A paperboard production wastewater zero discharge recycling process

Abstract

The application relates to the technical field of wastewater treatment, and discloses a paperboard production wastewater zero-discharge cyclic treatment process, which comprises the following steps: judging whether parameters in the paperboard production wastewater zero-discharge cyclic treatment process are abnormal, performing concentration multiple initial screening when the parameters are abnormal, and performing low-multiple analysis or high-multiple analysis according to the initial screening judgment conclusion after verification. The paperboard production wastewater zero-discharge cyclic treatment process realizes layer-by-layer accurate positioning of three faults of leakage, scaling and biological pollution through sub-second online table sampling and multiple fusion judgment of pressure difference-temperature-liquid level, greatly reduces blind film disassembly and repeated cleaning, realizes pressure difference stability control, continuously reaches the standard of water production and significantly prolongs the service life of the membrane, improves the overall operation efficiency, synchronously reduces the chemical and downtime costs, realizes self-adaptive drift of the threshold with seasons through closed-loop log self-learning, and long-term maintains low waste rate and high water production reliability.

Description

Technical Field

[0001] This invention relates to the field of wastewater treatment technology, specifically a zero-discharge recycling treatment process for wastewater from paperboard production. Background Technology

[0002] The paperboard industry is a major consumer of water resources and a major emitter of wastewater. Its traditional model faces enormous pressure to reduce carbon emissions. The background of the zero-emission recycling process for wastewater from paperboard production profoundly reflects the strategic transformation that the industry is undergoing: "from passive compliance to proactive pollution control, from cost consumption to value creation, and from environmental risks to development opportunities." Despite the challenges that remain, driven by policy, market, and technology, zero emissions are transforming from a "luxury" to a "standard configuration" and core competitiveness for more and more visionary paperboard companies. The zero-emission recycling process for wastewater from paperboard production is a profound transformation from "end-of-pipe treatment" to "resource reuse." It uses membrane technology as the backbone to achieve water resource recovery and evaporation crystallization as the final step to solidify pollutants. Although it faces cost and technological challenges, its strategic value in terms of water resource security, environmental compliance, and sustainable operation is irreplaceable. With technological progress and the increasing demand for resource utilization, especially the development of "fractional crystallization" technology, ZLD is gradually evolving from a high-cost environmental burden into a valuable resource recycling system. Existing zero-discharge recycling processes for wastewater from cardboard production largely rely on experience-based inspections or single-parameter alarms. They cannot differentiate between leaks, biofouling, and scaling online, often misdiagnosing it as "membrane fouling" and blindly disassembling and cleaning it. They lack real-time identification of the root causes of leaks, such as inverted expansion ratios, internal leakage in back pressure valves, and malfunctions in cooling water valves, leading to repeated increases in differential pressure. They also lack gentle, targeted treatment methods for EPS release, biofilm expansion, and CaCO3 thin layers, commonly employing high-concentration alkaline washing or acid soaking, which damages the membrane and increases chemical consumption. Furthermore, the thresholds are fixed and cannot adapt to seasonal changes, resulting in long-term problems of false alarms, missed alarms, and over-cleaning, thus limiting their practicality. Summary of the Invention

[0003] This invention provides a zero-discharge recycling process for wastewater from paperboard production, which helps to solve the problems mentioned in the background art.

[0004] This invention provides the following technical solution: a zero-discharge recycling treatment process for paperboard production wastewater, comprising: Determine if parameters are abnormal; if the system enters "triggered state", perform initial screening of concentration factor; extract the verified initial screening result; if the verified initial screening result is CH1, perform low-magnification analysis; if the verified initial screening result is CH2, perform high-magnification analysis; after performing high-magnification analysis, recalculate the instantaneous pressure difference; if it stabilizes within ±10% of the pressure difference baseline, the case is closed; if it does not stabilize within ±10% of the pressure difference baseline, trigger secondary diagnosis. The initial screening of concentration factor is performed as follows: simultaneous high-frequency acquisition with dual conductivity and timestamp alignment buffering is performed; conductivity data is cleaned and outlier values ​​are removed; dynamic baseline concentration factor and robust standard deviation are calculated; real-time concentration factor is calculated; the corresponding diversion channel is determined through three threshold initial screening judgments to form an initial screening judgment conclusion; and the duration and stability of the initial screening judgment conclusion are verified.

[0005] As an optional solution to the zero-discharge recycling treatment process for paperboard production wastewater described in this invention, the following steps are performed: Low-magnification analysis is conducted, specifically: S1. The historical curve is segmented, and sliding window features are generated; S2. All window features are traversed, and a rigid screening is performed to determine whether an event of "falling from the normal concentration zone to the short-circuit zone" has occurred; S3. The time of the fall is extracted. Regarding the moment of the fall The following six windows are used for rebound suppression verification; S4, determine whether there are periodic dip events during the night shift of the historical curve; S5, extract marked suspected events and perform dual verification of day and night rebound amplitude and stability for suspected events; S6, perform confidence-weighted scoring and channel final decision for events that are "inverted" or "periodic dip"; S7, if the channel final decision is that the inverted sub-channel is established, execute the inverted sub-channel; if the channel final decision is that the periodic dip is established, execute the periodic dip sub-channel; S8, after executing step S7, re-determine whether the parameters are abnormal; if the parameters are determined to be abnormal, execute the biocolloid determination with the label "false low ratio ruled out"; if the parameters are determined to be normal or the system enters "observation state", no further determination is executed.

[0006] As an optional solution to the zero-discharge recycling treatment process for paperboard production wastewater described in this invention, the following steps are taken: An inverted sub-channel is implemented, specifically: High-frequency data is collected synchronously using three parameters: permeate, concentrate, and pump pressure, and a dynamic sliding queue is established; baseline conditions are dynamically anchored, and real-time deviation is calculated; an internal leakage determination is performed to determine if an internal leakage fault exists; if the internal leakage determination output is 1, a suspected internal leakage is identified; if the internal leakage determination output is 0, normal fluctuation is identified; a bursting determination is performed to determine if a bursting fault exists; if the bursting determination output is 1, a suspected bursting is identified; if the bursting determination output is 0, normal fluctuation is identified; a dual-mode confidence weighted evaluation is performed on the internal leakage determination conclusion and the bursting determination conclusion; if a suspected internal leakage is identified, internal leakage fault handling is performed; if a suspected bursting is identified, bursting fault handling is performed; if a composite fault is identified, composite fault handling is performed; after fault handling, effect tracking and ratio recovery closed-loop verification are performed.

[0007] As an optional solution for the zero-discharge recycling treatment process of paperboard production wastewater described in this invention, the following steps are taken: A periodic low-head sub-channel is executed, specifically: high-frequency sampling is performed through dual channels by reading the product water tank level and concentrate ratio, and the timestamps are aligned; the night shift window is automatically marked, and the baseline level is calculated; the level sequence is negatively mapped, and the Pearson correlation coefficient within the night shift window is calculated; a mirror judgment is performed to determine if a mirror image is valid; if the mirror judgment output is 0, the mirror image is deemed invalid, marked as "weak mirror image," and downgraded to the manual observation channel; if the mirror judgment output is 1, the mirror image is deemed valid, the back pressure valve feedback opening during the same night shift is read, and the fault type is determined through linkage verification; if the linkage verification output is 1, the fault type is determined to be program-forced pressure relief; if the linkage verification output is 0, the fault type is determined to be valve internal leakage or diaphragm rupture; based on the fault type determination conclusion obtained from the linkage verification, corresponding fault handling measures are executed; after fault handling, the treatment effect is tracked and the mirror coefficient rebound is verified.

[0008] As an optional solution to the zero-discharge recycling treatment process for paperboard production wastewater described in this invention, the following steps are performed: Biocolloid determination is implemented by: at predetermined intervals, recording the same timestamp from four real-time data points—residual chlorine meter, influent pH meter, water temperature meter, and differential pressure gauge—and writing them into a 7-day rolling queue; obtaining the target stable segment; calculating the residual chlorine baseline, pH baseline, temperature baseline, and differential pressure baseline; segmenting the event window and recording the characteristics of each event segment; performing initial checks on EPS release colloid channels; performing initial checks on biofilm expansion channels; performing analysis on pure colloid or short fiber deposition channels; implementing corresponding treatment plans based on the fault type; and tracking the treatment effect in a closed loop.

[0009] As an optional solution for the zero-discharge recycling treatment process of paperboard production wastewater described in this invention, the following steps are performed: high-magnification analysis is conducted, including pH drift value analysis. Specifically, data is collected through dual pH high-frequency sampling and drift window cutting; pH baseline is dynamically analyzed; drift amplitude is calculated and synchronous initial screening is performed; secondary confirmation of scaling sentinel signal is performed; online hydrochloric acid washing is started immediately; and recovery rate is increased stepwise and effect is tracked.

[0010] As an optional solution for the zero-discharge recycling treatment process of paperboard production wastewater described in this invention, the following steps are included: performing high-magnification analysis, which also includes concentrated water temperature curve analysis. Specifically, this involves: data acquisition through dual-channel synchronous high-frequency sampling; determining the dynamic temperature baseline; performing initial screening with rigid thresholds for sudden drops and rises; confirming the linkage between temperature difference persistence and water-side pressure; conducting isolation and sheet pressure tests; replacing damaged sheets and resetting the system; and performing effect tracking and secondary protection.

[0011] As an optional solution for the zero-discharge recycling treatment process of paperboard production wastewater described in this invention, the high-magnification analysis further includes: data acquisition through dual-channel high-frequency sampling and night shift window labeling; calculation of cooling water valve opening baseline and temperature difference baseline; execution of automatic night shift opening and film temperature drop linkage for initial screening; execution of synchronous confirmation of rapid increase in ratio-pressure difference; execution of control strategy decoupling and immediate intervention; execution of effect tracking and ratio-pressure difference rebound verification.

[0012] The present invention has the following beneficial effects: 1. This zero-discharge recycling process for wastewater from cardboard production uses differential pressure rise as the inlet, automatically compares the concentrate ratio, and instantly divides the operating conditions into three channels: "low ratio high difference," "normal ratio high difference," and "grey zone." It then thoroughly investigates the root causes of leaks, such as inverted ratios, night shift oscillations, or malfunctioning cooling valves. It first repairs valves, replaces gaskets, and adjusts liquid levels to eliminate "false contamination." Then, it enters the biological or scale assessment with a clean label, avoiding blind membrane washing. This process enables early detection of leaks, reduces misjudgments, and minimizes the number of membrane washings and downtime.

[0013] 2. The zero-discharge recycling process for wastewater from paperboard production, after confirming no leakage, uses signals such as residual chlorine impact, overnight jump, and pH drift to distinguish between EPS release, biofilm expansion, and pure colloidal deposition. It employs gentle methods such as neutral enzyme washing, low-dose continuous chlorination, and surface activity enhancement to disperse and remove pollutants at the thin-layer stage without damaging the polyamide membrane or wasting chemicals. This makes cleaning more precise, extends membrane life, reduces chemical usage, and lowers total emissions.

[0014] 3. This zero-discharge recycling process for wastewater from paperboard production addresses scaling caused by actual concentration. The system detects CaCO3 thin layers, heat exchanger leaks, or false concentration surges in advance based on signals such as pH rise, sudden temperature increases, or blindly opening cooling water valves. It then performs online acid washing, isolates the heat exchanger, and locks the upper limit of the recovery rate, dissolving the scale layer on-site at the micron level. It quickly isolates the leak point, decouples the recovery rate from temperature, and rapidly reduces the membrane stack pressure difference to a safe range. This eliminates the need for membrane disassembly and line shutdown, allowing for early removal of true scale, stable pressure difference operation, and ensuring production line continuity and membrane element integrity. Attached Figure Description

[0015] Figure 1 This is a flow chart of the zero-discharge recycling treatment process for wastewater from paperboard production according to the present invention. Detailed Implementation

[0016] Example 1: A zero-discharge recycling treatment process for wastewater from paperboard production (see reference). Figure 1The process includes: determining whether parameters are abnormal; if the system enters a "triggered state," performing a preliminary screening based on the concentration factor; extracting the verified preliminary screening conclusion; if the verified preliminary screening conclusion is CH1, performing low-magnification analysis; if the verified preliminary screening conclusion is CH2, performing high-magnification analysis; after performing high-magnification analysis, recalculating the instantaneous pressure difference; if it stabilizes within ±10% of the pressure difference baseline, the case is closed; if it does not stabilize within ±10% of the pressure difference baseline, a secondary diagnosis is triggered, i.e., manual diagnosis and treatment. The initial screening for concentration factor is performed as follows: Simultaneous high-frequency acquisition with dual conductivity and timestamp alignment buffering is used. Instantaneous values ​​are read every 2 minutes from both the concentrate conductivity online meter and the influent conductivity online meter, and recorded as follows: and Since there may be a millisecond-level time difference between the two sampling meters, the system automatically uses the water inlet sampling time as a reference to pair the closest concentration conductivity values ​​within the last 3 seconds to generate a timestamp-aligned original rate sequence. Cache the ratio data for the most recent 7 days for subsequent dynamic baseline calculation: ; ;in, This is the conductivity value of the concentrated water. Due to the time difference in instrument sampling, it is permissible to... Take the closest value within 3 seconds before and after. The influent conductivity value is used as a reference. At the current sampling time, These are time deviation tolerance values ​​used for timestamp alignment between the two tables. These parameters are used to calculate the original concentration factor, reflecting the true concentration level of the membrane system. A 7-day rolling cache queue (5040 points = 7 days × 24 hours × 30 times / hour) is used to store recent historical odds for subsequent median, standard deviation calculation and trend analysis. The conductivity data is cleaned, and outlier values ​​are removed. The validity of each newly added pair of conductivity values ​​is assessed; if the influent conductivity is lower than the clean water threshold, then... Or the conductivity of the concentrated water exceeds the instrument's range, i.e. If the error is not found, it is determined to be an instrument calibration or disconnection anomaly. The magnification at that point is marked as invalid and replaced with the moving average of the previous 3 points to generate a cleaned magnification sequence. To avoid misjudging the channel due to magnification distortion caused by a single instrument failure: ;in, It is an arithmetic mean function used to smooth interpolation for missing or outlier points, preventing magnification distortion caused by a single instrument failure; To calculate the dynamic baseline fold and robust standard deviation, the 50th percentile of the most recent 72 hours (2160 points) of data from the cleaned sequence was used as the dynamic baseline fold for the current system. Simultaneously, the robust standard deviation of this window is calculated using the median absolute deviation (MAD) method. To avoid interference from extreme values: ; ;in, It is a median function. This is the cleaned rate sequence, i.e., after removing outliers. This is a constant used to convert the absolute median deviation to an approximate standard deviation. That is, the absolute median; if This indicates that the system is in an unstable period due to frequent load increases / decreases or frequent cleaning. The baseline calculation window is automatically extended to 120 hours. Repeat the calculation until... Only then is the baseline established, which reflects the "normal concentration level" at the current operating recovery rate; Calculate the real-time multiplier, while locking Then, calculate the current instantaneous multiplier in real time. And calculate its deviation from the baseline. : ; ;in, This is the instantaneous value of the cleaning magnification at the current moment, used to compare with the baseline and calculate the deviation. The initial screening process uses three threshold thresholds to determine the appropriate execution path and arrives at an initial screening conclusion. ;in, This indicates a suspicious passage with a "low elevation difference". This indicates a suspicious passage with a "normal height difference". This indicates that the area is considered a "gray zone" and requires parallel monitoring. The initial screening results are validated for duration and stability. To avoid misjudgments caused by instantaneous threshold crossing, each channel's judgment must last for 2 hours (60 consecutive points) before final confirmation. The system records the moment when a channel's condition is first met. If the same channel mark is maintained for 60 consecutive points and no cleaning operation occurs during this period (locked by the cleaning pump status signal), the channel is confirmed to be valid. If a channel jump or cleaning is started within 60 points, the counter is reset and the real-time magnification and the three-segment threshold initial screening judgment are recalculated to filter out "false low magnification" or "false high magnification" caused by load adjustment or instrument vibration.

[0017] This embodiment also provides a method for determining whether parameters are abnormal, specifically by: collecting target parameters, whereby the target parameters are instantaneous values ​​read every 5 minutes from the RO membrane inlet pressure gauge and the concentrate outlet pressure gauge, i.e., the inlet water pressure. and concentrated water pressure ; Calculate the instantaneous pressure difference: ; Implement sliding window caching, storing the most recent 30 days. Data is stored in a rolling queue, automatically discarding outdated data older than 30 days, ensuring the memory queue length remains constant at 8640 points. ; ;in, The sampling time is 300 seconds. The original pressure difference is the difference between the inlet pressure and the concentrate outlet pressure. For membrane inlet pressure, These parameters, representing the concentrate-side pressure, are used to reflect changes in transmembrane resistance caused by membrane fouling or scaling. A 30-day rolling cache queue (8640 points = 30 days × 24 hours × 12 times / hour) is used to store differential pressure history for trend slope calculation and baseline anchoring. Real-time data cleaning and abnormal pulse removal are performed on the cached data, and each new data entering the queue is processed. Outlier detection: If the difference between this value and the mean of the previous 3 points exceeds ±0.05MPa, it is determined to be an abnormal pulse caused by instrument vibration or pump start-up / stop impact, and is removed and replaced with the smoothed mean of the previous 3 points to generate the cleaned sequence. To avoid triggering false alarms from a single misreading: ;in, The moving average of the pressure difference at the first three points is used to smooth out abnormal pulses; if ,but ;like ,but ; Calculate the current dynamic baseline and determine if the system is functioning correctly. Take the data from the most recent 7 days in the cleaned sequence and calculate its 50th quantile, i.e., the median, as the current dynamic baseline. At the same time, calculate the standard deviation of these 7 days. ,like This indicates that the system is in an unstable period. The baseline calculation window should be extended to 14 days, until... Only then is the baseline locked, which represents the system's current "normal state": ; ;like The window then expands to arrive Recalculate to ;in, This indicates the starting time 1008 points (i.e., 7 days) prior, used to select the most recent 7-day stable period as the baseline window. This represents the standard deviation of the differential pressure baseline. This is a standard deviation function used to measure baseline stability. This indicates significant system fluctuations, requiring the window to be extended to 14 days. This indicates the starting point of pushing back 14 days, used to extend to 14 days to obtain a more robust trend when the 7-day baseline is unstable; The trend slope is continuously calculated, and the system state is determined. Starting from the time the baseline is locked, a 14-day sliding window is initiated at midnight every day to calculate the value within that window. Least squares fit slope over time If the slope on a certain day Calculate the current window endpoint value Compared with baseline relative rate of change : ; ;in, It is a linear regression slope function used to calculate the differential pressure rise rate and determine whether the 30% threshold is triggered; If the relative rate of change is ≥30%, that is If the parameter is abnormal, the system enters a "triggered state"; if the relative rate of change is between 25% and 30%, i.e. If the relative rate of change is below the early warning threshold, the system enters "observation mode" and a yellow light is displayed; if the relative rate of change is <25%, i.e. If the parameter is normal, the system will continue monitoring. Verify the duration and filter out false triggers. To prevent "false 30%" caused by a single cleaning cycle or temporary fluctuations in operating conditions, verification is required. Whether the state lasts for more than 24 hours, the system records the moment when it first exceeds 30%. If within the next 288 consecutive points (24 hours × 12 times / hour) If the spike rate is consistently ≥30% and no cleaning operation has been performed during the period, it is considered a valid trigger; otherwise, it is considered a false spike and the calculation is reverted to the baseline. Upon trigger confirmation, the system automatically retrieves relevant parameters from the 14 days prior to the trigger, packages them to generate a time-series comparison chart, and writes them to the trigger log. These relevant parameters include conductivity ratio, pH, residual chlorine, temperature, and flow rate. The trigger log includes the trigger time, , , Actual value Slope The number of outliers in associated parameters, etc.

[0018] Example 2 is an improvement on Example 1. This paperboard production wastewater zero-discharge recycling treatment process performs low-magnification analysis, specifically: S1, segmenting the historical curve and generating a sliding window feature, showing the most recent 7 days... The sequence was segmented into 42 data segments, each segment consisting of a 4-hour sub-window. Five core statistical features were calculated for each segment: median. Minimum value Standard deviation , drop range and the rate of fall This transforms time-series curves into quantifiable feature matrices, providing a computational foundation for subsequent pattern recognition. ; ; ; ;in, This represents the median multiplier for the current 4-hour window. This is the median of the leverage ratio for the next 4-hour window, used to quantify the instantaneous drop in leverage, determining whether it is ≥0.25, i.e., a drop from ≥1.35 to ≤1.10. For the first A 4-hour data segment, The multiple of the starting time of this segment. This is a multiple of the end time of the segment, used to divide the continuous time series into discrete windows, facilitating statistical feature extraction. For the first Window feature vectors The median. To be the minimum value, Standard deviation The magnitude of the drop. S2 is the drop rate, used to condense the data within the window into 5 statistics for pattern recognition; S3, iterate through all window features and rigidly screen for whether the "drop from the normal condensation area to the short-circuit area" event has occurred: if two consecutive windows satisfy the drop rate... Then for this window Record the moment the fall occurred. And mark it as a suspected inverted event. If no two consecutive windows meet the condition, it will be considered an inverted event. Then, step S4 is executed to perform periodic detection; where, Index of the current window, For the index of the next adjacent window, For the first The starting time of the window, The moment when the multiplier drop event occurs, i.e., the [number]th [event / time]. The window start point is used to record the time when the inversion occurred for subsequent tracking; S3, extract the time of the fall. Regarding the moment of the fall The subsequent six 24-hour windows were used to verify the rebound suppression, and the recovery suppression was calculated for these six windows. A sequence, if its median is... Always ≤1.12, and the maximum value If the value is ≤1.15, the inversion is confirmed. If it occurs within any window within 24 hours... If the value rises to >1.20, it is determined to be a "temporary fluctuation" rather than a structural short circuit. The event is marked as a pseudo-inversion, and step S1 is re-executed to eliminate instantaneous rate fluctuations caused by pump switching. ; ;in, It is a function with maximum value. That is, the maximum value This is used to check whether the leverage ratio has rebounded within 24 hours (6 windows) after the inversion. If the maximum value is still ≤1.15, it is confirmed that "it has not rebounded further"; if... If the inversion is true, then S6 is executed; if If the condition is not met, it is determined to be a false inversion, and S1 is executed again; S4, determine whether there are periodic head-down events during the night shift period of the historical curve, and binarize the historical curve according to the "night shift period" (22:00-06:00) and "day shift period" (06:00-22:00) to generate a binary sequence. The 24-hour periodic autocorrelation of the sequence was calculated to obtain the autocorrelation coefficient. : ; ; ;in, The Pearson correlation coefficient function is used to calculate binary sequences. (Night shift marker) and its 24-hour lag sequence periodicity; if >0.75, it indicates that the night shift head-down phenomenon repeats daily; if If a periodic head-down event is suspected, it is marked as a suspected event; if If the suspected events do not appear to be periodic head-down events, then S5, extract the marked suspected events, perform dual verification of the day-night rebound amplitude and stability of the suspected events, and extract three consecutive days of night-day shift data for verification, and calculate the median of each night shift. Compared with the median of the day shift Requirements must be met for all 3 days. ≤1.10 and ≥1.20, at the same time ≥0.15; where, ; ;in, Indicates the first Calculate the day respectively. The median magnification of the day and night shifts is used to determine the diurnal rebound amplitude. If all day and night pairs are satisfied, the periodic dip is determined to be valid, i.e., a periodic dip event exists, and step S6 is executed. If any day and night pair is not satisfied, it is determined to be "occasional low magnification during night shift" rather than structural leakage, and the event is marked as pseudo-period, and biocolloid determination is executed. S6: Confidence-weighted scoring and channel final decision are performed on events where "inversion is valid" or "periodic dip is valid". A confidence score of 0-100 is assigned to events where "inversion is valid" or "periodic dip is valid" to reduce misjudgments caused by instrument drift or single cleaning. ; ;in, Confidence score (0-100). It is a minimum value function. The duration of the event in hours. The post-event rate standard deviation is used to measure stability. The magnitude of the drop. The number of similar historical events is used to weight the credibility of an event; only those scoring above 85 are confirmed to avoid misjudgment. Then confirm the event and assign it the highest priority; if If so, it will be marked as "highly suspicious" and sent for manual secondary confirmation; if If the signal is not clear, it is considered low-confidence noise, and the process returns to step S1 for rescanning. S7: If the final channel decision is that the inverted position is valid, the inverted sub-channel is executed to locate physical leaks such as seal ring or filter bursts, avoiding misjudgment as membrane fouling. If the final channel decision is that the periodic downward movement is valid, the periodic downward movement sub-channel is executed to identify back pressure valve or PLC program problems, and eliminate "false low magnification" after repair. S8: After executing step S7, the parameters are re-evaluated for abnormality. If the parameters are determined to be abnormal, the biocolloid determination is executed with the "false low magnification eliminated" label to distinguish between EPS, biofilm, and pure colloid, using gentle cleaning to reduce membrane damage. If the parameters are determined to be normal or the system enters "observation state," subsequent determinations are not executed.

[0019] Specifically, the inverted sub-channel operation involves synchronous high-frequency data acquisition of three parameters: permeate flow meter, concentrate flow meter, and pump pressure. A dynamic sliding queue is established, and instantaneous values ​​are collected every minute from the permeate flow meter, concentrate flow meter, and high-pressure pump outlet pressure gauge, respectively, and recorded as follows: (Water production flow rate) (Concentrate flow rate) (High-pressure pump outlet pressure), establish three independent rolling queues to cache the most recent 24 hours of data (1440 points each), and automatically align the timestamps. If any point has a null value due to instrument communication interruption, it is filled with linear interpolation of the previous 5 points to ensure that the length of the three parameter sequences is strictly synchronized, providing a time-series consistency basis for subsequent linkage calculations: ; ; ;in, For the water production flow rolling queue, For the concentrate flow rolling queue, A pump pressure rolling queue is used to cache historical data and calculate baseline and deviation. The dynamic anchoring baseline condition is used, and the real-time deviation is calculated. The stable operating segment of the most recent 8 hours (480 points) in the three-parameter queue is taken, and its 50th quantile is calculated as the current dynamic baseline: , , Then, the percentage deviation of the current instantaneous value from the baseline is calculated. The baseline is refreshed every 4 hours to ensure that it adapts to minor adjustments in workshop load and avoids erroneous deviations caused by a fixed benchmark. ; ; ; ; ; ;in, This indicates the stable operating period of the last 8 hours in the water production parameter queue. The historical steady-state standard deviation, i.e., the standard deviation of the past 30 days of steady-state period, is used as a baseline stability criterion. <0.5× Only then can it be considered stable; Determine if an internal leakage fault exists by using internal leakage detection: ;in, This indicates a significant increase in permeate flow rate, meaning that a short circuit in the membrane column leads to a decrease in resistance on the permeate side. This indicates that the number of data points meeting this condition exceeds 25. This indicates that the concentrate flow rate is decreasing simultaneously, meaning the concentrate is being bypassed to the product water or pump inlet. This indicates that the pump outlet pressure is decreasing synchronously, meaning the total system resistance is decreasing and the pump load is reduced. If all three conditions are met—a significant increase in product water flow rate, a synchronous decrease in concentrate flow rate, and a synchronous decrease in pump outlet pressure—and the number of data points meeting each condition exceeds 25, then occasional instrument disturbances can be ruled out, and a suspected internal leak can be identified. If the internal leakage determination output is 1, that is... If so, it is determined to be a suspected internal leak; If the internal leakage determination output is 0, that is... If so, it is determined to be a normal fluctuation; Determine whether a bursting fault exists by detecting bursting: ;in, This indicates that the permeate flow rate remained essentially unchanged, meaning the rupture did not directly affect the membrane-side flow distribution. This indicates a sudden drop in pump outlet pressure, meaning that after the raw water bypass is short-circuited, the pump outlet resistance decreases sharply. This is an alarm signal for "sudden pressure drop at the top of the filter," triggered by on-site video monitoring or a pressure switch signal. If the alarm signal is triggered, the output is True; otherwise, the output is False. If the filter pressure gauge reading is based on the condition that the water flow rate remains basically unchanged and the pump outlet pressure drops suddenly, and the number of data points that meet the former condition exceeds 25 and the number of data points that meet the latter condition exceeds 20, and the filter pressure gauge reading drops below 0.05MPa at the same time as the "sudden pressure drop at the top of the filter" alarm is triggered, then it can be confirmed that the pump itself is not faulty, that is, it is suspected to be burst. If the burst detection output is 1, that is... If the result is positive, it is considered a suspected explosion; if the explosion determination output is 0, then it is considered a suspected explosion. If so, it is determined to be a normal fluctuation; A dual-mode confidence-weighted evaluation is performed on the internal leakage and burst failure conclusions. When both the internal leakage and burst failure conclusions output 1, the internal leakage mode weight and burst failure mode weight are calculated separately to determine whether it is a composite fault and to determine the fault mode to be prioritized. ; ;like Then choose and The fault mode corresponding to the larger of the two is the priority mode; if If the filter bursts, the circulation valve will leak internally due to water hammer impact, and the system will enter the manual confirmation channel, generating two emergency response plans at the same time. If a suspected internal leak is identified, internal leak troubleshooting is performed. Internal leak troubleshooting includes a planned 2-hour shutdown the next day, a single pressure vessel pressure holding test, locating the leak column number, replacing all O-rings and circulation valve cores of the column, and monitoring the rate for the first 4 hours after commissioning. If a suspected burst is detected, burst fault handling procedures are performed. These procedures include immediately switching to the backup filter, isolating the faulty filter, replacing the end cap and filter element within 30 minutes, and continuously monitoring the pump pressure after commissioning to restore it to normal operating level. ±2%; If the fault is determined to be a compound fault, compound fault handling shall be performed. The compound fault handling includes prioritizing the handling of bursting (emergency). After handling, observe for 1 hour. If the magnification is still inverted, then handle internal leakage (non-emergency). After handling, record the handling of both faults and archive the records. After fault handling, effect tracking and rate recovery closed-loop verification are performed. After executing the command, the system automatically enters a 6-hour recovery period for tracking: data is collected every minute. Calculate its relative to the baseline Rebound percentage ,like If the rate reaches ≥90% and remains stable within 6 hours, the treatment is considered successful; if If only 50-89% recovery occurs within 6 hours, it will be marked as "partially effective" and a prompt will be made to check for secondary leaks; if If the success rate is less than 50% within 6 hours, it is considered a "failure" and triggers a secondary diagnosis, i.e., manual diagnosis and treatment. ;in, This is the current multiplier. This is the lowest multiplier during the event, used to calculate the degree of multiplier recovery after the repair; a recovery rate of over 90% is considered successful. Each month, a statistical analysis of all successful and unsuccessful leak incidents is conducted. If the "failure rate" for internal leaks exceeds 20%, it indicates that the current situation is... and The threshold may be too sensitive, leading to over-maintenance. The system automatically relaxes the threshold. and And record the parameter tuning log. Conversely, if the failure rate of burst detection (i.e. bursting but no alarm triggered within 30 minutes) is >10%, then tighten the pressure drop threshold to This ensures a faster response next time, enabling self-learning from historical cases and adaptive threshold optimization.

[0020] This embodiment also provides a periodic low-head subchannel, specifically: by reading the product water tank level and concentrate ratio, performing dual-channel high-frequency sampling and aligning the timestamps, simultaneously reading the product water tank level every 2 minutes. With concentrate ratio A parallel queue with timestamp alignment is generated. If a liquid level or ratio is missing, it is filled in by linear interpolation of the previous 3 points. Data from the last 14 days (10080 points each) is cached to provide sufficient samples for subsequent circadian rhythm identification. Automatically label night shift windows and calculate baseline liquid levels. Designate 22:00-06:00 each day as a "night shift window," for a total of 8 hours (240 points), and calculate the median liquid level for each night shift window. Median of the ratio Similarly, the day shift from 06:00 to 22:00 is calculated. and If three consecutive night shifts ≤1.10 and day shift If the value is ≥1.20, the user will be initially included in the "suspect list of users looking down at their phones". ; ; ; ;in, It serves as a day index, used to distinguish different dates and extract the diurnal rhythm of multiple consecutive days; By performing a negative mapping on the liquid level sequence, the Pearson correlation coefficient within the night shift window is calculated: ; ;in, This is the negative mapping value of the liquid level. This represents the peak liquid level during the night shift, used to convert the peak liquid level into a sequence that moves in the same direction as the trough of the expansion ratio, facilitating the calculation of the mirror correlation coefficient. The Pearson correlation coefficient is used. This is a correlation coefficient calculation function used to quantify the mirror image of liquid level and expansion ratio; a value exceeding 0.85 indicates back pressure valve leakage. Determine whether the mirroring is valid by using mirroring criteria: ;in, For hypothesis testing The value is used to determine the significance of the correlation coefficient, i.e., the correlation coefficient. Is it not a coincidence? This indicates that, assuming the null hypothesis holds, random observations... The probability of ≥0.85 is less than 1%, and the evidence strongly rejects H0. Therefore, the level-expansion ratio mirror relationship is considered statistically significant, and the determination of back pressure valve leakage is reliable. The value was obtained through a significance test of the Pearson correlation coefficient, and the specific path is as follows: Hypothesis: The null hypothesis H0 is that "there is no linear correlation between liquid level and expansion ratio (overall correlation coefficient)". =0); the alternative hypothesis H1 is "there is a linear correlation between the two ( )". ≠0)”; The correlation coefficient r is calculated based on n sampling points within the night shift window (e.g., 8 hours × 30 times / hour = 240 points). First, the negative mapping sequence of liquid level is calculated. With multiple sequence The Pearson correlation coefficient r, which is the Pearson correlation coefficient in the algorithm. ; Constructing the test statistic t: Converting r to the t-statistic: ,in, For degrees of freedom ; To find the two-tailed probability in the t-distribution table or calculate the cumulative probability: Based on the t-value and degrees of freedom, find the two-tailed probability in the cumulative distribution function (CDF) of the t-distribution. ; The values ​​are automatically calculated by the built-in statistical module of the DCS system (Distributed Control System), eliminating the need for manual table lookups; only data needs to be read. It depends on whether the conditions are met; If the mirror result is 0, that is... If the mirroring is not valid, it is marked as a "weak mirror" and downgraded to the manual observation channel; if the mirroring determination output is 1, that is... If the mirroring is successful, the back pressure valve opening during the same night shift is read, and the fault type is determined through linkage verification. ; ;in, This represents the peak value of the maximum opening of the back pressure valve. The maximum opening within a 10-minute window before and after the peak liquid level is the moment when the liquid level in the product water tank reaches its highest point. This is used to determine whether the back pressure valve is forcibly opened by the liquid level. If the linkage verification output is 1, that is... If the fault type is determined to be forced pressure relief by the program, that is, the back pressure valve is forcibly opened by the liquid level, causing the permeate water to backflow into the concentrate side, and the ratio is instantly diluted; if the linkage verification output is 0, that is... If the fault type is determined to be internal leakage of the valve or diaphragm rupture, then it is suspected that the valve diaphragm is torn or the valve core is stuck. Based on the fault type determination conclusion obtained from the linkage verification, the corresponding fault handling measures are executed: If the fault type is determined to be forced pressure relief by the program, immediately reduce the setpoint of the product water tank by 10%, such as from 80% to 70%. Modify the PLC night shift interval to prohibit the forced opening of the back pressure valve during the night shift window, changing it to the logic of "open only when overpressure is 0.05MPa". Observe for 3 days. If the mirror coefficient is... Down to If the ratio is less than 0.6 and the rate drop disappears, the fault is considered successfully resolved. The rate drop disappearing indicates the night shift rate is being handled correctly. The pressure will no longer drop to ≤1.10, but will stabilize at ≥1.18. If the fault type is determined to be internal leakage of the valve or diaphragm rupture, the machine will be shut down for 30 minutes the next day. The back pressure valve end cover will be removed, the diaphragm will be checked for cracks or aging of the O-rings, the entire set of diaphragms and O-rings will be replaced, and a 2-hour pressure holding test will be performed after restarting to ensure that there is no backflow. After the fault handling was performed, the effectiveness of the handling was tracked and the mirror coefficient rebound was verified. After the handling was completed, 72 hours of night shift data were continuously tracked, and the Pearson correlation coefficient was recalculated. , recorded as ,like <0.50 and rebounded to If the value is ≥1.18 (close to the daytime level), then "mirror image removal" is determined; if ≥0.5 and If the value is <0.7, it will be marked "partially effective," prompting you to continue lowering the liquid level setting or to perform a second check of the valve body; if If the value is ≥0.7, the process is deemed "failed" and a secondary diagnosis is triggered, i.e., manual diagnosis and treatment. Monthly statistics are compiled on all "periodic downturn" events. If the proportion of "forced pressure relief" is greater than 70% and the average handling time is less than 0.5 days, the system will proactively tighten the night shift back pressure valve opening threshold from 0.05MPa to 0.03MPa to further reduce the probability of product water backflow. If the recurrence rate of "valve internal leakage or diaphragm rupture" is greater than 20%, an early warning of "valve lifespan reached" will be issued, and it is recommended to replace the entire back pressure valve body during the annual overhaul to avoid repeated disassembly and inspection, so as to achieve self-learning of historical performance and threshold self-adaptation.

[0021] This embodiment also provides that the biological colloid determination is performed as follows: at a set interval, the real-time data of the residual chlorine meter, the influent pH meter, the water temperature meter and the differential pressure gauge are stamped with the same timestamp and written into a 7-day rolling queue. The set interval is 5 minutes. If the instantaneous value of a certain probe is missing, it is filled in by linear interpolation of the previous 3 points to ensure that the subsequent comparison is uninterrupted. Obtain the target stable segment, which is the stable segment of the most recent 72 hours, i.e., the standard deviation is less than half of the historical mean; The residual chlorine baseline, pH baseline, temperature baseline, and pressure differential baseline are calculated separately. The baselines are automatically refreshed at midnight every day to prevent misjudgments caused by load adjustments or seasonal water changes. ; ; ;in, The baseline for residual chlorine is... As the pH baseline, As the temperature baseline, As the differential pressure baseline, This indicates the starting point 30 hours prior, representing a 30-hour period of stability. This indicates the starting point 7 days prior, representing a 7-day stable period. The event window is segmented and the characteristics of each event segment are recorded. The past 7 days are segmented into "day shift-night shift-shutdown-start". The residual chlorine peak, pH drift, average water temperature, and pressure difference rise slope of each segment are recorded. If a segment experiences "shock chlorination", that is, the residual chlorine is instantaneously higher than the baseline by more than 0.5 mg / L, it is marked in red. If an "overnight shutdown" occurs, it is also marked in red to provide time anchor points for subsequent bifurcation.

[0022] When performing an initial check on the EPS release colloidal channel, within a 3-day window centered on "shock chlorination", if the differential pressure slope changes from negative to positive and the rate of increase ranks first in the past 7 days, then "EPS release" is considered highly suspicious. At the same time, check the pH. If the change is not significant, i.e., the drift is <0.2, and the water temperature does not increase significantly, then the colloidal route is preferred. Here, drift is the difference between the median pH and the baseline (drift amount). When performing an initial inspection of the biofilm expansion channel, if the pressure difference increases by more than 10% compared to before shutdown within the first hour after shutdown each day, and the water temperature remains in the range of 25-30℃ with no residual chlorine peak, then "biofilm water absorption and expansion" is considered suspicious. If the pressure difference continues to rise steadily during daytime operation, but the multiplier does not increase, this further confirms biological contamination. When performing channel analysis on pure colloidal or short fiber deposits, if there is no chlorination shock, no overnight jump, and minimal pH drift (pH < 0.2), water temperature below 25°C, but the pressure differential increases slowly and uniformly, it is classified as "pure colloidal or short fiber deposits". This type usually occurs when the efficiency of the fiber recovery system decreases or after a recent replacement with a high-charge retention aid, resulting in insufficient neutralization of colloidal charges and accumulation on the membrane surface. Based on the type of fault, implement the corresponding handling plan: EPS-releasing type: Immediately stop using oxidant, wash with neutral enzyme for 2 hours; circulate clean water; put into normal operation; do not add chlorine within 4 hours after washing to prevent secondary EPS release; Biofilm expansion: Maintain low-dose continuous chlorination (0.1 mg / L) -1 4-hour sterilization; alkaline cleaning removes dead bacteria; clean water circulation; subsequently, continuous chlorination will be changed to weekly shock treatment to prevent biological regeneration; Pure colloidal type: After a single neutral enzymatic wash, add a small amount of nonionic surfactant to strengthen the process for 30 minutes; circulate clean water; at the same time, notify the paper machine staff to check the cylinder wire recovery efficiency to reduce short fibers entering the wastewater from the source; The treatment effect is tracked in a closed loop. After cleaning or chlorination, the pressure difference and permeate flux are recorded continuously for 6 hours. If the pressure difference drops to within ±5% of the baseline and the flux recovers to more than 95%, the treatment is considered successful. If the pressure difference is still high but the flux recovers, it is marked as "partially successful" and a second enzymatic wash is required. If neither of them improves, the treatment is considered a failure and a second diagnosis is triggered, i.e., manual diagnosis and treatment. Monthly statistics are compiled on the success rate of each type of treatment. If the failure rate of EPS type is >20%, the "observation period after shock chlorination" is extended from 3 days to 5 days to avoid premature misjudgment. If the failure rate of biological type is >20%, the "overnight jump" threshold is tightened from 10% to 8% to improve the capture sensitivity. All parameter adjustment records are written to the historical database to enable the threshold to adaptively drift with seasonal and slurry formula changes, so as to achieve monthly self-learning and threshold fine-tuning.

[0023] Example 3 is an improvement on Example 2. In this example, high-magnification analysis is performed, including pH drift analysis. Specifically, data is collected through dual pH high-frequency sampling and drift window segmentation. The instantaneous value of the influent pH meter is read every 2 minutes, and the pressure difference at the corresponding time is recorded. With concentrate ratio The pH sequence of the most recent 72 hours was sliced ​​into "sub-segments" of 4 hours each, resulting in a total of 18 data segments, which were used for subsequent drift calculation and trend judgment. Dynamically analyze the pH baseline, select several segments from the above 18 segments that are "stable in both ratio and pressure difference", and take their median as the "current pH baseline". The standard for judging whether it is stable is that the standard deviation within the segment is less than half of the average of the past 30 days. The baseline is automatically refreshed once a day at 02:00 AM to avoid baseline drift caused by seasonal water changes or chemical adjustments, so as to achieve dynamic pH baseline self-learning and stability locking. Calculate the drift amplitude and perform simultaneous initial screening. For each new slice, calculate the difference between its median pH and the baseline (drift amount). If the drift amounts of two consecutive slices both exceed 0.3 and are less than 0.5, and both slices contain... If an accelerated climb occurs, such as if the climb rate ranks in the top 20% of the past 7 days, it is marked as a "highly suspicious CaCO3 section" and a secondary confirmation of the scaling sentinel signal is performed; otherwise, it is marked as a "safe section". Secondary confirmation of scaling warning signals: In the "highly suspicious section", check whether "white microcrystals" visual signal appears on the concentrate side by uploading photos of online turbidity or photos taken by the duty officer. At the same time, observe the pump pressure. If the pump pressure rises slightly in the same section, it means that the membrane resistance has increased due to scaling, but the permeate flow rate has not decreased significantly, which further confirms that "CaCO3 thin layer" is forming. Initiate online hydrochloric acid washing immediately; if the difference between the median pH and the baseline (drift) is ≥0.3, continue for 2 hours. If the synchronous climb continues for 2 hours and white microcrystals are confirmed on the concentrate side, then the online hydrochloric acid wash is initiated to immediately wash the CaCO3 thin layer in the early stage of its formation, thus preventing thick scale buildup. After confirmation, first reduce the opening of the product water valve to decrease the system recovery rate by about 3% and reduce the supersaturation on the concentrate side. Then inject food-grade hydrochloric acid to maintain the pH of the concentrate side between 2 and 3 and circulate for 30 minutes. During the cleaning process, take a sample every 5 minutes for observation. If the concentrate changes from "milky white" to clear and the pump pressure begins to drop, the acid washing can be terminated in advance to avoid the potential impact of excessively low pH on the polyamide layer. The recovery rate was gradually increased and the effect was tracked. After pickling, the water was replaced for 15 minutes. Then, the recovery rate was gradually restored to the original set value at a rate of 1% every 10 minutes. The pressure difference was continuously monitored during the restoration process. If the pressure difference dropped to within "baseline + 5%" within 2 hours and the pH drift returned to within ±0.1, the treatment was considered successful. If the pressure difference only dropped partially, a second pickling was arranged after 24 hours. If there was no improvement, the treatment was considered a failure and a second diagnosis was triggered, i.e., manual diagnosis and treatment. Monthly statistics are compiled on all successful treatment events. If the pH drift is generally concentrated between 0.3 and 0.4, scaling is triggered, indicating that the alkalinity of the influent has increased seasonally. The system proactively advances the "drift warning threshold" from 0.3 to 0.25 to achieve preventive acid addition. At the same time, the "recovery rate reduction range" is slightly adjusted from 3% to 2% to reduce fluctuations in permeable water volume and maintain stable system operation, so as to achieve monthly threshold self-adaptation and preventive acid addition.

[0024] This embodiment also provides that, in performing high-magnification analysis, concentrated water temperature profile analysis is also included, specifically: Data acquisition is performed through dual-channel synchronous high-frequency sampling. The temperature sensor and the conductivity sensor of the concentrate outlet are read every 1 minute to obtain the instantaneous temperature value and the instantaneous rate value, respectively. At the same time, the inlet water temperature at the corresponding moment is recorded for subsequent elimination judgment of "whether the temperature rise is from the environment". After the data is stored in the database, the timestamp is automatically aligned and the history of the most recent 7 days is cached to provide sufficient background for transient capture. To determine the dynamic temperature baseline, the median of the concentrate temperature is calculated based on the period within the most recent 48 hours when the "ratio is stable and the influent temperature fluctuation is <1℃". The baseline is refreshed every 12 hours to avoid misjudgment caused by diurnal temperature differences or seasonal water changes. Perform initial screening using rigid thresholds for sudden drops and rises, and monitor a 10-minute sliding window. If the ratio drops by ≥0.25 from the start to the end of the window (i.e., close to the 1.0 region), and the concentrate temperature rises by ≥3℃ within the same window, it is immediately marked as "highly suspicious transient". At the same time, check the inlet water temperature. If the inlet water temperature changes by <1℃ within the same window, the possibility of "ambient temperature rise" is ruled out, and the temperature difference persistence is confirmed in conjunction with the water-side pressure. The temperature difference persistence and water-side pressure linkage were verified. For the moment when the "rigid threshold for sudden temperature drops and rises" was locked, observation was performed for 30 minutes. If the concentrate temperature consistently remained above the baseline + 3°C and the multiplier remained below 1.05, the "persistence" was confirmed. Simultaneously, the primary side (heating medium) pressure of the heat exchanger was read. If the primary side pressure experienced a synchronous drop greater than 0.05 MPa at the corresponding moment, it indicated that hot water or steam on the primary side might have leaked to the secondary side (raw water / concentrate), forming a "high temperature, low salinity" short circuit, thus confirming "internal leakage in the heat exchanger." The moment when the "rigid threshold for sudden temperature drops and rises" was locked was the initial moment when the multiplier dropped from ≥1.35 to ≤1.10 and the temperature rose by ≥3°C. This serves as the starting point for subsequent 30 minutes of continuous observation. The baseline at baseline +3°C is the concentrate temperature baseline, and the magnification is recorded as the magnification after washing. This is used to confirm that the temperature rise is not due to environmental fluctuations, but rather to internal leakage in the heat exchanger; Conduct isolation and plate pressure tests, immediately remotely close the inlet and outlet valves of the primary side of the heat exchanger, switch to the standby heat exchanger or bypass to ensure the production line does not stop, and conduct a water pressure holding test on the primary side after isolation, that is, pressurize room temperature clean water to 1.25 times the design pressure, hold the pressure for 30 minutes, and the pressure drop ≤0.05MPa is qualified. If the pressure drop exceeds the standard, disassemble the plates, visually look for cracks or misaligned gaskets, and take photos for record. Perform the replacement of damaged plates and system reset. Replace leaking plates or aged gaskets entirely, tightening each plate to the specified torque. After reassembly, perform a pressure test again. Only those that pass the test are allowed to be connected to the system. Slowly raise the primary side temperature to the operating temperature. Simultaneously, start the concentrate side circulation and observe for 1 hour. If the temperature returns to the baseline ±1℃ and the multiplier rises to ≥1.20, the reset is considered "successful." Here, the baseline is the concentrate temperature baseline. The ratio is the ratio after the current cleaning. After the heat exchanger is repaired, the temperature should return to normal. At ±1℃, the concentration ratio should rise to ≥1.20 before the system's concentration function can be confirmed to have recovered. The system tracks the performance and implements secondary protection, continuously monitoring the temperature and rate of change for the next 72 hours. If both temperature and rate of change remain stable within the normal range, a "successful closed loop" is recorded. If a "sudden drop + sudden rise" occurs again within 24 hours, it is determined that "a secondary leak still exists" or that the backup heat exchanger also has a problem, triggering secondary diagnosis. If the internal leakage of the same model heat exchanger exceeds 2 times per year, the system will automatically prompt "dismantle and inspect all plates and replace the gaskets during the annual overhaul" to prevent repeated leaks from causing the membrane module to operate at a low rate of change for a long time.

[0025] This embodiment also provides that, in performing high-magnification analysis, the following is specifically included: data acquisition is performed through dual-channel high-frequency sampling and night shift window annotation, and the cooling water valve opening feedback signal, membrane stack inlet water temperature, concentrate ratio, and pressure difference are read every 2 minutes. Four data points, with a unified timestamp, designate 22:00-06:00 each day as the "night shift window," providing a fixed time period for subsequent rhythm analysis; Calculate the cooling water valve opening baseline and temperature difference baseline. Take the data from the most recent 7 days of the "daytime window" and calculate the median cooling water valve opening as the "daytime baseline," i.e., the cooling water valve opening baseline. Simultaneously, the median membrane temperature is calculated as the "temperature difference reference," i.e., the temperature difference baseline. The baseline is automatically updated every Monday morning to prevent judgment drift caused by seasonal changes. ;in, This represents the actual percentage opening of the cooling water valve. The membrane stack feed water temperature and the two are linked to identify whether the night shift cooling valve is blindly opened, causing the membrane temperature to drop and the recovery rate to rise erroneously, which in turn triggers a false concentration surge. The baseline of the cooling water valve opening is also monitored. and temperature difference baseline Used to provide a comparison reference for night shifts and identify abnormally large openings of cooling valves; The system automatically increases the opening of the cooling water valve and links it with the membrane temperature drop during the night shift for initial screening. If, during the night shift window, the opening of the cooling water valve increases by ≥20% compared to the daytime baseline and continues for more than 30 minutes, while the membrane temperature drops by ≥2℃, it is immediately marked as "highly suspicious opening". Continue to observe. If the concentrate flow rate decreases and the permeate flow rate increases, it indicates that the system is automatically increasing the recovery rate to "maintain permeate". Then, the system performs synchronous confirmation of the surge in differential pressure. The system synchronously confirms a sharp increase in pressure differential during the execution ratio. For the time period locked by the "automatic increase in concentration during the night shift and the initial screening linked to membrane temperature decrease," it tracks for 1 hour. If the concentrate ratio instantly rises from the original 1.2-1.3 range to >1.5, and the pressure differential increases at a rate that ranks among the top 10% of the past 7 days within the same hour, then a "false concentration surge" is confirmed. At this point, the influent temperature has decreased, but due to the excessively high concentration ratio on the concentrate side, the CaCO3 supersaturation instantly exceeds the maximum, causing the pressure differential to rise at an accelerated rate. The time period locked by the "automatic increase in concentration during the night shift and the initial screening linked to membrane temperature decrease" is within the night shift window that meets the following conditions: Increase ≥20% and A continuous 30-minute period with a temperature drop of ≥2℃ is used to identify the starting point of a chain reaction that leads to a drop in membrane temperature and the need to improve system recovery rate and maintain production due to malfunction of the cooling valve. The control strategy is decoupled and intervened immediately to lock the upper limit of the recovery rate. The recovery rate setpoint is fixed at the "daytime operating value" through the PLC, and automatic adjustment due to temperature drop is prohibited. The PID control loop of the cooling water valve is changed from "temperature-feedback" to "fixed value opening" mode, and the opening is fixed near the daytime baseline to allow the membrane temperature to naturally rise to the original range and avoid further overcooling. At the same time, the small flow bypass of the concentrate discharge is opened for 10 minutes to quickly discharge the concentrated high brine and reduce the instantaneous supersaturation. After the above automatic intervention is completed, a human-machine interface prompt box will pop up. The operator needs to confirm that "the recovery rate is locked, the cooling valve is switched to manual, and the bypass is opened". If the operator clicks "confirm", the system will write the new "recovery rate limit" and "cooling valve set value opening" into the permanent parameter table. If "cancel" is clicked, the original PID strategy will be restored, but a "human over-control" will be recorded for subsequent auditing. The implementation effect is tracked and the rate-pressure difference rebound is verified. After the intervention, the monitoring continues for 4 hours. If the rate drops to below 1.3 and the pressure difference stops rising, the intervention is considered "successful". If the rate remains high, it indicates "there may be real scaling". Online hydrochloric acid washing is started immediately. If the pressure difference continues to rise, the load reduction operation is arranged in advance to prevent damage to the membrane element. Monthly data statistics are performed: If the "cooling water valve automatically opens higher during the night shift" event occurs ≥3 times, and all of them are due to "maintaining production water" resulting in a ratio >1.5, the system will automatically suggest "increasing the daytime baseline opening by 5%" to reduce excessive adjustment during the night shift. If there are zero events but the membrane temperature frequently exceeds the upper limit, the baseline opening will be appropriately reduced to prevent overcorrection. All adjustment records are archived to form a three-dimensional database of "temperature-ratio-pressure difference" to provide predictive adjustment basis for the next season, enabling monthly review and PID strategy optimization.

[0026] In this embodiment, sub-second online meter sampling and multi-spectral fusion judgment based on rate, differential pressure, temperature, and liquid level enable progressive and precise location of three major faults: leakage, scaling, and biological contamination. This significantly reduces blind membrane disassembly and repeated cleaning. By proactively sealing leaks and addressing issues through online acid washing and enzyme washing, differential pressure is stabilized, permeate production consistently meets standards, and membrane life is significantly extended. Overall operating efficiency is improved, while chemical and downtime costs decrease simultaneously. Through closed-loop log self-learning, thresholds are adaptively shifted seasonally, maintaining a low failure rate and high permeate reliability over the long term.

[0027] In the embodiments, all "All" means that all of the conditions must be met. "All" means "or", meaning that either condition must be met.

Claims

1. A zero-discharge recycling treatment process for wastewater from paperboard production, characterized in that: include: Determine if the parameters are abnormal; If the system enters the "triggered state", then perform the initial screening of the concentration factor; Extract and verify the initial screening results; If the initial screening result after verification is CH1, then perform low-magnification analysis; If the initial screening result after verification is CH2, then perform high-magnification analysis; After performing high-magnification analysis, the instantaneous pressure difference was recalculated; If the pressure difference stabilizes within ±10% of the baseline, the case is closed. If the pressure difference is not stabilized within ±10% of the baseline, a secondary diagnostic test will be triggered. The initial screening for concentration factor is performed as follows: Perform dual-conductance synchronous high-frequency acquisition and Timestamp alignment buffering; Clean the conductivity data and remove outlier values. Calculate the dynamic baseline fold and robust standard deviation; Calculate the real-time multiplier; The initial screening is conducted using three thresholds to determine the corresponding execution flow channel and form an initial screening conclusion. The duration and stability of the initial screening results were verified.

2. The zero-discharge recycling treatment process for paperboard production wastewater according to claim 1, characterized in that: To determine if a parameter is abnormal, the following steps are taken: Collect target parameters; Calculate the instantaneous pressure difference; Execute sliding window caching; Real-time data cleaning and abnormal pulse removal are performed on the cached data; Calculate the current dynamic baseline and determine if the system is functioning correctly; Continuously calculate the trend slope and determine the system state; If the relative rate of change is ≥30%, the parameter is determined to be abnormal and the system enters the "trigger state"; If the relative rate of change is between 25% and 30%, it indicates that the relative rate of change is at the early warning line, and the system enters "observation mode" with a yellow light warning. If the relative rate of change is less than 25%, the parameter is considered normal and the system continues to monitor. Verify the duration and filter out false triggers; After the trigger is confirmed, the system automatically captures the relevant parameters from the 14 days prior to the trigger, packages them to generate a time-series comparison chart, and writes it to the trigger log.

3. The zero-discharge recycling treatment process for paperboard production wastewater according to claim 1, characterized in that: Perform low-magnification analysis, specifically: S1. Segment the historical curve and generate a sliding window feature; S2. Traverse all window features and rigidly screen whether the "fall from the normal concentration area into the short circuit area" event has occurred; S3, Extract the time of the fall. Regarding the moment of the fall The subsequent 6 windows were used to verify the rebound suppression. S4. Determine whether there are periodic head-down events during the night shift in the historical curve; S5. Extract the marked suspected events and perform dual verification of the day and night rebound amplitude and stability of the suspected events; S6. Calculate confidence-weighted scores and make final channel decisions for events that are "inverted" or "periodicly downward". S7. If the final decision of the channel is that the inverted structure is valid, then the inverted sub-channel will be executed. If the final decision of the channel is that the periodic downward movement is valid, then the periodic downward movement sub-channel will be executed. S8. After executing step S7, re-evaluate whether the parameters are abnormal; If the parameters are found to be abnormal, the biocolloid determination will be performed with the label "false low ratio ruled out"; If the parameters are determined to be normal or the system enters "observation mode", then no further determination will be performed.

4. The zero-discharge recycling treatment process for paperboard production wastewater according to claim 3, characterized in that: Execute the inverted subchannel as follows: Data is collected synchronously at high frequency from three parameters: permeate, concentrate, and pump pressure, and a dynamic sliding queue is established. Dynamically anchor the baseline and calculate the real-time deviation. Internal leakage detection determines whether an internal leakage fault exists. If the internal leakage detection output is 1, then it is considered a suspected internal leakage. If the internal leakage detection output is 0, then it is considered normal fluctuation; The presence of a bursting fault is determined by bursting detection. If the burst detection output is 1, then it is determined to be a suspected burst. If the burst detection output is 0, then it is considered a normal fluctuation; A dual-mode confidence-weighted assessment was performed on the internal leakage and bursting determination conclusions. If an internal leak is suspected, internal leak troubleshooting will be performed. If a suspected burst is detected, burst fault handling shall be performed. If the fault is determined to be a complex fault, then complex fault handling shall be performed. After the fault handling is performed, effect tracking and closed-loop verification of the rate recovery are carried out.

5. The zero-discharge recycling treatment process for paperboard production wastewater according to claim 3, characterized in that: Execute the periodic low-head sub-channel, specifically as follows: By reading the product water tank level and concentrate ratio, dual-channel high-frequency sampling is performed and the timestamps are aligned. Automatically label the night shift window and calculate the baseline liquid level; The liquid level sequence is negatively mapped, and the Pearson correlation coefficient within the night shift window is calculated. The mirroring is used to determine whether the mirroring is valid. If the mirror judgment output is 0, the mirror judgment is invalid, it is marked as "weak mirror", and it is downgraded to the manual observation channel; If the mirror judgment output is 1, the mirror judgment is established. The back pressure valve feedback opening during the same night shift is read, and the fault type is determined through linkage verification. If the linkage verification output is 1, the fault type is determined to be forced pressure relief by the program. If the linkage verification output is 0, the fault type is determined to be internal leakage of the valve or diaphragm rupture. Based on the fault type determination conclusion obtained from the linkage verification, implement the corresponding fault handling measures. After the fault handling is performed, the effectiveness of the handling is tracked and the mirror coefficient rebound is verified.

6. The zero-discharge recycling treatment process for paperboard production wastewater according to claim 3, characterized in that: The biological colloid determination is performed as follows: At set intervals, the four real-time data points from the residual chlorine meter, influent pH meter, water thermometer, and differential pressure gauge are stamped with the same timestamp and written into a 7-day rolling queue. Obtain the target stable segment; Calculate the baseline for residual chlorine, pH, temperature, and pressure difference, respectively. Segment the event window and record the characteristics of each event segment; Perform initial inspection of EPS release colloidal channels; Perform initial inspection of the biofilm expansion channels; Perform pure colloidal or short fiber deposition channel analysis; Execute the corresponding handling plan according to the type of fault; Closed-loop tracking and handling effectiveness.

7. The zero-discharge recycling treatment process for paperboard production wastewater according to claim 1, characterized in that: Perform high-magnification analysis, including pH drift analysis, specifically: Data acquisition was performed using dual pH high-frequency sampling and drift window segmentation; Dynamic analysis of pH baseline; Calculate the drift amplitude and perform initial synchronization screening; Secondary confirmation of the scaling warning signal; Instantly initiate online hydrochloric acid washing; Implement a step-by-step recovery rate and track the results.

8. The zero-discharge recycling treatment process for paperboard production wastewater according to claim 7, characterized in that: High-magnification analysis also includes concentrate temperature profile analysis, specifically: Data acquisition is performed using dual-channel synchronous high-frequency sampling; Establish a dynamic temperature baseline; Perform initial screening using rigid thresholds for sudden drops and rises in price; Confirm the linkage between temperature difference persistence and water-side pressure; Conduct isolation and plate pressure tests; Perform damaged plate replacement and system reset; Execution effect tracking and secondary protection.

9. The zero-discharge recycling treatment process for paperboard production wastewater according to claim 7, characterized in that: Performing high-magnification analysis also includes, specifically: Data was collected using dual-channel high-frequency sampling and night shift window labeling. Calculate the baseline of cooling water valve opening and the baseline of temperature difference; The night shift system automatically activates the primary screening mechanism in conjunction with a decrease in membrane temperature. Simultaneous confirmation of a sharp increase in the pressure differential at the execution ratio; Decoupling from implementation control strategies and immediate intervention; Execution effect tracking and ratio-pressure difference rebound verification.

Images