Dynamic analysis method and system for chemical liquid medicine components based on multispectral fusion
By constructing an ideal component state database and adaptive spectral segmentation, combined with a component contribution optimization inversion model, the accuracy and efficiency issues of chemical drug component analysis were solved, enabling rapid and accurate dynamic analysis in complex industrial environments.
Patent Information
- Application Number
- CN202511019483.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-23
- Publication Date
- 2025-11-07
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Existing chemical drug component analysis technologies suffer from poor accuracy and low efficiency, especially in industrial settings where they struggle to meet the demands for rapid dynamic response. Furthermore, static spectral acquisition schemes cannot adapt to changes in composition and differences in equipment.
An ideal component state database is constructed, and adaptive segmentation is performed based on the performance of the target spectral acquisition device. The component-spectral segment contribution matrix is calculated, spectral acquisition commands are generated, the spectral acquisition device is controlled to acquire data of the selected spectral segments, and the component analysis results are obtained through iterative fusion inversion optimization.
It significantly improves the efficiency and accuracy of online analysis of multiple components in chemical solutions, adapts to complex industrial environments, enables rapid and accurate dynamic analysis, and reduces data redundancy and processing burden.
Smart Images
Figure CN120908113A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of component detection, and in particular to a chemical medicine water component dynamic analysis method and system based on multi-spectrum fusion. BACKGROUND
[0002] The accurate analysis of the components of chemical medicine water is crucial in the field of industrial production. At present, online spectrum detection technology, especially ultraviolet spectrum and infrared spectrum, is widely used in real-time monitoring of chemical medicine water components due to its non-contact and fast advantages. However, the existing technology faces significant challenges in practical application. In order to pursue accuracy, high-density data collection in the full spectrum range not only is inefficient and time-consuming, but also increases hardware costs and data processing burden, making it difficult to meet the demand for fast dynamic response in industrial sites. The static preset spectrum collection scheme cannot adapt to changes in medicine water components or environmental disturbances, and lacks optimal use of the actual collection performance of the spectrometer, further restricting the real-time and reliability of the analysis. SUMMARY
[0003] The present application provides a chemical medicine water component dynamic analysis method and system based on multi-spectrum fusion, which is used to solve the technical problems of poor accuracy and low efficiency of chemical medicine water component analysis in the prior art.
[0004] In view of the above problems, the present application provides a chemical medicine water component dynamic analysis method and system based on multi-spectrum fusion.
[0005] In the first aspect, the present application provides a chemical medicine water component dynamic analysis method based on multi-spectrum fusion, which comprises:
[0006] Constructing an ideal component state database, wherein the ideal component state database comprises a target component reference concentration and prior multi-spectrum data stored in association.
[0007] Based on the prior multi-spectrum data, an adaptive segmentation division is performed in the multi-spectrum space in combination with the collection performance of a target spectrum collection device to obtain a spectrum segment set, wherein each spectrum segment is a continuous spectrum range belonging to a type of spectrum.
[0008] In combination with the ideal component state database, the prior contribution degree of each target component to the spectrum segment set is calculated respectively to obtain a component-spectrum segment contribution degree matrix, and a spectrum collection scheme is defined according to the component-spectrum segment contribution degree matrix to generate corresponding spectrum collection instructions.
[0009] According to the spectrum collection instructions, the selected spectrum segment data of the target chemical medicine water is collected by the target spectrum collection device.
[0010] Based on the target component reference concentration, initialize the spectral inversion model, perform iterative fusion inversion optimization of the spectral inversion model with the selected spectral segment data as the target, and extract the target component concentration output corresponding to the optimal fusion inversion optimization result as the component analysis result.
[0011] In a second aspect, the present application provides a multi-spectral fusion-based chemical water component dynamic analysis system, comprising:
[0012] A database construction module is configured to construct an ideal component state database, wherein the ideal component state database comprises target component reference concentrations and prior multi-spectral data stored in association.
[0013] A spectral division module is configured to perform adaptive segmentation division in a multi-spectral space based on the prior multi-spectral data and in combination with the acquisition performance of a target spectrum acquisition device, to obtain a spectral segment set, wherein each spectral segment is a continuous spectral range belonging to a type of spectrum.
[0014] An acquisition instruction generation module is configured to calculate the prior contribution degree of each target component to the spectral segment set in combination with the ideal component state database, to obtain a component-spectral segment contribution degree matrix, and to define a spectrum acquisition scheme according to the component-spectral segment contribution degree matrix and generate corresponding spectrum acquisition instructions.
[0015] A spectrum data acquisition module is configured to control a target spectrum acquisition device to acquire selected spectral segment data of a target chemical water according to the spectrum acquisition instructions.
[0016] A component analysis module is configured to initialize a spectral inversion model based on the target component reference concentration, perform iterative fusion inversion optimization of the spectral inversion model with the selected spectral segment data as the target, and extract the target component concentration output corresponding to the optimal fusion inversion optimization result as the component analysis result.
[0017] One or more technical solutions provided in the present application have at least the following technical effects or advantages:
[0018] The present application provides a multi-spectral fusion-based chemical water component dynamic analysis method and system, which adaptively selects the spectral segment with the most information amount for efficient acquisition, and dynamically optimizes the inversion model based on the component contribution degree, significantly improving the efficiency, accuracy and adaptability to complex systems and equipment conditions of chemical water multi-component online analysis. Compared with traditional methods, the technical solutions provided by the present application significantly overcome the inherent defects of time-consuming and inefficient full-spectrum scanning and the inability of static schemes to adapt to component changes and equipment differences, achieving the technical effect of rapidly and accurately dynamically analyzing chemical water multi-components in complex industrial field environments. BRIEF DESCRIPTION OF DRAWINGS
[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0020] Figure 1 This is a flowchart illustrating a dynamic analysis method for chemical drug components based on multispectral fusion, provided in an embodiment of this application.
[0021] Figure 2 This is a schematic diagram of a dynamic analysis system for chemical drug components based on multispectral fusion, provided in an embodiment of this application.
[0022] The components represented by each number in the attached diagram are explained below:
[0023] The system includes a database construction module 100, a spectral segmentation module 200, an acquisition instruction generation module 300, a spectral data acquisition module 400, and a component analysis module 500. Detailed Implementation
[0024] This application provides a method and system for dynamic analysis of chemical drug components based on multispectral fusion, which addresses the technical problems of poor accuracy and low efficiency in the analysis of chemical drug components in existing technologies.
[0025] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.
[0026] It should be noted that the terms "comprising" and "having" are intended to cover non-exclusive inclusion. 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 that are explicitly listed, but may include other steps or modules that are not explicitly listed or that are inherent to these processes, methods, products, or devices.
[0027] Example 1, as Figure 1 As shown, this application provides a method for dynamic analysis of chemical drug components based on multispectral fusion, wherein the method includes:
[0028] S10: Construct an ideal component state database, wherein the ideal component state database includes the target component reference concentration and prior multispectral data stored in association.
[0029] Traditional chemical water composition analysis lacks standardized and structured prior knowledge base, resulting in an analysis process that relies on manual experience or scattered data, making it difficult to systematically correlate target component reference concentration with its corresponding multi-spectral features.
[0030] The step S10 in the method provided by the embodiments of the present application comprises:
[0031] The interactive production management terminal collects the component reference concentration of the target chemical water in the standard state, wherein the component reference concentration comprises multiple groups of associated target component-concentration fields.
[0032] Based on the preset spectral range, the multi-spectral data of each target component is extracted, and the component reference concentration and the multi-spectral data are stored in association.
[0033] Each target chemical water in the target scene is iteratively collected and extracted, and stored in association to form the ideal component state database.
[0034] In the embodiments of the present application, the interactive production management terminal collects the component reference concentration of the target chemical water in the standard state, wherein the component reference concentration comprises multiple groups of associated target component-concentration fields. For example, the interactive production management terminal knows that the produced chemical water is ethanol, and the reference concentration of the hydroxyl group in the standard state is 17.1 mol / L, and the reference concentration of the methyl group is 17.1 mol / L.
[0035] Based on the preset spectral range, the multi-spectral data of each target component is extracted, and the component reference concentration and the multi-spectral data are stored in association. For example, the preset spectral range is 2800 -1 to 3500 cm -1 , wherein the infrared absorption spectrum range of the hydroxyl group is 300 -1 0 to 3500 -1 , and the absorption spectrum range of the methyl group is 2800 -1 to 3000 -1 .
[0036] Each target chemical water in the target scene is iteratively collected and extracted, and stored in association to form the ideal component state database. The ideal component state database comprises multiple groups of component reference concentrations and multi-spectral data. For example, the ideal component state database may comprise a hydroxyl group-17.1 mol / L, and an absorption spectrum in the range of 3400 -1 to 3200 -1 .
[0037] By constructing the ideal component state database for correlating the storage of the target component reference concentration with the prior multispectral data, an authoritative and complete ideal state benchmark is provided for subsequent analysis. The ideal component state database systematically integrates the spectral response characteristics of each component under standard conditions, significantly improving the mapping reliability of multispectral data and target components, and laying a high-precision prior knowledge foundation for adaptive segmentation, contribution calculation and inversion optimization.
[0038] S20: Based on the prior multispectral data, adaptive segmentation is performed in the multispectral space in combination with the acquisition performance of the target spectrum acquisition device to obtain a spectrum segment set, wherein each spectrum segment is a continuous spectrum range belonging to a type of spectrum.
[0039] The fixed spectrum segmentation strategy cannot adapt to the performance differences of different spectrum acquisition devices, resulting in low efficiency of full-spectrum scanning. The existing methods often have problems such as collecting redundant data at high frequency or missing weak response signals due to the neglect of the actual acquisition capability of the device, which not only prolongs the detection time, but also may reduce the component resolution accuracy due to invalid spectral interference.
[0040] The step S20 in the method provided by the embodiments of the present application includes:
[0041] Based on the prior multispectral data, the absorption peaks of each target component are identified, and single-component segmentation is performed accordingly to obtain a plurality of single spectrum segment sets.
[0042] The segmentation division points are extracted by traversing the plurality of single spectrum segment sets to obtain a segmentation division point set.
[0043] An adaptive fusion constraint is defined according to the acquisition performance of the target spectrum acquisition device, and the segmentation division points with a distance less than the adaptive fusion constraint are adaptively fused to obtain an adaptive segmentation division point set.
[0044] The adaptive segmentation division point set is mapped to the multispectral space to obtain the spectrum segment set.
[0045] In the embodiments of the present application, based on the prior multispectral data, the absorption peaks of each target component are identified, and single-component segmentation is performed accordingly to obtain a plurality of single spectrum segment sets. For example, in the prior spectral data, the spectrum is an infrared spectrum, the target component is a hydroxyl group, and the absorption peak may be at 3000 -1 to 3200 -1 , then 3000 -1 to 3200 -1 is segmented for single-component segmentation. The same idea is used for spectrum segmentation to obtain a plurality of single spectrum segment sets. The spectrum may include infrared spectrum, ultraviolet spectrum, etc.
[0046] Traverse multiple single spectral segment sets, extract segmentation points, for example, infrared spectrum 3000 -1 to 3200 -1 Segmentation points of the segment are 3000 -1 and 3200 -1 Integrate the segmentation points to obtain a segmentation point set.
[0047] According to the acquisition performance of the target spectrum acquisition device, define an adaptive fusion constraint, and adaptively fuse the segmentation points with a distance less than the adaptive fusion constraint to obtain an adaptive segmentation point set. For example, the acquisition performance of the spectrum acquisition device is that the minimum acquisition range is 200 -1 , then adaptively fuse the segmentation points with a distance less than the adaptive fusion constraint, for example, several continuous segmentation points are 2750 -1 , 2950 -1 , 3000 -1 , then fuse the segmentation points into 2700 -1 , 2900 -1 and 3100 -1 .
[0048] Map the adaptive segmentation point set to the multispectral space to obtain a spectral segment set. The spectral segments in the spectral segment set can include infrared spectrum, ultraviolet spectrum, etc.
[0049] In combination with the actual performance of the target spectrum acquisition device, the multispectral space is adaptively segmented and divided to generate a continuous spectral segment set focusing on effective spectral types. By dynamically fusing similar segmentation points, the spectral information with distinguishability is maximized, the data amount to be acquired is significantly compressed while the integrity of key features is ensured, and an optimized input is provided for efficient and device-adapted spectrum acquisition scheme design.
[0050] S30: In combination with the ideal component state database, the prior contribution degree of each target component to the spectral segment set is calculated respectively to obtain a component-spectral segment contribution degree matrix, and a spectrum acquisition scheme is defined according to the component-spectral segment contribution degree matrix to generate corresponding spectrum acquisition instructions.
[0051] The traditional method can only rely on fixed wavelength or full spectrum acquisition, which leads to an acquisition scheme that is not optimized for target components. The blindness lies in not only introducing a large amount of interference data, but also causing a decrease in analysis sensitivity due to the failure to distinguish between key and non-key spectral segments.
[0052] The method provided in the embodiments of the present application comprises the following steps S30:
[0053] By combining the reference concentration of the target component with prior multispectral data, a weighted calculation method is applied to calculate and determine the contribution ratio of each target component in each spectral segment based on the characteristic absorption peak of each spectral segment and the response intensity of the target component, thereby obtaining an initial contribution matrix.
[0054] The initial contribution matrix is normalized by combining the total response intensity of each spectral band to obtain the component-spectral band contribution matrix.
[0055] Based on the component-spectral segment contribution matrix and combined with the preset significance constraint, the spectral segments whose total contribution is greater than the significance constraint are selected as the target acquisition spectral segments.
[0056] Based on the acquisition sequence and frequency of the multiple target spectral bands, and combined with the acquisition performance of the target spectral acquisition device, a spectral acquisition scheme is defined.
[0057] Input the spectral acquisition scheme into the host computer of the target spectral acquisition device to obtain the spectral acquisition command.
[0058] In this embodiment, by combining the reference concentration of the target component with prior multispectral data, a weighted calculation method is applied to calculate and determine the contribution ratio of each target component in each spectral segment based on the characteristic absorption peaks of each spectral segment and the response intensity of the target component, thus obtaining an initial contribution matrix. For example, the reference concentration of hydroxyl is 17.1 mol / L, the reference concentration of methyl is 17.1 mol / L, and the infrared spectrum in the prior multispectral data is 2800. -1 Up to 3500 -1 The infrared absorption spectrum of hydroxyl groups has a range of 3000. -1 Up to 3500 -1 The infrared absorption spectrum of methyl groups has a range of 2800. -1 Up to 3000 -1 The infrared spectral band is divided into 2700. -1 2900 -1 3100 -1 3300 -1 3500 -1 Then 2700 -1 Up to 2900 -1 In the segment, the methyl group showed the strongest response, while the hydroxyl group showed no response, therefore the response was at 2700. -1 Up to 2900 -1 The methyl group contributes 100% to the absorption peak appearing in the segment. At 2900... -1 Up to 3100 -1 In the segment, the methyl group is at 2900. -1 Up to 3000 -1 In response, the hydroxyl group at 3000 -1 to-1 3100 segment response, and the methyl and hydroxyl concentrations are both 17.1 mol / L, the contribution ratio = (the proportion of the response spectrum segment in the whole spectrum + the proportion of the concentration in the sum of the concentrations of all components) ÷ 2, in the 2900 -1 to 3100 -1 segment, the methyl contribution ratio = [(1000 ÷ 2000) + (17.1 ÷ 34.2)] ÷ 2 = 50%, and the hydroxyl contribution ratio = [(1000 ÷ 2000) + (17.1 ÷ 34.2)] ÷ 2 = 50%. In the 3100 -1 to 3300 -1 segment, the response intensity of the hydroxyl is the strongest, and the methyl has no response, so the contribution ratio of the hydroxyl in the absorption peak appearing in the 3100 -1 to 3300 -1 segment is 100%. In the 3300 -1 to 3500 -1 segment, the response intensity of the hydroxyl is the strongest, and the methyl has no response, so the contribution ratio of the hydroxyl in the absorption peak appearing in the 3300 -1 to 3500 -1 segment is 100%. The contribution ratios of the components in the various spectrum segments are integrated to obtain an initial contribution degree matrix.
[0059] The initial contribution degree matrix is normalized in combination with the total response intensity of each spectrum segment to obtain a component-spectrum segment contribution degree matrix. In an embodiment, in the 2900 -1 to 3100 -1 segment, the response intensity of the methyl is 40%, the response intensity of the methylene is 40%, and the response intensity of the hydroxyl is 50%, so after normalization, the methyl contribution degree = 40% ÷ (40% + 40% + 50%) = 30.7%, the methylene contribution degree = 40% ÷ (40% + 40% + 50%) = 30.7%, and the hydroxyl contribution degree = 50% ÷ (40% + 40% + 50%) = 38.4%, and the component-spectrum segment contribution degree matrix is integrated.
[0060] Based on the component-spectrum segment contribution degree matrix, in combination with a preset significance constraint, a spectrum segment with a contribution degree sum greater than the significance constraint is selected as a target acquisition spectrum segment. The significance constraint is a value representing the spectrum significance, and when the spectrum significance is too low, it can be difficult to identify the absorption peak. The significance constraint is exemplarily set to 60%. The spectrum segment with a contribution degree sum of the target component greater than 60% is selected as the target acquisition spectrum segment.
[0061] According to the acquisition sequence of the spectral segments of the multiple targets, the spectral segment frequency, and in combination with the acquisition performance of the target spectrum acquisition device, a spectrum acquisition scheme is defined. For example, the spectrum acquisition sequence is to acquire infrared spectrum first and then acquire ultraviolet spectrum, the acquisition performance of the target spectrum device is that the minimum acquisition range is 200 -1 , and the spectral segment frequency is 2700 -1 to 3500 -1 .
[0062] The spectrum acquisition scheme is input to the upper computer of the target spectrum acquisition device, and a spectrum acquisition instruction is obtained. The spectrum acquisition instruction is a programmed non-natural language instruction used by the spectrum acquisition device.
[0063] Based on the component-spectral segment contribution degree matrix, the contribution degree of each target component in each spectral segment is accurately quantified, and a spectrum acquisition scheme covering high-contribution-degree spectral segments is defined accordingly. By eliminating low-information-band, the device is guided to collect data with the most characteristic force, greatly improving the data quality and collection efficiency, and providing input data with strong anti-interference and high component discrimination for subsequent inversion.
[0064] S40: According to the spectrum acquisition instruction, the target spectrum acquisition device is controlled to acquire selected spectral segment data of the target chemical water.
[0065] In the embodiment of the application, according to the obtained spectrum acquisition instruction, the target spectrum acquisition device is controlled to acquire selected spectral segment data of the target chemical water.
[0066] By accurately controlling the target device through the spectrum acquisition instruction, only the key spectral segment data selected by the contribution degree matrix is obtained, the data redundancy is significantly reduced, the acquisition period is shortened, and accurate and targeted data source is provided for rapid and high-precision dynamic analysis.
[0067] S50: Based on the target component reference concentration, the spectrum inversion model is initialized, the iterative fusion inversion optimization of the spectrum inversion model is performed with the selected spectral segment data as the target, and the target component concentration output corresponding to the optimal fusion inversion optimization result is extracted as the component analysis result.
[0068] The traditional inversion method often directly uses a static model, and does not use prior concentration information to initialize the optimization direction, which is easy to fall into a local optimal solution. Especially in the scene of selected spectral segment data quantity simplification, the conventional inversion is difficult to stably converge to the true result.
[0069] The step S50 in the method provided in the embodiment of the application includes:
[0070] Random fluctuations are performed based on the target component reference concentration, and the inversion particle swarm is initialized according to the random fluctuation result.
[0071] The spectrum inversion model is activated to perform model inversion fitting on each inversion particle in the inversion particle group, and to calculate the similarity between each inversion fitting result and the selected spectral segment data as an inversion fitting evaluation value.
[0072] The inversion particle group is sequenced based on the inversion fitting evaluation value, top selection is performed on the inversion particle group according to a preset retention ratio, and random fluctuation is performed based on the top selection result to update the inversion particle group.
[0073] Iterative model inversion fitting and inversion fitting evaluation are performed on the updated inversion particle group until a preset iteration constraint is met, and the real-time inversion particle group is correspondingly output as a fusion inversion optimization result.
[0074] The fusion inversion optimization result is traversed, and the inversion particle corresponding to the optimal inversion fitting evaluation value is extracted as the optimal fusion inversion optimization result.
[0075] Based on the optimal fusion inversion optimization result, the target component concentration of each target component is extracted, and the component analysis result of the target chemical solution is correspondingly output.
[0076] In the embodiments of the present application, random fluctuation is performed based on the target component reference concentration, for example, the reference concentration of each target component is randomly increased or decreased within a range of ±10%, to obtain a plurality of inversion particles, each particle representing a group of possible component concentration values. According to the random fluctuation result, an initialized inversion particle group is obtained, wherein the inversion particle group contains a plurality of particles.
[0077] A single-layer linear combination model structure is used to construct a spectrum inversion model, and the concentration value is input, and the model can output predicted spectral data.
[0078] The spectrum inversion model is activated to perform model inversion fitting on each inversion particle in the inversion particle group, and to calculate the similarity between each inversion fitting result and the selected spectral segment data as an inversion fitting evaluation value. Similarity = 1 - |inversion fitting result - selected spectral segment data| ÷ [(inversion fitting result + selected spectral segment data) ÷ 2].
[0079] Based on the inversion fitting evaluation value, the inversion particle group is sequenced, specifically the inversion particles are sorted from high to low according to the evaluation value, top selection is performed on the inversion particle group according to a preset retention ratio, for example, the preset retention ratio is 50%, the top 50% inversion particles are selected, and based on the top selection result, random fluctuation is performed to update the inversion particle group.
[0080] The model inversion fitting and inversion fitting evaluation of the updated inversion particle swarm are performed iteratively until a preset iteration constraint is met, for example, the iteration constraint is preset to be that the inversion fitting evaluation is all above 0.85, or the iteration is 50 times. After the iteration constraint is met, the real-time inversion particle swarm corresponding to the output is the fusion inversion optimization result.
[0081] The fusion inversion optimization result is traversed, and the inversion particle corresponding to the optimal inversion fitting evaluation value is extracted as the optimal fusion inversion optimization result.
[0082] Based on the optimal fusion inversion optimization result, the target component concentration of each target component is extracted, and the component analysis result corresponding to the target chemical solution is output.
[0083] The present application initializes the model based on the target component reference concentration and performs iterative fusion inversion optimization. By dynamically adjusting the inversion particle swarm and fusing multiple rounds of optimization results, the spectral noise interference is effectively suppressed, the global optimal solution is quickly approached under the driving of a small amount of high-value data, and finally the high-accuracy component concentration result is output, ensuring the accuracy and stability of the dynamic analysis of the complex system.
[0084] Embodiment two, as Figure 2 shown, based on the same inventive concept of the chemical solution component dynamic analysis method based on multi-spectrum fusion provided in embodiment one, the present application embodiment further provides a chemical solution component dynamic analysis system based on multi-spectrum fusion, comprising:
[0085] The database construction module 100 is configured to construct an ideal component state database, wherein the ideal component state database comprises target component reference concentrations and prior multi-spectrum data stored in association.
[0086] The spectrum division module 200 is configured to perform adaptive segmentation division in a multi-spectrum space based on the prior multi-spectrum data and in combination with the acquisition performance of the target spectrum acquisition device, to obtain a spectrum segment set, wherein each spectrum segment is a continuous spectrum range belonging to a type of spectrum.
[0087] The acquisition instruction generation module 300 is configured to calculate the prior contribution degree of each target component to the spectrum segment set respectively in combination with the ideal component state database, to obtain a component-spectrum segment contribution degree matrix, and to define a spectrum acquisition scheme according to the component-spectrum segment contribution degree matrix, and to generate corresponding spectrum acquisition instructions.
[0088] The spectrum data acquisition module 400 is configured to control the target spectrum acquisition device to acquire selected spectrum segment data of the target chemical solution according to the spectrum acquisition instructions.
[0089] The component analysis module 500 is configured to initialize a spectrum inversion model based on the target component reference concentration, perform iterative fusion inversion optimization of the spectrum inversion model with the selected spectral range data as a target, and extract a target component concentration output corresponding to an optimal fusion inversion optimization result as a component analysis result.
[0090] In one embodiment, the database construction module 100 is further configured to:
[0091] The interactive production management end collects the component reference concentration of the target chemical medicine water in a standard state, wherein the component reference concentration includes multiple groups of associated target component-concentration fields.
[0092] Based on a preset spectral range, the multiple spectral data of each target component are extracted, and the component reference concentration and the multiple spectral data are associated and stored.
[0093] Iterative collection and extraction are performed on each target chemical medicine water of a target scene, and associated storage is performed to form the ideal component state database.
[0094] In one embodiment, the spectrum division module 200 is further configured to:
[0095] Based on the prior multiple spectral data, the absorption peak of each target component is identified, and single-component segmentation division is performed correspondingly to obtain multiple single spectral range sets.
[0096] Iterative extraction of segmentation division points is performed on multiple single spectral range sets to obtain a segmentation division point set.
[0097] An adaptive fusion constraint is defined according to the collection performance of a target spectrum collection device, and adaptive fusion is performed on segmentation division points with a distance less than the adaptive fusion constraint to obtain an adaptive segmentation division point set.
[0098] The adaptive segmentation division point set is mapped to the multiple spectral space to obtain the spectral range set.
[0099] In one embodiment, the collection instruction generation module 300 is further configured to:
[0100] In combination with the target component reference concentration and the prior multiple spectral data, a weighted calculation method is applied to calculate and determine the contribution proportion of each target component in each spectral range according to the characteristic absorption peak of each spectral range and the response intensity of the target component, and an initial contribution degree matrix is obtained.
[0101] In combination with the total response intensity of each spectral range, the initial contribution degree matrix is normalized to obtain the component-spectral range contribution degree matrix.
[0102] Based on the ingredient-spectrum contribution degree matrix, combined with a preset saliency constraint, a spectrum segment with a total contribution degree greater than the saliency constraint is selected as a target acquisition spectrum segment.
[0103] According to the acquisition sequence and spectrum segment frequency of the plurality of target acquisition spectrum segments, combined with the acquisition performance of a target spectrum acquisition device, a spectrum acquisition scheme is defined.
[0104] The spectrum acquisition scheme is input to a host computer of the target spectrum acquisition device, and the spectrum acquisition instruction is obtained.
[0105] In one embodiment, the ingredient analysis module 500 is further configured to:
[0106] Based on the target ingredient reference concentration, random fluctuations are performed, and the inversion particle swarm is initialized according to the random fluctuation results.
[0107] The spectrum inversion model is activated to perform model inversion fitting on each inversion particle in the inversion particle swarm, and the similarity of each inversion fitting result to the selected spectrum segment data is calculated as an inversion fitting evaluation value.
[0108] Based on the inversion fitting evaluation value, the inversion particle swarm is sequenced, the inversion particle swarm is top-selected according to a preset retention ratio, and random fluctuations are performed based on the top selection results to update the inversion particle swarm.
[0109] Iterative model inversion fitting and inversion fitting evaluation are performed on the updated inversion particle swarm until a preset iteration constraint is met, and the corresponding real-time inversion particle swarm is output as a fusion inversion optimization result.
[0110] The fusion inversion optimization result is traversed, and the inversion particle corresponding to the optimal inversion fitting evaluation value is extracted as the optimal fusion inversion optimization result.
[0111] Based on the optimal fusion inversion optimization result, the target ingredient concentration of each target ingredient is extracted, and the corresponding output is the ingredient analysis result of the target chemical solution.
[0112] In summary, the embodiments of the present application have at least the following technical effects:
[0113] The application provides a chemical medicine water component dynamic analysis method and system based on multispectral fusion. The method and system can improve the efficiency, accuracy and adaptability to complex systems and equipment conditions of online analysis of chemical medicine water multi-component by adaptively selecting the spectral segment with the most information for efficient collection and dynamically optimizing the inversion model based on component contribution. Specifically, first, an ideal component state database is constructed as prior knowledge; second, the multispectral space is adaptively segmented and divided according to the actual performance of the target spectrum acquisition device, and the redundant information is discarded to focus on the spectral segment set that can effectively represent the target component; then, the optimal spectrum acquisition scheme is intelligently defined based on the component-spectral segment contribution matrix to guide the device to collect key data, greatly shorten the collection time and reduce data redundancy and processing burden; in the inversion stage, the model is initialized by using the reference concentration of the target component, and only the selected efficient spectral segment data is used to perform iterative fusion inversion optimization to ensure the accuracy of the inversion of the concentration of each component in the complex mixed system. Meanwhile, the scheme provided by the application has a dynamic adjustment feature and can be self-optimized according to the actual fluctuation of the medicine water component and the equipment state, so that the stability and reliability of the analysis results can be maintained when facing different batches of chemical medicine water or using different performance spectrometers. Compared with the traditional method, the technical scheme provided by the application significantly overcomes the inherent defects of the traditional method, such as time-consuming and inefficient full-spectrum scanning and the static scheme that cannot adapt to component changes and equipment differences, and achieves the technical effect of realizing fast and accurate dynamic online analysis of chemical medicine water multi-component in a complex industrial site environment.
[0114] It should be noted that the above sequence of the embodiments of the application is only for description, and does not represent the advantages and disadvantages of the embodiments. Moreover, the above describes specific embodiments of the present application. In addition, the processes depicted in the 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 possible or can be advantageous.
[0115] The above only describes the preferred embodiments of the application and does not limit the application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the application shall be included in the protection scope of the application.
[0116] The present application and the drawings are only exemplary descriptions of the application, and any and all modifications, changes, combinations or equivalents within the scope of the application are considered to be covered by the present application. Obviously, those skilled in the art can make various modifications and changes to the application without departing from the scope of the application. Thus, if these modifications and changes of the application belong to the scope of the application and its equivalent technology, the application intends to include these modifications and changes.
Claims
1. A method for dynamic analysis of chemical water components based on multispectral fusion, characterized in that, The method comprises the steps of: constructing an ideal component state database, wherein the ideal component state database comprises a target component reference concentration and prior multispectral data stored in association; based on the prior multispectral data, performing adaptive segmentation division in the multispectral space in combination with the acquisition performance of a target spectrum acquisition device to obtain a spectrum segment set, wherein each spectrum segment is a continuous spectrum range belonging to a type of spectrum; in combination with the ideal component state database, calculating the prior contribution degree of each target component to the spectrum segment set respectively to obtain a component-spectrum segment contribution degree matrix, and defining a spectrum acquisition scheme according to the component-spectrum segment contribution degree matrix to generate corresponding spectrum acquisition instructions; controlling the target spectrum acquisition device to acquire selected spectrum segment data of the target chemical water according to the spectrum acquisition instructions; based on the target component reference concentration, initializing a spectrum inversion model, and performing iterative fusion inversion optimization on the spectrum inversion model with the selected spectrum segment data as the target, and extracting the target component concentration output corresponding to the optimal fusion inversion optimization result as the component analysis result.
2. The method according to claim 1, wherein the method is a multi-spectral fusion based dynamic analysis method for chemical water component. Constructing an ideal component state database, wherein the ideal component state database comprises a component reference concentration and multispectral data stored in association, comprising: an interactive production management terminal acquires the component reference concentration under the standard state of the target chemical water, wherein the component reference concentration comprises a plurality of associated target component-concentration fields; based on a preset spectrum range, extracting the multispectral data of each target component, and storing the component reference concentration and the multispectral data in association; iteratively collecting and extracting each target chemical water of a target scene, and storing in association to form the ideal component state database.
3. The method according to claim 2, wherein the method is a multi-spectral fusion based dynamic analysis method for chemical water component. Based on the prior multispectral data, adaptive segmentation division is performed in the multispectral space in combination with the acquisition performance of a target spectrum acquisition device to obtain a spectrum segment set, comprising: based on the prior multispectral data, identifying the absorption peak of each target component, and performing single-component segmentation division to obtain a plurality of single spectrum segment sets; traversing a plurality of single spectrum segment sets to extract segmentation division points to obtain a segmentation division point set; defining an adaptive fusion constraint according to the acquisition performance of the target spectrum acquisition device, and adaptively fusing the segmentation division points with a distance less than the adaptive fusion constraint to obtain an adaptive segmentation division point set; mapping the adaptive segmentation division point set to the multispectral space to obtain the spectrum segment set.
4. The method according to claim 3, wherein the method is a method for dynamically analyzing a chemical water component based on multispectral fusion. In combination with the ideal component state database, the prior contribution degree of each target component to the spectrum segment set is calculated respectively to obtain a component-spectrum segment contribution degree matrix, comprising: in combination with the target component reference concentration and prior multispectral data, applying a weighted calculation method to calculate and determine the contribution proportion of each target component in each spectrum segment according to the characteristic absorption peak of each spectrum segment and the response intensity of the target component to obtain an initial contribution degree matrix; in combination with the total response intensity of each spectrum segment, the initial contribution degree matrix is normalized to obtain the component-spectrum segment contribution degree matrix.
5. The method according to claim 4, wherein the method is a multi-spectral fusion based dynamic analysis method for chemical water component. According to the component-spectrum segment contribution degree matrix, a spectrum acquisition scheme is defined, and corresponding spectrum acquisition instructions are generated, including: Based on the component-spectrum segment contribution degree matrix, combined with a preset significance constraint, a spectrum segment with a total contribution degree greater than the significance constraint is selected as a target acquisition spectrum segment; According to the acquisition order and spectrum segment frequency of multiple target acquisition spectrum segments, combined with the acquisition performance of a target spectrum acquisition device, a spectrum acquisition scheme is defined; The spectrum acquisition scheme is input to the host computer of the target spectrum acquisition device, and the spectrum acquisition instructions are obtained.
6. The method according to claim 5, wherein the method is a multi-spectral fusion based dynamic analysis method for chemical water component. Based on the target component reference concentration, the spectrum inversion model is initialized, and the selected spectrum segment data is used as the target to perform iterative fusion inversion optimization of the spectrum inversion model, including: Based on the target component reference concentration, random fluctuations are performed, and the inversion particle swarm is initialized according to the random fluctuation results; The spectrum inversion model is activated to perform model inversion fitting on each inversion particle in the inversion particle swarm, and the similarity of each inversion fitting result to the selected spectrum segment data is calculated as an inversion fitting evaluation value; Based on the inversion fitting evaluation value, the inversion particle swarm is sequenced, the inversion particle swarm is top-selected according to a preset retention ratio, and the inversion particle swarm is updated based on the top selection result; Iterative model inversion fitting and inversion fitting evaluation are performed on the updated inversion particle swarm until a preset iteration constraint is met, and the real-time inversion particle swarm is correspondingly output as a fusion inversion optimization result.
7. The method according to claim 6, wherein the method is a multi-spectral fusion based dynamic analysis method for chemical water component. The target component concentration corresponding to the optimal fusion inversion optimization result is extracted and output as a component analysis result, including: Iterate through the fusion inversion optimization result, extract the inversion particle with the optimal inversion fitting evaluation value as the optimal fusion inversion optimization result; Based on the optimal fusion inversion optimization result, the target component concentration of each target component is extracted and output as the component analysis result of the target chemical solution.
8. A multi-spectral fusion-based dynamic analysis system for chemical water components, characterized in that, A chemical solution component dynamic analysis method based on multi-spectrum fusion for implementing any one of claims 1-7, the system comprising: A database construction module for constructing an ideal component state database, wherein the ideal component state database comprises a target component reference concentration and prior multi-spectrum data stored in association; A spectrum division module for performing adaptive segmentation division in a multi-spectrum space based on the prior multi-spectrum data and the acquisition performance of a target spectrum acquisition device to obtain a spectrum segment set, wherein each spectrum segment is a continuous spectrum range belonging to a type of spectrum; An acquisition instruction generation module for calculating the prior contribution degree of each target component to the spectrum segment set based on the ideal component state database to obtain a component-spectrum segment contribution degree matrix, and defining a spectrum acquisition scheme based on the component-spectrum segment contribution degree matrix to generate corresponding spectrum acquisition instructions; A spectrum data acquisition module for controlling a target spectrum acquisition device to acquire selected spectrum segment data of a target chemical solution according to the spectrum acquisition instructions; The component analysis module is configured to initialize a spectrum inversion model based on the target component reference concentration, perform iterative fusion inversion optimization on the spectrum inversion model with the selected spectral band data as a target, and extract a target component concentration output corresponding to an optimal fusion inversion optimization result as a component analysis result.