CN116578889A - Power generation fault diagnosis method - Google Patents
Power generation fault diagnosis method Download PDFInfo
- Publication number
- CN116578889A CN116578889A CN202310789953.XA CN202310789953A CN116578889A CN 116578889 A CN116578889 A CN 116578889A CN 202310789953 A CN202310789953 A CN 202310789953A CN 116578889 A CN116578889 A CN 116578889A
- Authority
- CN
- China
- Prior art keywords
- data
- fault
- fault diagnosis
- network
- sample data
- 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
- 238000003745 diagnosis Methods 0.000 title claims abstract description 110
- 238000000034 method Methods 0.000 title claims abstract description 52
- 238000010248 power generation Methods 0.000 title claims abstract description 16
- 230000009471 action Effects 0.000 claims description 27
- 238000012549 training Methods 0.000 claims description 26
- 238000004422 calculation algorithm Methods 0.000 claims description 13
- 238000004364 calculation method Methods 0.000 claims description 12
- 230000008569 process Effects 0.000 claims description 8
- 230000007246 mechanism Effects 0.000 claims description 6
- 230000004907 flux Effects 0.000 claims description 3
- 230000002787 reinforcement Effects 0.000 claims description 3
- 230000008859 change Effects 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005312 nonlinear dynamic Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F03—MACHINES OR ENGINES FOR LIQUIDS; WIND, SPRING, OR WEIGHT MOTORS; PRODUCING MECHANICAL POWER OR A REACTIVE PROPULSIVE THRUST, NOT OTHERWISE PROVIDED FOR
- F03D—WIND MOTORS
- F03D17/00—Monitoring or testing of wind motors, e.g. diagnostics
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Software Systems (AREA)
- Mechanical Engineering (AREA)
- Combustion & Propulsion (AREA)
- Chemical & Material Sciences (AREA)
- Sustainable Energy (AREA)
- Sustainable Development (AREA)
- Medical Informatics (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention discloses a power generation fault diagnosis method, which comprises the steps of collecting data, establishing a wind driven generator system model, establishing a fault diagnosis model, determining whether a fault exists, confirming the fault type and perfecting the fault diagnosis model. The invention belongs to the technical field of fault diagnosis, in particular to a power generation fault diagnosis method, which solves the problems that a fault diagnosis model cannot identify fault type data, the diagnosis efficiency is low, the diagnosis error rate is high and the sample number is insufficient due to the fact that the diagnosis method is too single.
Description
Technical Field
The invention relates to the technical field of fault diagnosis, in particular to a power generation fault diagnosis method.
Background
At present, wind energy becomes clean energy with huge development potential due to the occurrence of wind power generation technology, but because the working environment of a wind generating set is generally bad, a wind generating set is easy to fail due to various disturbances, the disturbance of the wind generating set is divided into internal and external disturbances, the internal disturbances mainly come from the aspects of parameter change, control coupling, current harmonic waves, modeling errors, nonlinear dynamics of factories and the like, the external disturbances mainly come from the aspects of wind speed change, torsional vibration, wind flow asymmetry and the like, the traditional power generation fault diagnosis method is based on a neural network and signal processing, but the problem that a fault diagnosis model cannot identify data of a fault type exists, the problem that the diagnosis efficiency is low and the diagnosis error rate is high due to the fact that the diagnosis method is too single, and the problem that the sample number is insufficient are solved.
Disclosure of Invention
Aiming at the situation, in order to overcome the defects of the prior art, the invention provides a power generation fault diagnosis method, and the method for establishing the combination of a wind driven generator system model and a fault diagnosis model is adopted to diversify the diagnosis method, improve the diagnosis efficiency and reduce the diagnosis error rate aiming at the problems of low diagnosis efficiency and high diagnosis error rate caused by the fact that the diagnosis method is too single; aiming at the problem of insufficient sample number, the scheme adopts an experience playback mechanism, takes training data as sample data to participate in training so as to increase the sample data number; aiming at the problem that fault data of which the fault type cannot be identified by the fault diagnosis model exist, the method adopts a multidimensional clustering algorithm to cluster sample data and real-time fault data of which the fault type cannot be identified by the fault diagnosis model, continuously adjusts an initial cluster center set, and selects the fault type with the highest clustering accuracy as the fault type corresponding to the operation fault data to be output.
The technical scheme adopted by the invention is as follows: the invention provides a power generation fault diagnosis method, which comprises the following steps:
step S1: collecting data, wherein the data comprise historical fault data and wind turbine system real-time operation data, and the historical fault data comprise fault types, a plurality of parameters and numerical values corresponding to the parameters;
step S2: establishing a wind driven generator system model;
step S3: establishing a fault diagnosis model;
step S4: determining whether a fault exists;
step S5: confirming the fault type;
step S6: and (5) perfecting a fault diagnosis model.
Further, in step S2, a rotor radius r and an polar logarithm n are preset F The specific method for establishing the wind driven generator system model comprises the following steps of:
step S21: the ratio μ of the shaft currents is calculated as follows:
；
where v is wind speed and δ is rotor speed;
step S22: the electromagnetic torque M is calculated as follows:
；
where τ is the flux linkage, i q Q-axis current, which is Dq-axis current;
step S23: the power F absorbed by the wind wheel is calculated, and the calculation formula is as follows:
；
where the parameter ε is the pitch angle, G is the moment of inertia,the method is to derive the rotation speed delta of the rotor;
step S24: the mechanical torque J is calculated as follows:
。
further, in step S3, the fault diagnosis model is initialized, the collected historical fault data is used as sample data by a fault diagnosis model building algorithm based on reinforcement learning, the sample data is randomly selected to train the P network and update the P network, an experience playback mechanism is adopted in the training process, and the data generated in the training process is used as sample data to participate in training, and the specific method comprises the following steps:
step S31: initializing a P network, namely initializing an n+3-dimensional P network based on collected historical fault data, wherein the first n-dimensional of the P network represents the data state of n parameters, the n+1-th dimension represents selected actions, the actions comprise normal data, unrecognizable data of type A, type B, type C, type D and type E, the n+2-th dimension represents the corresponding P value of the P network in different states, the n+3-th dimension represents rewards R obtained by selecting different actions in the current state, and the R is the largest when the action selection is correct;
step S32: updating the P network, wherein the calculation formula of the P network is as follows:
；
where s is the learning rate, k is the discount factor, P (z, d) is the P value corresponding to the updated P network when d is selected in z, P (z, d) is the P value corresponding to the original P network when d is selected in z, maxP (z, d) is the original P network when d is selected in zThe corresponding maximum P value in the actions which can be selected in the state;
step S33: given a loss threshold, calculating a loss L, wherein a calculation formula of the loss L is as follows:
；
wherein n is the number of actions that can be selected in the current state, R i Is the prize R corresponding to each action in the current state,the average value of rewards corresponding to all actions in the current state is obtained, and if L is lower than a loss threshold value, training in the current state is finished; if L is not lower than the loss threshold value, the current state training is continued, when L is 0, the accuracy of the fault diagnosis model identification sample data reaches 100%, and when L is not 0, the fault diagnosis model identification sample data has errors;
step S34: and (3) performing iterative training, training all states, and completing the establishment of the fault diagnosis model when the loss L in all states is lower than a loss threshold value.
Further, in step S4, the determining whether to fail sets a threshold value of each parameter of the wind turbine system model in advance, and based on the threshold value of each parameter, determining whether the wind turbine system real-time operation data is operation failure data.
Further, in step S5, the fault type is identified by using a fault diagnosis model, and if the identification is successful, the fault type is directly output; if the failure is not identified, the operation fault data and the collected historical fault data are used as sample data sets to be clustered by using a multidimensional clustering algorithm, an initial clustering center set is continuously adjusted, and the fault type with the highest clustering accuracy is selected to be output as the fault type corresponding to the operation fault data; the multidimensional clustering algorithm pre-acquires a sample data set, wherein the sample data set is X= { X 1 ，x 2 ，x 3 ，…，x n X, where x 1 ，x 2 ，x 3 ，…，x n Representing the 1 st, 2 nd, 3 rd, … th, n sample data in the sample data set X, a specific method comprises the steps of:
step S51: presetting a clustering ending threshold, and randomly selecting k sample data from a sample data set X as an initial clustering center set, wherein the initial clustering center set is C= { C 1 ，c 2 ，c 3 ，…，c k And (c), where c 1 ，c 2 ，c 3 ，…，c k Representing the 1 st, 2 nd, 3 rd, … th and k th initial cluster centers in the initial cluster center set C;
step S52: distributing data in all sample data sets to the periphery of cluster center data closest to the data to obtain k clusters; if the distribution times are more than two times, judging whether clusters obtained in two adjacent times are all the same, and if so, ending the clustering; if not, go to step S53;
step S53: calculating loss T, and if the loss T is lower than a set threshold value, ending clustering; if not, go to step S54; the calculation formula of the loss T is as follows:
；
wherein x is i Is the data in the sample data set, J xi Is x i Corresponding cluster center data;
step S54: for each cluster, taking the point with the smallest distance average value of the rest points in the cluster as a new cluster center;
step S55: go to step S52.
Further, in step S6, the perfect fault diagnosis model returns the operation fault data and the fault type corresponding to the operation fault data to the fault diagnosis model, so as to continuously perfect the fault diagnosis model.
By adopting the scheme, the beneficial effects obtained by the invention are as follows:
(1) Aiming at the problems of low diagnosis efficiency and high diagnosis error rate caused by the fact that the traditional diagnosis method is too single, the method for combining the wind driven generator system model and the fault diagnosis model is adopted, so that the diagnosis methods are diversified, the diagnosis efficiency is improved, and meanwhile, the diagnosis error rate is reduced;
(2) Aiming at the problem of insufficient sample number, the scheme adopts an experience playback mechanism, takes training data as sample data to participate in training so as to increase the sample data number;
(3) Aiming at the problem that fault data of which the fault type cannot be identified by the fault diagnosis model exist, the method adopts a multidimensional clustering algorithm to cluster the sample data and the fault data of which the fault type cannot be identified by the fault diagnosis model, continuously adjusts an initial cluster center set, and selects the fault type with the highest clustering accuracy as the fault type corresponding to the operation fault data to be output.
Drawings
FIG. 1 is a schematic flow chart of a power generation fault diagnosis method provided by the invention;
FIG. 2 is a flow chart of step S2;
FIG. 3 is a flow chart of step S3;
FIG. 4 is a flow diagram of a multidimensional clustering algorithm;
FIG. 5 is a graph showing the comparison of the diagnosis times of the conventional fault diagnosis model and the fault diagnosis model according to the present embodiment;
fig. 6 is a graph showing the comparison of the number of diagnostic errors between the conventional fault diagnosis model and the fault diagnosis model according to the present embodiment.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are only some, but not all embodiments of the invention; all other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Referring to fig. 1, fig. 2, fig. 3 and fig. 4, the method for diagnosing power generation faults provided by the present invention includes the following steps:
step S1: and collecting data, wherein the data comprise historical fault data and real-time operation data of the wind turbine system, and the historical fault data comprise fault types, a plurality of parameters and numerical values corresponding to the parameters.
Step S2: preset wind wheel radius r and polar logarithm n F The specific method for establishing the wind driven generator system model comprises the following steps of:
step S21: the ratio μ of the shaft currents is calculated as follows:
；
where v is wind speed and δ is rotor speed;
step S22: the electromagnetic torque M is calculated as follows:
；
where τ is the flux linkage, i q Q-axis current, which is Dq-axis current;
step S23: the power F absorbed by the wind wheel is calculated, and the calculation formula is as follows:
；
where the parameter ε is the pitch angle, G is the moment of inertia,the method is to derive the rotation speed delta of the rotor;
step S24: the mechanical torque J is calculated as follows:
。
step S3: the method comprises the steps of establishing a fault diagnosis model, namely initializing a P network, taking collected historical fault data as sample data through a fault diagnosis model establishment algorithm based on reinforcement learning, randomly selecting the sample data to train the P network and updating the P network, wherein an experience playback mechanism is adopted in the training process, and the data generated in the training process are taken as the sample data to participate in training, and the specific method comprises the following steps of:
step S31: initializing a P network, namely initializing an n+3-dimensional P network based on collected historical fault data, wherein the first n-dimensional of the P network represents the data state of n parameters, the n+1-th dimension represents selected actions, the actions comprise normal data, unrecognizable data of type A, type B, type C, type D and type E, the n+2-th dimension represents the corresponding P value of the P network in different states, the n+3-th dimension represents rewards R obtained by selecting different actions in the current state, and the R is the largest when the action selection is correct;
step S32: updating the P network, wherein the calculation formula of the P network is as follows:
；
where s is the learning rate, k is the discount factor, P (z, d) is the P value corresponding to the updated P network when d is selected in z, P (z, d) is the P value corresponding to the original P network when d is selected in z, maxP (z, d) is the original P network when d is selected in zThe corresponding maximum P value in the actions which can be selected in the state;
step S33: given a loss threshold, calculating a loss L, wherein a calculation formula of the loss L is as follows:
；
wherein n is the number of actions that can be selected in the current state, R i Is the prize R corresponding to each action in the current state,the average value of rewards corresponding to all actions in the current state is obtained, and if L is lower than a loss threshold value, training in the current state is finished; if L is not lower than the loss threshold value, the current state training is continued, when L is 0, the accuracy of the fault diagnosis model identification sample data reaches 100%, and when L is not 0, the fault diagnosis model identification sample data has errors;
step S34: and (3) performing iterative training, training all states, and completing the establishment of the fault diagnosis model when the loss L in all states is lower than a loss threshold value.
In the operation, aiming at the problems of low diagnosis efficiency and high diagnosis error rate caused by the fact that the traditional diagnosis method is too single, a method for establishing a combination of a wind driven generator system model and a fault diagnosis model is adopted, so that the diagnosis efficiency is improved in a diversified manner, and meanwhile, the diagnosis error rate is reduced; aiming at the problem of insufficient sample number, the scheme adopts an experience playback mechanism, takes training data as sample data to participate in training, and increases the sample data number.
Step S4: determining whether the wind driven generator system is faulty, presetting threshold values of all parameters of the wind driven generator system model, and judging whether the real-time operation data of the wind driven generator system is the operation fault data based on the threshold values of all the parameters.
Step S5: confirming a fault type, wherein the fault type is that operating fault data are identified by using a fault diagnosis model, and if the identification is successful, the fault type is directly output; if the failure is not identified, the operation fault data and the collected historical fault data are used as sample data sets to be clustered by using a multidimensional clustering algorithm, an initial clustering center set is continuously adjusted, and the fault type with the highest clustering accuracy is selected to be output as the fault type corresponding to the operation fault data; the multidimensional clustering algorithm pre-acquires a sample data set, wherein the sample data set is X= { X 1 ，x 2 ，x 3 ，…，x n X, where x 1 ，x 2 ，x 3 ，…，x n Representing the 1 st, 2 nd, 3 rd, … th, n sample data in the sample data set X, a specific method comprises the steps of:
step S51: presetting a clustering ending threshold, and randomly selecting k sample data from a sample data set X as an initial clustering center set, wherein the initial clustering center set is C= { C 1 ，c 2 ，c 3 ，…，c k And (c), where c 1 ，c 2 ，c 3 ，…，c k Representing the 1 st, 2 nd, 3 rd, … th and k th initial cluster centers in the initial cluster center set C;
step S52: distributing data in all sample data sets to the periphery of cluster center data closest to the data to obtain k clusters; if the distribution times are more than two times, judging whether clusters obtained in two adjacent times are all the same, and if so, ending the clustering; if not, go to step S53;
step S53: calculating loss T, and if the loss T is lower than a set threshold value, ending clustering; if not, go to step S54; the calculation formula of the loss T is as follows:
；
wherein x is i Is the data in the sample data set, J xi Is x i Corresponding cluster center data;
step S54: for each cluster, taking the point with the smallest distance average value of the rest points in the cluster as a new cluster center;
step S55: go to step S52.
In the operation, aiming at the problem that the fault diagnosis model cannot identify the fault data of the fault type, the method adopts a multidimensional clustering algorithm to cluster the sample data and the fault data of the fault type which cannot be identified by the fault diagnosis model, continuously adjusts an initial cluster center set, and selects the fault type with the highest clustering accuracy as the fault type corresponding to the operation fault data to output.
S6: and (3) perfecting the fault diagnosis model, wherein the perfecting the fault diagnosis model is to return the operation fault data and the fault types corresponding to the operation fault data to the fault diagnosis model, and continuously perfecting the fault diagnosis model.
Referring to fig. 5, 1000 sample data are randomly selected and respectively input into the conventional fault diagnosis model and the fault diagnosis model according to the scheme, 200 sample data are added each time to continuously input two fault diagnosis models, the diagnosis time of each fault diagnosis model is recorded, the difference between the diagnosis time of the two fault diagnosis models is not large when the number of the sample data is 1000 to 2000, and the diagnosis time of the conventional fault diagnosis model is obviously higher than that of the fault diagnosis model according to the scheme when the number of the sample data exceeds 2000.
Referring to fig. 6, 1000 sample data are randomly selected and respectively input into the conventional fault diagnosis model and the fault diagnosis model according to the scheme, 200 sample data are added each time to continuously input two fault diagnosis models, the number of the sample data of the fault diagnosis errors of each fault diagnosis model is recorded, the difference between the number of the sample data of the fault diagnosis errors of the two fault diagnosis models is not large when the number of the sample data is 1000 to 2200, and the number of the sample data of the fault diagnosis errors of the conventional fault diagnosis model is obviously higher than the number of the sample data of the fault diagnosis errors of the fault diagnosis model according to the scheme when the number of the sample data exceeds 2200.
It is noted that relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process or method that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process or method.
Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made therein without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
The invention and its embodiments have been described above with no limitation, and the actual construction is not limited to the embodiments of the invention as shown in the drawings. In summary, if one of ordinary skill in the art is informed by this disclosure, a structural manner and an embodiment similar to the technical solution should not be creatively devised without departing from the gist of the present invention.
Claims (5)
1. A power generation fault diagnosis method is characterized in that: the method comprises the following steps: step S1: collecting data, wherein the data comprise historical fault data and wind turbine system real-time operation data, and the historical fault data comprise fault types, a plurality of parameters and numerical values corresponding to the parameters;
step S2: establishing a wind driven generator system model;
step S3: establishing a fault diagnosis model;
step S4: determining whether a fault exists;
step S5: confirming the fault type;
step S6: perfecting a fault diagnosis model;
in step S2, the radius r and the pole pair number n of the wind wheel are preset F The specific method for establishing the wind driven generator system model comprises the following steps of:
step S21: the ratio μ of the shaft currents is calculated as follows:
；
where v is wind speed and δ is rotor speed;
step S22: the electromagnetic torque M is calculated as follows:
；
where τ is the flux linkage, i q Q-axis current, which is Dq-axis current;
step S23: the power F absorbed by the wind wheel is calculated, and the calculation formula is as follows:
；
where the parameter ε is the pitch angle, G is the moment of inertia,the method is to derive the rotation speed delta of the rotor;
step S24: the mechanical torque J is calculated as follows:
；
in step S3, the fault diagnosis model is initialized to the P network, the collected historical fault data is used as sample data by a fault diagnosis model building algorithm based on reinforcement learning, the sample data is randomly selected to train the P network and update the P network, the training process adopts an experience playback mechanism, and the data generated in the training process is used as sample data to participate in training, and the specific method comprises the following steps:
step S31: initializing a P network, namely initializing an n+3-dimensional P network based on collected historical fault data, wherein the first n-dimensional of the P network represents the data state of n parameters, the n+1-th dimension represents selected actions, the actions comprise normal data, unrecognizable data of type A, type B, type C, type D and type E, the n+2-th dimension represents the corresponding P value of the P network in different states, the n+3-th dimension represents rewards R obtained by selecting different actions in the current state, and the R is the largest when the action selection is correct;
step S32: updating the P network, wherein the calculation formula of the P network is as follows:
；
where s is the learning rate, k is the discount factor, P (z, d) is the P value corresponding to the updated P network when d is selected in z, P (z, d) is the P value corresponding to the original P network when d is selected in z, maxP (z, d) is the original P network when d is selected in zThe corresponding maximum P value in the actions which can be selected in the state;
step S33: given a loss threshold, calculating a loss L, wherein a calculation formula of the loss L is as follows:
；
wherein n is the number of actions that can be selected in the current state, R i Is the prize R corresponding to each action in the current state,the average value of rewards corresponding to all actions in the current state is obtained, and if L is lower than a loss threshold value, training in the current state is finished; if L is not lower than the loss threshold value, the current state training is continued, when L is 0, the accuracy of the fault diagnosis model identification sample data reaches 100%, and when L is not 0, the fault diagnosis model identification sample data has errors;
step S34: iterative training is carried out on all states, and when the loss L in all states is lower than a loss threshold value, the fault diagnosis model establishment is completed;
in step S5, the multidimensional clustering algorithm pre-acquires a sample dataset, where the sample dataset is x= { X 1 ，x 2 ，x 3 ，…，x n X, where x 1 ，x 2 ，x 3 ，…，x n Representing the 1 st, 2 nd, 3 rd, … th, n sample data in the sample data set X, a specific method comprises the steps of:
step S51:randomly selecting k sample data from a sample data set X as an initial clustering center set, wherein the initial clustering center set is C= { C 1 ，c 2 ，c 3 ，…，c k And (c), where c 1 ，c 2 ，c 3 ，…，c k Representing the 1 st, 2 nd, 3 rd, … th and k th initial cluster centers in the initial cluster center set C;
step S52: distributing data in all sample data sets to the periphery of cluster center data closest to the data to obtain k clusters; if the distribution times are more than two times, judging whether clusters obtained in two adjacent times are all the same, and if so, ending the clustering; if not, go to step S53;
step S53: calculating loss T, and if the loss T is lower than a set threshold value, ending clustering; if not, go to step S54; the calculation formula of the loss T is as follows:
；
wherein x is i Is the data in the sample data set, J xi Is x i Corresponding cluster center data;
step S54: for each cluster, taking the point with the smallest distance average value of the rest points in the cluster as a new cluster center;
step S55: go to step S52.
2. The power generation failure diagnosis method according to claim 1, characterized in that: in step S5, the fault type is identified by using a fault diagnosis model, and if the identification is successful, the fault type is directly output; if the failure is not identified, the operation fault data and the collected historical fault data are used as sample data sets to be clustered by using a multidimensional clustering algorithm, an initial clustering center set is continuously adjusted, and the fault type with the highest clustering accuracy is selected to be output as the fault type corresponding to the operation fault data.
3. The power generation failure diagnosis method according to claim 1, characterized in that: in step S1, the collected data includes historical fault data and wind turbine system real-time operation data, where the historical fault data includes a fault type, a plurality of parameters, and values corresponding to the parameters.
4. The power generation failure diagnosis method according to claim 1, characterized in that: in step S4, the determining whether to fail sets a threshold value of each parameter of the wind turbine system model in advance, and based on the threshold value of each parameter, determining whether the wind turbine system real-time operation data is operation failure data.
5. The power generation failure diagnosis method according to claim 1, characterized in that: in step S6, the perfect fault diagnosis model returns the operation fault data and the fault type corresponding to the operation fault data to the fault diagnosis model, so as to perfect the fault diagnosis model continuously.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310789953.XA CN116578889B (en) | 2023-06-30 | 2023-06-30 | Power generation fault diagnosis method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310789953.XA CN116578889B (en) | 2023-06-30 | 2023-06-30 | Power generation fault diagnosis method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116578889A true CN116578889A (en) | 2023-08-11 |
| CN116578889B CN116578889B (en) | 2023-11-10 |
Family
ID=87536044
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310789953.XA Active CN116578889B (en) | 2023-06-30 | 2023-06-30 | Power generation fault diagnosis method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116578889B (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108490353A (en) * | 2018-05-07 | 2018-09-04 | 东南大学 | Multiphase permanent magnet synchronous motor drive system method for diagnosing faults |
| CN109840548A (en) * | 2019-01-02 | 2019-06-04 | 广东工业大学 | One kind being based on BP neural network Diagnosis Method of Transformer Faults |
| CN111199252A (en) * | 2019-12-30 | 2020-05-26 | 广东电网有限责任公司 | Fault diagnosis method for intelligent operation and maintenance system of power communication network |
| CN111259927A (en) * | 2020-01-08 | 2020-06-09 | 西北工业大学 | Rocket engine fault diagnosis method based on neural network and evidence theory |
| CN111967618A (en) * | 2019-05-20 | 2020-11-20 | 武汉剑心科技有限公司 | Online diagnosis method for voltage regulator based on deep learning |
| CN113095367A (en) * | 2021-03-16 | 2021-07-09 | 华中科技大学 | Compressor fault diagnosis method based on deep reinforcement learning |
| CN113887729A (en) * | 2021-09-29 | 2022-01-04 | 广东电网有限责任公司电力调度控制中心 | Fault diagnosis method for low-voltage power line carrier communication system |
| CN114091553A (en) * | 2020-08-06 | 2022-02-25 | 长沙理工大学 | Diagnosis method for rolling bearing fault |
| CN115828745A (en) * | 2022-11-29 | 2023-03-21 | 西北工业大学 | Electromechanical actuator turn-to-turn short circuit fault diagnosis method based on integrated learning algorithm |
-
2023
- 2023-06-30 CN CN202310789953.XA patent/CN116578889B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108490353A (en) * | 2018-05-07 | 2018-09-04 | 东南大学 | Multiphase permanent magnet synchronous motor drive system method for diagnosing faults |
| CN109840548A (en) * | 2019-01-02 | 2019-06-04 | 广东工业大学 | One kind being based on BP neural network Diagnosis Method of Transformer Faults |
| CN111967618A (en) * | 2019-05-20 | 2020-11-20 | 武汉剑心科技有限公司 | Online diagnosis method for voltage regulator based on deep learning |
| CN111199252A (en) * | 2019-12-30 | 2020-05-26 | 广东电网有限责任公司 | Fault diagnosis method for intelligent operation and maintenance system of power communication network |
| CN111259927A (en) * | 2020-01-08 | 2020-06-09 | 西北工业大学 | Rocket engine fault diagnosis method based on neural network and evidence theory |
| CN114091553A (en) * | 2020-08-06 | 2022-02-25 | 长沙理工大学 | Diagnosis method for rolling bearing fault |
| CN113095367A (en) * | 2021-03-16 | 2021-07-09 | 华中科技大学 | Compressor fault diagnosis method based on deep reinforcement learning |
| CN113887729A (en) * | 2021-09-29 | 2022-01-04 | 广东电网有限责任公司电力调度控制中心 | Fault diagnosis method for low-voltage power line carrier communication system |
| CN115828745A (en) * | 2022-11-29 | 2023-03-21 | 西北工业大学 | Electromechanical actuator turn-to-turn short circuit fault diagnosis method based on integrated learning algorithm |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116578889B (en) | 2023-11-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112217202B (en) | Distributed new energy, energy storage and power distribution network planning method considering flexibility investment | |
| CN106682814B (en) | Wind turbine generator fault intelligent diagnosis method based on fault knowledge base | |
| CN108304623B (en) | Probability load flow online calculation method based on stack noise reduction automatic encoder | |
| CN110009135B (en) | Wind power prediction method based on width learning | |
| WO2021109644A1 (en) | Hybrid vehicle working condition prediction method based on meta-learning | |
| CN109255477B (en) | Wind speed prediction method based on depth limit learning machine, system and unit thereof | |
| CN111079343B (en) | Wind turbine generator effective wind speed estimation method based on width learning | |
| CN107045584A (en) | A kind of power frequency vibration abnormal failure diagnostic method suitable for water pump rotor system | |
| CN113988481B (en) | Wind power prediction method based on dynamic matrix prediction control | |
| CN110544184A (en) | fan performance analysis method based on operation parameters | |
| CN115146718A (en) | Depth representation-based wind turbine generator anomaly detection method | |
| CN116578889B (en) | Power generation fault diagnosis method | |
| CN111027229A (en) | Wind power curve fitting method based on sparse heteroscedastic multi-spline regression | |
| CN108563829B (en) | Multi-step wind speed forecasting method based on Bayes robust function regression | |
| CN114548444A (en) | Method and system for optimizing overall overhaul process of offshore wind turbine generator | |
| CN110674893B (en) | Self-adaptive correction method for diagnosis experience in rotary machine fault diagnosis knowledge base | |
| CN111736574B (en) | Universal thermal power plant fault diagnosis system and diagnosis method thereof | |
| CN114757111B (en) | Intelligent equipment health management method based on digital twinning technology | |
| CN114511158B (en) | Wind turbine power prediction method based on wake deflection effect and 2DJensen model | |
| CN115578084A (en) | Wind turbine generator set frequency converter fault early warning method based on deep convolution self-encoder | |
| CN114065618A (en) | Method for fitting fan power curve based on parameter model of differential evolution | |
| CN113991752A (en) | Power grid quasi-real-time intelligent control method and system | |
| Kumar et al. | Performance enhancement of wind power forecast model using novel pre‐processing strategy and hybrid optimization approach | |
| CN113609758B (en) | Power prediction method for newly-built wind power plant | |
| CN113031441B (en) | Rotary mechanical diagnosis network automatic search method based on reinforcement 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 | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20230920 Address after: 730050 no.628, Xijin East Road, Qilihe district, Lanzhou City, Gansu Province Applicant after: ECONOMY AND TECHNOLOGY Research Institute GANSU ELECTRIC POWER CORPORATION STATE GRID Address before: Room 1101, 1102, 1103, 1104, 1105, Building D, No.1 Gonghua Road, Huayuan Industrial Zone, Binhai New Area, Tianjin, 300192 Applicant before: GUOHENG NENGYUAN (TIANJIN) ELECTRIC POWER TECHNOLOGY DEVELOPMENT CO.,LTD. |
|
| TA01 | Transfer of patent application right | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |