CN108763729B - Process industry electromechanical system coupling state evaluation method based on network structure entropy - Google Patents
Process industry electromechanical system coupling state evaluation method based on network structure entropy Download PDFInfo
- Publication number
- CN108763729B CN108763729B CN201810508428.5A CN201810508428A CN108763729B CN 108763729 B CN108763729 B CN 108763729B CN 201810508428 A CN201810508428 A CN 201810508428A CN 108763729 B CN108763729 B CN 108763729B
- Authority
- CN
- China
- Prior art keywords
- coupling
- variables
- monitoring
- network structure
- dcca
- 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
Images
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
Abstract
The invention discloses a flow industrial electromechanical system coupling state evaluation method based on network structure entropy, which comprises the steps of firstly obtaining a quasi-period of a sequence through an FFT method so as to determine the time window width of coupling analysis, utilizing a DCCA algorithm to calculate the correlation between every two multivariable variables, constructing a weighted network model reflecting multivariable coupling relation, timely obtaining the change of the coupling relation between process monitoring variables, realizing the rapid and accurate scheduling of the upstream and downstream of a system, realizing the refined control of the system, utilizing a NSEn method to calculate the entropy value of a monitoring variable coupling relation network model in each time window, utilizing the coupling incidence relation among monitoring data, intuitively reflecting the dynamic coupling process of different parts of the system through the dynamic change of a network topological structure, quantitatively representing the state evolution process of the system through the network structure entropy, and providing comprehensive scheduling and maintenance decision information for system managers, the scientificity and the intellectualization level of safe and reliable operation decision of the complicated electromechanical system of the process industry under the complicated working condition are improved.
Description
Technical Field
The invention relates to the field of complex electromechanical system service safety state evaluation, in particular to a DCCA-NSEn-based process industrial electromechanical system coupling network modeling and evaluation method.
Background
The process industrial production system has various production equipment and needs various auxiliary systems, the structural units continuously exchange materials, information and energy, and the system has high internal association coupling degree and is a distributed complex electromechanical system. Equipment failure and process adjustment often cause systematic fluctuation, timely and accurately find operation failure in the industrial process and reasonably evaluate the recovery degree of the failure process, and are particularly important for reasonable regulation and control of the upstream and downstream of a flow system. The scheduling personnel frequently generates the situations of over-scheduling or untimely scheduling and the like for the sake of safety according to the previous scheduling experience for scheduling the upstream and the downstream of the system, thereby causing the interruption of production or the reduction of production load, and possibly causing huge economic loss. Therefore, an effective state evaluation method is found to evaluate the system running state timely and accurately, so that a reliable system real-time state is provided for dispatching personnel, and the urgent need of reducing economic loss caused by dispatching discomfort is reduced. In the aspect of comprehensive state evaluation of a complex electromechanical system, Li et al provide a method for evaluating the state of a power transformer according to information data association rules and a variable weight comprehensive concept of a factor space theory. The project, Yangqi and the like provide a state evaluation method based on monitoring data for an electric energy metering device. On the basis of evaluating the health state of equipment parameters, a health state evaluation model of the equipment based on an improved evidence theory is established, and the health states of all parameters are synthesized and decided. The researches are carried out on different specific objects such as transformers and the like, and good research results are obtained, but for distributed complex electromechanical systems which have complex working conditions, numerous monitoring variables and complex coupling relations and have redundant and insufficient monitoring information, the research methods are still deficient.
Disclosure of Invention
The invention aims to provide a process industrial electromechanical system coupling state evaluation method based on network structure entropy to overcome the defects of the prior art.
In order to achieve the purpose, the invention adopts the following technical scheme:
the process industry electromechanical system coupling state evaluation method based on the network structure entropy specifically comprises the following steps:
step 1), selecting a variable set of a monitoring target of a complex electromechanical system to be analyzed, acquiring a monitoring data set of a certain time course from a DCS monitoring system by the monitoring variable set, wherein the acquired monitoring time sequence data set is an n-dimensional monitoring time sequence matrix;
step 2), preprocessing the obtained monitoring data set, specifically comprising time series noise reduction, deleting variables with small information quantity, self-adaptive fusion of redundant features and determination of a sliding window;
step 3), performing trend-removing cross analysis on the preprocessed monitoring sequence data to determine whether a coupling relation exists between each pair of variables, and if so, calculating a trend-removing cross coefficient;
step 4), establishing a complex electromechanical system coupling network model by taking the monitoring variable as a node, taking the coupling relation as an edge and taking the magnitude of the coupling coefficient as the weight of the edge;
and 5) establishing a coupling network for each sliding window by setting sliding step length, and forming a system service performance state evolution curve by utilizing the coupling state of the network structure entropy quantitative analysis system so as to finish the coupling evaluation of the industrial electromechanical system.
Furthermore, the sampling frequency of the monitoring sequence needs to be set according to the sampling cost and the monitoring precision, the length of the sample is set, and a monitoring data set is obtained from historical data of the system operation process.
Further, the method for preprocessing the monitoring data set specifically comprises the following steps:
(1) carrying out data normalization on the data of different sources, and carrying out noise reduction on the normalized data by adopting a wavelet packet method;
(2) calculating the information quantity of each variable by using the information entropy of the time series, setting a threshold value R, and removing the variables of which the information quantity is less than R;
(3) the redundant variables are fused by using a self-adaptive weighting fusion method, so that the subsequent calculation complexity is reduced;
(4) the selection of the time window width is to analyze the pseudo-period of the time sequence through an algorithm to represent the change period of the time sequence, namely when the sequence length is greater than the pseudo-period of the variable, the time sequence can better reflect the characteristics of the variable.
Further, an FFT algorithm is adopted to calculate a plurality of quasi-periods with variable of chaotic characteristics to obtain n time series quasi-periods T1，T2，…,TnIn order to make the sequence length reflect the characteristics of each variable as much as possible, the variable with the longest pseudo-period in the n variables is taken as a reference, and 2 times T of the longest period is taken as 2max (T)1,T2,…,Tn) The length of the sequence is determined as the time window width of the sequence.
Further, a coupling relation between variables is qualitatively analyzed by a cross-trending analysis (DCCA) method, if coupling exists, DCCA coefficients of coupled variable pairs are calculated, and the DCCA coefficients are used as coupling strength between the variables.
Further, by utilizing the construction of DCCA coefficient network:
time series x for n variables1,x2,x3,…,xnRespectively calculating DCCA coefficient between two DCCA coefficients, DCCA (x)1,x1)，DCCA(x1,x2)，...，DCCA(xn,xn) Forming a DCCA matrix of n × n as follows:
in the above formula d11To dnnIs the DCCA coefficient, x, between variables1To xnFor selected n variables to be evaluated, where dijThe DCCA coefficient between variable i and variable j represents the correlation between two variables, the coupling degree between n variables forms a coupling degree network (DCCAnet) of n × n, and the DCCA method is symmetrical, so dijValue of and djiEqually, the DCCAnet matrix is a symmetric matrix.
Further, the advantage of the network structure entropy in the aspect of representing network heterogeneity is applied, and the network structure entropy of the system of the monitoring data in each sliding window is quantitatively analyzed; and determining and adjusting the slip STEP according to different precision requirements.
Further, a reasonable threshold value of the network structure entropy during normal operation of the system is calculated by using the data set obtained during normal operation of the system.
Further, the abnormal degree of the running state of the complex electromechanical system is quantitatively judged according to the magnitude value of a reasonable threshold value determined when the entropy change curve of the network structure obtained by real-time calculation exceeds the system normal state.
Compared with the prior art, the invention has the following beneficial technical effects:
the invention relates to a flow industry electromechanical system coupling state evaluation method based on network structure entropy, which comprises the steps of firstly solving a quasi-period of a sequence, determining the time window width of coupling analysis, calculating the correlation between every two multivariants, constructing a weighted network model reflecting the multivariable coupling relation, monitoring the slippage of the sequence time window in a certain step length to obtain a system coupling relation network dynamic evolution model, timely obtaining the change of the coupling relation between process monitoring variables, realizing the rapid and accurate scheduling of the upstream and downstream of a system, further realizing the fine control of the system and being beneficial to improving the safety service level of a production flow industry system; the method comprises the steps of calculating the entropy value of a monitoring variable coupling relation network model in each time window, comprehensively evaluating the service evolution state of the complex electromechanical system according to the change trend of entropy along with time, establishing a coupling network model reflecting the operation mechanism of the complex electromechanical system in the process industry by using the coupling incidence relation among monitoring data, generating different coupling network topological structures by the system at different service stages, visually reflecting the dynamic coupling process of different parts of the system through the dynamic change of the network topological structures, and quantitatively representing the state evolution process of the system by using the entropy of the network structures, thereby providing comprehensive scheduling and maintenance decision information for system managers, and improving the scientificity and intelligentization level of safe and reliable operation decision of the complex electromechanical system in the process industry under the complex working condition; the method not only realizes the comprehensive characterization of the service state of the system, but also can determine the abnormal part of the system through the dynamic evolution process of the network topology structure, realize the comprehensive perception of the service state of the system and further implement effective service safety control on the system.
Furthermore, the method can not only provide a comprehensive index change curve of the system, but also discover the evolution process of the inherent coupling relation network model of the system and the local network structure of the system in the dynamic evolution process of the constructed system coupling network topological structure, so as to provide comprehensive information for the operation decision of the complex electromechanical system, thereby providing a reliable basis for the accurate management and control of the service safety of the system.
Drawings
FIG. 1 is a flow chart of modeling and evaluating a DCCA-NSEn-based complex electromechanical system coupling network.
FIG. 2 is a schematic diagram of a process of modeling and evaluating a complex electromechanical system network in a process industry.
FIG. 3 is a diagram of variable 1-8 monitoring time series trend in normal service state.
Fig. 4 is a diagram of the effect before and after noise reduction of the system monitoring variable G _ AVIR _0401, (a) is a diagram of the effect before noise reduction of the system monitoring variable G _ AVIR _0401, and (b) is a diagram of the effect after noise reduction of the system monitoring variable G _ AVIR _ 0401.
Fig. 5 is a graph of the service safety state of a complex electromechanical system in the process industry.
FIG. 6 is a diagram of variable coupling networks in different states of a complex electromechanical system in the process industry.
FIG. 7 is a diagram of the CDFA method for extracting coupling characteristics of complex electromechanical systems in the process industry.
Detailed Description
The invention is described in further detail below with reference to the accompanying drawings:
as shown in fig. 1 to 7, in the method, a pseudo-period of a sequence is obtained through an FFT method to determine a time window width of coupling analysis, a DCCA algorithm is used to calculate a correlation between each two multivariants, a weighted network model reflecting a multivariable coupling relation is constructed, and a sequence time window is monitored to slip in a certain step size to obtain a system coupling relation network dynamic evolution model; the method not only realizes the comprehensive representation of the service state of the system, but also can determine the abnormal part of the system through the dynamic evolution process of the network topological structure, realize the comprehensive perception of the service state of the system and further implement effective service safety control on the system.
The process industry electromechanical system coupling state evaluation method based on the network structure entropy specifically comprises the following steps:
step 1), selecting a variable set of a monitoring target of a complex electromechanical system to be analyzed, acquiring a monitoring data set of a certain time course from a DCS monitoring system by the monitoring variable set, wherein the acquired monitoring time sequence data set is an n-dimensional monitoring time sequence matrix;
step 2), preprocessing the obtained monitoring data set, specifically comprising time-series noise reduction, deleting variables with small information quantity, setting a variable threshold L, deleting variables with a small information quantity residual variable threshold L, adaptively fusing redundant features and determining a sliding window;
step 3), performing trend-removing cross analysis on the preprocessed monitoring sequence data to determine whether a coupling relation exists between each pair of variables, and if so, calculating a trend-removing cross coefficient;
step 4), establishing a complex electromechanical system coupling network model by taking the monitoring variable as a node, taking the coupling relation as an edge and taking the magnitude of the coupling coefficient as the weight of the edge;
and 5) establishing a coupling network for each sliding window by setting sliding step length, and forming a system service performance state evolution curve by utilizing the coupling state of the network structure entropy quantitative analysis system so as to finish the coupling evaluation of the industrial electromechanical system.
The sampling frequency of the monitoring sequence needs to be set according to the sampling cost and the monitoring precision, the length of a sample is set, and a monitoring data set is obtained from historical data of the system operation process;
the pretreatment method for the monitoring data set specifically comprises the following steps:
(1) carrying out data normalization on the data of different sources, and carrying out noise reduction on the normalized data by adopting a wavelet packet method;
(2) calculating the information quantity of each variable by using the information entropy of the time series, setting a threshold value R, and removing the variables of which the information quantity is less than R;
(3) the redundant variables are fused by using a self-adaptive weighting fusion method, so that the subsequent calculation complexity is reduced;
(4) the selection of the time window width is to analyze the pseudo-period of the time sequence through an algorithm to represent the change period of the time sequence, namely when the sequence length is greater than the pseudo-period of the variable, the time sequence can better reflect the characteristics of the variable.
Calculating a plurality of quasi-periods with variable of chaos characteristic by adopting FFT algorithm to obtain n quasi-periods T of time sequence1，T2，…,TnIn order to make the sequence length reflect the characteristics of each variable as much as possible, the variable with the longest pseudo-period in the n variables is taken as a reference, and 2 times T of the longest period is taken as 2max (T)1,T2,…,Tn) The length of the sequence is determined as the time window width of the sequence.
And qualitatively analyzing the coupling relation between the variables by using a cross-over analysis on trend (DCCA) method, and if coupling exists, calculating DCCA coefficients of the coupling variable pairs, wherein the DCCA coefficients are used as the coupling strength between the variables.
Construction of networks using DCCA coefficients
Time series x for n variables1,x2,x3,…,xnRespectively calculating DCCA coefficient between two DCCA coefficients, DCCA (x)1,x1)，DCCA(x1,x2)，...，DCCA(xn,xn) Forming a DCCA matrix of n × n as follows:
in the above formula d11To dnnAs DCCA coefficients between variables，x1To xnFor the selected n variables to be evaluated. Wherein d isijThe coupling degree network (DCCAnet) of n × n is formed by the coupling degree between n variables, and d is symmetric due to DCCA methodijValue of and djiEqual, the DCCAnet matrix is a symmetric matrix.
And (3) applying the advantages of the network structure entropy in the aspect of representing network heterogeneity, and quantitatively analyzing the network structure entropy of the system of the monitoring data in each sliding window. In this case, the slip STEP is determined and adjusted according to different accuracy requirements.
And calculating to obtain a reasonable threshold value of the network structure entropy when the system normally operates by using the data set obtained when the system normally operates.
When a system fails, the coupling relation among variables is abnormal, and the formed system network topology structure changes, which can cause the change of the network structure entropy and exceed a reasonable threshold.
And quantitatively judging the abnormal degree of the running state of the complex electromechanical system according to the magnitude value of a reasonable threshold value determined when the network structure entropy change curve obtained by real-time calculation exceeds the system normal state.
Example (b):
a complex electromechanical system comprising n elements samples the elements to be produced according to a specific time period and records the sampled elements in system variables, and after m sampling periods, the system has m × n system variable data:
the method comprises the following steps: monitoring dataset preprocessing
The heterogeneous data is first normalized by the following formula for each value in variable X:
X(i)＝X(i)/mean(X)
and performing noise reduction processing on the normalized data, wherein a wavelet packet noise reduction method is adopted.
Step two: selection of time window width
The quasi-period of the time series is analyzed by an algorithm to represent the change period of the time series. Namely, when the length of the sequence is greater than the quasi-period of the variable, the time sequence can better reflect the characteristics of the variable.
Here, an FFT algorithm is adopted to calculate a plurality of quasi-periods with chaos characteristic variables to obtain n time-series quasi-periods T1，T2，…,Tn. In order to make the sequence length reflect the characteristics of each variable as much as possible, the variable with the longest period among the n variables is taken as a reference, and 2 times T of the longest period is taken as 2max (T)1,T2,…,Tn) As the time window width of the sequence, i.e. the length of the sequence is determined;
step three: construction of DCCA coefficient network
Time series x for n variables1,x2,x3,…,xnRespectively calculating DCCA coefficient between two DCCA coefficients, DCCA (x)1,x1)，DCCA(x1,x2)，...，DCCA(xn,xn) Forming a DCCA matrix of n × n as follows:
d in formula (11)11To dnnIs the DCCA coefficient, x, between variables1To xnSelecting n variables to be evaluated; wherein d isijThe DCCA coefficient between variable i and variable j represents the correlation between two variables, and the coupling degree between n variables forms a coupling degree network (DCCAnet) of n × nijValue of and djiEqual, so the DCCAnet matrix is a symmetric matrix;
step four: NSEn-based system service comprehensive state characterization
And solving the network structure entropy of the DCCAnet by using an algorithm of the network structure entropy, wherein the slippage STEP can be adjusted according to different monitoring precision requirements.
The method comprises the following steps of judging the overall operation situation of the complex electromechanical system according to a network structure entropy change curve, wherein the two steps are as follows:
① when the system is in normal operation, the entropy of the network structure fluctuates in a certain interval;
② when the system is in fault, the correlation of each variable is abnormal, the entropy of the network structure of DCCAnet will change greatly and exceed the reasonable threshold.
Monitoring time series variable selection
The compressor set is taken as a typical unit of the process industry represented by chemical enterprises, and the safe operation of the compressor set is crucial to the stable operation of the whole process industry production process. When the raw materials are stable, the equipment monitoring data in the general production process can indirectly reflect the operation state of the industrial process, and the fluctuation of the monitoring variable can indirectly reflect the fault of the industrial process, so that the typical process fault data can be applied to process fault early warning and process recovery evaluation research.
This application uses the fault monitoring data of compressor unit before the shutdown of a part of equipment trouble of coal chemical industry enterprise, and its trouble can be described as: the system is in a normal operation state at first, and then abnormal working conditions continuously appear, under the scheduling and control of operators, the service state of the system is improved to a certain extent, but the abnormal state is deteriorated, so that most of equipment must be stopped by being forced to break down and be repaired repeatedly. In the process, 8 monitoring point locations with high fault correlation degree are selected for fault analysis, and detailed information of the monitoring point locations is shown in table 1. The sampling interval of the DCS monitoring data set of the compressor set is 1min, and the effectiveness of the method for evaluating the system state is verified on the basis of fault data of 8 monitoring variables. Trends for these variable monitoring sequences are shown in fig. 3.
TABLE 1 monitoring variable meter for compressor set
2 data preprocessing
Before data analysis, the original monitoring data is converted into a time sequence capable of being analyzed uniformly, and the processing is divided into two steps, namely data normalization processing and data noise reduction processing.
The monitoring data set is normalized, so that the influence of non-uniform heterogeneous data units and non-uniform data scales on analysis results can be eliminated.
The data noise reduction process is divided into two steps of decomposition and reconstruction, ① adopts proper wavelet basis function and decomposition layer number for different variables, then soft closing value processing is carried out on each decomposed wavelet detail coefficient by using a fixed value closing method, ② reconstructs the last layer of approximate coefficient and detail coefficients of all layers to obtain a variable time sequence chart after noise reduction, and the noise reduction effect is shown in figure 4.
3, calculating the width T of a time sequence sliding window
In order to enable the data to reflect the characteristics of the data as much as possible and improve the accuracy of the coupling degree between the calculated variables of the DCCA algorithm, the FFT algorithm is adopted to calculate the quasi-period of each section of sequence, and the period of the time sequence with the maximum quasi-period is taken as a sliding time window T. Namely T ═ max (T (1), T (2), …, T (n)). In this document, 8 variables were selected for analysis in combination, so n-8. The quasi-periodic solution results for each variable are shown in the following table: the window width is 1667, i.e., the time window width is 1667 minutes.
TABLE 2 pseudo-period of variables 1-8
4 analysis and construction of coupling between monitoring variables
As can be seen from the analysis of the coupling change between every two variables, the coupling relation between the variables changes to different degrees in the whole process of the fault.
The DCCA coefficient variation trend among the variables can partially reflect the abnormal state information of the system. However, if the state of the system is reflected by using the coupling degree of some two variables, the system evaluation is too complete, and inaccurate or even wrong conclusions can be drawn. And the work of selecting variables is complex, and when the variables are more and the coupling relation among the variables is difficult to describe, a method capable of evaluating the state of the comprehensive information of the system is needed.
And calculating the correlation among the variables by using a DCCA algorithm, and constructing a network for monitoring the variable coupling degree. The following formula is the coupling network of the window of section 1 of the normal state curve.
In the coupling degree analysis process, the DCCA coefficient can be used for judging the coupling degree of the two variables. The criteria are given in the following table.
TABLE 3 DCCA coefficient-based coupling description
Here, in order to stabilize the analysis result, the degree of identification is high, and the weakly related elements are removed, the removal rule is as shown in the following formula.
NSEn-based system service state evolution rule analysis
The method selects normal and abnormal data of 8 variables to perform comparative analysis, a time window with a window width T1666 slides on a time sequence by a STEP with a certain STEP length, wherein the STEP value is 200, a coupling network of each time interval is obtained, and the network structure entropy is solved. And calculating the change of the network structure entropy and drawing a curve. And selecting points representing normal and abnormal states of the system to construct a coupling network model, and analyzing the change of the coupling relationship network among the variables caused by the state change of the system at different stages. The evaluation curves of the service safety state of different stages of the system are shown in figure 5.
According to analysis, the network structure entropy value of the system in a normal operation state fluctuates between 0.92 and 0.94. Before the fault parking, the system is just in an abnormal state, and the corresponding network structure entropy curve of the abnormal state is obviously deviated from a normal area. And after manual intervention and adjustment are carried out halfway, the service quality state of the system is temporarily improved, then the fault is further worsened, the NSEn value is obviously deviated from a normal region, even reaches 0.98 at the maximum and lasts for several days, and finally the fault is stopped.
In summary, the DCCA-NSEn method disclosed by the present invention is sensitive to the abnormal state perception of the system, and compared with the CDFA multi-characteristic curve of fig. 7, the method not only evaluates the normal state of the system stably, but also can clearly find the abnormal node of the system through the change of the network structure, which is shown in fig. 6, so that the abnormal state of the system can be warned comprehensively and timely, and a reference is provided for the maintenance decision of the scheduling personnel.
The invention aims to fully apply the coupling relation between monitoring sequences reflecting the running state of a system, combine a detrending fluctuation analysis method for analyzing the coupling of monitoring time sequences with a network structure entropy method, and provide a new multivariate-based complex electromechanical system modeling and comprehensive evaluation method, which has the following specific effects:
compared with single-variable and two-variable time series analysis methods such as DFA, DCCA, MFDF and MFDCCA, the method has the advantage of multivariate comprehensive evaluation.
Compared with multivariate time sequence analysis methods such as PCA, KPCA, CDFA and the like, the method has the advantages of stable effect on normal state evaluation and sensitivity on abnormal state perception.
The DCCA-NSEn method not only can provide a comprehensive index change curve of the system, but also can discover the evolution process of the inherent coupling relation network model of the system and the local network structure of the system in the dynamic evolution process of the constructed system coupling network topological structure, so as to provide comprehensive information for the operation decision of the complex electromechanical system, thereby providing a reliable basis for the accurate management and control of the service safety of the system.
Claims (9)
1. The process industry electromechanical system coupling state evaluation method based on the network structure entropy is characterized by specifically comprising the following steps of:
step 1), selecting a variable set of a monitoring target of a complex electromechanical system to be analyzed, acquiring a monitoring data set of a certain time course from a DCS monitoring system by the monitoring variable set, wherein the acquired monitoring time sequence data set is an n-dimensional monitoring time sequence matrix;
step 2), preprocessing the obtained monitoring data set, specifically comprising time series noise reduction, deleting variables with small information quantity, self-adaptive fusion of redundant features and determination of a sliding window;
step 3), performing trend-removing cross analysis on the preprocessed monitoring sequence data to determine whether a coupling relation exists between each pair of variables, and if so, calculating a trend-removing cross coefficient;
step 4), establishing a complex electromechanical system coupling network model by taking the monitoring variable as a node, taking the coupling relation as an edge and taking the magnitude of the coupling coefficient as the weight of the edge;
and 5) establishing a coupling network for each sliding window by setting sliding step length, and forming a system service performance state evolution curve by utilizing the coupling state of the network structure entropy quantitative analysis system so as to finish the coupling evaluation of the industrial electromechanical system.
2. The method for evaluating the coupling state of the process industrial electromechanical system based on the network structure entropy as claimed in claim 1, wherein the sampling frequency of the monitoring sequence needs to be set according to the sampling cost and the monitoring precision, the length of the sample is set, and the monitoring data set is obtained from historical data of the system operation process.
3. The process industry electromechanical system coupling state evaluation method based on the network structure entropy as claimed in claim 1, wherein the method for preprocessing the monitoring data set specifically comprises the following steps:
(1) carrying out data normalization on the data of different sources, and carrying out noise reduction on the normalized data by adopting a wavelet packet method;
(2) calculating the information quantity of each variable by using the information entropy of the time series, setting a threshold value R, and removing the variables of which the information quantity is less than R;
(3) the redundant variables are fused by using a self-adaptive weighting fusion method, so that the subsequent calculation complexity is reduced;
(4) the selection of the time window width is to analyze the pseudo-period of the time sequence through an algorithm to represent the change period of the time sequence, namely when the sequence length is greater than the pseudo-period of the variable, the time sequence can better reflect the characteristics of the variable.
4. The process industrial electromechanical system coupling state evaluation method based on network structure entropy of claim 3, characterized in that an FFT algorithm is adopted to calculate the pseudo-periods of a plurality of variables with chaotic characteristics to obtain n time series pseudo-periods T1，T2，…,TnIn order to make the sequence length reflect the characteristics of each variable as much as possible, the variable with the longest pseudo-period in the n variables is taken as a reference, and 2 times T of the longest period is taken as 2max (T)1,T2,…,Tn) The length of the sequence is determined as the time window width of the sequence.
5. The method for evaluating the coupling state of the process industrial electromechanical system based on the network structure entropy as claimed in claim 1, wherein the coupling relationship between the variables is qualitatively analyzed by a detrending cross analysis method, and if the coupling exists, the DCCA coefficient of the coupled variable pair is calculated and used as the coupling strength between the variables.
6. The process industry electromechanical system coupling state evaluation method based on the network structure entropy according to claim 1, characterized in that a DCCA coefficient network is constructed by:
time series x for n variables1,x2,x3,…,xnRespectively calculating DCCA coefficient between two DCCA coefficients, DCCA (x)1,x1)，DCCA(x1,x2)，...，DCCA(xn,xn) Forming a DCCA matrix of n × n as follows:
in the above formula d11To dnnIs the DCCA coefficient, x, between variables1To xnFor selected n variables to be evaluated, where dijThe DCCA coefficient between variable i and variable j represents the correlation between two variables, the coupling degree between n variables forms a n × n coupling degree network, i.e. DCCAnet matrix, and the DCCA method is symmetrical so that d isijValue of and djiEqually, the DCCAnet matrix is a symmetric matrix.
7. The method for evaluating the coupling state of the process industrial electromechanical system based on the network structure entropy as claimed in claim 1, wherein the advantage of the network structure entropy in characterizing network heterogeneity is applied to quantitatively analyze the network structure entropy of the system of the monitoring data in each sliding window; and determining and adjusting the slip STEP according to different precision requirements.
8. The process industry electromechanical system coupling state evaluation method based on network structure entropy of claim 1, wherein a reasonable threshold of the network structure entropy during normal operation of the system is obtained by calculation using a data set obtained during normal operation of the system.
9. The process industry electromechanical system coupling state evaluation method based on the network structure entropy as claimed in claim 1, wherein the degree of abnormality of the operation state of the complex electromechanical system is quantitatively determined according to the magnitude of the network structure entropy change curve obtained by real-time calculation exceeding a reasonable threshold determined when the system is normal.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810508428.5A CN108763729B (en) | 2018-05-24 | 2018-05-24 | Process industry electromechanical system coupling state evaluation method based on network structure entropy |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810508428.5A CN108763729B (en) | 2018-05-24 | 2018-05-24 | Process industry electromechanical system coupling state evaluation method based on network structure entropy |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108763729A CN108763729A (en) | 2018-11-06 |
| CN108763729B true CN108763729B (en) | 2020-07-28 |
Family
ID=64005597
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810508428.5A Active CN108763729B (en) | 2018-05-24 | 2018-05-24 | Process industry electromechanical system coupling state evaluation method based on network structure entropy |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108763729B (en) |
Families Citing this family (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109614451A (en) * | 2018-12-06 | 2019-04-12 | 南京邮电大学 | Industrial big data intellectual analysis decision making device |
| CN109634233A (en) * | 2018-12-06 | 2019-04-16 | 南京邮电大学 | Industrial big data intellectual analysis decision-making technique, readable storage medium storing program for executing and terminal |
| CN110032146B (en) * | 2019-04-24 | 2020-10-27 | 西安交通大学 | Complex machining process stability evaluation method based on multi-machine synergistic factors |
| CN111080074B (en) * | 2019-11-21 | 2022-07-12 | 西安交通大学 | System service security situation element obtaining method based on network multi-feature association |
| CN111008363B (en) * | 2019-11-21 | 2021-11-19 | 西安交通大学 | Multivariable causal-driven complex electromechanical system service safety situation evaluation method |
| CN111582603B (en) * | 2020-05-19 | 2022-09-13 | 中煤科工集团重庆研究院有限公司 | Intelligent early warning method for coal and gas outburst based on multi-source information fusion |
| CN111861272B (en) * | 2020-07-31 | 2022-12-09 | 西安交通大学 | Multi-source data-based complex electromechanical system abnormal state detection method |
| CN112116198B (en) * | 2020-08-04 | 2023-06-20 | 西安交通大学 | Data-driven process industrial state perception network key node screening method |
| CN112001295B (en) * | 2020-08-19 | 2023-12-08 | 北京航天飞行控制中心 | Performance evaluation method and device of high-speed rotor shaft system, storage medium and processor |
| CN113411821B (en) * | 2021-06-18 | 2021-12-03 | 北京航空航天大学 | System reconfiguration capability evaluation method and system for complex network |
| CN114637263B (en) * | 2022-03-15 | 2024-01-12 | 中国石油大学(北京) | Abnormal working condition real-time monitoring method, device, equipment and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8739081B2 (en) * | 2010-04-30 | 2014-05-27 | International Business Machines Corporation | Method and system for computing fourier series coefficients for mask layouts using FFT |
| CN106682835A (en) * | 2016-12-29 | 2017-05-17 | 西安交通大学 | Data-driven complex electromechanical system service quality state evaluation method |
| CN107248047A (en) * | 2017-08-01 | 2017-10-13 | 西安交通大学 | Process method for evaluating state based on process state entropy dynamic calculation |
| CN107292061A (en) * | 2017-07-28 | 2017-10-24 | 西安交通大学 | A kind of process industry complex electromechanical systems information modelling approach of data-driven |
-
2018
- 2018-05-24 CN CN201810508428.5A patent/CN108763729B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8739081B2 (en) * | 2010-04-30 | 2014-05-27 | International Business Machines Corporation | Method and system for computing fourier series coefficients for mask layouts using FFT |
| CN106682835A (en) * | 2016-12-29 | 2017-05-17 | 西安交通大学 | Data-driven complex electromechanical system service quality state evaluation method |
| CN107292061A (en) * | 2017-07-28 | 2017-10-24 | 西安交通大学 | A kind of process industry complex electromechanical systems information modelling approach of data-driven |
| CN107248047A (en) * | 2017-08-01 | 2017-10-13 | 西安交通大学 | Process method for evaluating state based on process state entropy dynamic calculation |
Non-Patent Citations (4)
| Title |
|---|
| 《Coupling analysis-based false monitoring information identification of production system in process industry》;GAO ZhiYong等;《SCIENCE CHINA》;20170512;第807-817页 * |
| 《Detrended Cross-Correlation Analysis: A New Method for Analyzing Two Nonstationary Time Series》;Boris Podobnik等;《PHYSICAL REVIEW LETTERS》;20080229;第1-4页 * |
| 《一种基于点和边差异性的网络结构熵》;蔡萌等;《物理学报》;20111130;第1-9页 * |
| 《网络结构熵及其在非标度网络中的应用》;谭跃进等;《系统工程理论与实践》;20040630;第1-3页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108763729A (en) | 2018-11-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108763729B (en) | Process industry electromechanical system coupling state evaluation method based on network structure entropy | |
| CN111222290B (en) | Multi-parameter feature fusion-based method for predicting residual service life of large-scale equipment | |
| CN109766334B (en) | Processing method and system for online monitoring abnormal data of power equipment | |
| Ko et al. | Assessment of achievable PI control performance for linear processes with dead time | |
| Lindemann et al. | Anomaly detection and prediction in discrete manufacturing based on cooperative LSTM networks | |
| CN107256449B (en) | State evaluation and assessment method for intelligent substation relay protection device | |
| CN112232447A (en) | Construction method of complete sample set of power equipment state monitoring data | |
| CN104390657A (en) | Generator set operating parameter measuring sensor fault diagnosis method and system | |
| CN112529320A (en) | Intelligent diagnosis system for air compressor cluster | |
| WO2022171788A1 (en) | Prediction model for predicting product quality parameter values | |
| Li et al. | Preventive maintenance decision model of urban transportation system equipment based on multi-control units | |
| CN113036913B (en) | Method and device for monitoring state of comprehensive energy equipment | |
| CN106682159A (en) | Threshold configuration method | |
| CN114580291A (en) | LSTM-VAE-based mobile equipment health condition assessment method | |
| Oberoi et al. | SPC (statistical process control): a quality control technique for confirmation to ability of process | |
| CN112668105A (en) | Helicopter transmission shaft abnormity judgment method based on SAE and Mahalanobis distance | |
| CN115409066A (en) | Method and device for detecting abnormality of time series data, and computer storage medium | |
| CN113902062A (en) | Transformer area line loss abnormal reason analysis method and device based on big data | |
| CN114749996A (en) | Tool residual life prediction method based on deep learning and time sequence regression model | |
| CN114118982A (en) | Cloud computing and edge computing combined production control system | |
| CN113591402A (en) | Digital power transformer health state fuzzy comprehensive evaluation method | |
| CN111367255A (en) | Performance evaluation test system and method for multi-variable control system | |
| CN108123436B (en) | Voltage out-of-limit prediction model based on principal component analysis and multiple regression algorithm | |
| CN117076260B (en) | Parameter and equipment abnormality detection method and device | |
| CN117631632B (en) | PTA mixed feeding control method and system |
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 |