CN112648140B - Fault tolerance method for wind turbine generator pitch angle encoder based on signal reconstruction - Google Patents
Fault tolerance method for wind turbine generator pitch angle encoder based on signal reconstruction Download PDFInfo
- Publication number
- CN112648140B CN112648140B CN202011522498.XA CN202011522498A CN112648140B CN 112648140 B CN112648140 B CN 112648140B CN 202011522498 A CN202011522498 A CN 202011522498A CN 112648140 B CN112648140 B CN 112648140B
- Authority
- CN
- China
- Prior art keywords
- pitch angle
- encoder
- neural network
- network model
- value
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 48
- 238000003062 neural network model Methods 0.000 claims abstract description 58
- 238000012546 transfer Methods 0.000 claims abstract description 29
- 238000012795 verification Methods 0.000 claims abstract description 17
- 230000007246 mechanism Effects 0.000 claims abstract description 15
- 238000007781 pre-processing Methods 0.000 claims abstract description 8
- 238000013528 artificial neural network Methods 0.000 claims description 14
- 238000012545 processing Methods 0.000 claims description 9
- 238000001914 filtration Methods 0.000 claims description 6
- 238000012549 training Methods 0.000 claims description 6
- 238000005457 optimization Methods 0.000 claims description 4
- 239000002245 particle Substances 0.000 claims description 4
- 230000009467 reduction Effects 0.000 claims description 4
- 230000001537 neural effect Effects 0.000 claims description 3
- 239000013589 supplement Substances 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 2
- 238000005259 measurement Methods 0.000 description 12
- 238000010248 power generation Methods 0.000 description 6
- 230000008569 process Effects 0.000 description 5
- 239000004576 sand Substances 0.000 description 5
- 238000012360 testing method Methods 0.000 description 5
- 230000008859 change Effects 0.000 description 4
- 238000011161 development Methods 0.000 description 2
- 238000003745 diagnosis Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 238000013178 mathematical model Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000010355 oscillation Effects 0.000 description 2
- 238000013461 design Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- PXUQTDZNOHRWLI-OXUVVOBNSA-O malvidin 3-O-beta-D-glucoside Chemical compound COC1=C(O)C(OC)=CC(C=2C(=CC=3C(O)=CC(O)=CC=3[O+]=2)O[C@H]2[C@@H]([C@@H](O)[C@H](O)[C@@H](CO)O2)O)=C1 PXUQTDZNOHRWLI-OXUVVOBNSA-O 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
Images
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
- F03D7/00—Controlling wind motors
-
- 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
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F05—INDEXING SCHEMES RELATING TO ENGINES OR PUMPS IN VARIOUS SUBCLASSES OF CLASSES F01-F04
- F05B—INDEXING SCHEME RELATING TO WIND, SPRING, WEIGHT, INERTIA OR LIKE MOTORS, TO MACHINES OR ENGINES FOR LIQUIDS COVERED BY SUBCLASSES F03B, F03D AND F03G
- F05B2270/00—Control
- F05B2270/30—Control parameters, e.g. input parameters
- F05B2270/328—Blade pitch angle
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F05—INDEXING SCHEMES RELATING TO ENGINES OR PUMPS IN VARIOUS SUBCLASSES OF CLASSES F01-F04
- F05B—INDEXING SCHEME RELATING TO WIND, SPRING, WEIGHT, INERTIA OR LIKE MOTORS, TO MACHINES OR ENGINES FOR LIQUIDS COVERED BY SUBCLASSES F03B, F03D AND F03G
- F05B2270/00—Control
- F05B2270/70—Type of control algorithm
- F05B2270/709—Type of control algorithm with neural networks
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F05—INDEXING SCHEMES RELATING TO ENGINES OR PUMPS IN VARIOUS SUBCLASSES OF CLASSES F01-F04
- F05B—INDEXING SCHEME RELATING TO WIND, SPRING, WEIGHT, INERTIA OR LIKE MOTORS, TO MACHINES OR ENGINES FOR LIQUIDS COVERED BY SUBCLASSES F03B, F03D AND F03G
- F05B2270/00—Control
- F05B2270/80—Devices generating input signals, e.g. transducers, sensors, cameras or strain gauges
- F05B2270/809—Encoders
-
- 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
- Y02E10/70—Wind energy
- Y02E10/72—Wind turbines with rotation axis in wind direction
Landscapes
- Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Sustainable Development (AREA)
- Sustainable Energy (AREA)
- Chemical & Material Sciences (AREA)
- Combustion & Propulsion (AREA)
- Mechanical Engineering (AREA)
- General Engineering & Computer Science (AREA)
- Wind Motors (AREA)
Abstract
The invention provides a fault tolerance method for a wind turbine generator encoder based on signal reconstruction, which comprises the following steps: the method comprises the steps of obtaining data information of a wind turbine generator within preset time, preprocessing the data information, identifying a transfer function and a neural network model of a servo mechanism according to the preprocessed data information, designing a state observer according to the transfer function, setting a gain vector, configuring a pole of the state observer within a preset range, verifying reconstruction performance of the state observer and accuracy of the neural network model when an encoder is normal, setting a threshold value of a residual error between a measured value of the encoder and a reconstructed value of the state observer, and substituting the measured value of the encoder and the reconstructed value of the state observer into the neural network model at a corresponding wind speed for verification if the residual error exceeds the threshold value so as to perform corresponding operation on the encoder. The invention effectively solves the problem that the pitch angle can still be reconstructed through an algorithm after hardware equipment of the pitch angle encoder of the wind turbine generator fails by utilizing a neural network model verification link.
Description
Technical Field
The invention belongs to the technical field of wind turbine generator control, and particularly relates to a fault tolerance method for a pitch angle encoder of a wind turbine generator based on signal reconstruction.
Background
The development of wind energy as a pollution-free renewable energy source has huge economic, social and environmental protection values and development prospects, and the utilization of the wind energy is highly valued by countries in the world. The pitch control technology is a control technology and a power system, and the pitch angle of blades on a rotor system is changed according to wind speed and the rotating speed of a generator so as to achieve the purpose of controlling the output power of the generator. The variable-pitch wind turbine generator system enables output power to be stable, reduces torque oscillation and engine room oscillation by controlling the pitch angle, optimizes the output power, effectively reduces noise, stabilizes the output power of the generator, improves the stress condition of blades and the whole machine, and has better wind energy capture characteristic than a fixed-pitch wind turbine generator.
According to the wind speed condition and the power generation characteristic of the wind turbine, the whole operation process of the wind turbine can be divided into four working conditions. When the wind speed is lower than the cut-in wind speed, the pitch angle beta is the starting optimal pitch angle designed by a manufacturer, the wind speed is waited to exceed the cut-in wind speed, and after the wind speed reaches the cut-in wind speed, the pitch angle is adjusted to 0 degree from the starting optimal pitch angle, and the phase is called as a starting phase; when the wind speed is between the cut-in wind speed and the rated wind speed and the generator runs below the rated rotating speed, the rotating speed of the rotor is adjusted along with the change of the wind speed without controlling the pitch angle, the wind energy is captured to the maximum extent and the electric energy is transmitted to the power grid, and the phase is called as an underpower phase; the electric power generated by the generator is increased along with the continuous increase of the wind speed, and when the wind speed reaches or exceeds the rated wind speed, the electric power generated by the generator also reaches the vicinity of the rated power, and then the pitch control system starts to control according to the power signal of the generator, so as to ensure that the output power of the unit is in the vicinity of the rated power and cannot exceed the power limit, and the stage is called as the rated power operation stage; when the wind speed increases to exceed the maximum wind speed that the unit can bear, the control system stops the unit safely in order to protect the unit. In the stage, the most important task of the variable pitch system is to timely retract the blades to achieve the purpose of reducing the rotating speed of the wind wheel, meanwhile, the brake system is also put into use, the grid-connected switch disconnects the converter from the power grid, and the stage is a protection switching-out stage. Only in the third stage, closed-loop control of the pitch angle is needed, and the encoder measurement value is particularly important, so that the fault-tolerant control of the pitch angle encoder in the stage is very important.
When the wind speed changes, the pitch angle encoder can transmit the information of the pitch angle change to the controller in real time according to the change of the wind speed. However, when the wind turbine generator system operates in a severe environment, the encoder is very easy to fail, and although a standby encoder is usually redundantly configured in the wind turbine generator system, the situation that the main encoder and the standby encoder are simultaneously failed due to the fact that the environment is too severe still occurs.
Therefore, under the condition that the pitch angle encoder of the wind power generation fails and cannot measure the pitch angle, a fault tolerance method for the pitch angle encoder of the wind power generation unit based on signal reconstruction is necessary, namely the pitch angle is reconstructed after the pitch angle encoder fails through a software fault tolerance control method, so that the reliability of the wind power generation is improved.
Disclosure of Invention
The invention aims to at least solve one of the technical problems in the prior art and provides a fault tolerance method for a pitch angle encoder of a wind turbine generator based on signal reconstruction.
The invention provides a fault tolerance method for a wind turbine generator encoder based on signal reconstruction, which comprises the following steps:
acquiring data information of a wind turbine generator within preset time;
preprocessing the data information;
according to the preprocessed data information, identifying a transfer function and a neural network model of the servo mechanism;
designing a state observer according to the transfer function, setting a gain vector, and configuring the pole of the state observer in a preset range;
verifying the reconstruction performance of the state observer and the accuracy of the neural network model when an encoder is normal;
setting a threshold value of a residual error between a measured value of the encoder and a reconstructed value of the state observer, and if the residual error exceeds the threshold value, substituting the measured value of the encoder and the reconstructed value of the state observer into a neural network model at a corresponding wind speed for verification so as to perform corresponding operation on the encoder.
Optionally, the data information includes a pitch angle command, an actual pitch angle, a wind speed, and a wind turbine rotation speed.
Optionally, the preprocessing the data information includes:
performing data supplement on the data information by adopting a linear interpolation method; and/or the presence of a gas in the gas,
performing ring value removing processing on the data information; and/or the presence of a gas in the gas,
carrying out noise reduction processing on the data information by adopting a filtering method; and/or the presence of a gas in the gas,
and carrying out alternate sampling processing on the data information.
Optionally, identifying a transfer function and a neural network model of the servo mechanism according to the preprocessed data information, including:
identifying a transfer function of a servo mechanism according to the preprocessed pitch angle instruction and the actual pitch angle; and the number of the first and second groups,
setting a wind speed range according to the preprocessed wind speed, and respectively training a neural network model between the actual pitch angle and the fan rotating speed in different wind speed ranges.
Optionally, identifying a transfer function of the servo mechanism according to the preprocessed pitch angle command and the actual pitch angle, including:
obtaining a transfer function G by adopting least square identification or particle swarm optimization algorithm according to the preprocessed pitch angle instruction and the actual pitch angle data1(s), the specific relation (1) is as follows:
β(s)＝G1(s)uβ(s) (1)
in the formula: β(s) represents actual pitch angle data;
uβ(s) represents a pitch command.
Optionally, the setting of a wind speed range according to the preprocessed wind speed, and training the neural network model between the actual pitch angle and the fan rotation speed in different wind speed ranges respectively includes:
setting wind speed range nodes v1 and v according to the preprocessed wind speed2、v3And Fitting a Neural network model between the actual pitch angle and the fan rotating speed by utilizing a matlab toolbox Neural Net Fitting, wherein the specific relation (2) is as follows:
in the formula: beta represents the actual pitch angle;
v represents wind speed;
ω represents the fan speed.
Optionally, after verifying the reconstruction performance of the state observer and the accuracy of the neural network model when the encoder is normal, the method further includes:
and if the accuracy of the neural network model is lower than a preset value, retraining the neural network model.
Optionally, the substituting the encoder measurement value and the state observer reconstruction value into a neural network model at a corresponding wind speed for verification to perform corresponding operations on the encoder includes:
and if the first fan rotating speed estimated value obtained by substituting the encoder measured value into the neural network is matched with the actual fan rotating speed, the state observer is mismatched, the transfer function of the servo mechanism between the pitch angle instruction and the actual pitch angle is further identified again, the state observer is redesigned, and pole allocation is completed.
Optionally, the substituting the encoder measurement value and the state observer reconstruction value into a neural network model at a corresponding wind speed for verification to perform corresponding operations on the encoder further includes:
and if the second fan rotating speed estimated value obtained after the state observer reconstructed value is substituted into the neural network is matched with the actual fan rotating speed, the encoder fails, the state observer reconstructed value is further adopted to replace the encoder measured value to serve as an actual pitch angle signal, and an alarm signal for replacing the encoder is sent.
Optionally, the substituting the encoder measurement value and the state observer reconstruction value into a neural network model at a corresponding wind speed for verification to perform corresponding operations on the encoder further includes:
if the measured value of the encoder and the reconstructed value of the state observer are substituted into the neural network model at the corresponding wind speed, and the obtained first fan rotating speed estimated value and the second fan rotating speed estimated value are not matched with the actual fan rotating speed, an alarm signal for detecting whether the encoder fails is sent out; and,
and further re-identifying a transfer function of the servo mechanism between the pitch angle instruction and the actual pitch angle, and re-training the actual pitch angle and fan rotating speed neural network model under different wind speeds.
The invention provides a fault tolerance method for a wind turbine generator encoder based on signal reconstruction, which comprises the following steps: the method comprises the steps of obtaining data information of a wind turbine generator within preset time, preprocessing the data information, identifying a transfer function and a neural network model of a servo mechanism according to the preprocessed data information, designing a state observer according to the transfer function, setting a gain vector, configuring a pole of the state observer within a preset range, verifying reconstruction performance of the state observer and accuracy of the neural network model when an encoder is normal, setting a threshold value of a residual error between a measured value of the encoder and a reconstructed value of the state observer, and substituting the measured value of the encoder and the reconstructed value of the state observer into the neural network model at a corresponding wind speed for verification if the residual error exceeds the threshold value so as to perform corresponding operation on the encoder. The method utilizes a neural network model verification link, further verification is needed when the residual error between the observer reconstruction value and the sensor measurement value is large, adjustment is correspondingly made, the problem that after hardware equipment of a pitch angle encoder of the wind turbine generator fails, a pitch angle can still be reconstructed through an algorithm, and extra economic cost is not needed can be effectively solved.
Drawings
FIG. 1 is a flow chart of a fault tolerance method for a wind turbine generator encoder based on signal reconstruction according to an embodiment of the present invention;
FIG. 2 is a schematic flow chart illustrating a method for fault tolerance of a wind turbine generator encoder based on signal reconstruction according to another embodiment of the present invention;
FIG. 3 is a schematic control structure diagram of a fault tolerance method for a wind turbine generator encoder based on signal reconstruction according to another embodiment of the present invention;
FIG. 4 is a graph of the fault tolerance effect of a pitch angle encoder according to another embodiment of the present invention;
FIG. 5 is a pitch angle reconstruction residual curve according to another embodiment of the present invention;
FIG. 6 is a diagram illustrating a neural network estimated residual error curve according to another embodiment of the present invention.
Detailed Description
In order to make the technical solutions of the present invention better understood, the present invention will be described in further detail with reference to the accompanying drawings and specific embodiments. It is to be understood that the embodiments described are only a few embodiments of the present invention, and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the described embodiments of the invention without any inventive step, are within the scope of protection of the invention.
As shown in fig. 1 to 3, the invention provides a fault tolerance method S100 for a wind turbine generator encoder based on signal reconstruction, which specifically includes the following steps S110 to S160:
s110, acquiring data information of the wind turbine generator within preset time;
specifically, in the process of the normal production operation of the unit, data information of the unit within a certain period of time is acquired, for example, a pitch angle instruction u recorded in an SCADA system within a T period of time is calledβAnd storing the data information into matlab for subsequent processing.
S120, preprocessing the data information;
specifically, each parameter obtained in step S110 is preprocessed in matlab, for example, in some embodiments, if some key data is missing, a linear interpolation method may be used to supplement data information. Alternatively, in other embodiments, if a certain data or some data has a too large deviation from the surrounding data, it is considered as a bad value, and the data information is processed by a ring value removing process to be replaced by an average value of the adjacent data. Alternatively, in other embodiments, if the noise information is too large, the data information may be subjected to noise reduction processing by using a filtering method. Or in other embodiments, if there is a large amount of repeated data, the data information is sampled at intervals, that is, one point is taken as valid data every fixed number of points, so as to reduce the amount of computation.
It should be noted that, in this embodiment, the filtering method is not specifically limited, and inertial filtering may be used, or kalman filtering may be used to perform noise reduction processing.
S130, identifying a transfer function and a neural network model of the servo mechanism according to the preprocessed data information;
specifically, identifying a transfer function of the servo mechanism according to the preprocessed pitch angle command and the actual pitch angle comprises the following steps:
obtaining a transfer function G by adopting least square identification or particle swarm optimization algorithm according to the preprocessed pitch angle instruction and the actual pitch angle data1(s), the specific relation (1) is as follows:
β(s)＝G1(s)uβ(s) (1)
in the formula: β(s) represents actual pitch angle data;
uβ(s) represents a pitch command.
And, because the relationship between the pitch angle and the fan rotating speed is greatly different under different wind speeds, in order to improve the neural network fitting degree, the wind speed range nodes v1 and v are set in the embodiment2、v3And Fitting the actual pitch angle beta with the wind by utilizing a matlab toolbox Neural Net FittingTraining a neural network model between the actual pitch angle and the fan rotating speed in different wind speed ranges, wherein the specific relation (2) is as follows:
in the formula: beta represents the actual pitch angle;
v denotes a fan
ω represents the fan speed.
It should be noted that, in the present embodiment, the deviation of the output values of two adjacent neural network models at the node of the wind speed range node should not exceed 0.15r/min, otherwise, the neural network models need to be re-fitted.
S140, designing a state observer according to the transfer function, setting a gain vector, and configuring the pole of the state observer in a preset range;
specifically, according to the transfer function G obtained in step S1301(s), firstly converting the state observer into a state space expression form, designing the state observer according to the state space expression, and obtaining a feedback gain vector K of the state observer by using a matlab pole allocation method, namely setting a proper state observer gain so as to allocate the observer pole in an ideal range to ensure the observer performance.
S150, verifying the reconstruction performance of the state observer and the accuracy of the neural network model when the encoder is normal. If the accuracy of the neural network model is lower than the preset value, namely the accuracy of the neural network model is lower, the neural network model needs to be retrained.
The designed state observer and neural network model are applied to the pitch angle control system as shown in fig. 3. Completing continuous 24h test before fault tolerance, and counting residual error e between reconstructed observer value and measured encoder valuesAnd residual error e between the output value of the neural network and the rotating speed of the real wind turbine generatorn. Definition of musAnd σsAre respectively within 24hsMean and standard deviation of (D), munAnd σnAre respectively within 24hnThe mean value and the standard deviation of (a) are given by the following relation (3):
note that if e is in the test processsAnd enAnd respectively not exceeding 5% of the real-time pitch angle beta of the unit and the rotating speed omega of the fan, and determining that the test is passed. If the test is finished, the fault-tolerant control method can be applied to the fault-tolerant control of the pitch angle encoder of the wind turbine generator. Otherwise, returning to step S130 to re-identify the transfer function model G between the pitch angle command and the actual pitch angle1(s), refitting the neural network model.
S160, setting a threshold value of a residual error between the measured value of the encoder and the reconstructed value of the state observer, if the residual error exceeds the threshold value, respectively substituting the measured value of the encoder and the reconstructed value of the state observer into the neural network model at the corresponding wind speed for verification so as to correspondingly operate the encoder,
specifically, definition Es＝μs+2.57σs 2Reconstructing a residual error e between an observer value and an encoder measurement valuesA threshold value of (D), definition En＝μn+2.57σn 2Is the residual e between the estimated value of the neural network and the actual rotating speedn1And en2Wherein μsAnd σsAre respectively within 24hsMean and standard deviation of (D), munAnd σnAre respectively within 24hnMean and standard deviation of. And en1＝ω1-，en2＝ω2-ω。
Based on the defined parameters, the measured value of the encoder and the reconstructed value of the state observer are respectively substituted into the neural network model under the corresponding wind speed for verification, and the specific operation of the encoder according to the verification condition has the following three conditions:
first, if during actual operation, the residual exceeds a threshold value, i.e. esTriggering a threshold value, respectively substituting the measured value of the encoder and the reconstructed value of the observer into the neural network model in the corresponding wind speed range, and if the encoder is in the state of the encoder, substituting the measured value of the encoder and the reconstructed value of the observer into the neural network model in the corresponding wind speed rangeSubstituting the measured value into the neural network to obtain a first fan rotating speed estimated value omega1Matching the actual fan speed ω, i.e. closer together, | en1|≤|en2And has: | en1|≤EnThen, the transfer function model G between the pitch angle command and the actual pitch angle needs to be identified again to account for the observer model mismatch1(s) and redesigning the state observer and completing the pole placement.
Second, if the reconstructed value of the state observer is substituted into the neural network, a second fan rotating speed estimated value omega is obtained2Matching the actual fan speed ω, i.e. closer together, | en2|≤|en1And has: | en2|≤EnIf the actual pitch angle signal is the same as the actual pitch angle signal, the reconstructed value of the state observer is used as the actual pitch angle signal, and the actual pitch angle signal is used as the actual pitch angle signal.
Third, if | es|＞Es，|en1|＞EnAnd | en2|＞EnAnd simultaneously, the requirements are met, namely the residual error between the measured value of the encoder and the reconstructed value of the state observer is larger, and the measured value of the encoder and the reconstructed value of the state observer are substituted into the neural network model at the corresponding wind speed to obtain the estimated value omega of the rotating speed of the first fan1And a second fan speed estimate ω2All mismatching with actual fan rotational speed omega, first fan rotational speed estimated value is great with actual fan rotational speed difference promptly, and second fan rotational speed estimated value also has actual fan rotational speed to differ greatly. That is, no matter the reconstructed observer value or the measured encoder value is substituted into the neural network model in the corresponding wind speed range, the obtained output has a larger difference with the actual rotating speed, the fault-tolerant control method is broken down, and an alarm signal for checking whether the encoder fails is sent out at the moment. That is, in the above case, an alarm signal is issued, and it is necessary to manually check whether or not the encoder has failed. And feeds back to step S130. Thereafter, the servomechanism between the pitch angle command and the actual pitch angle is further re-identifiedAnd (4) retraining the actual pitch angle and fan rotating speed neural network models at different wind speeds by using the transfer function.
It should be understood that when e is shown in fig. 2 and 3sAfter triggering the threshold, if not conform to | en1|≤EnAnd | en2|≤EnIn case of (3), the process returns to step S130. If yes, the next step is carried out, namely whether | e is satisfied is judgedn2|≤|en1If the fault is not met, returning to the step S130, if the fault is met, entering the next step, namely completing fault diagnosis, and replacing a pitch angle encoder with an observer reconstruction value.
With reference to fig. 4 to 6, a fault tolerance effect graph of the pitch angle encoder, a pitch angle reconstruction residual error curve and a neural network estimation residual error curve are shown. As can be seen from fig. 6, the blower rotation speed residual obtained by substituting the encoder measurement value into the neural network model triggers the threshold, and the blower rotation speed residual is within the threshold range after the observer reconstruction value is substituted into the neural network model, which indicates that the encoder is in failure, and it is reasonable to replace the encoder measurement value with the observer reconstruction value.
It should be noted that, in the conventional state observer-based sensor fault tolerance control method, when a residual error between a reconstructed value of the state observer and a measured value of the sensor is too large, the sensor fault is directly identified, which may cause a certain identification error. The method is based on a neural network model verification link, and two situations of sensor failure and observer model mismatch need to be verified when the residual error between the observer reconstructed value and the sensor measurement value is large, so that the reason of the overlarge residual error can be further determined, and adjustment can be correspondingly made.
According to the method, a neural network model verification link is utilized, and fault diagnosis of the pitch angle encoder of the wind turbine generator is achieved. And designing a state observer according to the variable pitch system and the mathematical model of the encoder, and reconstructing a pitch angle signal by matching with other normal sensor signals of the wind power to realize fault-tolerant control of the fault of the pitch angle encoder of the wind power generator so as to improve the reliability of wind power generation.
The fault tolerance method for the wind turbine generator encoder based on signal reconstruction will be further described with reference to the following specific embodiments:
example 1
In this example, the fault tolerance method for the wind turbine generator encoder based on signal reconstruction includes the following steps:
s1, on the basis of a certain 1.5MW wind generating set, acquiring historical data of the set in the process of normal production operation of the set, specifically: the tangential wind speed is 3m/s, the rated wind speed is 10.86m/s, the tangential wind speed is 25m/s, the radius of the impeller is 38.5m, the number of blades is 3, and the pitch angle change rate is not more than 3 degrees/s. To obtain G1(s) is:
and S2, preprocessing the parameter information in matlab.
S3, according to the pitch angle command uβAnd obtaining a transfer function G by adopting least square identification or particle swarm optimization algorithm identification with actual pitch angle beta data1(s), the specific relation (1) is as follows:
furthermore, the embodiment is suitable for fault tolerance of the encoder when closed-loop control is carried out on the pitch angle after the wind speed reaches the rated wind speed, so that the wind speed is studied within the range of 10.86-25 m/s. Namely: v is more than or equal to 10.86m/s and less than or equal to 25 m/s. Get v1＝12m/s，v2＝15m/s，v315 m/s. Respectively fitting a neural network model among the rotating speed, the wind speed and the pitch angle of the fan in different wind speed ranges: f ═ ωi(v，β)，i＝1，2，3，4 (2)。
S4, transfer function G obtained according to S31And(s), firstly converting the state space expression into a state space expression form A of-60, B of-1, C of-60 and D of-0, designing a state observer according to the state space expression, and obtaining a feedback gain vector K of 1 of the state observer by using a matlab pole configuration method.
S5-S6, according to the neural network model obtained by researching the wind speed within the range of 10.86-25 m/S, continuous 24-hour tests need to be completed before fault tolerance, and the reconstructed value of the statistical observer and the encoder need to be countedResidual error e of measured valuesAnd residual error e between the output value of the neural network and the rotating speed of the real wind turbine generatornAnd, Es＝μs+2.57σs 2Reconstructing a residual error e between an observer value and an encoder measurement valuesA threshold value of (D), definition En＝μn+2.57σn 2Is the residual e between the estimated value of the neural network and the actual rotating speedn1And en2The threshold value of (2). Further obtaining the observer reconstruction value and the encoder measurement value residual e of the embodiment according to the above formulassThreshold value E ofsIs 1.3 degrees. The neural network estimated value and the actual fan rotating speed threshold value are 1 r/min.
The invention provides a fault tolerance method for a wind turbine generator encoder based on signal reconstruction. Compared with the prior art, the method utilizes a neural network model verification link, further verification is needed when the residual error between the observer reconstruction value and the sensor measurement value is large, so that corresponding adjustment is made, the problem that the pitch angle can still be reconstructed through an algorithm after hardware equipment of a pitch angle encoder of the wind turbine generator fails can be effectively solved, and extra economic cost is not needed. In addition, the method designs the state observer according to the variable pitch system and the mathematical model of the encoder, and reconstructs a pitch angle signal by matching with other normal sensor signals of the wind power, so that the fault-tolerant control of the fault of the pitch angle encoder of the wind power generator is realized, and the reliability of wind power generation is improved.
It will be understood that the above embodiments are merely exemplary embodiments taken to illustrate the principles of the present invention, which is not limited thereto. It will be apparent to those skilled in the art that various modifications and improvements can be made without departing from the spirit and substance of the invention, and these modifications and improvements are also considered to be within the scope of the invention.
Claims (6)
1. A fault tolerance method for a wind turbine generator pitch angle encoder based on signal reconstruction is characterized by comprising the following steps:
acquiring data information of a wind turbine generator within preset time;
preprocessing the data information;
according to the preprocessed data information, identifying a transfer function and a neural network model of the servo mechanism, comprising: identifying a transfer function of a servo mechanism according to the preprocessed pitch angle instruction and the actual pitch angle; and the number of the first and second groups,
setting a wind speed range according to the preprocessed wind speed, and respectively training a neural network model between the actual pitch angle and the fan rotating speed in different wind speed ranges;
designing a state observer according to the transfer function, setting a gain vector, and configuring the pole of the state observer in a preset range;
verifying the reconstruction performance of the state observer and the accuracy of the neural network model when an encoder is normal;
setting a threshold value of a residual error between a measured value of an encoder and a reconstructed value of a state observer, and if the residual error exceeds the threshold value, respectively substituting the measured value of the encoder and the reconstructed value of the state observer into a neural network model at a corresponding wind speed for verification so as to perform corresponding operation on the encoder; wherein,
if the first fan rotating speed estimated value obtained by substituting the encoder measured value into the neural network is matched with the actual fan rotating speed, the state observer is mismatched, the transfer function of the servo mechanism between the pitch angle instruction and the actual pitch angle is further re-identified, the state observer is re-designed and the pole allocation is completed, and,
if the second fan rotating speed estimated value obtained after the state observer reconstructed value is substituted into the neural network is matched with the actual fan rotating speed, the encoder fails, the state observer reconstructed value is further adopted to replace the encoder measured value to serve as an actual pitch angle signal, and an alarm signal for replacing the encoder is sent out;
if the measured value of the encoder and the reconstructed value of the state observer are substituted into the neural network model at the corresponding wind speed, and the obtained first fan rotating speed estimated value and the second fan rotating speed estimated value are not matched with the actual fan rotating speed, an alarm signal for detecting whether the encoder fails is sent out; and,
and further re-identifying a transfer function of the servo mechanism between the pitch angle instruction and the actual pitch angle, and re-training the actual pitch angle and fan rotating speed neural network model under different wind speeds.
2. The method of claim 1, wherein the data information includes a pitch angle command, an actual pitch angle, a wind speed, and a wind turbine speed.
3. The method of claim 2, wherein the preprocessing the data information comprises:
performing data supplement on the data information by adopting a linear interpolation method; and/or the presence of a gas in the gas,
performing ring value removing processing on the data information; and/or the presence of a gas in the gas,
carrying out noise reduction processing on the data information by adopting a filtering method; and/or the presence of a gas in the gas,
and carrying out alternate sampling processing on the data information.
4. The method according to claim 1, wherein identifying a transfer function of a servo mechanism from the pre-processed pitch angle instructions and the actual pitch angle comprises:
obtaining a transfer function G by adopting least square identification or particle swarm optimization algorithm according to the preprocessed pitch angle instruction and the actual pitch angle data1(s), the specific relation (1) is as follows:
β(s)＝G1(s)uβ(s) (1)
in the formula: β(s) represents actual pitch angle data;
uβ(s) represents a pitch command.
5. The method of claim 1, wherein the setting of the wind speed range according to the preprocessed wind speed magnitude, training a neural network model between the actual pitch angle and the wind turbine speed in different wind speed ranges respectively, comprises:
setting a wind speed range node v according to the preprocessed wind speed1、v2、v3And Fitting a Neural network model between the actual pitch angle and the fan rotating speed by utilizing a matlab toolbox Neural Net Fitting, wherein the specific relation (2) is as follows:
in the formula: beta represents the actual pitch angle;
v represents wind speed;
ω represents the fan speed.
6. The method of claim 1, after verifying the state observer reconstruction performance and the accuracy of the neural network model when the encoder is normal, further comprising:
and if the accuracy of the neural network model is lower than a preset value, retraining the neural network model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011522498.XA CN112648140B (en) | 2020-12-21 | 2020-12-21 | Fault tolerance method for wind turbine generator pitch angle encoder based on signal reconstruction |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011522498.XA CN112648140B (en) | 2020-12-21 | 2020-12-21 | Fault tolerance method for wind turbine generator pitch angle encoder based on signal reconstruction |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112648140A CN112648140A (en) | 2021-04-13 |
| CN112648140B true CN112648140B (en) | 2022-03-18 |
Family
ID=75359014
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011522498.XA Active CN112648140B (en) | 2020-12-21 | 2020-12-21 | Fault tolerance method for wind turbine generator pitch angle encoder based on signal reconstruction |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112648140B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113389684B (en) * | 2021-08-18 | 2021-11-09 | 东方电气风电有限公司 | Fault-tolerant control method for yaw angle counter of wind generating set |
| CN114320773B (en) * | 2021-12-22 | 2023-09-22 | 大唐可再生能源试验研究院有限公司 | Wind turbine generator system fault early warning method based on power curve analysis and neural network |
| CN115407648B (en) * | 2022-11-01 | 2023-02-03 | 北京百脉朝宗科技有限公司 | Method, device and equipment for adjusting pitch angle of unmanned aerial vehicle and readable storage medium |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO1997049977A1 (en) * | 1996-06-24 | 1997-12-31 | Arcelik A.S. | Model-based fault detection system for electric motors |
| CN104850889A (en) * | 2014-11-19 | 2015-08-19 | 北京航空航天大学 | Airplane rotation actuator drive unit adaptive fault detection, isolation and confidences assessment method |
| CN105156270A (en) * | 2015-10-09 | 2015-12-16 | 上海电机学院 | Individual pitch control system and method for wind driven generator |
| CN108799012A (en) * | 2018-06-11 | 2018-11-13 | 江南大学 | Based on the wind energy conversion system pitch actuators method for diagnosing faults for improving more born of the same parents' type observation procedures |
| CN108828944A (en) * | 2018-06-21 | 2018-11-16 | 山东大学 | Based on the encoder fault diagnostic system and method for improving PSO and SVM |
| CN111241748A (en) * | 2020-01-13 | 2020-06-05 | 华北电力大学 | Wind driven generator fault diagnosis based on long-short term memory model recurrent neural network |
| CN111682593A (en) * | 2020-05-29 | 2020-09-18 | 黑龙江苑博信息技术有限公司 | Thermal power generating unit coordination optimization method based on neural network model state observer |
| CN111997833A (en) * | 2020-08-14 | 2020-11-27 | 山西大学 | Wind turbine generator set composite intelligent pitch control method based on laser wind measurement feedforward |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| DE10135586B4 (en) * | 2001-07-20 | 2007-02-08 | Eads Deutschland Gmbh | Reconfiguration method for a sensor system with two observers and sensor system for carrying out the method |
| US9234506B2 (en) * | 2011-10-14 | 2016-01-12 | Vestas Wind Systems A/S | Estimation of wind properties using a light detection and ranging device |
| CN106014877B (en) * | 2016-05-27 | 2019-03-15 | 江南大学 | Wind generator system multi-fault Diagnosis and faults-tolerant control |
| US11113395B2 (en) * | 2018-05-24 | 2021-09-07 | General Electric Company | System and method for anomaly and cyber-threat detection in a wind turbine |
| CN109814371A (en) * | 2019-01-08 | 2019-05-28 | 郑州大学 | Networking servo electrical machinery system Active Fault-tolerant Control Method based on Laplace transformation |
| CN110454329B (en) * | 2019-09-04 | 2021-03-30 | 风脉能源（武汉）股份有限公司 | Pitch angle control method for wind turbine generator |
-
2020
- 2020-12-21 CN CN202011522498.XA patent/CN112648140B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO1997049977A1 (en) * | 1996-06-24 | 1997-12-31 | Arcelik A.S. | Model-based fault detection system for electric motors |
| CN104850889A (en) * | 2014-11-19 | 2015-08-19 | 北京航空航天大学 | Airplane rotation actuator drive unit adaptive fault detection, isolation and confidences assessment method |
| CN105156270A (en) * | 2015-10-09 | 2015-12-16 | 上海电机学院 | Individual pitch control system and method for wind driven generator |
| CN108799012A (en) * | 2018-06-11 | 2018-11-13 | 江南大学 | Based on the wind energy conversion system pitch actuators method for diagnosing faults for improving more born of the same parents' type observation procedures |
| CN108828944A (en) * | 2018-06-21 | 2018-11-16 | 山东大学 | Based on the encoder fault diagnostic system and method for improving PSO and SVM |
| CN111241748A (en) * | 2020-01-13 | 2020-06-05 | 华北电力大学 | Wind driven generator fault diagnosis based on long-short term memory model recurrent neural network |
| CN111682593A (en) * | 2020-05-29 | 2020-09-18 | 黑龙江苑博信息技术有限公司 | Thermal power generating unit coordination optimization method based on neural network model state observer |
| CN111997833A (en) * | 2020-08-14 | 2020-11-27 | 山西大学 | Wind turbine generator set composite intelligent pitch control method based on laser wind measurement feedforward |
Non-Patent Citations (2)
| Title |
|---|
| Research on Wind Power Prediction Method Based on Convolutional Neural Network and Genetic Algorithm;Gang Chen，et al;《2019 IEEE PES Innovative Smart Grid Technologies Asia》;20191024;第3573-3578页 * |
| 基于自适应观测器的风电机组故障诊断研究;宋昌举;《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》;20190815;第41-52页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112648140A (en) | 2021-04-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112648140B (en) | Fault tolerance method for wind turbine generator pitch angle encoder based on signal reconstruction | |
| Hossain et al. | Control strategies for augmenting LVRT capability of DFIGs in interconnected power systems | |
| CN105041570B (en) | Yaw control method and device for wind turbine generator | |
| US20140103653A1 (en) | System and method of selecting wind turbine generators in a wind park for change of output power | |
| US10570883B2 (en) | System and method for protecting wind turbines during wind gusts | |
| CN107947228B (en) | Stochastic stability analysis method for power system containing wind power based on Markov theory | |
| CA2769274A1 (en) | Wind turbine | |
| CN111311021A (en) | Theoretical power prediction method, device, equipment and storage medium for wind power plant | |
| Bustan et al. | Adaptive interval type-2 fuzzy controller for variable-speed wind turbine | |
| Zhu et al. | Fault detection and isolation for wind turbine electric pitch system | |
| CN111712634B (en) | Power angle feed-forward signal for phase locked loop in wind turbine power system | |
| Sun et al. | MPPT adaptive controller of DC-based DFIG in resistances uncertainty | |
| CN203259351U (en) | Device for testing control system and pitch system of wind turbine generator set | |
| CN112283047B (en) | Load monitoring method and system based on wind turbine clearance monitoring | |
| CN105649896A (en) | Intelligent monitoring operation control system for wind turbine unit and control method thereof | |
| Bharti et al. | Reliability assessment and performance analysis of DFIG-based WT for wind energy conversion system | |
| CN113852114A (en) | System and method for controlling a power generation system | |
| CN117977554A (en) | Probability transient stability constraint optimal power flow solving method considering wind load uncertainty | |
| Badihi et al. | Fault-tolerant individual pitch control of a wind turbine with actuator faults | |
| Wilson et al. | Transient stability and performance based on nonlinear power flow control design of renewable energy systems | |
| US11739734B2 (en) | Fatigue loads in wind turbines and use of operational metadata | |
| CN109657380A (en) | A kind of double-fed fan motor field Dynamic Equivalence based on Extended Kalman filter | |
| CN114320773A (en) | Wind turbine generator fault early warning method based on power curve analysis and neural network | |
| US20200325874A1 (en) | System and Method for Preventing Catastrophic Damage in Drivetrain of a Wind Turbine | |
| Maaref et al. | A reliability model for a doubly fed induction generator based wind turbine unit considering auxiliary components |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |