A material particle size monitoring method and system for basic cobalt carbonate production

Abstract

The present invention discloses a material particle size monitoring method and system for basic cobalt carbonate production, relating to the technical field of material particle size monitoring. The present invention significantly improves the particle size monitoring accuracy under high solid content and strong turbulence conditions through multimodal sensor data fusion and intelligent decomposition algorithm. The present invention combines physical constraints with a dynamic splitting mechanism to effectively distinguish between seed particles and abnormal secondary nucleation particles, avoiding the problem of traditional methods misjudging double peaks as broad peaks. At the same time, based on the current reference threshold strategy of sliding window statistics, it adaptively matches different batches and working condition changes, reduces the need for manual intervention, and quantifies the adjustment of stirring parameters as a function of dispersion uniformity index, breaking through the limitations of empirical control. Frequency domain compensation and adaptive filtering design are additionally introduced to significantly suppress the influence of complex interference sources such as bubbles and mechanical vibrations.

Description

Technical Field

[0001] The present invention relates to the technical field of material particle size monitoring, and in particular to a material particle size monitoring method and system for basic cobalt carbonate production. Background Art

[0002] In the continuous production process of basic cobalt carbonate, uniform dispersion of seed crystals is the key to controlling the unimodal distribution of the product particle size. Current mainstream production lines generally use online laser particle size analyzers or ultrasonic attenuation spectroscopy analysis technology to dynamically adjust the stirring intensity and seed addition acceleration by real-time monitoring of particle size changes within the reactor. However, under conditions of high solids content >25% and strong turbulent stirring speeds >1000 r / min, seed crystals are prone to competing with newly formed particles due to local concentration gradients and uneven distribution of fluid shear forces, ultimately resulting in a significant bimodal particle size distribution curve. This phenomenon is particularly prominent in the preparation of high-tap-density lithium cobalt oxide precursors, directly affecting the compaction performance and cycle life of battery cathode materials.

[0003] Although machine learning-based light scattering signal analysis algorithms and multi-sensor fusion solutions have emerged in recent years, they still have essential defects in bimodal distribution scenarios. Traditional feature extraction methods have difficulty distinguishing between seed-induced peaks and abnormal secondary nucleation peaks, and mistakenly merge the two peaks into a broad peak. In addition, the particle size optical response model that relies on offline calibration cannot adapt to the real-time changing flow field environment, resulting in insufficient accuracy in the decoupling of the mixed distribution. For example, the patent disclosed by publication number CN104418388B uses eddy current enhanced dispersion technology to improve the uniformity of the initial distribution of seeds. However, during the continuous reaction process, the periodic occurrence of stirring dead zones will still cause the local seed concentration to fluctuate beyond the critical value, and the existing monitoring system lacks effective perception of such microscopic heterogeneity.

[0004] To address the above issues, some solutions have alleviated the double-peak phenomenon by combining a pulsed seed injection strategy with an adaptive optical compensation algorithm. These solutions include: using high-frequency, small-dose pneumatic injection instead of batch feeding to reduce the risk of local supersaturation; introducing a particle size distribution prediction model based on transfer learning to infer the seed dispersion state through historical data; however, such methods are limited by the hardware response speed and model generalization capabilities. When dealing with sudden flow field disturbances, such as vibration offset of the stirring shaft and blockage of the feed pipe, control lags or even false triggers will still occur, making it difficult to fundamentally suppress the formation of a double-peak distribution. Summary of the Invention

[0005] In view of the above existing problems, the present invention is proposed.

[0006] The present invention provides a material particle size monitoring method and system for basic cobalt carbonate production to solve the problem that existing monitoring methods are difficult to distinguish the mixed distribution of seed crystals and newly formed particles at high solid content, and control lag leads to frequent double peaks in particle size, which restricts product consistency.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0008] In a first aspect, an embodiment of the present invention provides a material particle size monitoring method for basic cobalt carbonate production, which comprises:

[0009] Step S1, real-time collection of multi-source sensor data in the reactor, wherein the multi-source sensor data includes a laser scattering intensity angular distribution signal, an ultrasonic attenuation spectrum signal, and a stirring shaft speed signal;

[0010] Step S2, generating a first particle size distribution curve based on the laser scattering intensity angular distribution signal, and generating a second particle size distribution curve based on the ultrasonic attenuation spectrum signal;

[0011] Step S3, inputting the first particle size distribution curve and the second particle size distribution curve into a multimodal distribution decomposition model, and outputting a decomposed seed particle distribution sub-curve and a newly formed particle distribution sub-curve;

[0012] Step S4, calculating the peak ratio of the seed particle distribution sub-curve, and generating a stirring parameter adjustment instruction when the peak ratio is lower than a preset threshold;

[0013] Step S5: sending the stirring parameter adjustment instruction to the stirring drive device to dynamically adjust the stirring blade inclination angle and speed gradient.

[0014] As a preferred embodiment of the material particle size monitoring method for basic cobalt carbonate production described in the present invention, the construction of the multimodal distribution decomposition model includes:

[0015] Gaussian mixture modeling is performed on the particle size distribution curve in historical production data to constrain the seed particle size distribution to conform to the preset nominal distribution characteristics;

[0016] Use KL divergence to dynamically evaluate the deviation between the real-time distribution curve and the historical distribution curve. When the deviation exceeds the set critical value, the distribution curve split operation is triggered.

[0017] The output includes the decomposition results of the seed particle distribution sub-curve and the new particle distribution sub-curve;

[0018] The multimodal distribution decomposition model includes a feature alignment layer, a cross attention layer, and a physical constraint layer, wherein:

[0019] The feature alignment layer performs time stamp synchronization and spatial registration on the laser scattering intensity angular distribution signal and the ultrasonic attenuation spectrum signal;

[0020] The cross attention layer calculates the correlation weight matrix of the two modal signals in the frequency-spatial domain;

[0021] The physical constraint layer enforces the particle volume conservation equation.

[0022] As a preferred embodiment of the material particle size monitoring method for basic cobalt carbonate production described in the present invention, the distribution curve splitting operation includes the following steps:

[0023] a) Detect the peak-to-valley ratio of the real-time distribution curve. When the valley bottom value is higher than 20% of the peak height on both sides, it is determined to be a potential double peak;

[0024] b) Perform second-order derivative zero-crossing detection on the suspected double-peak area to locate the center of the sub-peak;

[0025] c) Dynamically increase the number of components of the Gaussian mixture model according to the sub-peak spacing.

[0026] As a preferred embodiment of the material particle size monitoring method for basic cobalt carbonate production according to the present invention, in step S3, the Gaussian mixture modeling of the historical particle size distribution and the seed nominal distribution constraint are performed as follows:

[0027] The particle size variable after centering use Gaussian components to construct the mixture density function:

[0028] ,

[0029] in, Indicates the The weights of the Gaussian components, represents the total number of components of the Gaussian mixture model, Indicates the The variance of the components, Indicates the The mean of the components, represents the normalized and centered particle size variable, Indicates that the mixed model has a particle size The probability density at ;

[0030] The introduced pair number is The component and nominal distribution of seed crystals The KL divergence constraint of is used to make the component close to the preset feature. The overall optimization goal is set to be the sum of log-likelihood and constraint divergence:

[0031] ,

[0032] in, represents the total number of historical samples, Indicates the The particle size of the samples, represents the constraint strength coefficient, represents the Kullback-Leibler divergence between two normal distributions, represents the nominal mean value of the seed crystal, represents the nominal variance of the seed crystal, represents the component index assigned to the seed feature, and Respectively represent The mean and variance of each component;

[0033] , where the denominator and The variance of the distributions being compared before and after, represents the square of the mean deviation of the two distributions, and the constant 1 is used to standardize the divergence baseline;

[0034] Adopting improved EM iterative minimization ,include:

[0035] Step E: Calculate the posterior probability:

[0036] ,

[0037] M step:

[0038] ,

[0039] in, Indicates the The samples belong to The posterior probability of the components, Indicates the Gaussian components in The density value at Represents the first Particle size measurements of historical samples, Indicates the The weight of the component after the M-step update in this round, represents the updated mean, represents the updated variance;

[0040] right Component, additional gradient correction after M steps:

[0041] ,

[0042] in, Represents the learning rate used for regular gradient correction, and iterates until , Indicates that it belongs to the seed classification index The component of is updated in this round of M steps and the mean value after gradient correction, Indicates that it belongs to the seed classification index The variance of the component after being updated in this round of M steps and corrected by the gradient.

[0043] As a preferred embodiment of the material particle size monitoring method for basic cobalt carbonate production of the present invention, the step of generating the stirring parameter adjustment instruction includes:

[0044] Calculating a dispersion uniformity index based on the peak width of the seed particle distribution sub-curve;

[0045] Establishing a mapping relationship between the dispersion uniformity index and the inclination angle of the stirring blade to generate an inclination angle compensation amount;

[0046] Based on the current stirring speed and material viscosity, the speed gradient adjustment step is calculated.

[0047] As a preferred embodiment of the material particle size monitoring method for basic cobalt carbonate production described in the present invention, in which: in step S4, according to the peak ratio calculation and the current reference threshold value, the mixing weight of the seed component in the decomposition model is taken. As peak proportion , in response to operating condition fluctuations, design the current reference threshold based on sliding window statistics The process includes:

[0048] The mixing weight represents the proportion of seed particles in the overall distribution and is used to measure the abundance of seed particles. The current reference threshold is adaptively adjusted based on historical data.

[0049] ,

[0050] in, Represents the mixing weight of the seed component, that is, the weight of the component in the Gaussian mixture model;

[0051] The window statistics radius is set to Sampling times, calculate the mean and standard deviation of the historical peak ratio:

[0052] ,

[0053] in, Indicates the length of the historical sampling window, Indicates time The peak proportion of and are the mean and standard deviation within the window respectively;

[0054] Based on statistical results, set safety factors After that, the current reference threshold is:

[0055] ,

[0056] in, Indicates the safety factor, used to adjust the trigger sensitivity. Indicates the current reference threshold;

[0057] When the real-time peak ratio Below the current reference threshold When the stirring parameter adjustment instruction is generated immediately, the seed crystal abundance is restored.

[0058] In a second aspect, the present invention provides a material particle size monitoring system for basic cobalt carbonate production, comprising:

[0059] A multi-source sensing module, comprising a laser scattering probe array, an ultrasonic transceiver, and a stirring shaft encoder mounted on the side wall of the reactor;

[0060] a signal processing module connected to the multi-source sensing module and configured to perform noise reduction processing on the raw sensing data and generate a first particle size distribution curve and a second particle size distribution curve;

[0061] a distribution decomposition module, integrating the multimodal distribution decomposition model, receiving the first particle size distribution curve and the second particle size distribution curve and outputting a decomposition result;

[0062] a control decision module, connected to the distribution decomposition module, and configured to generate a stirring parameter adjustment instruction according to the peak proportion;

[0063] The actuator interface module converts the stirring parameter adjustment instruction into a driving signal and sends it to the stirring driving device.

[0064] As a preferred solution of the material particle size monitoring system for basic cobalt carbonate production described in the present invention, the arrangement of the laser scattering probe array is as follows:

[0065] Three detection planes are set up in the axial height direction of the reactor, with three probes distributed in a 120° ring in each plane;

[0066] The angle between the incident light path of each probe and the center line of the stirring shaft is 15°-30°, and the probes of adjacent layers are arranged with a 40° offset in the circumferential direction.

[0067] As a preferred solution of the material particle size monitoring system for basic cobalt carbonate production described in the present invention, the signal processing module includes:

[0068] Adaptive filtering unit, used to eliminate periodic mechanical vibration noise caused by the rotation of the mixing blade;

[0069] Frequency domain compensation unit dynamically adjusts the frequency band weight coefficient of the ultrasonic attenuation spectrum according to real-time solid content data;

[0070] The data synchronization unit is used for data synchronization.

[0071] As a preferred solution of the material particle size monitoring system for basic cobalt carbonate production described in the present invention, the control decision module further includes:

[0072] The emergency intervention unit automatically triggers the seed replenishment device to start when it detects that the peak particle size of the new particle distribution sub-curve exceeds 1.5 times the peak particle size of the seed particles;

[0073] The discharge port of the seed adding device is provided with an ultrasonic atomizing head, and the atomized particle size is controlled in the range of 50-100 μm.

[0074] The beneficial effects of the present invention are as follows: the present invention significantly improves the particle size monitoring accuracy under high solid content and strong turbulence conditions through multimodal sensor data fusion and intelligent decomposition algorithm; the present invention combines physical constraints with dynamic splitting mechanisms to effectively distinguish between seed particles and abnormal secondary nucleation particles, avoiding the problem of traditional methods misjudging double peaks as broad peaks; at the same time, a dynamic threshold strategy based on sliding window statistics adaptively matches different batches and working condition changes, reduces the need for manual intervention, and quantifies the adjustment of stirring parameters as a function of dispersion uniformity index, breaking through the limitations of empirical control, and additionally introducing frequency domain compensation and adaptive filtering design to significantly suppress the influence of complex interference sources such as bubbles and mechanical vibrations. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0076] Figure 1 Schematic diagram of the process of monitoring the material particle size for basic cobalt carbonate production in Example 1.

[0077] Figure 2 This is a schematic diagram of the framework of the material particle size monitoring system for basic cobalt carbonate production in Example 1. DETAILED DESCRIPTION

[0078] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0079] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0080] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0081] Example 1, reference Figure 1 and Figure 2 This embodiment provides a material particle size monitoring method for basic cobalt carbonate production, comprising the following steps:

[0082] Step S1, real-time collection of multi-source sensor data in the reactor, the multi-source sensor data including laser scattering intensity angular distribution signal, ultrasonic attenuation spectrum signal and stirring shaft speed signal;

[0083] Step S2, generating a first particle size distribution curve based on the laser scattering intensity angular distribution signal, and generating a second particle size distribution curve based on the ultrasonic attenuation spectrum signal;

[0084] Step S3, inputting the first particle size distribution curve and the second particle size distribution curve into a multimodal distribution decomposition model, and outputting a decomposed seed particle distribution sub-curve and a newly formed particle distribution sub-curve;

[0085] The construction of the multimodal distribution decomposition model includes:

[0086] Gaussian mixture modeling is performed on the particle size distribution curve in historical production data to constrain the seed particle size distribution to conform to the preset nominal distribution characteristics;

[0087] Use KL divergence to dynamically evaluate the deviation between the real-time distribution curve and the historical distribution curve. When the deviation exceeds the set critical value, the distribution curve split operation is triggered.

[0088] The output includes the decomposition results of the seed particle distribution sub-curve and the new particle distribution sub-curve;

[0089] The multimodal distribution decomposition model includes a feature alignment layer, a cross-attention layer, and a physical constraint layer, where:

[0090] The feature alignment layer performs time stamp synchronization and spatial registration on the laser scattering intensity angular distribution signal and the ultrasonic attenuation spectrum signal;

[0091] The cross attention layer calculates the correlation weight matrix of the two modal signals in the frequency-spatial domain;

[0092] The physical constraint layer forces the particle volume conservation equation to be satisfied, so that the total volume of the decomposed sub-curves is consistent with the original distribution;

[0093] The distribution curve splitting operation includes the following steps:

[0094] a) Detect the peak-to-valley ratio of the real-time distribution curve. When the valley bottom value is higher than 20% of the peak height on both sides, it is determined to be a potential double peak;

[0095] b) Perform second-order derivative zero-crossing detection on the suspected double-peak area to locate the center of the sub-peak;

[0096] c) dynamically increasing the number of components of the Gaussian mixture model according to the sub-peak spacing;

[0097] In step S3, the Gaussian mixture modeling of the historical particle size distribution and the seed nominal distribution constraint are as follows:

[0098] The particle size variable after centering use Gaussian components to construct the mixture density function:

[0099] ,

[0100] in, Indicates the The weights of the Gaussian components, represents the total number of components of the Gaussian mixture model, Indicates the The variance of the components, Indicates the The mean of the components, represents the normalized and centered particle size variable, Indicates that the mixed model has a particle size The probability density at ;

[0101] The introduced pair number is The component and nominal distribution of seed crystals The KL divergence constraint of is used to make the component close to the preset feature. The overall optimization goal is set to be the sum of log-likelihood and constraint divergence:

[0102] ,

[0103] in, represents the total number of historical samples, Indicates the The particle size of the samples, represents the constraint strength coefficient, represents the Kullback-Leibler divergence between two normal distributions, represents the nominal mean value of the seed crystal, represents the nominal variance of the seed crystal, represents the component index assigned to the seed feature, and Respectively represent The mean and variance of each component;

[0104] , where the denominator and The variance of the distributions being compared before and after, represents the square of the mean deviation of the two distributions, and the constant 1 is used to standardize the divergence baseline;

[0105] Adopting improved EM iterative minimization ,include:

[0106] Step E: Calculate the posterior probability:

[0107] ,

[0108] M step:

[0109] ,

[0110] in, Indicates the The samples belong to The posterior probability of the components, Indicates the Gaussian components in The density value at Represents the first Particle size measurements of historical samples, Indicates the The weight of the component after the M-step update in this round, represents the updated mean, represents the updated variance;

[0111] right Component, additional gradient correction after M steps:

[0112] ,

[0113] in, Represents the learning rate used for regular gradient correction, and iterates until , Indicates that it belongs to the seed classification index The component of is updated in this round of M steps and the mean value after gradient correction, Indicates that it belongs to the seed classification index The variance of the component after being updated in this round of M steps and corrected by the gradient;

[0114] Specifically, the Gaussian mixture model is tightly coupled with the nominal distribution of the target component through KL divergence, so that the selected component always expands around the preset peak position of the seed without deviation. In the EM iteration, an additional gradient correction is applied after the M step, so that the nominal distribution constraint is naturally integrated into the parameter update in the form of an analytical gradient, without the need for numerical approximation, and the optimization efficiency is improved. At the same time, it can be adjusted Controlling the model's reliance on priors;

[0115] Step S4, calculating the peak ratio of the seed particle distribution sub-curve, and generating a stirring parameter adjustment instruction when the peak ratio is lower than a preset threshold;

[0116] The steps for generating the stirring parameter adjustment instruction include:

[0117] The dispersion uniformity index is calculated based on the peak width of the seed particle distribution sub-curve;

[0118] Establish a mapping relationship between the dispersion uniformity index and the inclination angle of the stirring blade to generate the inclination compensation value;

[0119] Calculate the speed gradient adjustment step size based on the current stirring speed and material viscosity;

[0120] In step S4, the mixing weight of the seed component in the decomposition model is taken according to the peak ratio calculation and the current reference threshold to determine whether the stirring adjustment is triggered. As peak proportion , in response to operating condition fluctuations, design the current reference threshold based on sliding window statistics The process includes:

[0121] The mixing weight represents the proportion of seed particles in the overall distribution and is used to measure the abundance of seed particles. The current reference threshold is adaptively adjusted based on historical data.

[0122] ,

[0123] in, Represents the mixing weight of the seed component, that is, the weight of the component in the Gaussian mixture model;

[0124] The window statistics radius is set to Sampling times, calculate the mean and standard deviation of the historical peak ratio:

[0125] ,

[0126] in, Indicates the length of the historical sampling window, Indicates time The peak proportion of and are the mean and standard deviation within the window respectively;

[0127] Based on statistical results, set safety factors After that, the current reference threshold is:

[0128] ,

[0129] in, Indicates the safety factor, used to adjust the trigger sensitivity. Indicates the current reference threshold;

[0130] When the real-time peak ratio Below the current reference threshold When the stirring parameter adjustment instruction is generated immediately, the seed crystal abundance is restored;

[0131] Specifically, the Gaussian mixture model is used to directly output the mixing weight of the seed component. As the percentage of the seed peak, to avoid the computational overhead caused by the secondary numerical integration, the current reference threshold The historical trend and fluctuation information are introduced through sliding window statistics, and the safety factor Flexible settings allow the judgment rules to quickly respond to seed shortages while also resisting short-term abnormal interference. Compared to fixed thresholds, dynamic thresholds are more adaptive and suitable for different batches and working conditions, significantly reducing false triggering and missed triggering.

[0132] Step S5: Send the stirring parameter adjustment instruction to the stirring drive device to dynamically adjust the stirring blade inclination angle and speed gradient.

[0133] This embodiment also provides a material particle size monitoring system for basic cobalt carbonate production, comprising:

[0134] A multi-source sensing module, comprising a laser scattering probe array, an ultrasonic transceiver, and a stirring shaft encoder mounted on the side wall of the reactor;

[0135] The arrangement of the laser scattering probe array is as follows:

[0136] Three detection planes are set up in the axial height direction of the reactor, with three probes distributed in a 120° ring in each plane;

[0137] The angle between the incident light path of each probe and the center line of the stirring shaft is 15°-30°, and the adjacent layers of probes are staggered 40° in the circumferential direction;

[0138] a signal processing module connected to the multi-source sensing module and configured to perform noise reduction processing on the raw sensing data and generate a first particle size distribution curve and a second particle size distribution curve;

[0139] The signal processing module includes:

[0140] Adaptive filtering unit, used to eliminate periodic mechanical vibration noise caused by the rotation of the mixing blade;

[0141] Frequency domain compensation unit dynamically adjusts the frequency band weight coefficient of the ultrasonic attenuation spectrum according to real-time solid content data;

[0142] A data synchronization unit is used to synchronize data and keep the acquisition time stamp deviation of the laser scattering signal and the ultrasonic signal within 10ms;

[0143] In the frequency domain compensation unit, the design is based on the real-time solid content Dynamic adjustment mechanism of frequency band weights:

[0144] Divide the spectrum into frequency band, for the Solid content sensitivity coefficient corresponding to each frequency band , and generate unnormalized weights through exponential mapping:

[0145] ,

[0146] in, Indicates the The solid content sensitivity coefficient corresponding to each frequency band is: Indicates the current real-time measured solid content of the material. Indicates the reference solid content level, Indicates frequency band The unnormalized weights of , Indicates the total number of frequency bands;

[0147] Normalize all frequency bands to get dynamic weights:

[0148] ,

[0149] in, Indicates the The normalized weight of each frequency band,

[0150] represents the sum of all unnormalized weights;

[0151] Apply this weight to the original ultrasonic attenuation spectrum coefficient , get the compensated frequency band attenuation value:

[0152] ,

[0153] in, Indicates the Frequency band original ultrasonic attenuation spectrum coefficient, Indicates the frequency band attenuation value used for particle size inversion after compensation;

[0154] The sensitivity of the dynamic weight to solid content can be derived:

[0155] ,

[0156] in, Indicates the The rate of change of frequency band weight to solid content, It is the sum of weighted sensitivities of all frequency bands;

[0157] Specifically, the mechanism introduced here uses logarithmic exponential mapping to embed the information of the real-time solid content deviation from the reference value into the frequency band weight, ensuring that as the solid content increases or decreases, each frequency band automatically obtains gain or attenuation compensation. The exponential mapping and normalization steps ensure that the weights are always positive and sum to 1, thereby not changing the total energy of the spectrum and the solid content sensitivity coefficient. Based on experimental calibration, the response differences of different frequency bands to solid content can be flexibly adjusted through this parameter. The derivative formula of the quadratic expansion provides the gradient direction for online adaptive learning, which can be further used for subsequent sensitivity coefficient fine-tuning or closed-loop optimization. The overall solution is simple to calculate and has strong real-time performance. It does not require frequent retraining and only needs to read the current solid content to complete compensation, providing a more stable ultrasonic signal input for particle size distribution inversion.

[0158] a distribution decomposition module, integrating a multimodal distribution decomposition model, receiving a first particle size distribution curve and a second particle size distribution curve and outputting a decomposition result;

[0159] a control decision module, connected to the distribution decomposition module, and configured to generate stirring parameter adjustment instructions based on the peak proportion;

[0160] The control decision module also includes:

[0161] The emergency intervention unit automatically triggers the seed replenishment device to start when it detects that the peak particle size of the new particle distribution sub-curve exceeds 1.5 times the peak particle size of the seed particles;

[0162] The discharge port of the seed feeding device is equipped with an ultrasonic atomizing head, and the atomized particle size is controlled in the range of 50-100μm;

[0163] The actuator interface module converts the stirring parameter adjustment instruction into a driving signal and sends it to the stirring drive device.

[0164] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for monitoring the particle size of materials used in the production of basic cobalt carbonate, characterized in that: include, Step S1, real-time collection of multi-source sensor data in the reactor, wherein the multi-source sensor data includes a laser scattering intensity angular distribution signal, an ultrasonic attenuation spectrum signal, and a stirring shaft speed signal; Step S2, generating a first particle size distribution curve based on the laser scattering intensity angular distribution signal, and generating a second particle size distribution curve based on the ultrasonic attenuation spectrum signal; Step S3, inputting the first particle size distribution curve and the second particle size distribution curve into a multimodal distribution decomposition model, and outputting a decomposed seed particle distribution sub-curve and a newly formed particle distribution sub-curve; The construction of the multimodal distribution decomposition model includes: Gaussian mixture modeling is performed on the particle size distribution curve in historical production data to constrain the seed particle size distribution to conform to the preset nominal distribution characteristics; Use KL divergence to dynamically evaluate the deviation between the real-time distribution curve and the historical distribution curve. When the deviation exceeds the set critical value, the distribution curve split operation is triggered. The output includes the decomposition results of the seed particle distribution sub-curve and the new particle distribution sub-curve; The multimodal distribution decomposition model includes a feature alignment layer, a cross attention layer, and a physical constraint layer, wherein: The feature alignment layer performs time stamp synchronization and spatial registration on the laser scattering intensity angular distribution signal and the ultrasonic attenuation spectrum signal; The cross-attention layer calculates the correlation weight matrix between the laser scattering intensity angular distribution signal and the ultrasonic attenuation spectrum signal in the frequency-space domain; The physical constraint layer enforces the particle volume conservation equation; The distribution curve splitting operation comprises the following steps: a) Detect the peak-to-valley ratio of the real-time distribution curve. When the valley bottom value reaches more than 20% of the peak height on both sides, it is determined to be a potential double peak; b) performing second-order derivative zero-crossing detection on the potential double-peak region to locate the center of the sub-peak; c) dynamically increasing the number of components of the Gaussian mixture model according to the sub-peak spacing; Step S4, calculate the peak ratio of the seed particle distribution sub-curve, and when the peak ratio is lower than a preset threshold, generate a stirring parameter adjustment instruction; take the mixing weight of the seed component in the multimodal distribution decomposition model As peak proportion ; Step S5: sending the stirring parameter adjustment instruction to the stirring drive device to dynamically adjust the stirring blade inclination angle and speed gradient.

2. A material particle size monitoring method for basic cobalt carbonate production according to claim 1, characterized in that: In step S3, the Gaussian mixture modeling of the historical particle size distribution and the seed nominal distribution constraint are as follows: The particle size variable after centering use Gaussian components to construct the mixture density function: ， in, Indicates the The weights of the Gaussian components, represents the total number of components of the Gaussian mixture model, Indicates the The variance of the components, Indicates the The mean of the components, represents the normalized and centered particle size variable, Indicates that the mixed model has a particle size The probability density at ; The introduced pair number is The component and nominal distribution of seed crystals The KL divergence constraint of is used to make the component close to the preset feature. The overall optimization goal is set to be the sum of log-likelihood and constraint divergence: ， in, represents the total number of historical samples, Indicates the The particle size of the samples, represents the constraint strength coefficient, represents the Kullback-Leibler divergence between two normal distributions, represents the nominal mean value of the seed crystal, represents the nominal variance of the seed crystal, represents the number assigned to the seed feature, and Respectively represent The mean and variance of each component; , where the denominator and The variance of the distribution being compared before and after, represents the square of the mean deviation of the two distributions, and the constant 1 is used to standardize the divergence baseline; Adopting improved EM iterative minimization ,include: Step E: Calculate the posterior probability: ， M step: ， in, Indicates the The samples belong to The posterior probability of the components, Indicates the Gaussian components in The density value at represents the particle size measurement value of the i-th historical sample after normalization and centering, Indicates the The weight of the component after the M-step update in this round, represents the updated mean, represents the updated variance; right Component, additional gradient correction after M steps: ， in, Represents the learning rate used for regular gradient correction, and iterates until , Indicates that it belongs to the seed classification index The component of is updated in this round of M steps and the mean value after gradient correction, Indicates that it belongs to the seed classification index The variance of the component after being updated in this round of M steps and corrected by the gradient.

3. A material particle size monitoring method for basic cobalt carbonate production according to claim 2, characterized in that: The step of generating the stirring parameter adjustment instruction includes: Calculating a dispersion uniformity index based on the peak width of the seed particle distribution sub-curve; Establishing a mapping relationship between the dispersion uniformity index and the inclination angle of the stirring blade to generate an inclination angle compensation amount; Based on the current stirring speed and material viscosity, the speed gradient adjustment step is calculated.

4. A material particle size monitoring method for basic cobalt carbonate production according to claim 3, characterized in that: In step S4, the mixing weight of the seed component in the decomposition model is taken according to the peak ratio calculation and the current preset threshold to determine whether the stirring adjustment is triggered. As peak proportion , in response to operating condition fluctuations, design the current preset threshold based on sliding window statistics The process includes: The mixing weight represents the proportion of seed particles in the overall distribution and is used to measure the abundance of seed particles. The current preset threshold is adaptively adjusted based on historical data. ， in, Represents the mixing weight of the seed component, that is, the weight of the component in the Gaussian mixture model; The window statistics radius is set to Sampling times, calculate the mean and standard deviation of the historical peak ratio: ， in, Indicates the length of the historical sampling window, Indicates time The peak proportion of and are the mean and standard deviation within the window respectively; Based on statistical results, set safety factors After that, the current preset threshold is: ， in, Indicates the safety factor, used to adjust the trigger sensitivity. Indicates the current preset threshold; When the real-time peak ratio Below the current preset threshold When the stirring parameter adjustment instruction is generated immediately, the seed crystal abundance is restored.

5. A material particle size monitoring system for basic cobalt carbonate production, based on a material particle size monitoring method for basic cobalt carbonate production according to any one of claims 1 to 4, characterized in that: include: A multi-source sensing module, comprising a laser scattering probe array, an ultrasonic transceiver, and a stirring shaft encoder mounted on the side wall of the reactor; a signal processing module connected to the multi-source sensing module and configured to perform noise reduction processing on the raw sensing data and generate a first particle size distribution curve and a second particle size distribution curve; a distribution decomposition module, integrating the multimodal distribution decomposition model, receiving the first particle size distribution curve and the second particle size distribution curve and outputting a decomposition result; a control decision module, connected to the distribution decomposition module, and configured to generate a stirring parameter adjustment instruction according to the peak proportion; The actuator interface module converts the stirring parameter adjustment instruction into a driving signal and sends it to the stirring driving device.

6. A material particle size monitoring system for basic cobalt carbonate production according to claim 5, characterized in that: The arrangement of the laser scattering probe array is as follows: Three detection planes are set up in the axial height direction of the reactor, with three probes distributed in a 120° ring in each plane; The angle between the incident light path of each probe and the center line of the stirring shaft is 15°-30°, and the probes of adjacent layers are arranged with a 40° offset in the circumferential direction.

7. A material particle size monitoring system for basic cobalt carbonate production according to claim 5, characterized in that: The signal processing module includes: Adaptive filtering unit, used to eliminate periodic mechanical vibration noise caused by the rotation of the mixing blade; Frequency domain compensation unit dynamically adjusts the frequency band weight coefficient of the ultrasonic attenuation spectrum according to real-time solid content data; The data synchronization unit is used for data synchronization.

8. A material particle size monitoring system for basic cobalt carbonate production according to claim 5, characterized in that: The control decision module also includes: The emergency intervention unit automatically triggers the seed replenishment device to start when it detects that the peak particle size of the new particle distribution sub-curve exceeds 1.5 times the peak particle size of the seed particles; The discharge port of the seed adding device is provided with an ultrasonic atomizing head, and the atomized particle size is controlled in the range of 50-100 μm.

Images