A method for removing pollutants from papermaking wastewater based on nanofiltration staged separation

Abstract

This invention relates to the field of wastewater treatment technology, specifically disclosing a method for removing pollutants from papermaking wastewater based on nanofiltration staged separation. The method includes: collecting real-time conformational parameters of lignin in the papermaking wastewater from the nanofiltration system, calculating the conformational fluctuation rate to determine whether conformational imbalance analysis is needed; if needed, performing conformational imbalance analysis in conjunction with nanofiltration membrane characteristic parameters to determine whether lignin conformational dynamic trend tracking is triggered; extracting conformational dynamic trends and distribution shift characteristics to divide the nanofiltration system into imbalance intervals; if entering an imbalance interval, fitting a quantitative relationship between the lignin conformational transformation rate and the decay of retention efficiency, constructing a retention response model and conformational control boundary; when real-time conformational parameters reach the boundary, initiating targeted conformational adjustment according to the type of imbalance interval. This invention achieves efficient optimization of the compatibility between lignin in papermaking wastewater and the nanofiltration membrane, which is beneficial for improving pollutant removal efficiency and the stability of the nanofiltration system.

Description

Technical Field

[0001] This invention relates to the field of wastewater treatment technology, specifically to a method for removing pollutants from papermaking wastewater based on nanofiltration staged separation. Background Technology

[0002] In the field of wastewater treatment in the paper industry, nanofiltration technology is widely used because it can effectively remove large molecular pollutants such as lignin and achieve water resource recycling. However, existing nanofiltration systems still face the following technical bottlenecks in the treatment process:

[0003] Current technologies focus solely on the static physicochemical properties of lignin (such as total content and average molecular weight), failing to recognize that lignin undergoes conformational changes under conditions such as wastewater flow and pressure variations. These changes include sudden increases or decreases in aggregate size and transitions in molecular chains from contracted to extended states. Due to the lack of real-time monitoring methods for conformational parameters, current technologies cannot capture these dynamic changes. Consequently, when problems such as membrane pore blockage and retention rate fluctuations occur, it is difficult to trace the root cause at the conformational level, resulting in a delayed system response.

[0004] When nanofiltration membranes become clogged or their retention stability declines, existing technologies struggle to distinguish whether the cause is lignin conformational shift or membrane aging itself. There is a lack of quantitative correlation analysis between conformational dynamics and membrane performance deterioration types, making it impossible to determine whether conformational degradation or membrane deterioration is the core cause, leading to blind control measures.

[0005] Therefore, the present invention provides a method for removing pollutants from papermaking wastewater based on nanofiltration staged separation. Summary of the Invention

[0006] The purpose of this invention is to provide a method for removing pollutants from papermaking wastewater based on nanofiltration staged separation, so as to solve the problems mentioned above.

[0007] The objective of this invention can be achieved through the following technical solutions:

[0008] A method for removing pollutants from papermaking wastewater based on nanofiltration staged separation includes the following steps:

[0009] Real-time conformational parameters of lignin in papermaking wastewater in nanofiltration system are collected. Conformational dynamic analysis is performed on the real-time conformational parameters to obtain the conformational fluctuation rate, and it is determined whether it is necessary to trigger conformational imbalance analysis between lignin and nanofiltration membrane.

[0010] If conformational imbalance analysis is required, a conformational characteristic data set is established based on real-time conformational parameters, and conformational imbalance analysis is performed on the conformational characteristic data set to determine whether lignin conformational dynamic trend tracking is triggered.

[0011] If lignin conformation dynamic trend tracking is triggered, the conformation dynamic trend and distribution shift characteristics are extracted based on the conformation characteristic data set, and conformation shift analysis is performed to obtain the imbalance interval of the nanofiltration system.

[0012] If the nanofiltration system enters the imbalance region, the quantitative relationship between the lignin conformational transition rate and the decay of retention efficiency is fitted. Based on the quantitative relationship, a retention response model under conformational disturbance is constructed. The conformational regulation boundary is constructed through the retention response model. If the real-time conformational parameters reach the regulation boundary, conformational regulation is initiated.

[0013] As a further aspect of the present invention, the method for performing the conformational dynamic analysis is as follows:

[0014] Obtain the average particle size and particle size distribution width of aggregates from real-time conformational parameters;

[0015] Calculate the rate of change of the average particle size of the aggregates during the monitoring period to obtain the particle size change rate;

[0016] Calculate the rate of change of the average stretching of the molecular chain during the monitoring period to obtain the stretching change rate.

[0017] The conformational fluctuation rate is obtained by weighted summation of the particle size change rate and the stretching change rate.

[0018] Based on the comparative analysis of conformational volatility, the determination result of whether conformational imbalance analysis is triggered is obtained.

[0019] As a further aspect of the present invention, the conformational imbalance analysis is performed as follows:

[0020] Obtain real-time characteristic parameters of the nanofiltration membrane and combine them with real-time conformational parameters to establish a conformational characteristic data set;

[0021] Lignin retention rate was extracted from the conformational characteristic data set, and nonlinear amplification-triggered analysis was performed in combination with conformational change rate to obtain nonlinear response ratio;

[0022] A joint distribution model of conformation-nanofiltration membrane parameters was constructed, and the rate of change of adaptation entropy was obtained by performing ordered imbalance analysis.

[0023] The kurtosis of the adaptation entropy change rate is extracted, and a three-dimensional joint criterion is constructed based on the nonlinear response ratio, kurtosis, and adaptation entropy change rate to determine whether the dynamic trend tracking analysis of lignin conformation is triggered.

[0024] As a further aspect of the present invention, the method for performing the ordered imbalance analysis is as follows:

[0025] Based on the average particle size of aggregates, membrane flux and rejection rate in the conformational property data set as joint features, the three-dimensional distribution of conformation was fitted to obtain the conformational characteristics.

[0026] Based on the three-dimensional distribution of conformation, the conformational adaptation entropy is obtained through the entropy equation.

[0027] The rate of change of conformational adaptation entropy for the current and adjacent monitoring periods is obtained and calculated to obtain the rate of change of adaptation entropy.

[0028] As a further aspect of the present invention: the three-dimensional joint criterion is constructed as follows:

[0029] Nonlinear response ratio, distribution kurtosis and rate of change of fit entropy are used as three-dimensional joint features;

[0030] A baseline library of historical joint criteria for constructing nanofiltration systems;

[0031] Calculate the 3D deviation between the joint 3D features and the normal baseline library;

[0032] The collaborative deviation index is obtained by performing collaborative deviation analysis based on three-dimensional deviation.

[0033] Trajectory similarity is determined based on the co-deviation index, and if the trajectories are similar, the dynamic trend of lignin conformation is tracked.

[0034] As a further aspect of the present invention, the method for performing conformational shift analysis is as follows:

[0035] Extracting dynamic trend indicators for the quantification of lignin dynamic characteristics;

[0036] Extract the distribution migration index of conformational migration;

[0037] By combining dynamic trend indicators and distribution shift indicators, we can analyze the relationship between conformational shift and nanofiltration membrane performance degradation.

[0038] If the clogging is conformation-dominant, a stability analysis is performed on the target retention of the nanofiltration membrane to determine whether conformation-dominant retention is unstable.

[0039] Based on the results of stability analysis, imbalance intervals are divided by combining trajectory similarity and cooperative deviation index.

[0040] As a further aspect of the present invention: the method for analyzing the type of performance degradation of the nanofiltration membrane is as follows:

[0041] Obtain and calculate the second derivative of the membrane specific flux in the characteristic parameters of the nanofiltration membrane;

[0042] Using partial least squares regression algorithm, the rate of change of the average particle size of aggregates in the dynamic trend index during the continuous monitoring period and the rate of change of the particle size distribution width are used as independent variables, and the second derivative of the membrane specific flux is used as the dependent variable.

[0043] Obtain the absolute value of the regression coefficients output by the partial least squares regression algorithm, and the sum of the absolute values ​​of all coefficients;

[0044] The absolute value of the regression coefficient is compared with the sum of the absolute values ​​of all coefficients to obtain the contribution weight of the agglomerate particle size shift to the blockage.

[0045] Critical clogging analysis is performed based on the second derivative of contribution weight and membrane specific flux to determine whether it is conformation-dominated clogging.

[0046] As a further aspect of the present invention: the method for determining whether it is a conformation-dominant type of trapped instability is as follows:

[0047] If the blockage is conformation-dominant, calculate the coefficient of variation of the retention rate within the characteristic Dalton interval;

[0048] Based on the random regression forest algorithm, an extension-trap fluctuation correlation model is constructed to calculate the contribution weight of molecular chain extension offset to trap fluctuation.

[0049] Based on the coefficient of variation and contribution weight, a comparative analysis using a baseline library is conducted to determine whether it is a conformation-dominant type of retention instability.

[0050] As a further aspect of the present invention: the conformational regulation boundary is constructed in the following way:

[0051] Extract the average particle size and molecular chain extension of the aggregates from the conformational property data set and combine them to construct the core conformation set;

[0052] A retention response model is constructed. The core conformation set is input into the retention response model, and the retention response model outputs the predicted values ​​of membrane specific flux and retention rate for future monitoring cycles.

[0053] The regulatory boundary conditions are constructed based on conformational adaptation entropy, membrane specific flux prediction, retention rate prediction, and co-deviation index.

[0054] A two-dimensional control boundary was constructed using the average particle size and molecular chain extension of the aggregates as the coordinate axis;

[0055] The core conformation set that satisfies the regulatory boundary is marked as the boundary point outside the regulatory boundary;

[0056] A continuous control boundary curve is constructed by fitting the boundary points with a polynomial function.

[0057] As a further aspect of the present invention: the method for constructing the interception response model is as follows:

[0058] A cut-off response model is constructed using a long short-term memory network algorithm, and dynamic trend indicators, distribution offset indicators, and imbalance interval identifiers are input into the cut-off response model.

[0059] The beneficial effects of this invention are:

[0060] (1) The average particle size of lignin aggregates and the molecular chain extension are collected as conformational parameters. The conformational fluctuation rate is obtained by weighted calculation of the particle size change rate and the extension change rate, so as to realize the real-time capture and quantitative characterization of the dynamic changes in lignin conformation. This is helpful to identify conformational fluctuation anomalies, provide a triggering basis for whether to initiate conformational imbalance analysis, reduce the deterioration of membrane performance caused by the failure to detect conformational mutations in time, and improve the timeliness of system response.

[0061] (2) By constructing a conformational characteristic data set, the nonlinear response ratio is calculated by combining the nonlinear response equation, the adaptation entropy change rate is analyzed by the entropy equation, and the imbalance precursor analysis is performed based on the three-dimensional joint criteria. This achieves a quantitative assessment of the correlation between conformation and membrane performance. By using the normal baseline library and the three-dimensional deviation calculation, the accuracy of conformational dynamic trend tracking and trigger judgment is improved, and the reliability of imbalance early warning is enhanced.

[0062] (3) Dynamic trend indicators such as the rate of change in aggregate particle size and the slope of molecular chain extension shift, as well as distribution shift characteristics such as the rate of change in particle size distribution width, are extracted. Conformation-dominant blockage is analyzed by partial least squares regression, and conformation-dominant interception instability is determined by random forest algorithm. Finally, the imbalance interval is divided based on trajectory similarity and cooperative deviation index. This enables the attribution of the impact of conformation shift on membrane performance deterioration, clarifies the degree and type of membrane imbalance, and provides a basis for the targeted formulation of subsequent regulation strategies.

[0063] (4) The quantitative relationship between conformational transition rate and retention efficiency decay is fitted by nonlinear least squares method. A retention response model is constructed using long short-term memory network to predict future membrane performance. A two-dimensional control boundary is constructed with the average particle size of aggregates and the molecular chain extension as the core, and differentiated regulation is initiated according to the type of imbalance interval. This achieves predictive conformational regulation. By using dynamic boundary and prediction model, the control measures can be targeted to solve conformation-dominated problems, which is conducive to timely recovery of membrane performance and improves the stability and pollutant removal efficiency of nanofiltration system. Attached Figure Description

[0064] The invention will now be further described with reference to the accompanying drawings.

[0065] Figure 1 This is a flowchart of a method for removing pollutants from papermaking wastewater based on nanofiltration and hierarchical separation according to the present invention.

[0066] Figure 2 This is a flowchart of the determination process for conformational parameters of the present invention to reach the control boundary;

[0067] Figure 3 This is a block diagram of a papermaking wastewater pollutant removal system based on nanofiltration and graded separation, as described in this invention. Detailed Implementation

[0068] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0069] Example 1

[0070] Please see Figure 1 As shown, this invention is a method for removing pollutants from papermaking wastewater based on nanofiltration staged separation, comprising the following steps:

[0071] S1. Collect real-time conformational parameters of lignin in papermaking wastewater in nanofiltration system, perform conformational dynamic analysis on real-time conformational parameters to obtain conformational fluctuation rate, and determine whether it is necessary to trigger conformational imbalance analysis between lignin and nanofiltration membrane.

[0072] The method for collecting real-time conformational parameters of lignin in papermaking wastewater and characteristic parameters of the nanofiltration membrane in the nanofiltration system is as follows:

[0073] Preferably, online monitoring equipment is installed at the nanofiltration system inlet pipe, nanofiltration membrane module inlet, and concentrate outlet.

[0074] Includes: Dynamic light dispersion (DLS) sensor, used to measure in real time the lignin solution in the graded separation of papermaking wastewater by nanofiltration system. DLS uses laser to irradiate lignin aggregates and utilizes the fluctuation of scattered light intensity generated by the Brownian motion of particles to calculate the particle size distribution by cumulative analysis.

[0075] Among them, particle size distribution includes: the average particle size of lignin aggregates and the particle size distribution width;

[0076] Online UV-Vis spectrometer: Absorbance A at the lignin characteristic peak at 280 nm 280 A through lignin standards with known extension 280 The curve was used to establish a quantitative relationship between absorbance and molecular chain extension, and the average extension of the molecular chain was obtained.

[0077] For example, through the formula: Obtain the average extension Sz of the molecular chain, where , For calibration coefficients;

[0078] As those skilled in the art will understand, since the extension of the lignin molecular chain is directly related to the degree of exposure of its conjugated structure (benzene ring, double bond, etc.), the more extended the molecular chain segment, the easier it is for the conjugated groups to interact with ultraviolet light, and the higher the intensity of the characteristic absorption peak at 280 nm.

[0079] The average particle size and particle size distribution width of the aggregates are used as real-time conformational parameters.

[0080] Extract real-time conformational parameters collected during the monitoring period, calculate the rate of change of the average particle size of aggregates at the initial and final values ​​during the monitoring period, and obtain the particle size change rate.

[0081] The rate of change of the average stretching of the molecular chain at the initial and final values ​​within the monitoring period is simultaneously acquired and calculated to obtain the stretching change rate.

[0082] The conformational fluctuation rate is obtained by weighted summation of the particle size change rate and the stretching change rate.

[0083] Among them, the particle size change rate is weighted at 0.6 and the molecular chain extension change rate is weighted at 0.4 in the conformational fluctuation rate. This is because the particle size change is directly related to membrane pore blockage and has a more significant impact on the continuity of system operation, while the extension change mainly affects the retention stability and has a weaker immediate impact. This allocation can accurately quantify the actual risk of conformational fluctuation to the system and provide a basis for judging whether conformational imbalance analysis is triggered.

[0084] Understandably, conformational volatility quantifies the intensity of dynamic changes in lignin conformation. The physical meaning of conformational volatility is: by weighted fusion, it reflects the rate of change of the average particle size of aggregates that reflects the fluctuation of the aggregate state and the rate of change of the average extension of molecular chains that reflects the fluctuation of the spatial morphology of molecular chains. It characterizes the overall amplitude and rate of conformational volatility of lignin from macroscopic aggregates to microscopic molecular chain morphology within the monitoring period, and intuitively reflects the degree to which the conformation deviates from the stable state.

[0085] The purpose of calculating conformational volatility is as follows:

[0086] Function 1: As a trigger for conformational imbalance analysis, the conformational volatility is compared with a preset threshold to directly determine whether to initiate subsequent conformational imbalance analysis. When the volatility exceeds the threshold, it indicates that the conformational change has exceeded the normal stable range, and further analysis is needed to determine whether the conformation is mismatched with the nanofiltration membrane, providing a trigger criterion for subsequent processes.

[0087] Secondly, it enables real-time capture of abnormal conformational fluctuations, reducing membrane performance degradation. Conformational fluctuation rate allows for real-time quantification of dynamic changes in lignin conformation, promptly identifying sudden conformational changes such as abrupt increases in aggregate particle size and excessive molecular chain extension. If anomalies are not captured in time, they may lead to membrane pore blockage and a sharp drop in retention rate. Therefore, conformational fluctuation rate can provide early warning of potential risks and improve the system's responsiveness.

[0088] Thirdly, it provides basic data for subsequent conformational dynamic trend analysis. Conformational volatility is a core basic parameter characterizing the dynamic characteristics of conformation, supporting in-depth analysis of the direction, rate and impact of conformational shift, forming a logical closed loop of real-time monitoring-trend analysis-control.

[0089] The conformational volatility is compared with a preset conformational volatility threshold. If the conformational volatility is higher than the preset conformational volatility threshold, conformational imbalance analysis is triggered.

[0090] S2. If conformational imbalance analysis is required, a conformational characteristic data set is established based on real-time conformational parameters. Conformational imbalance analysis is performed on the conformational characteristic data set to determine whether lignin conformational dynamic trend tracking is triggered.

[0091] The method for obtaining real-time feature parameters and real-time conformational parameters to establish a conformational characteristic data set is as follows:

[0092] The interception rate, specific flux, and pore size of the nanofiltration membrane in the theoretical and real-time characteristic Dalton range are obtained as characteristic parameters of the nanofiltration membrane.

[0093] Preferably, the characteristic Dalton range is 200-1000;

[0094] Understandably, the theoretical characteristic parameters of nanofiltration membranes are verified through offline experiments using nominal parameters provided by the membrane manufacturer, such as nominal molecular weight cutoff, pore size distribution, and surface charge density, to determine the actual retention performance within the characteristic Dalton range. This involves preparing simulated solutions using standard substances (such as PEG200, PEG500, and PEG1000), testing the membrane's retention rate for substances of different molecular weights, plotting molecular weight-retention rate curves, and determining the retention rate range of 200-1000 Daltons and the critical retention point (e.g., the retention rate for substances with a molecular weight of 500 Daltons).

[0095] The real-time characteristic parameters of the nanofiltration membrane are collected by an online gel permeation chromatography (GPC) module, a pressure sensor, and a flow meter, including the real-time interception rate and specific flux within the characteristic Dalton range.

[0096] The real-time conformational parameters and the real-time characteristic parameters of the nanofiltration membrane are spatiotemporally aligned. The spatiotemporally aligned real-time conformational parameters and the characteristic parameters of the nanofiltration membrane are then combined to obtain a conformational characteristic data set.

[0097] It is understandable that by matching the time synchronization of the acquisition time of the same monitoring period and the spatial correlation of the same monitoring location of the corresponding nanofiltration system, the spatiotemporal alignment of real-time conformational parameters and nanofiltration membrane characteristic parameters can be achieved, and then the aligned data can be integrated into a conformational characteristic data set.

[0098] The method for performing conformational imbalance analysis on the conformational property data set to determine whether lignin conformational dynamic trend tracking is triggered is as follows:

[0099] S201. Obtain the cutoff rate of change and combine it with the conformational change rate to perform nonlinear amplification trigger analysis to obtain the nonlinear response ratio;

[0100] Obtain the rate of change of lignin retention rate in the conformational characteristic data set within the monitoring period to obtain the retention change rate;

[0101] Based on the cutoff rate of change and combined with the conformational change rate, nonlinear amplification trigger analysis is performed to obtain the nonlinear response ratio of the conformation;

[0102] Preferably, through a nonlinear response equation: Obtain the nonlinear response ratio of the conformation ;

[0103] in, For real-time conformational change rate, To capture the rate of change in real time, The rate of change of conformational baseline, To retain the benchmark rate of change;

[0104] It should be explained that the conformational reference rate of change refers to the normal rate of change of conformational parameters when the system is running at a stable state, corresponding to the cutoff of the dynamic reference value of the conformation in the stable state. The normal fluctuation rate of lignin retention rate during stable system operation is used as the dynamic benchmark value for stable retention performance, with the retention benchmark change rate as the benchmark.

[0105] S202. Construct a joint distribution model of conformation-nanofiltration membrane parameters and perform ordered imbalance analysis to obtain the rate of change of adaptation entropy;

[0106] Preferably, the three-dimensional conformational distribution is obtained by fitting the average particle size D of aggregates, membrane flux J, and rejection rate R in the conformational characteristic data set as joint features. ;

[0107] Based on the three-dimensional distribution of the conformation, through the entropy equation: Obtain the conformational adaptation entropy H;

[0108] Those skilled in the art will understand that data should first be collected during continuous operation of the nanofiltration system. Time-series data is then mapped to a particle size-flux-rejection space (each axis defines a physically feasible interval, such as...). , , Finally, the frequency of combined states of tiny intervals in three-dimensional space is calculated using a multivariate Gaussian distribution, which serves as the three-dimensional distribution of the conformation. The probability value within this range is used to quantify the probabilistic relationship between the three factors.

[0109] The domain of the three-dimensional integral in the entropy equation is... Cartesian product of the three physically feasible intervals;

[0110] For example: , For: the actual range of aggregate particle size: that is, from the initial particle size Critical particle size for membrane pore blockage ;

[0111] For: The actual range of membrane flux: from the initial flux To the acceptable lower limit of decay ;

[0112] To determine the actual range of the retention rate, from the design Retention rate to allowable fluctuation threshold :

[0113] Obtain and calculate the rate of change of conformational adaptation entropy H for the current and adjacent monitoring periods to obtain the rate of change of adaptation entropy.

[0114] Extract the kurtosis of the adaptation entropy change rate based on the nonlinear response ratio. A three-dimensional joint criterion is constructed using distribution kurtosis and adaptation entropy change rate to determine whether dynamic trend tracking analysis of lignin conformation is triggered.

[0115] Among them, by constructing the time series probability distribution of the adaptation entropy change rate, and then calculating the ratio of the fourth central moment to the square of the second central moment of the distribution, the kurtosis of the distribution is obtained.

[0116] The method for constructing the three-dimensional joint criterion is as follows:

[0117] nonlinear response ratio Kuness and rate of change of fitness entropy are used as three-dimensional joint features;

[0118] S211, a normal baseline library of historical joint criteria for framework nanofiltration systems;

[0119] Preferably, during the stable operation phase of the nanofiltration system, a clustering algorithm is used to identify low-fluctuation, high-fit samples and extract the nonlinear response ratio. Baseline distribution models of kurtosis and adaptation entropy change rate are used to construct a normal baseline library based on the three types of baseline distribution models.

[0120] Those skilled in the art will understand that when constructing a normal baseline library for historical joint criteria of nanofiltration systems, historical data from the stable operating period of the system are first screened, and samples with "low fluctuation (small fluctuations in conformation and membrane performance parameters)" and "high fit (high matching degree between conformation and membrane performance, such as low fit entropy)" are identified through clustering algorithms. For these samples, three types of indicators are extracted: nonlinear response ratio (the degree of nonlinearity of the dynamic correlation between conformation and retention), kurtosis (the steepness of the distribution of the rate of change of fit entropy, reflecting the concentration of change), and rate of change of fit entropy (the rate of change of fit entropy over time). Distribution models (such as statistical mean, standard deviation, or probability distribution parameters) are established for each of these indicators in the steady state. Finally, the three types of baseline models are integrated to form a normal baseline library, providing a steady-state reference threshold and distribution characteristics for subsequent judgment of conformational imbalance.

[0121] S212. Calculate the three-dimensional deviation between the three-dimensional joint features and the normal baseline library;

[0122] Preferably, according to formula one: Obtain the deviation of the nonlinear response ;

[0123] in, This is the cumulative distribution function of the standard normal distribution, used to calculate the probability of this value in the steady state. These are the nonlinear response ratio calculated in real time and the baseline in the normal operating environment library, respectively, during stable operation. The mean;

[0124] In the normal baseline library, during stable operation Standard deviation;

[0125] Through formula two: Obtain the entropy change deviation of conformational adaptation entropy ;

[0126] Where H is the rate of change of conformational adaptation entropy calculated in real time, In a steady state, the rate of change of the adaptation entropy is lower than the real-time value. The probability of;

[0127] in, The probability density function in the baseline distribution model to adapt to the rate of entropy change;

[0128] Through formula three: Get kurtosis deviation ;

[0129] in, These represent the minimum and maximum kurtosis values ​​during stable operation, respectively, in the normal baseline library.

[0130] Will , , As a three-dimensional deviation;

[0131] It should be noted that, , , The values ​​of are all in the range of [0,1]. A larger value indicates a more significant deviation from the normal state. A larger value indicates that the entropy decreases much faster than normal. This indicates that deviation occurs only when the kurtosis exceeds the upper limit of the normal range; the larger the value, the sharper the distribution.

[0132] S213. Based on the three-dimensional deviation degree, perform collaborative deviation analysis to obtain the collaborative deviation index;

[0133] Through the formula:

[0134] Obtain the co-dispersion index C;

[0135] in, The Pearson correlation coefficient is the sum of the nonlinear response ratio, kurtosis, and rate of change of fit entropy in the normal baseline library. In the normal baseline database, the Pearson correlation coefficient between the rate of change of adaptation entropy and the kurtosis of the distribution is... In the normal baseline library, the Pearson correlation coefficient between the nonlinear response ratio and the kurtosis of the distribution;

[0136] Used to reflect the synergistic relationship of the three characteristics under normal conditions;

[0137] S214. Based on the co-deviation index, determine the trajectory similarity. If the trajectories are similar, track the dynamic trend of lignin conformation.

[0138] Obtain the coordination deviation index over M monitoring periods and construct a coordination deviation sequence;

[0139] Unbalanced sample trajectories from the normal baseline database are obtained and subjected to dynamic time warping (DTW) analysis, with trajectory similarity C.

[0140] When the following conditions are met: Furthermore, if the trajectory similarity is lower than the preset trajectory similarity threshold, the tracking of the dynamic trend of lignin conformation will be triggered.

[0141] in, This is a trajectory similarity dataset of unimbalanced samples in a normal baseline database. for the median of It is the threshold for outlier detection.

[0142] Example 2

[0143] like Figure 1As shown, the present invention is a method for removing pollutants from papermaking wastewater based on nanofiltration staged separation, and further includes the following steps:

[0144] S3. If lignin conformation dynamic trend tracking is triggered, the conformation dynamic trend and distribution shift characteristics are extracted based on the conformation characteristic data set, and conformation shift analysis is performed to obtain the imbalance interval of the nanofiltration system.

[0145] The method for extracting conformational dynamic trends and distribution shift characteristics based on conformational property data sets, and then performing conformational shift analysis to obtain the imbalance interval of the nanofiltration system is as follows:

[0146] S301, Extracting dynamic trend indicators for quantifying the dynamic characteristics of lignin;

[0147] Preferably, by formula: Obtain the rate of change of the average particle size of aggregates within a continuous monitoring period. ,

[0148] Among them, the continuous monitoring period is higher than 3. This represents the change in the average particle size of aggregates due to the extension of the molecular chain over a continuous monitoring period. The duration of the continuous monitoring period;

[0149] The offset slope of the average extension of the molecular chain and the acceleration of the conformational fluctuation rate are obtained within a continuous monitoring period.

[0150] The slope of the average extension of the molecular chain, the acceleration of the conformational fluctuation rate, and the rate of change of the average particle size of the aggregates within a continuous monitoring period are used as dynamic trend indicators.

[0151] S302. Extract the distribution offset index of conformational shift;

[0152] Obtain the rate of change of particle size distribution width within a continuous monitoring period;

[0153] The rate of change of distribution width is obtained by calculating the difference between the particle size distribution width in the current monitoring period and the distribution width in the baseline library, and then processing the difference to the distribution width in the baseline library to obtain the distribution offset index.

[0154] S303. Combining dynamic trend indicators and distribution shift indicators, we analyze the type of conformational shift and nanofiltration membrane performance degradation.

[0155] The second derivative of the membrane specific flux, a characteristic parameter of the nanofiltration membrane, was calculated. Then, using partial least squares regression, the rate of change of the average particle size of aggregates, a dynamic trend indicator, over a continuous monitoring period was calculated. The rate of change of particle size distribution width is used as the independent variable, and the second derivative of membrane specific flux is used as the dependent variable.

[0156] Obtain the output of the partial least squares regression algorithm The absolute value of the regression coefficient, and the sum of the absolute values ​​of all coefficients;

[0157] Will The absolute value of the regression coefficient is compared with the sum of the absolute values ​​of all coefficients to obtain the contribution weight of the agglomerate particle size displacement to the blockage.

[0158] Critical clogging analysis is performed based on the second derivative of contribution weight and membrane specific flux to determine whether it is conformation-dominated clogging.

[0159] It should be noted that, firstly, the contribution weight of conformational changes (such as aggregate deformation and molecular chain conformational fluctuations) to membrane fouling is quantified through an attribution model. At the same time, the second derivative of membrane specific flux (a flux index that eliminates the influence of pressure) is calculated to characterize the accelerated characteristics of flux decay. The second derivative reflects the trend of flux change rate. When the conformational contribution weight exceeds the critical threshold, and the change pattern of the second derivative of membrane specific flux (such as sign change or amplitude exceeding the critical value) matches the nonlinear fouling dynamics caused by conformational dynamic imbalance (such as accelerated fouling caused by membrane pore contraction due to conformational disorder or aggregate blockage), it is determined to be conformation-dominated fouling.

[0160] S304. If the clogging is conformation-dominant, perform a stability analysis on the target retention of the nanofiltration membrane to determine whether the conformation-dominant retention is unstable.

[0161] If the blockage is conformation-dominant, calculate the coefficient of variation of the retention rate within the characteristic Dalton interval;

[0162] Among them, the coefficient of variation reflects the degree of dispersion of the retention rate fluctuation;

[0163] Based on the random regression forest algorithm, an extension-trap fluctuation correlation model is constructed to calculate the contribution weight of molecular chain extension offset to trap fluctuation.

[0164] Based on the coefficient of variation and contribution weight, comparative analysis is performed using a baseline library to determine whether it is a conformation-dominant type of retention instability.

[0165] Understandably, when constructing a stretch-cutoff fluctuation correlation model based on the random regression forest algorithm, real-time monitoring data of molecular chain stretch (such as stretch values ​​at different times) is used as input features, and the fluctuation amplitude of the cutoff rate in the corresponding cutoff fluctuation data is used as the output target. The nonlinear correlation between the two is fitted by training the model. After the model is trained, the weight of molecular chain stretch offset in all factors affecting cutoff fluctuation is calculated using a feature importance evaluation mechanism (such as the amount of impurity reduction when nodes split), thus obtaining its contribution weight to cutoff fluctuation.

[0166] When determining conformation-dominant retention instability based on the coefficient of variation and contribution weight, first calculate the coefficient of variation of the real-time retention fluctuation and combine it with the contribution weight of the molecular chain extension shift to the retention fluctuation obtained above; then compare the two with the retention fluctuation coefficient of variation threshold and conformation factor contribution weight threshold in the normal baseline library during stable operation. If the real-time coefficient of variation exceeds the baseline threshold and the contribution weight of the molecular chain extension shift also exceeds the critical weight of conformation dominance in the baseline, then it is determined to be conformation-dominant retention instability.

[0167] S305. Based on the results of stability analysis, the imbalance interval is divided by combining trajectory similarity and cooperative deviation index.

[0168] For example, the criteria for determining the warning interval are:

[0169] Cooperative deviation index C Trajectory similarity S When the co-deviation index and trajectory similarity are in the corresponding warning range, conformational shift begins to affect membrane performance, but has not yet formed a significant deterioration trend;

[0170] The unidirectional imbalance interval can be set as: the co-deviation index C Trajectory similarity S Furthermore, in nonlinear response, distribution kurtosis, and adaptation entropy deviation, a single dimension (such as conformation adaptation entropy deviation) is much higher than other dimensions (such as the deviation of this dimension accounting for more than 60% of the total deviation), reflecting a unidirectional imbalance dominated by conformation or membrane performance, and local deterioration of the system.

[0171] The global imbalance interval is defined as: the co-deviation index is higher than... The trajectory similarity is higher than 0.6, and the nonlinear response, distribution kurtosis, and adaptation entropy deviation simultaneously exceed their respective baseline thresholds. The coupling relationship between conformation and membrane performance is completely broken, and multiple feature trajectories simultaneously deviate from the steady state, and the system performance is on the verge of getting out of control.

[0172] Different imbalance intervals are numbered to obtain imbalance interval identifiers;

[0173] Understandably, the purpose of dividing the imbalance intervals is:

[0174] Function 1: Provides a hierarchical basis for tracing the root causes of conformation-membrane performance imbalances. Different imbalance intervals correspond to specific feature deviation patterns. For example, a warning interval may originate from a slight fluctuation in a single parameter, while a global imbalance involves the coordinated collapse of multiple features. By using interval labels, the dominant factor of the imbalance can be quickly identified (whether it is conformational dynamic anomaly, membrane performance degradation, or the coupling failure of the two), providing directional guidance for subsequent accurate diagnosis and reducing the blindness of traditional diagnosis.

[0175] Second, it supports the dynamic iteration and effectiveness evaluation of regulatory strategies. The division of imbalance intervals constructs a closed-loop relationship between state, intervention, and effect: by recording the response effects of regulatory measures in different intervals (such as whether fine-tuning of the warning interval prevents the entry into unidirectional imbalance), the suitability of the strategy can be quantitatively evaluated, thereby iteratively optimizing and adjusting the critical threshold and optimizing the timing of intervention, forming a continuously improving regulatory system.

[0176] S4. If the nanofiltration system enters the imbalance range, fit the quantitative relationship between the lignin conformational transformation rate and the decay of the retention efficiency, construct a retention response model under conformational disturbance based on the quantitative relationship, construct the conformational regulation boundary through the retention response model, and start conformational regulation if the real-time conformational parameters touch the regulation boundary.

[0177] The method for fitting the quantitative relationship between the lignin conformational transition rate and the decline in retention efficiency is as follows:

[0178] The conformational transition rate is obtained by summing the aggregate particle size change rate, the slope of the average extension, and the conformational fluctuation rate. ;

[0179] The decay rates of the rejection rate and the specific flux within the characteristic Dalton interval are obtained. These decay rates are then summed to obtain the decay coefficient of the rejection efficiency. ;

[0180] Using the nonlinear least squares method: Fit the functional relationship between retention efficiency and conformational transition rate;

[0181] in, , , The parameters are fitted using the nonlinear least squares method. Basic contribution coefficient, To characterize the exponential effect of the conformational transition rate on the retention rate. The offset correction term is determined by minimizing the sum of squared residuals between the predicted rejection rate and the actual value;

[0182] The method for constructing a interception response model under conformational disturbance based on quantitative relationships is as follows:

[0183] A cut-off response model is constructed using a long short-term memory network algorithm, and dynamic trend indicators, distribution offset indicators, and imbalance interval identifiers are input into the cut-off response model.

[0184] Those skilled in the art will understand that, when constructing a cutoff response model, the dynamic trend indicators, distribution offset indicators, and imbalance interval identifiers are first normalized and preprocessed to unify the data scale. Then, an LSTM network structure is designed, with the input layer dimension matching the total number of the three types of indicators. Temporal dependencies are captured through 1-2 layers of LSTM hidden layers, and the hidden layer output is mapped to the output layer via a fully connected layer. The output target is the cutoff rate or cutoff fluctuation value at future time. Next, the model is trained using historical operating data (input features and corresponding actual cutoff responses), with mean squared error as the loss function. The network weights are iteratively adjusted using the Adam optimizer, and a dropout layer is introduced to suppress overfitting. This enables the model to learn the dynamic correlation between input indicators and cutoff responses, achieving temporal prediction of cutoff behavior.

[0185] It should be noted that the retention response model outputs predicted values ​​of retention efficiency and membrane specific flux for future monitoring periods;

[0186] The method for constructing the conformational regulation boundary by intercepting the response model is as follows:

[0187] The average particle size D and molecular chain extension S of the aggregates are combined to construct the core conformation set;

[0188] The core conformation set is input into the retention response model, and the retention response model outputs the predicted values ​​of membrane specific flux and retention rate for future monitoring periods.

[0189] The regulatory boundary conditions are constructed based on conformational adaptation entropy, membrane specific flux prediction, retention rate prediction, and co-deviation index.

[0190] A two-dimensional control boundary was constructed using the average particle size and molecular chain extension of the aggregates as the coordinate axis;

[0191] The core conformation set that satisfies the regulatory boundary is marked as the boundary point outside the regulatory boundary;

[0192] Through polynomial functions: Fit the points outside the boundary to construct a continuous control boundary curve;

[0193] in, , , The quadratic polynomial fitting coefficients are obtained by regression analysis of the boundary out-of-bounds point data.

[0194] If the conformational parameters reach the control boundary, the conformational regulation is initiated in the following manner:

[0195] Obtain the real-time core conformation set and input the continuous boundary curve to calculate the real-time boundary;

[0196] Among them, the real-time core conformation set consists of real-time conformation parameters;

[0197] like Figure 2 As shown, if the real-time boundary exceeds the control boundary curve and the membrane specific flux prediction value of the cut-off response model is lower than the preset critical value, it is determined that the conformation parameter has reached the control boundary; otherwise, the real-time boundary is continuously monitored.

[0198] If the conformational parameters reach the control boundary, a corresponding conformational control strategy is established based on the type of imbalance interval.

[0199] For example, the corresponding conformational regulation strategy can be established as follows:

[0200] Total imbalance: Emergency addition of dispersant (0.3 g / L) and reduction of operating pressure by 10%, aiming to reduce D by 20% within 10 minutes;

[0201] Unidirectional imbalance (blockage-dominated): Conventional addition of dispersant (0.1-0.2 g / L) results in D returning to within the boundary within 30 minutes;

[0202] Unidirectional imbalance (dominated by interception instability): Adjust the influent pH to 4.0-4.5, with a target S decrease of 15% within 30 minutes;

[0203] Warning range: Preventive addition of 0.05 g / L dispersant to inhibit further conformational deterioration.

[0204] Example 3

[0205] like Figure 3 As shown, the present invention is a pollutant removal system for papermaking wastewater based on nanofiltration staged separation, and further includes the following modules:

[0206] Conformation acquisition module: used to acquire real-time conformational parameters of lignin in papermaking wastewater in nanofiltration system, perform conformational dynamic analysis on real-time conformational parameters to obtain conformational fluctuation rate, and determine whether to trigger conformational imbalance analysis between lignin and nanofiltration membrane.

[0207] Imbalance Analysis Module: If conformational imbalance analysis is required, a conformational characteristic data set is established based on real-time conformational parameters. Conformational imbalance analysis is performed on the conformational characteristic data set to determine whether lignin conformational dynamic trend tracking is triggered.

[0208] Interval Determination Module: If lignin conformation dynamic trend tracking is triggered, the conformation dynamic trend and distribution shift characteristics are extracted based on the conformation characteristic data set, and conformation shift analysis is performed to obtain the imbalance interval of the nanofiltration system.

[0209] Conformation optimization module: If the nanofiltration system enters the imbalance range, the quantitative relationship between the lignin conformational transition rate and the decay of retention efficiency is fitted. Based on the quantitative relationship, a retention response model under conformational disturbance is constructed. The conformational regulation boundary is constructed through the retention response model. If the real-time conformational parameters touch the regulation boundary, conformational regulation is initiated.

[0210] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the present invention should still fall within the scope of the present invention.

Claims

1. A method for removing pollutants from papermaking wastewater based on nanofiltration staged separation, characterized in that: Includes the following steps: Real-time conformational parameters of lignin in papermaking wastewater in nanofiltration system are collected. Conformational dynamic analysis is performed on the real-time conformational parameters to obtain the conformational fluctuation rate, and it is determined whether it is necessary to trigger conformational imbalance analysis between lignin and nanofiltration membrane. The method for performing the aforementioned conformational dynamic analysis is as follows: Obtain the average particle size and particle size distribution width of aggregates from real-time conformational parameters; The rate of change of the average particle size of the aggregates during the monitoring period was calculated to obtain the particle size change rate. Calculate the rate of change of the average stretching of the molecular chain during the monitoring period to obtain the stretching change rate. The conformational fluctuation rate is obtained by weighted summation of the particle size change rate and the stretching change rate. Based on the comparative analysis of conformational volatility, the determination result of whether conformational imbalance analysis is triggered is obtained; The average extension of the molecular chain was obtained by collecting the absorbance A of the lignin characteristic peak at 280 nm using an online ultraviolet-visible spectrometer. 280 A through lignin standards with known extension 280 The curve was used to establish a quantitative relationship between absorbance and molecular chain extension, and the average extension of the molecular chain was obtained. The rate of change of the average stretching of the molecular chain at the initial and final values ​​within the monitoring period is simultaneously acquired and calculated to obtain the stretching change rate. If conformational imbalance analysis is required, a conformational characteristic data set is established based on real-time conformational parameters, and conformational imbalance analysis is performed on the conformational characteristic data set to determine whether lignin conformational dynamic trend tracking is triggered. The conformational imbalance analysis is performed as follows: Obtain real-time characteristic parameters of the nanofiltration membrane and combine them with real-time conformational parameters to establish a conformational characteristic data set; Lignin retention rate was extracted from the conformational characteristic data set, and nonlinear amplification-triggered analysis was performed in combination with conformational change rate to obtain nonlinear response ratio; A joint distribution model of conformation-nanofiltration membrane parameters was constructed, and the rate of change of adaptation entropy was obtained by performing ordered imbalance analysis. Extract the kurtosis of the adaptation entropy change rate, and construct a three-dimensional joint criterion based on the nonlinear response ratio, kurtosis, and adaptation entropy change rate to determine whether dynamic trend tracking analysis of lignin conformation is triggered. If lignin conformation dynamic trend tracking is triggered, the conformation dynamic trend and distribution shift characteristics are extracted based on the conformation characteristic data set, and conformation shift analysis is performed to obtain the imbalance interval of the nanofiltration system. If the nanofiltration system enters the imbalance region, the quantitative relationship between the lignin conformational transition rate and the decay of retention efficiency is fitted. Based on the quantitative relationship, a retention response model under conformational disturbance is constructed. The conformational regulation boundary is constructed through the retention response model. If the real-time conformational parameters reach the regulation boundary, conformational regulation is initiated.

2. The method for removing pollutants from papermaking wastewater based on nanofiltration staged separation according to claim 1, characterized in that: The method for performing the ordered imbalance analysis is as follows: Based on the average particle size of aggregates, membrane flux and rejection rate in the conformational property data set as joint features, the three-dimensional distribution of conformation was fitted to obtain the conformational characteristics. Based on the three-dimensional distribution of conformation, the conformational adaptation entropy is obtained through the entropy equation. The rate of change of conformational adaptation entropy for the current and adjacent monitoring periods is obtained and calculated to obtain the rate of change of adaptation entropy.

3. The method for removing pollutants from papermaking wastewater based on nanofiltration staged separation according to claim 1, characterized in that: The method for constructing the three-dimensional joint criterion is as follows: Nonlinear response ratio, distribution kurtosis and rate of change of fit entropy are used as three-dimensional joint features; A baseline library of historical joint criteria for constructing nanofiltration systems; Calculate the 3D deviation between the joint 3D features and the normal baseline library; The collaborative deviation index is obtained by performing collaborative deviation analysis based on three-dimensional deviation. Trajectory similarity is determined based on the co-deviation index, and if the trajectories are similar, the dynamic trend of lignin conformation is tracked.

4. The method for removing pollutants from papermaking wastewater based on nanofiltration staged separation according to claim 1, characterized in that: The method for performing conformational shift analysis is as follows: Extracting dynamic trend indicators for the quantification of lignin dynamic characteristics; Extract the distribution migration index of conformational migration; By combining dynamic trend indicators and distribution shift indicators, we can analyze the relationship between conformational shift and nanofiltration membrane performance degradation. If the clogging is conformation-dominant, a stability analysis is performed on the target retention of the nanofiltration membrane to determine whether conformation-dominant retention is unstable. Based on the results of stability analysis, imbalance intervals are divided by combining trajectory similarity and cooperative deviation index.

5. The method for removing pollutants from papermaking wastewater based on nanofiltration staged separation according to claim 4, characterized in that: The method for analyzing the type of performance degradation of the nanofiltration membrane is as follows: Obtain and calculate the second derivative of the membrane specific flux in the characteristic parameters of the nanofiltration membrane; Using partial least squares regression algorithm, the rate of change of the average particle size of aggregates in the dynamic trend index during the continuous monitoring period and the rate of change of the particle size distribution width are used as independent variables, and the second derivative of the membrane specific flux is used as the dependent variable. Obtain the absolute value of the regression coefficients output by the partial least squares regression algorithm, and the sum of the absolute values ​​of all coefficients; The absolute value of the regression coefficient is compared with the sum of the absolute values ​​of all coefficients to obtain the contribution weight of the agglomerate particle size shift to the blockage. Critical clogging analysis is performed based on the second derivative of contribution weight and membrane specific flux to determine whether it is conformation-dominated clogging.

6. The method for removing pollutants from papermaking wastewater based on nanofiltration staged separation according to claim 4, characterized in that: The method for determining whether a structure is a conformation-dominant type of intercepted instability is as follows: If the blockage is conformation-dominant, calculate the coefficient of variation of the retention rate within the characteristic Dalton interval; Based on the random regression forest algorithm, an extension-trap fluctuation correlation model is constructed to calculate the contribution weight of molecular chain extension offset to trap fluctuation. Based on the coefficient of variation and contribution weight, a comparative analysis using a baseline library is conducted to determine whether it is a conformation-dominant type of retention instability.

7. The method for removing pollutants from papermaking wastewater based on nanofiltration staged separation according to claim 1, characterized in that: The conformational regulation boundary is constructed as follows: Extract the average particle size and molecular chain extension of the aggregates from the conformational property data set and combine them to construct the core conformation set; A retention response model is constructed. The core conformation set is input into the retention response model, and the retention response model outputs the predicted values ​​of membrane specific flux and retention rate for future monitoring cycles. The regulatory boundary conditions are constructed based on conformational adaptation entropy, membrane specific flux prediction, retention rate prediction, and co-deviation index. A two-dimensional control boundary was constructed using the average particle size and molecular chain extension of the aggregates as the coordinate axis; The core conformation set that satisfies the regulatory boundary is marked as the boundary point outside the regulatory boundary; A continuous control boundary curve is constructed by fitting the boundary points with a polynomial function.

8. The method for removing pollutants from papermaking wastewater based on nanofiltration staged separation according to claim 7, characterized in that: The interception response model is constructed as follows: A cut-off response model is constructed using a long short-term memory network algorithm, and dynamic trend indicators, distribution offset indicators, and imbalance interval identifiers are input into the cut-off response model.

Images