CN112949183B - Abnormal working condition detection system and method for cement raw material vertical mill system - Google Patents
Abnormal working condition detection system and method for cement raw material vertical mill system Download PDFInfo
- Publication number
- CN112949183B CN112949183B CN202110237393.8A CN202110237393A CN112949183B CN 112949183 B CN112949183 B CN 112949183B CN 202110237393 A CN202110237393 A CN 202110237393A CN 112949183 B CN112949183 B CN 112949183B
- Authority
- CN
- China
- Prior art keywords
- data
- module
- vertical mill
- neural network
- abnormal
- 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.)
- Active
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 122
- 230000002159 abnormal effect Effects 0.000 title claims abstract description 100
- 239000004568 cement Substances 0.000 title claims abstract description 90
- 239000002994 raw material Substances 0.000 title claims abstract description 66
- 238000000034 method Methods 0.000 title claims abstract description 56
- 238000013528 artificial neural network Methods 0.000 claims abstract description 44
- 238000000227 grinding Methods 0.000 claims abstract description 36
- 238000007781 pre-processing Methods 0.000 claims abstract description 28
- 238000003062 neural network model Methods 0.000 claims abstract description 16
- 239000011159 matrix material Substances 0.000 claims description 59
- 238000012549 training Methods 0.000 claims description 40
- 230000005856 abnormality Effects 0.000 claims description 38
- 239000013598 vector Substances 0.000 claims description 31
- 238000005070 sampling Methods 0.000 claims description 26
- 238000012545 processing Methods 0.000 claims description 22
- 238000012360 testing method Methods 0.000 claims description 21
- 238000006243 chemical reaction Methods 0.000 claims description 19
- 210000001787 dendrite Anatomy 0.000 claims description 18
- 230000002087 whitening effect Effects 0.000 claims description 15
- 238000004364 calculation method Methods 0.000 claims description 12
- 238000009826 distribution Methods 0.000 claims description 12
- 238000012880 independent component analysis Methods 0.000 claims description 12
- 238000006073 displacement reaction Methods 0.000 claims description 11
- 230000006870 function Effects 0.000 claims description 11
- 230000001965 increasing effect Effects 0.000 claims description 7
- 239000000843 powder Substances 0.000 claims description 7
- XEEYBQQBJWHFJM-UHFFFAOYSA-N Iron Chemical compound [Fe] XEEYBQQBJWHFJM-UHFFFAOYSA-N 0.000 claims description 6
- 235000019738 Limestone Nutrition 0.000 claims description 6
- 230000004913 activation Effects 0.000 claims description 6
- 230000010354 integration Effects 0.000 claims description 6
- 239000006028 limestone Substances 0.000 claims description 6
- 238000012847 principal component analysis method Methods 0.000 claims description 6
- 238000000605 extraction Methods 0.000 claims description 5
- 230000009466 transformation Effects 0.000 claims description 5
- 238000004458 analytical method Methods 0.000 claims description 4
- 230000005540 biological transmission Effects 0.000 claims description 4
- 238000000513 principal component analysis Methods 0.000 claims description 4
- 238000004891 communication Methods 0.000 claims description 3
- 238000000354 decomposition reaction Methods 0.000 claims description 3
- 230000009191 jumping Effects 0.000 claims description 3
- 238000000638 solvent extraction Methods 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 2
- 238000004519 manufacturing process Methods 0.000 abstract description 14
- 230000008569 process Effects 0.000 description 16
- 239000000463 material Substances 0.000 description 7
- 230000008859 change Effects 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 238000013178 mathematical model Methods 0.000 description 4
- 238000011217 control strategy Methods 0.000 description 3
- 238000005265 energy consumption Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 2
- 230000002708 enhancing effect Effects 0.000 description 2
- 235000012054 meals Nutrition 0.000 description 2
- 239000002245 particle Substances 0.000 description 2
- 238000002360 preparation method Methods 0.000 description 2
- 238000003556 assay Methods 0.000 description 1
- 238000010009 beating Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000004807 localization Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000004064 recycling Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000002893 slag Substances 0.000 description 1
- 238000005507 spraying Methods 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/02—Reliability analysis or reliability optimisation; Failure analysis, e.g. worst case scenario performance, failure mode and effects analysis [FMEA]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/22—Yield analysis or yield optimisation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Disintegrating Or Milling (AREA)
- Crushing And Grinding (AREA)
Abstract
The invention discloses an abnormal working condition detection system of a cement raw material vertical mill system, which relates to the technical field of raw material grinding industry of cement vertical mill systems, and comprises a system starting module, a data acquisition module, a data preprocessing module, an abnormal positioning module, a neural network module and a result output module; the invention also discloses a method for detecting abnormal working conditions of the cement raw material vertical mill system, which comprises the steps of S100, starting the cement vertical mill system, S200, collecting detection index data, S300, preprocessing the detection index data, S400, positioning abnormal detection indexes, S500, generating a neural network model, S600 and outputting abnormal working condition detection results. The invention realizes early discovery, early determination and early control on abnormal working conditions, and can improve production efficiency and reduce unqualified products and energy loss.
Description
Technical Field
The invention relates to the technical field of raw material grinding industry of cement vertical mill systems, in particular to a system and a method for detecting abnormal working conditions of a cement raw material vertical mill system.
Background
The cement production process flow is generally called as 'two grinding and one burning', and the raw material vertical grinding process is the first link in the cement preparation process. The cement raw material vertical mill is used for grinding raw materials in the cement preparation process, is an important procedure in the cement production process, and is also a process with more energy consumption. In the actual production process, due to reasons of material feeding quantity, grindability, water spraying quantity and the like, abnormal working conditions such as full grinding, empty grinding, circulation and the like of a raw material vertical grinding system often occur, and unplanned shutdown can be caused when the raw material vertical grinding system is severe, so that the continuity and stability of the grinding process of the vertical grinding machine are seriously affected. When abnormality occurs, a plurality of indexes such as current, power, vibration, roller displacement, sound and the like of the vertical mill are obviously changed. The stability and the operation efficiency of the system directly influence various economic indexes such as the production capacity, the energy consumption and the like of the whole cement production process. At present, the working condition judgment control of the cement raw material vertical mill system is mainly or manually detected and operated, the judgment of an operator on an abnormal working condition depends on personal experience, and when the operator finds the abnormality, the abnormal working condition is continued for a period of time. The adjustment of the control strategy by the operator can only be checked through the actual working condition change in a period of time, and has strong subjectivity and certain delay.
From the mathematical model of the controlled object, the characteristics of the cement raw material vertical mill system, such as nonlinearity, multimodality, multispeed, high dimensionality, intermittence and the like, make the traditional process monitoring method based on the process mechanism model difficult to adapt to the actual industrial process of the cement raw material vertical mill system, and the system is complicated to build an accurate mathematical model.
Cement raw materials enter a cement raw material vertical mill system through a raw material inlet, grinding pressure difference is formed through a grinding roller device and a workbench, raw materials are ground, and ground powdery particles are discharged from a finished product outlet through the pressure difference. The air is heated at the outlet, and the grinding particles are dried. At present, abnormal working conditions of a cement raw material vertical mill system are frequent, and due to the lack of a detection method for the flow of the abnormal working conditions of the system, the production efficiency is low, the energy consumption is seriously wasted, and the like.
Accordingly, those skilled in the art have focused on developing a system and method for detecting abnormal conditions in a cement raw material vertical mill system.
Disclosure of Invention
In view of the above-mentioned drawbacks of the prior art, the present invention aims to solve the technical problem of how to perform control strategy adjustment based on the recognition of abnormal working conditions of a cement raw material vertical mill system, so as to obtain a satisfactory grinding effect.
The inventors divide abnormal production conditions into three types:
first, full grinding, i.e., excessive material storage in the mill, may be caused by excessive feed, poor grindability of the material, or excessive moisture in the material. When saturation wear occurs, the following detection indexes generally change: the mill roller displacement increases, mill power increases, mill current increases, mill outlet temperature increases, mill differential pressure decreases, etc. The feeding amount should be reduced immediately after the full grinding phenomenon occurs, or the rotation speed of the powder concentrator is reduced, and the production is added after the system is normal.
Second, empty mills, i.e., with too little stock in the mill, may be caused by too little feed or good grindability of the material. After the empty grinding, the detection indexes of the grinding machine current, the power, the return hopper current, the grinding roller displacement, the vibration and the like can also obviously change, and the feeding quantity and the like should be increased immediately after the phenomenon occurs.
Third, the circulation, i.e. the grindability of the material is too poor, there is a repeated recycling, grinding and beating cycle of the material. The slag is thrown out from the grinding disc and then returned via the returning hopper, and the detection indexes of the grinding roller displacement, current, power, pressure difference, returning hopper current and the like can change obviously regularly and periodically. After the phenomenon occurs, the rotating speed of the powder selecting machine is reduced, so that the raw materials difficult to grind are discharged.
The inventors have discovered that as a number of new meters, networked meters, and sensing technologies are applied to a production manufacturing full process, a large amount of process data is collected and stored. In the running process of the cement raw material vertical mill system, different detection index values are acquired through various sensors, wherein the detection indexes comprise mill powder selecting machine current, feed back hopper current, circulating fan current, mill voltage, mill vibration, mill roller displacement, mill outlet temperature, mill body pressure difference, mill inlet gas temperature, mill inlet gas pressure, high-temperature fan outlet pressure, grinding pressure and the like. Under ideal working conditions, the detection index values can fluctuate within a certain range, and when abnormal working conditions occur, different detection index values can be distorted according to different types of the abnormal working conditions. Based on the collected historical data, a mathematical model is built, detection index values such as mill powder separator current, return hopper current, circulating fan current, mill voltage, mill vibration, mill roller displacement, mill outlet temperature, mill body pressure difference, mill inlet gas temperature, mill inlet gas pressure, high-temperature fan outlet pressure, mill pressure and the like at the current running time of the system are input into the model, whether abnormal working conditions occur at any position at the moment can be determined, and the abnormal type is judged.
The inventor researches and analyzes that satisfactory grinding effect can be obtained only by adjusting the control strategy based on the recognition of the abnormal working condition of the cement raw material vertical mill system.
In one embodiment of the invention, an abnormal working condition detection system of a cement raw material vertical mill system is provided, which comprises a system starting module, a data acquisition module, a data preprocessing module, an abnormal positioning module, a neural network module and a result output module;
the system starting module starts the cement raw material vertical mill system according to the set running parameters of the cement raw material vertical mill system and sends a starting signal to the data acquisition module;
the data acquisition module comprises a sensor and an upper computer, responds to a starting signal, acquires detection index data under different working conditions according to a set sampling frequency and sampling time, and feeds the detection index data back to the upper computer for recording to construct a historical data set for modeling;
the data preprocessing module performs centering processing and whitening processing on the historical data set, the data with the mean value of 0 is obtained through the centering processing, and the orthogonality of the base vectors is increased; the correlation among detection indexes is removed through the whitening treatment, the extraction process of independent components is simplified, and the convergence of an algorithm is enhanced;
the abnormality locating module uses an independent component analysis method to enable the data of the detection indexes to be independent; performing COX-BOX conversion to data having a Gaussian distribution; extracting data features by adopting a principal component analysis method, determining statistics and a control limit according to a historical dataset, analyzing the contribution value if the contribution value of the statistics exceeds the control limit, selecting a maximum value for reconstruction, determining an abnormal occurrence part of a detection index corresponding to the maximum value of the contribution value if the maximum value does not exceed the control limit, and repositioning if the maximum value still exceeds the control limit;
the neural network module reads in the data preprocessed by the data preprocessing module, divides a training set and a testing set, trains the neural network and generates a neural network model;
the result output module comprises a cement vertical mill system detection display screen, and the detection index values acquired in real time are input into the neural network module through the data preprocessing module and the abnormality positioning module to obtain the operation working condition of the cement vertical mill system, determine whether abnormality occurs, the abnormality occurrence part and the abnormality type and output the abnormality to the cement vertical mill system detection display screen;
the system starting module, the data acquisition module, the data preprocessing module, the abnormality positioning module, the neural network module and the result output module are sequentially in communication connection.
Optionally, in the system for detecting abnormal working conditions of the cement raw material vertical mill system in the above embodiment, the detection indexes include mill classifier current, return hopper current, circulating fan current, mill voltage, mill vibration, mill roller displacement, mill outlet temperature, mill body pressure difference, mill inlet gas temperature, mill inlet gas pressure, high-temperature fan outlet pressure and mill pressure.
Optionally, in the cement raw material vertical mill system abnormal condition detection system of any of the embodiments above, the operating parameters include mill frequency conversion given amount, limestone ratio amount, iron powder ratio amount, sandstone ratio amount, limestone given amount, sandstone given amount, and iron powder given amount.
Optionally, in the system for detecting abnormal working conditions of a cement raw material vertical mill system in any embodiment, the different working conditions include a normal working condition, a full-grinding working condition, an empty-grinding working condition and a circulation working condition.
Optionally, in the system for detecting abnormal working conditions of a cement raw material vertical mill system in any embodiment, training the neural network includes initializing the neural network, performing repeated adjustment training on weights and deviations of the network by applying an integral function and a back propagation algorithm of a dendrite portion in an artificial neural network with dendrites, so that the difference between an output vector and an expected vector is smaller than a given value, completing training when the square sum of errors of output layers of the network is smaller than a specified error, and saving the weights and the deviations of the network to obtain the learning rule.
Based on any one of the above embodiments, in another embodiment of the present invention, a method for detecting abnormal working conditions of a cement raw material vertical mill system is provided, including the following steps:
s100, starting a cement vertical mill system, wherein a system starting module starts the cement raw material vertical mill system according to set operation parameters of the cement raw material vertical mill system and sends a starting signal to a data acquisition module;
s200, acquiring detection index data, responding to a starting signal, and according to a set sampling frequency and sampling time, acquiring the detection index data under different working conditions by a sensor in a data acquisition module, and feeding the detection index data back to an upper computer for recording to construct a data set for modeling;
s300, preprocessing detection index data, performing centering processing and whitening processing on collected historical data by a data preprocessing module, obtaining data with a mean value of 0 through the centering processing, and increasing orthogonality of base vectors; whitening treatment, namely removing correlation among detection indexes, simplifying the extraction process of independent components, and enhancing the convergence of an algorithm;
s400, positioning abnormal detection indexes, wherein the abnormal positioning module uses an independent component analysis method to enable data of the detection indexes to be independent; performing COX-BOX conversion to data having a Gaussian distribution; extracting data characteristics by adopting a principal component analysis method, determining statistics and a control limit according to data of a data set, analyzing the contribution value if the contribution value of the statistics exceeds the control limit, selecting a maximum value for reconstruction, determining an abnormal occurrence part of detection index data corresponding to the maximum value of the contribution value if the maximum value does not exceed the control limit after reconstruction, and repositioning if the maximum value still exceeds the control limit;
s500, generating a neural network model, wherein the neural network module reads in the data preprocessed by the data preprocessing module, divides a training set and a testing set, trains the neural network and generates the neural network model;
s600, outputting an abnormal working condition detection result, wherein the detection index value acquired by the data acquisition module in real time is processed by the data preprocessing module and the abnormal positioning module and then is input into the neural network module to obtain the operation working condition of the cement vertical mill system, whether the abnormality occurs, the abnormality occurrence part and the abnormality type are determined, and the result output module outputs the detection index value to the detection display screen of the cement vertical mill system.
Optionally, in the method for detecting abnormal working conditions of the cement raw material vertical mill system in the above embodiment, the detection indexes include mill classifier current, return hopper current, circulating fan current, mill voltage, mill vibration, mill roller displacement, mill outlet temperature, mill body pressure difference, mill inlet gas temperature, mill inlet gas pressure, high-temperature fan outlet pressure and mill pressure.
Optionally, in the method for detecting abnormal working conditions of the cement raw material vertical mill system in any embodiment, in step S200, the sampling frequency is set to be once per second, the sampling time is set to be twenty-four hours, ten groups of data are collected together, and the ten groups of data contain different working conditions.
Optionally, in the method for detecting abnormal working conditions of the cement raw material vertical mill system in any embodiment, the different working conditions include a normal working condition, a full-grinding working condition, an empty-grinding working condition and a circulation working condition.
Optionally, in the method for detecting abnormal working conditions of a cement raw material vertical mill system in any of the foregoing embodiments, step S300 specifically includes:
s310, acquiring detection index data and writing the detection index data into a column vector X m×n ＝[x 1 ,x 2 ,…,x n ]The elements in the column vector represent the values obtained at the sampling instants, m represents the number of sampling instants, preferably m=10 groups×24 hours×60 minutes×60 seconds= 864000 (sampling frequency is set to once per second, twenty-four hours for ten groups together, m= 864000), where x i (i=1, 2, …, n) represents n detection indicators, preferably n=12;
s320, centering, wherein each acquired detection index data is centered, and the formula is thatObtaining a matrix X ', X' i Column vector X';
s330, whitening treatment, namely obtaining a left matrix U, a diagonal matrix D and a right matrix V by SVD singular value decomposition to obtain a whitened matrix
Optionally, in the method for detecting abnormal working conditions of a cement raw material vertical mill system in any of the foregoing embodiments, step S400 specifically includes:
s410, processing data by adopting an independent component analysis method to obtain a corresponding transformation matrix K;
s420, performing COX-BOX conversion on the conversion matrix K to obtain a COX-BOX data matrix Y, whereinWherein y is i Is the column vector, k, of matrix Y i Is the column vector of matrix K, K ji Represents the j-th row i element of the matrix K, m represents the sampling time number, lambda E (-1.5, 1), preferably lambda=0.8; the data matrix after COX-BOX conversion is Y;
s430, carrying out principal component analysis on the data with Gaussian distribution, and extracting data characteristics;
s440, determining statistics SPE and control limit delta α The processed matrix is S, the row vector is S, the statistics are selected from SPE statistics,the control limit is defined as delta α ，δ α Representing the control limit of SPE at a given confidence level (1-alpha). Times.100%, alpha represents the confidence level, and smaller alpha represents higher confidence; the calculation method comprises the following steps: />g＝θ 1 /θ 2 ，u＝θ 1 2 /θ 2 ，/> λ i The same meaning as above. />The representative confidence is alpha, and the degree of freedom is chi-square distribution of u;
s450 if SPE<δ α Representing that the system is normal; if SPE>δ α If the representative system is abnormal, the next abnormal positioning is carried out.
Optionally, in the method for detecting abnormal working conditions of a cement raw material vertical mill system in any of the foregoing embodiments, step S410 specifically includes:
s411, initializing an independent component matrix, wherein the unit matrix is preferably the independent component matrix;
s412, selecting the comparison function of the independent component analysis, preferably, selecting the comparison function asZ is the row vector corresponding to the whitened matrix Z;
s413, updating the independent component matrix W: the update rule of the independent component matrix W is w=e { zg (W T z)}-E{g′(W T z) } W, each time the update is completed, the decorrelation process is required
S414, setting an update stop condition, preferably, setting the iteration number n=10000, to obtain an independent component matrix W, where the processed transformation matrix is k=w·z.
Optionally, in the method for detecting abnormal working conditions of a cement raw material vertical mill system in any of the foregoing embodiments, step S430 specifically includes:
s431, solving a co-defense difference array COV (Y) of the Y matrix, and solving a characteristic value lambda of the co-defense difference array i (i=1, …, n) and feature vectors;
s432, the eigenvalue lambda i (i=1, …, n) from large to small, and selecting the eigenvectors corresponding to the first h (artificially given) eigenvalues to form a matrix P. The matrix formed by eigenvectors corresponding to the h+1th eigenvalue to n eigenvalues isDirectly performing matrix calculation to obtain s=p·y.
Optionally, in the method for detecting abnormal working conditions of a cement raw material vertical mill system in any of the foregoing embodiments, step S450 specifically includes:
s451, calculating RBC contribution values of each detection index, wherein the contribution value corresponding to the ith detection index is calculated as follows:
s is the same as the above, let us say>ξ i Representing the direction of occurrence of the abnormality, +.>The i element is 1, and the rest elements are 0;
s452, an ith detection index with the largest current contribution value is found out for reconstruction, and a reconstruction formula is as follows: z i ＝s-f i ·ξ i ，f i Representing the abnormal amplitude, the calculation formula is as follows:
s453, re-calculating after reconstructionStatistics and contribution value corresponding to ith detection indexThe calculation formula is as follows: />
S454, confirming abnormal positioning, adding the abnormal detection index into an abnormal variable set, ifThen the ith detection index is determined to be abnormal, the position of the cement vertical mill system corresponding to the ith detection index is determined to be abnormal, and the position of the cement vertical mill system corresponding to the ith detection index is added into an abnormal variable set; if->Then it can be determined that the ith detection index is abnormal and other abnormal detection indexes exist at the same time, the ith detection index is added into the abnormal variable set, then the jth detection index with the second largest contribution value is reconstructed, and S452 to S454 are repeated until +.>(represented as exceeding the control limit), the process ends.
Optionally, in the method for detecting abnormal working conditions of a cement raw material vertical mill system in any of the foregoing embodiments, step S500 specifically includes:
s510, importing and dividing data, partitioning the preprocessed data, wherein 70% of the data is used as a training set, and 30% of the data is used as a testing set;
s520, initializing an artificial neural network with dendrites, preferably setting the number of dendrite integration links to be M (M is more than or equal to 2 and less than or equal to 5), adopting a three-layer neural network structure, namely an input layer, a hidden layer and an output layer, setting 12 nodes of the input layer, 4 nodes of the output layer, setting 100 nodes of the hidden layer, and setting an activation functionx is the value of the input activation function, the connection weight is initialized, the random number between (-1, 1) is assigned by the connection weight, and the maximum learning times N are set max A precision threshold epsilon;
s530, inputting a training set, training an artificial neural network with dendrites, inputting training data, and performing forward transmission calculation through the network to obtainPrediction result Y k ' calculating the sum of squares of the predicted deviations(n 1 Number of training set samples), Y k For training set to expect result, if training times N>N max Taking m=m+1, and jumping to step S520; otherwise judge E 1 Whether or not it is smaller than the precision threshold epsilon, if E 1 Less than or equal to epsilon, training is completed, the step S540 is skipped, otherwise, the connection weight is adjusted by using an error back propagation algorithm, and the step S530 is restarted;
s540, inputting a test set, performing model test and optimization, inputting the test set into a trained artificial dendritic neural network model, and obtaining a prediction result Y through forward network transfer calculation p ' and calculates the sum of squares of the predicted deviations(n 2 For the number of test set samples), Y p To test the set of expected results, judge E 2 If the value of (2) is less than or equal to 1.5 times the precision threshold epsilon, if E 2 >1.5 epsilon, then further adjusting the number of dendrite integration segments m=m+1, otherwise executing S550;
s550, training is finished, the weight and the deviation of the network are saved, learning rules are obtained, and a neural network model is generated.
The invention provides a method for detecting the abnormality of a cement vertical mill system by using an artificial dendritic neural network, which utilizes a large amount of historical data accumulated in the actual production process and data acquired by a sensor to establish a mathematical model, thereby completing the problems of abnormality identification, abnormality positioning and abnormality classification and greatly reducing the influence of abnormality generation on the quality of the produced products in the grinding process of raw materials of the cement vertical mill system; before principal component analysis, independent component analysis is introduced, so that the relevance of data is reduced, the data becomes independent, the abnormality is positioned by using a maximum reconstruction contribution analysis method, and the positioning error or inaccuracy is reduced.
The invention advances the technology of detecting the abnormal working condition of the existing cement raw material vertical mill system, is oriented to the complex industrial working condition process, and can effectively process the problems of multi-mode, nonlinear, non-Gaussian and the like of process data. The detection method has the characteristics of flow and automatic processing, can effectively replace the defects of subjectivity, hysteresis and the like caused by manual detection, can realize early discovery, early determination and early control on abnormal working conditions, can improve the production efficiency, and can reduce unqualified products and energy loss.
The conception, specific structure, and technical effects of the present invention will be further described with reference to the accompanying drawings to fully understand the objects, features, and effects of the present invention.
Drawings
FIG. 1 is a schematic diagram illustrating the composition of a cement raw meal vertical mill system abnormal condition detection system according to an exemplary embodiment;
FIG. 2 is a flowchart illustrating a method of detecting abnormal conditions of a cement raw material vertical mill system according to an exemplary embodiment;
FIG. 3 is a flowchart illustrating an anomaly localization module in accordance with an exemplary embodiment;
fig. 4 is a flowchart illustrating a neural network module according to an example embodiment.
Detailed Description
The following description of the preferred embodiments of the present invention refers to the accompanying drawings, which make the technical contents thereof more clear and easy to understand. The present invention may be embodied in many different forms of embodiments and the scope of the present invention is not limited to only the embodiments described herein.
In the drawings, like structural elements are referred to by like reference numerals and components having similar structure or function are referred to by like reference numerals. The dimensions and thickness of each component shown in the drawings are arbitrarily shown, and the present invention is not limited to the dimensions and thickness of each component. The thickness of the components is schematically and appropriately exaggerated in some places in the drawings for clarity of illustration.
The inventor designs a cement raw meal vertical mill system abnormal condition detecting system, as shown in fig. 1, comprising:
the system starting module starts the cement raw material vertical mill system according to the set operation parameters of the cement raw material vertical mill system and sends a starting signal to the data acquisition module; wherein the operating parameters include mill frequency conversion of a given amount, limestone proportioning amount, iron powder proportioning amount, sandstone proportioning amount, limestone given amount, sandstone given amount, and iron powder given amount;
the data acquisition module comprises a sensor and an upper computer, responds to a starting signal, acquires detection index data under different working conditions according to a set sampling frequency and sampling time, and feeds the detection index data back to the upper computer for recording to construct a historical data set for modeling; wherein the different working conditions comprise a normal working condition, a full grinding working condition, an empty grinding working condition and a circulation working condition; the detection indexes comprise mill powder concentrator current, feed back hopper current, circulating fan current, mill voltage, mill vibration, mill roller displacement, mill outlet temperature, mill body pressure difference, mill inlet gas temperature, mill inlet gas pressure, high-temperature fan outlet pressure and mill pressure;
the data preprocessing module is used for carrying out centering processing and whitening processing on the historical data set, the data with the mean value of 0 is obtained through the centering processing, and the orthogonality of the base vectors is increased; the correlation among detection indexes is removed through the whitening treatment, the extraction process of independent components is simplified, and the convergence of an algorithm is enhanced;
the abnormality positioning module uses an independent component analysis method to enable the data of the detection indexes to be independent; performing COX-BOX conversion to data having a Gaussian distribution; extracting data features by adopting a principal component analysis method, determining statistics and a control limit according to a historical dataset, analyzing the contribution value if the contribution value of the statistics exceeds the control limit, selecting a maximum value for reconstruction, determining an abnormal occurrence part of a detection index corresponding to the maximum value of the contribution value if the maximum value does not exceed the control limit, and repositioning if the maximum value still exceeds the control limit;
the neural network module is used for reading the data preprocessed by the data preprocessing module, dividing a training set and a testing set, training the neural network to generate a neural network model, specifically, training the neural network comprises initializing the neural network, repeatedly adjusting and training the weight and deviation of the network by applying an integral function and a back propagation algorithm of a dendrite part in an artificial neural network with dendrites, so that the difference between the output vector and the expected vector is smaller than a given value, and when the square sum of errors of the output layer of the network is smaller than a specified error, finishing the training, and saving the weight and the deviation of the network to obtain a learning rule;
the result output module comprises a detection display screen of the cement vertical mill system, and the detection index values acquired in real time are input into the neural network module through the data preprocessing module and the abnormality positioning module to obtain the operation working condition of the cement vertical mill system, determine whether abnormality occurs, the abnormality occurrence part and the abnormality type and output the abnormality to the detection display screen of the cement vertical mill system;
the system starting module, the data acquisition module, the data preprocessing module, the abnormality positioning module, the neural network module and the result output module are sequentially in communication connection.
Based on the above embodiments, the inventor provides a method for detecting abnormal working conditions of a cement raw material vertical mill system, as shown in fig. 2, including the following steps:
s100, starting a cement vertical mill system, wherein a system starting module starts the cement raw material vertical mill system according to set operation parameters of the cement raw material vertical mill system and sends a starting signal to a data acquisition module;
s200, acquiring detection index data, responding to a starting signal, and according to a set sampling frequency and sampling time, acquiring the detection index data under different working conditions by a sensor in a data acquisition module, and feeding the detection index data back to an upper computer for recording to construct a data set for modeling; wherein the different working conditions comprise a normal working condition, a full grinding working condition, an empty grinding working condition and a circulation working condition; setting the sampling frequency to be once per second, setting the sampling time to be twenty-four hours, and collecting ten groups of data in total, wherein the ten groups of data comprise different working conditions; the detection indexes comprise mill powder selecting machine current, feed back hopper current, circulating fan current, mill voltage, mill vibration, mill roller displacement, mill outlet temperature, mill body pressure difference, mill inlet gas temperature, mill inlet gas pressure, high-temperature fan outlet pressure and grinding pressure;
s300, preprocessing detection index data, and performing centering processing and whitening processing on collected historical data by a data preprocessing module, wherein the centering processing is performed to obtain data with the mean value of 0, and orthogonality of base vectors is increased. And (3) carrying out centering processing to obtain data with the mean value of 0, and increasing the orthogonality of the base vectors. Whitening treatment, namely removing correlation among detection indexes, simplifying the extraction process of independent components, and enhancing the convergence of an algorithm; the method specifically comprises the following steps:
s310, acquiring detection index data and writing the detection index data into a column vector X m×n ＝[x 1 ,x 2 ,…,x n ]The elements in the column vector represent the values obtained at the sampling instants, m represents the number of sampling instants, preferably m=10 groups×24 hours×60 minutes×60 seconds= 864000 (sampling frequency is set to once per second, twenty-four hours for ten groups together, m= 864000), where x i (i=1, 2, …, n) represents n assays
A measure, preferably n=12;
s320, centering, wherein each acquired detection index data is centered, and the formula is thatObtaining a matrix X ', X' i Column vector X';
s330, whitening treatment, namely obtaining a left matrix U and a diagonal matrix by utilizing SVD singular value decomposition
D, right matrix V, get the matrix after whitening
S400, positioning an abnormal detection index, wherein an abnormal positioning module uses an independent component analysis method to enable data of the detection index to be independent, performs COX-BOX conversion to be converted into data with Gaussian distribution, adopts a principal component analysis method to extract data characteristics, determines statistics and a control limit according to data of a data set, performs contribution value analysis if a contribution value of the statistics exceeds the control limit, selects a maximum value for reconstruction, and can determine an abnormal occurrence part if the contribution value maximum value corresponds to abnormal detection index data and relocates if the contribution value exceeds the control limit; as shown in fig. 3, the method specifically includes:
s410, processing data by adopting an independent component analysis method to obtain a corresponding transformation matrix K; the method specifically comprises the following steps:
s411, initializing an independent component matrix, wherein the unit matrix is preferably the independent component matrix;
s412, selecting the comparison function of the independent component analysis, preferably, selecting the comparison function asZ is the row vector corresponding to the whitened matrix Z;
s413, updating the independent component matrix W: the update rule of the independent component matrix W is w=e { zg (W T z)}-E{g′(W T z) } W, each time the update is completed, the decorrelation process is required
S414, setting an update stop condition, preferably, setting the iteration number n=10000, to obtain an independent component matrix W, where the processed transformation matrix is k=w·z;
s420, performing COX-BOX conversion on the conversion matrix K to obtain a COX-BOX data matrix Y, whereinWherein y is i Is the column vector, k, of matrix Y i Is the column vector of matrix K, K ji Represents the j-th row i element of the matrix K, m represents the sampling time number, lambda E (-1.5, 1), preferably lambda=0.8; the data matrix after COX-BOX conversion is Y;
s430, carrying out principal component analysis on the data with Gaussian distribution, and extracting data characteristics; the method specifically comprises the following steps:
s431, solving a co-defense difference array COV (Y) of the Y matrix, and solving a characteristic value lambda of the co-defense difference array i (i=1, …, n) and feature vectors;
s432, the eigenvalue lambda i (i=1, …, n) from large to small arrangementAnd selecting feature vectors corresponding to the first h (artificially given) feature values to form a matrix P. The matrix formed by eigenvectors corresponding to the h+1th eigenvalue to n eigenvalues isDirectly performing matrix calculation to obtain S=P.Y;
s440, determining statistics SPE and control limit delta α The processed matrix is S, the row vector is S, the statistics are selected from SPE statistics,the control limit is defined as delta α ，δ α Representing the control limit of SPE at a given confidence level (1-alpha). Times.100%, alpha represents the confidence level, and smaller alpha represents higher confidence; the calculation method comprises the following steps: />g＝θ 1 /θ 2 ，λ i The same meaning as above. />The representative confidence is alpha, and the degree of freedom is chi-square distribution of u;
s450 if SPE<δ α Representing that the system is normal; if SPE>δ α If the system is abnormal, performing the next abnormal positioning; the method specifically comprises the following steps:
s451, calculating RBC contribution values of each detection index, wherein the contribution value corresponding to the ith detection index is calculated as follows:
s is the same as the above, let us say>ξ i Representing the direction in which the anomaly occurred,/>the i element is 1, and the rest elements are 0;
s452, an ith detection index with the largest current contribution value is found out for reconstruction, and a reconstruction formula is as follows:
z i ＝s-f i ·ξ i ，f i representing the abnormal amplitude, the calculation formula is as follows:
s453, re-calculating after reconstructionContribution value +.>The calculation formula is as follows: />
S454, confirming abnormal positioning, adding the abnormal detection index into an abnormal variable set, ifThen the ith detection index is determined to be abnormal, the position of the cement vertical mill system corresponding to the ith detection index is determined to be abnormal, and the position of the cement vertical mill system corresponding to the ith detection index is added into an abnormal variable set; if->Then it can be determined that the ith detection index is abnormal and other abnormal detection indexes exist at the same time, the ith detection index is added into the abnormal variable set, and then the jth detection index with the second largest contribution value is addedThe target is reconstructed, repeating S452 to S454 until +.>(represented by exceeding the control limit), the process ends;
s500, generating a neural network model, wherein the neural network module reads in the data preprocessed by the data preprocessing module, divides a training set and a testing set, trains the neural network and generates the neural network model; as shown in fig. 4, the method specifically includes:
s510, importing and dividing data, partitioning the preprocessed data, wherein 70% of the data is used as a training set, and 30% of the data is used as a testing set;
s520, initializing an artificial neural network with dendrites, preferably setting the number of dendrite integration links to be M (M is more than or equal to 2 and less than or equal to 5), adopting a three-layer neural network structure, namely an input layer, a hidden layer and an output layer, setting 12 nodes of the input layer, 4 nodes of the output layer, setting 100 nodes of the hidden layer, and setting an activation functionx is the value of the input activation function, the connection weight is initialized, the random number between (-1, 1) is assigned by the connection weight, and the maximum learning times N are set max A precision threshold epsilon;
s530, inputting a training set, training an artificial neural network with dendrites, and calculating through forward transmission of the network to obtain a prediction result Y k ' calculating the sum of squares of the predicted deviations(n 1 Number of training set samples), Y k For training set to expect result, if training times N>N max Taking m=m+1, and jumping to step S520; otherwise judge E 1 Whether or not it is smaller than the precision threshold epsilon, if E 1 Less than or equal to epsilon, training is completed, and the step S540 is skipped; otherwise, adjusting the connection weight by using the error back propagation algorithm, and restarting step S530;
s540, inputting a test set to perform model testAnd optimizing, namely inputting the test set into the trained artificial dendritic neural network model, and obtaining a prediction result Y through forward transmission calculation of the network p ' and calculates the sum of squares of the predicted deviations(n 2 For the number of test set samples), Y p To test the set of expected results, judge E 2 If the value of (2) is less than or equal to 1.5 times the precision threshold epsilon, if E 2 >1.5 epsilon, then further adjusting the number of dendrite integration segments m=m+1, otherwise executing S550;
s550, finishing training, saving the weight and the deviation of the network, obtaining a learning rule, and generating a neural network model;
s600, outputting an abnormal working condition detection result, wherein the detection index value acquired by the data acquisition module in real time is processed by the data preprocessing module and the abnormal positioning module and then is input into the neural network module to obtain the operation working condition of the cement vertical mill system at the moment, whether the abnormality occurs, the abnormality occurrence part and the abnormality type are determined, and the result output module outputs the detection index value to the detection display screen of the cement vertical mill system.
The foregoing describes in detail preferred embodiments of the present invention. It should be understood that numerous modifications and variations can be made in accordance with the concepts of the invention by one of ordinary skill in the art without undue burden. Therefore, all technical solutions which can be obtained by logic analysis, reasoning or limited experiments based on the prior art by the person skilled in the art according to the inventive concept shall be within the scope of protection defined by the claims.
Claims (10)
1. The abnormal working condition detection system of the cement raw material vertical mill system is characterized by comprising a system starting module, a data acquisition module, a data preprocessing module, an abnormal positioning module, a neural network module and a result output module;
the system starting module starts the cement raw material vertical mill system according to the set operation parameters of the cement raw material vertical mill system and sends a starting signal to the data acquisition module;
the data acquisition module comprises a sensor and an upper computer, responds to the starting signal, acquires detection index data under different working conditions according to a set sampling frequency and sampling time, and feeds the detection index data back to the upper computer for recording to construct a historical data set for modeling;
the data preprocessing module performs centering processing and whitening processing on the historical data set, the centering processing obtains data with the mean value of 0, and orthogonality of base vectors is increased; the whitening treatment removes the correlation between the detection indexes, simplifies the extraction process of independent components, and enhances the convergence of an algorithm;
the abnormality locating module uses an independent component analysis method to enable the data of the detection indexes to be independent; performing COX-BOX conversion to data having a Gaussian distribution; extracting data characteristics by adopting a principal component analysis method, determining statistics and a control limit according to the historical data set, analyzing the contribution value if the contribution value of the statistics exceeds the control limit, selecting a maximum value for reconstruction, determining an abnormal occurrence part of a detection index corresponding to the maximum value of the contribution value if the maximum value does not exceed the control limit, and repositioning if the maximum value of the contribution value still exceeds the control limit;
the neural network module reads in the data preprocessed by the data preprocessing module, divides a training set and a testing set, trains the neural network and generates a neural network model;
the result output module comprises a cement vertical mill system detection display screen, the detection index values acquired in real time are input into the neural network module through the data preprocessing module and the abnormality positioning module to obtain the operation working condition of the cement vertical mill system, and whether abnormality occurs, the abnormality occurrence position and the abnormality type are determined and output to the cement vertical mill system detection display screen;
the system starting module, the data acquisition module, the data preprocessing module, the abnormality positioning module, the neural network module and the result output module are sequentially in communication connection.
2. The system for detecting abnormal conditions of a cement raw material vertical mill system according to claim 1, wherein the detection indexes comprise mill powder separator current, return hopper current, circulating fan current, mill voltage, mill vibration, mill roller displacement, mill outlet temperature, mill body pressure difference, mill inlet gas temperature, mill inlet gas pressure, high-temperature fan outlet pressure and grinding pressure.
3. The system of claim 2, wherein the operating parameters include mill frequency conversion for a given amount, limestone dosing, iron powder dosing, sandstone dosing, limestone for a given amount, sandstone for a given amount, and iron powder for a given amount.
4. A cement raw material vertical mill system anomaly condition detection system as claimed in claim 3 wherein the different conditions include a normal condition, a full mill condition, an empty mill condition and a circulation condition.
5. The abnormal working condition detection system of the cement raw material vertical mill system according to claim 4, wherein the training of the neural network comprises the initialization of the neural network, the integral function of a dendrite part in the artificial neural network with dendrites and a back propagation algorithm are used for repeatedly adjusting and training the weight and deviation of the network, so that the difference between the output vector and the expected vector is smaller than a given value, and when the square sum of errors of the output layer of the network is smaller than a specified error, the training is completed, the weight and the deviation of the network are saved, and the learning rule is obtained.
6. A method for detecting abnormal conditions of a cement raw material vertical mill system using the abnormal conditions detection system of a cement raw material vertical mill system according to any one of claims 1 to 5, comprising the steps of:
s100, starting a cement vertical mill system, wherein the system starting module starts the cement raw material vertical mill system according to the set operation parameters of the cement raw material vertical mill system and sends a starting signal to the data acquisition module;
s200, acquiring detection index data, responding to a starting signal, and according to a set sampling frequency and sampling time, acquiring the detection index data under different working conditions by a sensor in the data acquisition module, and feeding the detection index data back to an upper computer for recording to construct a data set for modeling;
s300, preprocessing detection index data, wherein the data preprocessing module performs centering processing and whitening processing on collected historical data;
s400, positioning an abnormal detection index, wherein the abnormal positioning module uses an independent component analysis method to enable data of the detection index to be mutually independent, performs COX-BOX conversion to obtain data with Gaussian distribution, adopts a principal component analysis method to extract data characteristics, determines statistics and control limits according to data set data, performs contribution value analysis if the contribution value of the statistics exceeds the control limit, selects a maximum value for reconstruction, and if the contribution value exceeds the control limit after reconstruction, the detection index data corresponding to the maximum value of the contribution value is abnormal, and can determine an abnormal occurrence position, and if the contribution value still exceeds the control limit, repositioning;
s500, generating a neural network model, wherein the neural network module reads in the data preprocessed by the data preprocessing module, divides a training set and a testing set, trains the neural network and generates the neural network model;
s600, outputting an abnormal working condition detection result, wherein the detection index value acquired by the data acquisition module in real time is processed by the data preprocessing module and the abnormal positioning module and then is input into the neural network module to obtain the operation working condition of the cement vertical mill system, whether an abnormality occurs, an abnormality occurrence part and an abnormality type are determined, and the result output module outputs the detection index value to the detection display screen of the cement vertical mill system.
7. The method for detecting abnormal working conditions of a cement raw material vertical mill system according to claim 6, wherein the sampling frequency is set to be once per second in the step S200, the sampling time is set to be twenty-four hours, ten groups of data are collected together, and the ten groups of data comprise different working conditions.
8. The method for detecting abnormal conditions of a cement raw material vertical mill system according to claim 6 or 7, wherein the step S300 includes:
s310, acquiring detection index data and writing the detection index data into a column vector X m×n ＝[x 1 ,x 2 ,…,x n ]，m＝864000，n＝12；
S320, centering, wherein each acquired detection index data is centered, and the formula is thatObtaining a matrix X ', X' i Column vector X';
s330, whitening treatment, namely obtaining a left matrix U, a diagonal matrix D and a right matrix V by SVD singular value decomposition to obtain a whitened matrix
9. The method for detecting abnormal conditions of a cement raw material vertical mill system according to claim 8, wherein the step S400 specifically includes:
s410, processing data by adopting an independent component analysis method to obtain a corresponding transformation matrix K;
s420, performing COX-BOX conversion on the conversion matrix K, converting the conversion matrix K into data with Gaussian distribution, and enabling a data matrix after COX-BOX conversion to be Y;
s430, carrying out principal component analysis on the data with Gaussian distribution, and extracting data characteristics;
s440, determining statistics SPE and control limit delta α ，δ α Represents the control limit of SPE at a given confidence level (1- α). Times.100%;
s450 if SPE<δ α Representing that the system is normal; if SPE>δ α If the representative system is abnormal, the next abnormal positioning is carried out.
10. The method for detecting abnormal conditions of a cement raw material vertical mill system according to claim 9, wherein the step S500 specifically includes:
s510, importing and dividing data, partitioning the preprocessed data, wherein 70% of the data is used as a training set, and 30% of the data is used as a testing set;
s520, initializing an artificial neural network with dendrites, setting the number of dendrite integration links to be M (M is more than or equal to 2 and less than or equal to 5), adopting a three-layer neural network structure, namely an input layer, a hidden layer and an output layer, setting 12 nodes of the input layer, 4 nodes of the output layer, setting 100 nodes of the hidden layer, and setting an activation functionx is the value of the input activation function, initializing a connection weight, assigning a random number between (-1, 1), and setting the maximum learning times N max A precision threshold epsilon;
s530, inputting the training set, training the artificial neural network with dendrites, inputting the training set, and obtaining a prediction result Y through forward transmission calculation of the network k ' calculating the sum of squares of the predicted deviationsIf training times N>N max Taking m=m+1, and jumping to step S520; otherwise judge E 1 Whether or not it is smaller than the precision threshold epsilon, if E 1 Less than or equal to epsilon, training is completed, the step S540 is skipped, otherwise, the connection weight is adjusted by using an error back propagation algorithm, and the step S530 is restarted;
s540, inputting the test set, performing model test and optimization, and inputting the test set into the trained artificial dendritic neural network model to obtain a predicted value Y p ' and calculateJudgment E 2 If the value of (2) is less than or equal to 1.5 times the precision threshold epsilon, if E 2 >1.5 epsilon, then further adjusting the number of dendrite integration segments m=m+1, otherwise executing S550;
s550, training is finished, the weight and the deviation of the network are saved, learning rules are obtained, and a neural network model is generated.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110237393.8A CN112949183B (en) | 2021-03-04 | 2021-03-04 | Abnormal working condition detection system and method for cement raw material vertical mill system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110237393.8A CN112949183B (en) | 2021-03-04 | 2021-03-04 | Abnormal working condition detection system and method for cement raw material vertical mill system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112949183A CN112949183A (en) | 2021-06-11 |
| CN112949183B true CN112949183B (en) | 2024-01-16 |
Family
ID=76247513
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110237393.8A Active CN112949183B (en) | 2021-03-04 | 2021-03-04 | Abnormal working condition detection system and method for cement raw material vertical mill system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112949183B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113341870B (en) * | 2021-06-24 | 2022-08-16 | 上海交通大学宁波人工智能研究院 | System and method for recognizing control code exception |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102151605A (en) * | 2011-03-17 | 2011-08-17 | 浙江大学 | Advanced control method and system for vertical mill based on model identification and predictive control |
| CN102319612A (en) * | 2011-07-05 | 2012-01-18 | 中国科学院沈阳自动化研究所 | Method for intelligently controlling pressure difference of cement raw meal vertical mill |
| CN109446236A (en) * | 2018-10-18 | 2019-03-08 | 太原理工大学 | Cement-particle size distribution forecasting method based on random distribution |
| WO2020134801A1 (en) * | 2018-12-26 | 2020-07-02 | 厦门邑通软件科技有限公司 | Energy-saving optimization method of cement raw material vertical mill system |
-
2021
- 2021-03-04 CN CN202110237393.8A patent/CN112949183B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102151605A (en) * | 2011-03-17 | 2011-08-17 | 浙江大学 | Advanced control method and system for vertical mill based on model identification and predictive control |
| CN102319612A (en) * | 2011-07-05 | 2012-01-18 | 中国科学院沈阳自动化研究所 | Method for intelligently controlling pressure difference of cement raw meal vertical mill |
| CN109446236A (en) * | 2018-10-18 | 2019-03-08 | 太原理工大学 | Cement-particle size distribution forecasting method based on random distribution |
| WO2020134801A1 (en) * | 2018-12-26 | 2020-07-02 | 厦门邑通软件科技有限公司 | Energy-saving optimization method of cement raw material vertical mill system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112949183A (en) | 2021-06-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101630376B (en) | Soft-sensing modeling method and soft meter of multi-model neural network in biological fermentation process | |
| CN105278520B (en) | Based on T-KPRM complex industrial process evaluation of running status methods and application | |
| CN101169623B (en) | Non-linear procedure fault identification method based on kernel principal component analysis contribution plot | |
| KR101178235B1 (en) | Prediction and fault detection method and system for performance monitoring of plant instruments using principal component analysis, response surface method, Fuzzy Support Vector Regression and Generalized Likelihood Ratio Test | |
| CN108629091B (en) | Mill load parameter prediction method based on selective fusion multichannel mechanical signal frequency spectrum multi-feature subset | |
| CN110262450B (en) | Fault prediction method for cooperative analysis of multiple fault characteristics of steam turbine | |
| CN112036089A (en) | Coal mill fault early warning method based on DPC-MND and multivariate state estimation | |
| CN108664009A (en) | Divided stages based on correlation analysis and fault detection method | |
| Li et al. | Dynamic time features expanding and extracting method for prediction model of sintering process quality index | |
| CN112949183B (en) | Abnormal working condition detection system and method for cement raw material vertical mill system | |
| CN107239066B (en) | A kind of Vertical Mill operation closed-loop control device and method based on data mining | |
| CN113177358B (en) | Soft measurement method for cement quality based on fuzzy fine-grained feature extraction | |
| Dong et al. | Quality monitoring and root cause diagnosis for industrial processes based on Lasso-SAE-CCA | |
| CN116244625A (en) | Overflow type mill load indirect forecasting method based on multi-feature fusion neural network | |
| CN112257767A (en) | Product key part state classification method aiming at class imbalance data | |
| CN114757269A (en) | Complex process refined fault detection method based on local subspace-neighborhood preserving embedding | |
| CN107015541A (en) | The flexible measurement method being combined based on mutual information and least square method supporting vector machine | |
| CN116821695B (en) | Semi-supervised neural network soft measurement modeling method | |
| CN113858566A (en) | Injection molding machine energy consumption prediction method and system based on machine learning | |
| Soares et al. | Design and application of soft sensor using ensemble methods | |
| CN106407550A (en) | Soft sensor modeling method in industrial process | |
| CN113111588B (en) | NO of gas turbine X Emission concentration prediction method and device | |
| Zhang et al. | Application of neural network predictive control in cement combined grinding | |
| Pani et al. | A hybrid soft sensing approach of a cement mill using principal component analysis and artificial neural networks | |
| CN103994858B (en) | Dynamic balance detection control method based on biogeographical intelligent optimization support vector machine algorithm |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |