1. A gearbox oil temperature fault early warning method based on a multilayer perception neural network is characterized by comprising the following steps: the fault early warning method comprises the following steps:

(1) collecting SCADA data of a fan, inputting the data, analyzing the related characteristics and related characteristic data of the oil temperature of a fan gear box, extracting the data characteristics, designing a database required by data screening conditions, and storing the screening conditions according to a required format;

(2) analyzing and screening the acquired data, and removing data which is irrelevant to the oil temperature fault analysis of the gearbox, so that a cleaner data set can be obtained;

(3) establishing a gearbox oil temperature fault analysis model by utilizing a multilayer perception neural network, and inputting the cleaned data into the model for model training;

(4) applying the trained model to other analyses to be predicted or tested, obtaining a prediction result of the algorithm through model calculation, and determining that the analysis needs to be retested through judgment conditions;

(5) and judging the accuracy of the model by using the residual error of the actual value of the fan and the algorithm predicted value, and if the fault prediction model is accurate, outputting the fault diagnosis probability by using the probability distribution of the residual error.

2. The gearbox oil temperature fault early warning method based on the multilayer perceptive neural network of claim 1, characterized in that: and (2) collecting the SCADA data of the fan in the step (1) through an SCADA system of the fan, wherein the SCADA system of the fan is used for predicting the running state of the fan in real time.

3. The gearbox oil temperature fault early warning method based on the multilayer perceptive neural network of claim 1, characterized in that: and (2) comprehensively recording each predicted point of the wind turbine generator at intervals by the SCADA system of the fan in the step (1), and basically reflecting the real running condition of the fan through data generated by the SCADA system.

4. The gearbox oil temperature fault early warning method based on the multilayer perceptive neural network of claim 1, characterized in that: the method for analyzing and screening the collected data in the step (2) is a condition judgment and decision tree algorithm.

5. The gearbox oil temperature fault early warning method based on the multilayer perceptive neural network of claim 1, characterized in that: the base data of the SCADA received in the step (2) contains too much information, some of the base data do not influence and remove irrelevant data for judging the oil temperature fault of the gearbox, the construction of an analysis model of the gearbox is facilitated, and the removal of the irrelevant data enables the judgment of the fault result of the gearbox to be more accurate.

6. The gearbox oil temperature fault early warning method based on the multilayer perceptive neural network of claim 1, characterized in that: the multilayer perception neural network in the step (3) is a back propagation training algorithm and a feedback training algorithm, and is a gradient descent method using reverse automatic derivation, and a residual probability model with high prediction accuracy can be obtained through verification and iteration of a training set and a test set.

7. The gearbox oil temperature fault early warning method based on the multilayer perceptive neural network of claim 1, characterized in that: the training step in the step (3) is as follows: 1) calculating the feedforward output of each neuron at each connection layer, 2) calculating the output error of the network and calculating the contribution degree of each neuron of the last layer to the error, 3) continuing to calculate the contribution degree of each neuron of the last layer to the error, 4) until the algorithm reaches the input layer.

8. The gearbox oil temperature fault early warning method based on the multilayer perceptive neural network of claim 1, characterized in that: and (4) obtaining a predicted value in the model, then obtaining a real value in other fans, and then calculating a residual error, wherein the calculation of the residual error is to subtract the real value from the predicted value.

9. The gearbox oil temperature fault early warning method based on the multilayer perceptive neural network of claim 1, characterized in that: and (4) determining whether the calculation result meets a judgment condition according to the obtained residual error, performing the next fault diagnosis if the judgment condition is met, and restarting from the step (2) if the judgment condition is not met.

10. The gearbox oil temperature fault early warning method based on the multilayer perceptive neural network of claim 1, characterized in that: and (5) making a diagnosis result and the pre-mapping into a bar graph, so that the implementation effect of the model can be quickly known through the bar graph.

Background

With the increasingly stringent requirements of the international society on environmental protection, the consumption of traditional fossil fuels will further decrease, and in addition, China needs to achieve the goal of "carbon neutralization" in the future, which means that the proportion of renewable energy sources will further increase. Wind power generation is used as a clean and sustainable energy form and is increasingly applied worldwide. The environment in which the wind turbine generator operates is relatively complex and severe, and the influence from the environment brings great damage to the wind turbine generator.

In order to ensure that the wind turbine generator can reliably and stably operate and recover the cost for a fan operator as soon as possible, the method needs to predict the state of key parts of the wind turbine generator and the environment where the wind turbine generator operates in real time and predict and research the faults of the key parts of the wind turbine generator to be used as a key part gear box with very high maintenance cost and often high fault rate in the wind turbine generator, the function of the gear box cannot be replaced, and the estimated cost of the gear box accounts for about 16 percent of the total cost of the whole double-fed wind turbine generator, so that the method has great practical value for analyzing the faults of the gear box of the wind turbine generator. In addition, the method comprises the following steps. Therefore, a gearbox oil temperature fault early warning method based on a multilayer perceptive neural network is provided for solving the problems.

Disclosure of Invention

The embodiment provides a gearbox oil temperature fault early warning method based on a multilayer perception neural network, which is used for solving the problem that a wind turbine generator gearbox in the prior art is lack of fault prediction.

According to one aspect of the application, a gearbox oil temperature fault early warning method based on a multilayer perceptive neural network is provided, and the fault early warning method comprises the following steps:

(1) collecting SCADA data of a fan, inputting the data, analyzing the related characteristics and related characteristic data of the oil temperature of a fan gear box, extracting the data characteristics, designing a database required by data screening conditions, and storing the screening conditions according to a required format;

(2) analyzing and screening the acquired data, and removing data which is irrelevant to the oil temperature fault analysis of the gearbox, so that a cleaner data set can be obtained;

(3) establishing a gearbox oil temperature fault analysis model by utilizing a multilayer perception neural network, and inputting the cleaned data into the model for model training;

(4) applying the trained model to other analyses to be predicted or tested, obtaining a prediction result of the algorithm through model calculation, and determining that the analysis needs to be retested through judgment conditions;

(5) and judging the accuracy of the model by using the residual error of the actual value of the fan and the algorithm predicted value, and if the fault prediction model is accurate, outputting the fault diagnosis probability by using the probability distribution of the residual error.

Further, the SCADA data of the fan in the step (1) is collected through a SCADA system of the fan, and the SCADA system of the fan is a system used for predicting the running state of the fan in real time.

Further, the SCADA system of the wind turbine in the step (1) carries out comprehensive recording on each predicted point of the wind turbine at intervals, and the real operation condition of the wind turbine is basically reflected through data generated by the SCADA system.

Further, the method for analyzing and screening the collected data in the step (2) is a condition judgment and decision tree algorithm.

Further, the base data of the SCADA received in the step (2) contains too much information, some of the base data do not influence and remove irrelevant data for judging the gearbox oil temperature fault, and therefore the construction of a gearbox analysis model is facilitated, and the removal of the irrelevant data enables the judgment of the gearbox fault result to be more accurate.

Further, the multilayer perceptual neural network in the step (3) is a back propagation training algorithm and a feedback training algorithm, and is a gradient descent method using reverse automatic derivation, and the algorithm can obtain a residual probability model with high prediction accuracy through verification and iteration of a training set and a test set.

Further, the training step in the step (3) is: 1) calculating the feedforward output of each neuron at each connection layer, 2) calculating the output error of the network and calculating the contribution degree of each neuron of the last layer to the error, 3) continuing to calculate the contribution degree of each neuron of the last layer to the error, 4) until the algorithm reaches the input layer.

Furthermore, the predicted value in the model is obtained in the step (4), then the real value in other fans is obtained, and then the residual error is calculated, wherein the residual error is calculated by subtracting the real value from the predicted value.

Further, in the step (4), whether the calculation result meets the judgment condition is determined according to the obtained residual error, the next fault diagnosis is carried out when the judgment condition is met, and the step (2) is restarted when the judgment condition is not met.

Further, the diagnostic result and the pre-mapping are made into a bar graph in the step (5), so that the effect of the model implementation can be quickly known through the bar graph.

According to the embodiment of the application, the fault early warning method is adopted, the problem that the gearbox of the wind turbine generator is lack of fault prediction is solved, and the fault prediction and timely maintenance effect of the wind turbine generator can be achieved.

Drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art that other drawings can be obtained according to the drawings without inventive exercise.

Fig. 1 is a schematic flow chart of an embodiment of the present application.

Detailed Description

In order to make the technical solutions better understood by those skilled in the art, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only partial embodiments of the present application, but not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.

It should be noted that the terms "first," "second," and the like in the description and claims of this application and in the drawings described above are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It should be understood that the data so used may be interchanged under appropriate circumstances such that embodiments of the application described herein may be used. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.

In this application, the terms "upper", "lower", "left", "right", "front", "rear", "top", "bottom", "inner", "outer", "middle", "vertical", "horizontal", "lateral", "longitudinal", and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the drawings. These terms are used primarily to better describe the present application and its embodiments, and are not used to limit the indicated devices, elements or components to a particular orientation or to be constructed and operated in a particular orientation.

Moreover, some of the above terms may be used to indicate other meanings besides the orientation or positional relationship, for example, the term "on" may also be used to indicate some kind of attachment or connection relationship in some cases. The specific meaning of these terms in this application will be understood by those of ordinary skill in the art as appropriate.

Furthermore, the terms "mounted," "disposed," "provided," "connected," and "sleeved" are to be construed broadly. For example, it may be a fixed connection, a removable connection, or a unitary construction; can be a mechanical connection, or an electrical connection; may be directly connected, or indirectly connected through intervening media, or may be in internal communication between two devices, elements or components. The specific meaning of the above terms in the present application can be understood by those of ordinary skill in the art as appropriate.

It should be noted that the embodiments and features of the embodiments in the present application may be combined with each other without conflict. The present application will be described in detail below with reference to the embodiments with reference to the attached drawings.

The fault early warning method in this embodiment may be applicable to various wind turbines, for example, the following wind turbine is provided in this embodiment, and the fault early warning method in this embodiment may be used to predict the following wind turbine.

The wind turbine generator comprises an impeller, a transmission system and a generator; the generator comprises a stator and a rotor, and the windings of the stator and the windings of the rotor are used for being electrically connected with a power grid; the winding of the rotor is electrically connected with the power grid through a double-fed converter; the rotor is in transmission connection with the impeller through the transmission system;

the wind turbine generator system further comprises a stator winding short-circuit device, a stator power grid breaking device, an angle sensor, a first controller, a second controller and a calculation module; wherein: the stator power grid breaking device is connected between a winding of the stator and the power grid and is used for controlling the connection and disconnection of the electrical connection between the winding of the stator and the power grid; the stator winding short-circuit device is used for short-circuiting each winding of the stator when the winding of the stator is disconnected with the power grid; the angle sensor is arranged on a blade of the impeller and used for detecting the current deflection angle of the blade relative to an angle measurement base; the first controller is connected with the angle sensor and used for disconnecting the electric connection between the winding of the rotor and the power grid when the current deflection angle is a target deflection angle; the second controller is connected with the doubly-fed converter, the calculating module is connected with the second controller, the calculating module is used for calculating an absolute value of a difference between the current deflection angle and the target deflection angle, and the second controller is used for controlling the excitation quantity of the doubly-fed converter to the rotor so as to realize that the excitation quantity is reduced along with the reduction of the absolute value.

In the wind turbine generator set, the wind turbine generator set further comprises an impeller locking device; wherein: the impeller locking device is arranged on the transmission system and used for locking the rotation of the transmission system.

In the wind turbine generator, the stator winding short-circuit device or/and the stator power grid breaking device is a circuit breaker or a contactor.

In the wind turbine generator, the stator winding short-circuit device comprises a detection unit and an execution unit; wherein:

the detection unit is used for detecting the working state of the stator power grid breaking device; the working state comprises a disconnection state and a connection state;

and the execution unit is connected with the detection unit and is used for short-circuiting each winding of the stator when the detection result of the detection unit is in an off state.

In the wind turbine generator, the stator winding short-circuit device further comprises an alarm unit; wherein: and the alarm unit is connected with the detection unit and used for giving an alarm when the detection result of the detection unit is in a connection state.

In the wind turbine generator, the wind turbine generator is a double-fed wind turbine generator.

Of course, the embodiment can also be used for predicting wind turbines with other structures. Here, details are not repeated, and the fault early warning method according to the embodiment of the present application is described below.

Referring to fig. 1, a gearbox oil temperature fault early warning method based on a multilayer perceptive neural network includes the following steps:

(1) collecting SCADA data of a fan, inputting the data, analyzing the related characteristics and related characteristic data of the oil temperature of a fan gear box, extracting the data characteristics, designing a database required by data screening conditions, and storing the screening conditions according to a required format;

(2) analyzing and screening the acquired data, and removing data which is irrelevant to the oil temperature fault analysis of the gearbox, so that a cleaner data set can be obtained;

(3) establishing a gearbox oil temperature fault analysis model by utilizing a multilayer perception neural network, and inputting the cleaned data into the model for model training;

(4) applying the trained model to other analyses to be predicted or tested, obtaining a prediction result of the algorithm through model calculation, and determining that the analysis needs to be retested through judgment conditions;

(5) and judging the accuracy of the model by using the residual error of the actual value of the fan and the algorithm predicted value, and if the fault prediction model is accurate, outputting the fault diagnosis probability by using the probability distribution of the residual error.

And (2) collecting the SCADA data of the fan in the step (1) through an SCADA system of the fan, wherein the SCADA system of the fan is used for predicting the running state of the fan in real time.

And (2) comprehensively recording each predicted point of the wind turbine generator at intervals by the SCADA system of the fan in the step (1), and basically reflecting the real running condition of the fan through data generated by the SCADA system.

The method for analyzing and screening the collected data in the step (2) is a condition judgment and decision tree algorithm.

The base data of the SCADA received in the step (2) contains too much information, some of the base data do not influence and remove irrelevant data for judging the oil temperature fault of the gearbox, the construction of an analysis model of the gearbox is facilitated, and the removal of the irrelevant data enables the judgment of the fault result of the gearbox to be more accurate.

The multilayer perception neural network in the step (3) is a back propagation training algorithm and a feedback training algorithm, and is a gradient descent method using reverse automatic derivation, and a residual probability model with high prediction accuracy can be obtained through verification and iteration of a training set and a test set.

The training step in the step (3) is as follows: 1) calculating the feedforward output of each neuron at each connection layer, 2) calculating the output error of the network and calculating the contribution degree of each neuron of the last layer to the error, 3) continuing to calculate the contribution degree of each neuron of the last layer to the error, 4) until the algorithm reaches the input layer.

And (4) obtaining a predicted value in the model, then obtaining a real value in other fans, and then calculating a residual error, wherein the calculation of the residual error is to subtract the real value from the predicted value.

And (4) determining whether the calculation result meets a judgment condition according to the obtained residual error, performing the next fault diagnosis if the judgment condition is met, and restarting from the step (2) if the judgment condition is not met.

And (5) making a diagnosis result and the pre-mapping into a bar graph, so that the implementation effect of the model can be quickly known through the bar graph.

The application has the advantages that:

1. the invention researches and develops the oil temperature fault early warning method of the wind driven generator gearbox through a data screening algorithm and a multilayer perception neural network algorithm, and provides a specific implementation mode.

It is well within the skill of those in the art to implement, without undue experimentation, the present application is not directed to software and process improvements, as they relate to circuits and electronic components and modules.

The above description is only a preferred embodiment of the present application and is not intended to limit the present application, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, improvement and the like made within the spirit and principle of the present application shall be included in the protection scope of the present application.