A method and system for analyzing data of industrial wastewater treatment containing phosphorus ammonium

Abstract

The application provides a phosphorus-ammonium-containing industrial wastewater treatment data analysis method and system, which comprises the following steps: an electrochemical detection loop is arranged in a crystallization reactor, and a potential and current noise signal is collected. The ratio of the standard deviation of the potential and current noise is used to determine the crystallization starting time. In response to the crystallization starting, the physical morphology of the particles is evaluated by calculating the kurtosis value of the current noise, and the type of the particles is determined by combining the kurtosis value and the electrochemical noise change caused by the periodic change of the flow rate. The operation parameters of the reactor are adjusted based on the diagnosis result. Finally, when the kurtosis value is stable at a high level and the electrochemical noise index is stable, it is determined that the crystal growth is completed. The application solves the technical problem that the physical morphology and type of the growing particles in the reactor cannot be distinguished online by the traditional macroscopic parameter monitoring by analyzing the high-order statistical characteristics of the electrochemical noise and the response of the high-order statistical characteristics to the flow field change.

Description

Technical Field

[0001] This invention belongs to the field of data analysis technology, specifically relating to a data analysis method and system for treating industrial wastewater containing ammonium phosphate. Background Technology

[0002] Resource recovery of industrial wastewater containing ammonium phosphate typically involves inducing magnesium ammonium phosphate crystallization through chemical precipitation to recover phosphorus and nitrogen resources. The core of chemical precipitation lies in the precise control of operating parameters within the crystallization reactor (such as pH, dosing rate, and stirring intensity) to obtain a target crystalline product with uniform particle size, high purity, and easy separation. Currently, monitoring and control of this chemical reaction process mainly relies on online measurement of macroscopic parameters within the reactor, such as solution pH, temperature, conductivity, and the concentration of key ions. However, these macroscopic phase parameters reflect the average state of the liquid phase in the entire reaction system and cannot perceive or analyze the microscopic information of the solid particles themselves, which truly determine the quality of the final product. Specifically, even when maintaining stable pH and ion concentrations, it is not possible to directly determine whether the particles forming in the crystallization reactor are dense crystals or amorphous gel-like precipitates formed due to improper control of local supersaturation. These different physical forms and chemical types of particles exhibit drastically different performance characteristics in subsequent separation, dehydration, and as slow-release fertilizer applications. Traditional analytical control methods are difficult to use for targeted and precise regulation of crystallization reactors. Product quality can only be evaluated through offline finished product inspection after the reaction is completed. Therefore, identifying the physical morphology and type of particles growing in the reactor is a current technical challenge for improving the control level of ammonium phosphate crystallization process. Summary of the Invention

[0003] This invention provides a data analysis method and system for treating industrial wastewater containing ammonium phosphate, in order to solve the above-mentioned technical problems.

[0004] In a first aspect, the present invention provides a method for analyzing data from the treatment of industrial wastewater containing ammonium phosphate, the method comprising the following steps: A detection circuit consisting of two homogeneous inert electrodes and a reference electrode is set up in a crystallization reactor for treating industrial wastewater containing ammonium phosphate. A zero-resistance galvanometer is used to collect the potential noise signal and current noise signal of the detection circuit, and the polarization error in the potential noise signal and current noise signal is filtered out. The ratio of the standard deviation of the potential noise signal and the current noise signal after the filtering operation is used as the electrochemical noise index. By monitoring the electrochemical noise index, the crystallization initiation time when the reaction system in the crystallization reactor changes from a pure liquid phase to a solid-liquid mixed phase is determined. In response to the crystallization initiation moment, the physical morphology of particles impacting the electrode surface in the crystallization reactor is assessed by calculating the kurtosis value of the current noise signal; During the monitoring process, the flow field velocity in the crystallization reactor was periodically changed. The changes in the kurtosis value and electrochemical noise of the current noise signal before and after the flow field velocity change were combined, and the current particulate matter type was determined based on the fluid shear excitation principle. Adjust the dosing rate and reflux ratio parameters of the crystallization reactor based on the physical morphology and type of the particulate matter. The kurtosis value and electrochemical noise index are continuously monitored. The crystal growth in the crystallization reactor is determined to be complete only when the kurtosis value remains in the preset high range and no longer rises, and the electrochemical noise index is in the preset stable range.

[0005] Optionally, the step of acquiring the potential noise signal and current noise signal of the detection circuit using a zero-resistance galvanometer and filtering out the polarization error in the potential noise signal and current noise signal includes the following steps: The spontaneous current fluctuations between two homogeneous inert electrodes are collected using a zero-resistance galvanometer while maintaining zero potential difference between the two electrodes as current noise signals. The potential fluctuations of one of the homogeneous inert electrodes relative to the reference electrode are recorded synchronously as potential noise signals; The polarization error in potential noise signal and current noise signal is filtered out by using the time-series structure fragmentation and reconstruction method.

[0006] Optionally, the method of filtering polarization errors in potential noise signals and current noise signals using the time-series structure fragmentation and reconstruction method includes the following steps: Obtain the time index of the potential noise signal and the current noise signal; Based on the time index and using the Monte Carlo random algorithm, all discrete sampling points in the potential noise signal and current noise signal are randomly shuffled and rearranged to obtain potential noise signal and current noise signal with polarization error filtered out, which has monotonic evolution characteristics.

[0007] Optionally, the step of using the ratio of the standard deviations of the potential noise signal and the current noise signal after the filtering operation as an electrochemical noise index, and determining the crystallization initiation time when the reaction system in the crystallization reactor transforms from a pure liquid phase to a solid-liquid mixed phase by monitoring the electrochemical noise index, includes the following steps: Within a preset sliding window, the root mean square value of the potential noise signal after the filtering operation is calculated as the standard deviation of the potential noise, and the root mean square value of the current noise signal after the filtering operation is calculated as the standard deviation of the current noise. The instantaneous electrochemical noise index is obtained by dividing the standard deviation of the potential noise by the standard deviation of the current noise. The average value of the electrochemical noise index was recorded during the untreated stage of the reaction in the crystallization reactor as the pure liquid phase reference impedance. Calculate the deviation of the electrochemical noise index from the pure liquid phase reference impedance at the current moment. When the deviation exceeds the preset order of magnitude threshold for n consecutive sliding windows, it is confirmed that the double layer of the solid-liquid interface has been reconstructed. The current moment is determined to be the crystallization start moment when the reaction system in the crystallization reactor changes from the pure liquid phase to the solid-liquid mixed phase, n≥3.

[0008] Optionally, the step of evaluating the physical morphology of particles impacting the electrode surface in the crystallization reactor by calculating the kurtosis value of the current noise signal in response to the crystallization initiation time includes the following steps: Extract the current noise signal segment after the crystallization initiation time, and calculate the arithmetic mean and standard deviation of all discrete sampling points within the signal segment; For each discrete sampling point within a signal segment, the dimensionless kurtosis value of the signal segment is calculated by combining the arithmetic mean and standard deviation of the discrete sampling points. If the kurtosis value is less than the preset first morphology threshold, the impact event of particles hitting the electrode surface in the crystallization reactor is determined to follow a Gaussian distribution, and the physical morphology of the particles is evaluated as microcrystalline mode. If the kurtosis value is greater than the preset second morphology threshold, it is determined that the impact events of particles hitting the electrode surface in the crystallization reactor exhibit a long-tailed distribution, and the physical morphology of the particles is evaluated as an effective growth mode. The second morphology threshold is greater than the first morphology threshold and greater than 3. If the kurtosis value is greater than or equal to the first morphology threshold and less than or equal to the second morphology threshold, then the impact event of particles hitting the electrode surface in the crystallization reactor is determined to be in the transitional stage of non-Gaussian distribution, and the physical morphology of the particles is evaluated as a transitional growth mode.

[0009] Optionally, the step of periodically changing the flow field velocity in the crystallization reactor during the monitoring process, combining the changes in kurtosis and electrochemical noise of the current noise signal before and after the flow field velocity change, and determining the current particulate matter type based on the fluid shear excitation principle includes the following steps: Increase the preset base flow rate of the crystallization reactor to the preset disturbance flow rate; Simultaneously calculate the reference kurtosis value of the current noise signal at the base flow velocity and the disturbance kurtosis value of the current noise signal at the disturbance flow velocity, as well as the current noise standard deviation of the current noise signal at the reference flow velocity and the current noise standard deviation of the current noise signal at the disturbance flow velocity. By combining the baseline kurtosis value, the standard deviation of the current noise of the perturbation kurtosis value, and the standard deviation of the current noise, and based on the fluid shear excitation principle, the excess kurtosis energy flux ratio, which is used to quantify the competitive advantage of the growth rate of particulate impact characteristics relative to the growth rate of background turbulent energy, is calculated. If the excess kurtosis energy flux ratio is greater than 1, then the current particle type is determined to be dense crystal. If the excess kurtosis energy flow ratio is less than or equal to 1, then the current particulate matter type is determined to be either bubbles or soft flocs.

[0010] Optionally, after determining the current particulate matter type, the following steps may also be included: If the particulate matter is dense crystal, the kurtosis value of the current current noise signal is marked as an effective control variable, and the effective control variable is used as a reference parameter for adjusting the dosing rate and reflux ratio parameters. If the particulate matter is bubbles or soft flocs, the defoaming program of the crystallization reactor is started, and the flow field velocity is changed again after the bubble interference source is eliminated until the excess kurtosis energy-fluid ratio is greater than 1, and the update of the effective control variables is restored.

[0011] Optionally, adjusting the dosing rate and reflux ratio parameters of the crystallization reactor based on the physical morphology and type of the particles includes the following steps: When the physical morphology of the particulate matter is microcrystalline, the dosing pump of the crystallization reactor is controlled to reduce the dosing rate according to the exponential decay law to reduce local supersaturation, while the seed reflux pump of the crystallization reactor is controlled to operate at maximum power. When the physical morphology of the particulate matter is in the transitional growth mode, keep the current dosing rate and reflux ratio parameters of the crystallization reactor unchanged, and start the stirring device of the crystallization reactor to promote crystal suspension and mass transfer until the kurtosis value of the signal segment naturally evolves to above the second morphology threshold. When the physical form of the particles is in an effective growth mode, control the dosing pump to maintain the current dosing rate, and at the same time control the speed of the stirring device to match the critical flow rate of the particles to prevent the particles from settling and accumulating at the bottom.

[0012] In a second aspect, the present invention also provides a data analysis system for treating industrial wastewater containing ammonium phosphate, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the data analysis method for treating industrial wastewater containing ammonium phosphate as described in any one of the first aspects.

[0013] Thirdly, the present invention also provides a computer-readable storage medium storing instructions, characterized in that, when executed by a processor, the instructions cause the processor to be configured to perform the data analysis method for treating industrial wastewater containing ammonium phosphate according to any one of the first aspects.

[0014] The beneficial effects of this invention are: This invention extracts information about the microscopic dynamics of the crystallization process from raw potential and current noise signals that far exceeds what traditional macroscopic parameters can provide. Specifically, this invention does not treat noise as useless interference, but rather as an information carrier. By calculating the standard deviation ratio, it captures the crystallization initiation moment of the transition from a pure liquid phase to a solid-liquid mixed phase. Furthermore, it applies the kurtosis value of the current noise signal as a key indicator, which is highly sensitive to instantaneous impact events between particles and the electrode surface, thereby achieving quantitative characterization of the physical morphology of particles during growth. Combined with dynamic testing based on the fluid shear excitation principle, this method further gains the ability to distinguish different particle types. Finally, by directly using these physically significant microscopic particle characteristics for feedback control and combining the stable states of these characteristics, the crystallization endpoint is determined. Attached Figure Description

[0015] Figure 1 This is a schematic flowchart of a data analysis method for treating industrial wastewater containing ammonium phosphate in one embodiment of this application.

[0016] Figure 2 This is a schematic diagram of the flow topology of a data analysis method for treating industrial wastewater containing ammonium phosphate in one embodiment of this application. Detailed Implementation

[0017] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0018] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0019] Figure 1 This is a flowchart illustrating a data analysis method for treating industrial wastewater containing ammonium phosphate in one embodiment. It should be understood that, although... Figure 1The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order in which these steps are executed, and they can be performed in other orders. Figure 1 At least some steps in the process may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but may be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but may be executed in turn or alternately with other steps or at least a portion of the sub-steps or stages of other steps. (See reference...) Figure 1 and Figure 2 The data analysis method for treating industrial wastewater containing ammonium phosphate disclosed in this invention specifically includes the following steps: S101. A detection circuit comprising two homogeneous inert electrodes and a reference electrode is provided in a crystallization reactor for treating industrial wastewater containing ammonium phosphate.

[0020] The primary step in constructing a treatment system for phosphate-ammonium industrial wastewater is establishing a high-precision electrochemical monitoring environment. The internal environment of the crystallization reactor is complex, with high concentrations of phosphate and ammonium ions in a highly unstable thermodynamic state under the influence of the flow field, making it difficult for conventional monitoring methods to capture microscopic phase transitions. When setting up the detection loop in the core crystallization region of the reactor, materials with highly stable electrochemical properties must be selected. Platinum or boron-doped diamond can be used as the substrate for the homogeneous inert electrode to ensure that the electrode surface does not undergo a Faraday reaction in the highly corrosive wastewater, serving only as an inert interface for electron conduction. The two homogeneous inert electrodes must maintain completely identical geometric dimensions and surface roughness, and be fixed in parallel opposite positions in areas of intense fluid scouring to ensure a consistent hydrodynamic environment, thereby eliminating common-mode interference caused by uneven flow field distribution. The reference electrode should be a silver / silver chloride electrode or a saturated calomel electrode with a microporous ceramic liquid network, placed in a relatively stable bypass or protected area to provide a stable absolute potential reference for the homogeneous inert electrode. This three-electrode system is not constructed to apply an external polarization potential, but rather to form a highly sensitive microscopic electric double-layer detection array. By precisely controlling the electrode spacing and orientation, the detection circuit can keenly sense the perturbations in the electric double-layer capacitance caused by ion clusters and the formation of tiny crystal nuclei in the solution.

[0021] S102. Use a zero-resistance galvanometer to collect the potential noise signal and current noise signal of the detection circuit, and filter out the polarization error in the potential noise signal and current noise signal.

[0022] In the data acquisition and preprocessing stage, a zero-resistance galvanometer is used as the signal pickup device to measure the spontaneous coupling current generated between two homogeneous inert electrodes in real time to balance the tiny surface potential difference, without inputting any external energy into the system. This microampere-level current fluctuation directly reflects the transient changes in the electric double layer on the electrode surface. Simultaneously, a high-impedance voltage follower synchronously records the open-circuit potential fluctuation of any inert electrode relative to the reference electrode. The raw signal often contains a DC trend term, i.e., polarization error, caused by the slow growth of the oxide film on the electrode surface or the drift of the bulk solution concentration. This extremely low-frequency drift can mask the high-frequency crystal impact characteristics. Filtering out polarization error can completely preserve the random fluctuation characteristics contained in the signal, ensuring that the data source for subsequent statistical analysis is modulated only by the electrochemical processes and physical impact events inside the reactor, eliminating the interference of background environmental drift on the judgment of the microcrystalline process.

[0023] S103. The ratio of the standard deviation of the potential noise signal and the current noise signal after the filtering operation is used as the electrochemical noise index. The crystallization initiation time when the reaction system in the crystallization reactor changes from a pure liquid phase to a solid-liquid mixed phase is determined by monitoring the electrochemical noise index.

[0024] The core of determining the crystallization initiation time lies in the statistical analysis of the pure electrochemical noise signal, capturing the phase transition critical point by monitoring the relative intensity changes of potential and current fluctuations. The electrochemical noise index is defined as the ratio of the standard deviation of the potential noise signal (after filtering out polarization errors) to the standard deviation of the current noise signal. In the initial stage of the reaction, the system is in a pure liquid phase, and the noise on the electrode surface mainly originates from the thermal motion and diffusion of ions. At this time, potential fluctuations are relatively significant while current fluctuations are weak, resulting in the electrochemical noise index remaining at a high baseline level. As the induction period ends, tiny solid-phase nuclei begin to form explosively in the solution. These charged particles penetrate the diffusion layer on the electrode surface and disturb the electric double-layer structure. The physical interference of solid particles on the electrode interface causes a sharp increase in the standard deviation of the current noise signal, while the potential signal is relatively less affected, resulting in a significant step decrease in the electrochemical noise index. This abrupt change in the index precisely corresponds to the physical moment of the system's transition from a homogeneous solution to a solid-liquid two-phase mixture on the time axis. The ratio is calculated in real time by sliding window and compared with the pure liquid phase reference value. Once the index is detected to fall to the preset threshold range and show continuous low-level oscillation, it can be confirmed that the solid-liquid interface has been reconstructed and the reactor has officially entered the crystallization growth period.

[0025] S104. In response to the crystallization initiation moment, the physical morphology of particles impacting the electrode surface in the crystallization reactor is evaluated by calculating the kurtosis value of the current noise signal.

[0026] Once the crystallization stage is confirmed, the statistical characteristics of the current noise signal shift from measuring the overall fluctuation intensity to evaluating the waveform distribution morphology. Kurtosis is a key dimensionless parameter describing the steepness of the signal probability density distribution curve. When only tiny crystal nuclei exist in the crystallization reactor, their impact on the electrodes is weak and frequent, the signal amplitude distribution is close to a normal distribution, and the kurtosis is close to 3. As the crystal grows, the particle size increases, and the impact of large crystal particles on the electrode surface generates pulsed current spikes that significantly deviate from the average value, resulting in a thick-tailed characteristic in the signal distribution curve and a significant increase in kurtosis. Calculating the kurtosis of a signal segment of duration *t* requires using the arithmetic mean and standard deviation of the discrete points within that segment. By quantifying the kurtosis, the physical morphology of the particles impacting the electrode surface can be inferred: low kurtosis corresponds to an isotropic microcrystalline mode dominated by Brownian motion; while high kurtosis corresponds to an effective crystal growth mode dominated by inertia and with specific geometric angles. Kurtosis values ​​in the intermediate transition range indicate that the system is in a transitional stage of small particle agglomeration or secondary growth of crystal nuclei. This morphological assessment method based on higher-order statistics can achieve online soft measurement of the growth state of particulate matter in the dark and turbid environment inside the reactor without the need for an optical window.

[0027] S105. During the monitoring process, the flow field velocity in the crystallization reactor is periodically changed. The changes in the kurtosis value of the current noise signal and the changes in electrochemical noise before and after the change in flow field velocity are combined, and the current particulate matter type is determined based on the fluid shear excitation principle.

[0028] Based on the principle of fluid shear excitation, hard crystalline particles gain higher kinetic energy with increasing flow velocity, significantly enhancing the instantaneous pressure and double-layer compression effect upon impacting the electrode surface. This results in more pronounced spikes in the current signal and a positive correlation increase in kurtosis. Conversely, bubbles or soft flocs are prone to deformation, breakage, or removal from the electrode surface under high-speed flow shear, potentially weakening or smoothing their impact signals and leading to insignificant or even decreased kurtosis changes. By increasing the flow velocity within the reactor from the baseline velocity to the disturbed velocity, the kurtosis and energy increments of the current noise signal before and after the velocity change were simultaneously calculated. An excess kurtosis energy-fluidity ratio parameter was constructed to quantify the relative gain of kurtosis changes. A value significantly greater than 1 indicates that the injected flow energy is effectively converted into a non-Gaussian enhancement of the impact signal, confirming the particles as dense crystals. Conversely, a low ratio suggests that energy is mainly consumed in particle deformation or fluid turbulence, inferring that the current particles are primarily bubbles or amorphous flocs.

[0029] S106. Adjust the dosing rate and reflux ratio parameters of the crystallization reactor based on the physical morphology and type of the particulate matter.

[0030] Based on accurate identification of the physical morphology and material type of the particles, the control system performs refined closed-loop adjustment of the reactor operating parameters to maintain optimal crystal growth kinetics. Specifically, if the evaluation results show that the current mode is mainly microcrystalline, it indicates that the nucleation rate in the system is too fast or the supersaturation is too high. In this case, the dosing rate should be reduced immediately to suppress explosive nucleation and prevent the formation of fine, difficult-to-separate crystals. At the same time, the reflux ratio should be increased to bring the microcrystalline back to the reaction zone as seed crystals for induction of growth. If it is determined to be a transitional growth mode, it means that the crystals are in the growth stage. The current operating conditions should be maintained and the stirring device should be turned on to enhance mass transfer efficiency and promote the diffusion of solute molecules to the crystal surface. If it is confirmed to be an effective growth mode and the particles are dense crystals, it indicates that large crystal particles are forming stably. In this case, the dosing rate should be locked to maintain metastable supersaturation, and the stirring speed should be precisely adjusted to the critical suspension velocity of the particles to prevent large particles from settling and clogging, and to avoid excessive shear force that could cause crystal breakage. For bubbles or floc interference, defoaming or sludge removal procedures are triggered. This control strategy based on micromorphological feedback transforms traditional empirical and extensive regulation into adaptive regulation based on particle growth state, ensuring that the reagent dosage and hydraulic conditions are always matched with the current crystallization stage, maximizing ammonium phosphate recovery rate and crystal quality.

[0031] S107. Continuously monitor the kurtosis value and electrochemical noise index. If and only if the kurtosis value remains in the preset high range and no longer rises, and the electrochemical noise index is in the preset stable range, the crystal growth in the crystallization reactor is determined to be complete.

[0032] In this process, the endpoint determination of the crystallization process no longer relies on a fixed hydraulic residence time, but rather on the natural decay characteristics of crystal growth kinetics. As the reaction proceeds, the supersaturation in the solution gradually decreases, the solute concentration approaches the solubility equilibrium point, the crystal growth rate slows down, and the particle size reaches the limit of hydrodynamic constraints. At this point, even if the flow field remains constant, the signal characteristics generated by large particles impacting the electrode will tend to stabilize, and the kurtosis value of the current noise signal will remain within a relatively high range with slight fluctuations, no longer showing a monotonically increasing trend, indicating that the particle morphology no longer undergoes significant changes. Simultaneously, the electrochemical noise index, as a comprehensive parameter measuring the solid-liquid interface state, will also enter a relatively stable numerical range, meaning that the solid phase surface area and the liquid phase ion concentration have reached a dynamic equilibrium. Only when these two independent parameters simultaneously satisfy the stability criterion will the system determine that the crystal growth process of this batch or stage is completely complete. This dual verification mechanism effectively avoids false endpoint misjudgment caused by local flow field fluctuations, ensures the uniformity and maturity of the output crystal particle size distribution, provides the best material conditions for subsequent solid-liquid separation processes, and realizes precise control of the entire process of resource-based treatment of ammonium phosphate wastewater.

[0033] In one embodiment, the process of acquiring the potential noise signal and current noise signal of the detection circuit using a zero-resistance galvanometer and filtering out polarization errors in the potential noise signal and current noise signal includes the following steps: The spontaneous current fluctuations between two homogeneous inert electrodes are collected using a zero-resistance galvanometer while maintaining zero potential difference between the two electrodes as current noise signals. The potential fluctuations of one of the homogeneous inert electrodes relative to the reference electrode are recorded synchronously as potential noise signals; The polarization error in potential noise signal and current noise signal is filtered out by using the time-series structure fragmentation and reconstruction method.

[0034] In this embodiment, a zero-resistance galvanometer is selected as the signal acquisition device. The internal circuit design of the ZRA cleverly utilizes the virtual ground principle of the operational amplifier, dynamically adjusting the output voltage through a negative feedback mechanism. This forces the potential difference between the two homogeneous inert electrodes (such as platinum electrodes) connected to the two input terminals to remain at an extremely low level (theoretically zero volts). While maintaining this zero potential difference, the instantaneous unbalanced charge generated on the surfaces of the two electrodes due to local microenvironmental differences (such as crystal particle impacts or ion concentration perturbations) must be transferred through the external circuit. This spontaneously generated electron flow to balance the potential constitutes the current noise signal, denoted as . Meanwhile, in order to obtain thermodynamic state information of the electrode interface, a high-impedance voltage follower circuit is used at extremely high input impedance (typically greater than 1000 kJ / m²). Simultaneously measuring the open-circuit potential fluctuation of one of the homogeneous inert electrodes relative to a stable reference electrode (such as an Ag / AgCl electrode) is called the potential noise signal, denoted as . This synchronous acquisition mode not only avoids the induction or suppression of crystallization nucleation by the applied polarization current, but also ensures strict alignment of the current signal and the potential signal in the time dimension. Let the sampling time interval be... At any moment The collected discrete data point pairs are This data truly reflects the natural evolution trajectory of the double electric layer at the solid-liquid interface inside the reactor over time, without external energy interference.

[0035] Conventional low-frequency drift typically originates from the slow growth of the oxide film on the electrode surface or the monotonic change in the bulk electrolyte concentration. These processes exhibit strong sequential correlation over time, i.e., a memory effect. Therefore, the trend term can be stripped away by breaking this temporal continuity, retaining only the noise component reflecting random fluctuations. The specific implementation process first involves extracting the original potential noise signal. and current noise signal Time index set of all N discrete sampling points The Monte Carlo algorithm is driven by a computer-generated pseudo-random number sequence to generate a random index permutation of the same length as the original set. ,in Let T be a unique integer randomly selected from T. Then, the amplitude values ​​of the original signal are forcibly rearranged based on this random index to construct the recombined signal sequence. For any i-th data point, its magnitude is defined as... ,in This represents the unprocessed amplitude of the original signal. Through this thorough disordering operation, the slow upward or downward geometric trend formed by the accumulation of time in the original signal is physically broken up and transformed into a disordered background distribution. This eliminates the interference of polarization error on the calculation of signal statistical characteristics (such as variance and skewness) without setting a cutoff frequency, ensuring that the subsequent crystallization state assessment based on amplitude distribution is entirely based on real random fluctuation events.

[0036] In one embodiment, filtering polarization errors in potential noise signals and current noise signals using a time-series structure fragmentation and reconstruction method includes the following steps: Obtain the time index of the potential noise signal and the current noise signal; Based on the time index and using the Monte Carlo random algorithm, all discrete sampling points in the potential noise signal and current noise signal are randomly shuffled and rearranged to obtain potential noise signal and current noise signal with polarization error filtered out, which has monotonic evolution characteristics.

[0037] In this embodiment, obtaining the time index is a fundamental step in the digital processing of electrochemical noise data, transforming continuously changing analog physical quantities into discrete digital sequences that can be processed by computer algorithms. Since the electrochemical process within the crystallization reactor is a dynamic evolution occurring in a continuous time domain, while the subsequent Monte Carlo random algorithm requires operation based on discrete coordinate points, a strict temporal mapping relationship must first be established. Specifically, a high-precision analog-to-digital converter is used to sample the analog voltage signals from the zero-resistance galvanometer and voltage follower at equal intervals, and each data point is simultaneously assigned a unique and monotonically increasing integer identifier, i.e., the time index. The time index records the historical trajectory of the evolution of the double-layer state on the electrode surface. For polarization errors containing long-term drift trends, their monotonic evolution characteristic depends precisely on the strict temporal sequence.

[0038] In a preferred embodiment, the sampling frequency of the data acquisition system is set to... The time interval between two adjacent sampling points is After the acquisition command is initiated, potential noise and current noise signals are continuously captured within the total sampling duration. If the total number of discrete sampling points acquired is defined as M, then a time index sequence consisting of consecutive integers is constructed. This sequence can be represented as Any index value k precisely corresponds to a physical time. The original acquired potential signal amplitude sequence and current signal amplitude sequence A one-to-one mapping and binding is performed with this time index sequence to form a data packet with a strict timestamp.

[0039] Polarization error typically originates from the growth of passivation films on electrode surfaces or the macroscopic depletion of reactant concentrations. These physical processes exhibit a significant memory effect, manifesting as a strong temporal correlation between data points, exhibiting a smooth trend of monotonically increasing or decreasing over time (structural characteristics). Conversely, real electrochemical noise caused by crystal nucleation or shedding is a random, independent event. By completely randomizing the arrangement of all discrete sampling points using the Monte Carlo algorithm, the original temporal adjacency between data points is essentially forcibly disrupted at the physical level. This operation tears apart and discretizes the originally continuous and smooth polarization drift curve, causing it to appear as chaotic background fluctuations in the new disordered sequence, thus completely eliminating its monotonically evolving geometric characteristics.

[0040] In practice, it is based on time index sequences. A random permutation index sequence of length M is generated using Monte Carlo methods (such as the Fisher-Yates shuffle algorithm). This random sequence It contains all integers from 1 to M, but their order follows a uniformly distributed randomness. This random sequence is used as a new address pointer to reconstruct the original signal. For the rearranged potential noise signal... The data point at position j is taken from the original signal. The magnitude at each position, i.e., the mathematical relationship described as follows: Similarly, performing the same operation on the current noise signal yields... After this global random reorganization, any function f(t) in the original signal that depends on time t is transformed into a function f(rand) that depends on random variables. The autocorrelation function of the signal decays rapidly to zero at non-zero delay, and the trend attribute of polarization error is completely bleached. The output signal is the filtered signal that only contains amplitude statistical characteristics and has no trend structure.

[0041] In one embodiment, the ratio of the standard deviations of the potential noise signal and the current noise signal after the filtering operation is used as the electrochemical noise index. Determining the crystallization initiation time when the reaction system in the crystallization reactor transforms from a pure liquid phase to a solid-liquid mixed phase by monitoring the electrochemical noise index includes the following steps: Within a preset sliding window, the root mean square value of the potential noise signal after the filtering operation is calculated as the standard deviation of the potential noise, and the root mean square value of the current noise signal after the filtering operation is calculated as the standard deviation of the current noise. The instantaneous electrochemical noise index is obtained by dividing the standard deviation of the potential noise by the standard deviation of the current noise. The average value of the electrochemical noise index was recorded during the untreated stage of the reaction in the crystallization reactor as the pure liquid phase reference impedance. Calculate the deviation of the electrochemical noise index from the pure liquid phase reference impedance at the current moment. When the deviation exceeds the preset order of magnitude threshold for n consecutive sliding windows, it is confirmed that the double layer of the solid-liquid interface has been reconstructed. The current moment is determined to be the crystallization start moment when the reaction system in the crystallization reactor changes from the pure liquid phase to the solid-liquid mixed phase, n≥3.

[0042] In this embodiment, since the crystallization process is a dynamically evolving nonlinear system, global statistics cannot reflect instantaneous microscopic changes. Therefore, a sliding window mechanism with short-time memory is necessary. In practice, a sliding window with a fixed time length is set, moving forward on the time axis with sampling points as steps to capture local segments of the potential and current signals. For the signal after filtering out the DC trend term in the preceding steps, its mean theoretically approaches zero. At this point, the root mean square value of the signal is mathematically equivalent to the standard deviation, accurately characterizing the energy intensity of the signal fluctuations around the zero baseline. Calculating the ratio of the potential noise standard deviation to the current noise standard deviation essentially solves for the noise resistance in the frequency domain. This physical quantity reflects the ease of charge transfer at the electrode interface and the stability of the electric double layer. In a pure liquid phase, this ratio is usually dominated by the solution resistance and is relatively large; once a solid phase is formed, the interface charge transfer is disturbed, current fluctuations intensify, and the ratio changes.

[0043] In a specific implementation, the number of sampling points included in the sliding window is set to... For any time t, extract the data before that time. The filtered potential signal sequence v(k) and current signal sequence i(k) are obtained from each data point. First, the standard deviation of the potential noise within the window is calculated using statistical formulas. and current noise standard deviation Then, dividing the two yields the instantaneous electrochemical noise index at the current moment. As a comprehensive impedance parameter, this index eliminates the absolute bias of single signal amplitude caused by amplifier gain or electrode area, and provides normalized dimensionless observations that are highly sensitive to changes in the microstate of the interface.

[0044] In the initial stage before the addition of reactants to the crystallization reactor, the reaction system contains only the bottom liquid as a solvent and background ions. At this time, the system is in a thermodynamic metastable or equilibrium state, and there is no supersaturation-driven crystal nucleation process. The double-layer structure formed between the electrode surface and the solution has not yet been affected by the mechanical impact of solid particles or the adsorption interference of crystal growth. The electrochemical noise measured at this time mainly originates from the thermal motion of ions (thermal noise) and the small concentration fluctuations during diffusion (particle noise). This background noise level represents the inherent electrochemical impedance characteristics of the pure liquid system under this specific hydrodynamic and chemical environment. Recording the average noise index at this stage allows the establishment of a zero point that can cancel out systematic background interference, enabling subsequent monitoring to focus on the changes caused by phase transitions rather than the absolute value, thus avoiding long-term errors caused by electrode aging or temperature drift.

[0045] In practice, the period during which the reactor is started but the dosing pump has not yet been turned on is defined as the baseline calibration period, and its duration is denoted as [duration missing]. During this period, the system continuously calculates the instantaneous electrochemical noise index from the above steps. To eliminate outliers caused by random electronic noise or occasional turbulent flow, it is not advisable to directly select single-point values; instead, the arithmetic mean method is used to extract statistical characteristics. Assume that a total of [number missing] values ​​were obtained during the baseline calibration period. The effective instantaneous noise index data points are accumulated and averaged to calculate the pure liquid phase reference impedance. .Should The values ​​will be stored in the monitoring system's storage unit as a static comparison benchmark to determine whether the solid-liquid interface has undergone fundamental changes in subsequent reaction processes. Any signal that deviates significantly from this benchmark will be considered potential evidence of new phase formation.

[0046] The key to determining the crystallization initiation moment lies in capturing the double-layer reconstruction signal at the solid-liquid interface caused by the transition from a homogeneous to a multiphase reaction system. When the local supersaturation in the solution reaches a critical point, solute molecules begin to aggregate and nucleate. These tiny solid-phase nuclei adsorb onto the electrode surface or frequently collide with the electrode under the influence of the flow field, causing a sudden change in the double-layer capacitance at the electrode / solution interface and inducing a momentary spike in the Faraday current. This microscopic physical process manifests in macroscopic data as a significant deviation of the electrochemical noise index from the pure liquid phase reference impedance. To prevent false positives caused by bubble movement or electrical interference, the judgment logic must introduce a temporal continuity verification mechanism. Only when the deviation of the monitored index shows a certain persistence on the time axis, i.e., multiple consecutive observation windows show anomalies, can it be confirmed with statistical confidence that this is an irreversible process caused by a phase transition, rather than a transient interference. In a specific implementation, the electrochemical noise index at the current moment is calculated in real time. Relative to pure liquid phase reference impedance dimensionless deviation Set an empirical threshold of an order of magnitude. (For example, 0.5 or 1.0, depending on the system sensitivity). Continuous comparison. and The size relationship is determined, and a counter is started. If the condition is satisfied within n consecutive sliding windows (n ≥ 3), then... If the condition is met, it is determined that the double layer at the solid-liquid interface has undergone substantial reconstruction, and the time corresponding to the first window that meets the condition is confirmed as the crystallization initiation time.

[0047] In one embodiment, assessing the physical morphology of particles impacting the electrode surface in the crystallization reactor by calculating the kurtosis value of the current noise signal in response to the crystallization initiation time includes the following steps: Extract the current noise signal segment after the crystallization initiation time, and calculate the arithmetic mean and standard deviation of all discrete sampling points within the signal segment; For each discrete sampling point within a signal segment, the dimensionless kurtosis value of the signal segment is calculated by combining the arithmetic mean and standard deviation of the discrete sampling points. If the kurtosis value is less than the preset first morphology threshold, the impact event of particles hitting the electrode surface in the crystallization reactor is determined to follow a Gaussian distribution, and the physical morphology of the particles is evaluated as microcrystalline mode. If the kurtosis value is greater than the preset second morphology threshold, it is determined that the impact events of particles hitting the electrode surface in the crystallization reactor exhibit a long-tailed distribution, and the physical morphology of the particles is evaluated as an effective growth mode. The second morphology threshold is greater than the first morphology threshold and greater than 3. If the kurtosis value is greater than or equal to the first morphology threshold and less than or equal to the second morphology threshold, then the impact event of particles hitting the electrode surface in the crystallization reactor is determined to be in the transitional stage of non-Gaussian distribution, and the physical morphology of the particles is evaluated as a transitional growth mode.

[0048] In this embodiment, kurtosis, a higher-order statistic, can be used to extract the dynamic characteristics of crystalline particles impacting the electrode surface from the time-domain waveform. Unlike the mean or variance, which only describe the central tendency or dispersion of the signal, kurtosis can capture extreme events at the tail of the probability density distribution curve, i.e., sporadic large-amplitude pulse signals. In the crystallization reactor, particle impacts on the electrode generate current transients, the waveform characteristics of which are directly related to the particle size and physical morphology. To quantify this morphological difference, a signal segment of a specific duration needs to be extracted from the continuous current noise data stream. This segment must contain a sufficient number of sample points to ensure statistical significance. Subsequently, based on the reverse thinking of the central limit theorem, the degree to which the signal deviates from the normal distribution is determined by calculating the kurtosis. A normal distribution represents a superposition of a large number of small, random, and independent disturbances; while a non-normal distribution (especially with high kurtosis) means the existence of outlier events that are significantly different from the background noise, which physically corresponds to the strong impact of large crystal particles on the electrode.

[0049] In practice, the intercepted signal segment is set to include A discrete sampling point, represented as a sequence ( First, calculate the arithmetic mean of all points within the segment. and standard deviation Based on this, a dimensionless kurtosis value K is constructed, and its calculation formula is as follows: This formula significantly amplifies the contribution weight of large deviations to the final result by performing a fourth-power operation on the standardized deviation. Through this non-linear weighting, a few occasional high-peak pulses that might otherwise be ignored in the time-domain plot are transformed into significantly increased kurtosis values, thereby compressing complex waveform patterns into a single, highly sensitive quantitative indicator to particle morphology.

[0050] When the calculated kurtosis value is low, it physically reflects that the crystallization reactor is in the early stage of nucleation or the microcrystal formation stage. At this stage, the solution mainly contains a large number of nano- or micron-sized fine crystal nuclei. Due to the extremely small mass and weak inertia of these microcrystals, the momentum exchange generated when they impact the electrode surface under the influence of the flow field is very limited, resulting in small current fluctuations with extremely high frequency. According to the law of large numbers, the overall probability density function of this signal, composed of a massive superposition of tiny events, tends to be a Gaussian distribution (normal distribution), meaning that most data points are concentrated near the mean, with very few abnormal peaks far from the mean. The theoretical standard kurtosis value for a Gaussian distribution is 3. Therefore, if the actual monitored kurtosis value is close to or slightly lower than this theoretical benchmark, it can be inferred that the current noise mainly consists of background thermal noise and frequent slight disturbances from microcrystals, lacking significant impact characteristics from large particles.

[0051] In a specific implementation, an empirical first-morphological threshold is preset. (For example, set to 3.0 or 3.2). The real-time calculated kurtosis value K of the signal segment is then compared with... Compare them. If the conditions are met... This indicates that the distribution of the current noise amplitude does not exhibit a significant thick tail characteristic and the deviation is small. At this point, it is determined that the particle impact events within the crystallization reactor follow or approximately follow a Gaussian distribution. Based on this assessment, the current physical morphology of the particles is output as microcrystalline. This judgment implies that the nucleation rate within the reaction system may be dominant, or that the crystals have not yet grown to a size sufficient to produce a significant hydrodynamic impact effect. This suggests that the operator or control system is currently in the induction or early stage of crystal growth, and no strong stirring measures are needed for large particle suspension.

[0052] As the crystallization process progresses, the crystal size gradually increases, causing a qualitative change in the physical environment within the reactor. When large crystal particles form and move with the fluid, their significant kinetic energy impacts the electrode surface, instantly compressing the electric double layer or causing temporary changes in the geometry of the micro-region surface, resulting in large pulse spikes in the current signal. Statistically, these spikes exhibit a long-tail effect in the probability density curve, meaning that the probability of extreme values ​​occurring is significantly higher than predicted by a normal distribution, directly leading to a sharp increase in kurtosis. Therefore, by monitoring the degree of increase in kurtosis, the growth state of the crystals can be inferred. If the kurtosis is in the intermediate range, it indicates that the system contains both microcrystals and medium-sized particles, and the impact characteristics are evolving from random noise to a pulse sequence.

[0053] In practice, a second morphological threshold that is significantly higher than the Gaussian distribution benchmark is set. (Require and (For example, set to 5.0). The decision logic is divided into two branches: First, if the calculated kurtosis value... This indicates that the signal contains a large number of high-energy impact spikes, and the impact events exhibit a typical long-tail distribution. Based on this, it can be directly determined that the physical morphology of the particles is an effective growth mode, meaning that large-particle crystals have become the dominant phase. Secondly, if the kurtosis value falls between two thresholds, that is... This indicates that the signal distribution deviates from the Gaussian distribution but has not yet reached the state of extreme thick tails. It is determined that the impact event is in the transitional stage of non-Gaussian distribution, and the evaluation result is a transitional growth mode. This hierarchical evaluation strategy realizes dynamic tracking of the entire life cycle of crystal growth and can distinguish different stages of crystal growth, such as from nothing to something (microcrystals), from small to large (transition), or reaching maturity (effective growth).

[0054] In one embodiment, the flow rate in the crystallization reactor is periodically changed during monitoring. The changes in kurtosis and electrochemical noise before and after the flow rate change are combined with the fluid shear excitation principle to determine the current particulate matter type. This includes the following steps: Increase the preset base flow rate of the crystallization reactor to the preset disturbance flow rate; Simultaneously calculate the reference kurtosis value of the current noise signal at the base flow velocity and the disturbance kurtosis value of the current noise signal at the disturbance flow velocity, as well as the current noise standard deviation of the current noise signal at the reference flow velocity and the current noise standard deviation of the current noise signal at the disturbance flow velocity. By combining the baseline kurtosis value, the standard deviation of the current noise of the perturbation kurtosis value, and the standard deviation of the current noise, and based on the fluid shear excitation principle, the excess kurtosis energy flux ratio, which is used to quantify the competitive advantage of the growth rate of particulate impact characteristics relative to the growth rate of background turbulent energy, is calculated. If the excess kurtosis energy flux ratio is greater than 1, then the current particle type is determined to be dense crystal. If the excess kurtosis energy flow ratio is less than or equal to 1, then the current particulate matter type is determined to be either bubbles or soft flocs.

[0055] In this embodiment, to effectively identify the material properties of microparticles inside the crystallization reactor, passive static monitoring alone is insufficient; an active fluid dynamics excitation mechanism must be introduced. The core principle of this step lies in utilizing the differences in the response of substances of different densities to fluid shear forces: dense crystals possess a large inertial mass, and their kinetic energy increases significantly with increasing flow velocity, enabling them to break through the boundary layer on the electrode surface and generate more violent impacts; while bubbles or soft flocs have low density and are easily deformable, often following streamlines or breaking apart in high-speed flow fields, making it difficult to generate a significant impact enhancement effect. During implementation, a variable frequency drive controls the rotational speed of the circulating pump or agitator, raising the overall flow velocity inside the reactor from a lower, steady-state preset base velocity. Rapidly increase to a higher preset disturbance velocity This velocity step change needs to be completed in a very short time to ensure that sufficient transient shear force is generated.

[0056] During the synchronization window of flow velocity switching, the monitoring system locks onto and processes the current noise signal in real time. Using a sliding window algorithm, signal segments are extracted for both the base flow velocity stage and the disturbed flow velocity stage. For the signal segment at the base flow velocity, its fourth-order central moment statistic, i.e., the baseline kurtosis value, is calculated. Simultaneously, the root mean square value of this signal segment is calculated as the standard deviation of the current noise. This standard deviation physically characterizes the energy floor caused by background turbulence and Brownian motion at the current flow velocity. Similarly, for a signal segment at a disturbed flow velocity, the corresponding perturbation kurtosis value is calculated. and current noise standard deviation It must be ensured and , and The computation time windows are of consistent length, and the sampling frequency is sufficient to cover the high-frequency characteristics of particle impact. Through this paired feature extraction, the single-dimensional time-domain waveform is transformed into a four-element feature set reflecting the differences in the dynamic behavior of particles under different flow field energy levels. This provides complete data support for the subsequent decoupling of the physical properties of particles and avoids the risk of misjudgment caused by background noise fluctuations under a single flow velocity.

[0057] A dimensionless composite index, the excess kurtosis energy-fluidity ratio, is randomly constructed to quantify the hardness and inertia of particulate matter. The initial design of this index aims to isolate the interference of background turbulent energy growth and solely evaluate the net growth rate of non-Gaussian characteristics caused by particulate impacts. A normal increase in flow velocity inevitably leads to an increase in the fluid Reynolds number, which in turn causes a natural increase in background noise (standard deviation). If only the change in the absolute value of kurtosis is considered without considering the rise in background energy, false positives are highly likely. Therefore, a nonlinear mathematical model is needed to penalize the growth of background noise while rewarding the effective increase in kurtosis value.

[0058] In practical implementation, the formula for calculating the excess kurtosis energy flux ratio is as follows: In this mathematical model, the parameters and These represent the aforementioned base velocity and disturbance velocity, respectively. and For the corresponding kurtosis measurement value, and This represents the corresponding standard deviation of the current noise. (Function) Defined as an excess kurtosis extraction function, where the constant 3 represents the theoretical kurtosis value of the standard Gaussian distribution (i.e., pure liquid phase thermal noise). By subtracting 3 and taking the maximum value, this function eliminates the Gaussian components belonging to the background fluid noise, retaining only the non-Gaussian abrupt components caused by particle impacts; parameters For a very small positive number (such as This is used to prevent overflow caused by a zero denominator when the signal perfectly follows a Gaussian distribution. (Exponent) The turbulence energy suppression index is set between 1.5 and 2.0. This is used to nonlinearly amplify the penalty weight of background turbulence energy growth on the final index at a mathematical level; that is, if an increase in flow velocity leads to a significant increase in background noise without a significant increase in kurtosis, this ratio will drop sharply. (Logarithmic term) This serves as a normalization factor for the flow rate increment, eliminating the linear influence of the operational amplitude on the evaluation results, where e is a natural constant.

[0059] Based on the calculated excess kurtosis energy flux ratio The final particulate matter material identification is then performed. This judgment logic is based on the binary opposition characteristic of the fluid shear excitation principle: the fundamental difference in energy conversion efficiency between dense crystals and soft impurities. When A value greater than 1 indicates that during the flow velocity increase, the growth rate of the peak component representing non-Gaussian impacts in the current signal significantly exceeds the growth rate of the background turbulence energy. Physically, this indicates that the particles possess sufficient rigidity and density to effectively convert the additional kinetic energy imparted by the fluid into mechanical energy impacting the electrode surface, exhibiting a steeper pulse characteristic in the signal. This positive energy-signal conversion gain is a unique characteristic of hard particles. Therefore, when the condition is met... At that time, it can be confirmed that the main body of the current impact electrode is a dense crystal, such as magnesium ammonium phosphate crystal.

[0060] Conversely, if the calculation result shows This indicates that although the flow field energy increased, the non-Gaussian characteristics of the signal did not achieve a corresponding excess growth, and may even have been affected by background noise. The dramatic expansion (due to) The amplification effect of the exponent is masked. Physically, the corresponding situations are: particulate matter with a density close to that of a liquid phase (such as flocs), exhibiting good following behavior in an accelerating flow field, with low relative velocity and weak impact force; or the particulate matter being bubbles, undergoing deformation buffering or even breakup under high-speed shear, causing the original impact signal to be smoothed out. In these cases, the energy input to the flow field is mainly dissipated into the internal friction of the fluid or the deformation work of the soft particles, rather than being converted into impact pulses. Based on this, it can be determined that the currently dominant particle type is either bubbles or soft flocs.

[0061] In one embodiment, after determining the current particulate matter type, the following steps are further included: If the particulate matter is dense crystal, the kurtosis value of the current current noise signal is marked as an effective control variable, and the effective control variable is used as a reference parameter for adjusting the dosing rate and reflux ratio parameters. If the particulate matter is bubbles or soft flocs, the defoaming program of the crystallization reactor is started, and the flow field velocity is changed again after the bubble interference source is eliminated until the excess kurtosis energy-fluid ratio is greater than 1, and the update of the effective control variables is restored.

[0062] In this embodiment, when the predominant particulate matter type in the reactor is confirmed to be dense crystal through the aforementioned fluid shear excitation and excess kurtosis energy flow ratio calculation, it indicates that the current monitoring data truly reflects the kinetic state of crystal growth and is not affected by soft impurities. At this point, the kurtosis value of the current noise signal collected and calculated at the current moment is formally marked as an effective control variable. This marking process gives this data point a very high confidence weight in the control logic, distinguishing it from the raw data that has not been verified by the material. This effective control variable is then transmitted to the central control unit (PLC or DCS system) of the crystallization reactor as the core input parameter of the feedback loop. Based on the magnitude of this parameter and its evolution trend over time, the control unit dynamically calculates and adjusts the dosing rate (i.e., the amount of phosphorus and magnesium sources added) and the reflux ratio parameter. For example, if the effective control variable shows a continuous increase in kurtosis, it suggests that the crystal is on a benign growth trajectory, and the system can maintain the current dosing rate to maintain a stable supersaturation; if the kurtosis tends to level off, the residence time distribution of the crystal in the reaction zone may be changed by fine-tuning the reflux ratio.

[0063] If the calculated excess kurtosis energy flow ratio indicates that the current particulate matter type is bubbles or soft flocs, it means that the monitoring system is being severely disturbed by the amorphous solid phase. In this case, the collected kurtosis data is distorted and cannot be directly used for process control. Therefore, an anomaly handling mechanism will be triggered immediately. First, the automatic updates of the dosing rate and reflux ratio parameters will be suspended to prevent adjustments based on erroneous data that could worsen the process. Next, the pre-set defoaming program of the crystallization reactor will be initiated. This program typically includes activating the top mechanical defoamer, spraying defoamer, or adjusting the liquid level to overflow and remove surface scum. For soft flocs, this may involve adjusting the coagulant dosing point or changing the bottom sludge removal strategy. After a preset time for defoaming or sludge removal, a re-validation cycle will begin, where the flow field velocity will be actively changed again, i.e., the velocity will be increased from the baseline value to the disturbed value, and data will be collected again to calculate the excess kurtosis energy flow ratio. The cycle of "detection-cleaning-re-detection" will continue until the calculated excess kurtosis energy flux ratio is greater than 1 again, confirming that the interference source has been effectively eliminated, and the monitoring window will refocus on the dense crystal.

[0064] In one embodiment, adjusting the dosing rate and reflux ratio parameters of the crystallization reactor based on the physical morphology and type of the particles includes the following steps: When the physical morphology of the particulate matter is microcrystalline, the dosing pump of the crystallization reactor is controlled to reduce the dosing rate according to the exponential decay law to reduce local supersaturation, while the seed reflux pump of the crystallization reactor is controlled to operate at maximum power. When the physical morphology of the particulate matter is in the transitional growth mode, keep the current dosing rate and reflux ratio parameters of the crystallization reactor unchanged, and start the stirring device of the crystallization reactor to promote crystal suspension and mass transfer until the kurtosis value of the signal segment naturally evolves to above the second morphology threshold. When the physical form of the particles is in an effective growth mode, control the dosing pump to maintain the current dosing rate, and at the same time control the speed of the stirring device to match the critical flow rate of the particles to prevent the particles from settling and accumulating at the bottom.

[0065] In this embodiment, when the monitoring system determines that the physical morphology of the particulate matter is microcrystalline, it means that a large number of tiny crystal nuclei have formed within the reactor. Without intervention, excessive supersaturation can easily lead to the formation of fine, difficult-to-separate particles, affecting the particle size distribution of the final product. Therefore, the primary task of the control strategy is to suppress the nucleation rate and induce crystal growth. The system uses a frequency converter to reduce the frequency of the dosing pump, thereby decreasing the dosing rate. According to the pre-defined exponential decay law, it decreases with time t, and its governing equation is expressed as: ,in For the current dosing rate, The decay coefficient is denoted as . This nonlinear rate reduction mechanism rapidly pulls the local supersaturation in the reaction zone back to the metastable region, preventing the explosive generation of new crystal nuclei. Simultaneously, to provide sufficient growth sites to consume the remaining supersaturation, the control system forcibly increases the power of the seed reflux pump to its rated maximum. This operation increases the number of mature seeds refluxed from the precipitation zone to the reaction zone, utilizing the seed effect to preferentially deposit and grow the solute in the solution on the surface of existing seeds, rather than spontaneously forming new microcrystals. This reverses the microcrystallization trend on both thermodynamic and kinetic levels, guiding the system towards larger particle growth.

[0066] When the system identifies that the particles are in a transitional growth mode, it indicates that the crystals are in a critical intermediate stage of growth from micronuclei to macroscopic particles. At this point, the system is extremely sensitive, and excessive parameter adjustments can easily interrupt the growth process or cause crystal breakage. Therefore, the control logic switches to a steady-state maintenance strategy. First, the current dosing rate and reflux ratio are locked and kept constant to maintain the existing stoichiometry and hydraulic residence time in the reactor, providing a stable chemical environment for the crystals. To overcome the increasingly significant diffusion resistance as the crystal size increases, an auxiliary stirring device (such as a paddle agitator or a circulating flow agitator) is then simultaneously activated. The introduction of stirring is not for violent shearing, but to enhance the relative velocity between the liquid and solid phases, reduce the thickness of the liquid film diffusion layer on the crystal surface, and thus enhance the mass transfer efficiency of solute molecules to the crystal surface. This stirring state will be maintained continuously, during which the monitoring system continuously tracks the evolution of the kurtosis value of the current noise signal. Only when the kurtosis value naturally rises and stably exceeds the preset second morphology threshold, indicating that the crystals have passed the fragile transition period and completed the morphological transformation to large particles, will the system exit this steady-state maintenance mode and enter the next stage of control logic.

[0067] Once the particles are determined to have entered an effective growth mode, it means that a large number of dense crystals with large diameters and complete morphologies have formed within the reactor. Chemically, the dosing pump is controlled to maintain the current dosing rate to ensure a consistently moderate supersaturation in the solution, driving the crystals to continue growing to the target size. Physically, because large crystal particles tend to accumulate at the bottom of the reactor due to gravity, leading to dead zones or discharge blockages, the hydrodynamic conditions must be adjusted. The control system estimates the critical suspension velocity at the current average particle size using Stokes' theorem and precisely adjusts the rotation speed of the agitator accordingly. The goal is to ensure that the upward flow velocity generated by agitation is slightly greater than the final settling velocity of the crystals, resulting in a uniform fluidized suspension distribution of crystal particles within the reactor. This avoids bottom accumulation and prevents secondary collisions and breakage of large crystals due to the strong shear force generated by excessive rotation speed. Through this fluidized adaptive control, it is ensured that each crystal can fully contact the reaction liquid in a suspended state until the product discharge standard is reached, achieving efficient and stable operation in the later stages of the crystallization process.

[0068] The present invention also discloses a data analysis system for treating industrial wastewater containing ammonium phosphate, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the data analysis method for treating industrial wastewater containing ammonium phosphate as described in any of the above claims.

[0069] The processor can be a central processing unit (CPU). Of course, depending on the actual use, it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), off-the-shelf programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc., and this application does not limit it.

[0070] The memory can be an internal storage unit of a computer device, such as a hard disk or RAM, or an external storage device, such as a plug-in hard disk, smart memory card (SMC), secure digital card (SD), or flash memory card (FC) provided on the computer device. Furthermore, the memory can be a combination of internal storage units and external storage devices of a computer device. The memory is used to store computer programs and other programs and data required by the computer device. The memory can also be used to temporarily store data that has been output or will be output. This application does not limit this.

[0071] The present invention also discloses a computer-readable storage medium storing instructions that, when executed by a processor, configure the processor to perform the data analysis method and system method for treating industrial wastewater containing ammonium phosphate as described in any of the above embodiments.

[0072] The computer program can be stored in a machine-readable medium. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or certain middleware. The machine-readable medium includes any entity or device capable of carrying computer program code, recording media, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the machine-readable medium includes, but is not limited to, the above-mentioned components.

[0073] The data analysis method and system method for treating industrial wastewater containing ammonium phosphate in the above embodiments are stored in the computer-readable storage medium and loaded and executed on the processor to facilitate the storage and application of the above methods.

[0074] Those skilled in the art should understand that the discussion of any of the above embodiments is merely exemplary and is not intended to imply that the scope of protection of this application is limited to these examples; within the framework of this application, the technical features of the above embodiments or different embodiments can also be combined, the steps can be implemented in any order, and there are many other variations of different aspects of one or more embodiments of this application as described above, which are not provided in detail for the sake of brevity.

[0075] One or more embodiments in this application are intended to cover all such substitutions, modifications, and variations that fall within the broad scope of this application. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of one or more embodiments in this application should be included within the protection scope of this application.

Claims

1. A method for analyzing data from the treatment of industrial wastewater containing ammonium phosphate, characterized in that, Includes the following steps: A detection circuit consisting of two homogeneous inert electrodes and a reference electrode is set up in a crystallization reactor for treating industrial wastewater containing ammonium phosphate. A zero-resistance galvanometer is used to collect the potential noise signal and current noise signal of the detection circuit, and the polarization error in the potential noise signal and current noise signal is filtered out. The ratio of the standard deviation of the potential noise signal and the current noise signal after the filtering operation is used as the electrochemical noise index. By monitoring the electrochemical noise index, the crystallization initiation time when the reaction system in the crystallization reactor changes from a pure liquid phase to a solid-liquid mixed phase is determined. In response to the crystallization initiation moment, the physical morphology of particles impacting the electrode surface in the crystallization reactor is assessed by calculating the kurtosis value of the current noise signal; During the monitoring process, the flow field velocity in the crystallization reactor was periodically changed. The changes in the kurtosis value and electrochemical noise of the current noise signal before and after the flow field velocity change were combined, and the current particulate matter type was determined based on the fluid shear excitation principle. Adjust the dosing rate and reflux ratio parameters of the crystallization reactor based on the physical morphology and type of the particulate matter. The kurtosis value and electrochemical noise index are continuously monitored. The crystal growth in the crystallization reactor is determined to be complete only when the kurtosis value remains in the preset high range and no longer rises, and the electrochemical noise index is in the preset stable range.

2. The data analysis method for treating industrial wastewater containing ammonium phosphate according to claim 1, characterized in that, The process of acquiring the potential noise signal and current noise signal of the detection circuit using a zero-resistance galvanometer and filtering out the polarization error in the potential noise signal and current noise signal includes the following steps: The spontaneous current fluctuations between two homogeneous inert electrodes are collected using a zero-resistance galvanometer while maintaining zero potential difference between the two electrodes as current noise signals. The potential fluctuations of one of the homogeneous inert electrodes relative to the reference electrode are recorded synchronously as potential noise signals; The polarization error in potential noise signal and current noise signal is filtered out by using the time-series structure fragmentation and reconstruction method.

3. The data analysis method for treating industrial wastewater containing ammonium phosphate according to claim 2, characterized in that, The method for filtering polarization errors in potential noise signals and current noise signals using time-series structure fragmentation and reconstruction includes the following steps: Obtain the time index of the potential noise signal and the current noise signal; Based on the time index and using the Monte Carlo random algorithm, all discrete sampling points in the potential noise signal and current noise signal are randomly shuffled and rearranged to obtain potential noise signal and current noise signal with polarization error filtered out, which has monotonic evolution characteristics.

4. The data analysis method for treating industrial wastewater containing ammonium phosphate according to claim 1, characterized in that, The step of using the ratio of the standard deviations of the potential noise signal and the current noise signal after the filtering operation as an electrochemical noise index, and determining the crystallization initiation time when the reaction system in the crystallization reactor changes from a pure liquid phase to a solid-liquid mixed phase by monitoring the electrochemical noise index, includes the following steps: Within a preset sliding window, the root mean square value of the potential noise signal after the filtering operation is calculated as the standard deviation of the potential noise, and the root mean square value of the current noise signal after the filtering operation is calculated as the standard deviation of the current noise. The instantaneous electrochemical noise index is obtained by dividing the standard deviation of the potential noise by the standard deviation of the current noise. The average value of the electrochemical noise index was recorded during the untreated stage of the reaction in the crystallization reactor as the pure liquid phase reference impedance. Calculate the deviation of the electrochemical noise index from the pure liquid phase reference impedance at the current moment. When the deviation exceeds the preset order of magnitude threshold for n consecutive sliding windows, it is confirmed that the double layer of the solid-liquid interface has been reconstructed. The current moment is determined to be the crystallization start moment when the reaction system in the crystallization reactor changes from the pure liquid phase to the solid-liquid mixed phase, n≥3.

5. The data analysis method for treating industrial wastewater containing ammonium phosphate according to claim 1, characterized in that, The process of assessing the physical morphology of particles impacting the electrode surface in the crystallization reactor by calculating the kurtosis value of the current noise signal in response to the crystallization initiation time includes the following steps: Extract the current noise signal segment after the crystallization initiation time, and calculate the arithmetic mean and standard deviation of all discrete sampling points within the signal segment; For each discrete sampling point within a signal segment, the dimensionless kurtosis value of the signal segment is calculated by combining the arithmetic mean and standard deviation of the discrete sampling points. If the kurtosis value is less than the preset first morphology threshold, the impact event of particles hitting the electrode surface in the crystallization reactor is determined to follow a Gaussian distribution, and the physical morphology of the particles is evaluated as microcrystalline mode. If the kurtosis value is greater than the preset second morphology threshold, it is determined that the impact events of particles hitting the electrode surface in the crystallization reactor exhibit a long-tailed distribution, and the physical morphology of the particles is evaluated as an effective growth mode. The second morphology threshold is greater than the first morphology threshold and greater than 3. If the kurtosis value is greater than or equal to the first morphology threshold and less than or equal to the second morphology threshold, then the impact event of particles hitting the electrode surface in the crystallization reactor is determined to be in the transitional stage of non-Gaussian distribution, and the physical morphology of the particles is evaluated as a transitional growth mode.

6. The data analysis method for treating industrial wastewater containing ammonium phosphate according to claim 1, characterized in that, The process of periodically changing the flow field velocity in the crystallization reactor during monitoring, combining the changes in kurtosis and electrochemical noise of the current noise signal before and after the flow field velocity change, and determining the current particulate matter type based on the fluid shear excitation principle includes the following steps: Increase the preset base flow rate of the crystallization reactor to the preset disturbance flow rate; Simultaneously calculate the reference kurtosis value of the current noise signal at the base flow velocity and the disturbance kurtosis value of the current noise signal at the disturbance flow velocity, as well as the current noise standard deviation of the current noise signal at the reference flow velocity and the current noise standard deviation of the current noise signal at the disturbance flow velocity. By combining the baseline kurtosis value, the standard deviation of the current noise of the perturbation kurtosis value, and the standard deviation of the current noise, and based on the fluid shear excitation principle, the excess kurtosis energy flux ratio, which is used to quantify the competitive advantage of the growth rate of particulate impact characteristics relative to the growth rate of background turbulent energy, is calculated. If the excess kurtosis energy flux ratio is greater than 1, then the current particle type is determined to be dense crystal. If the excess kurtosis energy flow ratio is less than or equal to 1, then the current particulate matter type is determined to be either bubbles or soft flocs.

7. The data analysis method for treating industrial wastewater containing ammonium phosphate according to claim 6, characterized in that, After determining the current particulate matter type, the following steps are also included: If the particulate matter is dense crystal, the kurtosis value of the current current noise signal is marked as an effective control variable, and the effective control variable is used as a reference parameter for adjusting the dosing rate and reflux ratio parameters. If the particulate matter is bubbles or soft flocs, the defoaming program of the crystallization reactor is started, and the flow field velocity is changed again after the bubble interference source is eliminated until the excess kurtosis energy-fluid ratio is greater than 1, and the update of the effective control variables is restored.

8. The data analysis method for treating industrial wastewater containing ammonium phosphate according to claim 5, characterized in that, The adjustment of the dosing rate and reflux ratio parameters of the crystallization reactor based on the physical morphology and type of the particles includes the following steps: When the physical morphology of the particulate matter is microcrystalline, the dosing pump of the crystallization reactor is controlled to reduce the dosing rate according to the exponential decay law to reduce local supersaturation, while the seed reflux pump of the crystallization reactor is controlled to operate at maximum power. When the physical morphology of the particulate matter is in the transitional growth mode, keep the current dosing rate and reflux ratio parameters of the crystallization reactor unchanged, and start the stirring device of the crystallization reactor to promote crystal suspension and mass transfer until the kurtosis value of the signal segment naturally evolves to above the second morphology threshold. When the physical form of the particles is in an effective growth mode, control the dosing pump to maintain the current dosing rate, and at the same time control the speed of the stirring device to match the critical flow rate of the particles to prevent the particles from settling and accumulating at the bottom.

9. A data analysis system for treating industrial wastewater containing ammonium phosphate, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the data analysis method for treating industrial wastewater containing ammonium phosphate as described in any one of claims 1 to 8.

10. A computer-readable storage medium storing instructions thereon, characterized in that, When executed by a processor, this instruction causes the processor to be configured to perform a data analysis method for treating industrial wastewater containing ammonium phosphate according to any one of claims 1 to 8.

Images