CN111461551A - Electric submersible pump fault early warning method based on deep learning and SPC (statistical computer) criterion - Google Patents
Electric submersible pump fault early warning method based on deep learning and SPC (statistical computer) criterion Download PDFInfo
- Publication number
- CN111461551A CN111461551A CN202010252734.4A CN202010252734A CN111461551A CN 111461551 A CN111461551 A CN 111461551A CN 202010252734 A CN202010252734 A CN 202010252734A CN 111461551 A CN111461551 A CN 111461551A
- Authority
- CN
- China
- Prior art keywords
- submersible pump
- electric submersible
- value
- data
- parameter
- 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
- 238000000034 method Methods 0.000 title claims abstract description 48
- 238000013135 deep learning Methods 0.000 title claims abstract description 15
- 238000012544 monitoring process Methods 0.000 claims abstract description 63
- 238000012549 training Methods 0.000 claims abstract description 31
- 230000036541 health Effects 0.000 claims abstract description 28
- 238000012216 screening Methods 0.000 claims abstract description 15
- 230000007246 mechanism Effects 0.000 claims abstract description 11
- 230000002159 abnormal effect Effects 0.000 claims abstract description 9
- 230000015654 memory Effects 0.000 claims description 40
- 210000004027 cell Anatomy 0.000 claims description 22
- 230000008569 process Effects 0.000 claims description 14
- 230000006870 function Effects 0.000 claims description 13
- 238000011176 pooling Methods 0.000 claims description 12
- 230000004913 activation Effects 0.000 claims description 11
- 238000004140 cleaning Methods 0.000 claims description 7
- 210000002569 neuron Anatomy 0.000 claims description 7
- 238000010586 diagram Methods 0.000 claims description 6
- 238000012545 processing Methods 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 5
- 230000008859 change Effects 0.000 claims description 5
- 238000009826 distribution Methods 0.000 claims description 5
- 230000009471 action Effects 0.000 claims description 3
- 238000005065 mining Methods 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 claims description 2
- 238000003672 processing method Methods 0.000 claims description 2
- 238000003070 Statistical process control Methods 0.000 claims 10
- 238000011425 standardization method Methods 0.000 claims 1
- 238000004519 manufacturing process Methods 0.000 description 7
- 230000008901 benefit Effects 0.000 description 5
- 238000000605 extraction Methods 0.000 description 4
- 230000005856 abnormality Effects 0.000 description 3
- 238000013136 deep learning model Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 230000001960 triggered effect Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 238000007792 addition Methods 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 238000005553 drilling Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000004880 explosion Methods 0.000 description 1
- 230000007787 long-term memory Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000003129 oil well Substances 0.000 description 1
- 238000004321 preservation Methods 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000002759 z-score normalization Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Educational Administration (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (AREA)
- Tourism & Hospitality (AREA)
- Game Theory and Decision Science (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Testing And Monitoring For Control Systems (AREA)
- Control Of Non-Positive-Displacement Pumps (AREA)
Abstract
A method for early warning of electric submersible pump failure based on deep learning and SPC criteria includes obtaining data set, storing standardized model of training set, building CNN-L STM model, screening and standardizing monitoring parameters of electric submersible pump to be evaluated before applying model, applying CNN-L STM training model to estimate monitoring parameter values of electric submersible pump to be evaluated, calculating difference between current state and normal state of monitoring parameters of electric submersible pump, calculating health degree of electric submersible pump, judging whether electric submersible pump is abnormal or not according to health degree of electric submersible pump to trigger alarm mechanism.
Description
Technical Field
The invention relates to an electric submersible pump fault early warning method, in particular to an electric submersible pump fault early warning method based on deep learning and SPC (statistical computer) criteria.
Background
The electric submersible pump is oil extraction equipment of an offshore drilling platform, the performance of the electric submersible pump is guaranteed by the productivity of an oil well, the electric submersible pump needs to be maintained or replaced in time when an abnormality or a fault occurs, the operation and maintenance efficiency of an enterprise is improved, the fault rate of the enterprise is reduced, the well laying time is shortened, the production cost and the time cost are saved, the operation state of the electric submersible pump needs to be detected in real time, the abnormality is found in time, and effective early warning is carried out before the fault occurs. The current relatively mature fault monitoring method for the electric submersible pump mainly comprises two methods:
model based on electric submersible pump mechanism of operation: the method is closely combined with a control theory, a mathematical model is established through an electric submersible pump operation mechanism to predict the output of oil, gas and the like, and the output of oil, gas and the like is compared with an actual measured value to obtain a residual error; and analyzing the residual error to determine whether the process has a fault or not, and further identifying the fault type, wherein the method has the advantages that: the physical recognition and the parameter monitoring are combined, and the fault early warning is carried out by analyzing the residual error, so that the method is easy to understand; the disadvantages are as follows: most mechanism models are simplified linear systems; in the actual industrial process, the system is often a nonlinear system with high degree of freedom and multivariable coupling; the use effect is not ideal.
The method based on professional knowledge comprises the following steps: the method is based on human experience knowledge, the fault characteristics are deduced, namely after the electric submersible pump fails, an expert discovers problems by planning equipment, the connection relation, the fault propagation mode and the like among elements in the oil extraction process of the electric submersible pump are described qualitatively or quantitatively by combining the change condition of historical monitoring parameters before the failure, the qualitative or quantitative characteristics of the electric submersible pump failure are summarized, and the electric submersible pump failure early warning and monitoring are completed through the characteristics; wherein, the advantage: when the monitored object is simpler and the process knowledge and the production experience are more sufficient, the use effect is better; the disadvantages are as follows: the early warning accuracy has strong dependence on the richness of expert knowledge and the level of the expert knowledge; much experience has been difficult to describe with a reasonable formalized representation.
Disclosure of Invention
In order to overcome the defects of the technology, the invention provides an electric submersible pump fault early warning method based on deep learning and SPC (statistical processing) criteria.
In order to solve the technical problems, the invention adopts the technical scheme that: an electric submersible pump fault early warning method based on deep learning and SPC (statistical computer) criteria comprises the following steps:
the method comprises the following steps of firstly, acquiring a data set formed by all electric submersible pump operation monitoring parameters, analyzing an electric submersible pump operation mechanism and a data missing condition, determining the monitoring parameters capable of reflecting the performance of the electric submersible pump, and storing names of the parameter items;
intercepting part of data of monitoring parameters from historical operation data of the electric submersible pump as a training set; performing parameter screening, data cleaning and parameter standardization processing on the training set, and storing a standardized model of the training set; in the parameter screening process, the selection of the monitoring parameters is obtained from the data items saved in the step one; in the data cleaning process, only data in a normal operation time period is reserved, and data in a fault time period is removed; the data are subjected to standard normal distribution through standardization;
step three, constructing a CNN-L STM model, taking each monitoring parameter value of the electric submersible pump in the training set as a target of the model, taking other monitoring parameter values as input of the model, performing supervised training, and storing a training result model;
screening monitoring parameters of the electric submersible pump to be evaluated before the model is applied, and standardizing parameter values; acquiring monitoring data of the current submersible pump to be evaluated, screening monitoring parameter items of the pump, and standardizing a screening result set of the pump, wherein the two steps of treatment are based on a parameter item list and a standardized model stored in the first step and the second step;
fifthly, estimating the monitoring parameter value of the electric submersible pump to be evaluated by applying a CNN-L STM training model according to the fourth step;
and step six, calculating the difference between the current state and the normal state of the monitoring parameters of the electric submersible pump, wherein the difference between the current state and the normal state of the monitoring parameters is calculated as shown in a formula ①:
wherein y is an observed value of the monitoring parameter, y' is an estimated value of the monitoring parameter, and r is a difference of the monitoring parameter;
and step seven, calculating the health degree of the electric submersible pump according to the difference between the current state and the normal state of the electric submersible pump in the step six, wherein the health degree is calculated according to a formula ②:
wherein ,to monitor the mean value of the parameter differences, i.e.S is the health degree of the electric submersible pump;
and step eight, judging whether the operation of the electric submersible pump is abnormal or not according to the health degree of the electric submersible pump obtained in the step seven, thereby triggering an alarm mechanism.
Further, the parameter normalization processing method in the second step is a Z-score normalization method, wherein the variance of the new data set is 1, the mean value is 0, and the processed data conform to the standard normal distribution.
Further, the method for constructing the CNN-L STM model in the third step comprises the following steps:
a. setting parameters of the number of CNN layers, the number of filters and the convolution size, taking N-1 of N effective monitoring parameters of the electric submersible pump as the input of the CNN, traversing the whole input data sequence by using the convolution layer, the activation function and the pooling layer in the CNN, extracting local information of the N-1 monitoring parameters, and mining deep features;
b. and c, carrying out supervised training on the deep features acquired in the step a through an L STM network, and storing a result model.
Further, the operation of the convolutional layer in step a is shown in formula ③:
wherein ,is the jth' weight of the ith convolution kernel of the ith layer,is the jth quilt in the ith layerThe local area of convolution, V, is the width of the convolution kernel.
Further, the operation of the activation function in step a is shown in formulas ④ - ⑥:
al(i,j)=f(yl(i,j))=max{0,yl(i,j)equation ⑥
wherein ,al(i,j)To activate value, yl(i,j)Is the convolutional layer output value, where l (i, j) is the jth convolved local region of the ith convolution kernel of the ith layer.
Further, the pooling layer calculation is shown in equations ⑦ and ⑧:
wherein ,al(i,t)Activation value, p, output for the t-th neuron of the ith level i profilel(i,j)The characteristic value of the ith neuron output after pooling in the ith layer characteristic diagram is W, and W is the width of the pooling window.
Further, the calculation update state of L STM network is divided into the following steps:
i, temporary memory status information ct(ii) a In updating the memory cell ctBefore, a temporary memory cell c is generatedt(ii) a And ctThe input of the current time t and the output of the hidden layer unit of the last time t-1 are acted together and linearly combined with the respective weight matrix respectively to obtain the candidate memory unit value of the current time and update the state information of the memory unit, as shown in formula ⑨:
ct=tanh(Wxcxt+Whcht-1+bc) Equation ⑨
II, calculating an input gate value itCurrent data information is selectively stored in the memory cell through the input gate, thereby affecting the current memory cell state value, as shown in equation ⑩:
it=σ(Wxixt+Whiht-1+bi) Equation ⑩
III, calculating the value f of the forgetting gatet(ii) a The forgetting gate is mainly used for processing which information in the memory unit needs to be abandoned; such as formulaShown in the figure:
IV, calculating the state value c of the memory unit at the current momenttSuch as formulaShown in the figure:
in the formula ,representing a point-by-point product; it can be seen that the cell state is updated by the cell value c at the previous timet-1And temporarily memorize the state information ctThe information is selected and adjusted by utilizing the combined action of the forgetting gate and the input gate;
v, calculating output gate ot(ii) a The output gate mainly acts on the output of the state value of the memory unit; such as formulaShown in the figure:
wherein :Wxc、Wxi、Wxf、WxoRespectively the output layer x at the time ttAnd a hidden layer htThe connection weight between Whc、Whi、Whf、WhoRespectively the hidden layer connection weight between the time t-1 and the time t, bc、bi、bf、boOffset of input node, input gate, forget gate, output gate, ht-1And sigma is the sigmoid function output at the previous moment, and the value is (0, 1).
Further, the evaluation method for judging whether the operation of the electric submersible pump is abnormal in the step eight is to apply SPC criterion or to fit a health degree variation trend by using a polynomial.
The deep learning model based on the CNN-L STM network has the advantages that the deep learning model based on the L STM adept sequence structure analysis and the CNN adept feature extraction and transformation are considered, the model can quantify the health state of the electric submersible pump, namely health degree estimation, and judges the change trend of the health degree by combining with an SPC criterion, so that the early warning is carried out on the health state of the electric submersible pump, and the deep learning model has the following advantages:
the health degree of the method is calculated on the basis of comprehensively considering real-time residuals of a plurality of parameters, so that a simple mode of evaluating the health condition of the electric submersible pump only by abnormal parameters of a voucher is avoided;
the health degree extraction process of the method does not need any expert experience, does not need to manually set labels, greatly reduces manual participation, and saves a large amount of manpower.
Drawings
Fig. 1 is a logic flow diagram of the technical solution of the present invention.
FIG. 2 is a structural diagram of the STM model CNN-L.
FIG. 3 is a diagram of the structure of an L STM memory cell.
Detailed Description
The present invention will be described in further detail with reference to specific embodiments.
Fig. 1 shows a method for early warning of a fault of an electric submersible pump based on deep learning and SPC criteria, which comprises the following steps:
the method comprises the following steps: the method comprises the steps of collecting original data of the electric submersible pump, analyzing the operation mechanism and data missing condition of the electric submersible pump, determining monitoring parameters capable of reflecting the performance of the electric submersible pump, and storing names of the parameter items. The concrete contents are as follows: and acquiring a data set consisting of all the operation monitoring parameters of the electric submersible pump, analyzing the operation mechanism of the electric submersible pump in the data set, selecting the monitoring parameters capable of reflecting the performance of the electric submersible pump, and storing the names of the parameter items for a training set.
Step two: and intercepting part of data of the monitoring parameters from the historical operating data of the electric submersible pump to be used as a training set, carrying out operations such as data cleaning and standardization on the training set, and storing a standardized model. The concrete contents are as follows: performing operations such as parameter screening, data cleaning, parameter standardization and the like on the training set, and storing a standardized model of the training set, wherein in the parameter screening process, the selection of monitoring parameters is obtained from the data items stored in the step one; in the process of data cleaning, only data in a normal operation time period is reserved, and data in a fault time period needs to be removed; the normalization process uses the method Z-score, i.e. applying the formula X ═ (X- μ)/σ such that the new X data set variance is 1 and the mean is 0, so that the processed data will fit the standard normal distribution.
And thirdly, constructing a CNN-L STM model, respectively taking each monitoring parameter value of the electric submersible pump in the training set as a target of the model, taking other monitoring parameter values outside the training set as input of the model, performing supervised training, and storing the training result model, wherein the structure of the complete CNN-L STM model is shown in FIG. 2.
Suppose a training set shape like Dtra={Xt,yt}T, wherein yt∈R1Features, X, representing a certain monitored parameter p at time tt∈RN-1Representing the N-1 dimensional features outside the parameter p at time t, the specific modeling process is as follows:
setting parameters such as the number of CNN layers, the number of filters, the convolution size and the like, using N-1 of N effective monitoring parameters of the electric submersible pump as input of the CNN, traversing the whole input data sequence by applying formulas ③ - ⑧ of convolution layer, activation function and pooling layer in the CNN, extracting local information of the N-1 monitoring parameters, and mining deep features.
Convolution layer operation:
in the formula :is the jth' weight of the ith convolution kernel of the ith layer,is the jth convolved local region in the ith layer, and V is the width of the convolution kernel.
Activation function:
al(i,j)=f(yl(i,j))=max{0,yl(i,j)equation ⑥
in the formula :al(i,j)To activate value, yl(i,j)Is the convolutional layer output value, where l (i, j) is the jth convolved local region of the ith convolution kernel of the ith layer.
Pooling function:
in the formula :al(i,t)Activation value, p, output for the t-th neuron of the ith level i profilel(i,j)The characteristic value of the ith neuron output after pooling in the ith layer characteristic diagram is W, and W is the width of the pooling window.
Step b, inputting the deep features obtained in the step a into L STM network, and applying formula ⑨ EAnd L STM memory cell has long and short term memory advantages for time series data, and has supervised training and result model preservation.
L STM network is improved from RNN, adding cell memory unit structure in RNN hidden layer, which includes three gate controllers, input gate i, forget gate f and output gate o, able to allow the network to forget the history information, and able to use new information to update memory state, making the model have certain ability to learn long-term dependence information, effectively overcoming the problem of gradient disappearance or explosion, in the training and identification process, L STM hidden layer state value depends on the current input and previous hidden layer state value, and continuously circulating the process until the input is completed, wherein, the effect of new information on neuron is controlled by three gates, able to make L STM network able to store and transmit information for a longer time, effectively processing sequence data, L STM memory unit structure is shown in FIG. 3.
It can be seen from the figure that the three gates all use sigmoid functions and are all nonlinear summation units, and the activation functions for both inside and outside of the module are included, and the multiplication operation is used to control the activation functions of the units. The calculation updating state comprises the following steps:
i, temporary memory status information ct. In updating the memory cell ctBefore, a temporary memory cell c is generatedt. And ctThe input of the current time t and the output of the hidden layer unit of the last time t-1 act together and are linearly combined with respective weight matrixes respectively to obtain a candidate memory unit value of the current time and update the state information of the memory unit.
ct=tanh(Wxcxt+Whcht-1+bc) Equation ⑨
II, calculating an input gate value it. The current data information is selectively stored in the memory cell through the input gate, thereby affecting the current memory cell state value.
it=σ(Wxixt+Whiht-1+bi) Equation ⑩
III, calculating the value f of the forgetting gatet. The forgetting gate mainly processes which information in the memory unit needs to be discarded.
IV, calculating the state value c of the memory unit at the current momentt。
in the formula ,the dot-by-dot product is represented. It can be seen that the state of the memory cell is updated mainly by the cell value c at the previous timet-1And temporary memory status informationMessage ctAnd the information is selected and adjusted by utilizing the combined action of the forgetting gate and the input gate.
V, calculating output gate ot. The output gate primarily operates on the output of the memory cell state values.
VI, L STM unit memory output ht。
wherein :Wxc、Wxi、Wxf、WxoRespectively the output layer x at the time ttAnd a hidden layer htThe connection weight between Whc、Whi、Whf、WhoRespectively the hidden layer connection weight between the time t-1 and the time t, bc、bi、bf、boOffset of input node, input gate, forget gate, output gate, ht-1And sigma is the sigmoid function output at the previous moment, and the value is (0, 1).
Step four: before the model is applied, the monitoring parameters of the electric submersible pump to be evaluated are screened, and the parameter values are standardized. The concrete contents are as follows: acquiring monitoring data of the current submersible pump to be evaluated, screening monitoring parameter items of the pump, and standardizing a screening result set of the pump, wherein the two steps are based on the parameter item list and the standardized model stored in the step 1 and the step 2.
And fifthly, estimating the monitoring parameter values of the electric submersible pump to be estimated by applying the CNN-L STM model, wherein the concrete content is that each monitoring parameter value of the current electric submersible pump is estimated by applying the CNN-L STM training model according to the step 4.
Step six: and calculating the difference between the current state and the normal state of each monitoring parameter of the electric submersible pump. The difference between the current state and the normal state of the monitoring parameters is calculated according to the following formula:
in the formula, y is an observed value of the monitoring parameter, y' is an estimated value of the monitoring parameter, and r is a difference of the monitoring parameter.
Step seven: and calculating the health degree of the electric submersible pump according to the difference between the current state and the normal state of the electric submersible pump in the step six. The health degree calculation formula is as follows:
in the formula ,for the mean value of the difference of each monitored parameter, i.e.And S is the health degree of the electric submersible pump.
Step eight: defining an SPC rule capable of monitoring the health degree change of the electric submersible pump in real time by referring to eight different criteria of the SPC rule, wherein the specific contents are as follows:
SPC mainly refers to the application of statistical analysis technology to real-time monitoring of the production process, and scientifically distinguishes random fluctuation and abnormal fluctuation of product quality in the production process, so that early warning is provided for abnormal trends in the production process, production managers can take measures in time, abnormalities are eliminated, stability of the process is recovered, and the purpose of improving and controlling quality is achieved. The eight discriminant criteria for the SPC criteria are as follows:
rule 1: 1 point falls outside three standard deviations from the centerline;
rule 2: the continuous 9 points fall on the same side of the central line;
rule 3: successive 6 points increment or decrement;
rule 4: adjacent points in the continuous 14 points are alternately arranged up and down;
rule 5: 2 of the continuous 3 points fall outside two times of the standard deviation of the same measurement of the central line;
rule 6: 4 of the 5 consecutive points fall outside one standard deviation of the same side of the centerline;
rule 7: the continuous 15 points fall within one time of standard deviation of both sides of the central line;
rule 8: the continuous 8 points fall on two sides of the central line, and none of the continuous 8 points is within one time of the standard deviation;
the health degree sequence of the electric submersible pump calculated in the step seven can reflect the health condition of the electric submersible pump, so that an applicable SPC criterion is screened for judging whether the operation of the electric submersible pump is abnormal, and the specific selection rule is as follows:
rule 2. values for 9 consecutive points in the sequence are triggered below the centerline;
rule 3, triggering when continuous 6 points in the sequence stably descend;
rule 4. the value of 2 of the consecutive 3 points in the sequence is triggered below 2 standard deviations below the centerline;
rule 5. values of 4 of the 5 consecutive points in the sequence are triggered below 1 standard deviation below the centerline;
rule 6. trigger when the values of the continuous 14 points in the sequence appear alternately;
rule 7. values for consecutive 8 points in the sequence are on either side of the centerline and trigger when 1 standard deviation from the centerline.
And (4) applying the defined SPC criterion, monitoring the health degree sequence value of the electric submersible pump calculated in the step seven in real time, and judging whether the operation of the electric submersible pump is abnormal or not, thereby triggering an alarm mechanism.
In this step, the health condition of the electric submersible pump can be evaluated without adopting SPC (statistical computer test) criterion, and because the L STM layer considers the time characteristic of the sequence, the trend characteristic of the sequence is shown in the output result, a polynomial fitting can be adopted to fit the health degree change trend, and then early warning is carried out through a threshold value.
The above embodiments are not intended to limit the present invention, and the present invention is not limited to the above examples, and those skilled in the art may make variations, modifications, additions or substitutions within the technical scope of the present invention.
Claims (8)
1. A fault early warning method for an electric submersible pump based on deep learning and SPC (statistical computer) criteria is characterized by comprising the following steps: the method comprises the following steps:
the method comprises the following steps of firstly, acquiring a data set formed by all electric submersible pump operation monitoring parameters, analyzing an electric submersible pump operation mechanism and a data missing condition, determining the monitoring parameters capable of reflecting the performance of the electric submersible pump, and storing names of the parameter items;
intercepting part of data of monitoring parameters from historical operation data of the electric submersible pump as a training set; performing parameter screening, data cleaning and parameter standardization processing on the training set, and storing a standardized model of the training set; in the parameter screening process, the selection of the monitoring parameters is obtained from the data items saved in the step one; in the data cleaning process, only data in a normal operation time period is reserved, and data in a fault time period is removed; the data are subjected to standard normal distribution through standardization;
step three, constructing a CNN-L STM model, taking each monitoring parameter value of the electric submersible pump in the training set as a target of the model, taking other monitoring parameter values as input of the model, performing supervised training, and storing a training result model;
screening monitoring parameters of the electric submersible pump to be evaluated before the model is applied, and standardizing parameter values; acquiring monitoring data of the current submersible pump to be evaluated, screening monitoring parameter items of the pump, and standardizing a screening result set of the pump, wherein the two steps of treatment are based on a parameter item list and a standardized model stored in the first step and the second step;
fifthly, estimating the monitoring parameter value of the electric submersible pump to be evaluated by applying a CNN-L STM training model according to the fourth step;
and step six, calculating the difference between the current state and the normal state of the monitoring parameters of the electric submersible pump, wherein the difference between the current state and the normal state of the monitoring parameters is calculated as shown in a formula ①:
wherein y is an observed value of the monitoring parameter, y' is an estimated value of the monitoring parameter, and r is a difference of the monitoring parameter;
and step seven, calculating the health degree of the electric submersible pump according to the difference between the current state and the normal state of the electric submersible pump in the step six, wherein the health degree is calculated according to a formula ②:
wherein ,to monitor the mean value of the parameter differences, i.e.S is the health degree of the electric submersible pump;
and step eight, judging whether the operation of the electric submersible pump is abnormal or not according to the health degree of the electric submersible pump obtained in the step seven, thereby triggering an alarm mechanism.
2. The deep learning and SPC criteria based electric submersible pump fault pre-warning method as claimed in claim 1, wherein: and the parameter standardization processing method in the second step is a Z-score standardization method, wherein the variance of the new data set is 1, the mean value is 0, and the processed data conform to the standard normal distribution.
3. The electric submersible pump fault early warning method based on deep learning and SPC criteria as claimed in claim 1, wherein the method for constructing the CNN-L STM model in the third step is as follows:
a. setting parameters of the number of CNN layers, the number of filters and the convolution size, taking N-1 of N effective monitoring parameters of the electric submersible pump as the input of the CNN, traversing the whole input data sequence by using the convolution layer, the activation function and the pooling layer in the CNN, extracting local information of the N-1 monitoring parameters, and mining deep features;
b. and c, carrying out supervised training on the deep features acquired in the step a through an L STM network, and storing a result model.
4. The deep learning and SPC rule based electric submersible pump fault pre-warning method as claimed in claim 3, wherein the operation of the convolutional layer in the step a is shown as formula ③:
5. The deep learning and SPC criteria-based electric submersible pump fault pre-warning method as claimed in claim 4, wherein the operation of the activation function in the step a is as shown in formulas ④ - ⑥:
al(i,j)=f(yl(i,j))=max{0,yl(i,j)equation ⑥
wherein ,al(i,j)To activate value, yl(i,j)Is the convolutional layer output value, where l (i, j) is the jth convolved local region of the ith convolution kernel of the ith layer.
6. The deep learning and SPC criteria based electric submersible pump fault pre-warning method as claimed in claim 5, wherein the pooling layer is calculated as shown in formulas ⑦ and ⑧:
wherein ,al(i,t)Activation value, p, output for the t-th neuron of the ith level i profilel(i,j)The characteristic value of the ith neuron output after pooling in the ith layer characteristic diagram is W, and W is the width of the pooling window.
7. The deep learning and SPC criterion based electric submersible pump fault early warning method as claimed in claim 3, wherein the calculation updating state of the L STM network is divided into the following steps:
i, temporary memory status information ct(ii) a In updating the memory cell ctBefore, a temporary memory cell c is generatedt(ii) a And ctThe input of the current time t and the output of the hidden layer unit of the last time t-1 are acted together and linearly combined with the respective weight matrix respectively to obtain the candidate memory unit value of the current time and update the state information of the memory unit, as shown in formula ⑨:
ct=tanh(Wxcxt+Whcht-1+bc) Equation ⑨
II, calculating an input gate value itCurrent data information is selectively stored in the memory cell through the input gate, thereby affecting the current memory cell state value, as shown in equation ⑩:
it=σ(Wxixt+Whiht-1+bi) Equation ⑩
III, calculating the value f of the forgetting gatet(ii) a Forget door masterWhich information in the memory unit needs to be discarded for processing; such as formulaShown in the figure:
IV, calculating the state value c of the memory unit at the current momenttSuch as formulaShown in the figure:
in the formula ,representing a point-by-point product; it can be seen that the cell state is updated by the cell value c at the previous timet-1And temporarily memorize the state information ctThe information is selected and adjusted by utilizing the combined action of the forgetting gate and the input gate;
v, calculating output gate ot(ii) a The output gate mainly acts on the output of the state value of the memory unit; such as formulaShown in the figure:
wherein :Wxc、Wxi、Wxf、WxoRespectively the output layer x at the time ttAnd a hidden layer htThe connection weight between Whc、Whi、Whf、WhoRespectively the hidden layer connection weight between the time t-1 and the time t, bc、bi、bf、boOffset of input node, input gate, forget gate, output gate, ht-1And sigma is the sigmoid function output at the previous moment, and the value is (0, 1).
8. The deep learning and SPC criteria based electric submersible pump fault pre-warning method as claimed in claim 1, wherein: and step eight, adopting an SPC (statistical process control) rule or fitting a polynomial to fit the health degree change trend.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010252734.4A CN111461551B (en) | 2020-04-01 | 2020-04-01 | Deep learning and SPC criterion-based electric submersible pump fault early warning method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010252734.4A CN111461551B (en) | 2020-04-01 | 2020-04-01 | Deep learning and SPC criterion-based electric submersible pump fault early warning method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111461551A true CN111461551A (en) | 2020-07-28 |
CN111461551B CN111461551B (en) | 2023-05-02 |
Family
ID=71679338
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010252734.4A Active CN111461551B (en) | 2020-04-01 | 2020-04-01 | Deep learning and SPC criterion-based electric submersible pump fault early warning method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111461551B (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112132394A (en) * | 2020-08-21 | 2020-12-25 | 西安交通大学 | Power plant circulating water pump prediction state assessment method and system |
CN112506687A (en) * | 2020-11-24 | 2021-03-16 | 四川长虹电器股份有限公司 | Fault diagnosis method based on multi-period segmented sliding window standard deviation |
CN112785091A (en) * | 2021-03-04 | 2021-05-11 | 湖北工业大学 | Method for performing fault prediction and health management on oil field electric submersible pump |
CN113033663A (en) * | 2021-03-26 | 2021-06-25 | 同济大学 | Automatic container terminal equipment health prediction method based on machine learning |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2015158198A1 (en) * | 2014-04-17 | 2015-10-22 | 北京泰乐德信息技术有限公司 | Fault recognition method and system based on neural network self-learning |
CN110689075A (en) * | 2019-09-26 | 2020-01-14 | 北京工业大学 | Fault prediction method of self-adaptive threshold of refrigeration equipment based on multi-algorithm fusion |
CN110763929A (en) * | 2019-08-08 | 2020-02-07 | 浙江大学 | Intelligent monitoring and early warning system and method for convertor station equipment |
-
2020
- 2020-04-01 CN CN202010252734.4A patent/CN111461551B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2015158198A1 (en) * | 2014-04-17 | 2015-10-22 | 北京泰乐德信息技术有限公司 | Fault recognition method and system based on neural network self-learning |
CN110763929A (en) * | 2019-08-08 | 2020-02-07 | 浙江大学 | Intelligent monitoring and early warning system and method for convertor station equipment |
CN110689075A (en) * | 2019-09-26 | 2020-01-14 | 北京工业大学 | Fault prediction method of self-adaptive threshold of refrigeration equipment based on multi-algorithm fusion |
Non-Patent Citations (1)
Title |
---|
代杰杰等: "采用LSTM 网络的电力变压器运行状态预测方法研究" * |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112132394A (en) * | 2020-08-21 | 2020-12-25 | 西安交通大学 | Power plant circulating water pump prediction state assessment method and system |
CN112132394B (en) * | 2020-08-21 | 2024-03-29 | 西安交通大学 | Power plant circulating water pump predictive state evaluation method and system |
CN112506687A (en) * | 2020-11-24 | 2021-03-16 | 四川长虹电器股份有限公司 | Fault diagnosis method based on multi-period segmented sliding window standard deviation |
CN112506687B (en) * | 2020-11-24 | 2022-03-01 | 四川长虹电器股份有限公司 | Fault diagnosis method based on multi-period segmented sliding window standard deviation |
CN112785091A (en) * | 2021-03-04 | 2021-05-11 | 湖北工业大学 | Method for performing fault prediction and health management on oil field electric submersible pump |
CN113033663A (en) * | 2021-03-26 | 2021-06-25 | 同济大学 | Automatic container terminal equipment health prediction method based on machine learning |
Also Published As
Publication number | Publication date |
---|---|
CN111461551B (en) | 2023-05-02 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111461551A (en) | Electric submersible pump fault early warning method based on deep learning and SPC (statistical computer) criterion | |
CN109033450B (en) | Elevator equipment fault prediction method based on deep learning | |
CN110738360B (en) | Method and system for predicting residual life of equipment | |
CN108197845B (en) | Transaction index abnormity monitoring method based on deep learning model LSTM | |
CN112926273B (en) | Method for predicting residual life of multivariate degradation equipment | |
CN111813084B (en) | Mechanical equipment fault diagnosis method based on deep learning | |
CN110757510B (en) | Method and system for predicting remaining life of robot | |
CN110555230B (en) | Rotary machine residual life prediction method based on integrated GMDH framework | |
CN113642754B (en) | Complex industrial process fault prediction method based on RF noise reduction self-coding information reconstruction and time convolution network | |
CN111030889B (en) | Network traffic prediction method based on GRU model | |
CN109471698B (en) | System and method for detecting abnormal behavior of virtual machine in cloud environment | |
CN110705812A (en) | Industrial fault analysis system based on fuzzy neural network | |
CN113762329A (en) | Method and system for constructing state prediction model of large rolling mill | |
Garg et al. | A two-phase approach for reliability and maintainability analysis of an industrial system | |
CN112364560B (en) | Intelligent prediction method for working hours of mine rock drilling equipment | |
CN114282443B (en) | Residual service life prediction method based on MLP-LSTM supervised joint model | |
CN114118673A (en) | Workshop intelligent fault diagnosis early warning method based on digital twin technology | |
CN113779882A (en) | Method, device, equipment and storage medium for predicting residual service life of equipment | |
CN114841076A (en) | Power battery production process fluctuation abnormity detection method based on space-time diagram model | |
CN117290726A (en) | CAE-BiLSTM-based fault early warning method for mobile equipment | |
KR102474332B1 (en) | Prediction method of machine health stability in smart factory system, and recording medium thereof | |
CN113536671A (en) | Lithium battery life prediction method based on LSTM | |
CN112861422A (en) | Deep-learning coal bed gas screw pump well health index prediction method and system | |
CN114486262B (en) | CNN-AT-LSTM-based bearing residual service life prediction method | |
CN116881640A (en) | Method and system for predicting core extraction degree and computer-readable storage medium |
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 |