Process control-based concentration adjustment system for preparing liquid fertilizer from waste liquid of thiolutin

Abstract

This invention relates to the field of industrial process automation control and waste liquid resource utilization technology, specifically to a concentration regulation system for preparing liquid fertilizer from erythromycin waste liquid based on process control. The system includes: a data acquisition terminal acquiring fluid dynamics time-series data, concentration characterization time-series data, and actuator status data; an ideal reference reconstruction module reconstructing ideal concentration data based on the fluid dynamics time-series data and actuator status data; an interference simulation module injecting interference into the ideal concentration data based on preset fouling characteristics, rheological interference characteristics, and fluid dynamics time-series data to generate simulated concentration data; a coupling decision module generating actual concentration estimation data based on the concentration characterization time-series data, and accordingly generating regulation commands including closed-loop regulation, freeze regulation, cleaning, and temperature compensation commands; and a regulation execution module responding to the regulation commands and controlling the corresponding mechanisms. This invention ensures that the actions do not excessively rely on pollution observation values.

Description

Technical Field

[0001] This invention relates to the field of industrial process automation control and waste liquid resource utilization technology, specifically to a thioerythromycin waste liquid preparation liquid fertilizer concentration adjustment system based on process control. Background Technology

[0002] With the continuous development of bio-fermentation technology, the continuous process of preparing liquid fertilizer from thioerythromycin waste liquid is becoming increasingly complex. This waste liquid usually contains mycelial residue, organic suspended matter, and viscous components with shear-sensitive properties, which brings many challenges to the accurate monitoring and adjustment of concentration during the process.

[0003] Currently, concentration sensors are typically installed on the target pipeline for online sensing, and a traditional closed-loop control system is used to drive the dosing pump directly based on the single drop in the concentration sensor reading, so as to continuously increase the amount of replenishing solution added.

[0004] However, traditional concentration adjustment methods rely too heavily on a single detection reading. When residual substances in the waste liquid adhere to the sensor probe surface, causing sensor interface isolation, or when changes in temperature and flow rate cause a short-term enhancement of non-Newtonian rheological properties, the concentration detection signal will slowly decay or deviate from the true value. Traditional systems often mistake these detection artifacts caused by fouling or rheological disturbances for insufficient effective ingredients, thus blindly increasing the dosage, ultimately leading to excessively high concentrations of the finished liquid fertilizer, waste of raw materials, and loss of quality control.

[0005] Therefore, how to effectively separate the true concentration deviation from the sensor detection artifacts from the observed signal and achieve robust regulation of the fertilizer production process has become an urgent problem to be solved in this field. Summary of the Invention

[0006] The purpose of this invention is to provide a process-controlled system for adjusting the concentration of thioerythromycin waste liquid used in the preparation of liquid fertilizer, thereby solving the following technical problems:

[0007] By separating the true concentration deviation from the detection artifacts, robust regulation of the thioerythromycin waste liquid fertilizer production process can be achieved, avoiding excessive dosing, raw material waste, and loss of control over liquid fertilizer quality caused by sensor contamination or rheological disturbances.

[0008] The objective of this invention can be achieved through the following technical solutions:

[0009] A process control-based liquid fertilizer concentration regulation system for thioerythromycin waste liquid preparation includes a data acquisition terminal set in the target pipeline, an ideal reference reconstruction module, an interference simulation module, a coupling decision module, and a regulation execution module.

[0010] The target pipeline includes a main pipeline and connected dosing branches. The main pipeline is equipped with a regulating valve, a dosing point, a mixing section and a concentration detection point with a concentration sensor along the flow direction. The dosing branches are equipped with dosing pumps. The regulating execution module is connected to the regulating valve, the dosing pump, the cleaning mechanism and the temperature compensation mechanism.

[0011] The data acquisition terminal acquires fluid dynamics time-series data, concentration characterization time-series data, and actuator status data; the ideal reference reconstruction module reconstructs ideal concentration data based on the fluid dynamics time-series data and actuator status data.

[0012] The interference simulation module injects interference into the ideal concentration data based on preset fouling characteristics, rheological interference characteristics, and the fluid dynamics time series data to generate simulated concentration data;

[0013] The coupling decision module generates actual concentration estimation data based on the concentration characterization time series data, calculates the actual residual data between the actual data and the ideal concentration data, and the theoretical residual data between the simulated concentration data and the ideal concentration data, performs dynamic time warping calculation on the two residual data to generate time series similarity data, and generates adjustment instructions including closed-loop adjustment, freeze adjustment, cleaning and / or temperature compensation instructions accordingly.

[0014] The adjustment execution module responds to the adjustment command and controls the corresponding mechanism.

[0015] Preferably, the fluid dynamics time-series data includes the real-time flow rate of the main pipeline, the pressure difference between the front and rear ends of the mixing pipe section, and the fluid temperature; the concentration characterization time-series data includes the online conductivity sampling value and pH sampling value at the concentration detection point; and the actuator status data includes the dosing pump speed feedback value and the regulating valve opening feedback value.

[0016] Preferably, the ideal reference reconstruction module is further used to: perform time synchronization processing on the fluid dynamics timing data and the actuator state data;

[0017] Based on the synchronized inflow rate, dosing rate, and preset concentration parameters, the ideal concentration data corresponding to the target time is calculated; wherein, the preset concentration parameters include the effective component mass concentration base value of the erythromycin waste liquid to be treated in the main pipeline and the effective component mass concentration base value of the replenishment liquid in the dosing branch.

[0018] Preferably, the interference simulation module is also used to: perform time-series drift processing on the ideal concentration data based on the preset fouling attenuation rules and the pressure difference change rate at the front and rear ends of the mixed pipe section, to generate fouling simulation data;

[0019] Based on the preset rheological damping rules, the fluid temperature, and the real-time flow rate of the main pipeline, the ideal concentration data is subjected to transient perturbation processing to generate rheological simulation data; the fouling simulation data and the rheological simulation data are superimposed to generate the simulated concentration data.

[0020] The preset contamination characteristics and the preset rheological interference characteristics include at least the interference amplitude, time constant, and operating window.

[0021] Preferably, the coupling decision module is further configured to: convert the online conductivity sampling value and the pH sampling value into the actual concentration estimation data based on a preset calibration model;

[0022] Within a preset time window, dynamic time warping distance is calculated for the actual residual data and the theoretical residual data; the dynamic time warping distance is normalized to generate the time series similarity data, wherein the larger the value of the time series similarity data, the more similar the actual residual data and the theoretical residual data are.

[0023] Preferably, the coupling decision module is further configured to: determine the current state as an interference state in response to the temporal similarity data being greater than or equal to a preset second similarity threshold, and generate a freeze adjustment command;

[0024] In response to the time-series similarity data being less than or equal to a preset first similarity threshold, the current state is determined to be a true concentration deviation, and a closed-loop adjustment command is generated; wherein, the second similarity threshold is greater than the first similarity threshold.

[0025] Preferably, the coupling decision module is further configured to: in response to the time-series similarity data being located between the preset first similarity threshold and the preset second similarity threshold, obtain the current temperature change rate and the current pressure difference change rate;

[0026] In response to the temperature change rate being higher than a preset temperature change rate threshold or the pressure difference change rate being higher than a preset pressure difference change rate threshold, the disturbance state is determined to be a rheological disturbance, and a temperature compensation command is generated in the adjustment command.

[0027] In response to the temperature change rate not being higher than the preset temperature change rate threshold and the pressure difference change rate not being higher than the preset pressure difference change rate threshold, the interference state is determined to be sensor contamination interference, and a cleaning command is generated in the adjustment command.

[0028] Preferably, the adjustment execution module is further configured to: maintain the current adjustment valve opening and the current dosing pump speed in response to the freeze adjustment command;

[0029] In response to the cleaning command, the cleaning mechanism is started, wherein the cleaning mechanism is an ultrasonic cleaning mechanism; in response to the temperature compensation command, the temperature compensation mechanism is started; in response to the closed-loop adjustment command, the opening of the regulating valve and the speed of the dosing pump are adjusted to the target value.

[0030] Preferably, the data acquisition terminal is also used to: remove outliers and compensate for missing values ​​in the fluid dynamics time series data, the concentration characterization time series data, and the actuator status data;

[0031] The processed data is written into the cache queue according to a uniform sampling period and sent to the ideal benchmark reconstruction module, the interference simulation module and the coupling decision module.

[0032] Preferably, the system is applied to the process of preparing liquid fertilizer from thioerythromycin waste liquid; wherein, the preset fouling characteristics correspond to the mycelial residue adhesion characteristics and the organic suspended matter scaling characteristics; and the preset rheological interference characteristics correspond to the non-Newtonian fluid viscosity abrupt change characteristics.

[0033] The beneficial effects of this invention are:

[0034] 1. This invention generates simulation data by reconstructing ideal concentration data and injecting preset interference, and then compares it with the actual concentration residual through dynamic time normalization. This mechanism can effectively identify the similarity of evolutionary morphology, separate the real deviation from the artifacts such as contamination or rheological disturbances, and avoid blindly relying on distorted signals to cause excessive dosing and quality control failure.

[0035] 2. This invention establishes a dual-threshold decision mechanism to deal with complex working conditions; when the time sequence similarity is in the middle area between the two thresholds, the system further combines the current temperature change rate and pressure difference change rate to perform secondary diversion, accurately distinguishing rheological interference from sensor contamination; this achieves refined identification of the source of anomalies so as to take targeted intervention.

[0036] 3. The adjustment and execution module of this invention adopts multi-dimensional control actions based on the coupling decision results, breaking the limitations of traditional single closed-loop following; when a disturbance is confirmed, the system freezes the adjustment to maintain the current working condition, or starts the cleaning and temperature compensation mechanism; closed-loop adjustment is only performed when a real deviation is confirmed, ensuring that the action does not overly depend on the pollution observation value.

[0037] 4. At the data acquisition level, this invention introduces a time synchronization processing and abnormal data cleaning mechanism; by aligning multi-source time-series data, removing instantaneous noise and compensating for missing values, it eliminates baseline distortion caused by lag in upstream and downstream actions; this provides continuous and consistent high-reliability data support for subsequent residual analysis. Attached Figure Description

[0038] The invention will now be further described with reference to the accompanying drawings.

[0039] Figure 1 This is a schematic diagram of the module structure of the thioerythromycin waste liquid preparation liquid fertilizer concentration adjustment system based on process control of the present invention. Detailed Implementation

[0040] 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.

[0041] Please see Figure 1 A process control-based liquid fertilizer concentration regulation system for thioerythromycin waste liquid preparation includes a data acquisition terminal set in the target pipeline, an ideal reference reconstruction module, an interference simulation module, a coupling decision module, and a regulation execution module.

[0042] The target pipeline includes the main pipeline and the connected dosing branch pipelines. The main pipeline is equipped with a regulating valve, a dosing point, a mixing section and a concentration detection point with a concentration sensor along the flow direction. The dosing branch pipeline is equipped with a dosing pump. The regulating execution module is connected to the regulating valve, the dosing pump, the cleaning mechanism and the temperature compensation mechanism.

[0043] The data acquisition terminal acquires fluid dynamics time-series data, concentration characterization time-series data, and actuator status data; the ideal benchmark reconstruction module reconstructs ideal concentration data based on the fluid dynamics time-series data and actuator status data;

[0044] The interference simulation module injects interference into the ideal concentration data based on preset fouling characteristics, rheological interference characteristics, and fluid dynamics time series data to generate simulated concentration data;

[0045] The coupling decision module generates real concentration estimation data based on concentration characterization time series data, calculates the real residual data between the actual and ideal concentration data, and the theoretical residual data between the simulated concentration data and the ideal concentration data, respectively, performs dynamic time warping calculation on the two residual data to generate time series similarity data, and generates adjustment commands including closed-loop adjustment, freeze adjustment, cleaning and / or temperature compensation commands based on this.

[0046] The adjustment execution module responds to adjustment commands and controls the corresponding mechanism.

[0047] This embodiment provides a concentration regulation mechanism for the preparation of liquid fertilizer from thioerythromycin waste liquid; specifically, the system is arranged on the target pipeline where the fermentation waste liquid enters the liquid fertilizer finished product mixing section, and the target pipeline undertakes the continuous process tasks of raw liquid transportation, replenishment, mixing, detection and feedback regulation;

[0048] The main pipeline is responsible for transporting the thioerythromycin waste liquid to be treated. The dosing branch is responsible for injecting the replenishing liquid or concentrated conditioning liquid into the main pipeline. A mixing section is set up after the dosing point so that the mainstream waste liquid and the replenishing liquid can be turbulently mixed within a certain residence time. Then, the concentration sensor at the concentration detection point can sense the state of the mixed medium online.

[0049] The regulating valve is used to regulate the main flow rate of the main pipeline, the dosing pump is used to control the replenishment amount of the dosing branch, the cleaning mechanism is used to restore the surface activity of the contaminated sensor, and the temperature compensation mechanism is used to reduce the detection distortion and mixing deviation caused by fluid temperature changes.

[0050] In detail, the system does not directly take the single reading of the concentration sensor as the only control basis, but understands multiple physical processes in the industrial site separately; the thioerythromycin waste liquid comes from the fermentation by-product stream, which usually contains mycelial residue, organic suspended matter and viscous components with shear-sensitive characteristics. Therefore, a decrease in sensor reading does not necessarily mean that the effective ingredient has actually decreased.

[0051] First, fluctuations in feed, changes in main pipeline flow, or insufficient replenishment ratio can cause deviations in the true concentration. Second, if mycelial residue adheres to the surface of the sensor probe, the sensing interface will be isolated, resulting in a slow decay of conductivity or acid-base response. Third, if the temperature changes rapidly or the flow rate changes, the waste liquid may experience a short-term enhancement of non-Newtonian rheological properties, which will affect ion migration and mixing uniformity, causing the detection signal to deviate from the true concentration.

[0052] Therefore, this system first collects time-series information related to the above physical processes through the data acquisition terminal, and then the ideal reference reconstruction module establishes the ideal concentration that the concentration detection point should present under the condition of no contamination and no additional rheological interference based on the current fluid flow state and actuator state.

[0053] The disturbance simulation module maps typical fouling and rheological disturbances into industrial disturbance trajectories that can evolve over time. These disturbances are applied to the ideal concentration to form simulated concentration data showing how the reading should change if the current reading distortion is caused by fouling or rheology. The coupling decision module then converts the actual measured concentration characterization time series data into real concentration estimation data and constructs real residuals and theoretical residuals respectively. Dynamic time warping is used to examine whether the two are similar in terms of time evolution.

[0054] If the two are similar, it means that the current deviation is more likely caused by interference rather than an abnormal actual concentration; if the two are significantly different, it means that the current deviation is more likely to come from the actual process deviation, and closed-loop regulation should be implemented; the regulation execution module decides accordingly whether to maintain the current state of the valve and pump, perform cleaning, enable temperature compensation, or readjust the valve opening and the dosing pump speed.

[0055] As a safety precaution, the system adopts a conservative strategy in the following situations: if the data acquisition terminal finds that the key sensor is temporarily offline, the actuator feedback is interrupted, or the sampling timing is obviously misaligned, the coupling decision module will not directly output a forced dosing command, but will prioritize maintaining the current stable operating condition to avoid making incorrect dosing based on incomplete data.

[0056] If the ideal baseline reconstruction module cannot form a reliable ideal concentration baseline, for example, if there is severe pulsation in the main pipeline or the equipment is in the initial stage of startup and has not yet entered the stable mixing zone, the system can temporarily suspend the residual coupling decision and only retain basic safety control; if the cleaning mechanism or temperature compensation mechanism fails, the adjustment execution module will freeze the active correction action and output a maintenance prompt to the upper monitoring system to prevent the control error from continuing to amplify when the interference is not cleared.

[0057] For example, in a continuous production shift of liquid fertilizer, fermentation waste liquid enters the main pipeline from the temporary storage tank. In the morning, the operating conditions are relatively stable, with the flow rate, pressure difference, and temperature of the main pipeline within the normal range, and the output of the concentration detection point is also relatively stable. In the afternoon, the amount of mycelial residue in the raw liquid increases, and the reading of the concentration sensor begins to slowly decrease. If a traditional closed-loop control is used, the system will usually mistakenly believe that the effective ingredients are insufficient, thereby driving the dosing pump to continuously increase the amount of supplementation, ultimately resulting in a high concentration of the finished liquid fertilizer.

[0058] In this embodiment, the ideal reference reconstruction module judges based on the current flow rate and actuator status that, according to the mass conservation relationship, the concentration after mixing should not decrease by the same magnitude. At the same time, the fouling attenuation trajectory generated by the interference simulation module is highly consistent with the actual residual in terms of time form. The coupling decision module thus determines that the current state is closer to the probe fouling state, and outputs a freeze adjustment command and triggers a cleaning command at the same time, instead of blindly increasing the dosage. After cleaning, the sensor response recovers, and the system re-enters normal closed-loop adjustment.

[0059] The purpose of this step is to separate the true concentration deviation from the detection artifacts from the same observation signal, thereby achieving robust regulation of the thioerythromycin waste liquid fertilizer production process and avoiding excessive dosing, raw material waste, and loss of control over liquid fertilizer quality caused by sensor contamination or rheological disturbances.

[0060] In a preferred embodiment of the present invention, the fluid dynamics time-series data includes the real-time flow rate of the main pipeline, the pressure difference between the front and rear ends of the mixing pipe section, and the fluid temperature; the concentration characterization time-series data includes the online conductivity sampling value and pH sampling value at the concentration detection point; and the actuator status data includes the dosing pump speed feedback value and the regulating valve opening feedback value.

[0061] This embodiment provides a data acquisition configuration method for the above-mentioned concentration regulation system; specifically, the data acquisition terminal collects at least three types of field data: the first is fluid dynamics time-series data, including real-time flow rate of the main pipeline, pressure difference between the front and rear ends of the mixing pipe section, and fluid temperature; the second is concentration characterization time-series data, including online conductivity sampling value and pH sampling value; and the third is actuator status data, including dosing pump speed feedback value and regulating valve opening feedback value.

[0062] Specifically, the above data are not arbitrary combinations, but correspond to different physical levels in the liquid fertilizer preparation process; the real-time flow rate of the main pipeline reflects the amount of waste liquid passing through the system per unit time, which determines the basis for the ratio of replenishing liquid to mainstream waste liquid.

[0063] The pressure difference between the front and rear ends of the mixing section reflects the flow resistance and local structural changes of the fluid in the mixing section. When the viscosity of the waste liquid increases, the amount of flocculent material increases, or the flow regime changes abruptly, the pressure difference often shows abnormalities before the concentration reading. The fluid temperature directly affects the ion migration ability, sensor response sensitivity, and rheological properties of the waste liquid, and is a key factor in distinguishing whether the reading change comes from a change in the actual composition or a change in physical properties.

[0064] Online conductivity sampling values ​​can quickly characterize the overall activity level of conductive components in the system, while pH sampling values ​​reflect the acid-base environment of the system and the chemical equilibrium state of some effective components. The combination of the two helps to reduce misjudgments caused by the drift of a single sensor. The dosing pump speed feedback value and the regulating valve opening feedback value provide the real working conditions of the execution layer, avoiding the control system from ignoring field deviations such as mechanical execution lag, pump slippage, or valve jamming based only on the issued instructions.

[0065] It should be noted that the concentration sensor set at the concentration detection point in the embodiment can be understood as a detection unit for process concentration estimation. It can be an integrated composite sensor or a combination of a conductivity sensor and a pH sensor installed at the same concentration detection point and uniformly accessed by the same data acquisition terminal.

[0066] In other words, both conductivity and pH sampling values ​​are online characterization results of the mixed liquid state at the concentration detection point. They both serve the subsequent estimation of actual concentration and do not require to exist in the form of a single physical probe. After this processing, the concentration sensor setting is consistent with the bivariate acquisition method in this embodiment in engineering, avoiding the misunderstanding that the detection function of the concentration detection point can only correspond to a single measurement quantity.

[0067] As a safety precaution, if a certain type of data is temporarily missing, the system will not shut down immediately. For example, if the conductivity data is normal but the pH sensor is undergoing short-term maintenance, the reliability of the actual concentration estimate can be reduced, and the reliance on the trends of flow rate, pressure difference, and temperature changes can be temporarily increased.

[0068] If the flow meter experiences pulse interference, the system can briefly estimate the flow state based on feedback from valve opening and pump speed, but this alternative mode should not be maintained for a long time. If the differential pressure sensor is blocked or the temperature probe is inaccurate, the system should not directly attribute the current deviation to rheological interference, but instead maintain a strategy that prioritizes freezing or manual verification.

[0069] For example, during the same continuous production shift, the system detected a decrease in conductivity, but at the same time, the pH value changed very little, and the main pipeline flow rate and dosing pump speed remained stable; however, the pressure differential in the mixing section slowly increased, and the fluid temperature did not fluctuate significantly. In this scenario, the multi-source information provided by the data acquisition terminal indicates that the current situation is more likely to be sensor distortion caused by mycelial adhesion and local scaling, rather than the actual dilution of the mainstream material; without the pressure differential dimension, the system would have difficulty detecting the fouling process in a timely manner.

[0070] The purpose of this step is to establish a more reliable on-site condition profile using multi-source data that corresponds one-to-one with the process mechanism, thereby enabling stratified identification of the sources of concentration changes.

[0071] In a preferred embodiment of the present invention, the ideal reference reconstruction module is further used to: perform time synchronization processing on the fluid dynamics time series data and the actuator status data; calculate the ideal concentration data corresponding to the target time based on the synchronized inflow flow rate, dosing flow rate and preset concentration parameters; wherein, the preset concentration parameters include the effective component mass concentration base value of the erythromycin waste liquid to be treated in the main pipeline and the effective component mass concentration base value of the replenishment liquid in the dosing branch.

[0072] This embodiment provides an ideal reference reconstruction mechanism. Specifically, the ideal reference reconstruction module synchronizes the time-series data from the main pipeline, the dosing branch, and the actuator, and then combines the inflow flow rate, the dosing flow rate, and preset concentration parameters to form the ideal concentration data that should be presented at the concentration detection point at the target time. Specifically, the formula for calculating the ideal concentration data is:

[0073]

[0074] in, For inflow flow, To increase the drug flow rate, The base value for the effective component mass concentration of the thioerythromycin waste liquid to be treated. This is the baseline value for the effective component mass concentration of the replenishing solution;

[0075] Specifically, time synchronization is of direct engineering significance in this process. Since the waste liquid in the main pipeline flows from the dosing point through the mixing pipe section before reaching the concentration detection point, the concentration change seen at the detection point usually lags behind the valve action and the dosing pump action. If time synchronization is not performed, the system will incorrectly match the action that just occurred upstream with the response result that has a transmission lag downstream, resulting in distortion of the ideal concentration baseline.

[0076] To this end, the ideal baseline reconstruction module aligns the upstream flow rate changes and dosing behavior in time according to the pipeline structure, mixing section volume and current flow rate status, so that the ideal concentration at the same moment corresponds to the state of the mixed liquid that should have reached the detection point at this moment, rather than simply splicing together the original timestamps of each sensor.

[0077] Among the preset concentration parameters, the effective component mass concentration baseline value of the waste liquid in the main pipeline can be provided by the batch inspection of the raw liquid, the historical statistics of the feed tank, or the pre-analysis unit; the effective component mass concentration baseline value of the replenishment liquid in the dosing branch can be determined by the replenishment liquid preparation specification; based on this, the ideal reference reconstruction module reconstructs a reference concentration trajectory from the perspective of mass conservation that is not affected by probe adhesion, transient rheology, and sensor aging.

[0078] It should be noted that, in order to avoid the same physical quantity appearing under multiple names throughout the text and causing ambiguity, the inflow flow rate and the real-time flow rate of the main pipeline in this embodiment have the same meaning, both referring to the main flow rate entering the target pipeline main pipeline and passing through the regulating valve and the dosing point along the flow direction before entering the mixing pipe section per unit time;

[0079] If expressions such as mainstream flow rate or main pipeline flow rate appear in the following descriptions, they are all engineering descriptions of the inflow flow rate and do not constitute a new data type. Similarly, the dosing flow rate specifically refers to the volumetric or mass flow rate of the replenishing liquid transported by the dosing branch through the dosing pump and injected into the main pipeline at the dosing point. Its essence is used to determine the replenishment ratio, which is different from the dosing pump speed feedback value: the speed feedback value is the status quantity of the actuator, while the dosing flow rate is a process quantity that can be converted from the speed or obtained by actual measurement by the flow meter.

[0080] It should be noted that the dosing flow rate in this embodiment can be obtained from direct flow measurement on the dosing branch, or, in the case where no separate dosing flow meter is set up, it can be obtained by converting the dosing pump speed feedback value with the pump calibration curve, the current back pressure state, and the operating health state. The dosing pump speed feedback value listed in the embodiment is not only used for monitoring, but can also be used as the basic input for the ideal reference reconstruction module to obtain the dosing flow rate.

[0081] When both direct flow measurement and speed feedback are available on-site, the measured value is used first, and the speed feedback is used for consistency verification. This allows the calculation logic based on the synchronized inflow flow and dosing flow in this embodiment to form a closed correspondence with the aforementioned data acquisition, without requiring the addition of an independent hardware structure not specified in this embodiment.

[0082] Furthermore, during engineering implementation, the ideal benchmark reconstruction module can first obtain an initial estimate of the dosing flow rate based on the pump type and historical calibration relationship, and then correct the estimate by combining the valve opening, main pipeline flow rate changes and downstream mixing response;

[0083] When pump slippage, idling, air resistance, or changes in the viscosity of the replenishing fluid occur, the system does not rigidly equate rotational speed with flow rate. Instead, it cross-verifies the actuator status data with fluid dynamics time-series data to form the effective dosing flow rate for mass conservation calculations. This avoids the ideal concentration data from deviating from actual operating conditions due to relying solely on nominal rotational speed.

[0084] It should be noted that the ideal concentration data here does not pursue absolute chemical analysis values, but rather constructs the most reasonable engineering reference value under the current process configuration; its significance lies in providing a zero baseline for subsequent residual analysis, enabling the system to observe the true deviation and the interference deviation separately.

[0085] As a safety precaution, if the feed batch switching is too fast, causing the baseline value of the effective component mass concentration in the main pipeline to lose its representativeness in a short period of time, the ideal baseline reconstruction module should reduce its direct dependence on the baseline value and mark the current operating condition as a transition section. If there is pump idling, pipeline air resistance, or insufficient replenishment liquid in the dosing branch, the dosing flow rate cannot be inferred solely from the set speed, but must be corrected by feedback value and on-site flow status first; otherwise, the ideal baseline will be overly optimistic.

[0086] If the synchronized data still shows significant inconsistencies, such as the valve being opened significantly but the main pipeline flow not changing accordingly, the system can determine that there is a mechanical execution abnormality and suspend the inclusion of the data for that period into the ideal baseline update.

[0087] For example, when the liquid fertilizer production line switches from one fermentation waste liquid storage tank to another, the effective component base level of the raw liquid in the new tank is slightly different. The ideal baseline reconstruction module identifies a transition mixing zone based on the tank switching time, main pipeline flow rate, and mixing section residence time. During this period, although the detection point shows concentration fluctuations, the system interprets them as a natural result of upstream material replacement rather than directly identifying them as control failures. After the transition zone ends, the ideal concentration trajectory is reconstructed using the new raw liquid base value.

[0088] The purpose of this step is to provide a reference baseline consistent with the actual process propagation timeline for subsequent interference identification, thereby enabling active identification of downstream detection deviations rather than passively relying on single-point sensing signals.

[0089] In a preferred embodiment of the present invention, the disturbance simulation module is further configured to: perform time-series drift processing on the ideal concentration data based on a preset fouling attenuation rule and the differential pressure change rate between the front and rear ends of the mixed pipe section to generate fouling simulation data; and perform transient disturbance processing on the ideal concentration data based on a preset rheological damping rule, fluid temperature, and real-time flow rate of the main pipeline to generate rheological simulation data.

[0090] The fouling simulation data and the rheological simulation data are overlaid to generate simulated concentration data; wherein the preset fouling characteristics and preset rheological interference characteristics include at least the interference amplitude, time constant and action window.

[0091] This embodiment provides an interference simulation mechanism; specifically, the interference simulation module models the two most common non-target deviations in the thioerythromycin waste liquid process—sensor fouling and rheological disturbances—as disturbance processes that can evolve over time, and applies them to ideal concentration data to form simulated concentration data;

[0092] Specifically, the fouling attenuation rule adopts an exponential attenuation mapping algorithm, for example, the calculation formula is:

[0093]

[0094] in, For the simulation data of contamination, For ideal concentration data, This is a dimensionless disturbance amplitude coefficient, the value of which is obtained by normalizing the rate of change of pressure difference. The cumulative time constant, The time parameter is used; specifically, a linear mapping or a logistic nonlinear mapping function is used to map the current rate of change of pressure difference to the interval [0,1]. The larger the rate of change of pressure difference, the higher the value of the time parameter. The closer it is to 1;

[0095] Based on preset rheological damping rules and the fluid temperature and real-time flow rate of the main pipeline, transient perturbation processing is applied to the ideal concentration data to generate rheological simulation data. Specifically, the rheological damping rule adopts a transient decay damping model, for example, the calculation formula is as follows:

[0096]

[0097] in, For rheological simulation data, For ideal concentration data, This is a dimensionless damping amplitude coefficient, the value of which is obtained by normalizing the fluctuation rate of fluid temperature and real-time flow rate in the main pipeline. The parameter for the operating window has a time dimension to ensure that the exponential term is dimensionless. For time parameters;

[0098] Specifically, the fouling simulation targets the situation where the probe surface is gradually covered by mycelial residue, organic suspended matter and scale layer. Such interference in the industrial field usually does not occur instantaneously, but is cumulative and delayed: once the initial thin layer is attached to the probe surface, subsequent particles are more likely to remain, and the effective contact ability of the detection interface decreases over time. Therefore, the readings often show a slow drift rather than a sudden change.

[0099] The rate of change of pressure difference between the front and rear ends of the mixed pipe section can serve as evidence of the speed of fouling development, because the pressure difference often changes synchronously when the content of flocculents in the fluid increases, local deposition intensifies, or flow resistance rises; accordingly, the interference simulation module applies time-series drift to the ideal concentration data according to the preset fouling attenuation rules to generate fouling simulation data.

[0100] Rheological simulation targets short-term viscosity anomalies in waste liquids when temperature, flow rate, or local shear conditions change. High molecular weight substances and suspended particles in erythromycin fermentation waste liquid can cause the system to exhibit obvious non-Newtonian behavior. When the temperature suddenly drops, molecular motion weakens and the apparent viscosity of the fluid may increase. When the flow rate changes and the shear environment changes, the mixing uniformity and ion diffusion rate will also change.

[0101] Such interference often manifests as transient damping, that is, the sensor reading shows a short-term deviation that does not completely correspond to the actual concentration change; the interference simulation module uses a preset rheological damping rule, combined with fluid temperature and real-time flow rate of the main pipeline, to apply transient disturbances to the ideal concentration and generate rheological simulation data; the fouling simulation data is superimposed with the rheological simulation data to simulate the complex situation in the industrial field where the probe surface gradually adheres to fouling and the fluid properties change in a short time at the same time.

[0102] The preset fouling characteristics and rheological interference characteristics reflect the strength of the impact of such interference on the detection value; the time constant reflects whether the interference accumulates slowly or decays rapidly; the action window reflects the process period during which such interference usually lasts; the above parameters can be set through equipment debugging, historical batch summarization or on-site maintenance experience, and do not require remodeling every time;

[0103] As a safety precaution, if the differential pressure change rate is very low but the sensor reading continues to decline, the system cannot completely rule out fouling simply because the differential pressure does not respond. Especially when there is localized adhesion to the probe and the overall resistance of the main pipeline has not yet increased significantly, fouling simulation should be allowed to maintain a low intensity of participation.

[0104] If both temperature and flow rate change drastically, rheological disturbances and actual mixing deviations may be coupled. In this case, the simulated concentration data should only be used as a disturbance reference and should not be directly replaced by the actual control baseline. If the composition of the current batch of waste liquid is abnormal and significantly different from historical properties, the preset disturbance characteristics should allow manual revision to prevent old parameters from failing to cover the new material state.

[0105] For example, in the evening, the workshop switches to a lower temperature circulating cooling water to reduce energy consumption, causing the temperature of the waste liquid in the main pipeline to gradually decrease; at the same time, the pressure difference in the mixing section increases slightly; based on this, the system simulates a combined interference trajectory that is first caused by a short-term drop in reading due to rheological damping, and then by continuous drift caused by probe contamination; when the simulated trajectory is consistent with the subsequent evolution direction of the actual residual, the system is more inclined to determine that the current state is dominated by interference, rather than hastily increasing the dosage;

[0106] The purpose of this step is to identify the sources of distortion specific to the waste liquid process as identifiable reference patterns, thereby enabling the decomposition of the sources of abnormal readings on site.

[0107] In a preferred embodiment of the present invention, the coupling decision module is further configured to: convert online conductivity sampling values ​​and pH sampling values ​​into actual concentration estimation data based on a preset calibration model; calculate the dynamic time warping distance between the actual residual data and the theoretical residual data within a preset time window; normalize the dynamic time warping distance to generate time-series similarity data, wherein the larger the value of the time-series similarity data, the more similar the actual residual data and the theoretical residual data are.

[0108] This embodiment provides a concentration estimation and similarity generation mechanism in coupled decision-making. Specifically, the coupled decision-making module first converts the online conductivity sampling value and pH sampling value into actual concentration estimation data, and then compares the actual residual with the theoretical residual through dynamic time warping within a preset time window, and outputs time-series similarity data after normalization.

[0109] In summary, conductivity and pH response to concentration are complementary; conductivity is more sensitive to changes in mobile ions or conductive components in the system, and has a fast response speed, but is easily affected by temperature and surface adhesion.

[0110] Although pH is not a direct equivalent to the concentration of effective ingredients, it can reflect the chemical equilibrium environment and provide auxiliary constraints when the ratio of certain waste liquids fluctuates. The role of the pre-calibrated model is to transform these two types of original characterization quantities into a realistic concentration estimate that is more consistent with the process concentration.

[0111] The preset calibration model is a multivariate functional relationship fitted based on historical calibration data, such as using a multiple linear regression formula:

[0112]

[0113] in, For the estimated actual concentration data, These are online conductivity sampling values. for Sampled value, , They are conductivity and The calibration weighting coefficients, The basic compensation constant;

[0114] This model can be derived from calibration relationships formed during pre-production calibration, laboratory comparative analysis, and equipment installation and commissioning; the resulting real-world concentration estimates are closer to the actual process meaning than simply looking at conductivity.

[0115] The reason for using dynamic time warping in the residual comparison process is that the propagation of disturbances in industrial sites often varies in speed. For example, when the probe is contaminated, sometimes the fluid adheres quickly and causes a significant attenuation in a short period of time, while at other times the flow rate is high and the process is prolonged.

[0116] Even within the same rheological disturbance, there can be instances of premature or delayed events at different flow rates. If only point-to-point comparisons are made at the same instant, disturbances with similar shapes but slightly different velocities are easily misjudged as unrelated. Dynamic time warping focuses on the degree of convergence between two residual trajectories in the overall evolutionary pattern, making it more suitable for identifying industrial events with time-stretching characteristics, such as fouling and rheological disturbances.

[0117] Before performing dynamic time warping distance calculation, Z-Score standardization is performed on the actual residual data and the theoretical residual data to eliminate the difference in dimensional amplitude, so that the algorithm focuses on the comparison of time series patterns.

[0118] It should be further explained that, to ensure the reproducibility of the normalization process in engineering, the coupled decision module can first obtain the dynamic time-warped distance within a preset time window and denote it as D; the smaller the distance D, the closer the two residual trajectories are. Then, this distance is dimensionless according to the window length and scale, and time-series similarity data that satisfies the condition that the larger the value, the more similar the data. The following monotonic mapping can be used:

[0119]

[0120] in, For time-series similarity data, The distance is the dynamic time-normalized distance; when the two residual trajectories are completely identical. ,correspond When the difference increases Increase, corresponding Approaching 0;

[0121] This ensures that the semantics of larger values ​​for time-series similarity data represent greater similarity, maintaining consistency at the implementation level and avoiding ambiguity where the similarity direction is opposite to the threshold decision direction.

[0122] As a safety precaution, if the preset time window is too short, the system may only see a local fragment of the disturbance and it will be difficult to reliably identify the disturbance pattern; if the window is too long, different events may be superimposed together, reducing the sensitivity of the decision. Therefore, the window length can be selected according to the process cycle, the residence time of the mixing section, and the typical rate of fouling development, and the decision can be temporarily shortened or delayed during the transition period after start-up, shutdown, tank change or cleaning.

[0123] If conductivity and There are obvious contradictions, such as a significant decrease in conductivity. If the pressure difference and temperature remain stable over a long period without corresponding changes, the system should raise the suspicion level of local anomalies in the sensor, rather than simply outputting an actual concentration anomaly.

[0124] For example, in the aforementioned evening operating conditions, the conductivity continuously decreases, while Only slight fluctuations were observed; after the coupled decision module combined the calibration relationship to generate the actual concentration estimate, it was found that the actual residuals showed a pattern of slow decline followed by a plateau, which is similar to the trend formed by the accumulation of fouling in the theoretical residuals; even if the two occurred at different rates, dynamic time warping could still identify their common origin, thus providing a basis for subsequent decisions.

[0125] The purpose of this step is to map field deviations from different sources and at different paces into a comparable temporal morphological space, thereby achieving a robust distinction between interfering anomalies and true concentration deviations.

[0126] In a preferred embodiment of the present invention, the coupling decision module is further configured to: determine the current state as an interference state in response to the time series similarity data being greater than or equal to a preset second similarity threshold, and generate a freeze adjustment command; determine the current state as a true concentration deviation in response to the time series similarity data being less than or equal to a preset first similarity threshold, and generate a closed-loop adjustment command; wherein the second similarity threshold is greater than the first similarity threshold.

[0127] This embodiment provides a decision mechanism based on dual thresholds. Specifically, the coupled decision module sets a first similarity threshold and a second similarity threshold, wherein the second similarity threshold is higher than the first similarity threshold. When the sequential similarity reaches or exceeds the second similarity threshold, the system determines that the current deviation is highly consistent with the interference simulation trajectory, identifies the current state as the interference state, and generates a freeze adjustment command.

[0128] When the time sequence similarity is lower than or equal to the first similarity threshold, the system determines that the current deviation does not conform to the interference characteristics and is more likely to originate from the abnormal real concentration, and therefore generates a closed-loop adjustment command.

[0129] Specifically, setting a dual threshold instead of a single threshold is mainly to adapt to the complex interference conditions in industrial sites. The process of making fertilizer from thioerythromycin waste liquid is characterized by complex materials, multiple sources of disturbance and inconsistent propagation speed. At certain times, the reading deviation may include both the actual concentration change and the superimposed effects of slight contamination or temperature.

[0130] If only a single boundary is set, the system is prone to repeatedly switching control modes near the critical point, causing frequent operation of valves and pumps. By setting two thresholds, one high and one low, states that are highly like disturbances can enter the freeze zone, while states that are clearly not like disturbances can enter the closed-loop regulation zone, leaving the middle zone for subsequent more detailed discrimination logic processing.

[0131] The aforementioned preset first similarity threshold and second similarity threshold are obtained through statistical calibration based on historical steady-state batch data and manual cleaning records; the preset temperature change rate threshold and pressure difference change rate threshold are set offline based on the inflection point slope of the viscosity-temperature curve of the corresponding mixing pipe section and medium under different ambient temperatures.

[0132] The essence of freeze-loop control is to stabilize the dosing and flow control when there is sufficient evidence to show that the current deviation is mainly an observational illusion, so as to prevent the control system from chasing the false signal; closed-loop control is used when the deviation is more likely to come from the actual process changes, so that the system can correct the product concentration in a timely manner.

[0133] As a safety precaution, if the similarity exceeds the threshold in a short period of time and then quickly returns to its original state, the system can require several consecutive sampling windows to meet the same judgment direction before officially switching the control state, in order to avoid occasional noise triggering mode oscillations. If the current stage is equipment switching, post-cleaning recovery, or liquid replenishment and replacement, the threshold can adopt a more conservative strategy, prioritizing the observation of stability rather than immediate strong control.

[0134] If the similarity is extremely low but the actual concentration estimate itself is not reliable, for example... If the sensor has just been calibrated but is not yet stable, the system can also temporarily suspend the output of closed-loop control commands;

[0135] For example, in the same production shift, the system observed that the actual residual and the theoretical disturbance residual were highly similar in multiple consecutive windows, and this similarity was mainly manifested as the superposition of slow drift and short-term temperature-sensitive disturbance. Since the similarity continued to fall in a high range, the coupling decision module determined that the current state was a disturbance state, so it no longer increased the speed of the dosing pump, but issued a freeze adjustment command to keep the valve and the dosing pump in the existing operating conditions.

[0136] Conversely, after switching the stock solution tank at night, the actual residual suddenly deviated rapidly and was inconsistent with the simulation interference pattern. The similarity remained low. Based on this, the system determined that the actual concentration was deviated and initiated closed-loop adjustment to correct the replenishment ratio.

[0137] The purpose of this step is to establish a clear and stable control boundary between the credibility of the disturbance and the credibility of the true deviation, so as to ensure that the control action does not overly depend on the contaminated observations.

[0138] In a preferred embodiment of the present invention, the coupling decision module is further configured to: in response to the temporal similarity data being between a preset first similarity threshold and a preset second similarity threshold, obtain the current temperature change rate and the current pressure difference change rate; in response to the temperature change rate being higher than a preset temperature change rate threshold or the pressure difference change rate being higher than a preset pressure difference change rate threshold, determine the disturbance state as rheological disturbance, and generate a temperature compensation instruction in the adjustment instruction;

[0139] In response to the temperature change rate not exceeding the preset temperature change rate threshold and the pressure difference change rate not exceeding the preset pressure difference change rate threshold, the interference state is determined to be sensor contamination interference, and a cleaning command is generated in the adjustment command.

[0140] This embodiment provides a secondary diversion mechanism for non-cancellation control actions; specifically, when the timing similarity is between the first similarity threshold and the second similarity threshold, it indicates that the current deviation cannot be clearly identified as a true concentration deviation, nor can it be directly identified as a typical interference. Therefore, the coupling decision module further examines the temperature change rate and the pressure difference change rate to distinguish between rheological interference and sensor contamination interference.

[0141] Specifically, this additional feature is a reinforcement based on the previous layer of the scheme; while relying solely on the level of similarity can distinguish most cases that are clearly like interference from those that are clearly not, intermediate states often occur in industrial settings.

[0142] For example, when weather changes cause the material temperature to gradually decrease, and there is also slight adhesion on the sensor surface, the residual morphology may not be exactly like typical contamination, nor exactly like pure rheology when these two factors work together.

[0143] If the system simply freezes without further action, the source of interference will persist, and larger errors may accumulate. Therefore, this embodiment introduces temperature change rate and pressure difference change rate as the distinguishing criteria.

[0144] A high rate of temperature change usually indicates that fluid properties are changing rapidly, which can easily trigger changes in non-Newtonian viscosity, ion migration hindrance, or mixing efficiency. A high rate of pressure difference change often indicates that the flow resistance and local flow regime in the pipe are changing, which also supports the judgment of rheological interference. Conversely, if the temperature and pressure difference are relatively stable, and the time similarity is still in the middle range, it is more likely that the measurement interface is contaminated by the gradual adhesion of the probe surface. In this case, it is more appropriate to use the cleaning command first.

[0145] It should be noted that, in order to avoid the rate of change being interpreted as different calculation methods in different paragraphs, in this embodiment, the current temperature change rate and the current pressure difference change rate are both calculated within the same decision window based on the unified sampling period data output by the data acquisition terminal, and the synchronized physical quantity sequence is used for calculation first.

[0146] For example, the calculation method of the difference between the start and end values ​​of the window can be used to calculate the current temperature change rate separately. and the rate of change of current pressure difference :

[0147]

[0148]

[0149] The fluid temperature in the detection section forms a temperature sequence over time and is denoted as... The pressure difference between the front and rear ends of the mixed pipe section forms a pressure difference sequence over time and is denoted as . The current judgment time is recorded as ,and The time span corresponding to the current decision window or the equivalent time span of adjacent sampling points within the window;

[0150] By providing a consistent calculation method, it can be ensured that the temperature change rate threshold and the pressure difference change rate threshold of the embodiments have a clear dimensional basis, avoiding the same change rate being mistakenly treated as slope, variance or fluctuation amplitude in different implementations, which would lead to inconsistent judgments.

[0151] It should be noted that the differential pressure change rate has a stratified usage logic in this scheme and does not conflict with the aforementioned fouling simulation: In the interference simulation stage, the differential pressure change rate at the front and rear ends of the mixed pipe section is mainly used as low-frequency circumstantial evidence of the fouling formation trend, reflecting the slow evolution when flocculent matter increases, deposition is enhanced, or the risk of adhesion rises.

[0152] In the gray zone secondary diversion stage of this embodiment, the current differential pressure change rate under consideration emphasizes the intensity of the instantaneous change within a shorter decision window, which is used to identify whether there is a rapid flow change sufficient to cause abrupt changes in mixing conditions and physical property disturbances. In other words, a slowly rising differential pressure trend can support fouling risk modeling, while a short-term significant jump or rapid fluctuation in differential pressure change rate is more inclined to rheological disturbance discrimination. The two focus on the different meanings of the same physical quantity at different time scales.

[0153] As a safety precaution, if the rate of temperature change and the rate of pressure change give contradictory indications, such as a significant temperature change but almost no change in pressure, the system can first determine that it is an early stage of rheological disturbance and simultaneously observe subsequent windows to avoid premature cleaning and unnecessary pauses.

[0154] If both are close to the threshold and the sensor has been cleaned recently, the system can first perform a light temperature compensation, and then decide whether to perform additional cleaning based on the recovery situation; if the data from the temperature sensor or differential pressure sensor is not reliable enough, it is not advisable to force classification, but to keep the system frozen and request manual review.

[0155] For example, during night shift production, the outside temperature in the workshop drops rapidly, and the temperature of the circulating medium changes accordingly. The temperature of the waste liquid in the main pipeline drops continuously in a short period of time, and the pressure difference in the mixing section also fluctuates significantly. Although the temporal similarity is only in the middle zone and has not reached the high threshold of typical interference, the system combines the temperature change rate and the pressure difference change rate to determine that the current situation is more consistent with the rheological interference characteristics, so it prioritizes issuing a temperature compensation command.

[0156] In another scenario, if the temperature and pressure difference are relatively stable, but the conductivity reading still deviates slowly, the system will treat this as a more probable event of sensor contamination and trigger a cleaning command.

[0157] The purpose of this step is to further subdivide the sources of anomalies in the gray area, thereby enabling more targeted interventions, rather than adopting a single freezing strategy in all cases.

[0158] In a preferred embodiment of the present invention, the adjustment execution module is further configured to: maintain the current opening of the regulating valve and the current speed of the dosing pump in response to a freeze adjustment command; control the cleaning mechanism to start in response to a cleaning command, wherein the cleaning mechanism is an ultrasonic cleaning mechanism; control the temperature compensation mechanism to start in response to a temperature compensation command; and control the regulating valve opening and the speed of the dosing pump to adjust to the target value in response to a closed-loop adjustment command.

[0159] This embodiment provides an adjustment execution mechanism; specifically, the adjustment execution module takes four types of execution actions—holding, cleaning, temperature compensation, and closed-loop correction—according to different adjustment instructions issued by the coupling decision module.

[0160] Specifically, freezing adjustment does not cancel the control action, but rather, when it is confirmed that the current deviation is most likely caused by interference, it consciously maintains the current opening of the regulating valve and the speed of the dosing pump unchanged, so as to prevent the control system from amplifying the erroneous reading into process over-adjustment; the ultrasonic cleaning mechanism is suitable for situations where there are mycelial residues, organic deposits and slight scaling on the sensor surface;

[0161] The principle is to create high-frequency micro-vibration and local cavitation near the probe, causing the adhering layer to separate from the probe surface, thereby restoring the electrochemical contact capability of the sensing interface; the temperature compensation mechanism can use heat tracing, local heat exchange or algorithm linkage compensation to make the fluid temperature in the detection area and mixing section return to the stable working range more quickly, so as to reduce the viscosity increase, insufficient mixing and sensor response offset caused by temperature change.

[0162] The closed-loop control command is used in scenarios with actual concentration deviation. In this case, the control execution module adjusts the valve opening and the dosing pump speed synchronously or stepwise according to the target value to restore the target concentration of the finished liquid fertilizer.

[0163] It should be noted that in industrial settings, the adjustment rhythms of valves and dosing pumps should not be exactly the same; the main pipeline regulating valve has more of an impact on the mainstream flow rate and residence time, while the dosing pump directly affects the replenishment ratio; for slight deviations in actual operation, the dosing pump can be fine-tuned first; for deviations accompanied by changes in mixing conditions, the valve and pump can be adjusted in tandem to avoid simply increasing or decreasing the dosage while ignoring the mainstream flow state.

[0164] As a safety precaution, if the frozen state lasts for too long without recovery, a maintenance warning should be triggered to prevent the system from remaining in a seemingly stable state for a long time without addressing the root cause. If the sensor readings still do not show a recovery trend after ultrasonic cleaning, deeper problems such as probe aging, installation position misalignment, or large scale blockage should be considered. In this case, the automatic closed loop can be suspended and manual maintenance can be initiated.

[0165] If the temperature still cannot stabilize after temperature compensation is initiated, the system should not resume normal control simply because the compensation action has been performed. Instead, it should continue to determine whether freezing is still necessary based on real-time similarity and changes in physical quantities.

[0166] For example, in the aforementioned shift, the system first outputs a freeze adjustment command based on high similarity to maintain the current valve opening and dosing pump speed, preventing the liquid fertilizer concentration from being mistakenly increased; when the intermediate zone judgment further points to sensor contamination, the adjustment execution module activates the ultrasonic cleaning mechanism to briefly clean the adhering layer on the probe surface; in the case of rheological interference caused by a rapid temperature drop, the adjustment execution module activates the temperature compensation mechanism to bring the temperature of the detection section back to stability; after confirming that the original liquid component has truly decreased, the system then issues a closed-loop adjustment command to appropriately increase the dosing pump replenishment amount and adjust the valve accordingly to restore the target concentration;

[0167] The purpose of this step is to translate the judgment into physical actions that match the type of interference and process requirements, thereby achieving a closed loop from identification to handling.

[0168] In a preferred embodiment of the present invention, the data acquisition terminal is further used to: remove outliers and compensate for missing values ​​in the fluid dynamics time series data, concentration characterization time series data and actuator status data; write the processed data into a cache queue according to a uniform sampling period, and send it to the ideal benchmark reconstruction module, the interference simulation module and the coupling decision module.

[0169] This embodiment provides a data preprocessing mechanism for industrial noise environments. Specifically, before sending fluid dynamics time-series data, concentration characterization time-series data, and actuator status data to subsequent modules, the data acquisition terminal first performs outlier removal and missing value compensation, and writes the processed data into a cache queue according to a uniform sampling period.

[0170] In summary, although the previous layer scheme can improve the accuracy of judgment through multi-source data, if the original data itself has problems such as impulse noise, instantaneous communication interruption or inconsistent sampling period, the ideal benchmark reconstruction and residual coupling may still produce artifacts.

[0171] The following situations are particularly common in industrial settings: flow meters may experience spikes due to bubble interference; conductivity sensors may exhibit short-term transient pulse noise during cleaning switching; valve feedback may jump during mechanical vibration; or different data acquisition devices may experience time misalignment due to inconsistent refresh cycles.

[0172] The purpose of outlier removal is to exclude data points that clearly do not conform to the physical response law of the equipment from subsequent analysis. For example, if a sudden value appears in a sampling period that does not conform to the pump inertial characteristics, it should not be used directly to judge the true concentration change. Missing value compensation is used to handle short-term packet loss, communication delay or single sampling failure, so that subsequent modules can maintain continuous input and avoid residual trajectory distortion caused by data breakage.

[0173] A unified sampling period and buffer queue can organize data from different devices into the same cycle, ensuring that all types of data within the same analysis window correspond to the same process segment.

[0174] The cache queue here is not only a temporary data storage area, but also plays a role in balancing real-time performance and stability. For rheological interference identification that requires fast response, a shorter cache window can be used. For contamination identification that requires observing trends, continuous segments in the cache queue help to retain temporal evolution information.

[0175] As a safety precaution, if abnormal values ​​occur frequently and exceed the preset proportion, it is not possible to simply rely on elimination and compensation to maintain the system in the long term. Instead, the system should report to the superior system that there is a continuous fault in the sensor or communication channel. If critical data is missing continuously for more than the allowable duration, the system should stop outputting high-risk adjustment commands and only retain safety maintenance actions. If the time deviation between different sampling sources exceeds the correctable range, splicing should be rejected to avoid forming a process profile that appears complete but is actually distorted.

[0176] For example, after the production line enters the night shift, the flow meter experiences short-term fluctuations and spikes due to the switching of the circulating pump, and the conductivity probe also shows several abnormally high values ​​after ultrasonic cleaning; the data acquisition terminal removes these abnormal samples that do not conform to the actual process change inertia, continuously compensates for individual missing points, and then writes them into the cache queue according to a uniform sampling period.

[0177] In this way, the subsequent ideal reference reconstruction module and coupling decision module receive process data that is time-continuous and consistent in cycle time, and will not mistake the transient pulse noise spikes at the moment of cleaning for the real concentration transition.

[0178] The purpose of this step is to provide a stable and reliable data foundation for subsequent reconstruction, simulation, and decision-making, thereby enabling the automatic adjustment logic to be implemented in complex industrial settings.

[0179] In a preferred embodiment of the present invention, the system is applied to the process of preparing liquid fertilizer from thioerythromycin waste liquid; wherein, the preset fouling characteristics correspond to the mycelial residue adhesion characteristics and the organic suspended matter scaling characteristics; and the preset rheological interference characteristics correspond to the non-Newtonian fluid viscosity abrupt change characteristics.

[0180] This embodiment provides a parameter adaptation mechanism for a specific process object; specifically, this system is specifically applied to the process of preparing liquid fertilizer from thioerythromycin waste liquid. The preset fouling characteristics are established around the adhesion of mycelial residue and the scaling of organic suspended matter, and the preset rheological interference characteristics are established around the viscosity abrupt change of non-Newtonian fluid.

[0181] Specifically, this field-specific adaptation feature is a technological implementation of the aforementioned general control framework; thioerythromycin waste liquid is different from ordinary liquid fertilizer raw materials. It comes from a fermentation system and has a high content of residual mycelial fragments, metabolic byproducts and high molecular weight organic matter, so the sensor surface is more likely to form a flexible adhesion layer and a sticky film.

[0182] This fouling does not always cause sudden failures; more often, it gradually reduces the effective exchange between the probe and the liquid, leading to decreased conductivity and... Signal response passivation; in addition, this type of waste liquid often exhibits nonlinear viscosity changes when temperature, shear rate and solid content change, that is, the same flow rate change may lead to different degrees of mixing and sensing deviation at different temperature ranges.

[0183] Therefore, by specifically defining the fouling characteristics as mycelial residue adhesion characteristics and organic suspended matter scaling characteristics, and by specifically defining the rheological interference characteristics as non-Newtonian fluid viscosity abrupt change characteristics, the interference simulation module can be made closer to the real risk sources of the process.

[0184] Furthermore, under this process, the cleaning frequency, temperature compensation intensity, and threshold settings should be distinguished from ordinary fertilizer solution preparation systems; for example, mycelial adhesion is often more obvious after several hours of continuous operation, so the system can improve its sensitivity to slow drift without affecting production.

[0185] For workshops where ambient temperature changes significantly at night, the system should pay more attention to the rate of temperature change in order to identify false concentration deviations caused by sudden changes in viscosity in advance.

[0186] As a safety precaution, if the source of waste liquid changes, such as the fermentation stage being adjusted to result in a significant decrease in mycelial content or the addition of solid-liquid separation equipment in the pretreatment section, the existing fouling characteristics and rheological characteristics should be allowed to be recalibrated to avoid continuing to use old parameters and reducing identification accuracy.

[0187] If the production line is temporarily repurposed for other raw materials with similar properties but not erythromycin waste liquid, the system can retain its overall structure, but the risk of the same mycelial attachment and non-Newtonian viscosity mutation should be reassessed.

[0188] For example, in the continuous production scenario of this embodiment, a batch of waste liquid had a higher mycelial residue than usual because the upstream fermentation endpoint was later; after the system ran for several hours, the concentration reading gradually became dull, but the actual ratio in the main pipeline did not change synchronously.

[0189] Since the preset fouling characteristics take into account the gradual adhesion of mycelial residue, this phenomenon can be classified into the fouling channel for treatment earlier. For example, the workshop environment cools down quickly at night in winter, and the viscosity of waste liquid tends to change abruptly. The preset non-Newtonian fluid viscosity change characteristics enable the system to identify rheological interference and start temperature compensation in time, rather than misjudging it as insufficient actual concentration.

[0190] The purpose of this step is to accurately match the identification and handling logic of the control system with the physical property risks of the specific process object, thioerythromycin waste liquid, so as to achieve the engineering feasibility and stable applicability of the solution in the target industry.

[0191] 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 claims of this invention should still fall within the patent coverage of this invention.

Claims

1. A process-controlled liquid fertilizer concentration adjustment system for thioerythromycin waste liquid, characterized in that, This includes a data acquisition terminal installed in the target pipeline, an ideal reference reconstruction module, an interference simulation module, a coupling decision module, and an adjustment execution module; The target pipeline includes a main pipeline and connected dosing branches. The main pipeline is equipped with a regulating valve, a dosing point, a mixing section and a concentration detection point with a concentration sensor along the flow direction. The dosing branches are equipped with dosing pumps. The regulating execution module is connected to the regulating valve, the dosing pump, the cleaning mechanism and the temperature compensation mechanism. The data acquisition terminal acquires fluid dynamics time-series data, concentration characterization time-series data, and actuator status data; the ideal reference reconstruction module reconstructs ideal concentration data based on the fluid dynamics time-series data and actuator status data. The interference simulation module injects interference into the ideal concentration data based on preset fouling characteristics, rheological interference characteristics, and the fluid dynamics time series data to generate simulated concentration data; The coupling decision module generates actual concentration estimation data based on the concentration characterization time series data, calculates the actual residual data between the actual data and the ideal concentration data, and the theoretical residual data between the simulated concentration data and the ideal concentration data, performs dynamic time warping calculation on the two residual data to generate time series similarity data, and generates adjustment instructions including closed-loop adjustment, freeze adjustment, cleaning and / or temperature compensation instructions accordingly. The adjustment execution module responds to the adjustment command and controls the corresponding mechanism.

2. The thioerythromycin waste liquid preparation liquid fertilizer concentration adjustment system based on process control according to claim 1, characterized in that, The fluid dynamics time-series data includes the real-time flow rate of the main pipeline, the pressure difference between the front and rear ends of the mixing pipe section, and the fluid temperature; the concentration characterization time-series data includes the online conductivity sampling value and pH sampling value at the concentration detection point; the actuator status data includes the dosing pump speed feedback value and the regulating valve opening feedback value.

3. The thioerythromycin waste liquid preparation liquid fertilizer concentration adjustment system based on process control according to claim 1, characterized in that, The ideal reference reconstruction module is also used to: perform time synchronization processing on the fluid dynamics timing data and the actuator state data; Based on the synchronized inflow rate, dosing rate, and preset concentration parameters, the ideal concentration data corresponding to the target time is calculated; wherein, the preset concentration parameters include the effective component mass concentration base value of the erythromycin waste liquid to be treated in the main pipeline and the effective component mass concentration base value of the replenishment liquid in the dosing branch.

4. The thioerythromycin waste liquid preparation liquid fertilizer concentration adjustment system according to claim 1, characterized in that, The interference simulation module is also used to: perform time-series drift processing on the ideal concentration data based on the preset fouling attenuation rules and the pressure difference change rate at the front and rear ends of the mixed pipe section, and generate fouling simulation data; Based on the preset rheological damping rules, the fluid temperature, and the real-time flow rate of the main pipeline, transient perturbation processing is performed on the ideal concentration data to generate rheological simulation data. The pollution simulation data and the rheological simulation data are superimposed to generate the simulated concentration data; The preset contamination characteristics and the preset rheological interference characteristics include at least the interference amplitude, time constant, and operating window.

5. The thioerythromycin waste liquid preparation liquid fertilizer concentration adjustment system based on process control according to claim 1, characterized in that, The coupling decision module is also used to: convert the online conductivity sampling value and the pH sampling value into the actual concentration estimation data based on a preset calibration model; Within a preset time window, the dynamic time-warped distance between the actual residual data and the theoretical residual data is calculated. The dynamic time warp distance is normalized to generate the time series similarity data, wherein the larger the value of the time series similarity data, the more similar the actual residual data is to the theoretical residual data.

6. The thioerythromycin waste liquid preparation liquid fertilizer concentration adjustment system based on process control according to claim 5, characterized in that, The coupling decision module is further configured to: determine the current state as an interference state in response to the temporal similarity data being greater than or equal to a preset second similarity threshold, and generate a freeze adjustment command; In response to the time-series similarity data being less than or equal to a preset first similarity threshold, the current state is determined to be a true concentration deviation, and a closed-loop adjustment command is generated; wherein, the second similarity threshold is greater than the first similarity threshold.

7. The thioerythromycin waste liquid preparation liquid fertilizer concentration adjustment system based on process control according to claim 6, characterized in that, The coupling decision module is further configured to: in response to the temporal similarity data being located between the preset first similarity threshold and the preset second similarity threshold, obtain the current temperature change rate and the current pressure difference change rate; In response to the temperature change rate being higher than a preset temperature change rate threshold or the pressure difference change rate being higher than a preset pressure difference change rate threshold, the disturbance state is determined to be a rheological disturbance, and a temperature compensation command is generated in the adjustment command. In response to the temperature change rate not being higher than the preset temperature change rate threshold and the pressure difference change rate not being higher than the preset pressure difference change rate threshold, the interference state is determined to be sensor contamination interference, and a cleaning command is generated in the adjustment command.

8. The thioerythromycin waste liquid preparation liquid fertilizer concentration adjustment system based on process control according to claim 7, characterized in that, The adjustment execution module is also used to: in response to the freeze adjustment command, maintain the current adjustment valve opening and the current dosing pump speed; In response to the cleaning command, the cleaning mechanism is started, wherein the cleaning mechanism is an ultrasonic cleaning mechanism; in response to the temperature compensation command, the temperature compensation mechanism is started; in response to the closed-loop adjustment command, the opening of the regulating valve and the speed of the dosing pump are adjusted to the target value.

9. The thioerythromycin waste liquid preparation liquid fertilizer concentration adjustment system according to claim 1, characterized in that, The data acquisition terminal is also used to: remove outliers and compensate for missing values ​​in the fluid dynamics time series data, the concentration characterization time series data, and the actuator status data; The processed data is written into the cache queue according to a uniform sampling period and sent to the ideal benchmark reconstruction module, the interference simulation module and the coupling decision module.

10. The thioerythromycin waste liquid preparation liquid fertilizer concentration adjustment system according to claim 1, characterized in that, The system is applied to the process of preparing liquid fertilizer from thioerythromycin waste liquid; wherein, the preset fouling characteristics correspond to the mycelial residue adhesion characteristics and the organic suspended matter scaling characteristics; the preset rheological interference characteristics correspond to the non-Newtonian fluid viscosity abrupt change characteristics.

Images