A detection data analysis system and method for chemical wastewater treatment

Abstract

This application provides a detection data analysis system and method for chemical wastewater treatment, belonging to the field of chemical detection data analysis technology. The method includes: dividing the signal peaks in the detection signal of a chemical wastewater sample into symmetrical and asymmetrical peaks, and determining peak shape analysis parameters; using the peak shape analysis parameters to decompose the signal data corresponding to the asymmetrical peak to obtain corresponding virtual sub-peaks; obtaining residual signals from the detection signal based on the signal profiles of the virtual sub-peaks; using the signal profiles of the virtual sub-peaks as reconstructed signals; when the difference between the reconstructed signal and the stripped residual signal is below a threshold, determining the first component information corresponding to the virtual sub-peak; and generating a chemical composition analysis report of the chemical wastewater sample based on the first component information and the second component information corresponding to the symmetrical peak. The technical solution provided by this application can accurately analyze the detection signal of a chemical wastewater sample under the influence of peak shape distortion.

Description

Technical Field

[0001] This application relates to the field of chemical testing data analysis technology, and more specifically, to a testing data analysis system and method for chemical wastewater treatment. Background Technology

[0002] Chemical testing data analysis refers to the process of extracting information from raw chemical testing data. In modern chemical production processes, real-time and accurate testing data analysis is a core link in ensuring process safety, product quality, and environmental compliance. As the process industry transforms towards digitalization and intelligence, the traditional model relying on manual sampling and offline analysis can no longer meet the needs of dynamic monitoring and anomaly early warning of reaction processes. Based on technologies such as online spectral chromatography and intelligent sensors, combined with multivariate statistical analysis and machine learning algorithms, high-dimensional time-series data such as temperature, pressure, and component concentration can be deeply integrated and feature extracted, effectively promoting the transformation of chemical testing from passive judgment to proactive prediction.

[0003] In existing chemical testing data analysis, the process begins with acquiring raw process signals using online sensors, spectrometers, or chromatographs. This is followed by data cleaning and correction to eliminate noise, baseline drift, and environmental interference. Based on this, algorithms such as principal component analysis and partial least squares are used to reduce dimensionality and extract key feature variables related to component concentration or physical properties. Finally, a quantitative calibration model or pattern recognition rule is established, and the model's accuracy is evaluated through cross-validation. The analysis results are then fed back to the control system or process optimization stage to achieve real-time inference of material composition and quality. However, in the analysis of testing data from chemical wastewater treatment, the unanalyzed detection signals of chemical wastewater samples... Due to factors such as nonlinear isotherms between the stationary and mobile phases, slow mass transfer within the column, or secondary interactions, the peak shape of the target component often exhibits asymmetric characteristics (i.e., tailing or leading edge). This peak distortion can severely mask some overlapping trace component signals. Traditional fitting based on symmetric peak models fails in such scenarios, making it impossible to distinguish whether the peak distortion originates from abnormal kinetic processes of a single component or from the superposition of actual co-efferentiation of multiple components. Consequently, the analysis of the detection signal of chemical wastewater samples becomes distorted. Therefore, how to accurately analyze the detection signal of chemical wastewater samples under the influence of peak distortion has become a challenge for the industry. Summary of the Invention

[0004] This application provides a detection data analysis system and method for chemical wastewater treatment, which can accurately analyze the detection signals of chemical wastewater samples under the influence of peak distortion.

[0005] In a first aspect, this application provides a method for analyzing detection data in chemical wastewater treatment, comprising the following steps: The system acquires the detection signal of the chemical wastewater sample collected by the intelligent sensor, and divides the signal peaks in the detection signal into symmetrical peaks and asymmetrical peaks through the edge computing unit built into the intelligent sensor. Extract the peak shape deviation characteristics and local signal fluctuation characteristics of the asymmetric peak, and determine the peak shape analysis parameters of the asymmetric peak based on the peak shape deviation characteristics and the local signal fluctuation characteristics; The peak shape analysis parameters are used to iteratively decompose the signal data corresponding to the asymmetric peak, and corresponding virtual sub-peaks are generated in the iterative decomposition. Based on the signal profile of the virtual sub-peaks, the corresponding signal segments in the signal to be analyzed are stripped to obtain the stripped residual signal. When the difference between the signal profile of the virtual sub-peak and the residual signal after stripping is lower than a preset difference threshold, the detection information corresponding to the virtual sub-peak is used as the first component information. The second component information corresponding to the symmetrical peak is determined, and then a chemical composition analysis report of the chemical wastewater test sample is generated based on the first component information and the second component information and fed back to the network node of the smart sensor.

[0006] In some embodiments, dividing the signal peaks in the detected signal to be analyzed into symmetrical peaks and asymmetrical peaks using the edge computing unit built into the smart sensor specifically includes: The edge computing unit built into the smart sensor performs smoothing preprocessing on the detection signal to be analyzed to obtain a preprocessed detection signal. Based on the preprocessed detection signal, the peak apex coordinates and peak width range of each signal peak are located; For each signal peak, the signal intensity difference between corresponding positions on both sides within the peak width is compared with the peak apex as the axis of symmetry, and then the symmetry value of each signal peak is calculated. The symmetry value of each signal peak is compared with the preset symmetry threshold, and the corresponding signal peaks are divided into symmetric peaks and asymmetric peaks based on the comparison results.

[0007] In some embodiments, extracting the peak shape deviation characteristics and local signal fluctuation characteristics of the asymmetric peak specifically includes: The asymmetric peak is subjected to baseline correction processing to obtain the baseline-corrected asymmetric peak signal; The theoretical symmetric reference profile of the asymmetric peak is obtained by fitting the asymmetric peak signal after baseline correction. The deviation between the baseline-corrected asymmetric peak signal and the theoretical symmetric reference profile is calculated to extract the peak shape deviation feature; The local signal segment corresponding to the baseline-corrected asymmetric peak signal is divided from the signal to be analyzed, and the signal fluctuation coefficient of the local signal segment is calculated to extract the local signal fluctuation characteristics.

[0008] In some embodiments, determining the peak shape analysis parameters of the asymmetric peak based on the peak shape deviation characteristics and the local signal fluctuation characteristics specifically includes: The entropy weight method is used to calculate the weights of the peak deviation feature and the local signal fluctuation feature to obtain the weight coefficients corresponding to the peak deviation feature and the local signal fluctuation feature. Based on the weighting coefficients, the peak deviation feature and the local signal fluctuation feature are weighted and fused to generate an asymmetric peak feature vector; The asymmetric peak feature vector is calculated using a pre-trained peak shape parameter mapping model to output the peak shape analysis parameters of the asymmetric peak.

[0009] In some embodiments, using the peak shape analysis parameters to iteratively decompose the signal data corresponding to the asymmetric peak, and generating the corresponding virtual sub-peak in the iterative decomposition specifically includes: The signal decomposition model of the asymmetric peak is initialized based on the peak shape analysis parameters, and the iteration termination condition is set. The signal decomposition model described above is used to perform an initial decomposition of the signal data corresponding to the asymmetric peak to obtain preliminary sub-peak signals. Based on the peak shape characteristics of the preliminary sub-peak signal, a corresponding virtual sub-peak is generated, and the signal parameters of the virtual sub-peak are recorded. The signal parameters of the virtual sub-peaks are fed back to the signal decomposition model to determine whether the iteration termination condition is met. If the iteration termination condition is not met, the signal decomposition model parameters are adjusted based on the feedback results until a virtual sub-peak that meets the requirements is generated. If the condition is met, the iteration stops.

[0010] In some embodiments, determining the second component information corresponding to the symmetric peak specifically includes: Baseline correction is performed on the symmetrical peak to obtain the baseline-corrected symmetrical peak signal; The characteristic parameters of the symmetrical peak are calculated based on the baseline-corrected symmetrical peak signal. The characteristic parameters of the symmetric peak are matched with the standard component spectral library to obtain the spectral library matching results; The second component information corresponding to the symmetric peak is determined based on the spectral library matching results.

[0011] In some embodiments, the analytical detection signal of the chemical wastewater sample collected by the intelligent sensor is acquired by a chromatograph.

[0012] Secondly, this application provides a detection data analysis system for chemical wastewater treatment, used to execute a detection data analysis method for chemical wastewater treatment, the system comprising: The acquisition module is used to acquire the detection signal to be analyzed from the chemical wastewater sample collected by the intelligent sensor, and to divide the signal peaks in the detection signal to be analyzed into symmetrical peaks and asymmetrical peaks through the edge computing unit built into the intelligent sensor. The processing module is used to extract the peak shape deviation characteristics and local signal fluctuation characteristics of the asymmetric peak, and determine the peak shape analysis parameters of the asymmetric peak based on the peak shape deviation characteristics and the local signal fluctuation characteristics; The processing module is further configured to use the peak shape analysis parameters to iteratively decompose the signal data corresponding to the asymmetric peak, generate corresponding virtual sub-peaks in the iterative decomposition, and strip the corresponding signal segment in the signal to be analyzed and detected based on the signal profile of the virtual sub-peaks to obtain the stripped residual signal. The processing module is further configured to use the detection information corresponding to the virtual sub-peak as the first component information when the difference between the signal profile of the virtual sub-peak and the stripped residual signal is lower than a preset difference threshold. The execution module is used to determine the second component information corresponding to the symmetrical peak, and then generate a chemical composition analysis report of the chemical wastewater test sample based on the first component information and the second component information, and feed it back to the network node of the smart sensor.

[0013] Thirdly, this application provides a computer device, the computer device including a memory and a processor, the memory storing code, and the processor being configured to acquire the code and execute the above-described detection data analysis method for chemical wastewater treatment.

[0014] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method for analyzing detection data in chemical wastewater treatment.

[0015] The technical solutions provided by the embodiments disclosed in this application have the following beneficial effects: The detection data analysis system and method for chemical wastewater treatment provided in this application first acquires the detection signal of the chemical wastewater sample collected by an intelligent sensor, and divides the signal peaks in the detection signal into symmetrical peaks and asymmetrical peaks using the edge computing unit built into the intelligent sensor. Second, the peak shape deviation characteristics and local signal fluctuation characteristics of the asymmetrical peak are extracted, and the peak shape analysis parameters of the asymmetrical peak are determined based on these characteristics. Further, the peak shape analysis parameters are used to iteratively decompose the signal data corresponding to the asymmetrical peak, generating corresponding virtual sub-peaks during the iterative decomposition. Based on the signal contour of the virtual sub-peaks, the corresponding signal segments in the detection signal are stripped to obtain the stripped residual signal. Then, when the difference between the signal contour of the virtual sub-peak and the stripped residual signal is lower than a preset difference threshold, the detection information corresponding to the virtual sub-peak is used as the first component information. Finally, the second component information corresponding to the symmetrical peak is determined, and a chemical composition analysis report of the chemical wastewater sample is generated based on the first component information and the second component information and fed back to the network node of the intelligent sensor.

[0016] Therefore, this application can accurately analyze the detection signals of chemical wastewater samples under the influence of peak distortion. First, it acquires the detection signals of chemical wastewater samples collected by intelligent sensors and divides them into symmetrical and asymmetrical peaks, achieving precise differentiation of signal peaks with different shapes. This provides a data foundation for subsequent differentiated and precise analysis, avoiding the problem of poor adaptability of traditional unified analysis methods to different peak shapes. Second, it extracts the peak shape deviation characteristics and local signal fluctuation characteristics of asymmetrical peaks and determines peak shape analysis parameters accordingly. This allows for the quantitative characterization of the non-standard shape essence and local signal background characteristics of asymmetrical peaks, providing guidance for the subsequent iterative decomposition of asymmetrical peaks that fits their actual characteristics, avoiding the problem of poor parameter adaptability caused by blind decomposition without characteristic basis. Furthermore, it uses peak shape analysis parameters to iteratively decompose asymmetrical peaks to generate virtual sub-peaks and peel off residual signals, achieving precise separation of non-standard asymmetrical peaks and breaking down the overall aliased asymmetrical peaks. The solution is to standardize the virtual sub-peaks, and the residual signal obtained by stripping provides a concrete signal basis for determining the termination of the iteration, thereby effectively identifying the low-content component information hidden in the asymmetric peaks, so as to avoid the analysis distortion problem of the detection signal caused by the asymmetric characteristics of the target component peak shape. Then, the virtual sub-peak signal contour is used as the reconstructed signal and compared with the residual signal, and the first component information is determined according to the difference threshold. The accuracy of the virtual sub-peak splitting analysis is realized by the quantitative verification of the signal difference, ensuring the authenticity and validity of the first component information. The second component information corresponding to the symmetric peak is determined, and a chemical composition analysis report is generated based on the two types of component information, realizing the comprehensive coverage of all potential component information of chemical wastewater test samples, and providing reliable and complete component data support for the component identification and quality control of chemical wastewater test samples. In summary, the technical solution provided in this application can accurately analyze the detection signal of chemical wastewater test samples under the influence of peak shape distortion. Attached Figure Description

[0017] Figure 1 This is an exemplary flowchart of a detection data analysis method for chemical wastewater treatment according to some embodiments of this application; Figure 2 This is a schematic diagram of symmetrical and asymmetrical peaks according to some embodiments of this application; Figure 3 This is an exemplary flowchart illustrating the determination of peak shape analysis parameters according to some embodiments of this application; Figure 4 This is a schematic diagram of the structure of a detection data analysis system for chemical wastewater treatment, as shown in some embodiments of this application; Figure 5 This is a schematic diagram of the structure of a computer device for implementing a detection data analysis method for chemical wastewater treatment, according to some embodiments of this application. Detailed Implementation

[0018] To better understand the technical solution of this application, the technical solution of this application will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0019] refer to Figure 1 The figure is an exemplary flowchart of a detection data analysis method for chemical wastewater treatment according to some embodiments of this application. The figure mainly includes the following steps: In step S101, the detection signal to be analyzed of the chemical wastewater sample collected by the smart sensor is obtained, and the signal peaks in the detection signal to be analyzed are divided into symmetrical peaks and asymmetrical peaks by the edge computing unit built into the smart sensor.

[0020] It should be noted that the intelligent sensor in this application is a type of industrial-grade online analytical instrument that integrates advanced detection principles, edge computing, and intelligent algorithms. In terms of detection principles, intelligent sensors cover a variety of types, including electrochemical sensors (such as electrodes for heavy metal detection), optical sensors (such as COD analyzers based on ultraviolet-visible spectrophotometry), chromatographic sensors (used to acquire chromatographic signals of the sample being tested), and sensors using novel nanomaterials. Its core lies in the fact that the sensor integrates a microcontroller (MCU) or a dedicated intelligent sensor processing unit (ISPU) or other edge computing core, which can directly perform preprocessing such as digitization, filtering, and baseline correction on the raw signal, and even run machine learning models to perform localized peak identification, feature extraction, or anomaly judgment, and upload the processed key data to the network node, thereby realizing the transformation from a "data collector" to an intelligent monitoring node with "sensing" and "cognitive" capabilities.

[0021] It should also be noted that the chemical wastewater test samples in this application refer to representative liquid samples of a certain volume or mass collected from different discharge points in the chemical production process in accordance with regulations. These samples usually contain a variety of complex components, such as organic pollutants (benzene series compounds, phenols, polycyclic aromatic hydrocarbons, etc.), inorganic salts, heavy metal ions, suspended solids, oily substances, and possible microorganisms or toxic substances. By testing the chemical wastewater test samples, chemical wastewater can be effectively treated, reducing pollution to the environment.

[0022] In practice, the detection signals to be analyzed in chemical wastewater samples are collected by intelligent sensors. The details will not be elaborated here. The detection signals to be analyzed refer to the response data sequence that reflects the chemical characteristics of the components to be tested in the chemical wastewater samples. The acquisition of the detection signals to be analyzed provides a data basis for the subsequent analysis and detection of chemical wastewater samples.

[0023] In some embodiments, the division of signal peaks in the detected signal to be analyzed into symmetrical and asymmetrical peaks by the edge computing unit built into the smart sensor is achieved through the following steps: The edge computing unit built into the smart sensor performs smoothing preprocessing on the detection signal to be analyzed to obtain a preprocessed detection signal. Based on the preprocessed detection signal, the peak apex coordinates and peak width range of each signal peak are located; For each signal peak, the signal intensity difference between corresponding positions on both sides within the peak width is compared with the peak apex as the axis of symmetry, and then the symmetry value of each signal peak is calculated. The symmetry value of each signal peak is compared with the preset symmetry threshold, and the corresponding signal peaks are divided into symmetric peaks and asymmetric peaks based on the comparison results.

[0024] In specific implementation, firstly, the Savitzky-Gore smoothing method is used to smooth the detection signal to be analyzed by the edge computing unit built into the smart sensor, resulting in a pre-processed detection signal. This pre-processed detection signal refers to the smoothed version of the detection signal to be analyzed. Secondly, the slope of the pre-processed detection signal is calculated point-by-point using the first derivative method. The position where the slope is 0 is marked as the peak of the corresponding signal peak, and its horizontal and vertical coordinates are extracted to form the peak peak coordinates. Then, the curvature change of the pre-processed detection signal is calculated using the second derivative method. The position where the second derivative changes from positive to negative is marked as the starting point of the signal peak, and the position where it changes from negative to positive is marked as the ending point of the signal peak. The peak width range of the corresponding signal peak is constructed from the horizontal coordinate interval between the starting and ending points, thus obtaining the peak peak coordinates and peak width range of each signal peak. The peak peak coordinates refer to the combination of the horizontal and vertical coordinates corresponding to the maximum signal peak intensity, and the peak width range is the complete horizontal coordinate interval from the starting point to the ending point of the signal peak. Finally, for each signal peak, the obtained peak peak coordinates are used as the axis of symmetry to divide the peak... The positions of equal horizontal coordinate offsets on both sides of the symmetry axis within a wide range are matched one-to-one. The signal intensity values ​​of each matched position are extracted point by point, and the differences between each pair are calculated. Then, the absolute value averaging method is used to normalize the signal intensity differences of all corresponding positions within the peak width range, thereby completing the calculation of the symmetry value of each signal peak. The symmetry value refers to the index characterizing the symmetry of the signal intensity distribution on both sides of a single signal peak within the peak width range. Finally, the calculated symmetry values ​​of each signal peak are compared with the symmetry threshold preset in the field of chemical analysis according to the type of detection signal. If the symmetry value of the signal peak is less than or equal to the preset symmetry threshold, it is classified as a symmetrical peak. If the symmetry value of the signal peak is greater than the preset symmetry threshold, it is classified as an asymmetrical peak. The preset symmetry threshold is 0.15. In chromatographic analysis, peaks with a symmetry factor between 0.9 and 1.1 are usually considered symmetrical, and the corresponding symmetry value (calculated according to the normalized intensity difference) is approximately 0.1 to 0.2. Therefore, 0.15 is taken as the preset symmetry threshold, which is not limited here.

[0025] refer to Figure 2 This figure is a schematic diagram of symmetrical and asymmetrical peaks according to some embodiments of this application. The left side of the figure shows symmetrical peaks, and the right side shows asymmetrical peaks. In both figures, the vertical axis represents signal intensity, the horizontal axis represents retention time, the horizontal dashed line is the baseline, and the vertical dashed line at the peak P1 is the axis of symmetry of the symmetrical peak. In the symmetrical peak, the peak height values ​​at the same horizontal position on both sides of the axis of symmetry are equal. The asymmetrical peak on the right side takes the vertical dashed line corresponding to the theoretical peak P2 as the axis of symmetry. The actual peak peak is offset relative to the theoretical axis of symmetry, and the peak heights at the two positions at the same horizontal position of the axis of symmetry are inconsistent, resulting in a peak height difference ΔH. The peak height difference ΔH can be used to determine whether the chromatographic peak is a symmetrical peak or an asymmetrical peak.

[0026] It should be noted that, in this application, symmetrical peaks are signal peaks with consistent signal intensity distributions at corresponding positions on both sides of the peak width; asymmetric peaks are signal peaks with differing signal intensity distributions at corresponding positions on both sides of the peak width. Since symmetrical peaks are mostly characteristic peaks corresponding to a single pure component, their peak shapes fit the adaptation conditions of the standard peak shape fitting model, and accurate component information can be efficiently extracted directly through conventional standard peak shape fitting and spectral library matching. Asymmetric peaks, on the other hand, are mostly formed by factors such as overlapping characteristic peaks of multiple components, interference from the detection matrix, or differences in the response of the detection system, and cannot be directly analyzed by conventional fitting methods. Therefore, by determining symmetrical and asymmetric peaks, targeted signal peak decomposition can be performed, effectively avoiding the problems of redundant analysis process for symmetrical peaks and incomplete analysis of asymmetric peaks caused by using a unified analysis method for all signal peaks.

[0027] In step S102, the peak shape deviation characteristics and local signal fluctuation characteristics of the asymmetric peak are extracted, and the peak shape analysis parameters of the asymmetric peak are determined based on the peak shape deviation characteristics and the local signal fluctuation characteristics.

[0028] In some embodiments, the extraction of the peak shape deviation characteristics and local signal fluctuation characteristics of the asymmetric peak is achieved by the following steps: The asymmetric peak is subjected to baseline correction processing to obtain the baseline-corrected asymmetric peak signal; The theoretical symmetric reference profile of the asymmetric peak is obtained by fitting the asymmetric peak signal after baseline correction. The deviation between the baseline-corrected asymmetric peak signal and the theoretical symmetric reference profile is calculated to extract the peak shape deviation feature; The local signal segment corresponding to the baseline-corrected asymmetric peak signal is divided from the signal to be analyzed, and the signal fluctuation coefficient of the local signal segment is calculated to extract the local signal fluctuation characteristics.

[0029] In specific implementation, firstly, the signal data corresponding to the asymmetric peak is substituted into the adaptive iterative reweighted penalized least squares method for baseline correction processing to obtain the baseline-corrected asymmetric peak signal. The baseline-corrected asymmetric peak signal refers to the asymmetric peak detection signal that retains only the signal characteristics of the peak shape itself. Secondly, the abscissa, ordinate, and full peak width of the peak are extracted from the baseline-corrected asymmetric peak signal as initial fitting parameters. The baseline-corrected asymmetric peak signal is then fitted using Gaussian peak fitting (Gaussian peak fitting is a classic mathematical fitting model for characterizing standard symmetric signal peaks in chemical analysis detection signal processing. This model is based on the mathematical expression of the peak shape constructed by the Gaussian function, with the signal abscissa as the independent variable and the signal intensity as the dependent variable, which will not be elaborated here). That is, with the minimum of the sum of squared residuals as the optimization objective, the amplitude, center position, and half-peak width parameters of the Gaussian peak shape are iteratively optimized using the nonlinear least squares method until the fitting parameters tend to stabilize. The process involves several steps: First, the theoretical symmetry reference profile of the asymmetric peak obtained through fitting is output. This theoretical symmetry reference profile refers to the theoretical signal profile that serves as a reference for the deviation of the asymmetric peak shape. Then, the asymmetric peak signal after baseline correction is correlated with the theoretical symmetry reference profile point by point along the horizontal axis. The absolute deviation between the actual signal intensity and the theoretical profile signal intensity at each corresponding point is calculated. The root mean square deviation is then calculated for all absolute deviation values ​​within the peak width range. This root mean square deviation is used as a peak shape deviation feature, which is a characteristic parameter that characterizes the degree of deviation of the actual peak shape of the asymmetric peak from the theoretical symmetry profile. Finally, using the peak width range of the baseline-corrected asymmetric peak signal as the boundary, a local signal segment corresponding to the asymmetric peak signal is divided from the signal to be analyzed. The signal fluctuation coefficient of the local signal segment is obtained by calculating the ratio of the standard deviation to the mean of the signal intensity within the local signal segment. This signal fluctuation coefficient is then used as a local signal fluctuation feature.

[0030] It should be noted that, in this application, local signal fluctuation characteristics refer to characteristic parameters that characterize the random fluctuations of the signal within the local signal region where the asymmetric peak is located. Determining the peak shape deviation characteristics and local signal fluctuation characteristics can provide targeted characteristic basis for the subsequent determination of peak shape analysis parameters and accurate iterative decomposition of the asymmetric peak. The peak shape deviation characteristics, by quantifying the degree of deviation of the actual shape of the asymmetric peak from the theoretical symmetric reference profile, can provide an adaptability basis for the initialization and iterative update of peak shape analysis parameters that fits the actual shape of the asymmetric peak, so that the subsequent generation of virtual sub-peaks and iterative decomposition process matches the inherent peak shape characteristics of the asymmetric peak. The local signal fluctuation characteristics, by comprehensively characterizing the fluctuation distribution characteristics of the local signal segment corresponding to the asymmetric peak, can serve as an important reference basis for the determination of residual signal attributes and the quantitative setting of iteration termination conditions in the subsequent iterative decomposition.

[0031] In some embodiments, reference Figure 3As shown in the figure, this is an exemplary flowchart of determining peak shape analysis parameters according to some embodiments of this application. In this embodiment, determining the peak shape analysis parameters of the asymmetric peak based on the peak shape deviation characteristics and the local signal fluctuation characteristics can be achieved by the following steps: In step S1021, the entropy weight method is used to calculate the weights of the peak deviation feature and the local signal fluctuation feature to obtain the weight coefficients corresponding to the peak deviation feature and the local signal fluctuation feature. In step S1022, the peak deviation feature and the local signal fluctuation feature are weighted and fused based on the weighting coefficient to generate an asymmetric peak feature vector; In step S1023, the asymmetric peak feature vector is calculated using a pre-trained peak shape parameter mapping model to output the peak shape analysis parameters of the asymmetric peak.

[0032] In specific implementation, firstly, the extracted quantization indicators of peak deviation features and local signal fluctuation features are substituted into the entropy weight method. Range normalization is then performed on the quantization indicators of both types of features, mapping each quantization indicator to the [0,1] interval to eliminate differences in dimensions and orders of magnitude. Then, based on the normalized quantization indicators, the information entropy value of each quantization indicator is calculated according to the information entropy calculation formula. The difference coefficient of each quantization indicator is calculated based on the difference between the information entropy value and 1. The ratio of the difference coefficient of a single feature indicator to the total difference coefficient is used as the basic weight of the corresponding quantization indicator. Finally, the basic weight is allocated to the corresponding feature category based on the quantization indicator. The weighting coefficients corresponding to the peak deviation feature and the local signal fluctuation feature are obtained. These weighting coefficients represent the magnitude of the feature contribution of the peak deviation feature and the local signal fluctuation feature in the determination of asymmetric peak shape analysis parameters. Next, the obtained weighting coefficients are used to perform a weighted assignment operation on the peak deviation feature and the local signal fluctuation feature. The two types of features are then concatenated sequentially to generate an asymmetric peak feature vector. This asymmetric peak feature vector is a high-dimensional numerical vector representing the non-standard shape of the asymmetric peak and the background fluctuation features of the local signal. Finally, the peak shape parameter mapping model is constructed and pre-trained. The model is constructed as a three-layer fully connected neural network consisting of an input layer, a single hidden layer, and an output layer. The number of neurons in the input layer is consistent with the dimension of the asymmetric peak feature vector. The hidden layer uses the ReLU linear rectified activation function, and the number of neurons is set to twice the dimension of the input layer. The number of neurons in the output layer is consistent with the dimension of the asymmetric peak shape analysis parameters, and a linear activation function is used. The model is trained using a training set based on multi-type asymmetric peak sample data from the field of chemical analysis and detection. Each sample includes labeled peak shape deviation features, local signal fluctuation features, and corresponding measured peak shape analysis parameters. The training set is proportionally divided into a training subset and a validation subset. First, the connection weights and bias parameters of the model are initialized. Then, the mean squared error is used as the loss function and stochastic gradient descent is used as the optimizer. The fused feature vector of the samples is input into the model for forward propagation. The model parameters are updated by backpropagation based on the loss value of the model output value and the measured analysis parameters. The training is repeated iteratively until the loss value of the validation subset is less than the preset loss value to complete the model pre-training. The preset loss value can be set to 0.001 or set according to actual needs, which is not limited here. The generated asymmetric peak feature vector is input into the pre-trained peak shape parameter mapping model for forward propagation to output the peak shape analysis parameters of the asymmetric peak.

[0033] It should be noted that the peak shape analysis parameters in this application refer to the parameters used to guide the subsequent iterative decomposition of asymmetric peaks to generate virtual sub-peaks. Determining the peak shape analysis parameters can provide quantifiable operational guidance for the subsequent iterative decomposition of asymmetric peaks to generate virtual sub-peaks. These peak shape analysis parameters are specifically determined based on the peak shape deviation characteristics and local signal fluctuation characteristics of asymmetric peaks, which can make the decomposition process of asymmetric peaks conform to their actual non-standard shape and local signal background characteristics, avoiding the problems of distortion in the number of sub-peaks and poor peak shape adaptability caused by blind decomposition without parameter guidance.

[0034] In step S103, the peak shape analysis parameters are used to iteratively decompose the signal data corresponding to the asymmetric peak, and corresponding virtual sub-peaks are generated in the iterative decomposition. Based on the signal profile of the virtual sub-peaks, the corresponding signal segments in the signal to be analyzed are stripped to obtain the stripped residual signal.

[0035] In some embodiments, the iterative decomposition of the signal data corresponding to the asymmetric peak using the peak shape analysis parameters, and the generation of the corresponding virtual sub-peaks in the iterative decomposition, are achieved through the following steps: The signal decomposition model of the asymmetric peak is initialized based on the peak shape analysis parameters, and the iteration termination condition is set. The signal decomposition model described above is used to perform an initial decomposition of the signal data corresponding to the asymmetric peak to obtain preliminary sub-peak signals. Based on the peak shape characteristics of the preliminary sub-peak signal, a corresponding virtual sub-peak is generated, and the signal parameters of the virtual sub-peak are recorded. The signal parameters of the virtual sub-peaks are fed back to the signal decomposition model to determine whether the iteration termination condition is met. If the iteration termination condition is not met, the signal decomposition model parameters are adjusted based on the feedback results until a virtual sub-peak that meets the requirements is generated. If the condition is met, the iteration stops.

[0036] In specific implementation, firstly, based on the initial value of the number of virtual sub-peaks, the sub-peak matching peak type, and the iterative decomposition step size parameter in the peak shape analysis parameters, a nonlinear least squares fitting method is selected to construct a signal decomposition model. The aforementioned peak shape analysis parameters are directly assigned as the initial decomposition parameters of the signal decomposition model to complete the initialization of the asymmetric peak signal decomposition model. Simultaneously, according to actual needs, a dual iteration termination condition including residual index threshold constraints and maximum iteration round constraints is set, thereby obtaining the initialized signal decomposition model and iteration termination condition. The maximum iteration round can be set to 50 times according to actual needs. The initialized signal decomposition model refers to the signal decomposition algorithm model constructed and parameter assigned based on the peak shape analysis parameters as the initial basis. The iteration termination condition refers to the constraint condition used to determine whether the iteration process stops. Secondly, the signal data corresponding to the asymmetric peak is directly input into the initialized model. The signal decomposition model uses a nonlinear least squares fitting method with the goal of minimizing the sum of squared residuals to perform an initial fitting decomposition of multiple sub-peaks on the asymmetric peak signal data, outputting the preliminary sub-peak signal. The preliminary sub-peak signal refers to the initial sub-peak signal obtained after the signal decomposition model completes the initial fitting decomposition of the asymmetric peak signal. Further, the peak vertex coordinates, half-peak width, amplitude, and peak area are extracted from the preliminary sub-peak signal and combined to form the peak shape features of the preliminary sub-peak signal. The preliminary sub-peak signal is then reconstructed by standard peak shape fitting according to the sub-peak adaptation peak shape type specified by the peak shape analysis parameters to generate the corresponding virtual sub-peak. At the same time, the signal parameters of the virtual sub-peak are recorded. The signal parameters include peak vertex coordinates, half-peak width, amplitude, and peak area. The signal parameters of the virtual sub-peak refer to the set of peak shape quantification indicators of peak vertex coordinates, half-peak width, amplitude, and peak area corresponding to the virtual sub-peak.Then, the recorded signal parameters of the virtual sub-peaks are fed back to the signal decomposition model. First, the signal intensity residual between the original signal of the asymmetric peak and the superimposed signal of the virtual sub-peaks is calculated point-by-point using the signal decomposition model. The signal intensity residual across the entire peak width is normalized to obtain the signal residual index. Then, the currently calculated signal residual index is numerically compared with the residual signal judgment threshold in the iteration termination condition, and the current iteration round is quantitatively compared with the maximum iteration termination round (if the signal residual index is less than or equal to the residual signal judgment threshold, the iteration termination condition is met; if the signal residual index is still greater than the residual signal judgment threshold but the current iteration round has reached the preset maximum iteration termination round, the iteration termination condition is met; only when the signal residual index is greater than the preset residual signal judgment threshold and the current iteration round has not yet reached the maximum iteration termination round is the iteration termination condition not met). This completes the binary judgment of whether the iteration termination condition is met, and the iteration termination result is output. The iteration termination condition judgment result refers to the binary judgment result characterizing whether the current iteration process meets the termination requirements. The residual signal judgment threshold is set to 5% of the maximum intensity of the original signal. Since the noise level of the chromatographic signal is usually 1% to 3% of the maximum intensity, setting the termination threshold to 5% is slightly higher than the noise level, which can effectively separate the effective component signals without generating false sub-peaks due to overfitting noise. Finally, if the iteration termination condition judgment result is not met, the initial fitting parameters and decomposition step size of the sub-peak of the signal decomposition model are adjusted according to the preset gradient amplitude. Then, the adjusted signal decomposition model is re-decomposed to decompose the asymmetric peak signal and generate new virtual sub-peaks until the signal residual index corresponding to the generated virtual sub-peak is lower than the residual signal judgment threshold or the iteration round reaches the maximum iteration termination round, so as to generate a virtual sub-peak that meets the requirements. If the iteration termination condition judgment result is met, the iteration process is directly stopped, and the virtual sub-peak that meets the requirements is finally output.

[0037] It should be noted that the virtual sub-peaks that meet the requirements in this application refer to virtual sub-peaks that can match the real characteristics of each overlapping component within the asymmetric peak. In this embodiment, the non-standard asymmetric peak signal is decomposed into independent sub-peak signals that conform to the standard peak shape. The virtual sub-peaks correspond to the component characteristic peaks within the asymmetric peak, providing a standardized signal basis for the subsequent extraction of qualitative and quantitative characteristic information of the corresponding components. This avoids the problems of component information aliasing, qualitative deviation and insufficient quantitative accuracy caused by directly analyzing the asymmetric peak as a whole. It ensures that all potential component information within the peak, including low-content weak signals, is completely extracted as the first component information. From the signal analysis level, the real characteristics of each overlapping component behind the asymmetric peak are restored, ensuring that all component information corresponding to the symmetric peak and asymmetric peak in the chemical wastewater test sample can be fully extracted.

[0038] In some embodiments, the following steps are used to strip the corresponding signal segment in the signal to be analyzed and detected based on the signal profile of the virtual sub-peak to obtain the stripped residual signal: Feature matching is performed on the signal profile of the virtual sub-peak to locate the signal segment in the signal to be analyzed that corresponds to the virtual sub-peak; Align the signal profile of the virtual sub-peak with the corresponding signal segment along the wavelength dimension to generate a matched signal profile. Based on the matched signal profile, a signal subtraction algorithm is used to strip the signal segments to obtain the stripped residual signal.

[0039] In specific implementation, firstly, positional features including the abscissa of the peak apex, the abscissa of the peak width start, and the abscissa of the peak width end are extracted from the signal profile of the virtual sub-peak. A feature point matching method is then used to match these positional features point-by-point with the full-band positional features of the signal to be analyzed, thereby locating the signal segment in the signal to be analyzed that corresponds to the virtual sub-peak. This signal segment in the signal to be analyzed refers to the chemical analysis detection signal segment in the signal to be analyzed that matches the positional features of the virtual sub-peak. Secondly, using the wavelength coordinates of the signal to be analyzed as a reference system, linear interpolation is performed on the wavelength coordinates of the signal profile of the virtual sub-peak to achieve a one-to-one correspondence between the wavelength coordinate points of the virtual sub-peak signal profile and the wavelength coordinate points of the corresponding signal segment. Correspondingly, the signal profile of the virtual sub-peak is aligned with the wavelength dimension of the corresponding signal segment to generate a matched signal profile. The matched signal profile refers to the virtual sub-peak signal profile that is consistent with the wavelength coordinates of the corresponding signal segment in the signal to be analyzed and detected. Finally, the matched signal profile is matched with the corresponding signal segment in the signal to be analyzed and detected point by point according to the wavelength coordinates. A point-by-point signal subtraction algorithm is used, with the measured signal intensity of the corresponding signal segment as the minuend and the signal intensity of the matched signal profile as the subtrahend. The signal intensity at each wavelength coordinate point is calculated point by point. All the calculated signal intensity differences are arranged in order according to the wavelength coordinates to form continuous signal data, completing the stripping process of the corresponding signal segment and obtaining the stripped residual signal.

[0040] It should be noted that the residual signal after stripping in this application refers to the detection signal of the signal region corresponding to the stripped virtual sub-peak signal contour in the detection signal to be analyzed. The residual signal after stripping reflects the signal part of the corresponding segment in the detection signal to be analyzed that was not split and analyzed after deducting the generated virtual sub-peak. Its signal shape and intensity characteristics can be directly converted into residual indicators required for iteration termination judgment, becoming the core signal basis for judging whether the current virtual sub-peak splitting is in place and whether it is necessary to continue iterative adjustment. This allows the iteration termination judgment to return from numerical indicators to the signal itself, improving the accuracy of the judgment result.

[0041] In step S104, when the difference between the signal profile of the virtual sub-peak and the residual signal after stripping is lower than a preset difference threshold, the detection information corresponding to the virtual sub-peak is used as the first component information.

[0042] In some embodiments, when the difference between the signal profile of the virtual sub-peak and the stripped residual signal is less than a preset difference threshold, the detection information corresponding to the virtual sub-peak is used as the first component information by the following steps: The signal profile of the virtual sub-peak is used as the reconstructed signal and aligned with the signal dimension of the stripped residual signal to obtain the aligned dual signal. Calculate the signal difference between the aligned two signals; The signal difference is compared with a preset difference threshold. If the signal difference is lower than the preset difference threshold, the information corresponding to the virtual sub-peak is used as the first component information.

[0043] In specific implementation, firstly, the signal profile of the virtual sub-peak is used as the reconstructed signal. Using the horizontal coordinate dimension of the stripped residual signal as the reference frame, linear interpolation is performed on the horizontal coordinate data of the reconstructed signal to achieve a one-to-one correspondence between the horizontal coordinate points of the reconstructed signal and the stripped residual signal, completing the signal dimension alignment operation of the two signals. This aligned two signals are then directly used as input for the next step of calculating the signal difference. The aligned two signals refer to signal data pairs consisting of a reconstructed signal with consistent horizontal coordinate points and a stripped residual signal. Next, the aligned two signals are mapped point-by-point according to their horizontal coordinates. The root mean square error method is used to first calculate the absolute difference in signal intensity between the reconstructed signal and the stripped residual signal at each corresponding coordinate point. Then, all absolute differences across the entire dimension are calculated. The values ​​are squared and summed. The summation result is then compared with the total number of coordinate points, and the square root is calculated to obtain the signal difference degree. The signal difference degree refers to the index characterizing the overall deviation of the signal intensity between the reconstructed signal and the stripped residual signal across all dimensions. Finally, the calculated signal difference degree is compared with the difference threshold preset according to the signal analysis accuracy requirements in chemical analysis and detection. If the calculated signal difference degree is lower than the preset difference threshold, the splitting analysis result of the virtual sub-peak is determined to meet the accuracy requirements. The peak shape parameters, characteristic wavelengths, and signal intensities corresponding to the virtual sub-peak are then combined into the first component information. The preset difference threshold is 3% of the maximum intensity of the original signal. Taking 3% is equivalent to the upper limit of the noise level, which ensures the splitting quality while preventing the system from becoming overly sensitive to noise and unable to terminate due to an excessively strict threshold.

[0044] It should be noted that the first component information in this application refers to the information characterizing the corresponding components in the chemical wastewater test sample formed by various quantitative characteristics of virtual sub-peaks. Determining the first component information can provide qualitative and quantitative basic data obtained after differential and precise analysis for the chemical composition analysis of chemical wastewater test samples. The first component information integrates the characteristic information of all potential components of the sample obtained after iterative decomposition and accuracy verification of asymmetric peaks, which makes up for the problem that traditional unified analysis methods are incomplete in the analysis of asymmetric peaks and are prone to losing component information.

[0045] In step S105, the second component information corresponding to the symmetrical peak is determined, and then a chemical composition analysis report of the chemical wastewater test sample is generated based on the first component information and the second component information and fed back to the network node of the smart sensor.

[0046] In some embodiments, determining the second component information corresponding to the symmetric peak is achieved through the following steps: Baseline correction is performed on the symmetrical peak to obtain the baseline-corrected symmetrical peak signal; The characteristic parameters of the symmetrical peak are calculated based on the baseline-corrected symmetrical peak signal. The characteristic parameters of the symmetric peak are matched with the standard component spectral library to obtain the spectral library matching results; The second component information corresponding to the symmetric peak is determined based on the spectral library matching results.

[0047] In specific implementation, firstly, the original signal data of the symmetrical peak is substituted into the adaptive iterative reweighted penalized least squares method for baseline correction to obtain the baseline-corrected symmetrical peak signal. Secondly, feature parameters including the horizontal and vertical coordinates of the peak apex, half-peak width, characteristic wavelength range, and peak area are extracted from the baseline-corrected symmetrical peak signal. The peak area is calculated using numerical integration, the characteristic wavelength range is confirmed using extremum determination, and the half-peak width is calibrated using difference calculation. The feature parameters of the symmetrical peak refer to a set of indicators characterizing the qualitative and quantitative features of the symmetrical peak. Then, the feature parameters of the symmetrical peak are normalized according to the indicator dimensions of the standard component spectral library to eliminate dimensional and order-of-magnitude differences, transforming them into standardized feature vectors. Finally, a cosine similarity algorithm is used to calculate the similarity between the standardized feature vectors and the reference feature vectors of all standard components in the standard component spectral library. The degree of feature matching is characterized by the cosine value of the angle between the vectors. The matching results are sorted from high to low similarity and labeled with corresponding standard component information to obtain the spectral library matching results. The spectral library matching results refer to the result set containing standard component information and corresponding matching degrees formed by matching the standardized feature vector of the symmetric peak with the standard component spectral library using cosine similarity, and sorting them by similarity. Finally, based on the preset similarity threshold in the field of chemical analysis and detection, the spectral library matching results are screened, and the standard component matching results with the highest similarity and higher than the preset similarity threshold are selected as valid results. The qualitative and quantitative reference information of the standard components corresponding to the valid results are extracted as the second component information corresponding to the symmetric peak. The preset similarity threshold is 0.85 (cosine similarity). A cosine similarity of 0.85 is an empirical dividing value in the field of spectral library matching. A value of 0.85 can effectively filter false positive matches, while preventing the true components from being missed due to differences in spectra between instruments due to excessively high similarity.

[0048] It should be noted that the second component information in this application refers to the set of information characterizing the qualitative identification results and quantitative detection data of the corresponding components in the chemical wastewater test sample. In the intelligent analysis of chemical analysis and detection data, determining the second component information can provide standardized component qualitative and quantitative information corresponding to symmetrical peaks that has been verified by standard peak shape fitting and spectral library tracing for the chemical composition analysis of chemical wastewater test samples. It is a standardized data output of symmetrical peak differentiation analysis in the intelligent analysis process, which complements and synergizes with the first component information obtained from asymmetric peak analysis, and together constitutes the source of complete component information of the sample.

[0049] In some embodiments, the generation of a chemical composition analysis report of the chemical wastewater sample based on the first component information and the second component information, and its feedback to the network node of the smart sensor, is achieved through the following steps: The first component information and the second component information are combined to obtain the complete component information set; A chemical composition analysis report of the chemical wastewater test sample is generated based on the full set of component information and fed back to the network node of the smart sensor.

[0050] In specific implementation, firstly, the information of the first component and the information of the second component are structurally integrated according to the component characteristic index dimension to form a complete component information set containing all qualitative and quantitative indicators of the components. The complete component information set refers to the set of component information contained in the symmetrical and asymmetrical peaks in the signal to be analyzed and detected. Then, the component information in the complete component information set is arranged one by one using a standardized chemical composition analysis report template in chemical analysis and detection to obtain a chemical composition analysis report of the chemical wastewater test sample. Subsequently, the chemical composition analysis report is transmitted via the wireless communication module (such as LoRa, ZigBee or Wi-Fi) built into the smart sensor or an industrial fieldbus interface, based on Modbus, MQTT or OPC. Data is encapsulated using standard protocols such as UA and sent to designated network nodes, such as industrial IoT gateways or cloud platform network nodes, in an event-triggered manner. During the feedback process, the edge computing unit also implements breakpoint resume or data compression strategies based on network status to ensure that the report arrives completely. After receiving the report, the network node returns an acknowledgment frame and can further distribute the report to the central control room monitoring system, mobile applications, or historical databases, thereby completing the information loop from on-site analysis to the management end.

[0051] It should be noted that the chemical composition analysis report of the chemical wastewater test sample in this application refers to the conclusion that integrates the full characteristic information of the effective components of the chemical wastewater test sample.

[0052] In another aspect, in some embodiments, this application provides a detection data analysis system for chemical wastewater treatment, referencing... Figure 4 The figure is a schematic diagram of the structure of a detection data analysis system for chemical wastewater treatment according to some embodiments of this application. The detection data analysis system for chemical wastewater treatment includes: an acquisition module 201, a processing module 202, and an execution module 203, which are described below: The acquisition module 201 in this application is mainly used to acquire the detection signal to be analyzed from the chemical wastewater detection sample collected by the intelligent sensor, and to divide the signal peaks in the detection signal to be analyzed into symmetrical peaks and asymmetrical peaks through the edge computing unit built into the intelligent sensor. Processing module 202, in this application, is mainly used to extract the peak shape deviation characteristics and local signal fluctuation characteristics of the asymmetric peak, and to determine the peak shape analysis parameters of the asymmetric peak based on the peak shape deviation characteristics and the local signal fluctuation characteristics; The processing module 202 is further configured to use the peak shape analysis parameters to iteratively decompose the signal data corresponding to the asymmetric peak, generate corresponding virtual sub-peaks in the iterative decomposition, and strip the corresponding signal segment in the signal to be analyzed and detected based on the signal profile of the virtual sub-peaks to obtain the stripped residual signal. In addition, the processing module 202 is also used to take the detection information corresponding to the virtual sub-peak as the first component information when the difference between the signal profile of the virtual sub-peak and the stripped residual signal is lower than a preset difference threshold. The execution module 203 in this application is mainly used to determine the second component information corresponding to the symmetrical peak, and then generate a chemical composition analysis report of the chemical wastewater test sample based on the first component information and the second component information and feed it back to the network node of the smart sensor.

[0053] In addition, this application also provides a computer device, which includes a memory and a processor. The memory stores code, and the processor is configured to acquire the code and execute the above-described detection data analysis method for chemical wastewater treatment.

[0054] In some embodiments, reference Figure 5 The figure is a schematic diagram of the structure of a computer device for implementing a detection data analysis method for chemical wastewater treatment according to some embodiments of this application. The detection data analysis method for chemical wastewater treatment in the above embodiments can be achieved through... Figure 5 The computer device shown is used to implement this, and the computer device includes at least one processor 301, a communication bus 302, a memory 303, and at least one communication interface 304.

[0055] The processor 301 may be a general-purpose central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more devices used to control the execution of the detection data analysis method for chemical wastewater treatment in this application.

[0056] The communication bus 302 can be used to transmit information between the aforementioned components.

[0057] The memory 303 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disks or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto. The memory 303 may exist independently and be connected to the processor 301 via the communication bus 302. The memory 303 may also be integrated with the processor 301.

[0058] The memory 303 stores program code for executing the scheme of this application, and its execution is controlled by the processor 301. The processor 301 executes the program code stored in the memory 303. The program code may include one or more software modules. In the above embodiments, the determination of the detection data analysis method for chemical wastewater treatment can be implemented by the processor 301 and one or more software modules in the program code in the memory 303.

[0059] Communication interface 304 uses any transceiver-like device for communicating with other devices or communication networks, such as Ethernet, radio access network (RAN), wireless local area networks (WLAN), etc.

[0060] In a specific implementation, as one example, a computer device may include multiple processors, each of which may be a single-core (single-CPU) processor or a multi-core (multi-CPU) processor. Here, a processor may refer to one or more devices, circuits, and / or processing cores used to process data (e.g., computer program instructions).

[0061] The aforementioned computer device can be a general-purpose computer device or a special-purpose computer device. In specific implementations, the computer device can be a desktop computer, a portable computer, a network server, a handheld digital assistant (PDA), a mobile phone, a tablet computer, a wireless terminal device, a communication device, or an embedded device. This application does not limit the type of computer device.

[0062] In addition, this application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described detection data analysis method for chemical wastewater treatment.

[0063] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0064] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for analyzing detection data in chemical wastewater treatment, characterized in that, Includes the following steps: The system acquires the detection signal of the chemical wastewater sample collected by the intelligent sensor, and divides the signal peaks in the detection signal into symmetrical peaks and asymmetrical peaks through the edge computing unit built into the intelligent sensor. Extract the peak shape deviation characteristics and local signal fluctuation characteristics of the asymmetric peak, and determine the peak shape analysis parameters of the asymmetric peak based on the peak shape deviation characteristics and the local signal fluctuation characteristics; The peak shape analysis parameters are used to iteratively decompose the signal data corresponding to the asymmetric peak, and corresponding virtual sub-peaks are generated in the iterative decomposition. Based on the signal profile of the virtual sub-peaks, the corresponding signal segments in the signal to be analyzed are stripped to obtain the stripped residual signal. When the difference between the signal profile of the virtual sub-peak and the residual signal after stripping is lower than a preset difference threshold, the detection information corresponding to the virtual sub-peak is used as the first component information. The second component information corresponding to the symmetrical peak is determined, and then a chemical composition analysis report of the chemical wastewater test sample is generated based on the first component information and the second component information and fed back to the network node of the smart sensor.

2. The method as described in claim 1, characterized in that, The edge computing unit built into the smart sensor divides the signal peaks in the signal to be analyzed into symmetrical and asymmetrical peaks, specifically including: The edge computing unit built into the smart sensor performs smoothing preprocessing on the detection signal to be analyzed to obtain a preprocessed detection signal. Based on the preprocessed detection signal, the peak apex coordinates and peak width range of each signal peak are located; For each signal peak, the signal intensity difference between corresponding positions on both sides within the peak width is compared with the peak apex as the axis of symmetry, and then the symmetry value of each signal peak is calculated. The symmetry value of each signal peak is compared with the preset symmetry threshold, and the corresponding signal peaks are divided into symmetric peaks and asymmetric peaks based on the comparison results.

3. The method as described in claim 1, characterized in that, Extracting the peak shape deviation characteristics and local signal fluctuation characteristics of the asymmetric peak specifically includes: The asymmetric peak is subjected to baseline correction processing to obtain the baseline-corrected asymmetric peak signal; The theoretical symmetric reference profile of the asymmetric peak is obtained by fitting the asymmetric peak signal after baseline correction. The deviation between the baseline-corrected asymmetric peak signal and the theoretical symmetric reference profile is calculated to extract the peak shape deviation feature; The local signal segment corresponding to the baseline-corrected asymmetric peak signal is divided from the signal to be analyzed, and the signal fluctuation coefficient of the local signal segment is calculated to extract the local signal fluctuation characteristics.

4. The method as described in claim 1, characterized in that, The peak shape analysis parameters for determining the asymmetric peak based on the peak shape deviation characteristics and the local signal fluctuation characteristics specifically include: The entropy weight method is used to calculate the weights of the peak deviation feature and the local signal fluctuation feature to obtain the weight coefficients corresponding to the peak deviation feature and the local signal fluctuation feature. Based on the weighting coefficients, the peak deviation feature and the local signal fluctuation feature are weighted and fused to generate an asymmetric peak feature vector; The asymmetric peak feature vector is calculated using a pre-trained peak shape parameter mapping model to output the peak shape analysis parameters of the asymmetric peak.

5. The method as described in claim 1, characterized in that, The specific steps of using the peak shape analysis parameters to iteratively decompose the signal data corresponding to the asymmetric peak and generate the corresponding virtual sub-peaks during the iterative decomposition include: The signal decomposition model of the asymmetric peak is initialized based on the peak shape analysis parameters, and the iteration termination condition is set. The signal decomposition model described above is used to perform an initial decomposition of the signal data corresponding to the asymmetric peak to obtain preliminary sub-peak signals. Based on the peak shape characteristics of the preliminary sub-peak signal, a corresponding virtual sub-peak is generated, and the signal parameters of the virtual sub-peak are recorded. The signal parameters of the virtual sub-peaks are fed back to the signal decomposition model to determine whether the iteration termination condition is met. If the iteration termination condition is not met, the signal decomposition model parameters are adjusted based on the feedback results until a virtual sub-peak that meets the requirements is generated. If the condition is met, the iteration stops.

6. The method as described in claim 1, characterized in that, Determining the second component information corresponding to the symmetric peak specifically includes: Baseline correction is performed on the symmetrical peak to obtain the baseline-corrected symmetrical peak signal; The characteristic parameters of the symmetrical peak are calculated based on the baseline-corrected symmetrical peak signal. The characteristic parameters of the symmetric peak are matched with the standard component spectral library to obtain the spectral library matching results; The second component information corresponding to the symmetric peak is determined based on the spectral library matching results.

7. The method as described in claim 1, characterized in that, The analytical signals of chemical wastewater samples collected by intelligent sensors are obtained using a chromatograph.

8. A detection data analysis system for chemical wastewater treatment, used to execute the detection data analysis method for chemical wastewater treatment as described in any one of claims 1 to 7, characterized in that, The system includes: The acquisition module is used to acquire the detection signal to be analyzed from the chemical wastewater sample collected by the intelligent sensor, and to divide the signal peaks in the detection signal to be analyzed into symmetrical peaks and asymmetrical peaks through the edge computing unit built into the intelligent sensor. The processing module is used to extract the peak shape deviation characteristics and local signal fluctuation characteristics of the asymmetric peak, and determine the peak shape analysis parameters of the asymmetric peak based on the peak shape deviation characteristics and the local signal fluctuation characteristics; The processing module is further configured to use the peak shape analysis parameters to iteratively decompose the signal data corresponding to the asymmetric peak, generate corresponding virtual sub-peaks in the iterative decomposition, and strip the corresponding signal segment in the signal to be analyzed and detected based on the signal profile of the virtual sub-peaks to obtain the stripped residual signal. The processing module is further configured to use the detection information corresponding to the virtual sub-peak as the first component information when the difference between the signal profile of the virtual sub-peak and the stripped residual signal is lower than a preset difference threshold. The execution module is used to determine the second component information corresponding to the symmetrical peak, and then generate a chemical composition analysis report of the chemical wastewater test sample based on the first component information and the second component information, and feed it back to the network node of the smart sensor.

9. A computer device, characterized in that, The computer device includes a memory and a processor, the memory storing code, and the processor being configured to acquire the code and execute the detection data analysis method for chemical wastewater treatment as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the detection data analysis method for chemical wastewater treatment as described in any one of claims 1 to 7.

Images