A method and system for controlling the quality of crystallization in carbon source production

Abstract

The application relates to the technical field of crystallization quality control, in particular to a crystallization quality control method and a control system in carbon source production, which comprises the following steps: monitoring various related parameter data in each observation period in a crystallization process in real time during carbon source production; based on principal component analysis and an entropy weight method, key variables in all related parameter data are extracted and crystallization impurity stability is calculated; a sliding window mean-standard deviation method and dynamic time warping are used to quantify process fluctuation intensity and historical consistency, and dynamic coincidence degree scores are calculated; abnormal states in the crystallization process before the current collection moment are identified by using the dynamic coincidence degree scores, so that model predictive control is used to realize real-time optimization control of parameters and multivariate collaborative supervision control of the crystallization process. The application aims to solve the problems of uneven crystal particle size, unstable crystal form and impurity residue in the traditional crystallization process, and realizes multivariate collaborative supervision control of the crystallization process.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of crystallization quality control, in particular to a crystallization quality control method and a control system in carbon source production. BACKGROUND

[0002] Crystallization quality control in carbon source production is a core link in the chemical industry. Traditional crystallization process relies on experience to control parameters such as temperature and concentration, which is easily disturbed by kinetic and thermodynamic factors, resulting in problems such as uneven crystal size, unstable crystal form and impurity residues. With the development of fine industry and the upgrading of green manufacturing demand, crystallization process control technology has evolved from manual experience to intelligent supervision and control.

[0003] In the carbon source production process, the stability of crystal growth is disturbed by residual metal ions, organic by-products and incomplete removal of solvents in raw materials. These impurities destroy the integrity of the crystal structure through various mechanisms. For example, impurity molecules are easily adsorbed on active sites on the crystal surface, hindering the directional and orderly accumulation of solute molecules, resulting in unbalanced crystal face growth rate and abnormal crystal morphology. In addition, solvent residues may induce the formation of non-target crystal forms by changing the crystallization environment, or compete for adsorption sites with solute molecules, forcing the crystal to exhibit abnormal morphology such as needles and sheets. Conventional monitoring equipment can only provide static data and cannot accurately predict and control the dynamic interference of impurities. This limitation makes it difficult for existing technology to fundamentally solve the quality problems of morphology abnormalities and batch inconsistencies caused by impurity interference in the crystal growth process. SUMMARY

[0004] To solve the above technical problems, the application provides a crystallization quality control method and a control system in carbon source production, and the technical solutions are as follows:

[0005] In the first aspect, one embodiment of the application provides a crystallization quality control method in carbon source production, which comprises the following steps:

[0006] Real-time monitoring of various related parameter data in each observation period during crystallization in carbon source production;

[0007] Calculating the weight of each related parameter data allocated by the entropy weight method, and weighting and summing the variation coefficients of the corresponding related parameter data to obtain a comprehensive weight; extracting the principal component scores and variance contribution rates of two principal components in all related parameter data in an observation period by principal component analysis, and calculating the crystallization impurity stability in combination with the comprehensive weight;

[0008] The calculated crystallization impurity stability of all observation periods is composed into a sequence, and a sliding time window is set; the ratio of the mean value and the standard deviation in each time window, and the DTW distance between the elements in the corresponding time window and the elements in the sequence are fused, and the fusion result is taken as the dynamic fitness score;

[0009] The dynamic fitness score is used to identify the abnormal state in the crystallization process before the current collection time, so as to realize real-time optimization of the control parameters through model predictive control, and realize multivariate collaborative supervision control of the crystallization process.

[0010] Preferably, the observation period is a period within a preset time length before a collection time.

[0011] Preferably, the related parameter data includes temperature, solute concentration, metal ion concentration and solvent residual rate.

[0012] Preferably, the calculation method of the crystallization impurity stability is:

[0013] The variance contribution rate of each principal component after normalization is weighted and summed with the principal component score of the principal component;

[0014] The comprehensive weight is inversely proportionally mapped;

[0015] The result of the weighted sum of the principal component scores and the result of the inverse proportional mapping are positively fused to obtain the crystallization impurity stability.

[0016] Preferably, the acquisition method of the dynamic fitness score is further determined as:

[0017]

[0018] Wherein, B is the dynamic fitness score, M is the number of time windows, μ j and σ j are the mean value and the standard deviation in the jth time window, DTW j is the DTW distance between the elements in the jth time window and the elements in the sequence, Norm() is a normalization function, and e is a natural constant.

[0019] Preferably, the crystallization state includes a stable interval, a warning interval and an abnormal interval.

[0020] Preferably, the method for identifying the abnormal state in the crystallization process before the current collection time by using the dynamic fitness score is:

[0021] When the dynamic fitness score is less than a preset warning threshold, the crystallization process before the current collection time is an abnormal state.

[0022] Preferably, when the crystallization process before the current acquisition time is in an abnormal state, the method for real-time optimization of control parameters by model predictive control is:

[0023] If the standard deviation calculated by the observation period at the current acquisition time as a time window suddenly increases, the standard deviation calculated by the observation period at the current acquisition time as a time window and the stirring rate data at the current acquisition time are used as the input of the model predictive control, and the control signal is output to the stirring equipment of the crystallization kettle to control the stirring rate and the solution circulation flow rate;

[0024] If the DTW distance calculated by the observation period at the current acquisition time as a time window suddenly increases, the DTW distance value calculated by the observation period at the current acquisition time as a time window and the crystallization temperature gradient data at the current acquisition time are used as the input of the model predictive control, and the control signal is output to the temperature control equipment to dynamically control the crystallization temperature gradient;

[0025] If the mean value calculated by the observation period at the current acquisition time as a time window suddenly decreases, the mean value calculated by the observation period at the current acquisition time as a time window and the solvent residual rate data at the current acquisition time are used as the input of the model predictive control, and the control signal is output to the solvent recovery unit to control the evaporation pressure and the degassing time.

[0026] Preferably, the judgment criteria for sudden increase or sudden decrease are:

[0027] The standard deviation, DTW distance, and mean value are all recorded as dominant factors;

[0028] For the observation period at the current acquisition time as a time window, the historical data mean value of each acquisition time in the time window under various dominant factors is calculated, and linear fitting is performed on all historical data mean values of various corresponding dominant factors in the time window to obtain the linear regression slope of the corresponding dominant factor;

[0029] The historical data mean value of each acquisition time in the time window under various dominant factors is the average value of the corresponding dominant factor of the acquisition time and all previous acquisition times in the time window;

[0030] If the linear regression slope of various dominant factors is less than the preset decrease judgment threshold, it is determined that the corresponding dominant factor is suddenly decreased;

[0031] If the linear regression slope of various dominant factors is greater than the preset increase judgment threshold, it is determined that the corresponding dominant factor is suddenly increased.

[0032] In a second aspect, the other embodiment of the present application also provides a crystallization quality control system in carbon source production, which realizes the crystallization quality control method in carbon source production.

[0033] 1. An initialization configuration module for system initialization and parameter configuration;

[0034] 2. A data stream and functional module chain, including three sub-modules, namely: a process parameter real-time monitoring and signal optimization module, a multi-dimensional feature analysis and dominant factor identification module, and a process stability dynamic diagnosis and root cause tracking module;

[0035] The process parameter real-time monitoring and signal optimization module is used for real-time monitoring of various related parameter data in each observation period during the crystallization process during carbon source production;

[0036] The multi-dimensional feature analysis and dominant factor identification module is used for calculating the weight of each related parameter data by using the entropy weight method, and performing weighted summation on the variation coefficients of the corresponding related parameter data to obtain a comprehensive weight; the principal component scores and variance contribution rates of two principal components in all related parameter data in an observation period are extracted by using principal component analysis, and the crystalline impurity stability is calculated in combination with the comprehensive weight;

[0037] The process stability dynamic diagnosis and root cause tracking module is used for grouping all crystalline impurity stabilities calculated in all observation periods to form a sequence, and setting a sliding time window; the ratio of the mean value and the standard deviation in each time window, and the DTW distance between the elements in the corresponding time window and the elements in the sequence are fused, and the fusion result is taken as a dynamic fitness score;

[0038] 3. A supervisory control core module for identifying abnormal states in the crystallization process before the current collection time by using the dynamic fitness score, so as to realize real-time optimization control of the parameters by model predictive control, and to realize multivariate collaborative supervision control of the crystallization process.

[0039] The present application has at least the following beneficial effects:

[0040] (1) For the quality problems of crystal form abnormality and particle size unevenness caused by impurity adsorption and solvent residue in the crystallization process, the present application dynamically quantifies the crystallization quality stability characteristics by using principal component analysis combined with the entropy weight method, balances the local disturbance of the thermodynamic dominant factor and the impurity interference by fusing the principal component variance contribution rate and the real-time parameter state, and solves the problem of unbalanced crystal face growth rate;

[0041] (2) For batch instability caused by process fluctuation and historical mode deviation, the sliding window mean-standard deviation method and the dynamic time warping are used to comprehensively reflect the process stability and historical consistency characteristics, and to exclude the influence of process random fluctuation and non-target crystal form generation;

[0042] (3) Through model predictive control under the supervision control framework, real-time optimization of temperature, flow, pressure and other parameters is realized, multivariable collaborative closed-loop supervision control is realized, process stability is improved, and differential regulation strategies are triggered based on dynamic conformity score, realizing multivariable collaborative closed-loop control of carbon source crystallization process. Ultimately, the uniformity of crystal size, the stability of crystal form and the control precision of solvent residue are improved, and the batch consistency of products is ensured. BRIEF DESCRIPTION OF DRAWINGS

[0043] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present application or the prior art, the drawings needed in the description of the embodiments or the prior art will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and those skilled in the art can obtain other drawings according to these drawings without creative labor.

[0044] Figure 1 A flow chart of a crystallization quality control method in carbon source production provided by an embodiment of the present application;

[0045] Figure 2 A flow chart of a multivariable collaborative supervision control method of a crystallization process provided by an embodiment of the present application. DETAILED DESCRIPTION

[0046] An embodiment of the present application provides a crystallization quality control method in carbon source production, specifically referring to Figure 1 The method comprises the following steps:

[0047] Step one: Real-time monitoring of various related parameter data in each observation period in the crystallization process during carbon source production.

[0048] In the carbon source production crystallization quality control system, a high-precision temperature sensor is installed at the top of the crystallization kettle to collect real-time temperature data during crystallization for monitoring the thermodynamic changes in the crystallization process; an online concentration detector is integrated in the solution circulation pipeline to obtain real-time solute concentration data for evaluating the influence of solute supersaturation on crystal growth; an inductively coupled plasma (ICP) detector is used to continuously detect the raw material liquid to collect metal ion concentration data during the crystallization process for identifying the interference of impurity adsorption on crystal face growth; a flow meter and gas chromatography combined equipment are installed in the solvent recovery unit to collect solvent residual rate (S) data for analyzing the potential threat of solvent residue to crystal form stability; the data collection frequency of all collection equipment is set to 10 Hz, and real-time data collection is realized through high-precision sensors and online detectors to provide high-frequency input for supervision and control, ensuring comprehensive monitoring of process parameters. Among them, temperature, solute concentration, metal ion concentration and solvent residual rate are used as various related parameter data during the crystallization process during carbon source production.

[0049] For the raw data collected in the above, in the supervision and control process, a wavelet denoising algorithm is used to eliminate high-frequency noise, and Z-score standardization is used to unify the dimension. Each collection time is set as an observation period, and the preset time is set to 1 hour in this embodiment. For collection times with less than one hour of data, the mean filling method is used to fill the previous data to ensure the integrity of the observation period data. The preprocessed temperature sequence, solute concentration sequence, metal ion concentration sequence and solvent residual rate sequence in all observation periods are obtained to provide high signal-to-noise ratio input for the subsequent supervision and control model, thereby systematically solving the crystallization quality problems caused by impurity interference, crystal form abnormalities and solvent residue.

[0050] Step two: based on principal component analysis and entropy weight method, extracting key variables from all related parameter data and calculating crystallization impurity stability.

[0051] During the carbon source production process, impurities such as residual metal ions and organic by-products in the raw materials, as well as unremoved solvents, can be adsorbed on the active sites of the crystal surface or change the crystallization environment, leading to problems such as uneven crystal size, unstable crystal form and solvent residue.

[0052] Therefore, the entropy weight method is used to calculate the weight of each related parameter data, and the weighted sum of the variation coefficients of the corresponding related parameter data is obtained to obtain the comprehensive weight; principal component analysis is used to extract the principal component scores and variance contribution rates of the two principal components in all related parameter data in an observation period, and the comprehensive weight is combined to calculate the crystallization impurity stability.

[0053] The calculation method of the crystallization impurity stability is: the variance contribution rate of each principal component after normalization is weighted and summed with the principal component score of the principal component; the comprehensive weight is inversely proportionally mapped; the result of the weighted sum of the principal component scores and the result of the inverse proportional mapping are positively fused to obtain the crystallization impurity stability.

[0054] Optionally, the inverse proportional mapping can be realized by a negative linear mapping, a negative exponential mapping or a method of setting an adjustment parameter.

[0055] It can be understood that the positive fusion is a fusion method such as addition and multiplication between data, and the specific positive fusion method is determined by the implementer according to the actual situation, and the present application does not make special limitations.

[0056] In the embodiment, the temperature sequence, the solute concentration sequence, the metal ion concentration sequence and the solvent residual rate sequence after preprocessing in one observation period are taken as inputs, the principal component analysis (PCA) algorithm is used to extract key variables based on the demand of supervised control, the number of principal components is set to 2, the principal component scores PC1 and PC2 of two principal components and the variance contribution rates V1 and V2 corresponding thereto are output. Then, the four sequences after standardization are taken as inputs, the weight distribution results of each sequence are output by calculating the information entropy of each sequence using the entropy weight method, and the weight value quantifies the significance of the influence of the variation of each parameter on the quality. The principal component analysis algorithm is a known technology and will not be described again.

[0057] Based on the above analysis, the crystallization impurity stability A in each observation period is calculated, and the specific calculation relationship is:

[0058]

[0059] V1 and V2 are the variance contribution rates of the two principal components, respectively, which explain the proportion of the total variation in the original data set, V1 corresponds to the most significant parameter in the crystallization process, and V2 corresponds to the secondary but non-negligible factor. After normalization, V1 and V2 are used as the weights of PC1 and PC2, which can dynamically adjust the contribution of the crystallization impurity stability according to the proportion of the principal component explaining the data variation, so that the crystallization impurity stability focuses on the dominant factor (such as the thermodynamic state) while taking into account the influence of the secondary factor (such as impurity interference), thereby balancing the global trend and local disturbance of the crystallization quality stability; PC1 and PC2 are the principal component scores of the two principal components, and the greater the principal component score is, the higher the matching degree of the current process parameter combination with the main variation mode extracted by PCA is, which represents that the crystallization process is more stable, and vice versa.

[0060] wherein, L is the comprehensive weight, which is the weight ω i and the coefficient of variation Ki the result of weighted summation of ω i the result of weight distribution of the i th sequence, K i the coefficient of variation of the i th sequence, ω i ×K i indicates the contribution of the i th parameter to the comprehensive weight L, the greater the product, the more likely the i th parameter is the dominant factor affecting the quality, and the sum of the product represents the comprehensive influence of all parameters on the crystallization quality, the greater the sum, the greater the influence of parameter change on the crystallization quality, i is the summation index, which is 1 to 4, the coefficient of variation is a known technology and is not described here.

[0061] It should be understood that the crystallization impurity stability A is dynamically quantified by fusing the potential variation pattern extracted by PCA and the real-time parameter state weighted by entropy weight method, and the greater the value, the more stable the crystallization quality, and vice versa, which indicates the risk of crystal form abnormality or solvent residue.

[0062] Step three: using the sliding window mean-standard deviation method and dynamic time warping to quantify the process fluctuation intensity and historical consistency, and calculating the dynamic coincidence score.

[0063] Due to the imbalance of crystal growth rate and crystal form abnormality caused by impurity adsorption, solvent residue and other factors in the carbon source crystallization process, the monitoring parameters will mutate or deviate from the stable trend due to impurity adsorption, thermodynamic condition fluctuation or solvent residue induction, resulting in unstable crystal face growth rate and non-target crystal form generation, process fluctuation and target state deviation, which will reduce the principal component score and increase the comprehensive weight, resulting in the decrease of crystallization impurity stability A value, reflecting the deterioration of crystallization quality stability.

[0064] Therefore, all the crystallization impurity stabilities calculated in all observation periods are combined to form a sequence, and a sliding time window is set; the ratio of the mean and standard deviation in each time window, and the DTW distance between the elements in the corresponding time window and the elements in the sequence are fused, and the fusion result is taken as the dynamic coincidence score.

[0065] It can be understood that fusion can be divided into forward fusion and reverse fusion, and the reverse fusion method is adopted in this embodiment. Reverse fusion is a fusion method between data such as subtraction and division, and the specific reverse fusion method is determined by the implementer according to the actual situation, and the present application does not make special restrictions.

[0066] In the embodiment, for the calculated crystallization impurity stability A in each observation period, the crystallization impurity stability A of all observation periods is composed into a sequence, the sequence is taken as an input, and the process fluctuation intensity is quantified in real time by using a sliding window mean-standard deviation method. Due to the characteristics of short response period of crystallization kinetics and high-frequency monitoring, the sliding window length is set to 1 minute in the embodiment, and the mean value μ and the standard deviation σ in the output window are output, wherein the mean value μ reflects the average level of the current crystallization quality, and the standard deviation σ quantifies the process fluctuation intensity. The greater the standard deviation is, the worse the process stability is.

[0067] Subsequently, all crystallization impurity stabilities in each time window and all crystallization impurity stabilities in the sequence are taken as inputs, and a dynamic time warping distance is output between all crystallization impurity stabilities in each time window and all crystallization impurity stabilities in the sequence. The closer the DTW distance is to 0, the higher the matching degree of the data in the time window with the historical crystallization quality is.

[0068] Based on the above analysis, a dynamic coincidence degree score B is constructed. Specifically, the relationship is calculated as:

[0069]

[0070] Wherein, M is the number of time windows, μ j and σ j are the mean value and the standard deviation in the jth time window, the ratio between the two combines the quality level and the process stability, and represents the quality score after fluctuation adjustment. The greater the value is, the higher the crystallization quality in the time window is and the smaller the fluctuation is, and the more stable the crystallization process is; DTW j is the DTW distance between the elements in the jth time window and the elements in the sequence, which measures the shape matching degree of the crystallization impurity stability in the time window with the sequence composed of all crystallization impurity stabilities. The smaller the value is, the more similar the shape of the current window is to the historical data, that is, the current crystallization process is highly consistent with the historical crystallization process, the crystalline growth mode is more normal, and Norm() is a normalization function, and e is a natural constant.

[0071] It should be understood that the dynamic coincidence degree score B quantifies the process stability and the historical consistency, provides a dynamic decision basis for closed-loop control, quantifies the crystallization quality by the quality level and the matching degree, and the greater the value is, the more stable the crystallization process is, and the thermodynamic state, impurity control, and solvent residue all meet the expectations and the crystal form is uniform.

[0072] Step four: using the dynamic coincidence degree score, identifying the abnormal state in the crystallization process before the current collection time, to optimize the control parameters in real time by model predictive control, and realizing the multivariate collaborative supervision control of the crystallization process.

[0073] Then, according to historical stable production batch data, a reference range of the dynamic goodness-of-fit score B is determined, and specifically, a stable threshold and a warning threshold are preset, wherein the stable threshold is greater than the warning threshold, and in this embodiment, the stable threshold is 0.8, and the warning threshold is 0.5.

[0074] That is, when the dynamic goodness-of-fit score is greater than or equal to the stable threshold, the crystallization process before the current collection time is in a stable state; when the dynamic goodness-of-fit score is greater than or equal to the warning threshold and less than the stable threshold, the crystallization process before the current collection time is in a warning state; and when the dynamic goodness-of-fit score is less than the warning threshold, the crystallization process before the current collection time is in an abnormal state.

[0075] Accordingly, the present application uses the dynamic goodness-of-fit score to identify the abnormal state of the crystallization process before the current collection time.

[0076] Further, based on a supervisory control framework, the present application uses model predictive control to optimize control parameters in real time, forms a closed-loop supervisory control, solves the optimal control parameter combination in a future time window in real time, and realizes accurate regulation and control through an actuator according to the dominant factor of the deviation of the crystallization quality.

[0077] When the crystallization process before the current collection time is in an abnormal state, the dominant factor of the deviation of the current crystallization quality is located according to the composition elements of the dynamic goodness-of-fit score B, and the dominant factor is used to optimize control parameters in real time through model predictive control, specifically:

[0078] If the standard deviation calculated by taking the observation period at the current collection time as a time window suddenly increases, it indicates that the process fluctuation is intensified, and the standard deviation calculated by taking the observation period at the current collection time as a time window and the stirring rate data at the current collection time are used as the input of model predictive control, and a control signal is output to the crystallizer stirring device to control the stirring rate and the solution circulation flow rate, so as to improve the stirring uniformity and optimize the solution circulation flow rate.

[0079] If the DTW distance calculated by taking the observation period at the current collection time as a time window suddenly increases, the DTW distance value calculated by taking the observation period at the current collection time as a time window and the crystallization temperature gradient data at the current collection time are used as the input of model predictive control, and a control signal is output to the temperature regulation device to dynamically regulate the crystallization temperature gradient, that is, to reduce the cooling rate to inhibit the imbalance of crystal face growth and guide directional crystallization.

[0080] If the mean value calculated by the observation period of the current collection time as a time window suddenly decreases, the mean value calculated by the observation period of the current collection time as a time window and the solvent residual rate data of the current collection time are used as the input of model predictive control, and the control signal is output to the solvent recovery unit to control the evaporation pressure and degassing time, that is, to increase the evaporation pressure or prolong the degassing time, so as to reduce the influence of solvent residue on the crystal form.

[0081] It is worth noting that the "sudden increase" or "sudden decrease" in the above process can be determined by the implementer, such as using threshold method, sliding window trend analysis method and other methods for judgment.

[0082] In this application, the standard deviation, DTW distance and mean value are all recorded as dominant factors; for the observation period of the current collection time as a time window, the historical data mean value of each collection time in the time window under various dominant factors is calculated, and all historical data mean values calculated in the time window under various corresponding dominant factors are linearly fitted to obtain the linear regression slope of the corresponding dominant factor; wherein the historical data mean value of each collection time in the time window under various dominant factors is the average value of the corresponding dominant factor of all collection times in the time window and before the time window; if the linear regression slope of various dominant factors is less than the preset decrease judgment threshold, it is determined that the corresponding dominant factor is suddenly decreased; if the linear regression slope of various dominant factors is greater than the preset increase judgment threshold, it is determined that the corresponding dominant factor is suddenly increased.

[0083] Specifically, the mean value calculated by the observation period of the current collection time as a time window is taken as an example, the historical data mean value of each collection time in the time window is calculated, and all historical data mean values calculated in the time window are linearly fitted to obtain the linear regression slope. Wherein the historical data mean value of each collection time in the time window is the average value of the mean value of all collection times in the time window and before the time window.

[0084] If the linear regression slope is less than the preset decrease judgment threshold, it is determined that the mean value is suddenly decreased. Wherein the preset decrease judgment threshold of this embodiment is-0.1.

[0085] By precisely controlling the stirring rate, temperature gradient and evaporation pressure and other parameters for various dominant factors, the impurity adsorption interference is effectively suppressed, the crystal face growth rate is balanced, and the solvent residue risk is reduced, thereby improving the crystal particle size uniformity, crystal form stability and batch consistency, and providing reliable guarantee for large-scale production of high-purity carbon source products. Among them, model predictive control (MPC) as the core algorithm of supervisory control, through rolling optimization and feedback correction, real-time approximation of the optimal process state, specific implementation as a known technology, not much more described here.

[0086] In the present application, the multivariate collaborative supervisory control method flow chart of the crystallization process is shown in the accompanying Figure 2

[0087] In another embodiment of the present application, a crystallization quality control system in carbon source production is also provided, which is designed in a modular closed loop and consists of the following core modules, each module cooperates through standardized interfaces, and the system comprises:

[0088] 1. Initialization configuration module, used for system initialization and parameter configuration;

[0089] 2. Data stream and function module chain, including three sub-modules, which are respectively: process parameter real-time monitoring and signal optimization module, multi-dimensional feature analysis and dominant factor identification module, process stability dynamic diagnosis and root cause tracking module;

[0090] The process parameter real-time monitoring and signal optimization module is used for real-time monitoring of various related parameter data in each observation period during the crystallization process in carbon source production;

[0091] The multi-dimensional feature analysis and dominant factor identification module is used for calculating the weight of each related parameter data by using entropy weight method, and performing weighted summation on the variation coefficient of the corresponding related parameter data to obtain a comprehensive weight; the principal component scores and variance contribution rates of two principal components in all related parameter data in an observation period are extracted by using principal component analysis, and the crystalline impurity stability is calculated in combination with the comprehensive weight;

[0092] The process stability dynamic diagnosis and root cause tracking module is used for forming a sequence of all crystalline impurity stabilities calculated in all observation periods, and setting a sliding time window; the ratio of the mean value and the standard deviation in each time window, and the DTW distance between the elements in the corresponding time window and the elements in the sequence are fused, and the fusion result is taken as a dynamic fitness score;

[0093] 3. Supervisory control core module, used for identifying the abnormal state in the crystallization process before the current collection time by using the dynamic fitness score, so as to realize real-time optimization control of the control parameters by model predictive control, and realize multivariate collaborative supervisory control of the crystallization process.

[0094] Other embodiments of the present application will be apparent to those skilled in the art upon consideration of the specification and practice of the application disclosed herein. The present application is intended to cover any variations, uses, or adaptations of the application following, in general, the principles of the application and including such departures from the present disclosure as come within known or customary practice in the art to which the application pertains.

[0095] ​It is to be understood that the application is not limited to the precise construction already described above and shown in the drawings, and that various modifications and changes can be made by those skilled in the art without departing from the scope thereof.

Claims

1. A method of controlling the crystal quality in the production of a carbon source, characterized by, The method comprises the following steps: Real-time monitoring of various related parameter data in each observation period in the crystallization process during carbon source production; Calculating the weight of each related parameter data by using the entropy weight method, and performing weighted summation on the variation coefficients of the corresponding related parameter data to obtain a comprehensive weight; extracting the principal component scores and variance contribution rates of two principal components in all related parameter data in an observation period by using principal component analysis, and combining the comprehensive weight to calculate the crystallization impurity stability; All crystallization impurity stabilities calculated in all observation periods are combined to form a sequence, and a sliding time window is set; the ratio of the mean value and the standard deviation in each time window, and the DTW distance between the elements in the corresponding time window and the elements in the sequence are fused, and the fusion result is taken as the dynamic goodness of fit score; Using the dynamic goodness of fit score, the abnormal state in the crystallization process before the current collection time is identified, so as to realize real-time optimization of the control parameters by model predictive control, and realize multivariate collaborative supervision control of the crystallization process.

2. The method of claim 1, wherein the carbon source is a hydrocarbon. The observation period is a period within a preset time before the current collection time.

3. The method of claim 1, wherein the carbon source is a hydrocarbon. The related parameter data includes temperature, solute concentration, metal ion concentration and solvent residual rate.

4. The method of claim 1, wherein the carbon source is a hydrocarbon. The calculation method of the crystallization impurity stability is: The variance contribution rate of each principal component is normalized, and the principal component scores of the principal components are weighted and summed; The comprehensive weight is inversely proportional to the mapping result; The result of the weighted summation of the principal component scores and the inverse proportional mapping result are positively fused to obtain the crystallization impurity stability.

5. The method of claim 1, wherein the carbon source is a hydrocarbon. 5 The acquisition method of the dynamic goodness of fit score is further determined as: where B is the dynamic anastomosis score, M is the number of time windows, μ j and σ j are the mean and standard deviation within the jth time window, DTW j is the DTW distance between an element within the jth time window and an element within the sequence, Norm() is a normalization function, and e is the natural constant.

6. The method of claim 1, wherein the carbon source is a hydrocarbon. 5 The crystallization state includes: stable interval, warning interval and abnormal interval.

7. The method for controlling crystal quality in carbon source production according to claim 6, wherein: The method for identifying the abnormal state in the crystallization process before the current collection time by using the dynamic goodness of fit score is: When the dynamic goodness of fit score is less than a preset warning threshold, the crystallization process before the current collection time is an abnormal state.

8. The method for controlling crystal quality in carbon source production according to claim 1, wherein: When the crystallization process before the current collection time is an abnormal state, the method for real-time optimization of the control parameters by model predictive control is: If the standard deviation calculated by taking the observation period of the current collection time as a time window suddenly increases, the standard deviation calculated by taking the observation period of the current collection time as a time window and the stirring rate data of the current collection time are taken as the input of the model predictive control, and the control signal is output to the stirring equipment of the crystallization kettle to control the stirring rate and the solution circulation flow; If the DTW distance calculated by taking the observation period of the current collection time as a time window suddenly increases, the DTW distance value calculated by taking the observation period of the current collection time as a time window and the crystallization temperature gradient data of the current collection time are taken as the input of the model predictive control, and the control signal is output to the temperature control equipment to dynamically control the crystallization temperature gradient; If the mean value calculated by taking the observation period of the current collection time as a time window suddenly decreases, the mean value calculated by taking the observation period of the current collection time as a time window and the solvent residual rate data of the current collection time are taken as the input of the model predictive control, and the control signal is output to the solvent recovery unit to control the evaporation pressure and the degassing time.

9. The method for controlling crystal quality in carbon source production according to claim 8, wherein: The criterion for judging the sudden increase or sudden decrease is: The standard deviation, DTW distance and mean are recorded as the dominant factors; For the observation period of the current collection time as a time window, the mean of the historical data of each collection time under various dominant factors in the time window is calculated, and the linear fitting of all historical data means corresponding to various dominant factors in the time window is carried out to obtain the linear regression slope of the corresponding dominant factor; Wherein, the mean of the historical data of each collection time under various dominant factors in the time window is the average value of the corresponding dominant factor of the collection time and all collection times before the time window; If the linear regression slope of various dominant factors is less than the preset decrease judgment threshold, it is determined that the corresponding dominant factor is suddenly decreased; If the linear regression slope of various dominant factors is greater than the preset increase judgment threshold, it is determined that the corresponding dominant factor is suddenly increased.

10. A crystallization quality control system in carbon source production, which realizes a crystallization quality control method in carbon source production according to any one of claims 1-9, and the system comprises:

1. An initialization configuration module for system initialization and parameter configuration; 2. A data stream and function module chain, including three sub-modules, which are respectively: a process parameter real-time monitoring and signal optimization module, a multi-dimensional feature analysis and dominant factor identification module, and a process stability dynamic diagnosis and root cause tracking module; The process parameter real-time monitoring and signal optimization module is used for real-time monitoring of various related parameter data in each observation period during the crystallization process during carbon source production; The multi-dimensional feature analysis and dominant factor identification module is used for calculating the weight allocated to each related parameter data by entropy weight method, and performing weighted summation on the variation coefficient of the corresponding related parameter data to obtain a comprehensive weight; The principal component scores and variance contribution rates of two principal components in all related parameter data in an observation period are extracted by principal component analysis, and the crystalline impurity stability is calculated in combination with the comprehensive weight; The process stability dynamic diagnosis and root cause tracking module is used for forming a sequence of all crystalline impurity stabilities calculated in all observation periods, and setting a sliding time window; the ratio of the mean and the standard deviation in each time window, and the DTW distance between the elements in the corresponding time window and the elements in the sequence are fused, and the fusion result is taken as a dynamic coincidence degree score; 3. A supervisory control core module for identifying abnormal states in the crystallization process before the current collection time by using the dynamic coincidence degree score, so as to realize multivariate collaborative supervisory control of the crystallization process by real-time optimization of control parameters through model predictive control.

Images