CN116050130A - Multi-factor-based hydropower station generator deduction oil level prediction method - Google Patents
Multi-factor-based hydropower station generator deduction oil level prediction method Download PDFInfo
- Publication number
- CN116050130A CN116050130A CN202310032338.4A CN202310032338A CN116050130A CN 116050130 A CN116050130 A CN 116050130A CN 202310032338 A CN202310032338 A CN 202310032338A CN 116050130 A CN116050130 A CN 116050130A
- Authority
- CN
- China
- Prior art keywords
- oil level
- data
- oil
- matrix
- level 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 33
- 239000011159 matrix material Substances 0.000 claims abstract description 57
- 238000012549 training Methods 0.000 claims abstract description 28
- 238000013528 artificial neural network Methods 0.000 claims abstract description 9
- 239000003921 oil Substances 0.000 claims description 118
- 210000002569 neuron Anatomy 0.000 claims description 24
- 239000000498 cooling water Substances 0.000 claims description 17
- 239000010724 circulating oil Substances 0.000 claims description 10
- 238000004364 calculation method Methods 0.000 claims description 7
- 238000013507 mapping Methods 0.000 claims description 7
- 230000005484 gravity Effects 0.000 claims description 6
- 238000012887 quadratic function Methods 0.000 claims description 6
- 238000012546 transfer Methods 0.000 claims description 6
- 239000010723 turbine oil Substances 0.000 claims description 6
- 238000012545 processing Methods 0.000 claims description 5
- 238000011156 evaluation Methods 0.000 claims description 3
- 238000011478 gradient descent method Methods 0.000 claims description 3
- 230000010355 oscillation Effects 0.000 claims description 3
- 238000012423 maintenance Methods 0.000 description 3
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 2
- 238000005299 abrasion Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000008595 infiltration Effects 0.000 description 1
- 238000001764 infiltration Methods 0.000 description 1
- 238000005461 lubrication 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
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
-
- 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
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- 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
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2113/00—Details relating to the application field
- G06F2113/08—Fluids
-
- 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
- G06F2119/06—Power analysis or power optimisation
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Geometry (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Evolutionary Computation (AREA)
- Quality & Reliability (AREA)
- Development Economics (AREA)
- Computer Hardware Design (AREA)
- General Engineering & Computer Science (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
A hydroelectric power station generator deduction oil level prediction method based on multiple factors comprises the following steps: obtaining original data through a sensor, and then obtaining an original factor data matrix R and an original oil level matrix D; carrying out dimensionless treatment on the original factor data matrix R and the oil level original matrix D; obtaining a standardized factor data matrix X; weighting the standardized factor data matrix X; establishing an oil level prediction model, and training the oil level prediction model by using a three-layer back propagation neural network; inputting training samples to calculate errors; judging whether the oil level prediction model is trained; and if the training is completed, predicting by using the oil level prediction model. The method improves the accuracy and efficiency of deducing the oil level, uses multidimensional factors to establish a temperature prediction model, and avoids the influence on the historical trend analysis of the oil level; therefore, the oil level prediction result is more reasonable and accurate. The operation reliability of the hydropower station generator set can be effectively improved.
Description
Technical Field
The invention belongs to the technical field of oil level prediction deduction, and particularly relates to a multi-factor-based oil level prediction method for a hydropower station generator.
Background
The deduced oil level of the hydropower station generator can intuitively reflect the running state of a unit, and when the deduced oil level is too high (water seepage) or too low (oil leakage), the generator can burn tiles and even damage a large shaft due to insufficient lubrication. Therefore, the practical significance of timely and accurately predicting the variation trend of the deduced oil level of the generator of the water power station is great, early warning can be carried out in advance, and operation and maintenance personnel can take relevant emergency measures in sufficient time.
The existing hydropower station oil level deduction prediction method has a least square fitting deduction oil level trend curve. The method relies on measuring n derived oil level data with equal time intervals, finding the best function match of the data by minimizing the sum of squares of errors, and obtaining a function of derived oil level with respect to time such that the sum of distances of the n points to the function curve is minimized. And substituting the future t moment into the function to obtain the predicted value of the deduced oil level. Disadvantages of the prior art:
1) The independent variable is single, and the independent variable for deriving the oil level target function is only time. The deduced oil level is affected by a plurality of factors such as the rotating speed of a unit, the deduced external circulation oil flow and the like, and the predicted result is not reasonable enough.
2) The least square method ignores the influence of some noise points in the process of solving an objective function, and the noise points are probably caused by equipment faults, so that the least square method is important for oil level prediction. The objective function thus calculated is distorted to some extent.
Disclosure of Invention
In view of the technical problems in the background art, the multi-factor-based hydroelectric generator oil level deducing prediction method provided by the invention improves the accuracy and efficiency of oil level deduction, and a temperature prediction model is built by using multi-dimensional factors, so that the influence on historical trend analysis of the oil level is avoided; therefore, the oil level prediction result is more reasonable and accurate. The operation reliability of the hydropower station generator set can be effectively improved.
In order to solve the technical problems, the invention adopts the following technical scheme:
a hydroelectric power station generator deduction oil level prediction method based on multiple factors comprises the following steps:
step 1: the method comprises the steps of obtaining original data through a sensor, wherein the original data comprise m pieces of active power data, rotating speed data, oil flow data, turbine oil temperature data, cooling water flow data and oil tank oil level data which are obtained at equal time intervals t; then an original factor data matrix R and an oil level original matrix D are obtained;
step 2: carrying out dimensionless treatment on the original factor data matrix R and the oil level original matrix D; obtaining a standardized factor data matrix X;
step 3: weighting the standardized factor data matrix X;
step 4: establishing an oil level prediction model, and training the oil level prediction model by using a three-layer back propagation neural network;
step 5: inputting training samples to calculate errors;
step 6: judging whether the oil level prediction model is trained; if training is completed, entering a step 7;
step 7: the prediction is performed using an oil level prediction model.
Preferably, the operation of step 1 is as follows:
from the current time T, m active power data points r of the equal time interval T are measured by a unit power sensor 11 、r 21 、r 31 ......r m1 ;
From the current time T, m rotational speed data points r of the equal time interval T are measured by a rotational speed measuring device 12 、r 22 、r 32 、......r m2 ;
Starting from the current time T by deductionThe circulating oil flowmeter measures and obtains m deduction oil flow data points r of the equal time interval t 13 、r 23 、r 33 、......r m3 ;
From the current time T, m deduced turbine oil temperature data points r of the equal time interval T are measured by a deduced oil groove thermometer 14 、r 24 、r 34 、......r m4 ;
From the current time T, m deduced cooling water flow data points r of the equal time interval T are measured by a deduced cooling water flow meter 15 、r 25 、r 35 、......r m5 ;
From time t+t, m derived sump oil level data points d for the time interval T are measured by the derived oil level gauge 1 、d 2 、d 3 、......d m ;
Thereby obtaining an original factor data matrix R= (R) ij ) mx5 ,
And oil level primitive matrix d= [ D ] 1 ,d 2 ,…,d m ];
Where i=1, 2, m;
j=1,2,...,5。
preferably, in step 2, the active power, the deduced circulating oil flow and the deduced cooling water flow belong to forward indexes, dimensionless treatment is carried out according to the formula (2), and the positive indexes are mapped into the [0,1] interval;
where i=1, 2, m;
j=1,3,5;
deducing that the temperature of the oil groove is a negative index, carrying out dimensionless treatment according to a formula (3), and mapping the oil groove into a [0,1] interval;
where i=1, 2, m;
j=4;
the rotating speed of the unit is an oscillation index, and the ideal value is f; carrying out dimensionless treatment by taking the formula (4) and mapping the non-dimensionality treatment to the interval of [0,1 ];
where i=1, 2, m;
j=2;
f is a rated rotation speed value;
after dimensionless processing is carried out on the data, a standardized factor data matrix X= (X) can be obtained ij ) mx5 ,
Where i=1, 2, m;
j=1,2,...,5。
preferably, in step 3, according to the definition of entropy,
firstly, calculating the specific gravity O= (O) of the index occupied by different evaluation targets ij ) mx5 (6);
Re-calculating the entropy value e of each index based on specific gravity
Where i=1, 2, m;
j=1, 2,..5, and assuming wheno ij When=0, ln (o ij ) The value of (2) is 0;
finally, calculating the entropy weight of the jth index as omega j ;
Thereby obtaining the entropy weight matrix column vector of each index
ω=(ω 1 ,ω 2 ,...ω n ) T (10)
Finally weighting the standardized data matrix by utilizing the entropy weight matrix to obtain a weighted standardized matrix Z,
Z={z ij } m×5 ={x ij ω j } m×5 (11)
in the method, in the process of the invention, i=1, 2., m;
j=1,2,...,n。
preferably, in step 4, training of an oil level prediction model is performed using a three-layer counter-propagating neural network, in which the number of neurons of an input layer is 5, the number of neurons of an hidden layer is 3, and the number of neurons of an output layer is 1; the mth neuron of the input layer is denoted as x m Hidden layer ith neuron is denoted as k i ,x m To k i Weight of ω mi Ki to Y 1 Weight of ω i The hidden layer uses Log-Sigmoid transfer functionThe output layer uses a linear transfer function g (x) =x+b, where b is a constant;
training samples are m groups of samples in the weighted data matrix obtained by using the formula (11) and are omega mi ,ω i B is from the interval [ -0.48,0.48]Is assigned a value once.
Preferably, in step 5, let n be the calculated iteration number;
the output of the input layer is the training sample data of each group;
The input to the hidden layer i-th neuron is:
the output of the hidden layer i-th neuron is:
the output of the output layer neuron is then:
the model desired output is d (n) = [ d ]; if the difference between the actual value and the expected value after the nth iteration is recorded as: l (n) =d (n) -v (n);
the model error is noted as:
according to the gradient descent method, the error signal is co-propagated in the opposite direction i The adjustment amount of (2) is as follows:
wherein eta is the adjustment step length;
since S (n) is a quadratic function of L (n), L (n) is a quadratic function of v (n), g' (x) =1, the above formula is written:
similarly, the error signal is reversedPropagation time omega mi The adjustment amount of (2) is as follows:
wherein eta is the adjustment step length;
ω mi (n+1)=Δω mi (n)+ω mi (n) (19)
ω i (n+1)=Δω i (n)+ω i (n) (20)。
Preferably, in step 6, each set of sample data in the input layer is a row vector in Z, and m sets of samples are recorded together from the first row to the m-th row; setting an error margin, let n=1 (n.ltoreq.m);
(1) sample group n [ z ] n1 z n2 z n3 z n4 z n5 ]Carrying out formulas (12) - (15), and obtaining errors by using a formula (16);
(2) if the result calculated by the step (16) is smaller than the error tolerance, the algorithm is considered to be converged, and the model training is completed;
(3) otherwise, continue to pass ω through equations (17) - (20) mi (n) and ω i (n) updating to omega mi (n+1) and ω i (n+1);
(4) Then let n=n+1, if n < m, go back to step (1) to continue calculation; when n=m, then model training is complete.
Preferably, in step 7, the active power, the unit rotation speed and the pushing are measured at any time TGuiding circulation oil flow, oil groove oil temperature and cooling water flow data [ r ] 11 ’r 12 ’r 13 ’r 14 ’r 15 ’]Substituting the predicted oil level value into formulas (2) - (15) to obtain a predicted oil level value v 'at the moment T' +t.
The following beneficial effects can be achieved in this patent:
1. the method improves the accuracy and efficiency of deducing the oil level, uses multidimensional factors to establish a temperature prediction model, and avoids the influence on the historical trend analysis of the oil level; therefore, the oil level prediction result is more reasonable and accurate. The operation reliability of the hydropower station generator set can be effectively improved.
2. The invention provides a deduced oil level prediction method based on the actual running condition of a hydropower station generator set, and the deduced oil level at a certain moment in the future can be predicted more accurately. The invention is based on various dimensional factors affecting the deduced oil level, namely data of various different dimensions: active power, unit rotational speed, deriving circulating oil flow, oil sump oil temperature, and cooling water flow. And (3) carrying out weight processing on the factors, and establishing an oil level prediction model based on a neural network after weighting various factors. And after training the model by using the historical data, predicting the real-time data of the deduced oil level. Thereby making the prediction result more reasonable and accurate.
3. According to the invention, various oil level association factors with different dimensions are used for prediction calculation, and the model convergence speed and accuracy are improved by weighting the data matrix. These correlation factors not only facilitate measurement acquisition, but also promote the accuracy of deriving the oil level prediction. Can help operation and maintenance personnel acquire deducing the oil level suggestion in advance, discover the difficult trend of observing deducing oil groove oil leak or infiltration of naked eye. Therefore, operation and maintenance personnel can take measures in time, and huge economic losses caused by severe abrasion of the thrust tile surface are avoided.
Drawings
The invention is further illustrated by the following examples in conjunction with the accompanying drawings:
FIG. 1 is a flow chart of a prediction method of the present invention;
FIG. 2 is a neural network diagram of the oil level prediction model of the present invention.
Detailed Description
Example 1:
the optimal scheme is as shown in fig. 1-2, in the method for predicting the deduced oil level of the hydroelectric generator based on multiple factors, various deduced oil level association factors with different dimensions are used for participating in prediction calculation, and the data matrix is weighted to improve the convergence speed and accuracy of the model and the accuracy of deduced oil level prediction. The specific scheme is as follows:
step 1: the method comprises the steps of obtaining original data through a sensor, wherein the original data comprise m pieces of active power data, rotating speed data, oil flow data, turbine oil temperature data, cooling water flow data and oil tank oil level data which are obtained at equal time intervals t; then an original factor data matrix R and an oil level original matrix D are obtained; the specific operation is as follows:
from the current time T, m active power data points r of the equal time interval T are measured by a unit power sensor 11 、r 21 、r 31 ......r m1 ;
From the current time T, m rotational speed data points r of the equal time interval T are measured by a rotational speed measuring device 12 、r 22 、r 32 、......r m2 ;
From the current time T, m deduced oil flow data points r of the equal time interval T are measured by a deduced circulating oil flowmeter 13 、r 23 、r 33 、......r m3 ;
From the current time T, m deduced turbine oil temperature data points r of the equal time interval T are measured by a deduced oil groove thermometer 14 、r 24 、r 34 、......r m4 ;
From the current time T, m deduced cooling water flow data points r of the equal time interval T are measured by a deduced cooling water flow meter 15 、r 25 、r 35 、......r m5 ;
From time t+t, the isochrone is measured by deriving the oil level gaugeM derived sump oil level data points d at interval t 1 、d 2 、d 3 、......d m ;
Thereby obtaining an original factor data matrix R= (R) ij ) mx5 ,
And oil level primitive matrix d= [ D ] 1 ,d 2 ,…,d m ];
Where i=1, 2, m;
j=1,2,...,5。
step 2: carrying out dimensionless treatment on the original factor data matrix R and the oil level original matrix D; obtaining a standardized factor data matrix X; the specific operation is as follows:
because the data dimensions in the original factor data matrix and the oil level original data matrix are not all the same, the original factor data matrix and the oil level original data matrix cannot directly participate in calculation, and therefore standardized processing is required to be carried out on the original factor data matrix and the oil level original data matrix; the active power, the deduced circulating oil flow and the deduced cooling water flow belong to forward indexes, dimensionless treatment is carried out according to a formula (2), and the positive indexes are mapped into a [0,1] interval;
where i=1, 2, m;
j=1,3,5;
deducing that the temperature of the oil groove is a negative index, carrying out dimensionless treatment according to a formula (3), and mapping the oil groove into a [0,1] interval;
where i=1, 2, m;
j=4;
the rotation speed of the unit is an oscillation index, and the ideal value of the unit is f (f=50 Hz in general); carrying out dimensionless treatment by bringing the non-dimensional information into a formula (4), and mapping the non-dimensional information into a [0,1] interval;
where i=1, 2, m;
j=2;
f is a rated rotation speed value;
after dimensionless processing is carried out on the data, a standardized factor data matrix X= (X) can be obtained ij ) mx5 ,
Where i=1, 2, m;
j=1,2,...,5。
step 3: weighting the standardized factor data matrix X; the specific operation is as follows:
for the change of the unit deduced oil level, the active power, the unit rotating speed, the deduced circulating oil flow, the deduced oil tank oil temperature and the deduced cooling water flow factor have incomplete influence degrees, so that the influence weight of each factor is calculated, and then the standardized data matrix is weighted, so that the calculation convergence of a follow-up oil level prediction model can be better assisted, and the predicted value is more accurate. The step utilizes entropy weight to calculate the weight of each factor on the oil level. According to the definition of the entropy of the light,
firstly, calculating the specific gravity O= (O) of the index occupied by different evaluation targets ij ) mx5 (6);
Re-calculating the entropy value e of each index based on specific gravity
Where i=1, 2, m;
j=1, 2,..5, and assume when o ij When=0, ln (o ij ) The value of (2) is also 0;
finally, calculating the entropy weight of the jth index as omega j ;
Thereby obtaining the entropy weight matrix column vector of each index
ω=(ω 1 ,ω 2 ,...ω n ) T (10)
Finally weighting the standardized data matrix by utilizing the entropy weight matrix to obtain a weighted standardized matrix Z,
Z={z ij } m×5 ={x ij ω j } m×5 (11)
in the method, in the process of the invention, i=1, 2., m;
j=1,2,...,n。
step 4: establishing an oil level prediction model, and training the oil level prediction model by using a three-layer back propagation neural network; the specific operation is as follows:
the training of the oil level prediction model was performed using a three-layer back propagation neural network, the network structure of which is shown in fig. 2. In the prediction model, the number of neurons of an input layer is 5, the number of neurons of an implicit layer is 3, and the number of neurons of an output layer is 1; the mth neuron of the input layer is denoted as x m Hidden layer ith neuron is denoted as k i ,x m To k i Weight of ω mi Ki to Y 1 Weight of ω i The hidden layer uses Log-Sigmoid transfer functionThe output layer uses a linear transfer function g (x) =x+b, where b is a constant;
the training sample is obtained by using the formula (11)M groups of samples in the weighted data matrix and omega mi ,ω i B is from the interval [ -0.48,0.48]Is assigned a value once.
Step 5: inputting training samples to calculate errors; the specific operation is as follows:
setting n as the calculated iteration times;
The input to the hidden layer i-th neuron is:
the output of the hidden layer i-th neuron is:
then the output of the output layer neuron is
The model desired output is d (n) = [ d ]; if the difference between the actual value and the expected value after the nth iteration is recorded as: l (n) =d (n) -v (n)
The model error is noted as:
according to the gradient descent method, the error signal is co-propagated in the opposite direction i The adjustment amount of (2) is as follows:
wherein eta is the adjustment step length;
since S (n) is a quadratic function of L (n), L (n) is a quadratic function of v (n), g' (x) =1, the above formula can be written as:
similarly, ω is the error signal counter-propagating mi The adjustment amount of (2) is as follows:
wherein, eta is the adjustment step length and is generally 0.6;
ω mi (n+1)=Δω mi (n)+ω mi (n) (19)
ω i (n+1)=Δω i (n)+ω i (n) (20)。
Step 6: judging whether the model is trained; if training is completed, entering a step 7; the specific operation is as follows:
each group of sample data in the input layer is a row vector in Z, and m groups of samples are recorded from the first row to the m-th row; an error margin is set, typically taking 0.01. Let n=1 (n.ltoreq.m);
(1) sample group n [ z ] n1 z n2 z n3 z n4 z n5 ]Carrying out formulas (12) - (15), and obtaining errors by using a formula (16);
(2) if the result calculated by the step (16) is smaller than the error tolerance, the algorithm is considered to be converged, and the model training is completed;
(3) otherwise, continue to pass ω through equations (17) - (20) mi (n) and ω i (n) updating to omega mi (n+1) and ω i (n+1);
(4) Then let n=n+1, if n < m, go back to step (1) to continue calculation; when n=m, then model training is complete.
Step 7: the model is used for prediction. The specific operation is as follows:
after the model is trained through the step 6, only one active power, unit rotating speed, circulating oil flow, oil groove oil temperature and cooling water flow data [ r ] are measured at any time T ] 11 ’r 12 ’r 13 ’r 14 ’r 15 ’]Substituting the predicted oil level value into formulas (2) - (15) to obtain a predicted oil level value v 'at the moment T' +t.
The above embodiments are only preferred embodiments of the present invention, and should not be construed as limiting the present invention, and the scope of the present invention should be defined by the claims, including the equivalents of the technical features in the claims. I.e., equivalent replacement modifications within the scope of this invention are also within the scope of the invention.
Claims (8)
1. The method for predicting the deduced oil level of the hydroelectric power generator based on multiple factors is characterized by comprising the following steps of:
step 1: the method comprises the steps of obtaining original data through a sensor, wherein the original data comprise m pieces of active power data, rotating speed data, oil flow data, turbine oil temperature data, cooling water flow data and oil tank oil level data which are obtained at equal time intervals t; then an original factor data matrix R and an oil level original matrix D are obtained;
step 2: carrying out dimensionless treatment on the original factor data matrix R and the oil level original matrix D; obtaining a standardized factor data matrix X;
step 3: weighting the standardized factor data matrix X;
step 4: establishing an oil level prediction model, and training the oil level prediction model by using a three-layer back propagation neural network;
step 5: inputting training samples to calculate errors;
step 6: judging whether the oil level prediction model is trained; if training is completed, entering a step 7;
step 7: the prediction is performed using an oil level prediction model.
2. The multi-factor based hydroelectric generator derived oil level prediction method of claim 1, wherein: the operation mode of the step 1 is as follows:
from the current time T, m active power data points r of the equal time interval T are measured by a unit power sensor 11 、r 21 、r 31 ......r m1 ;
From the current time T, m rotational speed data points r of the equal time interval T are measured by a rotational speed measuring device 12 、r 22 、r 32 、......r m2 ;
From the current time T, m deduced oil flow data points r of the equal time interval T are measured by a deduced circulating oil flowmeter 13 、r 23 、r 33 、......r m3 ;
From the current time T, m deduced turbine oil temperature data points r of the equal time interval T are measured by a deduced oil groove thermometer 14 、r 24 、r 34 、......r m4 ;
From the current time T, m deduced cooling water flow data points r of the equal time interval T are measured by a deduced cooling water flow meter 15 、r 25 、r 35 、......r m5 ;
From time t+t, m derived sump oil level data at equal time intervals T are measured by the derived oil level gaugePoint d 1 、d 2 、d 3 、......d m ;
Thereby obtaining an original factor data matrix R= (R) ij ) mx5 ,
And oil level primitive matrix d= [ D ] 1 ,d 2 ,…,d m ];
Where i=1, 2, m;
j=1,2,...,5。
3. the multi-factor based hydroelectric generator derived oil level prediction method of claim 1, wherein: in the step 2, active power, deduced circulating oil flow and deduced cooling water flow belong to forward indexes, dimensionless treatment is carried out according to a formula (2), and mapping is carried out in a [0,1] interval;
where i=1, 2, m;
j=1,3,5;
deducing that the temperature of the oil groove is a negative index, carrying out dimensionless treatment according to a formula (3), and mapping the oil groove into a [0,1] interval;
where i=1, 2, m;
j=4;
the rotating speed of the unit is an oscillation index, and the ideal value is f; carrying out dimensionless treatment by taking the formula (4) and mapping the non-dimensionality treatment to the interval of [0,1 ];
where i=1, 2, m;
j=2;
f is a rated rotation speed value;
after dimensionless processing is carried out on the data, a standardized factor data matrix X= (X) can be obtained ij ) mx5 ,
Where i=1, 2, m;
j=1,2,...,5。
4. the multi-factor based hydroelectric generator derived oil level prediction method of claim 1, wherein: in step 3, according to the definition of entropy,
firstly, calculating the specific gravity O= (O) of the index occupied by different evaluation targets ij ) mx5 (6);
Re-calculating the entropy value e of each index based on specific gravity
Where i=1, 2, m;
j=1, 2,..5, and assume when o ij When=0, ln (o ij ) The value of (2) is 0;
finally, calculating the entropy weight of the jth index as omega j ;
Thereby obtaining the entropy weight matrix column vector of each index
ω=(ω 1 ,ω 2 ,...ω n ) T (10)
Finally weighting the standardized data matrix by utilizing the entropy weight matrix to obtain a weighted standardized matrix Z,
Z={z ij } m×5 ={x ij ω j } m×5 (11)
in the method, in the process of the invention, i=1, 2., m;
j=1,2,...,n。
5. the multi-factor based hydroelectric generator derived oil level prediction method of claim 1, wherein: in the step 4, training an oil level prediction model by using a three-layer counter propagation neural network, wherein in the prediction model, the number of neurons of an input layer is 5, the number of neurons of an hidden layer is 3, and the number of neurons of an output layer is 1; the mth neuron of the input layer is denoted as x m Hidden layer ith neuron is denoted as k i ,x m To k i Weight of ω mi Ki to Y 1 Weight of ω i The hidden layer uses Log-Sigmoid transfer functionThe output layer uses a linear transfer function g (x) =x+b, where b is a constant;
training samples are m groups of samples in the weighted data matrix obtained by using the formula (11) and are omega mi ,ω i B is from the interval [ -0.48,0.48]Is assigned a value once.
6. The multi-factor based hydroelectric generator derived oil level prediction method of claim 1, wherein: in step 5, setting n as the calculated iteration times;
the output of the input layer is the training sample data of each group;
The input to the hidden layer i-th neuron is:
the output of the hidden layer i-th neuron is:
the output of the output layer neuron is then:
the model desired output is d (n) = [ d ]; if the difference between the actual value and the expected value after the nth iteration is recorded as: l (n) =d (n) -v (n);
the model error is noted as:
according to the gradient descent method, the error signal is co-propagated in the opposite direction i The adjustment amount of (2) is as follows:
wherein eta is the adjustment step length;
since S (n) is a quadratic function of L (n), L (n) is a quadratic function of v (n), g' (x) =1, the above formula is written:
similarly, ω is the error signal counter-propagating mi The adjustment amount of (2) is as follows:
wherein eta is the adjustment step length;
ω mi (n+1)=Δω mi (n)+ω mi (n) (19)
ω i (n+1)=Δω i (n)+ω i (n) (20)。
7. The multi-factor based hydroelectric generator derived oil level prediction method of claim 1, wherein: in step 6, each group of sample data in the input layer is a row vector in Z, and m groups of samples are recorded from the first row to the m-th row; setting an error margin, let n=1 (n.ltoreq.m);
(1) sample group n [ z ] n1 z n2 z n3 z n4 z n5 ]Carrying out formulas (12) - (15), and obtaining errors by using a formula (16);
(2) if the result calculated by the step (16) is smaller than the error tolerance, the algorithm is considered to be converged, and the model training is completed;
(3) otherwise, continue to pass ω through equations (17) - (20) mi (n) and ω i (n) updating to omega mi (n+1) and ω i (n+1);
(4) Then let n=n+1, if n < m, go back to step (1) to continue calculation; when n=m, then model training is complete.
8. The multi-factor based hydroelectric generator derived oil level prediction method of claim 1, wherein: in step 7, one active power, unit rotation speed, circulating oil flow, oil groove oil temperature and cooling water flow data are measured at any time T [, r ] 11 ’r 12 ’r 13 ’r 14 ’r 15 ’]Substituting the predicted oil level value into formulas (2) - (15) to obtain a predicted oil level value v 'at the moment T' +t.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310032338.4A CN116050130B (en) | 2023-01-10 | 2023-01-10 | Multi-factor-based hydropower station generator deduction oil level prediction method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310032338.4A CN116050130B (en) | 2023-01-10 | 2023-01-10 | Multi-factor-based hydropower station generator deduction oil level prediction method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN116050130A true CN116050130A (en) | 2023-05-02 |
CN116050130B CN116050130B (en) | 2024-03-12 |
Family
ID=86116779
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310032338.4A Active CN116050130B (en) | 2023-01-10 | 2023-01-10 | Multi-factor-based hydropower station generator deduction oil level prediction method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116050130B (en) |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2004150375A (en) * | 2002-10-31 | 2004-05-27 | Honda Motor Co Ltd | Device for judging engine-oil degradation |
IL206232A0 (en) * | 2010-06-07 | 2010-11-30 | Izidor Liberman | System and method for planning of demand for power on an electrical power network |
US20130080022A1 (en) * | 2011-04-08 | 2013-03-28 | Gm Global Technology Operations Llc. | Engine oil level monitoring systems and methods |
CN112885047A (en) * | 2021-01-15 | 2021-06-01 | 湖南五凌电力科技有限公司 | Intelligent early warning method for state monitoring of unit, transformer and auxiliary machine |
CN114202112A (en) * | 2021-11-29 | 2022-03-18 | 国网福建省电力有限公司经济技术研究院 | Regional mode-considered power supply business hall service work order scale combined prediction method |
CN115204658A (en) * | 2022-07-12 | 2022-10-18 | 中国电力科学研究院有限公司 | Method and system for assessing health state of oil supply equipment of hydroelectric generating set |
-
2023
- 2023-01-10 CN CN202310032338.4A patent/CN116050130B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2004150375A (en) * | 2002-10-31 | 2004-05-27 | Honda Motor Co Ltd | Device for judging engine-oil degradation |
IL206232A0 (en) * | 2010-06-07 | 2010-11-30 | Izidor Liberman | System and method for planning of demand for power on an electrical power network |
US20130080022A1 (en) * | 2011-04-08 | 2013-03-28 | Gm Global Technology Operations Llc. | Engine oil level monitoring systems and methods |
CN112885047A (en) * | 2021-01-15 | 2021-06-01 | 湖南五凌电力科技有限公司 | Intelligent early warning method for state monitoring of unit, transformer and auxiliary machine |
CN114202112A (en) * | 2021-11-29 | 2022-03-18 | 国网福建省电力有限公司经济技术研究院 | Regional mode-considered power supply business hall service work order scale combined prediction method |
CN115204658A (en) * | 2022-07-12 | 2022-10-18 | 中国电力科学研究院有限公司 | Method and system for assessing health state of oil supply equipment of hydroelectric generating set |
Non-Patent Citations (1)
Title |
---|
春洪浩: "冶勒水电站发电机状态监测系统简介", 水力发电, no. 11 * |
Also Published As
Publication number | Publication date |
---|---|
CN116050130B (en) | 2024-03-12 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN101630376B (en) | Soft-sensing modeling method and soft meter of multi-model neural network in biological fermentation process | |
CN110807554B (en) | Generation method and system based on wind power/photovoltaic classical scene set | |
CN113281697B (en) | Operation error online analysis method and system | |
CN114109859B (en) | Centrifugal pump performance neural network prediction method without flow sensing | |
CN101793887A (en) | Construction method of fuzzy neural network expert system for water quality assessment in turbot culture | |
CN110619432A (en) | Feature extraction hydrological forecasting method based on deep learning | |
CN102419827B (en) | Radial basis function (RBF) neural network-based boiling heat exchanging prediction method | |
CN105824987A (en) | Wind field characteristic statistical distributing model building method based on genetic algorithm | |
CN113706337A (en) | Heat supply load prediction method based on similar time periods | |
CN113642259B (en) | River flow field reconstruction method based on multilayer flow velocity measurement | |
CN113991711B (en) | Capacity configuration method for energy storage system of photovoltaic power station | |
CN106405683B (en) | Wind speed forecasting method and device based on G-L mixed noise characteristic core ridge regression technology | |
CN113836703B (en) | Calculation method for resistance coefficient of river channel containing submerged vegetation | |
CN116050130B (en) | Multi-factor-based hydropower station generator deduction oil level prediction method | |
CN113392497B (en) | Method for measuring field performance degradation of photovoltaic module according to geographical region | |
CN108647817A (en) | A kind of prediction technique and forecasting system of energy consumption load | |
CN113159395A (en) | Deep learning-based sewage treatment plant water inflow prediction method and system | |
CN116757078A (en) | Method and system for measuring flow velocity of pulverized coal based on acting force | |
CN116663727A (en) | Photovoltaic power prediction method and system | |
CN113489072B (en) | Power distribution network state estimation method considering branch power pseudo measurement | |
Liu et al. | Improved dissipation rate estimation method based on time-varying glide parameter analysis of underwater gliders | |
CN109101759A (en) | A kind of parameter identification method based on forward and reverse response phase method | |
CN105205560A (en) | Photovoltaic power supply power prediction method based on positive and negative error variable weights | |
CN114781606A (en) | Method for predicting medium and long term electricity consumption in region based on federal learning | |
CN105699043B (en) | A kind of wind tunnel sensors that improve measure stability and the method for precision |
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 |