Method and system for detecting impurities in glucosamine production process

Abstract

The present invention discloses a method and system for detecting impurities in a glucosamine production process, and relates to the field of detection technology. The method comprises: obtaining quality control samples of multiple steps by sampling, extracting an impurity type set in combination with a production log, and configuring an impurity detection model and a call table. The cumulative regularity of sample impurities is analyzed to locate key steps and their corresponding impurity types, and based on this, process parameters are optimized and an online detection module is configured. The optimized process parameters are applied in key steps, and dynamic monitoring and early warning of impurity deviations are achieved through the online detection module. The technical problem of insufficient impurity control in the existing glucosamine production process is solved, and the accuracy and efficiency of impurity control in glucosamine production are improved, thereby ensuring the consistency and stability of product quality.

Description

Technical Field

[0001] The present application relates to the field of detection technology, and in particular to a method and system for detecting impurities in the production process of glucosamine. Background Art

[0002] Glucosamine is an important raw material for medicine and health products. The stability of product purity and quality during its production process is crucial to the safety and effectiveness of the final application. However, due to the accumulation, transformation or retention of various impurities during the production process, it is often difficult for the purity of the final product to meet high standards. Most existing technologies use single-point detection and offline analysis methods, which cannot fully reflect the changes in impurities during the production process and are difficult to respond in time, affecting production optimization and control. Especially in multi-step production, it is still a difficult problem to accurately locate key process steps and their corresponding impurity types to achieve efficient impurity removal. Therefore, there is an urgent need for a real-time monitoring and optimization control method that combines production process data to improve product quality and production efficiency while reducing production costs.

[0003] At present, relevant technologies still have the technical problem of insufficient impurity control during the production of glucosamine. Summary of the Invention

[0004] The present application solves the technical problem of insufficient impurity control in the existing glucosamine production process by providing a method and system for detecting impurities in the glucosamine production process.

[0005] This application provides a method for detecting impurities in the glucosamine production process, including:

[0006] By sampling during the trial production of glucosamine, multiple quality control samples of multiple production steps are obtained, wherein the multiple production steps have multiple execution order identifiers; after obtaining a sample impurity type set through an interactive production log, an impurity detection model is configured according to the sample impurity type set to obtain a detection model call table; an impurity accumulation law analysis is performed on the multiple quality control samples, and K groups of key impurity types of K key steps are located based on the analysis results, wherein the K key steps have K execution order identifiers; with the K groups of key impurity types as impurity removal guides, the K key steps are coupled optimized according to the K execution order identifiers, and K optimized process parameters are output; online detection functions are configured for the K key steps according to the K groups of key impurity types and the detection model call table to obtain K online detection modules; after updating the process parameters of the K key steps using the K optimized process parameters, dynamic detection and early warning of impurity deviations in the glucosamine production process are performed through the K online detection modules.

[0007] This application also provides an impurity detection system during the glucosamine production process, including:

[0008] A quality control sample acquisition module is used to obtain multiple quality control samples of multiple production steps by sampling during the trial production of glucosamine, wherein the multiple production steps have multiple execution order identifiers; an impurity detection model configuration module is used to obtain a sample impurity type set through interactive production logs, configure an impurity detection model according to the sample impurity type set, and obtain a detection model call table; a cumulative law analysis module is used to perform impurity accumulation law analysis on the multiple quality control samples, and locate K groups of key impurity types of K key steps based on the analysis results, wherein the K key steps have K execution order identifiers. order identifier; a coupling optimization module, the coupling optimization module is used to use the K groups of key impurity types as impurity removal guides, perform coupling optimization on the K key steps according to the K execution order identifiers, and output K optimized process parameters; an online detection function configuration module, the online detection function configuration module is used to perform online detection function configuration on the K key steps according to the K groups of key impurity types and the detection model call table, and obtain K online detection modules; a dynamic detection and early warning module, the dynamic detection and early warning module is used to perform dynamic detection and early warning of impurity deviations in the glucosamine production process through the K online detection modules after the process parameters of the K key steps are updated using the K optimized process parameters.

[0009] The proposed method and system for detecting impurities in the glucosamine production process, proposed in this application, first obtains quality control samples from multiple steps through sampling. A set of impurity types is extracted in conjunction with production logs, and an impurity detection model and call table are configured. Cumulative patterns of sample impurities are analyzed to identify key steps and their corresponding impurity types. Based on this, process parameters are optimized and an online detection module is configured. The optimized process parameters are applied to key steps, and dynamic monitoring and early warning of impurity deviations are achieved through the online detection module. This achieves the technical effect of improving the precision and efficiency of impurity control in glucosamine production and ensuring the consistency and stability of product quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings of the embodiments of the present invention are briefly introduced below. Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed in precise order. Instead, various steps may be processed in reverse order or simultaneously as needed. Furthermore, other operations may be added to these processes, or one or more operations may be removed from these processes.

[0011] Figure 1Schematic diagram of the process of detecting impurities in the glucosamine production process provided in the embodiments of the present application.

[0012] Figure 2 This is a schematic diagram of the structure of the impurity detection system in the glucosamine production process provided in the embodiment of the present application.

[0013] Explanation of the reference numerals: quality control sample acquisition module 10 , impurity detection model configuration module 20 , accumulation rule analysis module 30 , coupling optimization module 40 , online detection function configuration module 50 , dynamic detection warning module 60 . DETAILED DESCRIPTION

[0014] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below.

[0015] In order to make the purpose, technical solutions and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be regarded as limiting this application. All other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0016] In the following description, reference is made to “some embodiments”, which describes a subset of all possible embodiments, but it will be understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict, and the terms “first\second” involved are merely used to distinguish similar objects and do not represent a specific ordering of the objects. The terms “including” and “having” and any variations are intended to cover non-exclusive inclusions. For example, a process, method, system, product or server that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application belongs. The terms used herein are for the purpose of describing the embodiments of this application only.

[0017] The present invention provides a method and system for detecting impurities in the production process of glucosamine. Figure 1 As shown, the method includes:

[0018] Step S100, by sampling during the trial production of glucosamine, a plurality of quality control samples of a plurality of production steps are obtained, wherein the plurality of production steps have a plurality of execution order identifiers. Specifically, before the trial production of glucosamine, the production process is first planned and sorted out, each link is clarified and assigned an execution order identifier. Then, the sampling points are determined according to the production steps and key links, and liquid samples are collected as quality control samples using professional equipment specifications at different stages, such as after raw material processing, during the reaction, before and after purification of intermediate products, and after synthesis of the final product. After collection, the sample is labeled and recorded with information, including production step identification, sampling time, location and operator, etc., and an archive is established to record the appearance and preliminary physical and chemical properties. The sample is then properly stored and the pre-treatment reagents, instruments and equipment and operating procedures are prepared according to the test requirements to prevent the sample from deteriorating or being interfered with, and to ensure the accuracy and comparability of the test results.

[0019] Step S200: After obtaining a sample impurity type set through interactive production logs, an impurity detection model is configured according to the sample impurity type set to obtain a detection model call table. Specifically, the production logs of glucosamine are first interacted with to mine and organize the sample impurity type set. Then, for the M types of sample impurity types, network call data is integrated and classified to construct M sample impurity spectral atlases containing M sample impurity content identification sets. Then, with the M types of sample impurity types as constraints, composite and single-dimensional impurity spectra are obtained through network. These spectra are used to construct a layer separation model. Based on the spectral atlas set and the content identification set, M impurity content identification branches are constructed and the layer separation model and the identification branch are connected to form an impurity detection model. Finally, the M types of sample impurity types and the M impurity content identification branches are associated and stored to construct a detection model call table, so as to detect impurities efficiently and accurately during the production process.

[0020] In one possible implementation, after obtaining a set of sample impurity types from an interactive production log, an impurity detection model is configured based on the set of sample impurity types to obtain a detection model call table. Step S200 further includes step S210, in which networked data is called based on M sample impurity types in the set of sample impurity types to obtain M sample impurity spectral atlases for the M sample impurity types, wherein the M sample impurity spectral atlases have M sets of sample impurity content identifiers. Specifically, based on the M sample impurity types in the set of sample impurity types, a networked data call program is initiated. Connections are made to professional chemical databases, spectral data sharing platforms of scientific research institutions, and large-scale data repositories within the industry. Within these vast data resources, precise search algorithms and data screening techniques are utilized to search and match each of the M sample impurity types. For example, if one of the sample impurity types is a metal ion impurity, a database search is performed for spectral data of the metal ion in different chemical environments and at different concentrations. The spectral data corresponding to each sample impurity type found is organized and classified, thereby obtaining M sample impurity spectral atlases. Furthermore, during the construction of these atlases, M sets of sample impurity content identification sets were simultaneously established. This was achieved by analyzing and extracting a large amount of spectral data with precise impurity content annotations from the database, determining the quantitative relationship between different spectral features and specific impurity levels. This allowed the impurity content to be inferred from the spectral features when testing unknown samples.

[0021] Step S220, using the M types of sample impurity types as constraints, performs networked data calls to obtain multiple sample composite impurity spectra and multiple sets of sample single-dimensional impurity spectra. Specifically, using the M types of sample impurity types as constraints, the networked data resources are deeply explored, focusing on obtaining diverse spectra, including sample composite impurity spectra reflecting the mixing of multiple impurities, screening spectral data containing a mixture of multiple known impurities, and simulating the effects of impurity coexistence and interaction on spectral characteristics in glucosamine production; there are also multiple sets of sample single-dimensional impurity spectra focusing on the specific dimensional characteristics of a single impurity. The data is derived from information under a specific wavelength range, polarization direction, or spectral analysis technology. Due to the limitations of the impurity composition of the composite impurity spectrum, it can be decomposed from the composite spectrum or extracted from the single impurity spectrum resource to accurately present the specific dimensional spectral characteristics of a single impurity.

[0022] Step S230: Construct a layer separation model based on the multiple sample composite impurity spectra and the multiple groups of sample single-dimensional impurity spectra. Specifically, a layer separation model is constructed based on the multiple sample composite impurity spectra and the multiple groups of sample single-dimensional impurity spectra. Image analysis algorithms and chemometric principles are used to extract composite impurity spectra features, such as peak position, peak shape, peak intensity, and spectral curve trends. Cluster analysis, principal component analysis, and other methods are then used to separate the spectral information of different impurities in the composite spectra into different layers. The sample single-dimensional impurity spectra are used as a reference standard for verification and optimization. After repeated adjustment of model parameters and algorithm structure, the model can accurately separate the spectral information of M or fewer impurities in the composite impurity spectra into corresponding layers, providing a clear and independent spectral data source for impurity content identification.

[0023] Step S240, constructing M impurity content identification branches based on the M sample impurity spectrum atlases and the M sample impurity content identification sets. Specifically, constructing M impurity content identification branches based on the M sample impurity spectrum atlases and the M sample impurity content identification sets. For each impurity content identification branch, the corresponding sample impurity spectrum atlas is used as the main learning data source. Regression analysis algorithms in machine learning are adopted, such as linear regression, multiple regression or support vector regression. The spectral feature data in the sample impurity spectrum atlas is used as the input variable, and the impurity content data in the corresponding sample impurity content identification set is used as the output variable to perform model training. During the training process, the parameters of the model, such as regression coefficients, kernel function parameters, etc., are continuously optimized to improve the degree of fit of the model to the relationship between spectral features and impurity content. After training and verification with a large amount of data, each impurity content identification branch can accurately predict the content of the impurity based on the specific impurity spectral features of the input unknown sample, thereby realizing the content identification function for M types of sample impurity types.

[0024] Step S250 connects the M impurity content identification branches in parallel and identifies them using the M sample impurity types. Then, the output of the layer separation model is connected to the input of the M impurity content identification branches, completing the construction of the impurity detection model. Specifically, the M impurity content identification branches are connected in parallel so that they can independently identify the content of their corresponding impurity types. These M impurity content identification branches are then identified using the M sample impurity types, clarifying the function of each branch and the corresponding impurity type. The output of the layer separation model is connected to the input of the M impurity content identification branches to construct a complete impurity detection model. When spectral data of an unknown sample is input, the layer separation model is first used to separate the impurity spectral information into different layers according to the impurity type. The spectral data of each layer is then input into the corresponding impurity content identification branch. These branches then perform impurity content prediction and analysis, ultimately obtaining content information for the M sample impurity types in the unknown sample, achieving comprehensive and accurate detection of sample impurities.

[0025] Step S260, associate and store the M types of sample impurity types and the M impurity content identification branches to complete the construction of the detection model call table. Specifically, associate and store the M types of sample impurity types and the M impurity content identification branches to construct a detection model call table. In this process, the M types of sample impurity types are used as index keywords, and the corresponding M impurity content identification branches are used as storage content or link addresses pointing to branches. For example, it can be stored in the form of a database table, in which one column records the sample impurity type and the other column records the relevant information of the corresponding impurity content identification branch. In this way, when impurity detection is actually performed, when it is necessary to detect a certain specific sample impurity type, it is only necessary to query the detection model call table to quickly locate the corresponding impurity content identification branch, thereby calling the branch for impurity content detection. This method greatly improves the efficiency and flexibility of the impurity detection model in practical applications, ensuring that impurities in the glucosamine production process can be detected and monitored quickly and accurately.

[0026] Step S300: Analyze the impurity accumulation patterns of the multiple quality control samples and locate K groups of key impurity types of K key steps based on the analysis results, wherein the K key steps have K execution order identifiers. Specifically, first, for multiple liquid quality control samples, use the impurity detection model to divide the first quality control sample into N step subsamples and generate N real-time impurity spectra using an infrared spectrometer. These spectra are synchronized to the layer separation model to separate H real-time single-dimensional spectra of H sample impurity types. According to the corresponding relationship, the impurity content identification branch is input to obtain H real-time impurity contents to constitute the first real-time impurity information. The N real-time impurity spectra are processed accordingly to obtain N real-time impurity information. The extremely valuable first sample impurity content set is aggregated and extracted. The operation is repeated to obtain multiple sample impurity content sets. The impurity distribution matrix is ​​constructed with the sample impurity type set as a constraint. After cumulative amount calculation, cumulative rate analysis, cross-step impurity removal contribution analysis, and single-step weighted calculation, K key steps and K groups of key impurity types are located and their K execution order identifiers are clarified.

[0027] In one possible implementation, an impurity accumulation pattern analysis is performed on the multiple quality control samples, and K groups of key impurity types of K key steps are located based on the analysis results, wherein the K key steps have K execution order identifiers. Step S300 further includes step S310, using the impurity detection model to poll and detect the multiple quality control samples to obtain multiple sample impurity content sets. Specifically, after the quality control liquid samples collected from each step of glucosamine production are sorted out, the impurity detection model is enabled for polling detection, and a sample is divided into N step sub-samples according to key production factors. A high-precision infrared spectrometer is used to generate N real-time impurity spectra containing impurity molecule information. The first real-time impurity spectrum is sent to the layer separation model, and H real-time single-dimensional spectra of H types of sample impurity types are separated according to the spectral feature library and image analysis algorithm. Then, M impurity content identification branches are imported according to the corresponding relationship. With the help of regression analysis, H real-time impurity contents are obtained to constitute the first real-time impurity information. N real-time impurity spectra are processed in this way to obtain N real-time impurity information. The extreme value information is summarized, integrated and extracted to construct the first sample impurity content set. The operation is repeated to complete all sample detections to obtain multiple sample impurity content sets.

[0028] Step S320, using the sample impurity type set as a constraint, constructs and generates an impurity distribution matrix based on the multiple sample impurity content sets. Specifically, the impurity distribution matrix is ​​constructed based on the sample impurity type set as a constraint, wherein each row represents a specific impurity type such as heavy metal impurities, solvent residues, oxidation byproducts, etc., and each column corresponds to each production step from raw material pretreatment to final product molding. The matrix element values ​​are derived from multiple sample impurity content sets, that is, the detection concentration or mass data of a certain impurity in a specific production step. For example, the value of the column corresponding to the heavy metal impurity in the decalcification step is the average concentration or total mass of that step in all samples. This construction method integrates scattered data, makes the impurity distribution clear, and lays a solid data foundation for analyzing accumulation rules.

[0029] Step S330, perform cumulative regularity analysis based on the impurity distribution matrix, and locate the K key steps and K groups of key impurity types based on the analysis results. Specifically, based on the impurity distribution matrix, first calculate the impurity accumulation amount, and add the column element values ​​of each production step according to the impurity type in each row of the matrix, such as calculating the cumulative total amount of heavy metal impurities in multiple steps, so as to understand the overall accumulation and screen the key impurity types; then perform cumulative rate analysis, calculate the rate of change of impurity content in adjacent steps to determine the growth rate, such as the rate of change of heavy metal impurities between decalcification and deacidification steps, to find the key production interval; then carry out cross-step impurity removal contribution analysis, evaluate the effect of specific steps on impurity removal and subsequent impact, such as the effect of purification steps on solvent residues, and determine the key hub steps to improve impurity removal efficiency.

[0030] Step S340, according to the K key steps, K sample impurity content sets and the K execution order identifiers are obtained from the multiple execution order identifiers and multiple sample impurity content sets. Specifically, a single-step weighted calculation is carried out by comprehensively considering factors such as the impurity accumulation amount, accumulation rate and cross-step impurity removal contribution. According to the degree of influence of each factor on product quality and production process, weights are assigned to impurity types and production steps. High weights are assigned to those with high accumulation amount, fast rate and low impurity removal contribution. After weighted calculation, K key steps and K groups of key impurity types are accurately located from numerous steps and impurity types. For example, a production step with high accumulation amount, fast rate and low impurity removal contribution of a specific impurity is a key step, and the corresponding impurity is a key impurity type. At the same time, the execution order identifier is extracted according to the K key steps, and the corresponding sample impurity content set is called to provide targets and data support for process optimization and impurity control.

[0031] In one possible implementation, the impurity detection model is used to round-robinly detect multiple quality control samples to obtain multiple sample impurity content sets. Step S310 further includes step S311, where the first quality control sample is divided into N step sub-samples, and then an infrared spectrometer is used to generate N real-time impurity spectra for each of the N step sub-samples. Specifically, the first quality control sample is precisely divided into N step sub-samples based on key links or stage characteristics in the production process. Because each step has a corresponding liquid quality control sample, this division can meticulously reflect the changes in product and impurities at different production stages. After the division is completed, the N step sub-samples are individually tested using a high-precision infrared spectrometer. The infrared spectrometer generates N real-time impurity spectra for each of the N step sub-samples by emitting infrared light within a specific wavelength range and detecting the sample's absorption of this light. These spectra are presented as unique spectral curves, containing a wealth of information, such as the position and intensity of vibrational absorption peaks of specific chemical bonds in the impurity molecules. This information can serve as an important basis for identifying impurity types and levels.

[0032] In step S312, the first real-time impurity spectrum is synchronized with the layer separation model of the impurity detection model. Layer separation is performed based on impurity type, resulting in H real-time single-dimensional spectra for H sample impurity types, where H is a positive integer less than or equal to M. Specifically, the first real-time impurity spectrum is transmitted to the pre-built layer separation model in the impurity detection model. This model, leveraging powerful algorithms and a rich library of impurity spectral signatures, performs precise layer separation on the composite first real-time impurity spectrum based on impurity type. During the separation process, the model decomposes the first real-time impurity spectrum into H real-time single-dimensional spectra for H sample impurity types based on the characteristic differences exhibited by different impurity types in the spectra, such as the position, shape, and relative intensity of their characteristic absorption peaks. Here, H is a positive integer less than or equal to M, meaning that the number of separated impurity types does not exceed the predetermined total number of sample impurity types, M. This ensures the accuracy and validity of the separation results. Each single-dimensional spectrum focuses on a specific impurity type, laying a solid foundation for subsequent precise impurity content identification.

[0033] In step S313, based on the correspondence between the H sample impurity types and the M impurity content identification branches in the impurity detection model, the H real-time single-dimensional spectra are directed to H impurity content identification branches for targeted impurity content identification, thereby obtaining H real-time impurity contents, wherein the H real-time impurity contents constitute the first real-time impurity information. Specifically, based on the pre-set correspondence between the H sample impurity types and the M impurity content identification branches in the impurity detection model, the H real-time single-dimensional spectra are accurately directed to the corresponding H impurity content identification branches. These impurity content identification branches are intelligent modules trained based on a large amount of sample data, utilizing advanced machine learning algorithms such as linear regression, multivariate regression, or neural network regression. Upon inputting the single-dimensional spectra, the branching model extracts key feature data from the spectra and, based on the mathematical relationship between this data and samples with known impurity content, performs targeted impurity content identification calculations, ultimately obtaining the H real-time impurity contents. These real-time impurity contents together constitute the first real-time impurity information, which can reflect the impurity content of the first quality control sample in the current production step in real time, providing timely and accurate data support for quality monitoring during the production process.

[0034] Step S314, and so on, the impurity detection model is used to analyze the N real-time impurity spectra to obtain N real-time impurity information. Specifically, according to the same process and method as above, the impurity detection model is used to conduct in-depth analysis on the N real-time impurity spectra in turn. For each real-time impurity spectra, the operation steps of layer separation, single-dimensional spectrum acquisition and impurity content directional identification are repeated to obtain N complete real-time impurity information. These real-time impurity information covers various aspects of information such as the impurity type, content and related spectral characteristics of the first quality control sample in each subdivided production step. They are a comprehensive and detailed description of the impurity situation of the first quality control sample, and provide rich data resources for subsequent further analysis and processing.

[0035] Step S315, after aggregating the N real-time impurity information based on the impurity type, the impurity content extreme value extraction is performed to obtain the first sample impurity content set. Specifically, after obtaining N real-time impurity information, this information is aggregated based on the impurity type. The real-time impurity information of the same impurity type in different production steps is integrated to form an information sequence about each impurity type in the entire first quality control sample production process. Then, the impurity content extreme value extraction is performed from these information sequences. For example, the maximum content value, minimum content value and corresponding production step and other key information of a certain impurity in all production steps are found. In this way, a large amount of real-time impurity information is condensed into a more representative and instructive first sample impurity content set. This content set not only contains the content range information of various impurity types in the first quality control sample, but also indirectly reflects the changing trend and key control points of impurities in the production process, providing an important reference basis for the optimization of production processes and impurity control.

[0036] Step S316, and so on, adopt the impurity detection model to poll and detect the multiple quality control samples to obtain the multiple sample impurity content sets. Specifically, taking the detection and analysis process of the first quality control sample as an example, and so on, adopt the impurity detection model to perform comprehensive polling detection on all multiple quality control samples. For each quality control sample, the above-mentioned whole set of operation processes from sub-sample division, spectrum generation, layer separation, impurity content identification to impurity content set generation are repeated. After completing the detection of all quality control samples, the sample impurity content sets obtained from each sample are summarized and integrated, and finally multiple sample impurity content sets that can reflect the impurity conditions in different production steps of the entire glucosamine production process are successfully obtained. These data sets provide a comprehensive and detailed data basis for subsequent in-depth analysis of the impurity accumulation rules in the production process, positioning key steps and key impurity types, and formulating targeted impurity removal strategies, and are the key to achieving efficient quality control and process optimization in the glucosamine production process.

[0037] In one possible implementation, a cumulative regularity analysis is performed according to the impurity distribution matrix, and the K key steps and K groups of key impurity types are located based on the analysis results. Step S330 further includes step S331, in which cumulative amount calculation is performed according to the impurity distribution matrix to obtain M impurity cumulative amount sequences corresponding to the M sample impurity types. Specifically, the impurity cumulative amount calculation work is first carried out based on the constructed impurity distribution matrix. Each row of the impurity distribution matrix represents a sample impurity type, each column corresponds to a production step, and the matrix element value is the detected concentration or mass of the impurity in the step. For each of the M sample impurity types, along its corresponding row, the element values ​​(i.e., impurity concentration or mass data) in each production step column are accumulated and summed. For example, for the first sample impurity type, its impurity content in the first production step and the impurity content in the second production step are added in sequence until the accumulation of all production steps is completed, thereby obtaining the cumulative total amount of this impurity type. According to this method, the M types of sample impurity are calculated one by one, and finally M impurity accumulation sequences corresponding to the M types of sample impurity are obtained. These sequences clearly show the overall accumulation of each impurity in the entire production process.

[0038] Step S332, performing cumulative rate analysis based on the impurity distribution matrix to obtain M cumulative rate sequences. Specifically, after obtaining the impurity accumulation sequence, performing cumulative rate analysis based on the impurity distribution matrix. For each sample impurity type, calculate the rate of change of the impurity content between adjacent production steps, and the calculation method is to divide the concentration change of the impurity before and after the step by the execution time of the step. For example, for the second sample impurity type, calculate the cumulative rate from the third production step to the fourth production step, and express it as (the impurity concentration of the fourth production step - the impurity concentration of the third production step) / the execution time between these two steps. According to this calculation method, for each sample impurity type, calculations are performed between each adjacent production step to obtain a sequence of concentration change rates containing multiple steps, and finally obtain M cumulative rate sequences. These sequences can reflect the growth rate of impurities in different production stages, which helps to find the key intervals of abnormal impurity growth.

[0039] Step S333, performing a cross-step impurity removal contribution analysis based on the M impurity accumulation sequences, and obtaining M impurity removal contribution rate sequences. Specifically, a cross-step impurity removal contribution analysis is performed based on the existing M impurity accumulation sequences. For each sample impurity type, the effect of removing the impurity in a certain specific production step and the influence of this removal effect on the impurity content in the subsequent production steps are examined. For example, for the third sample impurity type, the amount of its removal in the purification step and the degree of reduction in the impurity content in the concentration step after removal are analyzed. By comparing the changes in the impurity content in different production steps, the contribution value of each step to the removal of specific impurities is quantified, and then M impurity removal contribution rate sequences are obtained. These sequences can clarify which production steps play a key role in the removal of specific impurities in the entire impurity removal process, and provide an important basis for subsequent process optimization.

[0040] Step S334, a single-step weighted calculation is performed on the M impurity accumulation sequences, M accumulation rate sequences and M impurity removal contribution rate sequences to obtain M impurity removal priority sequences. Specifically, a single-step weighted calculation is performed on the M impurity accumulation sequences, M accumulation rate sequences and M impurity removal contribution rate sequences. The weight is determined based on the importance of each factor to the product quality and production process. For example, a higher weight is given to the impurity type with a high accumulation amount, a fast accumulation rate and a low contribution to impurity removal across steps in the corresponding production step. For each production step and each impurity type, the data information in these three sequences is comprehensively considered, and the calculation is performed according to the set weight calculation method to obtain a value that comprehensively reflects the impurity removal priority of the impurity type in the production step. By calculating all impurity types and production steps, M impurity removal priority sequences are finally obtained. These sequences can more comprehensively and accurately measure the importance of each production step to the impurity removal of different impurity types.

[0041] Step S335 involves performing a top-level call from the M impurity removal priority sequences to obtain M impurity removal priority steps. Specifically, the top-level call is performed from the M impurity removal priority sequences, i.e., the production step corresponding to the maximum value in each impurity removal priority sequence is found. These production steps are those with the highest impurity removal priority for the corresponding impurity type. By extracting these steps, M impurity removal priority steps are obtained. These steps play a crucial role in controlling specific impurity types throughout the entire production process and are the focus of subsequent process optimization and impurity control.

[0042] Step S336: Aggregate the M priority impurity removal steps to obtain the K key steps. Specifically, the M priority impurity removal steps are aggregated, and any duplicate steps are removed to obtain the K key steps. These key steps cover the most critical production links for controlling various impurity types throughout the entire production process.

[0043] Step S337 aggregates the M sample impurity types based on the mapping relationship between the M priority impurity removal steps and the K key steps to obtain the K groups of key impurity types. Specifically, based on the mapping relationship between the M priority impurity removal steps and the K key steps, the sample impurity types associated with the K key steps are identified and aggregated to obtain the K groups of key impurity types. The identification of these key steps and key impurity types provides clear goals and directions for further targeted process optimization and impurity control in the glucosamine production process, helping to improve product quality and production efficiency.

[0044] Step S400: Using the K groups of key impurity types as the impurity removal guide, the K key steps are coupled and optimized according to the K execution order identifiers, and K optimized process parameters are output. Specifically, first, using the K groups of key impurity types as the impurity removal guide and determining the information of each key step based on the K execution order identifiers, for each key step and its corresponding impurity type, the target impurity concentrations of M sample impurity types are interactively obtained and the impurity concentration constraints of the key steps are determined accordingly. At the same time, the initial process parameters are interactively obtained, and the sample impurity concentration set, sample process parameter set, and sample concentration change set are collected online with the key steps and impurity types as constraints. An impurity prediction model is constructed through multivariate regression analysis, and the initial process parameters are expanded using a preset parameter adjustment scale to obtain updated process parameters. Reliable process parameters are screened out using the prediction model combined with the impurity concentration constraints. In this way, the K key steps are optimized one by one to obtain K groups of reliable process parameters. Then, through coupled optimization such as permutation and combination enumeration, simulated production, and cost analysis, K optimized process parameters are finally output.

[0045] In one possible implementation, the K groups of key impurity types are used as impurity removal guides, the K key steps are coupled and optimized according to the K execution order identifiers, and K optimized process parameters are output. Step S400 further includes step S410, interactively obtaining M target impurity concentrations of the M sample impurity types. Specifically, first, by interacting with a professional database, an industry standard data interface, or a large amount of internally accumulated experimental data records, M target impurity concentrations of M sample impurity types are obtained. These target impurity concentrations are determined based on in-depth research on the quality requirements of glucosamine products and the purity standards generally recognized in the industry. For example, for a certain metal ion impurity, its target impurity concentration may be set to an extremely low one part per million level to ensure the safety and effectiveness of glucosamine products in pharmaceutical or other high-end application fields. These data provide a clear impurity control direction for the optimization of process parameters in subsequent key steps and are an important benchmark for the entire optimization process.

[0046] Step S420, extracting the first key step and the first group of key impurity types from the K key steps and K groups of key impurity types respectively according to the K execution order identifiers. Specifically, based on the K execution order identifiers, the first key step and the first group of key impurity types are accurately extracted from the determined K key steps and K groups of key impurity types. Taking the glucosamine production process as an example, if the K key steps are distributed in the raw material pretreatment, reaction synthesis, intermediate product purification and other links, then the first key step determined according to the execution order identifier may be a specific step in the reaction synthesis link, and the corresponding first group of key impurity types may be by-product impurities that are easily generated during the reaction process or residual impurities that are difficult to remove in the raw materials. This step clarifies the specific object that currently needs to be optimized for process parameters, making subsequent data collection and model construction more targeted.

[0047] Step S430, based on the first group of key impurity types, the first impurity concentration constraint is obtained by calling from the M target impurity concentrations. Specifically, based on the first group of key impurity types, the concentration data that matches it is called from the M target impurity concentrations to obtain the first impurity concentration constraint. This constraint clarifies the concentration range allowed for the first group of key impurity types in the first key step. For example, if the first group of key impurity types contains two impurities, then the first impurity concentration constraint will specify the maximum allowable concentration values ​​of each of the two impurities at the end of the first key step. This provides strict restrictions for the adjustment of subsequent process parameters, ensuring that when optimizing process parameters, the control of impurity concentration will not be neglected due to excessive pursuit of other indicators, and ensuring that product quality is always within a controllable range.

[0048] Step S440, interactively obtain the first initial process parameters of the first key step. Specifically, the first initial process parameters of the first key step are obtained by interacting with the production equipment control system, the process parameter record document or the operator's experience data. These initial process parameters cover the various process conditions involved in this step, such as reaction temperature, reaction pressure, reaction time, reactant concentration ratio, etc. For example, when the first key step is a chemical reaction step, the initial process parameters may include a reaction temperature set to 50 degrees Celsius, a reaction pressure of 2 atmospheres, a reaction time of 3 hours, and a concentration ratio of reactant A and reactant B of 2:1, etc. These data reflect the actual operating parameter settings of the first key step in the current production process and are the starting point for process parameter optimization.

[0049] In step S450, networked data collection is performed using the first key step and the first group of key impurity types as data call constraints to obtain a first sample impurity concentration set, a first sample process parameter set, and a first sample concentration change set. Specifically, using the first key step and the first group of key impurity types as data call constraints, access is made to a wide range of chemical industry data networks, shared data platforms of scientific research institutions, and historical production data warehouses within enterprises to conduct comprehensive networked data collection. In this process, a first sample impurity concentration set is obtained, which contains the actual impurity concentration measurements of the first group of key impurity types in many production processes similar to the first key step; a first sample process parameter set collects various process parameter combinations used in these similar production processes; and a first sample concentration change set records the concentration changes of the first group of key impurity types as the process parameters change. For example, from experimental data of multiple different production batches but similar to the first key step, concentration change data of the first group of key impurity types under different reaction temperature, pressure, and time combinations are extracted. This data will provide sufficient material for building an accurate impurity prediction model.

[0050] Step S460, by performing a multivariate regression analysis on the first sample impurity concentration set, the first sample process parameter set and the first sample concentration change set, a first impurity prediction model is fitted and generated. Specifically, an in-depth multivariate regression analysis is performed on the collected first sample impurity concentration set, the first sample process parameter set and the first sample concentration change set. By using advanced statistical analysis software and data modeling algorithms, a first impurity prediction model is fitted and generated. The model establishes a mathematical relationship between process parameters and impurity concentrations. For example, it can predict the concentration values ​​of the first group of key impurity types at the end of the first key step based on the input process parameters such as reaction temperature, pressure and time. This model is a core tool for subsequent evaluation and optimization of process parameters. It can quickly evaluate the impact of different process parameter combinations on impurity concentrations without conducting actual production tests, greatly improving the efficiency of process parameter optimization.

[0051] Step S470, use the preset parameter adjustment scale to de-expand the first initial process parameters to obtain a first set of updated process parameters. Specifically, the preset parameter adjustment scale is used to de-expand the first initial process parameters to obtain a first set of updated process parameters. The preset parameter adjustment scale is determined based on the understanding of the production process, industry experience and the mastery of equipment performance. For example, if the initial reaction temperature is 50 degrees Celsius, and the preset parameter adjustment scale stipulates that the temperature adjustment range is ±10 degrees Celsius, then the reaction temperature in the updated process parameters may take values ​​of 40 degrees Celsius, 45 degrees Celsius, 55 degrees Celsius, 60 degrees Celsius, etc. In this way, the limitations of the initial process parameters are broken, a wider range of process parameter space is explored, and the possibility of finding a better combination of process parameters is increased.

[0052] Step S480, use the first impurity prediction model to evaluate the first set of updated process parameters to obtain a first set of reliable process parameters. Specifically, the first impurity prediction model is used to evaluate the first set of updated process parameters. The first set of updated process parameters is input into the first impurity prediction model to obtain a first set of predicted concentration change sets, which contains the concentration changes of the first group of key impurity types predicted under different updated process parameter combinations. Then, the first impurity concentration constraint is used to traverse and compare the first set of predicted concentration change sets, and the first set of reliable process parameters whose impurity concentration meets the first impurity concentration constraint are screened out from the first set of updated process parameters. For example, if the first impurity concentration constraint stipulates that the concentration of a certain impurity shall not exceed 0.1%, then in the first set of predicted concentration change sets, only those updated process parameter combinations that make the impurity concentration below 0.1% will be screened out as the first set of reliable process parameters. This step ensures that the optimized process parameters can not only meet the impurity concentration control requirements, but also find the optimal solution within a wider parameter range.

[0053] In step S490, the K groups of key impurity types are used as impurity removal guides, and process parameters of the K key steps are optimized according to the K execution order identifiers, thereby outputting K groups of reliable process parameters. Specifically, according to the above steps, the K groups of key impurity types are used as impurity removal guides, and process parameters of the K key steps are optimized in sequence according to the K execution order identifiers, ultimately outputting K groups of reliable process parameters.

[0054] Step S4100, couples and optimizes the K groups of reliable process parameters, and outputs the K optimized process parameters. Specifically, after completing the independent optimization of each key step, couples and optimizes the K groups of reliable process parameters. Considering that each key step does not exist in isolation, but rather influences and restricts each other, for example, the change in the process parameters of the previous key step may affect the raw material quality or reaction conditions of the next key step, it is necessary to comprehensively coordinate the process parameters of each key step through methods such as permutation and combination enumeration, simulated production and cost analysis. For example, in the simulated production process, try different K groups of reliable process parameter combinations, calculate the indicators such as reagent dosage, energy consumption, impurity removal cycle and production efficiency under each combination, and finally determine a set of optimal process parameter combinations, that is, output K optimized process parameters, which can achieve the best production benefits while achieving effective control of impurities as a whole.

[0055] In one possible implementation, the first set of updated process parameters is evaluated using the first impurity prediction model to obtain a first set of reliable process parameters. Step S480 further includes step S481, using the first key step and the first set of key impurity types as search constraints to search and obtain a first set of key impurity concentrations from the impurity distribution matrix. Specifically, using the first key step and the first set of key impurity types as search constraints, a precise search is performed within the constructed impurity distribution matrix. Each row of the impurity distribution matrix represents an impurity type, and each column corresponds to a production step, storing the detected concentration or mass information of each impurity in the corresponding step. Since the first key step and the first set of key impurity types have been determined, the corresponding cell data can be quickly located using row and column indexing to obtain the first set of key impurity concentrations. For example, if the first key step is the "mid-reaction" step in the glucosamine production process, and the first set of key impurity types includes "side reaction product A" and "residual raw material impurity B", then the data in the cells where the "mid-reaction" column intersects with the "side reaction product A" row and the "residual raw material impurity B" row in the impurity distribution matrix is ​​found; these data are the first set of key impurity concentrations. These concentration data reflect the actual concentration levels of these key impurities at this critical step in past production practices, providing an important reference for subsequent model evaluation.

[0056] Step S482: Synchronize the first set of key impurity concentrations and the first set of updated process parameters to the first impurity prediction model for evaluation, obtaining a first set of predicted concentration change sets. Specifically, the acquired first set of key impurity concentrations and the first set of updated process parameters are synchronously input into the first impurity prediction model for evaluation. The first impurity prediction model is constructed using a large amount of sample data through multivariate regression analysis, establishing a mathematical relationship between process parameters and impurity concentrations. Upon inputting the first set of key impurity concentrations and the first set of updated process parameters, the model performs complex calculations based on its internal algorithms and data relationships to predict how the first set of key impurity concentrations will change when these updated process parameters are adopted, thereby obtaining a first set of predicted concentration change sets. For example, if the first set of updated process parameters includes adjustments to parameters such as reaction temperature, pressure, and reaction time, the model calculates the concentration changes of "side reaction product A" and "residual raw material impurity B" under the new parameters based on these parameter changes and their correlation with impurity concentrations. These values ​​constitute the first set of predicted concentration change sets, which demonstrate the possible trends in impurity concentrations when the updated process parameters are adopted.

[0057] In step S483, the first set of updated process parameters is screened for the first set of reliable process parameters whose impurity concentrations satisfy the first impurity concentration constraint by traversing and comparing the first set of predicted concentration variations using the first impurity concentration constraint. Specifically, the first set of updated process parameters is screened for the first set of reliable process parameters that meet the requirements by traversing and comparing the first set of predicted concentration variations using the first impurity concentration constraint. The first impurity concentration constraint specifies the allowable concentration range for the first set of key impurity types in the first key step, which is set based on product quality standards and production process requirements. During the comparison process, each predicted concentration value in the first set of predicted concentration variations is individually checked to see if it satisfies the first impurity concentration constraint. For example, if the first impurity concentration constraint stipulates that the concentration of "side reaction product A" must not exceed 0.5% and the concentration of "residual raw material impurity B" must not exceed 0.3%, then only if the predicted concentration values ​​of the corresponding impurities in the first set of predicted concentration variations are both below these limits will the corresponding first set of updated process parameters be considered to meet the requirements and thus be screened as the first set of reliable process parameters. These reliable process parameters can not only theoretically ensure that the concentration of key impurities is controlled within a reasonable range, but also reflect the effectiveness verification of the update and adjustment of process parameters, providing feasible parameter selection for the optimization of subsequent production processes.

[0058] In one possible implementation, the K groups of reliable process parameters are coupled and optimized, and the K optimized process parameters are output. Step S4100 further includes step S4101, performing permutation and combination enumeration on the K groups of reliable process parameters to obtain multiple groups of coupled process parameters. Specifically, the K groups of reliable process parameters are first permuted and combined. Because each group covers optimization parameters for multiple key steps, comprehensive permutation and combination can generate a large number of multiple groups of coupled process parameters. For example, if there are three key steps and each has three reliable parameter values, 27 groups can be generated. This method can systematically explore the effects of parameter combinations in different steps, discover more optimal combinations, avoid falling into local optimality and missing the global optimality, and lay a solid foundation for screening the optimal process parameter combination.

[0059] Step S4102, performing a glucosamine production simulation on the multiple sets of coupled process parameters to obtain multiple process performance information, wherein the process performance information includes reagent dosage, impurity removal cycle, production efficiency, and production energy consumption. Specifically, for multiple sets of coupled process parameters, chemical production simulation software or an established mathematical model is used to simulate the glucosamine production process. After inputting the parameters, the conditions of each link are recorded in detail to obtain process performance information, such as reagent dosage, impurity removal cycle, production efficiency, and production energy consumption data under different combinations, wherein the reagent dosage is determined according to the reaction consumption, the impurity removal cycle is calculated according to the simulation steps and the removal rate, the production efficiency is reflected in the output per unit time, and the production energy consumption is obtained by integrating the consumption of multiple operations. This information can reflect the actual performance of the coupled process parameters and provide quantitative data support for cost analysis.

[0060] Step S4103: A target coupled process parameter combination is obtained by performing a cost analysis on the multiple process performance information. The target coupled process parameter combination includes the K optimized process parameters. Specifically, by performing a cost analysis on the multiple process performance information, various cost factors are comprehensively weighed. Reagent usage directly affects production cost; higher reagent usage increases costs. Long impurity removal cycles can result in longer equipment usage and increased labor costs. Low production efficiency leads to higher fixed costs per unit of product. High production energy consumption also increases energy costs. Cost weights are assigned based on these factors. For example, for companies with a high proportion of energy costs, energy consumption may be weighted relatively higher. The total cost corresponding to each set of coupled process parameters is then calculated. By comparing the total costs, one or more coupled process parameter combinations with the lowest cost are selected. These combinations become the target coupled process parameter combinations, which include the K optimized process parameters. These optimized process parameters minimize production costs while ensuring product quality (through effective control of key impurity types), thereby improving the economic efficiency and competitiveness of the glucosamine production process.

[0061] Step S500, configure the online detection function for the K key steps according to the K groups of key impurity types and the detection model call table to obtain K online detection modules. Specifically, according to the K groups of key impurity types, K impurity content identification branches that match them are screened out in the detection model call table. For the K key steps, these branches are integrated and configured and equipped with hardware equipment such as high-precision spectrometers, sample collection and transmission devices, and data processing units to construct K online detection modules. Each module corresponds to a specific key step and can accurately detect the content of key impurity types therein. After that, each module is calibrated and optimized using standard impurity samples of known concentrations, the spectrometer parameters are adjusted, and the algorithm model is optimized to ensure detection accuracy and stability. At the same time, the sample collection and transmission efficiency and data processing speed are improved, so that the K online detection modules can effectively perform dynamic detection and early warning of impurity deviations for the K key steps in glucosamine production to ensure stable and consistent product quality.

[0062] Step S600: After the process parameters of the K key steps are updated using the K optimized process parameters, the K online detection modules are used to perform dynamic detection and early warning of impurity deviations in the glucosamine production process. Specifically, during glucosamine production, the process parameters of the K key steps are first updated according to the K optimized process parameters. After production is started, the K online detection modules collect samples according to predetermined conditions for each key step using a specialized sample collection device and send them to a high-precision spectrometer to obtain spectral data. The data processing unit inputs the data into the corresponding impurity content identification branch to calculate the key impurity content, and then compares it with the preset standard content range to calculate the deviation value. If the deviation value is within the normal range, production continues. If it exceeds the normal range, the module immediately triggers an early warning signal, which is conveyed to monitoring personnel through various means along with detailed deviation information. Based on this information, the monitoring personnel evaluate the production process and take adjustment measures such as checking process parameters or equipment conditions to correct the deviation, ensuring stable, efficient, and standard-compliant production.

[0063] The embodiment of the present application adopts the quality control samples of multiple steps obtained by sampling, and extracts the impurity type set in combination with the production log, configures the impurity detection model and call table. The sample impurities are analyzed for cumulative regularity, the key steps and their corresponding impurity types are located, and based on this, the process parameters are optimized and the online detection module is configured. The optimized process parameters are applied in the key steps, and the dynamic monitoring and early warning of impurity deviations are realized by the online detection module, thereby achieving the technical effect of improving the accuracy and efficiency of impurity control in glucosamine production and ensuring the consistency and stability of product quality.

[0064] In the above, refer to Figure 1 The method for detecting impurities in the glucosamine production process according to the embodiment of the present invention is described in detail. Figure 2 The following describes an impurity detection system in a glucosamine production process according to an embodiment of the present invention.

[0065] The impurity detection system for glucosamine production according to an embodiment of the present invention addresses the technical problem of insufficient impurity control in existing glucosamine production processes, thereby improving the accuracy and efficiency of impurity control in glucosamine production and ensuring the consistency and stability of product quality. The impurity detection system for glucosamine production includes a quality control sample acquisition module 10, an impurity detection model configuration module 20, a cumulative pattern analysis module 30, a coupling optimization module 40, an online detection function configuration module 50, and a dynamic detection and early warning module 60.

[0066] The quality control sample acquisition module 10 is used to obtain a plurality of quality control samples of a plurality of production steps by sampling during the trial production of glucosamine, wherein the plurality of production steps have a plurality of execution order identifiers.

[0067] The impurity detection model configuration module 20 is used to obtain a sample impurity type set through interactive production logs, configure the impurity detection model according to the sample impurity type set, and obtain a detection model call table.

[0068] The accumulation rule analysis module 30 is used to perform impurity accumulation rule analysis on the multiple quality control samples, and locate K groups of key impurity types of K key steps based on the analysis results, wherein the K key steps have K execution order identifiers.

[0069] The coupling optimization module 40 is used to take the K groups of key impurity types as impurity removal guides, perform coupling optimization on the K key steps according to the K execution order identifiers, and output K optimized process parameters.

[0070] The online detection function configuration module 50 is used to perform online detection function configuration on the K key steps according to the K groups of key impurity types and the detection model call table to obtain K online detection modules.

[0071] The dynamic detection and early warning module 60 is used to perform dynamic detection and early warning of impurity deviations in the glucosamine production process through the K online detection modules after the process parameters of the K key steps are updated using the K optimized process parameters.

[0072] The specific configuration of the impurity detection model configuration module 20 will be described in detail below. As described above, after the interactive production log obtains the sample impurity type set, the impurity detection model is configured according to the sample impurity type set to obtain the detection model call table. The impurity detection model configuration module 20 further includes: a sample impurity spectrum atlas acquisition unit, the sample impurity spectrum atlas acquisition unit is used to perform network data call based on the M types of sample impurity types in the sample impurity type set, and obtain M sample impurity spectrum atlases of the M types of sample impurity types, wherein the M sample impurity spectrum atlases have M sample impurity content identification sets; an online data call unit, the online data call unit is used to perform network data call based on the M types of sample impurity types as constraints, and obtain multiple sample composite impurity spectra and A plurality of groups of sample single-dimensional impurity spectra; a layer separation model construction unit, the layer separation model construction unit being used to construct a layer separation model based on the plurality of sample composite impurity spectra and the plurality of groups of sample single-dimensional impurity spectra; an identification branch construction unit, the identification branch construction unit being used to construct M impurity content identification branches based on the M sample impurity spectrum sets and the M sample impurity content identification sets; an end connection unit, the end connection unit being used to connect the M impurity content identification branches in parallel, and after identifying the M impurity content identification branches with the M types of sample impurity types, connecting the output end of the layer separation model to the input end of the M impurity content identification branches to complete the construction of the impurity detection model;

[0073] The detection model calling table construction unit is used to associate and store the M sample impurity types and the M impurity content identification branches to complete the construction of the detection model calling table.

[0074] The specific configuration of the accumulation rule analysis module 30 will be described in detail below. As described above, the impurity accumulation rule analysis is performed on the multiple quality control samples, and based on the analysis results, K groups of key impurity types of K key steps are located, wherein the K key steps have K execution order identifiers. The accumulation rule analysis module 30 further includes: a sample impurity content set acquisition unit, the sample impurity content set acquisition unit is used to use the impurity detection model to poll the multiple quality control samples to obtain multiple sample impurity content sets; an impurity distribution matrix generation unit, the impurity distribution matrix generation unit is used to construct an impurity distribution matrix based on the multiple sample impurity content sets with the sample impurity type set as a constraint; an impurity type positioning unit, the impurity type positioning unit is used to perform accumulation rule analysis based on the impurity distribution matrix and locate the K key steps and K groups of key impurity types based on the analysis results; and a content set calling unit, the content set calling unit is used to call and obtain K sample impurity content sets and the K execution order identifiers from the multiple execution order identifiers and multiple sample impurity content sets according to the K key steps.

[0075] Wherein, the impurity detection model is adopted to poll and detect the multiple quality control samples to obtain multiple sample impurity content sets, and the sample impurity content set acquisition unit further includes: a real-time impurity spectrum generating subunit, the real-time impurity spectrum generating subunit is used to divide the first quality control sample into N step subsamples, and then use an infrared spectrometer to generate N real-time impurity spectrum graphs of the N step subsamples; a layer separation subunit, the layer separation subunit is used to synchronize the first real-time impurity spectrum graph to the layer separation model of the impurity detection model, perform layer separation based on the impurity type, and obtain H real-time single-dimensional spectrum graphs of H types of sample impurity types, where H is a positive integer less than or equal to M; a directional identification subunit, the directional identification subunit is used to identify the H types of sample impurity types based on the M impurity types in the impurity detection model. The impurity content identification branch is guided to input the H real-time single-dimensional spectra into H impurity content identification branches for directional identification of impurity content to obtain H real-time impurity contents, wherein the H real-time impurity contents constitute the first real-time impurity information; an impurity information acquisition subunit, the impurity information acquisition subunit is used to analyze the N real-time impurity spectra by analogy using the impurity detection model to obtain N real-time impurity information; an impurity content extreme value extraction subunit, the impurity content extreme value extraction subunit is used to aggregate the N real-time impurity information based on impurity type, perform impurity content extreme value extraction, and obtain a first sample impurity content set; a polling detection subunit, the polling detection subunit is used to poll and detect the multiple quality control samples by analogy using the impurity detection model to obtain the multiple sample impurity content sets.

[0076] Wherein, a cumulative regularity analysis is performed according to the impurity distribution matrix, and the K key steps and K groups of key impurity types are located based on the analysis results. The impurity type location unit further includes: a cumulative amount calculation subunit, the cumulative amount calculation subunit is used to perform cumulative amount calculation according to the impurity distribution matrix, and obtain M impurity cumulative amount sequences corresponding to the M sample impurity types; a cumulative rate analysis subunit, the cumulative rate analysis subunit is used to perform cumulative rate analysis according to the impurity distribution matrix, and obtain M cumulative rate sequences; an impurity removal contribution analysis subunit, the impurity removal contribution analysis subunit is used to perform cross-step impurity removal contribution analysis according to the M impurity cumulative amount sequences, and obtain M impurity removal contribution rate sequences; and a weight calculation subunit, the weighted calculation subunit being used to perform single-step weighted calculation on the M impurity accumulation amount sequences, the M accumulation rate sequences, and the M impurity removal contribution rate sequences to obtain M impurity removal priority sequences; a highest-level call subunit, the highest-level call subunit being used to perform the highest-level call from the M impurity removal priority sequences to obtain M impurity removal priority steps; an aggregation step subunit, the aggregation step subunit being used to aggregate the M impurity removal priority steps to obtain the K key steps; and a key impurity type acquisition subunit, the key impurity type acquisition subunit being used to aggregate the M sample impurity types according to the mapping relationship between the M impurity removal priority steps and the K key steps to obtain the K groups of key impurity types.

[0077] The specific configuration of the coupling optimization module 40 will be described in detail below. As described above, with the K groups of key impurity types as the impurity removal guide, the K key steps are coupled optimized according to the K execution order identifiers, and K optimized process parameters are output. The coupling optimization module 40 further includes: an interactive acquisition unit, the interactive acquisition unit is used to interactively obtain the M target impurity concentrations of the M sample impurity types; a type extraction unit, the type extraction unit is used to extract the first key step and the first group of key impurity types from the K key steps and the K groups of key impurity types according to the K execution order identifiers; an impurity concentration constraint acquisition unit, the impurity concentration constraint acquisition unit is used to call and obtain the first impurity concentration constraint from the M target impurity concentrations according to the first group of key impurity types; a parameter step acquisition unit, the parameter step acquisition unit is used to interactively obtain the first initial process parameter of the first key step; a networked data acquisition unit, the networked data acquisition unit is used to perform networked data acquisition with the first key step and the first group of key impurity types as data call constraints. a set, obtaining a first sample impurity concentration set, a first sample process parameter set and a first sample concentration change set; a multiple regression analysis unit, the multiple regression analysis unit being used to perform multiple regression analysis on the first sample impurity concentration set, the first sample process parameter set and the first sample concentration change set, and fitting and generating a first impurity prediction model; a de-expansion unit, the de-expansion unit being used to de-expand the first initial process parameters using a preset parameter adjustment scale to obtain a first group of updated process parameters; an evaluation unit, the parameter evaluation unit being used to evaluate the first group of updated process parameters using the first impurity prediction model to obtain a first group of reliable process parameters; a process parameter optimization unit, the process parameter optimization unit being used to, by analogy, use the K groups of key impurity types as impurity removal guides, optimize the process parameters of the K key steps according to the K execution order identifiers, and output K groups of reliable process parameters; a coupling optimization unit, the coupling optimization unit being used to perform coupling optimization on the K groups of reliable process parameters and output the K optimized process parameters.

[0078] Among them, the first impurity prediction model is used to evaluate the first group of updated process parameters to obtain a first group of reliable process parameters, and the parameter evaluation unit further includes: a search constraint subunit, the search constraint subunit is used to use the first key step and the first group of key impurity types as search constraints to search and obtain a first group of key impurity concentrations from the impurity distribution matrix; a parameter synchronization subunit, the parameter synchronization subunit is used to synchronize the first group of key impurity concentrations and the first group of updated process parameters to the first impurity prediction model evaluation to obtain a first group of predicted concentration change sets; a constraint traversal subunit, the constraint traversal subunit is used to compare the first group of predicted concentration change sets by using the first impurity concentration constraint traversal to screen from the first group of updated process parameters the first group of reliable process parameters whose impurity concentrations meet the first impurity concentration constraint.

[0079] Wherein, the K groups of reliable process parameters are coupled and optimized, and the K optimized process parameters are output. The coupling optimization unit further includes: a permutation and combination enumeration subunit, which is used to perform permutation and combination enumeration on the K groups of reliable process parameters to obtain multiple groups of coupled process parameters; a production simulation subunit, which is used to perform glucosamine production simulation on the multiple groups of coupled process parameters to obtain multiple process performance information, wherein the process performance information includes reagent usage, impurity removal cycle, production efficiency and production energy consumption; a cost analysis subunit, which is used to obtain a target coupled process parameter combination by performing cost analysis on the multiple process performance information, wherein the target coupled process parameter combination includes the K optimized process parameters.

[0080] The impurity detection system for the glucosamine production process provided by the embodiment of the present invention can execute the impurity detection method for the glucosamine production process provided by any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0081] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules may be used and run on the user terminal and / or server, and the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other and are not used to limit the scope of protection of the present invention.

[0082] The above specific embodiments do not constitute a limitation to the scope of protection of this application. It should be understood by those skilled in the art that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent replacements and improvements made within the spirit and principles of this application should be included in the scope of protection of this application. In some cases, the actions or steps recorded in this application can be performed in an order different from that in the embodiments and can still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. A method for detecting impurities in the glucosamine production process, characterized in that: The method comprises: A plurality of quality control samples of a plurality of production steps are obtained by sampling during the trial production of glucosamine, wherein the plurality of production steps have a plurality of execution order identifiers; After obtaining a sample impurity type set through interactive production logs, an impurity detection model is configured according to the sample impurity type set to obtain a detection model call table; Performing impurity accumulation analysis on the plurality of quality control samples, and locating K groups of key impurity types of K key steps based on the analysis results, wherein the K key steps have K execution order identifiers; Taking the K groups of key impurity types as impurity removal guides, performing coupled optimization on the K key steps according to the K execution order identifiers, and outputting K optimized process parameters; Performing online detection function configuration on the K key steps according to the K groups of key impurity types and the detection model call table to obtain K online detection modules; After the K optimized process parameters are used to update the process parameters of the K key steps, dynamic detection and early warning of impurity deviations in the glucosamine production process are performed through the K online detection modules; After obtaining a sample impurity type set through interactive production logs, an impurity detection model is configured according to the sample impurity type set to obtain a detection model call table. The method includes: Performing online data call based on M sample impurity types in the sample impurity type set to obtain M sample impurity spectrum atlases of the M sample impurity types, wherein the M sample impurity spectrum atlases have M sample impurity content identification sets; Performing online data calls based on the M sample impurity types as constraints to obtain multiple sample composite impurity spectra and multiple groups of sample single-dimensional impurity spectra; Constructing a layer separation model based on the multiple sample composite impurity spectra and multiple groups of sample single-dimensional impurity spectra; M impurity content identification branches are constructed based on the M sample impurity spectrum atlases and the M sample impurity content identification sets; The M impurity content identification branches are connected in parallel, and after using the M sample impurity types to identify the M impurity content identification branches, the output end of the layer separation model is connected to the input end of the M impurity content identification branches to complete the construction of the impurity detection model; The M sample impurity types and the M impurity content identification branches are stored in association to complete the construction of the detection model call table.

2. The method for detecting impurities in the glucosamine production process according to claim 1, wherein Performing impurity accumulation analysis on the multiple quality control samples, and locating K groups of key impurity types in K key steps based on the analysis results, the method comprising: Using the impurity detection model, polling and detecting the plurality of quality control samples to obtain a plurality of sample impurity content sets; Taking the sample impurity type set as a constraint, constructing an impurity distribution matrix according to the plurality of sample impurity content sets; Performing cumulative regularity analysis according to the impurity distribution matrix, and locating the K key steps and K groups of key impurity types based on the analysis results; K sample impurity content sets and the K execution order identifiers are obtained by calling from the multiple execution order identifiers and the multiple sample impurity content sets according to the K key steps.

3. The method for detecting impurities in the glucosamine production process according to claim 2, wherein: The impurity detection model is used to poll and detect the multiple quality control samples to obtain multiple sample impurity content sets. The method includes: After dividing the first quality control sample into N step sub-samples, generating N real-time impurity spectra of the N step sub-samples using an infrared spectrometer; Synchronizing the first real-time impurity spectrum to the layer separation model of the impurity detection model, performing layer separation based on impurity type, and obtaining H real-time single-dimensional spectra of H types of sample impurity types, where H is a positive integer less than or equal to M; According to the correspondence between the H sample impurity types and the M impurity content identification branches in the impurity detection model, the H real-time single-dimensional spectra are guided and input into the H impurity content identification branches for directional identification of impurity content to obtain H real-time impurity contents, wherein the H real-time impurity contents constitute first real-time impurity information; Similarly, the impurity detection model is used to analyze the N real-time impurity spectra to obtain N real-time impurity information; After aggregating the N real-time impurity information based on impurity type, extracting extreme values ​​of impurity content to obtain a first sample impurity content set; Similarly, the impurity detection model is adopted to poll and detect the multiple quality control samples to obtain the multiple sample impurity content sets.

4. The method for detecting impurities in the glucosamine production process according to claim 2, wherein: Performing cumulative regularity analysis according to the impurity distribution matrix, and locating the K key steps and K groups of key impurity types based on the analysis results, the method comprising: Performing cumulative amount calculation according to the impurity distribution matrix to obtain M impurity cumulative amount sequences corresponding to the M sample impurity types; Performing cumulative rate analysis according to the impurity distribution matrix to obtain M cumulative rate sequences; Performing a cross-step impurity removal contribution analysis based on the M impurity accumulation amount sequences to obtain M impurity removal contribution rate sequences; Performing a single-step weighted calculation on the M impurity accumulation amount sequences, the M accumulation rate sequences, and the M impurity removal contribution rate sequences to obtain M impurity removal priority sequences; Performing a highest-level call from the M impurity removal priority sequences to obtain M impurity removal priority steps; Aggregating the M impurity removal priority steps to obtain the K key steps; The M sample impurity types are aggregated according to the mapping relationship between the M impurity removal priority steps and the K key steps to obtain the K groups of key impurity types.

5. The method for detecting impurities in the glucosamine production process according to claim 2, wherein: Taking the K groups of key impurity types as impurity removal guides, performing coupled optimization on the K key steps according to the K execution order identifiers, and outputting K optimized process parameters, the method includes: Interactively obtaining M target impurity concentrations of the M sample impurity types; extracting a first key step and a first group of key impurity types from the K key steps and the K groups of key impurity types respectively according to the K execution order identifiers; Obtaining a first impurity concentration constraint from the M target impurity concentrations according to the first group of key impurity types; interactively obtaining first initial process parameters of the first key step; Using the first key step and the first group of key impurity types as data call constraints, performing networked data acquisition to obtain a first sample impurity concentration set, a first sample process parameter set, and a first sample concentration change set; Performing a multivariate regression analysis on the first sample impurity concentration set, the first sample process parameter set, and the first sample concentration change set to generate a first impurity prediction model; De-expanding the first initial process parameters using a preset parameter adjustment scale to obtain a first set of updated process parameters; evaluating the first set of updated process parameters using the first impurity prediction model to obtain a first set of reliable process parameters; Similarly, taking the K groups of key impurity types as the impurity removal guide, the process parameters of the K key steps are optimized according to the K execution order identifiers, and K groups of reliable process parameters are output; The K groups of reliable process parameters are coupled optimized to output the K optimized process parameters.

6. The method for detecting impurities in the glucosamine production process according to claim 5, wherein: The method includes evaluating the first set of updated process parameters using the first impurity prediction model to obtain a first set of reliable process parameters. Using the first key step and the first group of key impurity types as search constraints, searching and obtaining a first group of key impurity concentrations from the impurity distribution matrix; Synchronizing the first set of key impurity concentrations and the first set of updated process parameters to the first impurity prediction model for evaluation to obtain a first set of predicted concentration change sets; By using the first impurity concentration constraint to traverse and compare the first set of predicted concentration change sets, the first set of reliable process parameters whose impurity concentrations meet the first impurity concentration constraint are screened from the first set of updated process parameters.

7. The method for detecting impurities in the glucosamine production process according to claim 5, wherein: Performing coupled optimization on the K groups of reliable process parameters and outputting the K optimized process parameters, the method comprising: Performing permutation and combination enumeration on the K groups of reliable process parameters to obtain multiple groups of coupled process parameters; Performing a glucosamine production simulation on the multiple sets of coupled process parameters to obtain a plurality of process performance information, wherein the process performance information includes reagent dosage, impurity removal cycle, production efficiency, and production energy consumption; A target coupled process parameter combination is obtained by performing a cost analysis on the plurality of process performance information, wherein the target coupled process parameter combination includes the K optimized process parameters.

8. Impurity detection system in glucosamine production process, characterized in that: The system is used to implement the method for detecting impurities in the glucosamine production process according to any one of claims 1 to 7, and the system comprises: A quality control sample acquisition module, wherein the quality control sample acquisition module is used to obtain a plurality of quality control samples of a plurality of production steps by sampling during the trial production of glucosamine, wherein the plurality of production steps have a plurality of execution order identifiers; An impurity detection model configuration module is configured to obtain a sample impurity type set through interactive production logs, configure an impurity detection model according to the sample impurity type set, and obtain a detection model call table; a cumulative regularity analysis module, configured to perform impurity cumulative regularity analysis on the plurality of quality control samples and locate K groups of key impurity types of K key steps based on the analysis results, wherein the K key steps have K execution order identifiers; a coupling optimization module, the coupling optimization module being configured to perform coupling optimization on the K key steps according to the K execution order identifiers, taking the K groups of key impurity types as impurity removal guides, and outputting K optimized process parameters; An online detection function configuration module, configured to perform online detection function configuration on the K key steps according to the K groups of key impurity types and the detection model call table, to obtain K online detection modules; A dynamic detection and early warning module, wherein the dynamic detection and early warning module is used to perform dynamic detection and early warning of impurity deviations in the glucosamine production process through the K online detection modules after the process parameters of the K key steps are updated using the K optimized process parameters; The impurity detection model configuration module further includes: a sample impurity spectrum atlas acquisition unit, the sample impurity spectrum atlas acquisition unit is used to perform network data call based on M types of sample impurity types in the sample impurity type set, and obtain M sample impurity spectrum atlases of the M types of sample impurity types, wherein the M sample impurity spectrum atlases have M sample impurity content identification sets; an online data call unit, the online data call unit is used to perform network data call based on the M types of sample impurity types as constraints, and obtain multiple sample composite impurity spectrum maps and multiple groups of sample single-dimensional impurity spectrum maps; a layer separation model construction unit, the layer separation A model construction unit is used to construct a layer separation model based on the multiple sample composite impurity spectra and the multiple groups of sample single-dimensional impurity spectra; an identification branch construction unit is used to construct M impurity content identification branches based on the M sample impurity spectrum sets and the M sample impurity content identification sets; an end connection unit is used to connect the M impurity content identification branches in parallel, and after identifying the M impurity content identification branches using the M sample impurity types, connect the output end of the layer separation model to the input end of the M impurity content identification branches to complete the construction of the impurity detection model; The detection model calling table construction unit is used to associate and store the M sample impurity types and the M impurity content identification branches to complete the construction of the detection model calling table.

Images