Variable working condition abnormal data processing system and method for electric wheel reducer of mine car

Abstract

The application provides a variable working condition abnormal data processing system and method for a mine car electric wheel reducer, comprising a data acquisition device (100) arranged on the mine car electric wheel and used for acquiring mine car electric wheel running state data A and surrounding environment working condition data T, a data classification device (200) used for classifying the running state data A according to time effect and giving a suspicious value, a data processing device used for judging and processing abnormal data F in processing data B and processing data C through a GAN network model, and a display device (400) used for displaying analysis results to a user. The variable working condition abnormal data processing system and method for the mine car electric wheel reducer effectively ensure the accuracy of data and improve the authenticity of fault diagnosis by cleaning abnormal data under a variable working condition.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of abnormal data processing of electric wheel reduction gear, in particular to a variable working condition abnormal data processing system and method for mine car electric wheel reduction gear. BACKGROUND

[0002] As one of the most important assembly parts of mine cars, electric wheels have high cost and high repair cost. The electric wheel reduction gear currently uses an offline periodic test method for monitoring, which has a large gap period. If the overhaul is not timely, it is easy to cause bearing damage and gear tooth damage and other serious mechanical accidents, resulting in large losses. In order to ensure safety, the overhaul cycle is shortened, which increases the cost.

[0003] Variable working condition refers to other working conditions that do not belong to the design working condition. During the variable working condition process, abnormal data is usually generated, but most of the abnormal data is abnormal. The generation of such data will interfere with the fault diagnosis of the electric wheel, and in severe cases, it will cause diagnostic errors. How to clean these abnormal data in time is crucial to the accuracy of the fault diagnosis of the equipment.

[0004] Application No. CN202311060547.6 discloses a mechanical rotating equipment variable working condition anomaly detection method based on feature alignment residual GAN. The method comprises: constructing a time-frequency graph of rotating machinery data, using an encoder GE(x) and a decoder GD(x) to respectively combine a residual module ResNet for time-frequency graph reconstruction, using a discriminator D to distinguish between reconstructed data and source data to determine abnormal data, and the loss function of the network contains an adversarial loss Ladv, a context loss Lcon of generator reconstruction data error, an encoder loss Lenc, and a Wasserstein distance loss Lw-gan of feature value alignment constructed by ranking odd-even data classification. SUMMARY

[0005] The present application is to overcome the technical problem of the influence of abnormal data generated by variable working condition on the accuracy of fault diagnosis in the fault diagnosis process of the electric wheel reduction gear in the prior art. A variable working condition abnormal data processing system and method for mine car electric wheel reduction gear is provided. By cleaning the abnormal data under the variable working condition environment, the accuracy of the data is effectively guaranteed, and the authenticity of the fault diagnosis is improved.

[0006] The present application provides a variable working condition abnormal data processing system for mine car electric wheel reduction gear, comprising:

[0007] The data acquisition device is arranged on the mine car electric wheel and is used for acquiring mine car electric wheel running state data A and surrounding environment working condition data T, and is used for transmitting the running state data A and the working condition data T to the data classification device.

[0008] The data classification device is used for receiving the operation state data A and the working condition data T transmitted by the data acquisition device, for classifying the operation state data A into operation data A1 irrelevant to the change of time t and time effect data A2 related to the time t according to the time effect, for assigning the operation data A1 to a suspicious value S1 to generate processing data B, for assigning the time effect data A2 to a suspicious value S2 to generate processing data C, and for transmitting the processing data B, the processing data C and the working condition data T to the data processing device; the suspicious value S1 and the suspicious value S2 are calculated as follows:

[0009] S1 = f (t) * f (a, E)

[0010] S2 = g (t) * g (b, E)

[0011] Wherein, f (t) and g (t) are time constant curves, f (a, E) and g (b, E) are functions related to the measured data, E is the measured data value, and a and b are deviation coefficients of the measured data and the theoretical data.

[0012] The data processing device is used for receiving the processing data B, the processing data C and the working condition data T transmitted by the data classification device, for calculating a working condition parameter curve S according to the working condition data, for determining whether a working condition change occurs according to the working condition parameter curve S, for training a GAN network model through processing data D with the suspicious value S2 less than a set threshold value in the case of the working condition change, for judging abnormal data F in the processing data B and the processing data C through the GAN network model, for analyzing the abnormal data F and generating an analysis result, and for transmitting the analysis result to the display device.

[0013] The display device is used for receiving the analysis result transmitted by the data processing device and displaying the analysis result to the user.

[0014] The abnormal data processing system for the variable working condition of the mine car electric wheel reduction machine provided by the application comprises the data acquisition device, the data classification device, the data processing device and the display device.

[0015] The application provides an abnormal data processing method for a variable working condition of a mine car electric wheel reduction machine, comprising the following steps:

[0016] S1, data acquisition: the data acquisition device acquires the operation state data A and the working condition data T of the mine car electric wheel in real time.

[0017] S2, data classification: the data classification device classifies the running state data into running data A1 irrelevant to time t change and time-dependent data A2 according to time effect;

[0018] S3, suspicious value assignment: the data classification device assigns running data A1 and time-dependent data A2 with suspicious value S1 and suspicious value S2, respectively;

[0019] S4, variable working condition judgment: the data processing device judges whether a variable working condition behavior occurs according to working condition data T, if yes, step S5 is entered, if no, step S7 is entered;

[0020] S5, abnormal data judgment: the data processing device screens out abnormal data F through a GAN network model;

[0021] S6, abnormal data confirmation: the data processing device cleans abnormal data F according to suspicious value S of abnormal data F and generates an analysis report;

[0022] S7, abnormal data processing system normal operation.

[0023] The variable working condition abnormal data processing method for the electric wheel reducer of the mine car, as a preferred mode, step S4 further comprises the following steps:

[0024] S41, calculating the working condition parameter curve S according to working condition data T

[0025] S=cxf(t)xf(T,P,rh,v)+d

[0026] Wherein, f(t) is a function related to time, f(T,P,rh,v) is a function related to temperature, air pressure, humidity and wind speed, c is an influence factor, and d is a correction deviation;

[0027] S42, variable working condition behavior judgment: the derivative of the working condition parameter curve S is calculated

[0028]

[0029] If S' is greater than the variable working condition threshold, it is determined that the variable working condition behavior occurs.

[0030] The variable working condition abnormal data processing method for the electric wheel reducer of the mine car, as a preferred mode, the data processing device in step S6 can correct suspicious value S of abnormal data F according to historical data or data of other monitoring points, and judge whether the abnormal data F is real abnormal data.

[0031] The abnormal data processing method for variable working condition of the electric wheel reducer of the mine car, as a preferred mode, the data processing device carries out fault diagnosis on the real abnormal data and carries out data cleaning on the non-real abnormal data.

[0032] The abnormal data processing method for variable working condition of the electric wheel reducer of the mine car, as a preferred mode, further comprises the following steps:

[0033] S8, GAN network model training: normal data E under variable working condition can be used for training the GAN network model, and the normal data E:

[0034] E=D-F U D.

[0035] Compared with the prior art, the present application has the following advantages:

[0036] (1) The present application assigns a suspicious value S to the running state data A, which is used to verify the abnormal data screened by the GAN network model, and can effectively complete data cleaning.

[0037] (2) The present application combines different historical data and different collected data under similar working condition environment to make secondary determination on abnormal data, improves the accuracy of abnormal data cleaning, and ensures the normal operation of the fault diagnosis system.

[0038] (3) The present application trains the GAN network model by using the actual running data, and further improves the accuracy of abnormal data screening. BRIEF DESCRIPTION OF DRAWINGS

[0039] Figure 1 It is a variable working condition abnormal data processing system composition diagram for an electric wheel reducer of a mine car;

[0040] Figure 2 It is a variable working condition abnormal data processing system electric wheel outer side sensor installation position schematic diagram for an electric wheel reducer of a mine car;

[0041] Figure 3 It is a variable working condition abnormal data processing system electric wheel inner side sensor installation position schematic diagram for an electric wheel reducer of a mine car;

[0042] Figure 4 It is a variable working condition abnormal data processing system main generator sensor installation position schematic diagram for an electric wheel reducer of a mine car;

[0043] Figure 5 It is a variable working condition abnormal data processing method flow chart for an electric wheel reducer of a mine car.

[0044] REFERENCE NUMERALS:

[0045] 100, data acquisition device; 200, data classification device; 300, data processing device; 400, display device. DETAILED DESCRIPTION

[0046] 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. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments.

[0047] Embodiment 1

[0048] As shown in the figure, a variable working condition abnormal data processing system for a mine car electric wheel reducer includes: Figure 1

[0049] Data acquisition device 100: provided on the mine car electric wheel, used for collecting mine car electric wheel running state data A and surrounding environment working condition data T, used for transmitting running state data A and working condition data T to data classification device 200; as shown in the figure, each mine car includes at least 3 collection points, respectively provided on the outside of the electric motor, the inside of the electric motor (on the gear box) and the main generator; Figures 2 to 4

[0050] Data classification device 200: used for receiving running state data A and working condition data T transmitted by data acquisition device 100, used for dividing running state data A into running data A1 irrelevant to time t change and time effect data A2 related to time t according to time effect, used for assigning running data A1 to suspicious value S1 to generate processing data B, used for assigning time effect data A2 to suspicious value S2 to generate processing data C, used for transmitting processing data B, processing data C and working condition data T to data processing device 300; running data A1 includes real-time speed, deviation value of real-time speed and set speed, vibration frequency, amplitude, phase; time effect data A2 includes detection point travel, detection point travel deviation value, time domain feature, frequency domain feature, time-frequency domain feature, energy spectrum, time domain feature includes variance, mean square value, effective value, skewness and kurtosis, peak value index, margin index, waveform index, pulse index, frequency domain feature includes FFT spectrum, power spectrum, envelope spectrum;

[0051] Suspicious value S1 and suspicious value S2 are calculated as follows:

[0052] S1=f(t)×f(α,E)

[0053] S2=g(t)×g(β,E)

[0054] Wherein, f(t), g(t) are time constant curves, f(α,E), g(β,E) are functions related to measured data, E is the measured data value α, β is the deviation coefficient of measured data and theoretical data; ​​

[0055] The data processing device 300 is configured to receive the processing data B and the processing data C and the working condition data T transmitted by the data classification device 200, to calculate a working condition parameter curve S according to the working condition data, to determine whether a working condition change occurs according to the working condition parameter curve S, to train a GAN network model by using the processing data D whose suspicious value S2 is less than a set threshold in the case of the working condition change, to determine abnormal data F in the processing data B and the processing data C by using the GAN network model, to analyze the abnormal data F and generate an analysis result, and to transmit the analysis result to the display device 400.

[0056] The display device 400 is configured to receive the analysis result transmitted by the data processing device 300 and display the analysis result to a user.

[0057] As shown in Figure 5 The embodiment provides a working condition change abnormal data processing method for a mine car electric wheel reducer.

[0058] S1, data acquisition: the data acquisition device 100 acquires the running state data A and the working condition data T of the mine car electric wheel in real time.

[0059] S2, data classification: the data classification device 200 classifies the running state data A into running data A1 irrelevant to time t and time-dependent data A2 according to time effect.

[0060] S3, suspicious value assignment: the data classification device 200 assigns the running data A1 and the time-dependent data A2 with suspicious values S1 and S2 respectively.

[0061] S4, working condition change judgment: the data processing device 300 judges whether a working condition change occurs according to the working condition data T, if yes, the step S5 is entered, if no, the step S7 is entered; the step S4 further includes the following steps.

[0062] S41, calculating the working condition parameter curve S according to the working condition data T

[0063] S=cxf(t)xf(T,P,rh,v)+d

[0064] Wherein, f(t) is a function related to time, f(T,P,rh,v) is a function related to temperature, air pressure, humidity and wind speed, c is an influence factor, and d is a correction deviation.

[0065] S42, working condition change behavior judgment: the working condition parameter curve S is derived

[0066]

[0067] If S' is greater than the variable working condition threshold, it is determined that the variable working condition behavior occurs;

[0068] S5, Abnormal data judgment: The data processing device 300 screens out abnormal data F through the GAN network model;

[0069] S6, Abnormal data confirmation: The data processing device 300 cleans and generates an analysis report according to the abnormal data F according to the suspicious value S of the abnormal data F; The data processing device 300 can correct the suspicious value S of the abnormal data F according to the historical data or the data of other monitoring points, judge whether the abnormal data F is real abnormal data, the data processing device 300 carries out fault diagnosis to the real abnormal data, carries out data cleaning to the non-real abnormal data;

[0070] S7, Abnormal data processing system normal operation;

[0071] S8, GAN network model training: Normal data E under variable working condition can be used for training GAN network model, normal data E:

[0072] E=D-F∪D.

[0073] The above is only the preferred specific embodiment of the present application, but the protection scope of the present application is not limited to this, any person skilled in the art in the technical range disclosed by the present application, according to the technical scheme and the invention concept of the present application, equivalent replacement or change, should be covered in the protection scope of the present application.

Claims

1. A variable operating condition abnormal data processing system for an electric motor wheel reduction gear of a mine car, characterized by: The application relates to a data processing method for mining electric wheels, which comprises the following steps: A data acquisition device (100) is arranged on a mining electric wheel and is used for acquiring running state data A of the mining electric wheel and surrounding environment working condition data T, and is used for transmitting the running state data A and the working condition data T to a data classification device (200); The data classification device (200) is used for receiving the running state data A and the working condition data T transmitted by the data acquisition device, is used for classifying the running state data A into running data A1 irrelevant to time t change and time-effect data A2 related to the time t according to time effect, is used for assigning the running data A1 to suspicious value S1 to generate processing data B, is used for assigning the time-effect data A2 to suspicious value S2 to generate processing data C, and is used for transmitting the processing data B, the processing data C and the working condition data T to a data processing device (300); the suspicious value S1 and the suspicious value S2 are calculated as follows: S1=f(t)xf(alpha, E) S2=g(t) x g(beta, E) Wherein, f(t) and g(t) are time constant curves, f(alpha, E) and g(beta, E) are functions related to measured data, E is a measured data value, and alpha and beta are deviation coefficients of measured data and theoretical data; The data processing device (300) is used for receiving the processing data B, the processing data C and the working condition data T transmitted by the data classification device (200), is used for calculating a working condition parameter curve S according to the working condition data, is used for determining whether a working condition change occurs according to the working condition parameter curve S, is used for training a GAN network model through processing data D of the suspicious value S2 being smaller than a set threshold value in the case of the working condition change, is used for judging abnormal data F in the processing data B and the processing data C through the GAN network model, is used for analyzing the abnormal data F and generating an analysis result, and is used for transmitting the analysis result to a display device (400); The display device (400) is used for receiving the analysis result transmitted by the data processing device and displaying the analysis result to a user.

2. The variable operating condition abnormal data processing system for a mine car electric wheel reduction gear according to claim 1, characterized in that: The running data A1 comprises real-time rotating speed, a deviation value of real-time rotating speed and a set rotating speed, vibration frequency, amplitude and phase; the time-effect data A2 comprises detection point travel, a detection point travel deviation value, time domain characteristics, frequency domain characteristics, time-frequency domain characteristics and energy spectrum; the time domain characteristics comprise variance, mean square value, effective value, skewness and kurtosis, peak value index, margin index, waveform index and pulse index; the frequency domain characteristics comprise FFT spectrum, power spectrum and envelope spectrum.

3. The variable working condition abnormal data processing method for the electric wheel reducer of the mine car according to any one of claims 1-2, characterized in that: The application further discloses a data processing method for mining electric wheels, which comprises the following steps: S1, data acquisition: the data acquisition device (100) acquires the running state data A and the working condition data T of the mining electric wheel in real time; S2, data classification: the data classification device (200) classifies the running state data A into the running data A1 irrelevant to time t change and the time-effect data A2 related to the time t according to time effect. S3. Assigning doubtful values: The data classification device (200) assigns doubtful values ​​S1 and doubtful values ​​S2 to the running data A1 and the time-sensitive data A2, respectively. S4. Change of working condition judgment: The data processing device (300) determines whether a change of working condition has occurred based on the working condition data T. If the judgment is yes, proceed to step S5; otherwise, proceed to step S7. S5. Abnormal data judgment: The data processing device (300) filters out the abnormal data F through the GAN network model; S6. Abnormal data confirmation: The data processing device (300) cleans the abnormal data F according to the doubt value S of the abnormal data F and generates an analysis report. S7. The abnormal data processing system is operating normally.

4. The variable working condition abnormal data processing method for the electric wheel reduction machine of the mine car according to claim 3, characterized in that: Step S4 further includes the following steps: S41. Calculate the operating condition parameter curve S based on the operating condition data T. S=c×f(t)×f(T,P,rh,v)+d Where f(t) is a time-dependent function, f(T,P,rh,v) is a function related to temperature, air pressure, humidity and wind speed, c is an influencing factor, and d is a correction bias; S42. Variable Operating Condition Behavior Judgment: Differentiate the operating condition parameter curve S. If S' is greater than the variable operating condition threshold, then the variable operating condition behavior is determined to have occurred.

5. The variable working condition abnormal data processing method for the electric wheel reduction machine of the mine car according to claim 3, characterized in that: In step S6, the data processing device (300) can correct the doubt value S of the abnormal data F based on historical data or data from other monitoring points, and determine whether the abnormal data F is real abnormal data.

6. The variable working condition abnormal data processing method for the electric wheel reduction machine of the mine car according to claim 5, characterized in that: The data processing device (300) performs fault diagnosis on real abnormal data and cleans up non-real abnormal data.

7. The variable working condition abnormal data processing method for the electric wheel reduction machine of the mine car according to claim 3, characterized in that: It also includes the following steps: S8. GAN Network Model Training: Normal data E under varying operating conditions can be used to train the GAN network model. Normal data E: E = DF∪D.

Images