CN115829422A - Industrial equipment operation abnormal state identification method based on big data - Google Patents
Industrial equipment operation abnormal state identification method based on big data Download PDFInfo
- Publication number
- CN115829422A CN115829422A CN202310138955.2A CN202310138955A CN115829422A CN 115829422 A CN115829422 A CN 115829422A CN 202310138955 A CN202310138955 A CN 202310138955A CN 115829422 A CN115829422 A CN 115829422A
- Authority
- CN
- China
- Prior art keywords
- data
- index
- matrix
- industrial equipment
- detection
- 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.)
- Granted
Links
- 230000002159 abnormal effect Effects 0.000 title claims abstract description 47
- 238000000034 method Methods 0.000 title claims abstract description 32
- 238000001514 detection method Methods 0.000 claims abstract description 207
- 239000011159 matrix material Substances 0.000 claims abstract description 106
- 238000009826 distribution Methods 0.000 claims abstract description 53
- 238000012216 screening Methods 0.000 claims abstract description 10
- 239000013598 vector Substances 0.000 claims description 14
- 238000004364 calculation method Methods 0.000 claims description 11
- 230000005856 abnormality Effects 0.000 claims description 7
- 238000012545 processing Methods 0.000 abstract description 4
- 238000004458 analytical method Methods 0.000 description 11
- 230000008569 process Effects 0.000 description 6
- 238000010606 normalization Methods 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 239000006185 dispersion Substances 0.000 description 3
- 230000009467 reduction Effects 0.000 description 3
- 238000012512 characterization method Methods 0.000 description 2
- 230000001186 cumulative effect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000010354 integration Effects 0.000 description 2
- 238000000513 principal component analysis Methods 0.000 description 2
- 206010063385 Intellectualisation Diseases 0.000 description 1
- 230000001133 acceleration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000000227 grinding Methods 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 238000009776 industrial production Methods 0.000 description 1
- 238000003801 milling Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000000547 structure data Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Landscapes
- Testing Or Calibration Of Command Recording Devices (AREA)
- Testing And Monitoring For Control Systems (AREA)
Abstract
The invention relates to the technical field of electric digital data processing, in particular to a method for identifying an abnormal operation state of industrial equipment based on big data; acquiring data corresponding to a plurality of detection parameters of the industrial equipment, and calculating the relevance between any two detection parameters; screening a plurality of key detection parameters from the plurality of detection parameters according to the relevance; acquiring data corresponding to the key detection parameters, constructing a key detection parameter matrix and reducing the dimension of the key detection parameter matrix to obtain a reconstructed data matrix; calculating a first distribution index and a second distribution index corresponding to each element in the reconstructed data matrix, and then calculating a state index corresponding to each element according to the first distribution index and the second distribution index; calculating a data comprehensive index based on the reconstructed data matrix and the standard reconstructed data matrix; and obtaining an operation state index according to the state index and the data comprehensive index, and judging whether the industrial equipment is abnormal or not according to the operation state index. The invention can accurately judge whether the running state of the industrial equipment is abnormal.
Description
Technical Field
The invention relates to the technical field of electric digital data processing, in particular to a method for identifying an abnormal operation state of industrial equipment based on big data.
Background
In recent years, with the advance of industry 4.0, the intellectualization and complexity of industrial equipment are greatly increased, and the traditional method for acquiring the running state of the industrial equipment based on a simple model can not meet the requirement of health evaluation of the industrial equipment; that is, when the performance of the industrial equipment slightly changes, the traditional simple model is difficult to detect the slight change, and only can determine that the industrial equipment normally operates, thereby causing a great loss of productivity. Therefore, the anomaly detection of the industrial equipment has great scientific and engineering value.
In order to meet the detection requirement of industrial equipment, data such as frequency, acceleration or pressure of the industrial equipment are collected by adopting a piezoelectric sensor or a handheld data collection tool and the like to detect the industrial equipment at present, on one hand, the two modes need manpower cooperation, on the other hand, the collected data cannot intuitively reflect whether the industrial equipment is abnormal or not, the data needs to be further analyzed to obtain a detection result, most of the data utilizes single data to analyze the abnormal condition of the industrial equipment, and the problem that the detection precision of data sheet and single-source single-structure data is low exists.
Disclosure of Invention
In order to solve the above technical problems, an object of the present invention is to provide a method for identifying an abnormal operation state of an industrial device based on big data, wherein the adopted technical scheme is as follows:
acquiring data corresponding to a plurality of detection parameters of the industrial equipment in a historical time period to obtain a data sequence corresponding to each detection parameter, and calculating the relevance between any two detection parameters based on the data sequence;
constructing a correlation matrix based on the correlation; screening a plurality of key detection parameters from a plurality of detection parameters according to the relevance matrix;
acquiring data corresponding to each detection moment of the key detection parameters in a set time period, constructing a key detection parameter matrix and reducing the dimension of the key detection parameter matrix to obtain a reconstructed data matrix;
randomly selecting an element in the reconstruction data matrix, calculating the difference absolute value of the element and all the elements, and calculating a first distribution index of the element according to the difference absolute value and a difference threshold;
acquiring an element corresponding to a first distribution index of which the first distribution index is larger than the element in a window area with the element as the center, calculating the absolute value of the difference value of the element corresponding to all the acquired elements, and taking the minimum value of the absolute value of the difference value as a second distribution index of the element; the size of the window area is n multiplied by n, wherein n is more than or equal to 3;
calculating a state index corresponding to each element according to the first distribution index and the second distribution index corresponding to each element;
calculating a data comprehensive index according to the maximum value and the minimum value of the elements in the reconstructed data matrix and the maximum value and the minimum value of the elements in the standard reconstructed data matrix;
setting a state threshold, acquiring state indexes corresponding to the state indexes smaller than the state threshold, calculating the accumulated sum of the reciprocals of all the state indexes smaller than the state threshold, taking the product of the accumulated sum and the data comprehensive index as an operation state index, and judging whether the industrial equipment is abnormal or not according to the operation state index.
Preferably, the detection parameters include vibration frequency, power, bearing temperature, motor rotation speed and feeding speed corresponding to the industrial equipment.
Preferably, the relevance calculation method includes: acquiring the frequency of simultaneous failures of any two detection parameters in a historical period, and performing linear fitting on each data sequence to obtain the slope of the corresponding line of each data sequence; calculating discrete coefficients corresponding to the data sequences, and calculating Pearson correlation coefficients of any two data sequences; calculating the relevance between any two detection parameters according to the frequency, the slope, the discrete coefficient and the Pearson correlation coefficient;
the relevance is as follows:
wherein,
the correlation between the ith detection parameter and the jth detection parameter is obtained;
the frequency of abnormality occurring in the ith detection parameter and the jth detection parameter in the historical time period at the same time,
for the ith examinationThe Pearson correlation coefficient between the data sequence corresponding to the detection parameter and the data sequence corresponding to the jth detection parameter is measured;
the slope of the line corresponding to the data sequence for the ith detection parameter,
the slope of the line corresponding to the data sequence for the jth detection parameter,
discrete coefficients corresponding to the data sequence of the ith detection parameter,
discrete coefficients corresponding to the data sequence of the jth detection parameter,
is a vector consisting of a data sequence of the ith detection parameter,
is a vector consisting of a data sequence of the jth detection parameter,
is composed of
And
the inner product of (a) is,
、
are the weight parameters respectively, and are the weight parameters,
in order to be the parameters of the model,
is a natural constant.
Preferably, the element in the ith row in the correlation matrix is a value obtained by normalizing the correlation between the ith detection parameter and the rest of the other detection parameters;
the method for screening out a plurality of key detection parameters from a plurality of detection parameters according to the relevance matrix specifically comprises the following steps: obtaining the maximum value of each row element in the relevance matrix, and recording as the maximum relevance; and when the maximum relevance is smaller than the relevance threshold, both the two detection parameters corresponding to the maximum relevance are key detection parameters.
Preferably, the first distribution index is:
wherein,
to reconstruct the first distribution index corresponding to the element z in the data matrix,
in order to be the difference threshold value,
to reconstruct the absolute value of the difference between the element z and the element s in the data matrix,
to reconstruct the number of elements in the data matrix.
Preferably, the status index is:
wherein,
to reconstruct the state index corresponding to the element z in the data matrix,
to reconstruct the first distribution index corresponding to element z in the data matrix,
to reconstruct the maximum of the first distribution index corresponding to all elements in the data matrix,
a second distribution index corresponding to an element z in the reconstruction data matrix;
to reconstruct the maximum value of the second distribution index corresponding to all elements in the data matrix,
，
respectively, the adjustment parameters.
Preferably, the data comprehensive index is:
wherein,
is a comprehensive index of the data,
to reconstruct the maximum value of the elements in the data matrix,
the maximum value of the elements in the standard reconstruction data matrix,
to reconstruct the minimum of the elements in the data matrix,
is the minimum of the elements in the standard reconstructed data matrix,
is an exponential function with e as the base.
Preferably, the method for judging whether the industrial equipment is abnormal according to the operation state index specifically comprises the following steps: setting an index threshold, comparing the operation state index with the index threshold, and when the operation state index is larger than the index threshold, determining that the operation state of the industrial equipment is abnormal; and when the operation state index is smaller than the index threshold value, the operation state of the industrial equipment is normal.
The embodiment of the invention at least has the following beneficial effects:
according to the invention, the relevance between any two detection parameters is calculated by acquiring the data corresponding to a plurality of detection parameters of the industrial equipment; screening a plurality of key detection parameters from the plurality of detection parameters according to the relevance; the method and the device have the advantages that the calculated amount is reduced, the problems that subjectivity is too strong and redundancy is high when each detection parameter is manually selected are solved, and meanwhile, the problems that data amount is increased and detection efficiency is low when the detection parameters with high relevance are analyzed, so that a plurality of key detection parameters are extracted based on the relevance analysis among the detection parameters, repeated analysis among the detection parameters with too high relevance is avoided, the data analysis precision is improved, and the detection efficiency of the operation state of the industrial equipment is improved. Meanwhile, the state indexes corresponding to all elements in the reconstruction data matrix are calculated according to the first distribution index and the second distribution index, the first distribution index is an overall analysis result, the second distribution index is a local analysis result, the overall and local analysis results are fused through the calculation of the state indexes, and the possibility that all elements in the reconstruction data matrix are abnormal elements can be represented more accurately; and further, whether the running state of the industrial equipment is abnormal or not is accurately judged, and the problems of one side of data and low detection precision of the traditional detection method are solved. The invention judges whether the industrial equipment is abnormal or not through the operation state index, does not participate in the detection of the industrial equipment manually, and has the advantage of automation.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions and advantages of the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without creative efforts.
Fig. 1 is a flowchart illustrating steps of an embodiment of a method for identifying an abnormal operation state of an industrial device based on big data according to the present invention.
Detailed Description
To further explain the technical means and effects of the present invention adopted to achieve the predetermined objects, the following detailed description of the proposed solution, its specific implementation, structure, features and effects will be made with reference to the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" refers to not necessarily the same embodiment. Furthermore, the particular features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
Referring to fig. 1, a flowchart illustrating steps of a big data based identification method for abnormal operation state of industrial equipment according to an embodiment of the present invention is shown, where the method includes the following steps:
step 1, acquiring data corresponding to a plurality of detection parameters of the industrial equipment in a historical time period, acquiring a data sequence corresponding to each detection parameter, and calculating the relevance between any two detection parameters based on the data sequence.
The industrial equipment refers to industrial production equipment and various machine tools, such as lathes, milling machines, grinding machines, planing machines and the like.
Specifically, data corresponding to a plurality of detection parameters of the lathe in a historical time period are obtained, and a data sequence corresponding to each detection parameter is obtained, wherein the detection parameters comprise the vibration frequency, the power, the bearing temperature, the motor rotating speed and the feeding speed of the lathe; the practitioner can adjust the detection parameters according to the particular industrial equipment.
In the embodiment, the vibration frequency is obtained through a vibration detector, the vibration detector is arranged at the side position of the lathe, and an implementer sets the position of the vibration detector according to the actual condition; the power acquires data of voltage and current of the lathe through voltage and current equipment, and calculates the power according to the acquired voltage and current data; the rotating speed and the feeding speed of the motor are collected through a Hall sensor; the bearing temperature is collected by a patch type temperature sensor arranged on the bearing.
Preferably, for each detection parameter, data corresponding to the detection parameter in a history period is obtained, where the history period in this embodiment is half a year, and the data corresponding to each detection parameter in the history period is massive, and in order to reduce the amount of computation, M data of each detection parameter are randomly selected from the data corresponding to each detection parameter in the history period to form a data sequence corresponding to each detection parameter, that is, the length of the data sequence is M. In the actual operation process, the implementer can adjust the value of the historical time period according to specific conditions, for example, the historical time period is set to be one year or three months.
Then calculating the relevance between any two detection parameters according to the data sequence; the relevance calculation method comprises the following steps: acquiring the frequency of simultaneous failures of any two detection parameters in a historical period, and performing linear fitting on each data sequence to obtain the slope of the corresponding line of each data sequence; calculating discrete coefficients corresponding to the data sequences, and calculating Pearson correlation coefficients of any two data sequences; and calculating the relevance between any two detection parameters according to the frequency, the slope, the dispersion coefficient and the Pearson correlation coefficient.
The calculation formula of the relevance is specifically as follows:
wherein,
the correlation between the ith detection parameter and the jth detection parameter is obtained;
the frequency of abnormality occurring in the ith detection parameter and the jth detection parameter in the historical period at the same time,
the Pearson correlation coefficient between the data sequence corresponding to the ith detection parameter and the data sequence corresponding to the jth detection parameter is obtained;
the slope of the line corresponding to the data sequence for the ith detection parameter,
the slope of the line corresponding to the data sequence for the jth detection parameter,
discrete coefficients corresponding to the data sequence of the ith detection parameter,
discrete coefficients corresponding to the data sequence of the jth detection parameter,
is a vector formed by the data sequence of the ith detection parameter,
is a vector consisting of a data sequence of the jth detection parameter,
is composed of
And
the internal product of (a) is,
、
respectively, the weight value parameters are the weight value parameters,
are the parameters of the model and are used as the parameters,
are natural constants. Wherein, the model parameter is more than 1, the weight parameter is more than 0 and less than 1, and the sum of the two weight parameters is 1; model parameters in this example
Weight parameter
Weight parameter
The implementer can adjust the values of the weight parameters and the model parameters according to the actual situation.
The relevance represents the similarity degree between the two detection parameters, and the greater the relevance, the higher the similarity degree between the two corresponding detection parameters.
As can be seen from the above-mentioned analysis,
、
and
the greater the value of (A), the greater the degree of similarity between the corresponding two detection parameters, i.e.
、
And
is in positive correlation with the degree of similarity between the corresponding two detection parameters, and thus
、
And
the correlation between the size of the detection parameter(s) and the corresponding two detection parameters presents a positive correlation;
and
the greater the value of (A), the smaller the degree of similarity between the corresponding two detection parameters, i.e.
And with
Is in a negative correlation with the degree of similarity between the corresponding two detection parameters, and thus
And
the size of the detection parameter(s) and the relevance between the corresponding two detection parameters present a negative correlation; based on this, the calculation formula of the relevance is obtained by the mathematical modeling method, and the positive correlation and the negative correlation of each factor and the relevance are satisfied.
It should be noted that the calculation methods of the discrete coefficient and the pearson correlation coefficient are all known techniques, and are not in the protection scope of the present invention and are not described again; for a vector composed of data sequences, the length of the data sequence is the dimension of the corresponding vector, i.e. in the present embodiment, the length of the data sequence is M, and the dimension of the vector is M.
Step 2, constructing a relevance matrix based on the relevance; and screening a plurality of key detection parameters from the plurality of detection parameters according to the correlation matrix.
Further, the relevance is normalized to obtain a value after the relevance normalization, the value after the relevance normalization is between 0 and 1, and then a relevance matrix is constructed based on the value after the relevance normalization.
The correlation matrix is
Wherein
normalized values for the correlation between the 1 st detected parameter and the 2 nd detected parameter,
normalized values for the correlation between the 1 st detection parameter and the qth detection parameter,
normalizing the value of the correlation between the 2 nd detection parameter and the 1 st detection parameter, and
'；
' is a normalized value of the correlation between the 2 nd detection parameter and the Q-th detection parameter,
' is a normalized value of the correlation between the Q-th detection parameter and the 1 st detection parameter, and
；
normalizing the value of the correlation between the Q & ltth & gt detection parameter and the Q & lt-1 & gt detection parameter; q is the number of the detection parameters, the number of the detection parameters in this embodiment is 5, i.e. Q =5, corresponding to the 5 detection parameters of vibration frequency, power, bearing temperature, motor rotation speed and feeding speed, during the actual operation, the implementer can adjust the detection parameters according to the actual situationThe number of parameters is detected.
In the context of the correlation matrix,
，
' and
' each represents a value obtained by normalizing the relevance between the ith detection parameter and the jth detection parameter, and the element of the ith row in the relevance matrix is the value obtained by normalizing the relevance between the ith detection parameter and the rest other detection parameters; that is, the element in the 1 st row in the correlation matrix is the value obtained by normalizing the correlation between the 1 st detection parameter and the rest of the other detection parameters.
The method for screening out a plurality of key detection parameters from a plurality of detection parameters according to the relevance matrix specifically comprises the following steps: obtaining the maximum value of each row element in the relevance matrix, and recording as the maximum relevance; and when the maximum relevance is greater than the relevance threshold, randomly selecting one detection parameter from the two detection parameters corresponding to the maximum relevance as a key detection parameter, and when the maximum relevance is less than the relevance threshold, both the two detection parameters corresponding to the maximum relevance are key detection parameters. For example, if the maximum correlation corresponding to the first row is
When is coming into contact with
If greater than the relevance threshold, then it will
Taking the corresponding 1 st detection parameter or 2 nd detection parameter as a key detection parameter; when in use
If less than the relevance threshold, then it will
And taking the corresponding 1 st detection parameter and the corresponding 2 nd detection parameter as key detection parameters. Wherein, the relevance threshold is 0.95, and the implementer can set itself according to the actual situation in the actual operation process.
It should be noted that the purpose of screening out a plurality of detection parameters from a plurality of detection parameters is to reduce the amount of calculation, and considering that the problem of high subjectivity, high redundancy and the like of artificially selecting each detection parameter is considered, and meanwhile, if the detection parameters with high relevance are all analyzed, the problems of increased data volume and low detection efficiency are caused.
And 3, acquiring data corresponding to each detection moment of the key detection parameters in a set time period, constructing a key detection parameter matrix and reducing the dimension of the key detection parameter matrix to obtain a reconstructed data matrix, randomly selecting an element in the reconstructed data matrix, calculating the absolute value of the difference between the element and all the elements, and calculating a first distribution index of the element according to the absolute value of the difference and a difference threshold.
In consideration of the fact that most of data corresponding to the key detection parameters are continuous in the operation process of the industrial equipment and analysis of the data is inconvenient, the embodiment discretizes the data, specifically, for each key detection parameter, data corresponding to each detection time of the key detection parameter in a set time period is obtained, and the operation state of the industrial equipment in the set time period is analyzed based on the data in the set time period, that is, the operation condition of the industrial equipment is detected and identified once at intervals (time intervals between two adjacent set time periods).
The present embodiment takes a set time period as an example, and a process of analyzing the operation state of the industrial equipment in the set time period based on data in the set time period is described. The time interval between two adjacent detection moments in the set time interval is 1s, the set time interval is 1 minute, the time interval between two adjacent set time intervals is 5 minutes, and an implementer can set the time interval between two adjacent detection moments and the time interval between two adjacent set time intervals according to specific situations.
The key detection parameter matrix in the above is:
wherein
data corresponding to the mth detection time of the mth key detection parameter in a set time period; m is the number of key detection parameters, an
Q is the number of detection parameters; t is the number of detection instants within a set period, t =60 in the present embodiment.
Further, in order to avoid the influence of different dimensions between data corresponding to each key detection parameter, the key detection parameter data matrix is normalized, and each value in the key detection parameter matrix is ensured to be between 0 and 1.
In order to reduce the complexity of data processing and increase the detection speed of the industrial equipment in the operating state within a set time period, the embodiment performs dimension reduction on the normalized key detection parameter matrix through a principal component analysis algorithm, and records the matrix after dimension reduction as a reconstructed data matrix, and further performs normalization processing on the reconstructed data matrix to ensure that each value in the reconstructed data matrix is between 0 and 1. The dimension reduction of the matrix through the principal component analysis algorithm is a known technology, and is not in the protection scope of the invention, and the specific process is not described again.
Then randomly selecting an element in the reconstructed data matrix, calculating the difference absolute value of the element and all the elements, and calculating a first distribution index of the element according to the difference absolute value and a difference threshold;
the first distribution index is:
wherein,
to reconstruct the first distribution index corresponding to element z in the data matrix,
in order to be the difference threshold value,
to reconstruct the absolute value of the difference between the element z and the element s in the data matrix,
to reconstruct the number of elements in the data matrix. In this embodiment, the difference threshold is 0.3, and in the specific operation process, an implementer can adjust the difference threshold according to the actual situation.
Step 4, randomly selecting an element in the reconstruction data matrix, acquiring an element corresponding to a first distribution index of which the first distribution index is larger than the element in a window area taking the element as a center, calculating the absolute value of the difference value of the element and all the acquired elements, and taking the minimum value of the absolute value of the difference value as a second distribution index of the element; the size of the window area is n × n, and n is equal to or greater than 3.
The second distribution index represents the distribution condition of elements in a window area with any element as the center of the reconstructed data matrix, and the larger the second distribution index is, the higher the difference degree between the element in the center of the window area and other elements in the window area is, the more probable the element in the window area is to be an abnormal element. I.e. the second distribution index expresses the analysis results obtained from locally analyzing each element in the reconstructed data matrix.
The size of the window area in this embodiment is 7 × 7, which can be adjusted by the implementer according to the actual situation.
And 5, calculating the state indexes corresponding to the elements according to the first distribution indexes and the second distribution indexes corresponding to the elements.
The state indexes are as follows:
wherein,
to reconstruct the state index corresponding to the element z in the data matrix,
to reconstruct the first distribution index corresponding to element z in the data matrix,
to reconstruct the maximum value of the first distribution index corresponding to all elements in the data matrix,
a second distribution index corresponding to an element z in the reconstruction data matrix;
to reconstruct the maximum value of the second distribution index corresponding to all elements in the data matrix,
，
respectively are adjusting parameters, and the adjusting parameters are more than 0; in this example
，
The operator can adjust the adjusting parameter to satisfy the adjusting parameter larger than 0.
The calculation of the state index combines the first step index and the second distribution index, the overall and local analysis results are fused, and the possibility that each element in the reconstructed data matrix is an abnormal element can be represented more accurately; the smaller the status indicator, the higher the confidence that the corresponding element is an abnormal element, i.e., the more likely the corresponding element is to be an abnormal element.
Furthermore, in order to facilitate subsequent calculation and further more accurately judge the operating state of the industrial equipment, the state indexes corresponding to all elements in the reconstruction data matrix are normalized, and the value of each state index is ensured to be between 0 and 1.
And 6, calculating a data comprehensive index according to the maximum value and the minimum value of the elements in the reconstructed data matrix and the maximum value and the minimum value of the elements in the standard reconstructed data matrix.
Specifically, the data comprehensive index is as follows:
wherein,
is a comprehensive index of the data,
to reconstruct the maximum of the elements in the data matrix,
is the maximum value of the elements in the standard reconstruction data matrix,
to reconstruct the minimum of the elements in the data matrix,
is the minimum value of the elements in the standard reconstruction data,
is an exponential function with e as the base.
The data comprehensive index represents the difference between the reconstructed data matrix and the standard reconstructed data matrix, and the larger the difference is, the more unstable the operation of the industrial equipment is considered, that is, the higher the possibility that the operation condition of the industrial equipment is abnormal is. Therefore, the larger the data integration index is, the higher the possibility that the operation state of the industrial equipment is abnormal.
The standard reconstruction data matrix is a corresponding reconstruction data matrix when the operation state of the industrial equipment is normal in a set time period.
And 7, setting a state threshold, acquiring state indexes corresponding to the state indexes smaller than the state threshold, calculating the accumulated sum of the inverses of all the state indexes corresponding to the state threshold, taking the product of the accumulated sum and the data comprehensive index as an operation state index, and judging whether the industrial equipment is abnormal or not according to the operation state index.
Specifically, the state threshold is 0.5, and an implementer can set the value of the state threshold according to the actual situation, then obtain the state index smaller than the state threshold, calculate the cumulative sum of the inverses of all the state indexes smaller than the state threshold, and take the product of the cumulative sum and the data comprehensive index as the running state index.
The formula of the running state index is specifically as follows:
wherein,
in order to be an index of the operating state,
is a comprehensive index of the data,
to reconstruct the state index for the xth state less than the state threshold in the data matrix,
the number of state indicators that are less than the state threshold.
Screening out the state indexes which are smaller than the state threshold value and are corresponding to the elements with higher possibility of being abnormal elements, and analyzing the state indexes which are smaller than the state threshold value and are corresponding to the elements with higher possibility of being abnormal elements, so that the calculation amount is reduced; from the analysis in step 5, it is found that the smaller the state index is, the higher the confidence that the element is an abnormal element is, and the more likely the element is to be an abnormal element, and that
The larger the element is, the more likely the element is considered to be an abnormal element, and the more likely the operation state of the industrial equipment is abnormal; data integration index
The larger the operation state of the industrial equipment is, the higher the possibility that the operation state is abnormal is; therefore, the greater the operation state index is, the higher the possibility that the operation state of the industrial equipment is abnormal.
Further, the operation state index is normalized, so that the value of the operation state index is between 0 and 1.
Then, the method for judging whether the industrial equipment is abnormal or not according to the running state index specifically comprises the following steps: setting an index threshold value, judging the sizes of the operation state index and the index threshold value, and when the operation state index is larger than the index threshold value, judging that the operation state of the industrial equipment is abnormal; when the operation state index is smaller than the index threshold value, the operation state of the industrial equipment is normal; the index threshold in this embodiment is 0.45, and an implementer can adjust the value of the index threshold according to actual needs.
The above-mentioned embodiments are only used for illustrating the technical solutions of the present application, and not for limiting the same; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; the modifications or substitutions do not make the essence of the corresponding technical solutions deviate from the technical solutions of the embodiments of the present application, and are included in the protection scope of the present application.
Claims (8)
1. A big data-based industrial equipment operation abnormal state identification method is characterized by comprising the following steps:
acquiring data corresponding to a plurality of detection parameters of the industrial equipment in a historical time period to obtain a data sequence corresponding to each detection parameter, and calculating the relevance between any two detection parameters based on the data sequence;
constructing a correlation matrix based on the correlation; screening a plurality of key detection parameters from a plurality of detection parameters according to the relevance matrix;
acquiring data corresponding to each detection moment of the key detection parameters in a set time period, constructing a key detection parameter matrix and reducing the dimension of the key detection parameter matrix to obtain a reconstructed data matrix;
randomly selecting an element in the reconstruction data matrix, calculating the difference absolute value of the element and all the elements, and calculating a first distribution index of the element according to the difference absolute value and a difference threshold;
acquiring an element corresponding to a first distribution index of which the first distribution index is larger than the element in a window area with the element as the center, calculating the absolute value of the difference value of the element corresponding to all the acquired elements, and taking the minimum value of the absolute value of the difference value as a second distribution index of the element; the size of the window area is n multiplied by n, wherein n is more than or equal to 3;
calculating a state index corresponding to each element according to the first distribution index and the second distribution index corresponding to each element;
calculating a data comprehensive index according to the maximum value and the minimum value of the elements in the reconstructed data matrix and the maximum value and the minimum value of the elements in the standard reconstructed data matrix;
setting a state threshold, acquiring state indexes corresponding to the state indexes smaller than the state threshold, calculating the accumulated sum of the inverses of all the state indexes corresponding to the state threshold, taking the product of the accumulated sum and the data comprehensive index as an operation state index, and judging whether the industrial equipment is abnormal or not according to the operation state index.
2. The method for identifying the abnormal operation state of the industrial equipment based on the big data as claimed in claim 1, wherein the detection parameters comprise vibration frequency, power, bearing temperature, motor rotation speed and feed speed corresponding to the industrial equipment.
3. The method for identifying the abnormal operation state of the industrial equipment based on the big data as claimed in claim 1, wherein the correlation calculation method comprises: acquiring the frequency of simultaneous failures of any two detection parameters in a historical period, and performing straight line fitting on each data sequence to obtain the slope of a straight line corresponding to each data sequence; calculating discrete coefficients corresponding to the data sequences, and calculating Pearson correlation coefficients of any two data sequences; calculating the correlation between any two detection parameters according to the frequency, the slope, the discrete coefficient and the Pearson correlation coefficient;
the relevance is as follows:
wherein,
for the ith detection parameter and the jth detection parameterMeasuring the correlation among the parameters;
the frequency of abnormality occurring in the ith detection parameter and the jth detection parameter in the historical period at the same time,
the Pearson correlation coefficient between the data sequence corresponding to the ith detection parameter and the data sequence corresponding to the jth detection parameter is obtained;
the slope of the line corresponding to the data series for the ith detection parameter,
the slope of the line corresponding to the data sequence for the jth detection parameter,
discrete coefficients corresponding to the data sequence of the ith detection parameter,
discrete coefficients corresponding to the data sequence of the j-th detection parameter,
is a vector formed by the data sequence of the ith detection parameter,
is a vector consisting of a data sequence of the jth detection parameter,
is composed of
And
the inner product of (a) is,
、
are the weight parameters respectively, and are the weight parameters,
in order to be the parameters of the model,
are natural constants.
4. The method for identifying the abnormal operation state of the industrial equipment based on the big data as claimed in claim 1, wherein the element of the ith row in the relevance matrix is a value obtained by normalizing the relevance between the ith detection parameter and the rest of the other detection parameters;
the method for screening out a plurality of key detection parameters from a plurality of detection parameters according to the relevance matrix specifically comprises the following steps: obtaining the maximum value of each row element in the relevance matrix, and recording as the maximum relevance; and when the maximum relevance is smaller than the relevance threshold, both the two detection parameters corresponding to the maximum relevance are key detection parameters.
5. The method for identifying the abnormal operation state of the industrial equipment based on the big data as claimed in claim 1, wherein the first distribution index is as follows:
wherein,
to reconstruct the first distribution index corresponding to element z in the data matrix,
in order to be the difference threshold value,
to reconstruct the absolute value of the difference between the element z and the element s in the data matrix,
to reconstruct the number of elements in the data matrix.
6. The method for identifying the abnormal operation state of the industrial equipment based on the big data as claimed in claim 1, wherein the state indexes are as follows:
wherein,
to reconstruct the state index corresponding to the element z in the data matrix,
to reconstruct the first distribution index corresponding to element z in the data matrix,
to reconstruct the maximum of the first distribution index corresponding to all elements in the data matrix,
a second distribution index corresponding to an element z in the reconstruction data matrix;
to reconstruct the maximum value of the second distribution index corresponding to all elements in the data matrix,
，
respectively, the adjustment parameters.
7. The method for identifying the abnormal operation state of the industrial equipment based on the big data as claimed in claim 1, wherein the data comprehensive indexes are as follows:
wherein,
is a comprehensive index of the data,
to reconstruct the maximum value of the elements in the data matrix,
is the maximum value of the elements in the standard reconstruction data matrix,
to reconstruct the minimum of the elements in the data matrix,
is the minimum of the elements in the standard reconstructed data matrix,
is an exponential function with e as the base.
8. The method for identifying the abnormal operation state of the industrial equipment based on the big data as claimed in claim 1, wherein the method for judging whether the industrial equipment is abnormal or not according to the operation state index specifically comprises the following steps: setting an index threshold, comparing the operation state index with the index threshold, and when the operation state index is larger than the index threshold, determining that the operation state of the industrial equipment is abnormal; and when the operation state index is smaller than the index threshold value, the operation state of the industrial equipment is normal.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310138955.2A CN115829422B (en) | 2023-02-21 | 2023-02-21 | Industrial equipment operation abnormal state identification method based on big data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310138955.2A CN115829422B (en) | 2023-02-21 | 2023-02-21 | Industrial equipment operation abnormal state identification method based on big data |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115829422A true CN115829422A (en) | 2023-03-21 |
| CN115829422B CN115829422B (en) | 2024-01-02 |
Family
ID=85521927
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310138955.2A Active CN115829422B (en) | 2023-02-21 | 2023-02-21 | Industrial equipment operation abnormal state identification method based on big data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115829422B (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116011686A (en) * | 2023-03-28 | 2023-04-25 | 山东创宇环保科技有限公司 | Charging shed photovoltaic power generation reserve prediction method based on multi-data fusion |
| CN116257963A (en) * | 2023-05-16 | 2023-06-13 | 华能济南黄台发电有限公司 | Power plant pipeline state parameter confirmation method and device |
| CN116295539A (en) * | 2023-05-18 | 2023-06-23 | 山东省地质矿产勘查开发局八〇一水文地质工程地质大队(山东省地矿工程勘察院) | Underground space monitoring method based on urban underground space exploration data |
| CN116306931A (en) * | 2023-05-24 | 2023-06-23 | 典基网络科技(上海)有限公司 | Knowledge graph construction method applied to industrial field |
| CN116628616A (en) * | 2023-07-20 | 2023-08-22 | 山东万辉新能源科技有限公司 | Data processing method and system for high-power charging energy |
| CN116773645A (en) * | 2023-08-24 | 2023-09-19 | 中国特种设备检测研究院 | Alloy pipeline carburization damage degree determination method and system and electronic equipment |
| CN116933210A (en) * | 2023-09-15 | 2023-10-24 | 山东荷唯美食品有限公司 | Food processing filling equipment monitoring method and system based on multidimensional sensor |
| CN117193432A (en) * | 2023-09-15 | 2023-12-08 | 浙江迪弗莱包装科技股份有限公司 | Electromagnetic heating roller heating control method and system |
| CN118133210A (en) * | 2024-05-07 | 2024-06-04 | 浙江新再灵科技股份有限公司 | Method and device for extracting abnormal characteristics of elevator human body induction sensor data |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2004165216A (en) * | 2002-11-08 | 2004-06-10 | Matsushita Electric Ind Co Ltd | Production control method and production control apparatus |
| CN110008565A (en) * | 2019-03-28 | 2019-07-12 | 浙江大学 | A kind of industrial process unusual service condition prediction technique based on operating parameter association analysis |
| CN111259730A (en) * | 2019-12-31 | 2020-06-09 | 杭州安脉盛智能技术有限公司 | State monitoring method and system based on multivariate state estimation |
| CN115392408A (en) * | 2022-10-31 | 2022-11-25 | 山东鲁抗医药集团赛特有限责任公司 | Method and system for detecting abnormal operation of electronic particle counter |
-
2023
- 2023-02-21 CN CN202310138955.2A patent/CN115829422B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2004165216A (en) * | 2002-11-08 | 2004-06-10 | Matsushita Electric Ind Co Ltd | Production control method and production control apparatus |
| CN110008565A (en) * | 2019-03-28 | 2019-07-12 | 浙江大学 | A kind of industrial process unusual service condition prediction technique based on operating parameter association analysis |
| CN111259730A (en) * | 2019-12-31 | 2020-06-09 | 杭州安脉盛智能技术有限公司 | State monitoring method and system based on multivariate state estimation |
| CN115392408A (en) * | 2022-10-31 | 2022-11-25 | 山东鲁抗医药集团赛特有限责任公司 | Method and system for detecting abnormal operation of electronic particle counter |
Non-Patent Citations (2)
| Title |
|---|
| 洪斌;邓波;彭甫阳;包阳;冯学伟;: "基于PCA降维的云资源状态监控数据压缩技术" * |
| 王焘;顾泽宇;张文博;徐继伟;魏峻;钟华;: "一种基于自适应监测的云计算系统故障检测方法" * |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116011686A (en) * | 2023-03-28 | 2023-04-25 | 山东创宇环保科技有限公司 | Charging shed photovoltaic power generation reserve prediction method based on multi-data fusion |
| CN116257963A (en) * | 2023-05-16 | 2023-06-13 | 华能济南黄台发电有限公司 | Power plant pipeline state parameter confirmation method and device |
| CN116295539A (en) * | 2023-05-18 | 2023-06-23 | 山东省地质矿产勘查开发局八〇一水文地质工程地质大队(山东省地矿工程勘察院) | Underground space monitoring method based on urban underground space exploration data |
| CN116295539B (en) * | 2023-05-18 | 2023-08-11 | 山东省地质矿产勘查开发局八〇一水文地质工程地质大队(山东省地矿工程勘察院) | Underground space monitoring method based on urban underground space exploration data |
| CN116306931A (en) * | 2023-05-24 | 2023-06-23 | 典基网络科技(上海)有限公司 | Knowledge graph construction method applied to industrial field |
| CN116306931B (en) * | 2023-05-24 | 2023-08-04 | 典基网络科技(上海)有限公司 | Knowledge graph construction method applied to industrial field |
| CN116628616A (en) * | 2023-07-20 | 2023-08-22 | 山东万辉新能源科技有限公司 | Data processing method and system for high-power charging energy |
| CN116628616B (en) * | 2023-07-20 | 2023-09-26 | 山东万辉新能源科技有限公司 | Data processing method and system for high-power charging energy |
| CN116773645A (en) * | 2023-08-24 | 2023-09-19 | 中国特种设备检测研究院 | Alloy pipeline carburization damage degree determination method and system and electronic equipment |
| CN116773645B (en) * | 2023-08-24 | 2023-11-10 | 中国特种设备检测研究院 | Alloy pipeline carburization damage degree determination method and system and electronic equipment |
| CN116933210A (en) * | 2023-09-15 | 2023-10-24 | 山东荷唯美食品有限公司 | Food processing filling equipment monitoring method and system based on multidimensional sensor |
| CN116933210B (en) * | 2023-09-15 | 2023-11-24 | 山东荷唯美食品有限公司 | Food processing filling equipment monitoring method and system based on multidimensional sensor |
| CN117193432A (en) * | 2023-09-15 | 2023-12-08 | 浙江迪弗莱包装科技股份有限公司 | Electromagnetic heating roller heating control method and system |
| CN117193432B (en) * | 2023-09-15 | 2024-10-22 | 浙江迪弗莱包装科技股份有限公司 | Electromagnetic heating roller heating control method and system |
| CN118133210A (en) * | 2024-05-07 | 2024-06-04 | 浙江新再灵科技股份有限公司 | Method and device for extracting abnormal characteristics of elevator human body induction sensor data |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115829422B (en) | 2024-01-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115829422A (en) | Industrial equipment operation abnormal state identification method based on big data | |
| CN112505549A (en) | New energy automobile battery abnormity detection method based on isolated forest algorithm | |
| CN111353482A (en) | LSTM-based fatigue factor recessive anomaly detection and fault diagnosis method | |
| CN111761409A (en) | Multi-sensor numerical control machine tool cutter wear monitoring method based on deep learning | |
| CN116881745B (en) | Pressure transmitter abnormality monitoring method based on big data | |
| CN112284440B (en) | Sensor data deviation self-adaptive correction method | |
| CN116070163B (en) | Indoor harmful gas concentration anomaly monitoring data processing method | |
| CN112904810B (en) | Process industry nonlinear process monitoring method based on effective feature selection | |
| TWI645275B (en) | Abnormality diagnostic apparatus for manufacturing facility | |
| CN107877262A (en) | A kind of numerical control machine tool wear monitoring method based on deep learning | |
| CN105259895A (en) | Method and monitoring system for detecting and separating micro fault in industrial process | |
| CN115409131A (en) | Production line abnormity detection method based on SPC process control system | |
| CN113579851B (en) | Non-stationary drilling process monitoring method based on adaptive segmented PCA | |
| CN114800040B (en) | Cutter wear monitoring method and system related to process-state data | |
| CN111723850A (en) | Automatic verification equipment state evaluation method based on intelligent inspection system | |
| CN114091525A (en) | Rolling bearing degradation trend prediction method | |
| CN112207631B (en) | Method for generating tool detection model, method, system, device and medium for detecting tool detection model | |
| CN113757093A (en) | Fault diagnosis method for flash steam compressor unit | |
| CN117786399A (en) | Remote online monitoring method and system for faults of communication equipment | |
| CN114354184A (en) | Deep learning-based method and device for establishing health early warning model of main shaft of large-scale rotating equipment | |
| CN115034137A (en) | RVM and degradation model-based two-stage hybrid prediction method for residual life of bearing | |
| CN110751217A (en) | Equipment energy consumption ratio early warning analysis method based on principal component analysis | |
| CN117309042A (en) | Intelligent manufacturing data real-time monitoring method and system based on Internet of things technology | |
| CN117783745A (en) | Data online monitoring method and system for battery replacement cabinet | |
| CN114326593B (en) | Cutter life prediction system and method |
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 | ||
| TA01 | Transfer of patent application right | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20231207 Address after: Room 2561, Building 2, East Ring Road, Yanqing Garden, Zhongguancun, Yanqing District, Beijing, 100000 Applicant after: Beijing Hanhai Lanshan Intelligent Technology Co.,Ltd. Address before: Room 102, Building 57, No. 218, Yongxing Road, Yongxing Street, Chongchuan District, Nantong City, Jiangsu Province, 226000 Applicant before: Chuangyin Technology (Nantong) Co.,Ltd. |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant |