Hydrocyclone working state intelligent monitoring system and method based on data analysis

Abstract

The invention relates to the technical field of data analysis, and discloses a hydrocyclone working state intelligent monitoring system and method based on data analysis. According to the method, firstly, working parameters of the hydrocyclone are obtained, correlation analysis is conducted on the working parameters of the hydrocyclone, and related working parameters of the hydrocyclone are obtained; secondly, after abnormal value deletion is carried out on the working parameters of the related hydrocyclone, missing data are filled through a cubic spline difference method; training a twin neural network to obtain a twin neural network model, outputting similarity, and judging whether the working parameters of the hydrocyclone are abnormal or not according to the similarity, so as to monitor the working state of the hydrocyclone; and finally, the working parameters of the related hydrocyclone are optimized by using a fishing optimization algorithm. The purpose of intelligently monitoring the working state is achieved by judging the abnormal condition of the working parameters of the related hydrocyclone, and the method is objective and accurate.

Description

Technical Field

[0001] The present invention relates to the technical field of data analysis, and in particular to an intelligent monitoring system and method for the working state of a hydrocyclone based on data analysis. Background Art

[0002] The traditional gravity stratification process has unstable separation performance and is easily affected by factors such as pressure and equipment wear. Hydrocyclones can be used to separate heterogeneous mixtures and have good separation performance. However, traditional methods for monitoring the working status of hydrocyclones rely mostly on manual monitoring and cannot monitor whether there are any abnormalities in the working status in real time. High-tech technologies such as neural networks are not used, which leads to errors in the monitoring of the working status of hydrocyclones, causing inconvenience to production work. Summary of the invention

[0003] In view of the problems in the related art, the present invention provides an intelligent monitoring system and method for the working status of a hydrocyclone based on data analysis to overcome the above-mentioned technical problems existing in the existing related art.

[0004] In order to solve the above technical problems, the present invention is achieved through the following technical solutions:

[0005] The present invention is an intelligent monitoring method for the working state of a hydrocyclone based on data analysis, comprising the following steps:

[0006] S1. Obtaining the working parameters of the hydrocyclone, performing correlation analysis on the working parameters of the hydrocyclone, and obtaining a correlation coefficient matrix, wherein the correlation coefficient matrix obtains relevant working parameters of the hydrocyclone according to positive correlation and negative correlation;

[0007] S2, selecting relevant hydrocyclone working parameters to form a relevant hydrocyclone working parameter set, performing outlier deletion on the relevant hydrocyclone working parameter set to obtain an outlier-deleted relevant hydrocyclone working parameter set; then performing data filling on the outlier-deleted relevant hydrocyclone working parameter set to obtain a processed relevant hydrocyclone working parameter set;

[0008] S3, labeling the processed set of related hydrocyclone working parameters, training a twin neural network, obtaining a twin neural network model, outputting similarity, determining whether the related hydrocyclone working parameters are abnormal through similarity, and realizing monitoring of the working status of the hydrocyclone;

[0009] S4. After the working state monitoring of the hydrocyclone is realized, the relevant working parameters of the hydrocyclone are optimized using the fishing optimization algorithm to obtain the best relevant working parameters of the hydrocyclone in the processed relevant working parameter set of the hydrocyclone.

[0010] The invention first obtains the working parameters of the hydrocyclone, establishes the working parameter matrix of the hydrocyclone, calculates the correlation coefficient of the working parameters of the hydrocyclone, and performs correlation analysis on the working parameters of the hydrocyclone based on the Pearson method to obtain the relevant working parameters of the hydrocyclone; it can effectively reduce the complexity of the data and the amount of calculation; secondly, based on the Thompson The tau method and quartile method were used to perform the first and second outlier removal on the relevant hydrocyclone working parameters, respectively, and then the missing data was filled by the cubic spline difference method to obtain the processed relevant hydrocyclone working parameters; the normal operation data was retained to improve the data processing efficiency; the processed relevant hydrocyclone working parameters were then used to train the twin neural network to determine whether the hydrocyclone working parameters were abnormal, so as to realize the hydrocyclone working status monitoring; the hydrocyclone working status was monitored by whether the relevant hydrocyclone working parameters were abnormal, and the twin neural network converged quickly and achieved good results; finally, the fishing optimization algorithm was used to optimize the relevant hydrocyclone working parameters to optimize the hydrocyclone working efficiency, and the optimal solution was found by updating the fishermen's position. The fishing optimization algorithm had good optimization effect and strong development capability.

[0011] Preferably, the S1 comprises the following steps:

[0012] S11, obtaining the working parameters of the hydrocyclone, wherein the working parameters of the hydrocyclone include slurry density, slurry particle size, slurry concentration, slurry pressure, sand settling port diameter, cone angle, size of the feed pipe and overflow pipe, and material particle size, etc., and generating a set of working parameters of the hydrocyclone A = {a1, a2, a3, ..., a i}, where a i represents the i-th hydrocyclone operating parameter, each hydrocyclone operating parameter contains j sample data, and the hydrocyclone operating parameter matrix C is generated as follows:

[0013]

[0014] Among them, a ij represents the jth sample data of the operating parameters of the i-th hydrocyclone;

[0015] Select the first and second columns in the hydrocyclone operating parameter matrix to represent the hydrocyclone operating parameters a1 and a2, respectively, and set Indicates the size of the i1th sample data of the hydrocyclone operating parameter a1, Indicates the size of the i1th sample data of the hydrocyclone operating parameter a2, It represents the average value of the sample data of the working parameter a1 of the hydrocyclone. represents the average value of the sample data of the hydrocyclone working parameter a2, and the correlation coefficient calculation formula is as follows:

[0016]

[0017] Where B represents the correlation coefficient;

[0018] The correlation coefficients of the hydrocyclone operating parameter a1 and other hydrocyclone operating parameters in the hydrocyclone operating parameter matrix are calculated respectively, which are recorded as the first correlation coefficient set; the correlation coefficients of the hydrocyclone operating parameter a2 and other hydrocyclone operating parameters in the hydrocyclone operating parameter matrix are calculated, which are recorded as the second correlation coefficient set, and the correlation coefficients of all hydrocyclone operating parameters in the hydrocyclone operating parameter matrix are obtained. All correlation coefficient sets are combined to obtain the correlation coefficient matrix C′ as follows:

[0019]

[0020] Among them, a ij Indicates the working parameters of the hydrocyclone a i And hydrocyclone working parameters a i The jth correlation coefficient of

[0021] S12. Compare the correlation coefficients in the correlation coefficient matrix, set a correlation threshold ω, and when the correlation coefficient in the correlation coefficient matrix is ​​greater than the correlation threshold ω, retain the correlation coefficient in the correlation coefficient matrix; otherwise, set the correlation coefficient in the correlation coefficient matrix to 0 to obtain a processed correlation coefficient matrix, and find the hydrocyclone operating parameters corresponding to the positive correlation or negative correlation of the hydrocyclone operating parameters in the processed correlation coefficient matrix, and record them as the relevant hydrocyclone operating parameters.

[0022] The invention obtains the working parameters of the hydrocyclone, establishes a hydrocyclone working parameter matrix, performs correlation analysis on the hydrocyclone working parameters, finds relevant hydrocyclone working parameters after comparing with thresholds, and monitors whether the relevant hydrocyclone working parameters are abnormal to achieve the purpose of hydrocyclone working status monitoring, which can effectively reduce data complexity and reduce the amount of calculation.

[0023] Preferably, S2 comprises the following steps:

[0024] S21, select the first e samples of the relevant hydrocyclone operating parameters to form a relevant hydrocyclone operating parameter set, denoted as A′={(b1, b2, .., b e ),(c1,c2,..,c e ),...,(d1,d2,..,d e )}, where b e 、c e and d eAll represent the e-th sample data of the relevant hydrocyclone working parameters. The Thompsontau method and the quartile method are used to perform the first outlier removal and the second outlier removal on the relevant hydrocyclone working parameter set, respectively, to obtain the outlier-removed relevant hydrocyclone working parameter set. The specific steps are as follows:

[0025] S211, dividing the set of relevant hydrocyclone operating parameters into β parameter intervals according to the relevant hydrocyclone operating parameters, where β represents the number of relevant hydrocyclone operating parameters, calculating the average value of the sample data in the parameter interval, calculating the standard deviation of the sample data in the parameter interval according to the average value of the sample data in the parameter interval, and obtaining the absolute value of the deviation of the sample data in the parameter interval; setting

[0026] represents the distribution value of sample data in the parameter interval, χ represents the significance level, and the calculation formula of parameter δ in Thompson tau method is as follows:

[0027]

[0028] in, The significance level is The distribution value of sample data in the time parameter interval;

[0029] When the absolute value of the deviation of the sample data in the parameter interval is less than the product of the standard deviation of the sample data in the parameter interval and the parameter δ, the sample data in the corresponding parameter interval is a normal value, otherwise the sample data in the corresponding parameter interval is an abnormal value, and the abnormal value is deleted, completing the first abnormal value deletion of the relevant hydrocyclone working parameter set; repeat S211 until there is no abnormal value in the relevant hydrocyclone working parameter set, obtaining the first abnormal value deletion parameter set, and completing the first abnormal value deletion;

[0030] S212: record the first outlier deletion parameter set as A″={(b1, b2, .., b e′ ),(c1,c2,..,c e′ ),...,(d1,d2,..,d e′ )}, where b e′ 、c e′ and d e′ All represent the e′th sample of the first outlier removal parameter set. The first outlier removal parameter set is divided into β1 parameter intervals. The parameter interval is sorted in ascending order according to the sample data and then divided into the parameter interval three times. The sample data in the parameter interval is divided into four equal parts. The sample data at the three divisions are recorded as the first quantile, the second quantile and the third quantile. The second quantile is recorded as the median of the parameter interval.

[0031] When the number of sample data in the parameter interval is even, the second quantile divides the parameter interval into the same sub-intervals, and the first quantile and the third quantile are the medians of the same sub-intervals respectively;

[0032] When the number of sample data in the parameter interval is 4e′+3, the calculation formulas for the first quantile and the third quantile are as follows:

[0033]

[0034] Among them, B1 represents the first quantile, B3 represents the third quantile, and b e″+1 represents the e″+1th sample data in the parameter interval, b e″+2 represents the e″+2th sample data in the parameter interval, b 3e″+2 Indicates the 3e″+2th sample data in the parameter interval, b 3e″+3 Represents the 3e″+3th sample data in the parameter interval;

[0035] When the number of sample data in the parameter interval is 4e′+1, the calculation formulas for the first quantile and the third quantile are as follows:

[0036]

[0037] Among them, b e″ represents the e″th sample data in the parameter interval, b 3e″+1 Represents the 3e″+1th sample data in the parameter interval;

[0038] Interquartile range Upper limit of interval Interval lower limit The sample data in the parameter interval that are located at the upper limit and the lower limit of the interval are retained, and the sample data in the parameter interval that are not located at the upper limit and the lower limit of the interval are deleted; S212 is repeated until there are no outliers in the first outlier deletion parameter set, and the second outlier deletion is completed;

[0039] S22, dividing the set of hydrocyclone working parameters related to the abnormal value deletion into β2 interpolation intervals, and finding a cubic function in the interpolation interval that satisfies the sample data of the interpolation interval, has a continuous second derivative and a smooth curve, and recording it as an interpolation function; substituting the time point corresponding to the interpolation interval into the interpolation function to obtain the interpolation function value, which is the missing sample data of the interpolation interval, and supplementing the interpolation function value to the interpolation interval; filling the processed set of related hydrocyclone working parameters with data in turn to obtain the processed set of related hydrocyclone working parameters.

[0040] The invention performs the first outlier deletion and the second outlier deletion on the relevant hydrocyclone working parameters based on the Thompson tau method and the quartile method respectively, and then uses the cubic spline difference method to fill in the missing data to obtain the processed relevant hydrocyclone working parameters; the valid data is retained and the data quality is improved.

[0041] Preferably, S3 comprises the following steps:

[0042] S31, record the processed related hydrocyclone working parameter set as A″′={(b1, b2, .., b e″′ ),(c1,c2,..,c e″′ ),...,(d1,d2,..,d e″′ )}, where b e″′ 、c e″′ and d e″′ All represent the e″′th sample data of the processed related hydrocyclone working parameter set, model the sample data in the processed related hydrocyclone working parameter set, set data labels C1 and C2, where C1 represents the sample data in the processed related hydrocyclone working parameter set, and C2 represents the theoretical sample data in the processed related hydrocyclone working parameter set; set the category label to C3, where C3=0 represents that the sample data in the processed related hydrocyclone working parameter set is normal, and C3=1 represents that the sample data in the processed related hydrocyclone working parameter set is abnormal;

[0043] Normalizing the sample data in the processed related hydrocyclone working parameter set to obtain a normalized related hydrocyclone working parameter set;

[0044] S32, obtaining new working parameters of the hydrocyclone, generating a new normalized set of related working parameters of the hydrocyclone, dividing the new normalized set of related working parameters of the hydrocyclone into a working parameter training set and a working parameter test set, inputting the working parameter training set into the twin neural network, mapping C1 and C2 in the working parameter training set into the vector space through the feedforward neural network weight matrix, taking the Euclidean distance in the vector space as the similarity between C1 and C2 in the working parameter training set, setting the loss function f, and continuously updating the weights through back propagation until the twin neural network converges to obtain a trained twin neural network; then inputting the working parameter test set into the trained twin neural network, setting the accuracy threshold as ξ, and obtaining the twin neural network model when the output result accuracy is less than ξ, otherwise repeating S32 until the output result accuracy is less than ξ;

[0045] S33. Input the normalized related hydrocyclone working parameter set into the twin neural network model. The twin neural network model outputs a similarity result, which is recorded as a similarity set. When the similarity in the similarity set is greater than or equal to 0.6, the sample data category label C3 corresponding to the similarity in the similarity set is 0, otherwise the sample data category label C3 corresponding to the similarity in the similarity set is 1; whether the related hydrocyclone working parameters are abnormal is obtained according to the similarity in the similarity set, so as to realize the working status monitoring of the hydrocyclone.

[0046] The invention obtains a twin neural network model by training the twin neural network, and uses the output similarity to judge whether the working parameters of the hydrocyclone are abnormal, thereby realizing the working status monitoring of the hydrocyclone. The twin neural network has a fast convergence speed and a good implementation effect.

[0047] Preferably, S4 comprises the following steps:

[0048] S41, assigning weights to the relevant hydrocyclone working parameters in the processed relevant hydrocyclone working parameter set, calculating the working efficiency of the hydrocyclone, and using the hydrocyclone working efficiency calculation formula as the fitness function; setting g search spaces, where g represents the number of relevant hydrocyclone working parameters in the processed relevant hydrocyclone working parameter set, in each search space, the sample data is regarded as fishermen, the fisherman dimension is j′, and the fisherman matrix C″ is as follows:

[0049]

[0050] Among them, a e ″ ″j′ represents the e″′th fisherman with dimension j′;

[0051] The fishermen are in the search space. Set the upper bound of the search space to g′, the lower bound of the search space to g″, h1 represents a random number and h1∈(0,1), then the initial position of the hth fisherman is The calculation formula is as follows:

[0052]

[0053] S42. In the exploration stage, fishermen adjust their positions according to the fitness function value, constantly looking for the best sample data, and evaluate the catch. The current catch of fishermen is set to l, the maximum catch of fishermen is set to L, and the capture speed parameter is ε. The calculation formula of the capture speed parameter is as follows:

[0054]

[0055] Set the capture probability to φ. When φ < ε, the fisherman chooses to capture independently. Set the number of iterations to k. The best fitness function value of the kth iteration is f′(k). The worst fitness function value of the kth iteration is f″(k). The current fitness function value is f1. The fitness function value of any previous period is f2. The position of the hth fisherman in the kth iteration is The position of the hth fisherman in the k+1th iteration is The position of the h′th fisherman in the k+1th iteration is h2 represents a random number and h2∈(0,1), h3 represents a random unit vector, and the detection range is D, then the capture coefficient The calculation formula for the hth fisherman's position in the k+1th iteration is as follows:

[0056]

[0057] When φ≥ε, the fishermen select a group to capture, set the target point surrounded by the group to x′, the speed of the fishermen approaching the target point is v1, h4 represents the moving offset, and the position of the hth fisherman in the kth iteration is updated to In the k+1th iteration, the hth fisherman’s position is updated to The calculation formula is as follows:

[0058]

[0059] S43. During the collective capture phase, fishermen adjust their positions and continuously search for the position of fishermen corresponding to the best fitness function. The position corresponding to the current best fitness function is set as x″, the Gaussian distribution function is F, the function variance is k′, h5 represents a random number and h5∈(0,1), and the average position of fishermen in different dimensions is In the k+1th iteration, the hth fisherman’s position is updated to The calculation formula is as follows:

[0060]

[0061] The fisherman continuously iterates his own position, and sets the maximum number of iterations to K. When the current number of iterations reaches the maximum number of iterations, the final fisherman position and the best fitness function value are output. In the g search spaces, the dimension of the final fisherman position in the fisherman matrix corresponds to the best relevant hydrocyclone working parameter in the processed relevant hydrocyclone working parameter set.

[0062] The invention optimizes the working parameters of the relevant hydrocyclones by using a fishing optimization algorithm, thereby optimizing the working efficiency of the hydrocyclones, and finding the optimal solution by updating the positions of fishermen. The fishing optimization algorithm has good optimization effects and strong development capabilities.

[0063] This embodiment also discloses a system for intelligent monitoring method of working state of a hydrocyclone based on data analysis, which specifically includes: a hydrocyclone working parameter correlation analysis module, a relevant hydrocyclone working parameter preprocessing module, a hydrocyclone working state monitoring module and a relevant hydrocyclone working parameter optimization module;

[0064] The hydrocyclone operating parameter correlation analysis module is used to perform correlation analysis on the hydrocyclone operating parameters;

[0065] The relevant hydrocyclone operating parameter preprocessing module is used to delete abnormal values ​​and fill data of the relevant hydrocyclone operating parameters;

[0066] The hydrocyclone working state monitoring module is used to use the twin neural network model to determine whether the relevant hydrocyclone working parameters are abnormal;

[0067] The relevant hydrocyclone operating parameter optimization module is used to optimize the relevant hydrocyclone operating parameters using a fishing optimization algorithm.

[0068] The present invention has the following beneficial effects:

[0069] 1. The invention obtains the working parameters of the hydrocyclone, establishes the working parameter matrix of the hydrocyclone, performs correlation analysis on the working parameters of the hydrocyclone by the Pearson method, and selects relevant parameters for subsequent processing, which can effectively reduce data complexity and reduce the amount of calculation.

[0070] 2. The invention deletes outliers based on the working parameters of relevant hydrocyclones. Deleting outliers can retain valid data and increase the accuracy of processing results. The missing data is filled in using the cubic spline interpolation method, thereby improving data quality.

[0071] 3. The invention obtains a twin neural network model by training the twin neural network, and uses the output similarity to determine whether the working parameters of the hydrocyclone are abnormal, thereby realizing the working status monitoring of the hydrocyclone. The twin neural network has a fast convergence speed and a good implementation effect.

[0072] 4. The invention optimizes the working parameters of the relevant hydrocyclones by using a fishing optimization algorithm, thereby optimizing the working efficiency of the hydrocyclones. The positions of fishermen are updated through three capture stages to find the optimal solution. The fishing optimization algorithm has good optimization effects and strong development capabilities.

[0073] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] In order to more clearly illustrate the technical solutions of the embodiments of the invention, the drawings required for describing the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the invention. For ordinary technicians in this field, they can also obtain drawings based on these drawings without paying creative work.

[0075] Figure 1 The present invention provides a flow chart of intelligent monitoring of the working state of a hydraulic cyclone by an intelligent monitoring system of the working state of a hydraulic cyclone based on data analysis. DETAILED DESCRIPTION

[0076] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0077] In the description of the present invention, it is necessary to understand that the terms "opening", "upper", "lower", "top", "middle", "inside" and the like indicating orientation or positional relationship are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the components or elements referred to must have a specific orientation, be constructed and operate in a specific orientation, and therefore cannot be understood as limiting the invention.

[0078] Example 1

[0079] This embodiment discloses a method for intelligent monitoring of the working status of a hydrocyclone based on data analysis, which specifically includes the following contents:

[0080] S1. Obtaining the working parameters of the hydrocyclone, performing correlation analysis on the working parameters of the hydrocyclone, and obtaining a correlation coefficient matrix, wherein the correlation coefficient matrix obtains relevant working parameters of the hydrocyclone according to positive correlation and negative correlation;

[0081] The S1 comprises the following steps:

[0082] S11, obtaining the working parameters of the hydrocyclone, wherein the working parameters of the hydrocyclone include slurry density, slurry particle size, slurry concentration, slurry pressure, sand settling port diameter, cone angle, size of the feed pipe and overflow pipe, and material particle size, etc., and generating a set of working parameters of the hydrocyclone A = {a1, a2, a3, ..., a i}, where a i represents the i-th hydrocyclone operating parameter, each hydrocyclone operating parameter contains j sample data, and the hydrocyclone operating parameter matrix C is generated as follows:

[0083]

[0084] Among them, a ij represents the jth sample data of the operating parameters of the i-th hydrocyclone;

[0085] Select the first and second columns in the hydrocyclone operating parameter matrix to represent the hydrocyclone operating parameters a1 and a2, respectively, and set Indicates the size of the i1th sample data of the hydrocyclone operating parameter a1, Indicates the size of the i1th sample data of the hydrocyclone operating parameter a2, It represents the average value of the sample data of the working parameter a1 of the hydrocyclone. represents the average value of the sample data of the hydrocyclone working parameter a2, and the correlation coefficient calculation formula is as follows:

[0086]

[0087] Where B represents the correlation coefficient;

[0088] The correlation coefficients of the hydrocyclone operating parameter a1 and other hydrocyclone operating parameters in the hydrocyclone operating parameter matrix are calculated respectively, which are recorded as the first correlation coefficient set; the correlation coefficients of the hydrocyclone operating parameter a2 and other hydrocyclone operating parameters in the hydrocyclone operating parameter matrix are calculated, which are recorded as the second correlation coefficient set, and the correlation coefficients of all hydrocyclone operating parameters in the hydrocyclone operating parameter matrix are obtained. All correlation coefficient sets are combined to obtain the correlation coefficient matrix C′ as follows:

[0089]

[0090] Among them, a ij Indicates the working parameters of the hydrocyclone a i And hydrocyclone working parameters a i The jth correlation coefficient of

[0091] S12, comparing the correlation coefficients in the correlation coefficient matrix, setting a correlation threshold ω, when the correlation coefficient in the correlation coefficient matrix is ​​greater than the correlation threshold ω, retaining the correlation coefficient in the correlation coefficient matrix; otherwise, setting the correlation coefficient in the correlation coefficient matrix to 0, obtaining a processed correlation coefficient matrix, finding the hydrocyclone operating parameters corresponding to the positive correlation or negative correlation of the hydrocyclone operating parameters in the processed correlation coefficient matrix, and recording them as the relevant hydrocyclone operating parameters;

[0092] S2, selecting relevant hydrocyclone working parameters to form a relevant hydrocyclone working parameter set, performing outlier deletion on the relevant hydrocyclone working parameter set to obtain an outlier-deleted relevant hydrocyclone working parameter set; then performing data filling on the outlier-deleted relevant hydrocyclone working parameter set to obtain a processed relevant hydrocyclone working parameter set;

[0093] The S2 comprises the following steps:

[0094] S21, select the first e samples of the relevant hydrocyclone operating parameters to form a relevant hydrocyclone operating parameter set, denoted as A′={(b1, b2, .., b e ),(c1,c2,..,c e ),...,(d1,d2,..,d e )}, where b e 、c e and d e All represent the e-th sample data of the relevant hydrocyclone working parameters. The Thompson tau method and the quartile method are used to perform the first outlier removal and the second outlier removal on the relevant hydrocyclone working parameter set, respectively, to obtain the outlier-removed relevant hydrocyclone working parameter set. The specific steps are as follows:

[0095] S211, dividing the set of relevant hydrocyclone operating parameters into β parameter intervals according to the relevant hydrocyclone operating parameters, where β represents the number of relevant hydrocyclone operating parameters, calculating the average value of the sample data in the parameter interval, calculating the standard deviation of the sample data in the parameter interval according to the average value of the sample data in the parameter interval, and obtaining the absolute value of the deviation of the sample data in the parameter interval; setting

[0096] represents the distribution value of sample data in the parameter interval, χ represents the significance level, and the calculation formula of parameter δ in Thompson tau method is as follows:

[0097]

[0098] in, The significance level is The distribution value of sample data in the time parameter interval;

[0099] When the absolute value of the deviation of the sample data in the parameter interval is less than the product of the standard deviation of the sample data in the parameter interval and the parameter δ, the sample data in the corresponding parameter interval is a normal value, otherwise the sample data in the corresponding parameter interval is an abnormal value, and the abnormal value is deleted, completing the first abnormal value deletion of the relevant hydrocyclone working parameter set; repeat S211 until there is no abnormal value in the relevant hydrocyclone working parameter set, obtaining the first abnormal value deletion parameter set, and completing the first abnormal value deletion;

[0100] S212: record the first outlier deletion parameter set as A″={(b1, b2, .., b e′ ),(c1,c2,..,c e′ ),...,(d1,d2,..,d e′ )}, where b e′ 、c e′ and d e′ All represent the e′th sample of the first outlier removal parameter set. The first outlier removal parameter set is divided into β1 parameter intervals. The parameter interval is sorted in ascending order according to the sample data and then divided into the parameter interval three times. The sample data in the parameter interval is divided into four equal parts. The sample data at the three divisions are recorded as the first quantile, the second quantile and the third quantile. The second quantile is recorded as the median of the parameter interval.

[0101] When the number of sample data in the parameter interval is even, the second quantile divides the parameter interval into the same sub-intervals, and the first quantile and the third quantile are the medians of the same sub-intervals respectively;

[0102] When the number of sample data in the parameter interval is 4e′+3, the calculation formulas for the first quantile and the third quantile are as follows:

[0103]

[0104] Among them, B1 represents the first quantile, B3 represents the third quantile, and b e″+1 represents the e″+1th sample data in the parameter interval, b e″+2 represents the e″+2th sample data in the parameter interval, b 3e″+2 Indicates the 3e″+2th sample data in the parameter interval, b 3e″+3 Represents the 3e″+3th sample data in the parameter interval;

[0105] When the number of sample data in the parameter interval is 4e′+1, the calculation formulas for the first quantile and the third quantile are as follows:

[0106]

[0107] Among them, b e″represents the e″th sample data in the parameter interval, b 3e″+1 Represents the 3e″+1th sample data in the parameter interval;

[0108] Interquartile range Upper limit of interval Interval lower limit The sample data in the parameter interval that are located at the upper limit and the lower limit of the interval are retained, and the sample data in the parameter interval that are not located at the upper limit and the lower limit of the interval are deleted; S212 is repeated until there are no outliers in the first outlier deletion parameter set, and the second outlier deletion is completed;

[0109] S22, dividing the set of hydrocyclone working parameters related to the abnormal value deletion into β2 interpolation intervals, finding a cubic function that satisfies the sample data of the interpolation interval, has a continuous second derivative and a smooth curve in the interpolation interval, and recording it as an interpolation function; substituting the time point corresponding to the interpolation interval into the interpolation function to obtain an interpolation function value, which is the missing sample data in the interpolation interval, and supplementing the interpolation function value to the interpolation interval; sequentially filling the processed set of hydrocyclone working parameters to obtain a processed set of hydrocyclone working parameters;

[0110] S3, labeling the processed set of related hydrocyclone working parameters, training a twin neural network, obtaining a twin neural network model, outputting similarity, determining whether the related hydrocyclone working parameters are abnormal through similarity, and realizing monitoring of the working status of the hydrocyclone;

[0111] The S3 comprises the following steps:

[0112] S31, record the processed related hydrocyclone working parameter set as A″′={(b1, b2, .., b e″′ ),(c1,c2,..,c e″′ ),...,(d1,d2,..,d e″′ )}, where b e″′ 、c e″′ and d e″′ All represent the e″′th sample data of the processed related hydrocyclone working parameter set, model the sample data in the processed related hydrocyclone working parameter set, set data labels C1 and C2, where C1 represents the sample data in the processed related hydrocyclone working parameter set, and C2 represents the theoretical sample data in the processed related hydrocyclone working parameter set; set the category label to C3, where C3=0 represents that the sample data in the processed related hydrocyclone working parameter set is normal, and C3=1 represents that the sample data in the processed related hydrocyclone working parameter set is abnormal;

[0113] Normalizing the sample data in the processed related hydrocyclone working parameter set to obtain a normalized related hydrocyclone working parameter set;

[0114] S32, obtaining new working parameters of the hydrocyclone, generating a new normalized set of related working parameters of the hydrocyclone, dividing the new normalized set of related working parameters of the hydrocyclone into a working parameter training set and a working parameter test set, inputting the working parameter training set into the twin neural network, mapping C1 and C2 in the working parameter training set into the vector space through the feedforward neural network weight matrix, taking the Euclidean distance in the vector space as the similarity between C1 and C2 in the working parameter training set, setting the loss function f, and continuously updating the weights through back propagation until the twin neural network converges to obtain a trained twin neural network; then inputting the working parameter test set into the trained twin neural network, setting the accuracy threshold as ξ, and obtaining the twin neural network model when the output result accuracy is less than ξ, otherwise repeating S32 until the output result accuracy is less than ξ;

[0115] S33, input the normalized related hydrocyclone working parameter set into the twin neural network model, the twin neural network model outputs a similarity result, recorded as a similarity set, when the similarity in the similarity set is greater than or equal to 0.6, the sample data category label C3 corresponding to the similarity in the similarity set is 0, otherwise the sample data category label C3 corresponding to the similarity in the similarity set is 1; according to the similarity in the similarity set, whether the related hydrocyclone working parameters are abnormal is obtained to realize the working status monitoring of the hydrocyclone;

[0116] S4. After the working state monitoring of the hydrocyclone is realized, the relevant working parameters of the hydrocyclone are optimized using the fishing optimization algorithm to obtain the best relevant working parameters of the hydrocyclone in the processed relevant working parameter set of the hydrocyclone;

[0117] The S4 comprises the following steps:

[0118] S41, assigning weights to the relevant hydrocyclone working parameters in the processed relevant hydrocyclone working parameter set, calculating the working efficiency of the hydrocyclone, and using the hydrocyclone working efficiency calculation formula as the fitness function; setting g search spaces, where g represents the number of relevant hydrocyclone working parameters in the processed relevant hydrocyclone working parameter set, in each search space, the sample data is regarded as fishermen, the fisherman dimension is j′, and the fisherman matrix C″ is as follows:

[0119]

[0120] Among them, a e ″ ″j′ represents the e″′th fisherman with dimension j′;

[0121] The fishermen are in the search space. Set the upper bound of the search space to g′, the lower bound of the search space to g″, h1 represents a random number and h1∈(0,1), then the initial position of the hth fisherman is The calculation formula is as follows:

[0122]

[0123] S42. In the exploration stage, fishermen adjust their positions according to the fitness function value, constantly looking for the best sample data, and evaluate the catch. The current catch of fishermen is set to l, the maximum catch of fishermen is set to L, and the capture speed parameter is ε. The calculation formula of the capture speed parameter is as follows:

[0124]

[0125] Set the capture probability to φ. When φ < ε, the fisherman chooses to capture independently. Set the number of iterations to k. The best fitness function value of the kth iteration is f′(k). The worst fitness function value of the kth iteration is f″(k). The current fitness function value is f1. The fitness function value of any previous period is f2. The position of the hth fisherman in the kth iteration is The position of the hth fisherman in the k+1th iteration is The position of the h′th fisherman in the k+1th iteration is h2 represents a random number and h2∈(0,1), h3 represents a random unit vector, and the detection range is D, then the capture coefficient The calculation formula for the hth fisherman's position in the k+1th iteration is as follows:

[0126]

[0127] When φ≥ε, the fishermen select a group to capture, set the target point surrounded by the group to x′, the speed of the fishermen approaching the target point is v1, h4 represents the moving offset, and the position of the hth fisherman in the kth iteration is updated to In the k+1th iteration, the hth fisherman’s position is updated to The calculation formula is as follows:

[0128]

[0129] S43. During the collective capture phase, fishermen adjust their positions and continuously search for the position of fishermen corresponding to the best fitness function. The position corresponding to the current best fitness function is set as x″, the Gaussian distribution function is F, the function variance is k′, h5 represents a random number and h5∈(0,1), and the average position of fishermen in different dimensions is In the k+1th iteration, the hth fisherman’s position is updated to The calculation formula is as follows:

[0130]

[0131] The fisherman continuously iterates his own position, and sets the maximum number of iterations to K. When the current number of iterations reaches the maximum number of iterations, the final fisherman position and the best fitness function value are output. In the g search spaces, the dimension of the final fisherman position in the fisherman matrix corresponds to the best relevant hydrocyclone working parameter in the processed relevant hydrocyclone working parameter set.

[0132] Example 2

[0133] This embodiment also discloses a system for intelligent monitoring method of working state of a hydrocyclone based on data analysis, which specifically includes: a hydrocyclone working parameter correlation analysis module, a relevant hydrocyclone working parameter preprocessing module, a hydrocyclone working state monitoring module and a relevant hydrocyclone working parameter optimization module;

[0134] The hydrocyclone operating parameter correlation analysis module is used to perform correlation analysis on the hydrocyclone operating parameters;

[0135] The relevant hydrocyclone operating parameter preprocessing module is used to delete abnormal values ​​and fill data of the relevant hydrocyclone operating parameters;

[0136] The hydrocyclone working state monitoring module is used to use the twin neural network model to determine whether the relevant hydrocyclone working parameters are abnormal;

[0137] The relevant hydrocyclone operating parameter optimization module is used to optimize the relevant hydrocyclone operating parameters using a fishing optimization algorithm.

[0138] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0139] The preferred embodiments of the invention disclosed above are only used to help explain the invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the invention, so that those skilled in the art can understand and use the invention well.

Claims

1. An intelligent monitoring method for the working status of a hydrocyclone based on data analysis, characterized in that: The steps include: S1. Obtaining the working parameters of the hydrocyclone, performing correlation analysis on the working parameters of the hydrocyclone, and obtaining a correlation coefficient matrix, wherein the correlation coefficient matrix obtains relevant working parameters of the hydrocyclone according to positive correlation and negative correlation; S2, selecting relevant hydrocyclone working parameters to form a relevant hydrocyclone working parameter set, performing outlier deletion on the relevant hydrocyclone working parameter set to obtain an outlier-deleted relevant hydrocyclone working parameter set; then performing data filling on the outlier-deleted relevant hydrocyclone working parameter set to obtain a processed relevant hydrocyclone working parameter set; S3, labeling the processed set of related hydrocyclone working parameters, training a twin neural network, obtaining a twin neural network model, outputting similarity, determining whether the related hydrocyclone working parameters are abnormal through similarity, and realizing monitoring of the working status of the hydrocyclone; S4. After the working state monitoring of the hydrocyclone is realized, the relevant working parameters of the hydrocyclone are optimized to obtain the best relevant working parameters of the hydrocyclone in the processed relevant working parameter set of the hydrocyclone.

2. The method for intelligently monitoring the working state of a hydrocyclone based on data analysis according to claim 1 is characterized in that: The S1 comprises the following steps: S11, obtaining the working parameters of the hydrocyclone, and using the Pearson method to perform correlation calculation on the working parameters of the hydrocyclone to obtain a correlation coefficient matrix; S12. Compare the correlation coefficients in the correlation coefficient matrix, and obtain relevant hydrocyclone operating parameters according to positive correlation and negative correlation.

3. According to the method for intelligent monitoring of the working state of a hydrocyclone based on data analysis in claim 2, the correlation coefficient matrix obtains the relevant hydrocyclone working parameters according to the positive correlation and negative correlation. The specific process is as follows: A correlation threshold ω is set. When the correlation coefficient in the correlation coefficient matrix is ​​greater than the correlation threshold ω, the correlation coefficient in the correlation coefficient matrix is ​​retained; otherwise, the correlation coefficient in the correlation coefficient matrix is ​​set to 0 to obtain a processed correlation coefficient matrix. In the processed correlation coefficient matrix, the hydrocyclone operating parameters corresponding to the positive correlation or negative correlation of the hydrocyclone operating parameters are found and recorded as the relevant hydrocyclone operating parameters.

4. The method for intelligently monitoring the working state of a hydrocyclone based on data analysis according to claim 3 is characterized in that: The S2 comprises the following steps: S21, selecting the first e samples of the relevant hydrocyclone working parameters to form a relevant hydrocyclone working parameter set, and performing outlier deletion on the relevant hydrocyclone working parameter set to obtain an outlier-deleted relevant hydrocyclone working parameter set; S22, removing the abnormal value from the relevant hydrocyclone operating parameter set, filling the missing data based on the cubic spline difference method, and obtaining the processed relevant hydrocyclone operating parameters.

5. The method for intelligently monitoring the working state of a hydrocyclone based on data analysis according to claim 4 is characterized in that: The S21 comprises the following steps: S211, using the Thompson tau method to delete outliers from the relevant hydrocyclone working parameter set, completing the first outlier deletion, and obtaining the first outlier deletion parameter set; S212: Perform outlier removal again on the first outlier removal parameter set based on the quartile method to complete the second outlier removal.

6. The method for intelligently monitoring the working state of a hydrocyclone based on data analysis according to claim 5 is characterized in that: The S3 comprises the following steps: S31, setting data labels C1 and C2 for the processed set of related hydrocyclone working parameters, wherein C1 represents the sample data in the processed set of related hydrocyclone working parameters, and C2 represents the theoretical sample data in the processed set of related hydrocyclone working parameters; setting the category label to C3, wherein C3=0 represents that the sample data in the processed set of related hydrocyclone working parameters is normal, and C3=1 represents that the sample data in the processed set of related hydrocyclone working parameters is abnormal; Normalizing the sample data in the processed related hydrocyclone working parameter set to obtain a normalized related hydrocyclone working parameter set; S32, obtaining new working parameters of the hydrocyclone, generating a new normalized set of related working parameters of the hydrocyclone, dividing the new normalized set of related working parameters of the hydrocyclone into a working parameter training set and a working parameter test set, inputting the working parameter training set into the twin neural network, setting the loss function f, and continuously updating the weights through back propagation until the twin neural network converges to obtain a trained twin neural network; then inputting the working parameter test set into the trained twin neural network, setting the accuracy threshold to ξ, and when the output result accuracy is less than ξ, obtaining the twin neural network model, otherwise repeating S32 until the output result accuracy is less than ξ; S33. Input the normalized related hydrocyclone working parameter set into the twin neural network model. The twin neural network model outputs a similarity result, which is recorded as a similarity set. When the similarity in the similarity set is greater than or equal to 0.6, the sample data category label C3 corresponding to the similarity in the similarity set is 0, otherwise the sample data category label C3 corresponding to the similarity in the similarity set is 1; whether the related hydrocyclone working parameters are abnormal is obtained according to the similarity in the similarity set, so as to realize the working status monitoring of the hydrocyclone.

7. The method for intelligently monitoring the working state of a hydrocyclone based on data analysis according to claim 6 is characterized in that: The S4 comprises the following steps: S41, assigning weights to the relevant hydrocyclone working parameters in the processed set of relevant hydrocyclone working parameters, calculating the working efficiency of the hydrocyclone, and using the hydrocyclone working efficiency calculation formula as the fitness function; setting g search spaces, where g represents the number of relevant hydrocyclone working parameters in the processed set of relevant hydrocyclone working parameters, in each search space, treating sample data as fishermen, the fisherman dimension is j′, and establishing a fisherman matrix; The fishermen are in the search space. Set the upper bound of the search space to g′, the lower bound of the search space to g″, h1 represents a random number and h1∈(0,1), then the initial position of the hth fisherman is S42, in the exploration phase, fishermen adjust their positions according to the fitness function value, constantly looking for the best sample data, and evaluate the catch, setting the capture speed parameter as ε and the capture probability as φ; When φ<ε, the fishermen choose to capture independently, and the number of iterations is set to k. The best fitness function value of the kth iteration is f′(k), the worst fitness function value of the kth iteration is f″(k), the current fitness function value is f1, and the fitness function value of any previous period is f2. The position of the hth fisherman in the kth iteration is The position of the hth fisherman in the k+1th iteration is The position of the h′th fisherman in the k+1th iteration is h2 represents a random number and h2∈(0,1), h3 represents a random unit vector, and the detection range is D, then the capture coefficient The calculation formula for the hth fisherman's position in the k+1th iteration is as follows: When φ≥ε, the fishermen select a group to capture, set the target point surrounded by the group to x′, the speed of the fishermen approaching the target point is v1, h4 represents the moving offset, and the position of the hth fisherman in the kth iteration is updated to In the k+1th iteration, the hth fisherman’s position is updated to The calculation formula is as follows: S43. During the collective capture phase, fishermen adjust their positions and continuously search for the position of fishermen corresponding to the best fitness function. The position corresponding to the current best fitness function is set as x″, the Gaussian distribution function is F, the function variance is k′, h5 represents a random number and h5∈(0,1), and the average position of fishermen in different dimensions is In the k+1th iteration, the hth fisherman’s position is updated to The calculation formula is as follows:

8. The method for intelligently monitoring the working state of a hydrocyclone based on data analysis according to claim 7 is characterized in that: The fisherman continuously iterates his own position, and sets the maximum number of iterations to K. When the current number of iterations reaches the maximum number of iterations, the final fisherman position and the best fitness function value are output. In the g search spaces, the dimension of the final fisherman position in the fisherman matrix corresponds to the best relevant hydrocyclone working parameter in the processed relevant hydrocyclone working parameter set.

9. A system for realizing the method for intelligently monitoring the working state of a hydrocyclone based on data analysis as claimed in any one of claims 1 to 8, characterized in that: Specifically include: A hydrocyclone working parameter correlation analysis module, a related hydrocyclone working parameter preprocessing module, a hydrocyclone working state monitoring module and a related hydrocyclone working parameter optimization module; The hydrocyclone operating parameter correlation analysis module is used to perform correlation analysis on the hydrocyclone operating parameters; The relevant hydrocyclone operating parameter preprocessing module is used to delete abnormal values ​​and fill data of the relevant hydrocyclone operating parameters; The hydrocyclone working state monitoring module is used to use the twin neural network model to determine whether the relevant hydrocyclone working parameters are abnormal; The relevant hydrocyclone operating parameter optimization module is used to optimize the relevant hydrocyclone operating parameters using a fishing optimization algorithm.

Images