CN117514664A - Wind turbine generator blade real-time icing early warning method and system based on Scada data - Google Patents
Wind turbine generator blade real-time icing early warning method and system based on Scada data Download PDFInfo
- Publication number
- CN117514664A CN117514664A CN202311541590.4A CN202311541590A CN117514664A CN 117514664 A CN117514664 A CN 117514664A CN 202311541590 A CN202311541590 A CN 202311541590A CN 117514664 A CN117514664 A CN 117514664A
- Authority
- CN
- China
- Prior art keywords
- data
- wind turbine
- real
- time
- icing
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 40
- 238000004140 cleaning Methods 0.000 claims abstract description 27
- 238000012544 monitoring process Methods 0.000 claims abstract description 26
- 238000010248 power generation Methods 0.000 claims abstract description 21
- 238000012549 training Methods 0.000 claims abstract description 14
- 230000008859 change Effects 0.000 claims description 13
- 230000002159 abnormal effect Effects 0.000 claims description 8
- 238000007781 pre-processing Methods 0.000 claims description 2
- 238000012850 discrimination method Methods 0.000 claims 1
- 230000008901 benefit Effects 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 238000012795 verification Methods 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000007405 data analysis Methods 0.000 description 2
- 230000007423 decrease Effects 0.000 description 2
- 238000013461 design 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
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000005856 abnormality Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000007667 floating Methods 0.000 description 1
- 230000008014 freezing Effects 0.000 description 1
- 238000007710 freezing Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 230000006798 recombination Effects 0.000 description 1
- 238000005215 recombination Methods 0.000 description 1
- 230000009467 reduction Effects 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
- F03D80/00—Details, components or accessories not provided for in groups F03D1/00 - F03D17/00
- F03D80/40—Ice detection; De-icing means
-
- 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
- G06F18/10—Pre-processing; Data cleansing
- G06F18/15—Statistical pre-processing, e.g. techniques for normalisation or restoring missing data
-
- 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
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Chemical & Material Sciences (AREA)
- Combustion & Propulsion (AREA)
- Sustainable Energy (AREA)
- Mechanical Engineering (AREA)
- Sustainable Development (AREA)
- Probability & Statistics with Applications (AREA)
- Wind Motors (AREA)
Abstract
The invention belongs to the field of wind turbine blade icing early warning, and particularly relates to a wind turbine blade real-time icing early warning method and system based on Scada data. Firstly, determining data acquisition points through an SCADA system and configuring data acquisition tasks to acquire wind turbine generator set acquisition point data. And then, carrying out overrun and dead value cleaning on the data, and judging the normal power generation state by using an LOF algorithm so as to obtain accurate wind turbine generator set acquisition point data. These data are then input into a multi-layer perceptron model for training to fit the normal power generation conditions under different conditions and to deploy onnxuntime in the trained model. And finally, constructing a real-time data stream by using Apache Flink, transmitting the real-time data stream to a multi-layer sensor model deployed with ONNXRuntime, and carrying out real-time reasoning to realize real-time monitoring and early warning on icing states of blades of the wind turbine. The method can provide high-efficiency and accurate icing early warning service for the wind turbine blade, and is expected to play an important role in the wind power industry.
Description
Technical Field
The invention belongs to the field of wind turbine blade icing early warning, and particularly relates to a wind turbine blade real-time icing early warning method and system based on Scada data.
Background
The blades are key components for energy conversion of the wind turbine generator. In an actual wind power plant, wind power generation sets are distributed in areas with relatively cold climates such as high altitude, high latitude and the like for capturing more wind energy, so that the wind power generation set blades are easily affected by low-temperature weather such as snowfall and the like to generate icing. Blade icing can alter the airfoil profile of the blade thereby reducing wind energy capture capacity while increasing the energy required for blade rotation, resulting in power loss. In addition, the ice layer generated by the icing of the blade can change the modal parameters of the blade, and the breakage of the blade can be seriously induced.
The current common blade icing monitoring and early warning methods include, but are not limited to, the following:
monitoring a temperature sensor: the temperature sensor arranged on the surface of the blade is used for monitoring the temperature change of the blade, and when the temperature drops below the freezing point, an early warning signal is sent out.
Monitoring by a camera: the icing condition of the surface of the blade is monitored through a camera arranged on the top of the fan tower or the blade, and the icing detection is carried out by utilizing an image processing technology.
Vibration monitoring: the vibration characteristics of the blade are monitored through the vibration sensor arranged on the blade, and whether the blade is frozen is judged by combining vibration signal analysis.
Wind generating set operation data analysis: whether the blades are frozen or not is indirectly judged by analyzing operation data of the wind generating set, such as wind speed, rotating speed, power and the like.
However, the above method has disadvantages: 1) Temperature sensor monitoring may be affected by factors such as ambient temperature and sunlight, and the monitoring result is not accurate enough. 2) The image head monitoring requires processing and analysis of a large amount of image data, has high computational resource requirements, and may be affected in severe weather conditions. 3) The vibration monitoring method needs to have deep knowledge on the vibration characteristics of the blade, and the vibration characteristics of the blade during normal operation and icing can be insufficient and difficult to accurately judge; 4) The wind generating set operation data analysis method needs a large amount of historical data support, is influenced by the operation state of the wind generating set, and can influence the prediction accuracy.
Therefore, the current wind turbine generator blade icing has certain limitations in the aspects of accuracy, instantaneity, applicability and the like, and a more accurate and reliable monitoring and early warning method is required to improve the safety and reliability of wind power generation equipment.
Disclosure of Invention
In view of the above, the invention provides a wind turbine blade real-time icing early warning method and system based on Scada data.
In order to achieve the above purpose, the specific technical scheme is as follows:
the first aspect provides a wind turbine blade real-time icing early warning method based on Scada data, which comprises the following steps: determining a wind turbine generator set data acquisition point in the SCADA system, and configuring a data acquisition task in the SCADA system to acquire wind turbine generator set acquisition point data; performing overrun data cleaning and dead value data cleaning on wind turbine generator collection point data, and judging wind turbine generator collection point data in a normal power generation state of the wind turbine generator based on an LOF algorithm to obtain accurate wind turbine generator collection point data; inputting accurate wind turbine generator set acquisition point data into a multi-layer sensor model for training, and fitting normal power generation states of the wind turbine generator under different working conditions; deploying ONNXRunintime in the trained multi-layer perceptron model; and transmitting the data of the wind turbine generator into a real-time data stream constructed by using Apache Flink to form a real-time data stream of the Flink, transmitting the real-time data stream of the Flink into a multi-layer sensor model deployed with ONNXRuntime, and carrying out real-time reasoning by a reasoning service to carry out real-time reasoning and real-time monitoring and early warning on the icing state of the blades of the wind turbine generator.
Firstly, determining data acquisition points through an SCADA system and configuring data acquisition tasks to acquire wind turbine generator set acquisition point data. And then, carrying out overrun and dead value cleaning on the data, and judging the normal power generation state by using an LOF algorithm so as to obtain accurate wind turbine generator set acquisition point data. These data are then input into a multi-layer perceptron model for training to fit the normal power generation conditions under different conditions and to deploy onnxuntime in the trained model. And finally, constructing a real-time data stream by using Apache Flink, transmitting the real-time data stream to a multi-layer sensor model deployed with ONNXRuntime, and carrying out real-time reasoning to realize real-time monitoring and early warning on icing states of blades of the wind turbine. According to the method, accurate wind turbine generator set data can be effectively extracted through data acquisition and cleaning of the SCADA system and data discrimination based on an LOF algorithm. Meanwhile, the training of the multi-layer perceptron model and the deployment of ONNXRuntime ensure efficient model reasoning service. By combining with the real-time data stream constructed by the Apache Flink, the method can realize accurate monitoring and real-time early warning of the icing state of the wind turbine blade, and improves the safety and reliability of the wind turbine. The method comprehensively utilizes data processing, machine learning and real-time reasoning, and provides a high-efficiency and accurate icing early warning solution for the wind power industry.
The second aspect provides a wind turbine blade real-time icing early warning system based on Scada data, comprising: the data acquisition module is configured to acquire wind turbine generator set acquisition point data by using the SCADA system; the data preprocessing module is configured to perform overrun data cleaning and dead value data cleaning on the data and judge the data of the collection points of the wind turbine in the normal power generation state; and an icing prediction module: the method comprises the steps of utilizing an ONNXRuntime multi-layer sensor model to monitor and pre-warn icing conditions of blades of a wind turbine generator in real time; and a display alarm module: configured as a display and alarm
Compared with the prior art, the invention has the beneficial effects that:
1) According to the method, the SCADA system is utilized to collect the data of the wind turbine in real time, and the Apache Flink is combined to construct a real-time data stream, so that the real-time monitoring and early warning of the icing state of the wind turbine blade are realized. The real-time performance is helpful for timely finding out abnormal conditions and taking corresponding measures, and the safety and reliability of the wind turbine generator are improved.
2) By means of over-limit data cleaning, dead value data cleaning and LOF algorithm-based normal power generation state judgment on the wind turbine generator set acquisition point data, accurate wind turbine generator set acquisition point data can be obtained. This helps to reduce false alarms and false misses, improving the accuracy of the pre-warning.
3) The whole early warning method utilizes automatic data acquisition, cleaning and model reasoning processes, reduces the need of manual intervention, improves the efficiency and reduces the human error.
4) The method trains wind turbine data by using a multi-layer perceptron model to fit normal power generation states under different working conditions. And by deploying ONNXRuntime, efficient model reasoning service is realized. The application of the machine learning model is beneficial to improving the accuracy and the real-time performance of early warning.
In summary, the method comprehensively utilizes the SCADA system, the data cleaning technology, the machine learning model training and the real-time reasoning service, forms a complete real-time monitoring and early warning system, and provides an efficient and accurate icing early warning solution for the wind power industry.
Drawings
FIG. 1 is a flow chart of a wind turbine blade real-time icing early warning method based on Scada data;
FIG. 2 is a flowchart of a wind turbine blade real-time icing early warning method S200 based on Scada data;
FIG. 3 is a flowchart of a wind turbine blade real-time icing early warning method S500 based on Scada data;
Detailed Description
The following description of the technical solutions in the embodiments of the present application will be made clearly and completely with reference to the drawings in the embodiments of the present application, and it is apparent that the described embodiments are only some embodiments of the present application, not all embodiments. All other embodiments, which can be made by one of ordinary skill in the art based on the embodiments herein without making any inventive effort, are intended to be within the scope of the present application.
The following is an exemplary wind turbine blade real-time icing early warning method based on Scada data
The overall steps are as shown in fig. 1:
s100: and determining a wind turbine generator set data acquisition point in the SCADA system, and configuring a data acquisition task in the SCADA system to acquire wind turbine generator set acquisition point data.
In this step, determining the data acquisition points to be monitored in the SCADA system generally needs to be based on the specific design and operation conditions of the wind turbine, and first, needs to refer to the design document and the equipment parameter table of the wind turbine to know the positions of various sensors and monitoring points and the parameters monitored by the sensors and monitoring points. These parameters include blade angle, wind turbine active power, wind speed, off-board temperature, etc. And then analyzing the running state of the wind turbine generator, and determining and judging the parameters which are the most critical to the icing state of the blade. Through practical judgment, blade icing generally causes abnormal decline of blade angles, abnormal decline of active power of a wind turbine, reduction of wind speed and conditions of temperature change outside a cabin. And determining the data acquisition points to be monitored, and configuring corresponding data acquisition tasks in the SCADA system to realize real-time monitoring of the parameters.
The SCADA system is utilized to collect data, firstly an administrator or a user account with configuration authority is used for logging in the SCADA system, configuration or setting options are found in an interface of the SCADA system after logging in, and a data collection task configuration interface is opened. In the data acquisition task configuration interface, there is typically a button, such as "add data points" or "add tasks", that is clicked to begin adding the data acquisition points for which monitoring has been determined. For each monitored data acquisition point, parameters thereof, including data point names, data types, data sources, acquisition frequencies, and the like, need to be configured. For example, for a blade angle data point, it is necessary to configure what is known as a "blade angle", the data type is a floating point number, the data source is the corresponding sensor or device, and the acquisition frequency is once per second. After confirming the configuration parameters of each data acquisition point, the configuration is saved or confirmed to be effective. After configuration is completed, data acquisition tests can be carried out, whether the numerical value of each data acquisition point can be correctly acquired is monitored in real time, and once the configuration is completed and the tests are passed, the system starts to monitor the data acquisition points in real time, so that the data such as the blade angle, the active power of the wind turbine, the wind speed and the outside temperature of the cabin of the wind turbine can be acquired in real time.
S200: and performing overrun data cleaning and dead value data cleaning on the wind turbine collecting point data, and judging the wind turbine collecting point data in the normal power generation state of the wind turbine based on the LOF algorithm to obtain accurate wind turbine collecting point data.
In wind turbine generator system monitoring, overrun data cleaning can limit the range of collected data, and data exceeding a reasonable range can be cleaned or marked. And (5) performing overrun data cleaning by setting a reasonable value range of the data points in the SCADA system.
The value range of the blade angle data is 0-90 degrees, the value range of the wind motor active power data is 0-rated power, the value range of the wind speed data is 0-25 m/s, and the value range of the outdoor temperature data is-40-60 ℃. And defining data exceeding the range of the blade angle data, the wind power active power data, the wind speed data and the outdoor temperature data as overrun data, and cleaning or marking.
In wind turbine generator monitoring, dead value data cleaning can be performed on dead values or repeated values in collected data, and dead value data cleaning is performed by setting the change rate or fluctuation range of data points in a SCADA system
For blade angle data, a rate of change threshold for blade angle is set in the SCADA system, i.e., between 5 °/min and 10 °/min. For wind motor active power data, a power change rate threshold is set in the SCADA system, namely, 50 kW/min-100 kW/min. For wind speed data, a change rate threshold of wind speed, i.e. 5m/s to 10m/s, is set in the SCADA system. For the out-of-cabin temperature data, setting a change rate threshold of the out-of-cabin temperature in the SCADA system, namely, 2-5 ℃/min. And defining data exceeding the range of the blade angle data, the wind power active power data, the wind speed data and the outdoor temperature data as dead value data, and cleaning or marking.
And judging the data of the wind turbine generator set acquisition points by using a LOF algorithm-based method so as to identify the data of the normal power generation state. The LOF algorithm is an algorithm for detecting abnormal values, which determines the degree of abnormality by calculating the ratio of the local density of each data point to the local density of the adjacent points, so that the data points in a normal state can be identified.
As shown in fig. 2, the detailed discrimination is as follows:
s210, inputting collection point data of the wind turbine generator after the overrun data are cleaned and the dead value data are cleaned as data points of an LOF algorithm;
s220, for each data point, calculating the distance between the data point and other data points, and for each data point, determining k nearest neighbor data points;
s230, calculating the reachable distance of each data point, wherein the reachable distance is a larger value in the direct distance from the data point a to the data point b and the k-adjacent distance of the data point b, and the calculation formula is as follows:
D K (a,b)＝max{D(a,b),K-D(b)}
wherein D is K (a, b) an achievable distance from data point a to data point b, D (a, b) a direct distance from data point a to data point b, and K-D (b) a K-adjacent distance from data point b;
s240, calculating local reachable density of each data point, wherein the local reachable density is the average reciprocal of reachable distance of k-adjacent points of the data point a, and the calculation formula is as follows:
wherein ρ is K (a) Representing the local reachable density of data point a, D K (a, b) the reachable distance from data point a to data point b, N K (a) A k-neighbor set representing data point a;
s250, calculating a local outlier factor LOF of each data point, wherein the local outlier factor is an average value of the ratio of the local reachable density of the k-adjacent point of a to the local reachable density of a, and the calculation formula is as follows:
wherein LOF K (a) Local outlier factor of data point a, ρ K (a) Representing the local reachable density of data point a, ρ K (b) Representing the local reachable density of data point b, N K (a) Representing the k-neighbor set of data points a.
S260, setting a threshold value of data for judging the normal power generation state, and marking data points with LOF values lower than the threshold value as normal state data and data points with LOF values higher than the threshold value as abnormal state data.
The threshold value is set by comprehensively considering factors such as data distribution, domain knowledge, actual demand and the like, and performing actual debugging and verification. In general, an appropriate threshold can be determined by analyzing historical data, communicating with a domain expert, and actual testing and adjustment to achieve efficient discrimination between normal and abnormal state data.
S300, inputting accurate wind turbine generator set acquisition point data into a multi-layer sensor model for training, and fitting normal power generation states of the wind turbine generator under different working conditions;
the method comprises the steps of training a multi-layer perceptron model by using accurate data, and constructing an icing early-warning model. The multi-layer perceptron model has stronger nonlinear modeling capability and can adapt to the change of the running state and icing condition of the complex wind turbine generator. By using accurate data to train the model, the accuracy and the robustness of the icing pre-warning model can be improved.
And training a multi-layer perceptron (MLP) model by utilizing accurate wind turbine acquisition point data, dividing the data into a training set and a verification set in the training process, training the MLP model by utilizing the training set, evaluating the performance of the model by utilizing the verification set, and checking whether the model can accurately predict the normal state and icing condition of the wind turbine.
If the model performs well on the verification set, the normal state and the icing state can be accurately identified, and the trained multi-layer perceptron model can be saved as a deployable format for subsequent deployment into ONNXRuntime for real-time reasoning.
S400, deploying ONNXRuntime in the trained multi-layer perceptron model.
Onnxuntime is designed to enable high performance reasoning, and can provide fast and efficient reasoning capabilities on a variety of hardware platforms. After the multi-layer perceptron model training is completed, the trained model is exported to an ONNX format. The inference service with ONNXRuntime deployed is integrated into the system, and corresponding code is written to load and run the ONNX multi-tier perceptron model.
S500, transmitting the data of the wind turbine generator into a real-time data stream constructed by using Apache Flink to form the real-time data stream of the Flink, transmitting the real-time data stream of the Flink to a multi-layer sensor model deployed with ONNXRUNtime, and carrying out real-time reasoning by a reasoning service to carry out real-time reasoning and real-time monitoring and early warning on the icing state of the blades of the wind turbine generator.
The method can realize the rapid processing and transmission of the wind turbine generator data acquired in real time by utilizing the real-time data stream constructed by the Apache Flink, and ensures the instantaneity of ice-making early warning. As shown in fig. 3, the specific operation mode is as follows:
s510, transmitting the wind turbine generator data acquired in real time into a real-time data stream constructed by Apache Flink through data acquisition equipment or a sensor;
s502, transmitting the real-time data stream of the Flink to an inference service of a multi-layer sensor model deployed with ONNXRuntime by using wind turbine data acquired in real time;
s503, the ONNXRuntime multi-layer sensor model utilizes the trained multi-layer sensor model to carry out real-time reasoning on the input wind turbine generator data so as to monitor and early warn the icing state of the wind turbine generator blades.
And an ONNXRUNtime multilayer sensor model is adopted, and the model can be used for carrying out real-time reasoning according to the data of wind speed, temperature, humidity and the like so as to monitor the icing state of the wind turbine blade. Once the model detects abnormal conditions, such as low wind speed and low temperature, icing may occur, the system will immediately send out early warning, and notify relevant personnel to perform overhaul and maintenance, so as to avoid potential safety hazards and losses caused by icing of the wind turbine blades. Thus, through the operation of the whole S500 stage, the real-time monitoring and early warning of the icing state of the wind turbine blade are realized, and the safe operation of the wind turbine is ensured.
The basic principles of the present application have been described above in connection with specific embodiments, however, it should be noted that the advantages, benefits, effects, etc. mentioned in the present application are merely examples and not limiting, and these advantages, benefits, effects, etc. are not to be considered as necessarily possessed by the various embodiments of the present application. Furthermore, the specific details disclosed herein are for purposes of illustration and understanding only, and are not intended to be limiting, as the application is not intended to be limited to the details disclosed herein as such.
The block diagrams of the devices, apparatuses, devices, systems referred to in this application are only illustrative examples and are not intended to require or imply that the connections, arrangements, configurations must be made in the manner shown in the block diagrams. As will be appreciated by one of skill in the art, the devices, apparatuses, devices, systems may be connected, arranged, configured in any manner. Words such as "including," "comprising," "having," and the like are words of openness and mean "including but not limited to," and are used interchangeably therewith. The terms "or" and "as used herein refer to and are used interchangeably with the term" and/or "unless the context clearly indicates otherwise. The term "such as" as used herein refers to, and is used interchangeably with, the phrase "such as, but not limited to.
It is also noted that in the apparatus, devices and methods of the present application, the components or steps may be disassembled and/or assembled. Such decomposition and/or recombination should be considered as equivalent to the present application.
The previous description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other aspects without departing from the scope of the application. Thus, the present application is not intended to be limited to the aspects shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
The foregoing description of the preferred embodiments of the present invention is not intended to limit the invention to the precise form disclosed, and any modifications, equivalents, and alternatives falling within the spirit and principles of the present invention are intended to be included within the scope of the present invention.
Claims (8)
1. A wind turbine generator blade real-time icing early warning method based on Scada data is characterized by comprising the following steps:
determining a wind turbine generator set data acquisition point in the SCADA system, and configuring a data acquisition task in the SCADA system to acquire wind turbine generator set acquisition point data;
performing overrun data cleaning and dead value data cleaning on wind turbine collecting point data, and judging wind turbine collecting point data in a normal power generation state of the wind turbine based on an LOF algorithm to obtain accurate wind turbine collecting point data;
inputting accurate wind turbine generator set acquisition point data into a multi-layer sensor model for training, and fitting normal power generation states of the wind turbine generator under different working conditions;
deploying ONNXRunintime in the trained multi-layer perceptron model;
and transmitting the data of the wind turbine generator into a real-time data stream constructed by using Apache Flink to form a real-time data stream of the Flink, transmitting the real-time data stream of the Flink into a multi-layer sensor model deployed with ONNXRuntime, and carrying out real-time reasoning by a reasoning service to carry out real-time reasoning and real-time monitoring and early warning on the icing state of the blades of the wind turbine generator.
2. The method for early warning of real-time icing of a wind turbine blade based on Scada data according to claim 1, wherein the wind turbine acquisition point data comprises: and (5) data of blade angles, active power of the wind turbine, wind speed and outside-cabin temperature of the wind turbine.
3. The method for early warning the real-time icing of the wind turbine blade based on Scada data according to claim 2 is characterized by comprising the following steps of:
the blade angle data is set to be 0-90 degrees in the SCADA system;
the value range of the active power data of the wind power generator is set to be 0-rated power in the SCADA system;
the wind speed data is set in the SCADA system with the value range of 0 m/s-25 m/s;
the value range of the outdoor temperature data is set to be-40 ℃ to 60 ℃ in the SCADA system;
and defining data exceeding the blade angle data, wind power active power data, wind speed data and outside-cabin temperature data range as overrun data, and cleaning or marking.
4. The method for early warning the real-time icing of the wind turbine blade based on Scada data according to claim 2 is characterized by comprising the following steps of:
the paddle angle data sets the change rate threshold range of the paddle angle in the SCADA system to be 5 degrees/min-10 degrees/min;
the power change rate threshold value range is set to be 50 kW/min-100 kW/min in the SCADA system according to the active power data of the wind power generator;
the wind speed data sets a change rate threshold value of wind speed to be 5 m/s-10 m/s in the SCADA system;
the temperature data outside the cabin is provided with a change rate threshold range of the temperature outside the cabin in the SCADA system of 2-5 ℃/min;
and defining data exceeding the blade angle data, wind power active power data, wind speed data and outside-cabin temperature data range as dead value data, and cleaning or marking.
5. The method for early warning the real-time icing of the wind turbine blade based on Scada data according to claim 1 is characterized by comprising the following steps of: the wind turbine generator set acquisition point data discrimination method based on LOF algorithm in the normal power generation state of the wind turbine generator set is as follows:
the collected point data of the wind turbine generator after the overrun data cleaning and dead value data cleaning are used as data points of an LOF algorithm to be input; for each data point, calculate the distance between it and the other data points, and for each data point, it is necessary to determine its k nearest neighbor data points; and gradually calculating the reachable distance of each data point, the local reachable density of each data point and the local outlier factor LOF of each data point, and marking the data points with LOF values lower than a threshold value as normal state data and the data points with LOF values higher than the threshold value as abnormal state data by setting a threshold value of the data for judging the normal power generation state.
6. The method for early warning the real-time icing of the wind turbine blade based on Scada data according to claim 1 is characterized by comprising the following steps of: the method for deploying ONNXRuntime in the multi-layer perceptron model is as follows:
after the multi-layer perceptron model is trained, the trained model is exported to an ONNX format;
the inference service deployed with ONNX multiple layer perceptron model is integrated into the system and code is written to load and run the ONNX multiple layer perceptron model.
7. The method for early warning the real-time icing of the wind turbine blade based on Scada data according to claim 1 is characterized by comprising the following steps of: the real-time monitoring and early warning mode of the icing state of the wind turbine blade of the ONNXRUNtime multi-layer sensor model is as follows:
the method comprises the steps that wind turbine generator data acquired in real time are transmitted into a real-time data stream constructed by Apache Flink through data acquisition equipment or a sensor;
s502, transmitting the real-time data stream of the Flink to an inference service of a multi-layer sensor model deployed with ONNXRuntime by using wind turbine data acquired in real time;
s503, the ONNXRuntime multi-layer sensor model utilizes the trained multi-layer sensor model to carry out real-time reasoning on the input wind turbine generator data so as to monitor and early warn the icing state of the wind turbine generator blades.
8. Wind turbine generator system blade real-time icing early warning system based on Scada data, which is characterized by comprising:
the data acquisition module is configured to acquire wind turbine generator set acquisition point data by using the SCADA system;
the data preprocessing module is configured to perform overrun data cleaning and dead value data cleaning on the data and judge the data of the collection points of the wind turbine in the normal power generation state;
and an icing prediction module: the method comprises the steps of utilizing an ONNXRuntime multi-layer sensor model to monitor and pre-warn icing conditions of blades of a wind turbine generator in real time;
and a display alarm module: configured as a display and an alarm.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311541590.4A CN117514664A (en) | 2023-11-17 | 2023-11-17 | Wind turbine generator blade real-time icing early warning method and system based on Scada data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311541590.4A CN117514664A (en) | 2023-11-17 | 2023-11-17 | Wind turbine generator blade real-time icing early warning method and system based on Scada data |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117514664A true CN117514664A (en) | 2024-02-06 |
Family
ID=89747325
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311541590.4A Pending CN117514664A (en) | 2023-11-17 | 2023-11-17 | Wind turbine generator blade real-time icing early warning method and system based on Scada data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117514664A (en) |
-
2023
- 2023-11-17 CN CN202311541590.4A patent/CN117514664A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Wang et al. | SCADA data based condition monitoring of wind turbines | |
| US7895018B2 (en) | Event monitoring via combination of signals | |
| Kreutz et al. | Machine learning-based icing prediction on wind turbines | |
| KR20160073945A (en) | System and method for managing wind plant | |
| KR20160017681A (en) | System and method for managing wind plant | |
| CN108599724A (en) | A kind of photovoltaic module on-line monitoring system and monitoring method | |
| CN114061743A (en) | Vibration monitoring method, device, equipment and medium for wind generating set | |
| Zhang et al. | A novel fault diagnosis method for wind turbine based on adaptive multivariate time-series convolutional network using SCADA data | |
| CN115314235A (en) | System and method for network attack detection in wind turbine control systems | |
| CN116306139A (en) | Intelligent monitoring method and system for service life of wind turbine blade | |
| CN114593020A (en) | Blade icing monitoring and early warning method and device for wind turbine generator and electronic equipment | |
| CN113606099A (en) | Method and system for monitoring icing of blades of wind turbine generator | |
| CN114021906A (en) | Unattended wind power generation operation and maintenance management method and system | |
| CN114442543A (en) | Computer monitoring method suitable for early warning of hydropower station fault | |
| CN117514664A (en) | Wind turbine generator blade real-time icing early warning method and system based on Scada data | |
| CN112464151B (en) | Abnormal sound diagnosis method for yaw system of wind turbine generator based on acoustic diagnosis | |
| CN114320773B (en) | Wind turbine generator system fault early warning method based on power curve analysis and neural network | |
| Ashwini et al. | Wireless sensor network for condition monitoring of remote wind mill | |
| CN114607571A (en) | Offshore wind power gear box fault identification method and system based on lubricating system monitoring | |
| CN112710918A (en) | Wireless data acquisition method and system based on edge calculation | |
| CN115711206B (en) | Wind driven generator blade icing state monitoring system based on clustering weight | |
| CN214330806U (en) | Wind driven generator system and fault analysis equipment of wind driven generator | |
| CN117894133B (en) | Icing alarm method and system for power transmission line | |
| CN220434942U (en) | Panoramic fault diagnosis system for wind turbine generator | |
| CN118008729B (en) | Beidou-based offshore wind turbine structure state monitoring method and system |
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 |