CN117238113A - Early warning method for hydroelectric component working condition measurement value based on probability function - Google Patents
Early warning method for hydroelectric component working condition measurement value based on probability function Download PDFInfo
- Publication number
- CN117238113A CN117238113A CN202311429993.XA CN202311429993A CN117238113A CN 117238113 A CN117238113 A CN 117238113A CN 202311429993 A CN202311429993 A CN 202311429993A CN 117238113 A CN117238113 A CN 117238113A
- Authority
- CN
- China
- Prior art keywords
- generating set
- hydroelectric
- working condition
- hydroelectric generating
- time domain
- 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 61
- 238000005259 measurement Methods 0.000 title claims description 26
- 230000006870 function Effects 0.000 claims abstract description 72
- 238000012544 monitoring process Methods 0.000 claims abstract description 47
- 238000003062 neural network model Methods 0.000 claims abstract description 30
- 238000012549 training Methods 0.000 claims abstract description 18
- 238000001514 detection method Methods 0.000 claims abstract description 16
- 238000013528 artificial neural network Methods 0.000 claims description 9
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 9
- 238000004590 computer program Methods 0.000 claims description 5
- 238000012360 testing method Methods 0.000 claims description 5
- 230000011218 segmentation Effects 0.000 claims description 4
- 238000009825 accumulation Methods 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 abstract description 11
- 238000012545 processing Methods 0.000 description 9
- 230000000694 effects Effects 0.000 description 7
- 238000013527 convolutional neural network Methods 0.000 description 4
- 238000005086 pumping Methods 0.000 description 4
- 230000002159 abnormal effect Effects 0.000 description 3
- 230000004913 activation Effects 0.000 description 3
- 230000001186 cumulative effect Effects 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 230000033228 biological regulation Effects 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000011002 quantification Methods 0.000 description 2
- 238000007493 shaping process Methods 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 230000014759 maintenance of location Effects 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 238000010248 power generation Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
Classifications
-
- 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
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
- Y02E10/00—Energy generation through renewable energy sources
- Y02E10/20—Hydro energy
Abstract
The invention discloses a hydroelectric component working condition measuring value early warning method based on a probability function. Wherein the method comprises the following steps: collecting historical time domain waveform vibration data of each part of the hydroelectric generating set, and eliminating noise data; constructing and training a neural network model based on the historical time domain waveform vibration data, and predicting a working condition judgment result of the hydroelectric generating set based on the trained neural network model; calculating alarm thresholds of all parts of the hydroelectric generating set based on probability functions and working condition judgment results; and carrying out-of-limit alarm detection and shutdown protection judgment on real-time vibration monitoring values of all parts of the hydroelectric generating set based on the alarm threshold. According to the invention, a reasonable probability function is used for defining a calculation method of the runout early warning threshold value, and the shutdown protection logic under different working conditions is determined, so that the accuracy of early warning signal output is ensured.
Description
Technical Field
The invention relates to the technical field of hydroelectric generating set fault early warning, in particular to a hydroelectric generating set component working condition measuring value early warning method based on probability functions.
Background
Vibration is a common phenomenon of the hydroelectric generating set, and strong vibration can influence the normal operation of the hydroelectric generating set and reduce the service life of the generating set and parts. Therefore, the runout signal is an important index for evaluating the running state of the hydroelectric generating set and a fault judgment basis. Various industry rules and power station operation regulations are issued by the existing national and industry standards for determining unit monitoring and early warning thresholds. However, some thresholds of the indexes are changed along with the change of working conditions, relevant regulations are not directly given, and meanwhile, the monitoring value is greatly influenced by factors such as the monitoring installation position, different working conditions of the unit and the like, so that the early warning threshold of each position of the unit cannot be effectively determined only by means of industry standards.
Based on the problem, most of the existing hydropower station runout early warning threshold determining methods mainly comprise statistics of historical data threshold values of the hydropower station and expert experience judgment, and different runout alarm values or runout values are set according to working conditions. And the expert determines the alarm constant value coefficient of each working condition through experience, and multiplies the alarm constant value coefficient with the peak-peak value of the historical data of the hydroelectric generating set to obtain the grading alarm threshold value of each working condition. In order to fully consider the actual situation of the unit and accumulated massive historical data, some students propose to use probability function methods applicable to normal distribution such as 3 sigma criteria and Gaussian distribution to calculate unit monitoring early warning values by analyzing the distribution situation of unit monitoring samples. Because the hydroelectric generating set monitoring data has statistical characteristics obeying the rule of the bias distribution, but is not completely normal distribution, the final output early warning value of the existing scheme is not accurate. .
Disclosure of Invention
Therefore, the invention aims to provide the early warning method for the hydroelectric component working condition measurement value based on the probability function, which can determine shutdown protection logic under different working conditions by using a reasonable calculation method for the probability function to determine the runout early warning threshold value and ensure the accuracy of early warning signal output.
According to one aspect of the invention, there is provided a method for pre-warning of hydroelectric component operating condition measurements based on probability functions, comprising:
collecting historical time domain waveform vibration data of each part of the hydroelectric generating set, and eliminating noise data;
constructing and training a neural network model based on the historical time domain waveform vibration data, and predicting a working condition judgment result of the hydroelectric generating set based on the trained neural network model;
calculating alarm thresholds of all parts of the hydroelectric generating set based on probability functions and working condition judgment results;
and carrying out-of-limit alarm detection and shutdown protection judgment on real-time vibration monitoring values of all parts of the hydroelectric generating set based on the alarm threshold.
In the technical scheme, the working condition of the water outlet motor set is constructed and calculated by utilizing the historical data, and the purpose is that the running condition of the hydroelectric generating set is complex. Therefore, the traditional method for calculating the runout early warning threshold through expert experience and historical data is not completely abandoned, a part of analysis work is constructed by the traditional method for calculating the runout early warning threshold through expert experience and historical data, subsequent work can be completed more quickly, and meanwhile scheme adaptation of different hydroelectric generating sets is facilitated. Further, after a working condition result is obtained by a traditional method, a concept of a probability function is introduced, an alarm threshold value is calculated by the probability function, and out-of-limit alarm detection and shutdown protection judgment are carried out through the alarm threshold value. The probability function is adopted in the scheme, and the hydroelectric generating set monitoring data has statistical characteristics obeying the bias distribution rule, but is not completely normal distribution, so that the Pearson III type frequency curve is used as the probability function suitable for the bias distribution, the monitoring signal can be better fitted, and the accuracy and the reliability of environment assessment are improved. Therefore, the method for calculating the pearson III-type frequency curve clear runout early warning threshold is used for determining shutdown protection logic under different working conditions, and the accuracy of early warning signal output is guaranteed.
In some embodiments, historical time domain waveform vibration data of various components of the hydroelectric generating set is collected, in particular:
measuring large-shaft radial time domain waveform vibration data by adopting an eddy current sensor;
and measuring time domain waveform vibration data of each supporting part of the water turbine set by adopting a low-frequency speed sensor.
In the above technical solution, the purpose of this arrangement is: setting different vibration monitoring sensors according to different positions can collect vibration data more accurately.
In some embodiments, noise data is culled, specifically:
and calculating the average value and standard deviation of each section of historical time domain waveform vibration data, and utilizing the Laida criterion to propose the noise data contained in the historical time domain waveform vibration data.
In the above technical solution, the reason for selecting the rada criterion is that the use is simple, and the time domain waveform data is in a state similar to normal distribution, so that the abnormal data can be quickly removed by using the method. Although the rada rule requires enough measurement data to reject abnormal values, otherwise the error is large, a common hydroelectric generating set has a large amount of historical data, so that the situation cannot occur.
In some embodiments, a neural network model is constructed and trained based on historical time domain waveform vibration data, and the working condition judgment result of the hydroelectric generating set is predicted based on the trained neural network model, specifically:
the time domain waveform data is subjected to equal-length segmentation, and divided into a plurality of groups according to the rotation period, and the time domain characteristic value of each group is calculated;
and taking the calculated time domain characteristic values as a training set and a testing set, constructing and training a neural network model, and predicting the working condition judgment result of the hydroelectric generating set by using the trained neural network model.
In the above technical solution, since there are various time domain feature values of waveform data, such as maximum value, maximum absolute value, minimum value, average value, peak-to-peak value, absolute average value, root mean square value, root amplitude, standard deviation, kurtosis, skewness, margin, waveform, pulse, and peak value. Therefore, a neural network model is adopted, and the relationship of different input characteristic values can be well coordinated by utilizing the strong information comprehensive capability of the neural network model and simultaneously processing quantification and shaping. Compared with the traditional machine learning scheme, the self-learning, self-organization and self-adaptability of the neural network are better.
In some embodiments, the alarm threshold of each component of the hydroelectric generating set is calculated based on the probability function and the working condition determination result, and specifically:
based on the working condition judgment result, selecting time domain waveform vibration data of the hydroelectric generating set components, arranging the time domain waveform vibration data according to the descending order of amplitude values, and calculating the modulus coefficient and the discrete coefficient of each component of the hydroelectric generating set;
setting deviation coefficients of all parts of the hydroelectric generating set, respectively drawing a Pearson-III type theoretical frequency curve graph, respectively fitting a corresponding theoretical frequency curve according to the curve graph, and calculating experience accumulation probability and modulus coefficient of all the parts of the hydroelectric generating set;
and calculating the maximum amplitude value of each part of the hydroelectric generating set based on the modulus coefficient, and setting the alarm threshold value of each part of the hydroelectric generating set according to the maximum amplitude value.
In the above technical solution, the purpose of this arrangement is to: the probability function is adopted in the scheme, and the hydroelectric generating set monitoring data has statistical characteristics obeying the bias distribution rule, but is not completely normal distribution, so that the Pearson III type frequency curve is used as the probability function suitable for the bias distribution, the monitoring signal can be better fitted, and the accuracy and the reliability of environment assessment are improved. Therefore, the method for calculating the pearson III-type frequency curve clear runout early warning threshold is used for determining shutdown protection logic under different working conditions, and the accuracy of early warning signal output is guaranteed.
In some embodiments, out-of-limit alarm detection and shutdown protection determination are performed on real-time vibration monitoring values of all components of the hydroelectric generating set based on alarm thresholds, specifically:
i. out-of-limit alarm detection
If the real-time vibration monitoring value of each component of the hydroelectric generating set accords with the out-of-limit alarm condition, releasing a corresponding level alarm signal, wherein the alarm signal comprises a primary alarm signal and a secondary alarm signal;
determination conditions for issuing a primary alarm signal: the secondary alarm value of the working condition is more than or equal to the primary alarm value of the working condition, and the real-time vibration monitoring value is more than or equal to the primary alarm value of the working condition;
determination conditions for issuing a secondary alarm signal: the real-time vibration monitoring value under the working condition is more than or equal to the secondary alarm value of the working condition;
shutdown protection determination
When the hydroelectric generating set releases two or more secondary alarm signals, if the signals are positioned at two monitoring points of the same hydroelectric generating set component, the hydroelectric generating set is in shutdown protection, and if the signals are respectively positioned at the monitoring points of different components, the hydroelectric generating set is not in shutdown protection.
In the above-described embodiments, the purpose of this arrangement is to set the primary and secondary alarm thresholds for the individual components of the hydroelectric generating set using the maximum amplitude value obtained by the probability function. When the primary and secondary alarm thresholds are used as the judgment standard, the probability function obtains the maximum amplitude value under different accumulated probabilities, and the advantage is that historical data are more fully considered. On the basis, a set of reasonable shutdown logic is established by setting a primary alarm threshold and a secondary alarm threshold by using the empirical cumulative probabilities of 0.1 and 0.01 as indexes, so that the accuracy of early warning signal output is guaranteed.
According to another aspect of the present invention, an early warning device for a hydroelectric component working condition measurement based on a probability function is provided, and an early warning method for a hydroelectric component working condition measurement based on a probability function is provided, which comprises an acquisition module, a neural network module, a probability function module and a determination module, which are electrically connected in sequence:
the acquisition module is used for acquiring historical time domain waveform vibration data of each part of the hydroelectric generating set and eliminating noise data;
the neural network module is used for constructing and training a neural network model based on the historical time domain waveform vibration data and predicting the working condition judgment result of the hydroelectric generating set based on the trained neural network model;
the probability function module is used for calculating alarm thresholds of all parts of the hydroelectric generating set based on probability functions and working condition judgment results;
and the judging module is used for carrying out-of-limit alarm detection and shutdown protection judgment on the real-time vibration monitoring values of all parts of the hydroelectric generating set based on the alarm threshold.
In the technical scheme, in order to better apply the method, different modules are sequentially established in different steps, and the modules are connected in series, so that the method can be more efficiently trained and used. It should be noted that the principle and effect of each step have been described above, and will not be explained here.
According to still another aspect of the present invention, there is provided an early warning apparatus for hydroelectric component operating condition measurement based on probability function, comprising:
at least one processor; the method comprises the steps of,
a memory communicatively coupled to the at least one processor; wherein,
the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of pre-warning hydroelectric component operating conditions measurements based on probability functions described above.
In the above technical solution, for better running and processing of the method, the above method is stored in a memory, and the stored method is executed by a processor. It should be noted that the principle and effect of each step have been described above, and will not be explained here.
According to another aspect of the present invention, a computer readable storage medium is provided, in which a computer program is stored, the computer program, when executed by a processor, implements the foregoing method for early warning of hydroelectric component operating condition measurements based on probability functions.
In the above technical solution, for better operation and use of the method, the above method is stored in a computer readable storage medium and implemented by a processor. It should be noted that the principle and effect of each step have been described above, and will not be explained here.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic flow chart of one embodiment of a method for pre-warning hydroelectric component operating condition measurements based on probability functions according to the present invention;
FIG. 2 is a flowchart of an implementation of an early warning method of an embodiment of one of the embodiments of the early warning method of hydroelectric component working condition measurements based on probability functions of the present invention;
FIG. 3 is a graph of time domain vibration waveforms of horizontal and vertical vibration of a stator core of a specific embodiment of one embodiment of a method for pre-warning of hydroelectric component working condition measurements based on probability functions according to the present invention;
fig. 4 is a flowchart of calculation of a runout warning threshold according to a specific embodiment of one of the embodiments of the method for warning the hydroelectric component working condition measurement based on the probability function.
Detailed Description
The invention is described in further detail below with reference to the drawings and examples. It is specifically noted that the following examples are only for illustrating the present invention, but do not limit the scope of the present invention. Likewise, the following examples are only some, but not all, of the examples of the present invention, and all other examples, which a person of ordinary skill in the art would obtain without making any inventive effort, are within the scope of the present invention.
The invention provides a hydroelectric component working condition measurement early warning method based on a probability function, which can determine shutdown protection logic under different working conditions by using a reasonable probability function to determine a runout early warning threshold calculation method and ensure the accuracy of early warning signal output.
Referring to fig. 1, fig. 1 is a flow chart of an embodiment of a method for early warning of hydroelectric component working condition measurements based on probability functions according to the present invention. It should be noted that, if there are substantially the same results, the method of the present invention is not limited to the flow sequence shown in fig. 1. As shown in fig. 1, the method comprises the steps of:
s101, collecting historical time domain waveform vibration data of each part of a hydroelectric generating set, and eliminating noise data;
in the embodiment, historical time domain waveform vibration data of each part of the hydroelectric generating set is collected, and specifically: measuring large-shaft radial time domain waveform vibration data by adopting an eddy current sensor; and measuring time domain waveform vibration data of each supporting part of the water turbine set by adopting a low-frequency speed sensor.
In the present embodiment, setting different vibration monitoring sensors according to different positions can collect vibration data more accurately.
In this embodiment, it is to be understood that, in order to facilitate the processing of data, the type of data collected locates waveform data first.
In this embodiment, noise data is removed, specifically:
and calculating the average value and standard deviation of each section of historical time domain waveform vibration data, and utilizing the Laida criterion to propose the noise data contained in the historical time domain waveform vibration data.
In this embodiment, the reason for selecting the rada criterion is that the use is simple, and the time domain waveform data is in a state similar to normal distribution. Although the rada rule requires enough measurement data to reject abnormal values, otherwise the error is large, a common hydroelectric generating set has a large amount of historical data, so that the situation cannot occur.
In this embodiment, it should be understood that, in this embodiment, the distance is only described by using the rada criterion, which is a preferred scheme adopted for the purpose of approaching the convenience of the present case for processing data and subsequent processing, and other rejection schemes may also be adopted in a targeted manner.
S102, constructing and training a neural network model based on historical time domain waveform vibration data, and predicting a working condition judgment result of the hydroelectric generating set based on the trained neural network model;
in this embodiment, a neural network model is constructed and trained based on historical time domain waveform vibration data, and a working condition judgment result of the hydroelectric generating set is predicted based on the trained neural network model, specifically:
the time domain waveform data is subjected to equal-length segmentation, and divided into a plurality of groups according to the rotation period, and the time domain characteristic value of each group is calculated;
and taking the calculated time domain characteristic values as a training set and a testing set, constructing and training a neural network model, and predicting the working condition judgment result of the hydroelectric generating set by using the trained neural network model.
In this embodiment, since there are various time domain feature values of waveform data, such as maximum value, maximum absolute value, minimum value, average value, peak-to-peak value, absolute average value, root mean square value, root amplitude, standard deviation, kurtosis, skewness, margin, waveform, pulse, peak value. Therefore, a neural network model is adopted, and the relationship of different input characteristic values can be well coordinated by utilizing the strong information comprehensive capability of the neural network model and simultaneously processing quantification and shaping. Compared with the traditional machine learning scheme, the self-learning, self-organization and self-adaptability of the neural network are better.
In this embodiment, it should be noted that, the present application only defines a scheme of using a neural network, and other schemes may be used to implement the embodiment without limitation. It will be appreciated that the model employed in this step needs to meet the following two points:
1. processing of big data
2. Coordination and rapid processing of multi-feature inputs.
S103, calculating alarm thresholds of all parts of the hydroelectric generating set based on probability functions and working condition judgment results;
in the embodiment, the alarm threshold value of each component of the hydroelectric generating set is calculated based on the probability function and the working condition judgment result, and the method is specifically as follows:
based on the working condition judgment result, selecting time domain waveform vibration data of the hydroelectric generating set components, arranging the time domain waveform vibration data according to the descending order of amplitude values, and calculating the modulus coefficient and the discrete coefficient of each component of the hydroelectric generating set;
setting deviation coefficients of all parts of the hydroelectric generating set, respectively drawing a Pearson-III type theoretical frequency curve graph, respectively fitting a corresponding theoretical frequency curve according to the curve graph, and calculating experience accumulation probability and modulus coefficient of all the parts of the hydroelectric generating set;
and calculating the maximum amplitude value of each part of the hydroelectric generating set based on the modulus coefficient, and setting the alarm threshold value of each part of the hydroelectric generating set according to the maximum amplitude value.
In the embodiment, the probability function is adopted in the scheme, so that the Pearson III type frequency curve is used as the probability function suitable for the bias distribution, the monitoring signal can be better fitted, and the accuracy and the reliability of environmental assessment are improved because the monitoring data of the hydroelectric generating set have statistical characteristics obeying the bias distribution law instead of the complete normal distribution. Therefore, the proposal uses the calculation method of the Piercon III type frequency curve to determine the shutdown protection logic under different working conditions and ensure the accuracy of the output of the early warning signal,
s104, carrying out-of-limit alarm detection and shutdown protection judgment on real-time vibration monitoring values of all parts of the hydroelectric generating set based on the alarm threshold.
In the embodiment, out-of-limit alarm detection and shutdown protection judgment are carried out on real-time vibration monitoring values of all parts of the hydroelectric generating set based on an alarm threshold, and the method is specifically as follows:
i. out-of-limit alarm detection
If the real-time vibration monitoring value of each component of the hydroelectric generating set accords with the out-of-limit alarm condition, releasing a corresponding level alarm signal, wherein the alarm signal comprises a primary alarm signal and a secondary alarm signal;
determination conditions for issuing a primary alarm signal: the secondary alarm value of the working condition is more than or equal to the primary alarm value of the working condition, and the real-time vibration monitoring value is more than or equal to the primary alarm value of the working condition;
determination conditions for issuing a secondary alarm signal: the real-time vibration monitoring value under the working condition is more than or equal to the secondary alarm value of the working condition;
shutdown protection determination
When the hydroelectric generating set releases two or more secondary alarm signals, if the signals are positioned at two monitoring points of the same hydroelectric generating set component, the hydroelectric generating set is in shutdown protection, and if the signals are respectively positioned at the monitoring points of different components, the hydroelectric generating set is not in shutdown protection.
In this embodiment, the purpose of this is to set the primary and secondary alarm thresholds for the components of the hydroelectric generating set using the maximum amplitude value obtained by the probability function. When the primary and secondary alarm thresholds are used as the judgment standard, the probability function obtains the maximum amplitude value under different accumulated probabilities, and the advantage is that historical data are more fully considered. On the basis, a set of reasonable shutdown logic is established by setting the primary and secondary alarm thresholds by using the empirical cumulative probabilities of 0.1 and 0.01 as indexes, which is helpful for ensuring the accuracy of early warning signal output.
Referring to fig. 2, 3 and 4, the method for the division prediction and early warning of a hydroelectric generating set according to one of the embodiments comprises the following steps:
step one, arranging vibration sensors at all parts of the hydroelectric generating set, and collecting and uploading vibration data of monitoring points. The acquired runout data is time domain waveform data, as shown in fig. 3.
The invention is further provided with: the runout data of the monitoring points in the step 1) are respectively as follows: upper guide X-Y direction swing degree, lower guide X-Y direction swing degree, water guide X-Y direction swing degree,
the upper frame X-Y-Z direction swing degree, the stator frame X-Y direction swing degree, the lower frame X-Y-Z direction swing degree, the X-Y-Z direction swing degree at the top cover, the horizontal and vertical vibration of the stator core and key phase signals.
The vibration sensor for measuring the radial vibration (upper guide, lower guide and water guide) of the large shaft is an eddy current sensor, and the sensor for measuring the vibration of each supporting component (upper frame, lower frame and top cover) of the water turbine unit is a low-frequency speed sensor.
The invention is further provided with: the time domain waveform data in the step 1) refers to waveform of vibration signal changing along with time, and represents vibration data of normal operation of the hydroelectric generating set under each working condition, as shown in fig. 2.
And step two, calculating an average value and a standard deviation of each section of waveform data, and eliminating noise data of the monitoring points.
The invention is further provided with: the average value and standard deviation of each section of waveform data in the step 2) are calculated, noise data of monitoring points are removed, specifically,
2-1) taking the key phase signal as a starting point, selecting data containing n rotation periods as a calculation sample (n is more than or equal to 2).
2-2) calculating the average value mu and the standard deviation sigma of the sample, carrying out confidence analysis, and eliminating noise data exceeding [ mu-3 sigma, mu+3 sigma ] intervals in waveform sample data.
2-3) right shifting by 1 rotation period in the next calculation interval, obtaining the next data sample containing n rotation periods. And (3) repeating the step (2-2) to finally obtain the time domain waveform data with the data noise removed.
And thirdly, judging working conditions.
The invention is further provided with: the working condition judgment in the step 3) comprises water pumping starting working condition judgment, power generation starting working condition judgment, water pumping phase modulation working condition judgment, stop working condition judgment and steady-state working condition judgment. The current operation working condition states of the unit can be respectively obtained by a one-dimensional convolutional neural network method, namely a pumping start-up working condition, a generating start-up working condition, a pumping phase modulation working condition, a stopping working condition and a steady-state working condition, and specifically:
3-1) performing equal-length segmentation on the time domain waveform data obtained in the step two, dividing the time domain waveform data into a plurality of groups according to the rotation period, and calculating the time domain characteristic value of each group to be used as the input of the convolutional neural network model. The time domain eigenvalues include: maximum, maximum absolute, minimum, average, peak-to-peak, absolute average, root mean square, square root amplitude, standard deviation, kurtosis, skewness, margin, waveform, pulse, peak.
3-2) marking the time domain characteristic value to the corresponding working condition category, and storing the csv format as an experimental data set.
3-3) the experimental dataset is divided into a training set and a test set, wherein the training set and test set division ratio is 7:3.
3-4) building a one-dimensional convolutional neural network model. The network model structure is divided into 7 layers, specifically:
c1& C2: a one-dimensional convolution layer with a convolution kernel length of 64, and the activation function is a relu function;
and C3: a maximum pooling layer with a unit length of 2;
c4& C5: a one-dimensional convolution layer with a convolution kernel length of 256, and the activation function is a relu function;
c6: to prevent overfitting, a Dropout layer with a retention probability of 0.5 is set;
c7: the fully connected layer, the output layer uses Softmax activation functions.
3-5) inputting the training set into the model for training until the model converges and the performance meets the requirement, and obtaining the working condition judgment model based on the one-dimensional convolutional neural network.
3-6) collecting vibration data of the hydroelectric generating set monitored in real time, inputting the trained working condition judgment model in the step 3-5), and outputting a working condition judgment result.
And step four, calculating primary and secondary alarm thresholds of all parts of the hydroelectric generating set by using probability functions.
The invention is further provided with: in the step 4), the probability function is used to calculate the primary and secondary alarm thresholds of each component of the hydroelectric generating set, as shown in fig. 3, which means that the Pearson-iii probability function is used to draw a frequency curve of the working condition amplitude, and the primary and secondary alarm thresholds of each component of the vibration system of the hydroelectric generating set are determined, specifically:
4-1) under the judging working condition, selecting the vibration swing data of the hydroelectric generating set components to be arranged according to the descending order of amplitude values,
4-2) calculating the calculated average amplitude value of the calculated vibration and swing dataModulus coefficient K i Discrete coefficient C v The calculation formula is as follows:
in the formula (1), x i For the amplitude data of the i-th sequence number arranged in descending order,an average amplitude value that is the amplitude data;
in the formula (2), K i The modulus coefficient of the ith amplitude data is given, and n is the total number of the runout data.
4-3) setting a coefficient of deviation C s (0<C s Less than or equal to 3), respectively drawing a Pearson-III theoretical frequency curve graph, analyzing a theoretical frequency curve with the best fitting effect based on a least square method, and then obtaining a deviation coefficient C corresponding to the curve s As an estimate of the overall parameter.
4-4) calculating the empirical cumulative probability f of the amplitude data, wherein the calculation formula is as follows:
in the formula (3), m is the mth amplitude data arranged in descending order, and n is the total number of runout data.
4-5) according to the coefficient of deviation C s Consult PeaThe average value table of the rson-III curve is used for searching the average value when f is 0.1 and 0.01Calculating modulus coefficient K under different f values i The calculation formula is as follows:
4-6) based on the modulus factor K in step 4-5) i Calculating the corresponding maximum amplitude value P when f is 0.1 and 0.01 i The first-level alarm threshold value and the second-level alarm threshold value are set as judgment conditions for out-of-limit alarm detection. The calculation formula is as follows:
4-7) selecting horizontal or vibration signals of all parts of the hydroelectric generating set, repeating the steps 4-1) to 4-6), and respectively calculating the primary and secondary alarm thresholds of all parts of the split-working-condition hydroelectric generating set.
And fifthly, detecting out-of-limit alarm and judging shutdown protection.
The invention is further provided with: and 5) the out-of-limit alarm detection in the step is to release the alarm signal of the corresponding grade if the monitoring value of the runout system meets the out-of-limit alarm condition. Determination conditions for issuing a primary alarm signal: the second-level alarm value of the working condition is more than or equal to the first-level alarm value of the working condition; determination conditions for issuing a secondary alarm signal: the monitoring value of the runout system under the working condition is more than or equal to the secondary alarm value of the working condition.
The shutdown protection judgment in the step 5) means that when the unit releases two or more secondary alarm signals, if the signals are positioned at two monitoring points of the same component, shutdown protection logic is met, the shutdown protection signals are started, and the unit mechanically jumps. If the signals are respectively positioned at monitoring points of different components, the shutdown protection logic judgment is not satisfied.
Second embodiment
An early warning device for hydroelectric component working condition measurement based on probability function, based on the early warning method for hydroelectric component working condition measurement based on probability function according to one of the embodiments, comprises a collecting module, a neural network module, a probability function module and a judging module which are electrically connected in sequence:
the acquisition module is used for acquiring historical time domain waveform vibration data of each part of the hydroelectric generating set and eliminating noise data;
the neural network module is used for constructing and training a neural network model based on the historical time domain waveform vibration data and predicting the working condition judgment result of the hydroelectric generating set based on the trained neural network model;
the probability function module is used for calculating alarm thresholds of all parts of the hydroelectric generating set based on probability functions and working condition judgment results;
and the judging module is used for carrying out-of-limit alarm detection and shutdown protection judgment on the real-time vibration monitoring values of all parts of the hydroelectric generating set based on the alarm threshold.
In this embodiment, in order to better apply the method described in one of the embodiments, different modules are sequentially built by different steps, and each module is connected in series, so that the method can be more efficiently trained and used. It should be noted that the principle and effect of each step have been described above, and will not be explained here.
Third embodiment
An early warning device for hydroelectric component working condition measurement based on probability function, comprising:
at least one processor; the method comprises the steps of,
a memory communicatively coupled to the at least one processor; wherein,
the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of probability function based hydroelectric component operating profile pre-warning of one embodiment.
In the present embodiment, the above-described method is stored to the memory for better operation and processing of the method, and the stored method is executed by the processor. It should be noted that the principle and effect of each step have been described above, and will not be explained here.
Fourth embodiment
A computer readable storage medium storing a computer program which when executed by a processor implements the above-described method for early warning of hydroelectric component condition measurements based on probability functions.
In this embodiment, for better operation and use of the method described in one of the embodiments, the above method is stored in a computer-readable storage medium and implemented with a processor. It should be noted that the principle and effect of each step have been described above, and will not be explained here.
The foregoing description is only a partial embodiment of the present invention, and is not intended to limit the scope of the present invention, and all equivalent devices or equivalent processes using the descriptions and the drawings of the present invention or directly or indirectly applied to other related technical fields are included in the scope of the present invention.
Claims (9)
1. The utility model provides a hydroelectric component working condition measurement early warning method based on probability function which is characterized by comprising the following steps:
collecting historical time domain waveform vibration data of each part of the hydroelectric generating set, and eliminating noise data;
constructing and training a neural network model based on the historical time domain waveform vibration data, and predicting a working condition judgment result of the hydroelectric generating set based on the trained neural network model;
calculating alarm thresholds of all parts of the hydroelectric generating set based on probability functions and working condition judgment results;
and carrying out-of-limit alarm detection and shutdown protection judgment on real-time vibration monitoring values of all parts of the hydroelectric generating set based on the alarm threshold.
2. A method for pre-warning of hydroelectric component operating conditions based on a probability function according to claim 1,
collecting historical time domain waveform vibration data of each part of the hydroelectric generating set, and specifically:
measuring large-shaft radial time domain waveform vibration data by adopting an eddy current sensor;
and measuring time domain waveform vibration data of each supporting part of the water turbine set by adopting a low-frequency speed sensor.
3. A method for pre-warning of hydroelectric component operating conditions based on a probability function according to claim 1,
noise data is removed, specifically:
and calculating the average value and standard deviation of each section of historical time domain waveform vibration data, and utilizing the Laida criterion to propose the noise data contained in the historical time domain waveform vibration data.
4. A method for pre-warning of hydroelectric component operating conditions based on a probability function according to claim 1,
building and training a neural network model based on historical time domain waveform vibration data, and predicting a working condition judgment result of the hydroelectric generating set based on the trained neural network model, and specifically:
the time domain waveform data is subjected to equal-length segmentation, and divided into a plurality of groups according to the rotation period, and the time domain characteristic value of each group is calculated;
and taking the calculated time domain characteristic values as a training set and a testing set, constructing and training a neural network model, and predicting the working condition judgment result of the hydroelectric generating set by using the trained neural network model.
5. A method for pre-warning of hydroelectric component operating conditions based on a probability function according to claim 1,
calculating alarm thresholds of all parts of the hydroelectric generating set based on probability functions and working condition judgment results, and specifically:
based on the working condition judgment result, selecting time domain waveform vibration data of the hydroelectric generating set components, arranging the time domain waveform vibration data according to the descending order of amplitude values, and calculating the modulus coefficient and the discrete coefficient of each component of the hydroelectric generating set;
setting deviation coefficients of all parts of the hydroelectric generating set, respectively drawing a Pearson-III type theoretical frequency curve graph, respectively fitting a corresponding theoretical frequency curve according to the curve graph, and calculating experience accumulation probability and modulus coefficient of all the parts of the hydroelectric generating set;
and calculating the maximum amplitude value of each part of the hydroelectric generating set based on the modulus coefficient, and setting the alarm threshold value of each part of the hydroelectric generating set according to the maximum amplitude value.
6. A method for pre-warning a measured value of a hydroelectric component's condition based on a probability function as claimed in claim 5,
and carrying out-of-limit alarm detection and shutdown protection judgment on real-time vibration monitoring values of all parts of the hydroelectric generating set based on an alarm threshold, and specifically:
i. out-of-limit alarm detection
If the real-time vibration monitoring value of each component of the hydroelectric generating set accords with the out-of-limit alarm condition, releasing a corresponding level alarm signal, wherein the alarm signal comprises a primary alarm signal and a secondary alarm signal;
determination conditions for issuing a primary alarm signal: the secondary alarm value of the working condition is more than or equal to the primary alarm value of the working condition, and the real-time vibration monitoring value is more than or equal to the primary alarm value of the working condition;
determination conditions for issuing a secondary alarm signal: the real-time vibration monitoring value under the working condition is more than or equal to the secondary alarm value of the working condition;
shutdown protection determination
When the hydroelectric generating set releases two or more secondary alarm signals, if the signals are positioned at two monitoring points of the same hydroelectric generating set component, the hydroelectric generating set is in shutdown protection, and if the signals are respectively positioned at the monitoring points of different components, the hydroelectric generating set is not in shutdown protection.
7. The early warning device for the hydroelectric component working condition measurement based on the probability function is characterized by comprising an acquisition module, a neural network module, a probability function module and a judging module which are electrically connected in sequence, wherein the early warning device is based on the hydroelectric component working condition measurement based on the probability function, and the early warning device is based on any one of claims 1-6:
the acquisition module is used for acquiring historical time domain waveform vibration data of each part of the hydroelectric generating set and eliminating noise data;
the neural network module is used for constructing and training a neural network model based on the historical time domain waveform vibration data and predicting the working condition judgment result of the hydroelectric generating set based on the trained neural network model;
the probability function module is used for calculating alarm thresholds of all parts of the hydroelectric generating set based on probability functions and working condition judgment results;
and the judging module is used for carrying out-of-limit alarm detection and shutdown protection judgment on the real-time vibration monitoring values of all parts of the hydroelectric generating set based on the alarm threshold.
8. An early warning device for hydroelectric component working condition measurement based on probability function, comprising:
at least one processor; the method comprises the steps of,
a memory communicatively coupled to the at least one processor; wherein,
the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of pre-warning of probability function based hydroelectric component operating metrics as claimed in any one of claims 1 to 6.
9. A computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the method for early warning of hydroelectric component operating conditions measurement values based on a probability function according to any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311429993.XA CN117238113A (en) | 2023-10-31 | 2023-10-31 | Early warning method for hydroelectric component working condition measurement value based on probability function |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311429993.XA CN117238113A (en) | 2023-10-31 | 2023-10-31 | Early warning method for hydroelectric component working condition measurement value based on probability function |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117238113A true CN117238113A (en) | 2023-12-15 |
Family
ID=89088209
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311429993.XA Pending CN117238113A (en) | 2023-10-31 | 2023-10-31 | Early warning method for hydroelectric component working condition measurement value based on probability function |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117238113A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117691752A (en) * | 2024-02-01 | 2024-03-12 | 国网吉林省电力有限公司白山供电公司 | Automatic low-voltage station power failure alarm device with communication function |
-
2023
- 2023-10-31 CN CN202311429993.XA patent/CN117238113A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117691752A (en) * | 2024-02-01 | 2024-03-12 | 国网吉林省电力有限公司白山供电公司 | Automatic low-voltage station power failure alarm device with communication function |
| CN117691752B (en) * | 2024-02-01 | 2024-04-26 | 国网吉林省电力有限公司白山供电公司 | Automatic low-voltage station power failure alarm device with communication function |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108375476B (en) | Hydroelectric generating set health assessment method | |
| CN111539553B (en) | Wind turbine generator fault early warning method based on SVR algorithm and off-peak degree | |
| CN108069308A (en) | A kind of electric staircase failure diagnosis method based on sequential probability | |
| CN111709465B (en) | Intelligent identification method for rough difference of dam safety monitoring data | |
| CN117238113A (en) | Early warning method for hydroelectric component working condition measurement value based on probability function | |
| CN110596506A (en) | Converter fault diagnosis method based on time convolution network | |
| CN110110768A (en) | Fault Diagnosis of Roller Bearings based on Concurrent Feature study and multi-categorizer | |
| CN103103570B (en) | Based on the aluminium cell condition diagnostic method of pivot similarity measure | |
| CN108802535B (en) | Screening method, main interference source identification method and device, server and storage medium | |
| CN103034170A (en) | Numerical control machine tool machining performance prediction method based on intervals | |
| CN116380445B (en) | Equipment state diagnosis method and related device based on vibration waveform | |
| CN111504647A (en) | AR-MSET-based performance degradation evaluation method for rolling bearing | |
| CN105930900A (en) | Method and system for predicting hybrid wind power generation | |
| CN111458149A (en) | Method and system for predicting performance and service life of rolling bearing | |
| CN115614292A (en) | Vibration monitoring device and method for vertical water pump unit | |
| CN114330096B (en) | Rail transit motor bearing fault degree classification model construction method based on deep learning | |
| CN114757365A (en) | High-speed railway roadbed settlement prediction and early warning method based on deep learning | |
| CN113237619B (en) | Fault early warning method, device, equipment and storage medium for variable-speed rotating machinery vibration | |
| CN114881087A (en) | Building robot bearing performance degradation assessment method | |
| CN111291918A (en) | Rotating machine degradation trend prediction method based on stationary subspace exogenous vector autoregression | |
| CN113377750B (en) | Hydrological data cleaning method and system | |
| CN105224733B (en) | The method that wind power abandons wind data feature recognition | |
| CN113742993A (en) | Method, device, equipment and storage medium for predicting life loss of dry-type transformer | |
| CN112926656A (en) | Method, system and equipment for predicting state of circulating water pump of nuclear power plant | |
| CN110826600B (en) | Engine surge prediction method based on adaptive resonance network online incremental learning |
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 |