CN118193997A - Chlor-alkali electrolytic cell process abnormality detection method based on self-adaptive threshold - Google Patents
Chlor-alkali electrolytic cell process abnormality detection method based on self-adaptive threshold Download PDFInfo
- Publication number
- CN118193997A CN118193997A CN202410358974.0A CN202410358974A CN118193997A CN 118193997 A CN118193997 A CN 118193997A CN 202410358974 A CN202410358974 A CN 202410358974A CN 118193997 A CN118193997 A CN 118193997A
- Authority
- CN
- China
- Prior art keywords
- chlor
- variable
- alkali
- detected
- electrolytic cell
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 133
- 239000003513 alkali Substances 0.000 title claims abstract description 83
- 238000001514 detection method Methods 0.000 title claims abstract description 63
- 230000005856 abnormality Effects 0.000 title claims abstract description 53
- 230000002159 abnormal effect Effects 0.000 claims abstract description 55
- HEMHJVSKTPXQMS-UHFFFAOYSA-M Sodium hydroxide Chemical compound [OH-].[Na+] HEMHJVSKTPXQMS-UHFFFAOYSA-M 0.000 claims description 121
- 235000011121 sodium hydroxide Nutrition 0.000 claims description 42
- 239000011159 matrix material Substances 0.000 claims description 36
- HPALAKNZSZLMCH-UHFFFAOYSA-M sodium;chloride;hydrate Chemical compound O.[Na+].[Cl-] HPALAKNZSZLMCH-UHFFFAOYSA-M 0.000 claims description 31
- 239000012267 brine Substances 0.000 claims description 30
- 238000010606 normalization Methods 0.000 claims description 29
- 238000012549 training Methods 0.000 claims description 27
- 238000012795 verification Methods 0.000 claims description 23
- 238000012360 testing method Methods 0.000 claims description 14
- 230000003044 adaptive effect Effects 0.000 claims description 12
- 238000012545 processing Methods 0.000 claims description 12
- 150000002500 ions Chemical class 0.000 claims description 8
- 239000012528 membrane Substances 0.000 claims description 8
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 7
- VEXZGXHMUGYJMC-UHFFFAOYSA-N Hydrochloric acid Chemical compound Cl VEXZGXHMUGYJMC-UHFFFAOYSA-N 0.000 claims description 6
- 239000003518 caustics Substances 0.000 claims description 6
- 229910001424 calcium ion Inorganic materials 0.000 claims description 3
- 229910001425 magnesium ion Inorganic materials 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 abstract description 4
- 238000009776 industrial production Methods 0.000 abstract description 2
- 238000004519 manufacturing process Methods 0.000 description 11
- 238000005868 electrolysis reaction Methods 0.000 description 9
- 239000000460 chlorine Substances 0.000 description 6
- ZAMOUSCENKQFHK-UHFFFAOYSA-N Chlorine atom Chemical compound [Cl] ZAMOUSCENKQFHK-UHFFFAOYSA-N 0.000 description 4
- 229910052801 chlorine Inorganic materials 0.000 description 4
- UFHFLCQGNIYNRP-UHFFFAOYSA-N Hydrogen Chemical compound [H][H] UFHFLCQGNIYNRP-UHFFFAOYSA-N 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 238000005265 energy consumption Methods 0.000 description 2
- 239000001257 hydrogen Substances 0.000 description 2
- 229910052739 hydrogen Inorganic materials 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 230000035945 sensitivity Effects 0.000 description 2
- 238000005299 abrasion Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 238000012824 chemical production Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000007797 corrosion Effects 0.000 description 1
- 238000005260 corrosion Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 239000003792 electrolyte Substances 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 238000004880 explosion Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000002035 prolonged effect Effects 0.000 description 1
- 238000004904 shortening Methods 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/217—Validation; Performance evaluation; Active pattern learning techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/09—Supervised learning
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Optimization (AREA)
- Computing Systems (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Software Systems (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computational Mathematics (AREA)
- Mathematical Analysis (AREA)
- Biophysics (AREA)
- Health & Medical Sciences (AREA)
- Algebra (AREA)
- Databases & Information Systems (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Electrolytic Production Of Non-Metals, Compounds, Apparatuses Therefor (AREA)
Abstract
The invention relates to the technical field of industrial production, in particular to a chlor-alkali cell process abnormality detection method based on a self-adaptive threshold value, which is used for constructing a chlor-alkali cell abnormality detection model based on the historical normal working condition data of a chlor-alkali cell, does not need to establish a mechanism model and is easy to engineer; abnormal samples are not needed, and meanwhile, the data acquisition and marking cost is reduced; the chlor-alkali electrolytic cell abnormality detection model is used for analyzing real-time data of the electrolytic cell to detect and judge whether the chlor-alkali electrolytic cell is abnormal, so that the calculation difficulty is reduced, the judgment efficiency is improved, the detection cost is reduced, and the actual detection requirement is met.
Description
Technical Field
The invention relates to the technical field of industrial production, in particular to a chlor-alkali electrolytic cell process abnormality detection method based on a self-adaptive threshold value.
Background
The abnormal detection of the chlor-alkali industrial electrolytic cell is critical to the safety and efficiency of the production process. The chlor-alkali industry is the chemical production process of electrolysis brine, which takes chlorine, hydrogen and 32% alkali liquor (such as sodium hydroxide) as main products, and ensures the safety of chlor-alkali electrolysis tanks has the following significance:
① And (3) ensuring safety: the chlor-alkali industrial electrolytic tank is a high-temperature, high-pressure and highly corrosive process device. Abnormal conditions such as excessive current, abnormal voltage, excessive temperature of the electrolytic tank, leakage of electrolyte and the like can cause accidents, and even cause important safety accidents such as fire, explosion and the like. Through the abnormal detection of the electrolytic cell, the problems can be timely found and solved, and the safety of the production process is ensured.
② The production cost is reduced: abnormal conditions often lead to reduced performance of the electrolyzer, affecting product quality and yield. By timely detecting and eliminating abnormality, production interruption and equipment failure can be avoided, production efficiency and yield are improved, and production cost is reduced.
③ Energy and resource are saved: the chlor-alkali industrial electrolytic tank is a process device with larger energy consumption. Abnormal conditions may lead to reduced efficiency of the electrolysis process and increased energy consumption. Through anomaly detection, the process parameters can be timely adjusted, the energy utilization efficiency is improved, and the resource waste is reduced.
④ The product quality is improved: abnormality of the electrolytic cell may cause degradation of product quality such as decrease in purity of chlorine, change in concentration of lye, etc. The process conditions can be timely adjusted through abnormality detection, and the stable product quality is ensured.
⑤ Prolonging the service life of equipment: the abnormal conditions can cause the problems of aggravation of the internal corrosion of the electrolytic tank, abrasion of the electrode and the like, thereby shortening the service life of the equipment. Through anomaly detection, maintenance measures can be timely taken, the service life of equipment is prolonged, and the replacement and maintenance cost of the equipment is reduced.
In conclusion, the abnormal detection of the chlor-alkali industrial electrolytic cell has important significance for ensuring production safety, improving efficiency, reducing cost and ensuring product quality, and the abnormal condition can be timely found and solved by monitoring and analyzing the running state of the electrolytic cell in real time, so that the production process can be optimized, and sustainable development can be realized.
However, since the analysis model is built in the production process of the chlor-alkali electrolytic cell, an expert is required to summarize rules of the production process, the cost is high, the period is long, customization is required to be carried out for a specific scene, engineering is difficult to realize, and therefore, the abnormality of the chlor-alkali electrolytic process is difficult to detect by using the analysis model method.
The method based on supervised learning in the machine learning method requires a large number of training with negative samples (abnormal samples), training data are manually marked, the cost is high, the applicability is poor, meanwhile, part of algorithm calculation is complex, and more hardware resources are occupied; the important difficulty faced by machine learning is insufficient data, namely that the lack of measurement means is difficult to monitor the abnormality of the ion membrane.
Disclosure of Invention
The invention provides a chlor-alkali electrolyzer process abnormality detection method based on a self-adaptive threshold value, which solves the problem that the existing chlor-alkali electrolyzer is difficult to detect.
The invention solves the technical problems as follows:
a chlor-alkali electrolyzer process abnormality detection method based on an adaptive threshold comprises the following steps:
S1, acquiring history working condition data of a chlor-alkali electrolytic cell, and determining a process variable and a variable to be detected in the history working condition data of the chlor-alkali electrolytic cell;
S2, dividing the history working condition data of the chlor-alkali electrolytic cell into a training set and a verification set;
S3, constructing a chlor-alkali electrolytic cell abnormality detection model through a training set;
S4, verifying the chlor-alkali electrolytic cell abnormality detection model through a verification set, judging whether the chlor-alkali electrolytic cell abnormality detection model is qualified, if so, executing the step S5, and if not, executing the steps S3-S4 to reconstruct the chlor-alkali electrolytic cell abnormality detection model and verifying;
s5, calculating the self-adaptive threshold value of the detection model according to the chlor-alkali electrolytic cell abnormality detection model obtained in the step S4;
S6, collecting working condition data of the chlor-alkali electrolytic cell in real time, obtaining a process variable through the chlor-alkali electrolytic cell abnormality detection model obtained in the step S4, and judging whether the chlor-alkali electrolytic cell is abnormal or not through comparison of the obtained process variable and the self-adaptive threshold value obtained in the step S5.
Further defined, the process variable includes an amount of brine supplied, a concentration of Ca/Mg ions, a temperature of brine supplied, an amount of pure water supplied, a concentration of brine supplied, a flow rate of brine supplied, a hydrochloric acid flow rate of the electrolyzer, and a current level of the electrolyzer;
The variables to be detected comprise caustic soda current efficiency, cell voltage, cell temperature, pH value and catholyte alkali concentration.
Further defined, the step of obtaining the caustic soda current efficiency is as follows:
a. The daily yield of caustic soda M Caustic soda was obtained:
M Caustic soda ＝V Catholyte ×C Caustic soda ×ρ Catholyte ×t×10-3
Wherein V Catholyte is the total amount of catholyte generated in the time interval of the cell collection, and C Caustic soda is the average caustic soda concentration of the catholyte generated in the cell; ρ Catholyte is the average density of the catholyte produced by the cell, t is the time;
b. Caustic current efficiency η Caustic soda was calculated from the caustic daily yield:
Wherein M Caustic soda is the daily yield of caustic soda, I is the current of a single electrolytic cell, N is the number of electrolytic cells, and t is the time.
Further defined, said step S3 comprises the steps of:
s31, acquiring a process variable X and a variable Y to be detected in a training set;
s32, respectively carrying out normalization processing on the process variable X and the variable Y to be detected to obtain corresponding process variable normalization values And the normalized value/>, of the variable to be detected
S33, defining a matrix K according toEach element K a,b in the matrix K is determined:
Wherein, Normalizing values for process variables/>In a%Normalizing values for process variables/>C is an adjustable gaussian kernel constant;
s34, performing centering treatment on the definition matrix K to obtain
S35, through a KPLS algorithm, obtaining according to the step S34And the normalized value/>, obtained in the step S32, of the variable to be detectedObtain/>Score matrix T and variable to be detected normalized value/>The scoring matrix U of the chlor-alkali electrolytic cell abnormality detection model is built;
S36, calculating statistics according to the principal component number A of the KPLS and the sample number n of the training set
Where F A,n-A,α is the confidence interval and α is the confidence level.
Further defined, said step S4 comprises the steps of:
S41, acquiring a process variable X test and a variable Y test to be detected in the verification set, and respectively carrying out normalization processing to obtain a process normalization value And detecting normalized values/>
S42, normalizing the value according to the processThe process variable normalization value obtained in step S32/>And an adjustable gaussian kernel constant c, determining each element/>, in the matrix K test
Wherein,Value normalized for procedure/>In a%Normalizing values for process variables/>The b element of (b);
S43, centering the matrix K test to obtain Wherein 1 n is a column vector with n dimensions of 1, and K is a definition matrix obtained in the step S33;
S44, normalizing the variable to be detected obtained in the step S32 S34 is obtained/>And the scoring matrix T and the scoring matrix U obtained in the step S35 are calculated to obtain a predicted value/>
S45, performing inverse normalization on the predicted value to obtain a predicted resultI is the i-th variable in the verification set I to be detected;
S46, judging the prediction result Corresponding to the variables to be detected in the verification set/>If the deviation accords with the set value, obtaining the optimal number A best of the principal components of the Gaussian kernel constant c best and the KPLS, executing the step S5, otherwise, adjusting the adjustable number A of the principal components of the Gaussian kernel constant c and the KPLS, and executing the steps S33-S46 again.
Further defined, said step S5 comprises the steps of:
S51, calculating statistics corresponding to continuous h variables to be detected in the verification set according to the Gaussian kernel optimal constant c best and the principal component optimal number A best of the KPLS And collecting to obtain/>
Wherein,I=1, 2, …, l; l is the number of variables to be detected in the verification set, and n is the number of samples in the training set;
s52, calculating statistics Corresponding adaptive threshold/>And any process variable in the training set/>Corresponding weighted moving average/>
Where n is the number of samples of the training set, j= (1, 2, …, h), i > h, λ represents the weight parameter, λ j is the power of λ to j
When (when)Abnormality alarm is performed at the time, and the process variable/>, when the abnormality alarm is judgedIf the corresponding actual working conditions are abnormal, determining the optimal h and the optimal lambda; if not, readjust h and λ, and re-execute step S52.
Further defined, said step S6 comprises the steps of:
S61, respectively acquiring a current process variable X new and a current variable Y new to be detected in an actual working process;
S62, respectively carrying out normalization processing on the current process variable X new and the current variable Y new to be detected to obtain a current process variable normalization value And the current variable to be detected is normalized to a value/>
S63, normalizing the value according to the current process variableThe process variable normalization value obtained in step S32/>And the gaussian kernel optimal constant c best obtained in step S46, each element/>, in the matrix K new is determined
Wherein,For the f-th current process variable X new,/>, acquiredNormalizing values for process variables/>G element of (b);
S64, centering the matrix K new to obtain Wherein 1 n is a column vector with n dimensions of 1;
S65, calculating the f current process variable in the actual working process according to the Gaussian kernel optimal constant c best, the principal component optimal number A best of the KPLS and the optimal h obtained in the step S52
Where n is the number of samples of the training set, j= (1, 2, …, h);
S66, calculating statistics according to the best h and the best lambda obtained in the step S52 Corresponding adaptive thresholdWeighted moving average/>, corresponding to the current process variable
Where n is the number of samples of the training set, j= (1, 2, …, h), i > h, λ represents the weight parameter, λ j is the power j of λ;
When (when) If not, executing step S67;
s67, judging whether the current variable Y new to be detected is abnormal, if so, carrying out ion membrane abnormality early warning, and if not, keeping silence.
The invention has the beneficial effects that:
1. The method constructs the chlor-alkali electrolytic cell abnormality detection model based on the chlor-alkali electrolytic cell history normal working condition data, does not need to establish a mechanism model, and is easy to engineer; abnormal samples are not needed, and meanwhile, the data acquisition and marking cost is reduced; the chlor-alkali electrolytic cell abnormality detection model is used for analyzing real-time data of the electrolytic cell to detect and judge whether the chlor-alkali electrolytic cell is abnormal, so that the calculation difficulty is reduced, the judgment efficiency is improved, the detection cost is reduced, and the actual detection requirement is met.
2. The invention adopts KPLS algorithm to construct the abnormal detection model of the chlor-alkali electrolytic cell, realizes the processing of nonlinear data, and can ensure the accurate fault detection when a plurality of process variables are mutually influenced; by detecting the current electrolysis efficiency in real time, the defect of overlong counting period in the prior art is overcome, and the detection result is more accurate and timely.
3. The detection of the ion membrane abnormality is realized by a fault elimination method, so that the method is simpler and more efficient; meanwhile, the abnormal detection threshold value can be adaptively adjusted, so that the adjustment of the abnormal detection sensitivity is realized, the adjustment can be carried out according to different use requirements, and the sensitivity of the abnormal detection of the chlor-alkali electrolytic cell is met.
Detailed Description
In the chlor-alkali industry, electrolysis of saturated brine is realized through a chlor-alkali electrolytic tank to obtain chlorine, hydrogen and sodium hydroxide lye, and meanwhile, fresh brine is also present, during the working process, the concentration of the fresh brine is reduced due to the fact that the amount of supplied brine is too small when the concentration of the brine is unchanged, the concentration of the fresh brine is reduced due to the fact that the concentration is reduced when the amount of supplied brine is unchanged, the concentration of the fresh brine is increased due to the fact that the amount of supplied brine is too large when the concentration of the brine is unchanged, the temperature of the electrolytic tank is increased due to the fact that the temperature of the electrolytic tank is high when the temperature of the supplied brine is high, the concentration of the lye is reduced due to the fact that the temperature of the electrolytic tank is low when the temperature of the supplied brine is low, the concentration of the lye is reduced due to the fact that the pure water is supplied to keep stable when the concentration of the output lye is kept, the abnormal increase of lye is caused when the pure water is supplied, abnormal increase of the tank voltage of the electrolytic tank and abnormal increase of the tank temperature of the electrolytic tank when the supply of the brine is stopped when the film is broken, and the rise of H 2 and Cl 2 are caused when the film is present.
When the concentration of the weak brine is reduced, the concentration of the alkali liquor is reduced to below 30 percent, or the temperature of the tank is increased to above 92 ℃, the expansion is generated, namely the water content is overlarge.
The concentration of the weak brine rises, the temperature of the tank drops below 65 ℃ or the concentration of the alkali liquor rises above 36%, so that shrinkage is caused, namely, the water content is too small, and the current efficiency in the electrolysis process is reduced due to expansion or shrinkage.
At the same time, high concentrations of Ca + and Mg 2+ in brine also lead to reduced current efficiency, which ultimately leads to increased amounts of O 2 in Cl 2, reduced chlorine purity, increased ClO - concentration in brackish brine, and increased pH of brackish brine.
And when the concentration of the alkali liquor is increased, the abnormal increase of the tank voltage and the tank temperature can be directly caused, so that various or multiple factors exist for the reason of the abnormal electrolysis, and the abnormal detection of the electrolytic tank can not be accurately realized by singly detecting any process variable.
Example 1
The embodiment provides a chlor-alkali electrolytic cell process abnormality detection method based on a self-adaptive threshold value, which comprises the following steps:
S1, acquiring history working condition data of a chlor-alkali electrolytic cell, and determining a process variable and a variable to be detected in the history working condition data of the chlor-alkali electrolytic cell;
S2, dividing the history working condition data of the chlor-alkali electrolytic cell into a training set and a verification set;
S3, constructing a chlor-alkali electrolytic cell abnormality detection model through a training set;
S4, verifying the chlor-alkali electrolytic cell abnormality detection model through a verification set, judging whether the chlor-alkali electrolytic cell abnormality detection model is qualified, if so, executing the step S5, and if not, executing the steps S3-S4 to reconstruct the chlor-alkali electrolytic cell abnormality detection model and verifying;
s5, calculating the self-adaptive threshold value of the detection model according to the chlor-alkali electrolytic cell abnormality detection model obtained in the step S4;
S6, collecting working condition data of the chlor-alkali electrolytic cell in real time, obtaining a process variable through the chlor-alkali electrolytic cell abnormality detection model obtained in the step S4, and judging whether the chlor-alkali electrolytic cell is abnormal or not through comparison of the obtained process variable and the self-adaptive threshold value obtained in the step S5.
Further, the process variable in step S1 includes the amount of brine supplied, the concentration of Ca/Mg ions, the temperature of brine supplied, the amount of pure water supplied, the concentration of brine supplied, the flow rate of hydrochloric acid in the electrolytic cell, and the current level in the electrolytic cell, and the amount of brine supplied may be selected from a certain period of time, for example, a collection interval period set at 24 hours, 1 hour, 15 minutes, or the like.
The variables to be detected include caustic soda current efficiency, cell voltage, cell temperature, alkaline pH and catholyte caustic soda concentration.
For example, the caustic soda current efficiency calculation includes the steps of:
a. Daily yield of caustic soda (100% naoh) ton/day, M Caustic soda :
M Caustic soda ＝V Catholyte ×C Caustic soda ×ρ Catholyte ×t×10-3
wherein V Catholyte is the total amount of catholyte produced during the time interval of cell collection (m 3),C Caustic soda is the average caustic soda concentration of the catholyte produced by the cell, i.e. kg. NaOH/kg catholyte; ρ Catholyte is the average density of the catholyte produced by the cell (kg/m 3), t is time;
b. caustic current efficiency (%) η Caustic soda was calculated from caustic daily yield:
Where I is the current of the individual cells (DC-kA), N is the number of cells (e.g. 20) and t is the time (MT/day in 100% caustic soda).
Further illustratively, step S3 includes the steps of:
s31, acquiring a process variable X and a variable Y to be detected in a training set;
s32, respectively carrying out normalization processing on the process variable X and the variable Y to be detected to obtain corresponding process variable normalization values And the normalized value/>, of the variable to be detected
S33, defining a matrix K according toEach element K a,b in the matrix K is determined:
Wherein, Normalizing values for process variables/>In a%Normalizing values for process variables/>C is an adjustable gaussian kernel constant;
s34, performing centering treatment on the definition matrix K to obtain
S35, through a KPLS algorithm, obtaining according to the step S34And the normalized value/>, obtained in the step S32, of the variable to be detectedObtain/>Score matrix T and variable to be detected normalized value/>The scoring matrix U of the chlor-alkali electrolytic cell abnormality detection model is built;
S36, calculating statistics according to the principal component number A of the KPLS and the sample number n of the training set
Where F A,n-A,α is the confidence interval and α is the confidence level.
Further illustratively, step S4 includes the steps of:
S41, acquiring a process variable X test and a variable Y test to be detected in the verification set, and respectively carrying out normalization processing to obtain a process normalization value And detecting normalized values/>Because the verification set substitutes the process variable and the variable to be detected into the chlor-alkali electrolytic cell abnormality detection model one by one in the execution process, the mean square value of the process variable and the mean square value of the variable to be detected, which are obtained by using the normalization processing of the step S32 when the normalization processing is carried out on the process variable X test and the variable to be detected Y test.
S42, normalizing the value according to the processThe process variable normalization value obtained in step S32/>And an adjustable gaussian kernel constant c, determining each element/>, in the matrix K test
Wherein,Value normalized for procedure/>In a%Normalizing values for process variables/>The b element of (b);
S43, centering the matrix K test to obtain Wherein 1 n is a column vector with n dimensions of 1, and K is a definition matrix obtained in the step S33;
S44, normalizing the variable to be detected obtained in the step S32 S34 is obtained/>And the scoring matrix T and the scoring matrix U obtained in the step S35 are calculated to obtain a predicted value/>
S45, performing inverse normalization on the predicted value to obtain a predicted resultI is the i-th variable in the verification set I to be detected;
S46, judging the prediction result Corresponding to the variables to be detected in the verification set/>If the deviation accords with the set value, obtaining the optimal number A best of the principal components of the Gaussian kernel constant c best and the KPLS, executing the step S5, otherwise, adjusting the adjustable number A of the principal components of the Gaussian kernel constant c and the KPLS, and executing the steps S33-S46 again.
The construction of the chlor-alkali electrolytic cell abnormality detection model is completed through the step S3, and then the chlor-alkali electrolytic cell abnormality detection model is verified through the step S4, so that the optimal parameters are obtained: the optimum number of principal components A best of the Gaussian kernel optimum constant c best and KPLS, namely a chlor-alkali electrolytic cell abnormality detection model containing optimum parameters is used in the later period.
Further illustratively, step S5 includes the steps of:
S51, calculating statistics corresponding to continuous h variables to be detected in the verification set according to the Gaussian kernel optimal constant c best and the principal component optimal number A best of the KPLS And collecting to obtain/>
Wherein,I=1, 2, …, l; l is the number of variables to be detected in the verification set, and n is the number of samples in the training set;
s52, calculating statistics Corresponding adaptive threshold/>And any process variable in the training set/>Corresponding weighted moving average/>
Where n is the number of samples of the training set, j= (1, 2, …, h), i > h, λ represents the weight parameter, λ j is the power j of λ;
When (when) Abnormality alarm is performed at the time, and the process variable/>, when the abnormality alarm is judgedIf the corresponding actual working conditions are abnormal, determining the optimal h and the optimal lambda; if not, readjust h and λ, and re-execute step S52.
In the historical working condition data of the chlor-alkali electrolytic cell counted in the actual production process, abnormal variable to be detected is not obtained immediately when the process variable is abnormal, and the variable to be detected is usually abnormal after the process variable is abnormal by h values, so that early warning timeliness is realized by calculating the self-adaptive threshold value in order to improve the early warning timeliness.
Further illustratively, step S6 includes the steps of:
S61, respectively acquiring a current process variable X new and a current variable Y new to be detected in an actual working process;
S62, respectively carrying out normalization processing on the current process variable X new and the current variable Y new to be detected to obtain a current process variable normalization value And the current variable to be detected is normalized to a value/>
S63, normalizing the value according to the current process variableThe process variable normalization value obtained in step S32/>And the gaussian kernel optimal constant c best obtained in step S46, each element/>, in the matrix K new is determined
Wherein,For the f-th current process variable X new,/>, acquiredNormalizing values for process variables/>G element of (b);
S64, centering the matrix K new to obtain Wherein 1 n is a column vector with n dimensions of 1;
S65, calculating the f current process variable in the actual working process according to the Gaussian kernel optimal constant c best, the principal component optimal number A best of the KPLS and the optimal h obtained in the step S52
S66, calculating statistics according to the best h and the best lambda obtained in the step S52Corresponding adaptive thresholdWeighted moving average/>, corresponding to the current process variable
Where n is the number of samples of the training set, j= (1, 2, …, h), i > h, λ represents the weight parameter, λ j is the power j of λ;
When (when) If not, executing step S67;
s67, judging whether the current variable Y new to be detected is abnormal, if so, carrying out ion membrane abnormality early warning, and if not, keeping silence.
Since the abnormal state of the non-process variable is not monitored because the abnormal state of the non-process variable is used as the process variable when the chlor-alkali electrolytic cell abnormal detection model is constructed, in the actual abnormal detection process, the abnormal state of the electrolysis caused by the abnormal state of the process variable can be timely pre-warned through the step S66, the abnormal state of the process variable can not be pre-warned when the process variable is normal, even if the variable to be detected is abnormal, in order to avoid the missing report caused by the abnormal state of the electrolysis caused by the abnormal state of the non-process variable, the step S67 is also executed to detect whether the variable to be detected is abnormal or not through a conventional detection mode, at the moment, the data abnormality represents the fault of the ion membrane, the abnormal state of the ion membrane is pre-warned, and the ion membrane fault can be judged when the abnormal state of the detected data result is normal, so that the abnormal reason can be distinguished through the step S67, and the pre-warning effect is further optimized.
The above is an embodiment of the present application. The foregoing embodiments and the specific parameters of the embodiments are only for clarity of the verification process of the application, and are not intended to limit the scope of the application, which is defined by the claims, and all equivalent structural changes made by applying the content of the present application are included in the scope of the application.
Claims (7)
1. The chlor-alkali electrolyzer process abnormality detection method based on the self-adaptive threshold is characterized by comprising the following steps of:
S1, acquiring history working condition data of a chlor-alkali electrolytic cell, and determining a process variable and a variable to be detected in the history working condition data of the chlor-alkali electrolytic cell;
S2, dividing the history working condition data of the chlor-alkali electrolytic cell into a training set and a verification set;
S3, constructing a chlor-alkali electrolytic cell abnormality detection model through a training set;
S4, verifying the chlor-alkali electrolytic cell abnormality detection model through a verification set, judging whether the chlor-alkali electrolytic cell abnormality detection model is qualified, if so, executing the step S5, and if not, executing the steps S3-S4 to reconstruct the chlor-alkali electrolytic cell abnormality detection model and verifying;
s5, calculating the self-adaptive threshold value of the detection model according to the chlor-alkali electrolytic cell abnormality detection model obtained in the step S4;
S6, collecting working condition data of the chlor-alkali electrolytic cell in real time, obtaining a process variable through the chlor-alkali electrolytic cell abnormality detection model obtained in the step S4, and judging whether the chlor-alkali electrolytic cell is abnormal or not through comparison of the obtained process variable and the self-adaptive threshold value obtained in the step S5.
2. The adaptive threshold-based chlor-alkali cell process anomaly detection method of claim 1, wherein the process variables include an amount of feed brine, a concentration of Ca/Mg ions, a temperature of feed brine, an amount of feed pure water, a concentration of feed brine, a flow rate of feed brine, a hydrochloric acid flow rate of the cell, and a current level of the cell;
The variables to be detected comprise caustic soda current efficiency, cell voltage, cell temperature, pH value and catholyte alkali concentration.
3. The method for detecting abnormal chlor-alkali cell process based on the self-adaptive threshold according to claim 2, wherein the step of obtaining the caustic soda current efficiency is as follows:
a. The daily yield of caustic soda M Caustic soda was obtained:
M Caustic soda ＝V Catholyte ×C Caustic soda ×ρ Catholyte ×t×10-3
Wherein V Catholyte is the total amount of catholyte generated in the time interval of the cell collection, and C Caustic soda is the average caustic soda concentration of the catholyte generated in the cell; ρ Catholyte is the average density of the catholyte produced by the cell, t is the time;
b. Caustic current efficiency η Caustic soda was calculated from the caustic daily yield:
Wherein M Caustic soda is the daily yield of caustic soda, I is the current of a single electrolytic cell, N is the number of electrolytic cells, and t is the time.
4. A chlor-alkali cell process anomaly detection method based on an adaptive threshold according to claim 2 or 3, wherein step S3 comprises the steps of:
s31, acquiring a process variable X and a variable Y to be detected in a training set;
s32, respectively carrying out normalization processing on the process variable X and the variable Y to be detected to obtain corresponding process variable normalization values And the normalized value/>, of the variable to be detected
S33, defining a matrix K according toEach element K a,b in the matrix K is determined:
Wherein, Normalizing values for process variables/>In a%Normalizing values for process variables/>C is an adjustable gaussian kernel constant;
s34, performing centering treatment on the definition matrix K to obtain
S35, through a KPLS algorithm, obtaining according to the step S34And the normalized value/>, obtained in the step S32, of the variable to be detectedObtainingScore matrix T and variable to be detected normalized value/>The scoring matrix U of the chlor-alkali electrolytic cell abnormality detection model is built;
S36, calculating statistics according to the principal component number A of the KPLS and the sample number n of the training set
Where F A,n-A,α is the confidence interval and α is the confidence level.
5. The method for detecting abnormal chlor-alkali cell process based on adaptive threshold according to claim 4, wherein said step S4 comprises the steps of:
S41, acquiring a process variable X test and a variable Y test to be detected in the verification set, and respectively carrying out normalization processing to obtain a process normalization value And detecting normalized values/>
S42, normalizing the value according to the processThe process variable normalization value obtained in step S32/>And an adjustable gaussian kernel constant c, determining each element/>, in the matrix K test
Wherein,Value normalized for procedure/>In a%Normalizing values for process variables/>The b element of (b);
S43, centering the matrix K test to obtain Wherein 1 n is a column vector with n dimensions of 1, and K is a definition matrix obtained in the step S33;
S44, normalizing the variable to be detected obtained in the step S32 S34 is obtained/>And the scoring matrix T and the scoring matrix U obtained in the step S35 are calculated to obtain a predicted value/>
S45, performing inverse normalization on the predicted value to obtain a predicted resultI is the i-th variable in the verification set I to be detected;
S46, judging the prediction result Corresponding to the variables to be detected in the verification set/>If the deviation accords with the set value, obtaining the optimal number A best of the principal components of the Gaussian kernel constant c best and the KPLS, executing the step S5, otherwise, adjusting the adjustable number A of the principal components of the Gaussian kernel constant c and the KPLS, and executing the steps S33-S46 again.
6. The method for detecting abnormal chlor-alkali cell process based on adaptive threshold according to claim 5, wherein said step S5 comprises the steps of:
S51, calculating statistics corresponding to continuous h variables to be detected in the verification set according to the Gaussian kernel optimal constant c best and the principal component optimal number A best of the KPLS And collecting to obtain/>
Wherein,I=1, 2, …, l; l is the number of variables to be detected in the verification set, and n is the number of samples in the training set;
s52, calculating statistics Corresponding adaptive threshold/>And any process variable in the training set/>Corresponding weighted moving average/>
Where n is the number of samples of the training set, j= (1, 2, …, h), i > h, λ represents the weight parameter, λ j is the power j of λ;
When (when) Abnormality alarm is performed at the time, and the process variable/>, when the abnormality alarm is judgedIf the corresponding actual working conditions are abnormal, determining the optimal h and the optimal lambda; if not, readjust h and λ, and re-execute step S52.
7. The method for detecting abnormal chlor-alkali cell process based on adaptive threshold according to claim 6, wherein said step S6 comprises the steps of:
S61, respectively acquiring a current process variable X new and a current variable Y new to be detected in an actual working process;
S62, respectively carrying out normalization processing on the current process variable X new and the current variable Y new to be detected to obtain a current process variable normalization value And the current variable to be detected is normalized to a value/>
S63, normalizing the value according to the current process variableThe process variable normalization value obtained in step S32/>And the gaussian kernel optimal constant c best obtained in step S46, each element/>, in the matrix K new is determined
Wherein,For the f-th current process variable X new,/>, acquiredNormalizing values for process variables/>G element of (b);
S64, centering the matrix K new to obtain Wherein 1 n is a column vector with n dimensions of 1;
S65, calculating the f current process variable in the actual working process according to the Gaussian kernel optimal constant c best, the principal component optimal number A best of the KPLS and the optimal h obtained in the step S52
S66, calculating statistics according to the best h and the best lambda obtained in the step S52Corresponding adaptive threshold/>Weighted moving average/>, corresponding to the current process variable
Where n is the number of samples of the training set, j= (1, 2, …, h), i > h, λ represents the weight parameter, λ j is the power j of λ;
When (when) If not, executing step S67;
s67, judging whether the current variable Y new to be detected is abnormal, if so, carrying out ion membrane abnormality early warning, and if not, keeping silence.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410358974.0A CN118193997A (en) | 2024-03-27 | 2024-03-27 | Chlor-alkali electrolytic cell process abnormality detection method based on self-adaptive threshold |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410358974.0A CN118193997A (en) | 2024-03-27 | 2024-03-27 | Chlor-alkali electrolytic cell process abnormality detection method based on self-adaptive threshold |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN118193997A true CN118193997A (en) | 2024-06-14 |
Family
ID=91403218
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410358974.0A Pending CN118193997A (en) | 2024-03-27 | 2024-03-27 | Chlor-alkali electrolytic cell process abnormality detection method based on self-adaptive threshold |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118193997A (en) |
-
2024
- 2024-03-27 CN CN202410358974.0A patent/CN118193997A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US8152987B2 (en) | Method for ensuring and monitoring electrolyzer safety and performances | |
| US8114265B2 (en) | Efficiency optimization and damage detection of electrolysis cells | |
| US6591199B2 (en) | Method and system for acquisition, monitoring, display and diagnosis of operational parameters of electrolyzers | |
| CN113403645B (en) | Method and device for determining working state of electrolytic cell and controller | |
| CN113136592B (en) | Method and system for detecting faults in an electrolysis device having an electrolysis cell | |
| EP2226411B1 (en) | Method for ensuring and monitoring electrolyzer safety and performances | |
| EP1910588A1 (en) | Method and system for electrolyzer diagnosis based on curve fitting analysis and efficiency optimization | |
| CN114369849B (en) | Method and device for monitoring health degree of electrolytic cell and electrolytic cell monitoring system | |
| CN112563542B (en) | Fuel cell online detection method and detection system | |
| CN116894663B (en) | Intelligent AEM electrolytic tank control system | |
| CN117874666A (en) | Operation state evaluation system of renewable energy hydrogen production system | |
| CA2794737C (en) | Method for ensuring and monitoring electrolyzer safety and performances | |
| CN216386866U (en) | Evaluation device for electrolytic hydrogen production alkali liquor | |
| CN118193997A (en) | Chlor-alkali electrolytic cell process abnormality detection method based on self-adaptive threshold | |
| US9453286B2 (en) | Method and system for electrolyser single cell current efficiency | |
| CN117587457A (en) | Fault early warning system of PEM (PEM) water electrolysis hydrogen production equipment | |
| JP2025505734A (en) | Method for operating a multiple electrolyzer stack | |
| CN115354351A (en) | Electrolyzer control device, control system and method for electrolytic hydrogen production | |
| CN112080754B (en) | High-density low-power-consumption electrolytic cell and testing method thereof | |
| CN118884124B (en) | Water electrolysis hydrogen production system connected to the power grid test system | |
| JPH0874082A (en) | Operation of ion-exchange membrane electrtolytic cell | |
| CN118480820A (en) | Electrolytic tank sudden stop treatment method and device based on chlor-alkali industrial process | |
| CN117538391A (en) | Method, device and system for detecting running state of electrolytic tank | |
| CN113984858A (en) | Evaluation device and evaluation method for electrolytic hydrogen production alkali liquor | |
| CN118858814A (en) | An early warning system for the combined operation of water electrolysis hydrogen production and power grid |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |