CN112883639A - GIS equipment service life prediction device and method based on machine learning - Google Patents
GIS equipment service life prediction device and method based on machine learning Download PDFInfo
- Publication number
- CN112883639A CN112883639A CN202110151383.2A CN202110151383A CN112883639A CN 112883639 A CN112883639 A CN 112883639A CN 202110151383 A CN202110151383 A CN 202110151383A CN 112883639 A CN112883639 A CN 112883639A
- Authority
- CN
- China
- Prior art keywords
- gis equipment
- data
- service life
- gis
- life prediction
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 47
- 238000010801 machine learning Methods 0.000 title claims abstract description 24
- 238000011161 development Methods 0.000 claims abstract description 10
- 238000012360 testing method Methods 0.000 claims description 44
- 238000012545 processing Methods 0.000 claims description 41
- 238000004458 analytical method Methods 0.000 claims description 28
- 238000004422 calculation algorithm Methods 0.000 claims description 23
- 230000036541 health Effects 0.000 claims description 20
- 238000012423 maintenance Methods 0.000 claims description 17
- 238000012706 support-vector machine Methods 0.000 claims description 16
- 238000004364 calculation method Methods 0.000 claims description 15
- 239000011159 matrix material Substances 0.000 claims description 15
- 238000012549 training Methods 0.000 claims description 15
- 230000008569 process Effects 0.000 claims description 12
- 230000009467 reduction Effects 0.000 claims description 12
- 238000000926 separation method Methods 0.000 claims description 9
- 238000000513 principal component analysis Methods 0.000 claims description 8
- 238000013499 data model Methods 0.000 claims description 6
- 238000012216 screening Methods 0.000 claims description 6
- 238000007781 pre-processing Methods 0.000 claims description 5
- 230000008030 elimination Effects 0.000 claims description 4
- 238000003379 elimination reaction Methods 0.000 claims description 4
- 230000009466 transformation Effects 0.000 claims description 4
- 238000002955 isolation Methods 0.000 claims description 2
- 238000006467 substitution reaction Methods 0.000 claims description 2
- 230000005484 gravity Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 238000013500 data storage Methods 0.000 description 2
- 229910018503 SF6 Inorganic materials 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000009472 formulation Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012821 model calculation Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- SFZCNBIFKDRMGX-UHFFFAOYSA-N sulfur hexafluoride Chemical compound FS(F)(F)(F)(F)F SFZCNBIFKDRMGX-UHFFFAOYSA-N 0.000 description 1
- 229960000909 sulfur hexafluoride Drugs 0.000 description 1
- 238000010998 test method Methods 0.000 description 1
- 239000013598 vector Substances 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
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
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/04—Ageing analysis or optimisation against ageing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a GIS equipment service life prediction device and a method based on machine learning. The service life prediction method specifically comprises the following steps: setting a data acquisition cycle, acquiring index data of the GIS equipment according to the data acquisition cycle, establishing a GIS equipment service life prediction model, then calculating to obtain the fault probability of the GIS equipment in each data acquisition cycle, thereby constructing a fault probability development trend graph, setting a fault probability threshold, and finally obtaining the time of the GIS equipment reaching the fault threshold through the fault probability development trend graph prediction to obtain the residual service life of the GIS equipment. The GIS equipment service life prediction device and method provided by the invention can be used for comprehensively evaluating and predicting GIS equipment, and the prediction result is accurate and reliable.
Description
Technical Field
The invention relates to the technical field of operation and maintenance of power equipment, in particular to a GIS equipment service life prediction device and method based on machine learning.
Background
Sulfur hexafluoride closed type combined electrical apparatus (GIS) is widely applied and operated in various fields of power transmission and distribution of power grid by virtue of excellent performance, strong starting capability and safe reliability of the closed combined electrical apparatus. With the rapid development of market economy and the continuous increase of electrical loads, the conventional GIS equipment maintenance cannot meet the increasing requirements. At present, the running condition of the GIS equipment is mostly detected by methods such as factory acceptance, first inspection, calibration and the like, but whether the actual running state of the GIS equipment is healthy or not cannot be determined, whether safety or equipment hidden dangers exist in a verification period or not cannot be determined, and the safety of a power system and the running reliability of a power grid are greatly influenced. Whether the actual running state of the GIS equipment is healthy or not and the safety or hidden equipment danger in the detection period are shown in the service life of the GIS equipment. Therefore, a method for predicting the service life of the GIS equipment to formulate a reasonable maintenance scheme to improve the safety of the power system and the reliability of the operation of the power grid is provided. However, in the existing GIS equipment life prediction technical means, the labor cost management cost is high, the intelligentization and automation level is low, and the GIS equipment life prediction method cannot adapt to the development of a new generation of intelligent transformer substations. The service life prediction method IS complex and various, data collection, arrangement and comprehensive analysis of the GIS equipment are not complete, the service life prediction objects of the GIS equipment are in a one-sided mode, only local devices of the GIS equipment are evaluated and predicted, the test method IS independent and limited, comprehensive evaluation and prediction cannot be performed on the GIS equipment, the obtained prediction result IS not accurate enough, effective and reliable service life prediction cannot be performed on the GIS equipment, the overhaul efficiency of the GIS equipment IS influenced, and the operation reliability and economy of a power grid are also influenced.
Disclosure of Invention
The invention aims to overcome the defects in the prior art and provides a GIS equipment service life prediction device and method based on machine learning.
The purpose of the invention is realized by the following technical scheme:
a GIS equipment service life prediction method based on machine learning comprises the following steps:
setting a data acquisition period, acquiring GIS equipment index data in real time by a data acquisition module according to the data acquisition period, and storing the GIS equipment index data into a database;
secondly, a model building module builds a GIS equipment service life prediction model according to historical GIS equipment index data in a database;
thirdly, the analysis processing module calls all GIS equipment index data in each data acquisition period in the database, inputs the GIS equipment index data into a GIS equipment service life prediction model for analysis and calculation to obtain the fault probability of the GIS equipment in each data acquisition period, and constructs a fault probability development trend graph according to the time of the data acquisition period and the corresponding fault probability;
and step four, setting a fault probability threshold, and predicting the time of the GIS equipment reaching the fault threshold by the analysis processing module according to the fault probability development trend graph so as to obtain the residual service life of the GIS equipment.
Through carrying out data acquisition to GIS equipment, and be different from common life prediction method, the index data of gathering a plurality of devices, in order to guarantee the real-time nature of the data of gathering, the time interval of fixed data acquisition has been set up, can carry out real-time life prediction to GIS equipment, in the follow-up maintenance, can overhaul the scheme and formulate according to the life prediction data of real-time update, it is that the maintenance scheme is more laminated reality, improve life prediction's accuracy, prevent because the unreasonable cost loss that causes of maintenance scheme.
Further, after a GIS equipment service life prediction model is constructed, the GIS equipment service life prediction model is trained through a machine learning algorithm; the machine learning algorithm is specifically a Support Vector Machine (SVM) algorithm, the SVM algorithm obtains a test data set through training data processing, a GIS equipment service life prediction model is trained through the test data set to obtain a separation hyperplane required for judging the health condition of the GIS equipment, and the separation hyperplane is used for distinguishing the health condition of the GIS equipment.
The GIS equipment life prediction model is trained through a Support Vector Machine (SVM) algorithm, so that when the GIS equipment life prediction model is subsequently calculated, the obtained result is closer to the fact, the health condition of the GIS equipment is judged through the obtained separation hyperplane, the health condition of the GIS equipment is determined through the relative position of index data and the separation hyperplane, the health condition of the GIS equipment is directly defined as 1 or-1 data according to the relative position block, and the subsequent data classification and fault probability calculation work is simplified.
Further, the specific process of the training data processing is as follows: selecting a plurality of test parameters in a data acquisition period, wherein the test parameters are GIS equipment index data acquired by a data acquisition module, forming a sample matrix x according to the selected test parameters, and judging the state y of the tested GIS equipment in the data acquisition period, wherein if the state of the tested GIS equipment is healthy, y is 1; if the state of the GIS equipment to be tested is unhealthy, y is-1; and then forming a test data set D according to the sample matrix x and the state Y of the GIS equipment to be tested.
The GIS equipment service life prediction model is subjected to data training test through the test data set D, index data and health conditions of the GIS equipment are used as training test data, when fault probability calculation of the GIS equipment is carried out through the index data acquired in real time, calculation results are more accurate, and the calculation results of the residual service life of the GIS equipment acquired subsequently are more reliable.
Further, the expression of the test data set D is specifically:
D＝{(xi，yi)|xi∈Rp，yi∈{-1，1}}，i＝1，...，n；
wherein: d is the test data set, xiSample matrix, y, for selected i-th test parameteriFor the ith group of GIS equipment to be tested in the state of RpFor a test parameter set comprising p test parameters, n is the number of sample matrices.
Furthermore, the analysis processing module in the fourth step predicts the time of the GIS equipment reaching the fault threshold value through a least square method.
Because a dynamic process is used for calculating the residual service life of the GIS equipment, the time is taken as an independent variable, the relative position of GIS equipment index data and a separation hyperplane is taken as a dependent variable, and the functional relation between the time and the relative position can be found by a least square method so as to obtain the fault occurrence time.
Further, when the time that the GIS equipment reaches the fault threshold value is predicted by the least square method, data rotation substitution is adopted to store the predicted data, and the method specifically comprises the following steps:
1.1, setting the capacity of the prediction data;
1.2, after the prediction of the analysis processing module is started, sequentially storing N data { H in time sequence1，H2，…，HNIn which H1For the oldest prediction data, HNIs the latest prediction data;
1.3, after the storage space is full, setting the replaced data as ith data and initializing i;
1.4, after acquiring new prediction data, replacing H with new prediction dataiIs marked as HNAnd the original H is combinedi+1To HNData relabeling as H betweeniTo Hi-1；
1.5, modulo i by N, add i by one, and go back to step 1.4 after the new data acquisition.
In the process of predicting the service life of the GIS equipment, the relation between the performance of the GIS equipment and the time does not have a stable change, but is staggered and fluctuated, so that the final predicted service life has the characteristic of irregular change, and the calculation result probably displays the false image of advanced fault, thereby causing difficulty to the formulation of a maintenance scheme and increasing the maintenance expense. When fitting a curve using the least squares method, the influence of the stepwise nature of the data on the trend of the fitted curve is enormous. In order to reduce the influence caused by the stage characteristics of the data, part of older historical data is used in curve fitting, so that the fitting calculation result is more real and reliable. The invention adopts a rotation replacement mode to replace old data with new data, leaves a little historical old data not to be replaced in the new data replacement process, and leads the older historical data to participate in the calculation process when fitting the curve.
Furthermore, in the second step, before constructing a GIS equipment life prediction model according to the historical GIS equipment index data, the historical GIS equipment index data needs to be preprocessed, and the preprocessing comprises irrelevant item interference elimination and noise reduction.
After the index data of the historical GIS equipment is preprocessed, the accuracy of a GIS equipment service life prediction model can be further improved.
Further, the elimination of the interference of the irrelevant item specifically comprises the steps of screening out a data dimension with the highest relevance through a PCA dimension reduction algorithm, and carrying out dimension reduction processing on historical GIS equipment index data according to the data dimension; the noise reduction processing specifically comprises the steps of carrying out orthogonal linear transformation on historical GIS equipment index data through principal component analysis, sorting coordinates of the data according to the variance of the data to obtain each principal component, carrying out data screening on the historical GIS equipment index data according to the obtained principal component, discarding the historical GIS equipment index data which do not belong to the principal component, and finally carrying out fuzzy processing on the historical GIS equipment index data left after the principal component analysis by means of a KNN algorithm.
The dimensionality of the historical GIS equipment index data can be effectively reduced through the PCA dimension reduction algorithm, so that the historical GIS equipment index data are simplified. And the characteristic value with smaller specific gravity can be effectively removed through principal component analysis, a data part with large specific gravity is left, and historical GIS equipment index data is further optimized. And when training is carried out subsequently through a Support Vector Machine (SVM) algorithm, the influence of irrelevant data is reduced, and the calculation accuracy of a GIS equipment service life prediction model is improved subsequently.
Further, the GIS equipment service life prediction device based on machine learning comprises a data acquisition module, a model establishing module, an analysis processing module and a database, wherein the data acquisition module, the model establishing module and the analysis processing module are all connected with the database, and the data acquisition module is used for acquiring GIS equipment index data and storing the GIS equipment index data in the database; the model establishing module is used for establishing a data model required for predicting the service life of the GIS equipment according to the index data of the GIS equipment; the analysis processing module is also connected with the model establishing module and is used for carrying out GIS equipment service life prediction calculation according to the data model.
Furthermore, the GIS equipment index data comprises state parameters and maintenance parameters of the isolation grounding switch, state parameters and maintenance parameters of the breaker switch and state parameters and maintenance parameters of the switch chamber arc extinguish chamber corrugated pipe.
The invention has the beneficial effects that:
the service life prediction model is built by collecting index data of the GIS equipment, the collected index data of the GIS equipment comprises various parameter data, the parameter data of various devices influencing the service life of the GIS equipment is collected not only aiming at the parameter of one device of the GIS equipment, and the comprehensiveness of the service life prediction model of the GIS equipment is ensured. And not only the state parameter data of the GIS device is collected, but also the overhaul parameter data is collected as GIS device index data, the comprehensiveness of the GIS device index data is ensured, and the accuracy of service life prediction is improved. A GIS equipment service life prediction model established through GIS equipment index data is trained by utilizing a Support Vector Machine (SVM) algorithm, and the accuracy of a subsequent model calculation result is further improved.
Drawings
FIG. 1 is a schematic flow diagram of the present invention;
FIG. 2 is a schematic diagram of an alternative data storage process for data rotation according to the present invention;
FIG. 3 is a schematic diagram of an embodiment of the present invention;
wherein: 1. the system comprises a data acquisition module 2, a model establishing module 3, an analysis processing module 4 and a database.
Detailed Description
The invention is further described below with reference to the figures and examples.
Example (b):
a method for predicting the service life of a GIS device based on machine learning, as shown in fig. 1, includes the following steps:
step one, setting a data acquisition period, acquiring GIS equipment index data in real time by a data acquisition module 1 according to the data acquisition period, and storing the GIS equipment index data into a database 4;
the index data specifically comprise state parameters and maintenance parameters of a grounding switch of the GIS equipment, state parameters and maintenance parameters of a breaker switch and state parameters and maintenance parameters of a switch chamber arc extinguish chamber corrugated pipe. Before GIS equipment index parameters are collected, the possible changes of state parameters and overhaul parameters of each device of the GIS equipment during performance degradation need to be considered, wherein the parameters with obvious changes are preferably research parameters and can be used as historical GIS equipment index data for model building.
Step two, the model building module 2 builds a GIS equipment service life prediction model according to historical GIS equipment index data in the database 4;
after a GIS equipment life prediction model is constructed, training the GIS equipment life prediction model through a machine learning algorithm; the machine learning algorithm is specifically a Support Vector Machine (SVM) algorithm, the SVM algorithm obtains a test data set through training data processing, a GIS equipment service life prediction model is trained through the test data set to obtain a separation hyperplane required for judging the health condition of the GIS equipment, and the separation hyperplane is used for distinguishing the health condition of the GIS equipment. In the training process, data overlapping of training data is a common phenomenon, and the SVM algorithm can still obtain a linear classification hyperplane for distinguishing two types under the condition of training data overlapping.
The specific process of the training data processing is as follows: selecting a plurality of test parameters in a data acquisition period, wherein the test parameters are GIS equipment index data acquired by a data acquisition module 1, forming a sample matrix x according to the selected test parameters, and judging the state y of the tested GIS equipment in the data acquisition period, wherein if the state of the tested GIS equipment is healthy, y is 1; if the state of the GIS equipment to be tested is unhealthy, y is-1; and then forming a test data set D according to the sample matrix x and the state Y of the GIS equipment to be tested.
The expression of the test data set D is specifically:
D＝{(xi，yi)|xi∈Rp，yi∈{-1，1}}，i＝1，...，n；
wherein: d is the test data set, xiSample matrix, y, for selected i-th test parameteriFor the ith group of GIS equipment to be tested in the state of RpFor a test parameter set comprising p test parameters, n is the number of sample matrices.
The test parameters are linked with the health state of the GIS equipment, after model training is carried out by using the test data set, the relation between the index data and the health state of the GIS equipment can be determined, and the calculation result of the GIS equipment service life prediction model is more reliable when the fault probability is judged by the GIS equipment index data acquired in real time.
Thirdly, the analysis processing module 3 calls all GIS equipment index data in each data acquisition period in the database 4, inputs the index data into a GIS equipment service life prediction model for analysis and calculation to obtain the fault probability of the GIS equipment in each data acquisition period, and constructs a fault probability development trend graph according to the time of the data acquisition period and the corresponding fault probability;
and step four, setting a fault probability threshold, and predicting the time of the GIS equipment reaching the fault threshold by the analysis processing module 3 according to the fault probability development trend graph so as to obtain the residual service life of the GIS equipment.
In the fourth step, the analysis processing module 3 predicts the time of the GIS equipment reaching the fault threshold value through a least square method.
When the time that the GIS equipment reaches the fault threshold is predicted by the least square method, data rotation is adopted to replace the time to store the predicted data, as shown in fig. 2, the method specifically comprises the following steps:
1.1, setting the capacity of the prediction data;
1.2, after the prediction of the analysis processing module 3 is started, sequentially storing N data { H }according to the time sequence1，H2,…,HNIn which H1For the oldest prediction data, HNIs the latest prediction data;
1.3, after the storage space is full, setting the replaced data as ith data and initializing i;
1.4, after acquiring new prediction data, replacing H with new prediction dataiIs marked as HNAnd the original H is combinedi+1To HNData relabeling as H betweeniTo Hi-1；
1.5, modulo i by N, add i by one, and go back to step 1.4 after the new data acquisition.
As the GIS equipment service life prediction result is closely related to the stage characteristics of the GIS equipment health state data, the GIS equipment health state data storage method is improved, and the stored GIS equipment health state data is ensured to include GIS equipment health state data in different stages. The method simultaneously saves the dense recent GIS equipment health state data and the sparse GIS equipment health state data with longer storage time, and obtains the health value with longer time span under the condition of not increasing the GIS equipment health state data capacity and not reducing the test frequency, so that the service life prediction result of the GIS equipment is more stable and reliable.
And secondly, preprocessing historical GIS equipment index data before constructing a GIS equipment service life prediction model according to the historical GIS equipment index data, wherein the preprocessing comprises removing irrelevant item interference and denoising.
Specifically, the elimination of the interference of the irrelevant item is to screen out the data dimension with the highest relevance through a PCA dimension reduction algorithm, and to perform dimension reduction processing on historical GIS equipment index data according to the data dimension; the noise reduction processing specifically comprises the steps of carrying out orthogonal linear transformation on historical GIS equipment index data through principal component analysis, sorting coordinates of the data according to the variance of the data to obtain each principal component, carrying out data screening on the historical GIS equipment index data according to the obtained principal component, discarding the historical GIS equipment index data which do not belong to the principal component, and finally carrying out fuzzy processing on the historical GIS equipment index data left after the principal component analysis by means of a KNN algorithm.
The main processes of principal component analysis performed on the index data are as follows:
x represents the characteristic data of GIS equipment index data, Y represents the GIS equipment state label, and data set D1The expression is as follows:
D1＝{(Xi，Yi)|Xi∈Rp，Y＝{1，2，...，m}}，i＝1，...，n；
firstly, a data set D is calculated and extracted1The characteristic data X of all GIS equipment index data form a data set D2The expression is as follows:
D2＝{Xi|Xi∈Rp}，i＝1，...，n；
and further calculating the average value, wherein the expression is as follows:
then, carrying out feature centralization processing on the X, translating each test parameter in the X to the mean value of 0, and obtaining a data set D3The expression is as follows:
D3＝{zi|zi∈Rp}，i＝1，...，n；
then, a covariance matrix C is calculated, which is expressed as:
and C, the characteristic value generated in the characteristic decomposition process represents the information amount of a certain dimension after transformation. And discarding some characteristic values with smaller specific gravity generated in the process, leaving a part with large specific gravity, and selecting characteristic vectors corresponding to the characteristic values left after screening to form a matrix M. And calculating a sample matrix formed by the test parameters in the training data set according to the matrix M, wherein the expression is as follows:
xi＝zi*M，i＝1，...，n；
wherein: rpFor a test parameter set comprising p test parameters, n is the number of sample matrices, XiCharacteristic data of the ith GIS equipment index data, YiThe GIS equipment state label corresponding to the characteristic data of the ith GIS equipment index data, m is the number of the GIS equipment state labels,
is the mean value, z, of the characteristic data X of the GIS equipment index dataiThe data obtained by translating the characteristic data of the ith GIS equipment index data in the X is C, and C is a data set D3M is a matrix formed by eigenvectors corresponding to the eigenvalues left after the screening, xiAnd forming a sample matrix for the selected ith test parameter.
A GIS equipment service life prediction device based on machine learning is shown in figure 3 and comprises a data acquisition module 1, a model building module 2, an analysis processing module 3 and a database 4, wherein the data acquisition module 1, the model building module 2 and the analysis processing module 3 are all connected with the database 4, and the data acquisition module 1 is used for acquiring GIS equipment index data and storing the GIS equipment index data into the database 4; the model establishing module 2 is used for establishing a data model required for predicting the service life of the GIS equipment according to the index data of the GIS equipment; the analysis processing module 3 is also connected with the model establishing module 2, and the analysis processing module 3 is used for carrying out GIS equipment service life prediction calculation according to the data model.
The above-described embodiments are only preferred embodiments of the present invention, and are not intended to limit the present invention in any way, and other variations and modifications may be made without departing from the spirit of the invention as set forth in the claims.
Claims (10)
1. A GIS equipment service life prediction method based on machine learning is characterized by comprising the following steps:
step one, setting a data acquisition period, acquiring GIS equipment index data in real time by a data acquisition module (1) according to the data acquisition period, and storing the GIS equipment index data into a database (4);
secondly, a model building module (2) builds a GIS equipment service life prediction model according to historical GIS equipment index data in the database (4);
thirdly, the analysis processing module (3) calls all GIS equipment index data in each data acquisition period in the database (4), inputs the GIS equipment index data into a GIS equipment service life prediction model for analysis and calculation to obtain the fault probability of the GIS equipment in each data acquisition period, and constructs a fault probability development trend graph according to the time of the data acquisition period and the corresponding fault probability;
and step four, setting a fault probability threshold, and predicting the time of the GIS equipment reaching the fault threshold by the analysis processing module (3) according to the fault probability development trend graph so as to obtain the residual service life of the GIS equipment.
2. The GIS equipment life prediction method based on machine learning of claim 1 is characterized in that after a GIS equipment life prediction model is constructed, the GIS equipment life prediction model is trained through a machine learning algorithm; the machine learning algorithm is specifically a Support Vector Machine (SVM) algorithm, the SVM algorithm obtains a test data set through training data processing, a GIS equipment service life prediction model is trained through the test data set to obtain a separation hyperplane required for judging the health condition of the GIS equipment, and the separation hyperplane is used for distinguishing the health condition of the GIS equipment.
3. The method for predicting the service life of the GIS device based on the machine learning as claimed in claim 2, wherein the specific process of the training data processing is as follows: selecting a plurality of test parameters in a data acquisition period, wherein the test parameters are GIS equipment index data acquired by a data acquisition module (1), forming a sample matrix x according to the selected test parameters, and judging the state y of the tested GIS equipment in the data acquisition period, wherein if the state of the tested GIS equipment is healthy, y is 1; if the state of the GIS equipment to be tested is unhealthy, y is-1; and then forming a test data set D according to the sample matrix x and the state Y of the GIS equipment to be tested.
4. The method for predicting the service life of the GIS device based on machine learning according to claim 3, wherein the expression of the test data set D is specifically as follows:
D＝{(xi，yi)|xi∈Rp，yi∈{-1，1}}，i＝1，...，n；
wherein: d is the test data set, xiSample matrix, y, for selected i-th test parameteriFor the ith group of GIS equipment to be tested in the state of RpFor a test parameter set comprising p test parameters, n is the number of sample matrices.
5. The GIS device service life prediction method based on machine learning of claim 1 is characterized in that the analysis processing module (3) in the fourth step predicts the time of the GIS device reaching the fault threshold value through a least square method.
6. The method for predicting the service life of the GIS equipment based on the machine learning as claimed in claim 5, wherein when the time that the GIS equipment reaches the fault threshold is predicted by the least square method, the prediction data is saved by adopting data rotation substitution, and the method specifically comprises the following steps:
1.1, setting the capacity of the prediction data;
1.2, after the prediction of the analysis processing module (3) is started, sequentially storing N data { H }according to the time sequence1,H2,…,HNIn which H1For the oldest prediction data, HNIs the latest prediction data;
1.3, after the storage space is full, setting the replaced data as ith data and initializing i;
1.4, after acquiring new prediction data, replacing H with new prediction dataiIs marked as HNAnd the original H is combinedi+1To HNData relabeling as H betweeniTo Hi-1；
1.5, modulo i by N, add i by one, and go back to step 1.4 after the new data acquisition.
7. The GIS equipment life prediction method based on machine learning of claim 1, wherein in step two, before constructing a GIS equipment life prediction model according to historical GIS equipment index data, preprocessing is further required to be performed on the historical GIS equipment index data, and the preprocessing includes removing irrelevant item interference and noise reduction processing.
8. The GIS equipment life prediction method based on machine learning as claimed in claim 7, wherein the elimination of the interference of the irrelevant item is specifically to screen out the data dimension with the highest relevance through a PCA dimension reduction algorithm, and to perform dimension reduction processing on historical GIS equipment index data according to the data dimension; the noise reduction processing specifically comprises the steps of carrying out orthogonal linear transformation on historical GIS equipment index data through principal component analysis, sorting coordinates of the data according to the variance of the data to obtain each principal component, carrying out data screening on the historical GIS equipment index data according to the obtained principal component, discarding the historical GIS equipment index data which do not belong to the principal component, and finally carrying out fuzzy processing on the historical GIS equipment index data left after the principal component analysis by means of a KNN algorithm.
9. The GIS equipment service life prediction device based on machine learning is characterized by comprising a data acquisition module (1), a model building module (2), an analysis processing module (3) and a database (4), wherein the data acquisition module (1), the model building module (2) and the analysis processing module (3) are all connected with the database (4), and the data acquisition module (1) is used for acquiring GIS equipment index data and storing the GIS equipment index data into the database (4); the model establishing module (2) is used for establishing a data model required for predicting the service life of the GIS equipment according to the index data of the GIS equipment; the analysis processing module (3) is also connected with the model establishing module (2), and the analysis processing module (3) is used for carrying out GIS equipment service life prediction calculation according to the data model.
10. The device of claim 9, wherein the GIS equipment index data includes status and maintenance parameters of the isolation grounding switch, status and maintenance parameters of the breaker switch, and status and maintenance parameters of the interrupter bellows of the switchgear room.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110151383.2A CN112883639A (en) | 2021-02-03 | 2021-02-03 | GIS equipment service life prediction device and method based on machine learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110151383.2A CN112883639A (en) | 2021-02-03 | 2021-02-03 | GIS equipment service life prediction device and method based on machine learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112883639A true CN112883639A (en) | 2021-06-01 |
Family
ID=76057089
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110151383.2A Pending CN112883639A (en) | 2021-02-03 | 2021-02-03 | GIS equipment service life prediction device and method based on machine learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112883639A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116780782A (en) * | 2023-08-24 | 2023-09-19 | 成都易电云商工程服务有限公司 | Comprehensive maintenance system for electrical equipment of power distribution network |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106528975A (en) * | 2016-11-01 | 2017-03-22 | 电子科技大学 | Fault prognostics and health management method applied to circuits and systems |
| US20190005400A1 (en) * | 2016-09-20 | 2019-01-03 | Southwest Petroleum University | A fuzzy evaluation and prediction method for running status of mechanical equipment with occurrence probability of failure modes |
| CN109308522A (en) * | 2018-09-03 | 2019-02-05 | 河海大学常州校区 | A kind of GIS failure prediction method based on Recognition with Recurrent Neural Network |
| CN110348615A (en) * | 2019-06-27 | 2019-10-18 | 西安工程大学 | Cable line fault probability forecasting method based on ant group optimization support vector machines |
| CN110995384A (en) * | 2019-11-27 | 2020-04-10 | 杭州联汇科技股份有限公司 | Broadcast master control fault trend prejudging method based on machine learning |
| CN111274737A (en) * | 2020-02-25 | 2020-06-12 | 山东大学 | Method and system for predicting remaining service life of mechanical equipment |
| CN112069930A (en) * | 2020-08-20 | 2020-12-11 | 国网山西省电力公司电力科学研究院 | Vibration signal processing method and device for improving GIS equipment fault diagnosis accuracy |
| CN112257333A (en) * | 2020-09-24 | 2021-01-22 | 浙江工业大学 | Mechanical equipment internal component service life prediction method based on deep learning |
-
2021
- 2021-02-03 CN CN202110151383.2A patent/CN112883639A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20190005400A1 (en) * | 2016-09-20 | 2019-01-03 | Southwest Petroleum University | A fuzzy evaluation and prediction method for running status of mechanical equipment with occurrence probability of failure modes |
| CN106528975A (en) * | 2016-11-01 | 2017-03-22 | 电子科技大学 | Fault prognostics and health management method applied to circuits and systems |
| CN109308522A (en) * | 2018-09-03 | 2019-02-05 | 河海大学常州校区 | A kind of GIS failure prediction method based on Recognition with Recurrent Neural Network |
| CN110348615A (en) * | 2019-06-27 | 2019-10-18 | 西安工程大学 | Cable line fault probability forecasting method based on ant group optimization support vector machines |
| CN110995384A (en) * | 2019-11-27 | 2020-04-10 | 杭州联汇科技股份有限公司 | Broadcast master control fault trend prejudging method based on machine learning |
| CN111274737A (en) * | 2020-02-25 | 2020-06-12 | 山东大学 | Method and system for predicting remaining service life of mechanical equipment |
| CN112069930A (en) * | 2020-08-20 | 2020-12-11 | 国网山西省电力公司电力科学研究院 | Vibration signal processing method and device for improving GIS equipment fault diagnosis accuracy |
| CN112257333A (en) * | 2020-09-24 | 2021-01-22 | 浙江工业大学 | Mechanical equipment internal component service life prediction method based on deep learning |
Non-Patent Citations (2)
| Title |
|---|
| 李东妍等: "超高压GIS剩余寿命评估方法综述", 《高压电器》 * |
| 王均华等: "变电设备状态评估及故障诊断系统设计", 《农村电气化》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116780782A (en) * | 2023-08-24 | 2023-09-19 | 成都易电云商工程服务有限公司 | Comprehensive maintenance system for electrical equipment of power distribution network |
| CN116780782B (en) * | 2023-08-24 | 2023-10-31 | 成都易电云商工程服务有限公司 | Comprehensive maintenance system for electrical equipment of power distribution network |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111596604B (en) | Intelligent fault diagnosis and self-healing control system and method for engineering equipment based on digital twinning | |
| CN113762329A (en) | Method and system for constructing state prediction model of large rolling mill | |
| CN108304567B (en) | Method and system for identifying working condition mode and classifying data of high-voltage transformer | |
| CN112069930A (en) | Vibration signal processing method and device for improving GIS equipment fault diagnosis accuracy | |
| CN111325410A (en) | General fault early warning system based on sample distribution and early warning method thereof | |
| CN114386537A (en) | Lithium battery fault diagnosis method and device based on Catboost and electronic equipment | |
| CN113379252A (en) | Bayesian network high-voltage switch cabinet health system evaluation method based on multi-scale arrangement entropy | |
| CN110703078A (en) | GIS fault diagnosis method based on spectral energy analysis and self-organizing competition algorithm | |
| CN110794254B (en) | Power distribution network fault prediction method and system based on reinforcement learning | |
| CN114091549A (en) | Equipment fault diagnosis method based on deep residual error network | |
| CN116559650A (en) | Automatic change-over switch fault identification method based on multidimensional entropy distance | |
| CN108445382A (en) | A kind of rapid failure detection method of high-voltage circuitbreaker | |
| CN117607672A (en) | Intelligent monitoring method and system for GIS circuit breaker | |
| CN112883639A (en) | GIS equipment service life prediction device and method based on machine learning | |
| CN115828129A (en) | Intelligent power utilization system fault detection method and system based on convolution feature clustering | |
| CN113988205B (en) | Method and system for judging electric precipitation working condition | |
| CN115292820A (en) | Method for predicting residual service life of urban rail train bearing | |
| CN114137915A (en) | Fault diagnosis method for industrial equipment | |
| Gao et al. | Fault detection of electric vehicle charging piles based on extreme learning machine algorithm | |
| CN113722659B (en) | Fault mode-based substation main equipment diagnosis method and system | |
| Zhao et al. | Research on machine learning-based correlation analysis method for power equipment alarms | |
| CN115684835B (en) | Power distribution network fault diagnosis method, device, equipment and storage medium | |
| CN118312734B (en) | Power system equipment state monitoring and fault diagnosis method | |
| CN117691756B (en) | Safety early warning management method and system for power distribution cabinet | |
| CN118035923B (en) | Power grid wave recording abnormal signal identification method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210601 |