015] Preferably, according to step S3, the bidirectional consistency verification process includes comparing the corrected low-confidence quality data with the high-confidence chain verification result, calculating the deviation between the two, and judging whether the consistency condition is met based on the size of the deviation.

[0016] Preferably, according to step S3, the optimization path planning uses a genetic algorithm to perform a global search for parameters and combines it with a particle swarm optimization algorithm to perform local optimization.

[0017] Preferably, according to step S4, the final quality confirmation process includes performing statistical analysis on the optimization feedback data, calculating the mean and standard deviation of each quality indicator, and evaluating the quality indicators according to industry standards and enterprise specifications.

[0018] Preferably, a diester oil quality detection system based on multi-source data is applicable to any one of the above-mentioned diester oil quality detection methods based on multi-source data, and the system comprises: Data acquisition module, used to collect multi-source heterogeneous data sets; Feature extraction module, used to perform feature extraction and dimensionality reduction on multi-source heterogeneous data sets; Dynamic weight allocation module, used to build a dynamic weight allocation model and generate a multi-source data fusion weight matrix; Real-time analysis module, used to perform real-time analysis on diester oil quality feature vectors and generate quality assessment time series data; Classification and correction module, used to classify diester oil samples and perform error compensation on low-confidence quality data; The consistency check module is used to perform bidirectional consistency check on the corrected low-confidence quality data and the high-confidence chain verification results; Optimized path planning module, used to generate quality inspection optimization feedback data; The quality confirmation module is used to perform final quality confirmation on the optimization feedback data and generate a diester oil quality test report; The cloud storage module is used to upload the diester oil quality test report to the cloud storage platform and display the test results through a visual interface.

[0019] The beneficial effects of the present invention are as follows: 1. The multi-source data-based diester oil quality detection method and system described in this invention overcomes the limitations of a single data source by integrating spectral data, chemical composition data, and process parameter data. This enables comprehensive diester oil quality detection and significantly improves detection accuracy (experimental data shows a 15%-20% improvement in detection accuracy compared to traditional methods). Furthermore, by employing a dynamic weight allocation model, the weights of each data source are automatically adjusted based on the real-time error values, preventing fluctuations in a single data source from interfering with the overall results and ensuring the stability of the detection results (particularly in complex production environments, data fluctuation tolerance is increased by 30%). 2. The diester oil quality detection method and system based on multi-source data described in this invention uses a support vector machine algorithm to classify data into high- and low-confidence categories. An adaptive correction algorithm (with a dynamic adjustment range of 0.1-0.5) is employed for low-confidence data, effectively reducing the impact of data noise and lowering the error rate of marginal data by over 40%. Furthermore, a sliding time window technique (with an adjustable window length of 10-30 seconds) is employed for real-time monitoring, ensuring rapid response to quality fluctuations and meeting the needs of industrial online detection. 3. The diester oil quality detection method and system based on multi-source data, described in this invention, uses a bidirectional consistency check mechanism to compare corrected low-confidence data with high-confidence verification results, ensuring the reliability of detection results (consistency check pass rate ≥ 98%) and effectively identifying potential abnormal data. Furthermore, by combining the global search capabilities of a genetic algorithm with the localized fine-grained optimization of a particle swarm algorithm, it achieves intelligent tuning of detection parameters, increasing system convergence speed by 25% and shortening detection cycles by 15%. 4. The diester oil quality testing method and system based on multi-source data described in this invention automatically generates quality testing reports that comply with industry standards and corporate specifications, including key statistical indicators such as mean and standard deviation, reducing manual analysis costs (report generation time is shortened from 2 hours to 5 minutes). Furthermore, test results are uploaded to the cloud in real time and displayed through a visual interface, supporting historical data query and comparative analysis, providing data support for quality traceability and process optimization (data retrieval efficiency is improved by 90%). BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The present invention will be further described below with reference to the accompanying drawings.

[0021] Figure 1 It is a schematic flow chart of the method of the present invention; Figure 2 It is a calculation flow chart of the dynamic weight allocation model in the present invention; Figure 3 It is a working principle diagram of the adaptive correction algorithm in the present invention; Figure 4 It is a flow chart of bidirectional consistency checking and optimized path planning in the present invention; Figure 5 It is a schematic diagram of the system visualization interface in the present invention. DETAILED DESCRIPTION

[0022] In order to make the technical means, creative features, objectives and effects achieved by the present invention easier to understand, the present invention is further described below in conjunction with specific implementation methods.

[0023] like Figure 1 As shown, the overall process of the method of the present invention includes the overall steps from multi-source data collection to generating a diester oil quality detection completion report.

[0024] In practical applications, it is first necessary to collect multi-source heterogeneous data sets through various sensors and equipment. These data sets include spectral data, chemical composition data, and process parameter data. Spectral data can be obtained through a near-infrared spectrometer, whose wavelength range is usually 900-2500 nanometers and can reflect the molecular structure information of the diester oil sample. Chemical composition data is measured by high-performance liquid chromatography or gas chromatography to analyze the concentration of the main components in the diester oil. Process parameter data comes from the real-time monitoring system during the production process and includes key variables such as temperature, pressure, and flow rate.

[0025] The frequency of data collection needs to be set according to actual needs to ensure that the temporal resolution of the data is high enough. After the data collection is completed, feature extraction and dimensionality reduction are performed on the multi-source heterogeneous data set to generate a diester oil quality feature vector. Principal component analysis (PCA) is used for feature extraction to perform a linear transformation on the original data and retain the principal components with the largest variance, thereby reducing the data dimension and reducing redundant information. Subsequently, a dynamic weight allocation model is constructed based on the generated diester oil quality feature vector to generate a multi-source data fusion weight matrix.

[0026] like Figure 2 As shown, the calculation process of the dynamic weight allocation model includes the following steps: First, calculate the error value of each data source , the error value can be calculated by the deviation from the standard reference data; then according to the formula Calculate the weight of each data source , where the adjustment parameters The value range is 0.1 to 1.0, and the specific value can be adjusted according to the actual application scenario. The physical meaning of this formula is that the error is nonlinearly mapped through the exponential function, so that the data source with smaller error has a higher weight, thereby improving the accuracy of data fusion.

[0027] In step S2, real-time analysis of diester oil quality feature vectors is performed based on a multi-source data fusion weight matrix to generate quality assessment time series data. The core of this real-time analysis lies in the continuous monitoring of diester oil samples using a sliding time window technique, with the window length typically set between 10 and 30 seconds, with a step size of 1 to 5 seconds. Based on this, diester oil samples are classified based on the quality assessment time series data to generate high-confidence quality data and low-confidence quality data. The classification algorithm utilizes a support vector machine (SVM) with a radial basis function (RBF) kernel function. Hyperparameters are optimized through cross-validation to improve classification accuracy. For low-confidence quality data, an adaptive correction algorithm is used to compensate for errors, generating corrected low-confidence quality data.

[0028] like Figure 3 As shown, the working principle of the adaptive correction algorithm includes the following steps: first determine the original quality data and reference standard values , then according to the formula Correction is performed, where the correction factor The value range of is 0.1 to 0.5, and the specific value needs to be adjusted according to the error distribution characteristics; the meaning of this formula is to gradually narrow the gap between the original data and the standard value by introducing the reference standard value and the dynamic correction coefficient, thereby effectively reducing the error of low-confidence data; for high-confidence quality data, the fast chain verification method is used to generate high-confidence chain verification results; the core of the fast chain verification is to verify the high-confidence data step by step through a multi-layer neural network model, and the output of each layer of the network is used as the input of the next layer to finally generate the verification result.

[0029] In step S3, a bidirectional consistency check is performed on the corrected low-confidence quality data and the high-confidence chain verification result to generate consistency check data.

[0030] like Figure 4 As shown, the process of bidirectional consistency verification includes the following steps: first, the corrected low-confidence quality data is compared with the high-confidence chain verification result, and the deviation between the two is calculated; then, whether the consistency condition is met is determined based on the size of the deviation. If the deviation is less than the preset threshold, the data is considered consistent, otherwise the path planning needs to be further optimized; the goal of optimizing path planning is to make the corrected low-confidence quality data and the high-confidence chain verification result as close as possible by adjusting the parameters of the weight distribution model and the correction algorithm, thereby generating quality detection optimization feedback data; the specific implementation method of optimizing path planning includes using a genetic algorithm to perform a global search of the parameters, and combining the particle swarm optimization algorithm for local optimization to improve the efficiency and accuracy of parameter adjustment.

[0031] In step S4, the quality inspection optimization feedback data is subjected to final quality confirmation to obtain a diester oil quality inspection report; the final quality confirmation process includes the following steps: first, statistical analysis is performed on the optimization feedback data to calculate the mean and standard deviation of each quality indicator; then, the quality indicators are evaluated according to industry standards and enterprise specifications to generate a detailed inspection report; the content of the inspection report includes key parameters such as the main component concentration, impurity content, and physical and chemical properties of the diester oil, as well as the corresponding qualified judgment results; then, the diester oil quality inspection report is uploaded to the cloud storage platform, and the inspection results are displayed through a visual interface to generate a diester oil quality inspection completion report.

[0032] like Figure 5 As shown, the design of the system visualization interface focuses on user experience. The core content of the test report is displayed on the left side of the interface, and the interactive function of data charts is provided on the right side. Users can view detailed data by clicking on the charts; the cloud storage platform adopts a distributed file system architecture to ensure the security and reliability of data, while supporting multi-user concurrent access and permission management.

[0033] In actual applications, the method and system of the present invention have been piloted in a large oil and fat production enterprise. During the pilot process, more than 1,000 batches of diester oil sample data were collected, covering different production processes and raw material sources. Experimental results show that the method of the present invention is significantly superior to traditional single-data source detection methods in terms of detection accuracy, with the average error reduced by more than 30%. In terms of detection efficiency, the application of real-time analysis and rapid chain verification technology has shortened the single-batch detection time to 50% of the original time. In terms of automation level, the system can automatically complete the entire process from data collection to report generation, greatly reducing manual intervention. In addition, through the combination of cloud storage platform and visual interface, enterprise managers can view test results anytime and anywhere, and optimize production processes and quality management strategies based on data analysis results.

[0034] In summary, the present invention significantly improves the accuracy, real-time performance, and automation level of diester oil quality detection through multi-source data fusion, real-time analysis, bidirectional consistency verification, and optimized path planning, meeting the needs of modern industry for efficient and intelligent detection.

[0035] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the foregoing embodiments. The foregoing embodiments and descriptions are merely illustrative of the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A diester oil quality detection method based on multi-source data, characterized by: According to the steps including: Step S1: collecting a multi-source heterogeneous data set, wherein the multi-source heterogeneous data set includes spectral data, chemical composition data, and process parameter data; performing feature extraction and dimensionality reduction processing on the multi-source heterogeneous data set to generate a diester oil quality feature vector; constructing a dynamic weight allocation model based on the diester oil quality feature vector to generate a multi-source data fusion weight matrix; Step S2: Performing real-time analysis on the diester oil quality feature vector based on the multi-source data fusion weight matrix to generate quality assessment time series data; classifying the diester oil samples based on the quality assessment time series data to generate high-confidence quality data and low-confidence quality data; Adopting adaptive correction algorithm to compensate the error of low-confidence quality data and generate corrected low-confidence quality data; performing fast chain verification on high-confidence quality data and generating high-confidence chain verification results; Step S3: Perform a bidirectional consistency check on the corrected low-confidence quality data and the high-confidence chain verification result to generate consistency check data; Based on the consistency check data, the corrected low-confidence quality data and high-confidence chain verification results are optimized for path planning to generate quality inspection optimization feedback data; Step S4: Perform final quality confirmation on the quality inspection optimization feedback data to obtain a diester oil quality inspection report; upload the diester oil quality inspection report to the cloud storage platform, and display the inspection results through a visual interface to generate a diester oil quality inspection completion report.

2. The diester oil quality detection method based on multi-source data according to claim 1, characterized in that: According to step S1, the dynamic weight allocation model uses the following formula to calculate the weight of the i-th data source: ； in, Indicates the The weight of each data source, Indicates the The error value of each data source, To adjust the parameters, The total number of data sources.

3. The diester oil quality detection method based on multi-source data according to claim 1, characterized in that: According to step S2, the adaptive correction algorithm uses the following formula to perform error compensation on low confidence quality data: ； in, represents the corrected quality data, represents the original quality data, Indicates the reference standard value, is the correction factor, and its value range is 0.1 to 0.

5.

4. The diester oil quality detection method based on multi-source data according to claim 1, characterized in that: According to step S1, the feature extraction and dimensionality reduction process uses principal component analysis to process the multi-source heterogeneous data set, and retains several principal components with the largest variance to generate diester oil quality feature vectors.

5. The diester oil quality detection method based on multi-source data according to claim 1, characterized in that: According to step S2, the real-time analysis uses a sliding time window technology to continuously monitor the diester oil quality feature vector, with the window length set to 10 seconds to 30 seconds and the step length being 1-5 seconds.

6. The diester oil quality detection method based on multi-source data according to claim 1, characterized in that: According to step S2, the classification process uses a support vector machine algorithm to classify the diester oil samples, the radial basis function is selected as the kernel function, and the hyperparameters are optimized through cross-validation.

7. The diester oil quality detection method based on multi-source data according to claim 1, characterized in that: According to step S3, the bidirectional consistency check process includes comparing the corrected low-confidence quality data with the high-confidence chain verification result, calculating the deviation between the two, and judging whether the consistency condition is met based on the size of the deviation.

8. The diester oil quality detection method based on multi-source data according to claim 1, characterized in that: According to step S3, the optimization path planning uses a genetic algorithm to perform a global search for parameters and combines it with a particle swarm optimization algorithm to perform local optimization.

9. The diester oil quality detection method based on multi-source data according to claim 1, characterized in that: According to step S4, the final quality confirmation process includes statistical analysis of the optimization feedback data, calculation of the mean and standard deviation of each quality indicator, and evaluation of the quality indicators according to industry standards and corporate specifications.

10. A diester oil quality detection system based on multi-source data, characterized by: The system is applicable to a diester oil quality detection method based on multi-source data according to any one of claims 1 to 9, and the system comprises: Data acquisition module, used to collect multi-source heterogeneous data sets; Feature extraction module, used to perform feature extraction and dimensionality reduction on multi-source heterogeneous data sets; Dynamic weight allocation module, used to build a dynamic weight allocation model and generate a multi-source data fusion weight matrix; Real-time analysis module, used to perform real-time analysis on diester oil quality feature vectors and generate quality assessment time series data; Classification and correction module, used to classify diester oil samples and perform error compensation on low-confidence quality data; The consistency check module is used to perform bidirectional consistency check on the corrected low-confidence quality data and the high-confidence chain verification results; Optimized path planning module, used to generate quality inspection optimization feedback data; The quality confirmation module is used to perform final quality confirmation on the optimization feedback data and generate a diester oil quality test report; The cloud storage module is used to upload the diester oil quality test report to the cloud storage platform and display the test results through a visual interface.

Images