RV reducer output disc rack with sound absorption and oil storage function

Abstract

The present invention relates to the technical field of RV reducers, and in particular to an RV reducer output disc rack with a sound-absorbing and oil-storing function, comprising a reducer housing, a needle roller steel wheel is arranged in the reducer housing, and a self-lubricating system is also arranged on the reducer housing for lubricating the needle roller steel wheel, and the self-lubricating system comprises: a storage tube, which is arranged on the reducer housing and is used to store lubricating oil, and the storage tube is composed of an arc portion and a collecting portion; a micro pump, which is fixed on the collecting portion through a bracket. The RV reducer output disc rack with a sound-absorbing and oil-storing function provided by the present invention realizes intelligent judgment of whether lubricating grease is lacking based on the working state of the RV reducer, and when it is lacking, it realizes automatic quantitative addition of lubricating grease without manual addition.

Description

Technical Field

[0001] The invention relates to the technical field of RV reducers, in particular to an RV reducer output disc frame with sound absorption and oil storage functions. Background Art

[0002] RV reducers (Rotary Vector reducers) are widely used in industrial robots, automation equipment, precision machine tools and other fields. They are characterized by high rigidity, high reduction ratio, high torque capacity and high transmission efficiency. The output disc frame of the RV reducer is an important component connecting the output shaft and the load, and usually needs to have good torsional rigidity and strength. However, in practical applications, RV reducers often face problems with noise and lubricant management.

[0003] A Chinese patent with publication number CN108253121A discloses a self-lubricating RV reducer, including a reducer housing and a crankshaft arranged in the reducer housing, and also includes an automatic greasing system; the automatic greasing system includes a grease storage barrel for storing grease, a nozzle for spraying grease into the crankshaft, a grease delivery pipe for connecting the grease storage barrel and the nozzle, an oil pump connected to the grease delivery pipe for driving the grease to flow, a drive motor for driving the oil pump and a controller for controlling the drive motor, a grease flow channel for lubricating the moving pair in the reducer housing is provided in the crankshaft, the nozzle is positioned on the reducer housing and the outlet end of the nozzle corresponds to the inlet end of the grease flow channel; the invention can automatically add grease.

[0004] As mentioned in the above application, although the existing RV reducer is equipped with a self-lubricating system when in use, the lubricating grease is generally added manually under control, and its intelligence level is low, that is, it is impossible to intelligently judge whether the lubricating grease is lacking based on the working status of the RV reducer, and automatically and quantitatively add the lubricating grease when it is lacking. Summary of the invention

[0005] In order to solve the above problems, the present invention provides an RV reducer output disc frame with sound absorption and oil storage functions.

[0006] The present invention adopts the following technical scheme: an RV reducer output disc frame with sound-absorbing and oil-storing functions comprises a reducer housing, a needle roller steel wheel is arranged in the reducer housing, and a self-lubricating system is also arranged on the reducer housing for lubricating the needle roller steel wheel. The self-lubricating system comprises:

[0007] A storage pipe, which is arranged on the reducer housing and is used to store lubricating oil, and the storage pipe is composed of an arc portion and a collecting portion;

[0008] A micro pump is mounted and fixed on the collecting part through a bracket;

[0009] An annular oil delivery cavity is provided in the reducer housing. A plurality of oil injection nozzles are provided between the annular oil delivery cavity and the needle roller steel wheel. Lubricating oil is injected into the preset arc groove on the needle roller steel wheel through the oil injection nozzles.

[0010] A controller, which is used to control the operation of the micro pump, and the controller comprises:

[0011] The data receiving module collects in advance the historical training data required for predicting the lubricating oil change of the RV reducer;

[0012] The data analysis module trains a machine learning model for predicting the real-time lubricant shortage based on historical training data, collects the working condition data and status data of the RV reducer during operation at preset unit time T, and predicts the lubricant shortage based on the trained machine learning model;

[0013] The data processing module controls the micro pump to pump the corresponding amount of lubricating oil into the annular oil delivery cavity based on the predicted amount of lubricating oil shortage, and then injects it into the needle steel wheel through the oil injection nozzle.

[0014] As a further description of the above technical solution: the historical training data is training data collected in an experimental environment, and the experimental environment is that the test personnel control the change of the amount of lubricating oil in the RV reducer to continuously collect the operating condition data and status data of the RV reducer during operation, the operating condition data includes load data and speed data, and the status data includes standard noise data and standard vibration data.

[0015] As a further description of the above technical solution: the method for collecting standard noise data in the state data of the RV reducer during operation at every preset unit time T includes:

[0016] Step P1: Set noise collection points at the housing, input shaft, and output disk of the RV reducer in advance, install high-sensitivity microphones at the noise collection points, and collect noise data at each noise collection point every T / 4;

[0017] Step P2: extracting the timbre features in each noise data, wherein the timbre features include fundamental frequency, harmonic content and amplitude envelope;

[0018] Step P3: inputting the fundamental frequency, harmonic content, amplitude envelope and noise data into a pre-trained noise recognition machine learning model, and outputting a noise type, wherein the noise type includes a lubricating oil lacking noise type and a lubricating oil not lacking noise type;

[0019] Step P4: marking the noise data output as the type of noise lacking lubricating oil as available noise data, and establishing a noise data set from the available noise data;

[0020] Step P5: preset a lubricating oil missing noise data decibel threshold, obtain the decibel values ​​of all available noise data in the set, and when the decibel value of the available noise data is greater than or equal to the lubricating oil missing noise data decibel threshold, remove the available noise data;

[0021] Step P6: Obtain all available noise data in the noise data set after removal, and calculate the average value as the standard noise data.

[0022] As a further description of the above technical solution: the method of collecting standard vibration data in the state data of the RV reducer during operation at every preset unit time T includes:

[0023] Step K1: Install acceleration sensors at the housing, input shaft, and output disc of the RV transmission, and collect vibration data through the acceleration sensors;

[0024] Step K2: Analyze the frequency spectrum of the vibration signal using fast Fourier transform and draw a spectrum diagram;

[0025] Step K3: By analyzing the increase value of the high-frequency component in the spectrum diagram and comparing it with the preset threshold, when the increase value of the high-frequency component is greater than or equal to the preset threshold, the vibration data is marked as available vibration data; when the increase value of the high-frequency component is less than the preset threshold, the vibration data is marked as unavailable vibration data.

[0026] Step K4: Calculate and obtain the average value of the available vibration data as standard vibration data.

[0027] As a further description of the above technical solution: the method of using fast Fourier transform to analyze the spectrum of the vibration signal and draw a spectrum diagram includes:

[0028] Convert the time domain signal into the frequency domain signal using Fast Fourier Transform;

[0029] Calculate the frequency axis to determine the frequency corresponding to each frequency component;

[0030] Plot a spectrogram to display the frequency content in a signal.

[0031] As a further description of the above technical solution: the training method of the noise recognition machine learning model includes:

[0032] Collecting a noise data set, the noise data set includes timbre characteristics and noise types of the noise data, wherein the timbre characteristics of the noise data include fundamental frequency, harmonic content and amplitude envelope, and the noise types include a lack of lubricating oil noise type and a non-lack of lubricating oil noise type;

[0033] Convert the timbre features of the collected noise data into a corresponding set of feature vectors;

[0034] Each group of feature vectors is used as the input of the noise recognition machine learning model. The noise recognition machine learning model takes the noise type corresponding to the timbre characteristics of each group of noise data as output, takes the noise type actually corresponding to the timbre characteristics of each group of noise data as the prediction target, and takes minimizing the loss function value of the noise recognition machine learning model as the training goal; training stops when the loss function value of the noise recognition machine learning model is less than or equal to the preset target loss value.

[0035] As a further description of the above technical solution: a first oil delivery pipe is connected between the micro pump and the collecting part, and a second oil delivery pipe is connected between the micro pump and the annular oil delivery cavity.

[0036] As a further description of the above technical solution: the arc portion is welded and fixed on the outer wall of the reducer housing, and the two ends of the arc portion are respectively connected to the two ends of the collecting portion, so that the arc portion and the internal space of the collecting portion form a closed storage space for storing lubricating oil, and an oil filling port is arranged at the top center of the arc portion, and a sealing cover is threadedly connected to the oil filling port.

[0037] As a further description of the above technical solution: output disks and disk rack bases are provided at both ends of the reducer housing, and the output disks and disk rack bases are connected together by screws to achieve connection and fixation with the reducer housing, and a protrusion is provided on the side of the output disk opposite to the disk rack base, and a threaded hole for matching with the screw is opened on the protrusion.

[0038] As a further description of the above technical solution: a flange is arranged on the outer side of the reducer housing, which is used for installing and fixing the reducer housing. The reducer housing is a cylindrical structure, and a sound-absorbing cotton layer is arranged on the reducer housing. Beneficial Effects

[0039] The RV reducer output disc rack with a sound-absorbing oil storage function provided by the present invention is provided with a self-lubricating system. The historical training data required for the change of the lubricating oil of the RV reducer is collected in advance, and a machine learning model for predicting the real-time lack of lubricating oil is trained based on the historical data. By collecting the working condition data and state data of the RV reducer during operation and inputting them into the trained machine learning model, the lack of lubricating oil can be predicted, and based on the predicted lack of lubricating oil, the micro pump is controlled to pump in the corresponding lack of lubricating oil for use, thereby realizing intelligent judgment of whether lubricating oil is lacking based on the working state of the RV reducer. When lacking, the lubricating oil is automatically and quantitatively added without manual addition;

[0040] Furthermore, by collecting noise data from multiple points on the RV transmission, it is ensured that the measurement point can represent the overall noise level of the transmission, and by using a pre-trained noise recognition machine learning model, it is determined whether the noise type is a noise type caused by lack of lubricating oil, thereby obtaining available noise data, and then by presetting the decibel threshold of lubricating oil missing noise data, the available noise data is screened and eliminated for a second time, and finally all available noise data in the eliminated noise data set are obtained, and the average value is calculated as the standard noise data, thereby ensuring the accuracy of the standard noise data, overcoming the influence of other external noise data on the collected standard noise data, and improving the accuracy of the machine learning model prediction of the real-time lubricating oil shortage;

[0041] Furthermore, vibration data is collected at multiple points to ensure that the measurement points can represent the overall vibration level of the transmission, and the frequency spectrum of the vibration signal is analyzed using fast Fourier transform (FFT) and a spectrum diagram is drawn. The increase in the high-frequency component in the spectrum diagram is compared with a preset threshold. When the increase in the high-frequency component is greater than or equal to the preset threshold, the vibration data is marked as available vibration data, that is, the vibration data is identified, and vibration data that is not caused by lack of lubricant is removed. Then, the average value of the available vibration data is calculated as the standard vibration data, thereby ensuring the accuracy of the standard vibration data, overcoming the influence of vibration caused by other reasons on the collected standard vibration data, and improving the accuracy of the machine learning model for predicting the real-time lack of lubricant. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The present invention will be further explained below in conjunction with the accompanying drawings and embodiments:

[0043] Figure 1 A schematic diagram of the disassembled structure of an RV reducer output disc frame with sound-absorbing and oil-storing functions provided in Example 1 of the present invention;

[0044] Figure 2 A front view of an output disc frame of an RV reducer with a sound-absorbing and oil-storing function provided in Example 1 of the present invention;

[0045] Figure 3 A schematic cross-sectional view of an RV reducer output disc frame with sound-absorbing and oil-storing functions provided in Example 1 of the present invention;

[0046] Figure 4 For the present invention Figure 3 A magnified view of area A in;

[0047] Figure 5 A schematic diagram of a controller provided by an embodiment of the present invention;

[0048] Figure 6 A flow chart of a method for collecting standard noise data provided in Example 2 of the present invention;

[0049] Figure 7 This is a flow chart of the method for obtaining standard vibration data provided in Example 3 of the present invention.

[0050] In the figure: 1. reducer housing; 11. sound-absorbing cotton layer; 2. output disk; 21. protrusion; 22. threaded hole; 3. disk rack base; 31. screw; 4. needle roller; 5. flange; 6. storage tube; 61. arc-shaped part; 611. oil filling port; 62. collecting part; 7. micro pump; 71. bracket; 72. first oil pipeline; 73. second oil pipeline; 8. annular oil delivery chamber; 81. oil filling nozzle; 9. controller. DETAILED DESCRIPTION

[0051] In order to make the technical means, creative features, objectives and effects of the present invention easy to understand, the present invention is further described below with reference to specific diagrams. It should be noted that the embodiments and features in the embodiments of the present application can be combined with each other without conflict.

[0052] Example 1

[0053] See also Figure 1-Figure 5 The embodiment of the present invention provides a technical solution: an RV reducer output disc rack with a sound-absorbing and oil-storing function, comprising a reducer housing 1, a needle steel wheel 4 is arranged in the reducer housing 1, a flange 5 is arranged on the outer side of the reducer housing 1, and the flange 5 is used for installing and fixing the reducer housing 1. The reducer housing 1 is a cylindrical structure, and a sound-absorbing cotton layer 11 is sleeved on the reducer housing 1. An output disc 2 and a disc rack base 3 are arranged at both ends of the reducer housing 1, and the output disc 2 and the disc rack base 3 are connected together by screws 31 to achieve connection and fixation with the reducer housing 1. A protrusion 21 is arranged on the side of the output disc 2 opposite to the disc rack base 3, and a threaded hole 22 used for matching the screw 31 is opened on the protrusion 21, wherein the output disc 2 and the disc rack base 3 are used to set and install the RV reducer crankshaft;

[0054] The reducer housing 1 is also provided with a self-lubricating system for lubricating the needle roller steel wheel 4. The self-lubricating system includes:

[0055] A storage pipe 6, which is arranged on the reducer housing 1 and is used to store lubricating oil. The storage pipe 6 is composed of an arc portion 61 and a collecting portion 62;

[0056] The micro pump 7 is mounted and fixed on the collecting portion 62 via a bracket 71;

[0057] An annular oil delivery chamber 8 is provided in the reducer housing 1. A plurality of oil injection nozzles 81 are provided between the annular oil delivery chamber 8 and the needle roller steel wheel 4. Lubricating oil is injected into the preset arc groove on the needle roller steel wheel 4 through the oil injection nozzles 81. A first oil delivery pipe 72 is connected between the micro pump 7 and the collecting portion 62. A second oil delivery pipe 73 is connected between the micro pump 7 and the annular oil delivery chamber 8.

[0058] The arc portion 61 is welded and fixed on the outer wall of the reducer housing 1, and the two ends of the arc portion 61 are respectively connected to the two ends of the collecting portion 62, so that the internal space of the arc portion 61 and the collecting portion 62 forms a closed storage space for storing lubricating oil. An oil filling port 611 is provided at the top center of the arc portion 61, and a sealing cover is threadedly connected to the oil filling port 611.

[0059] The controller 9 is used to control the operation of the micro pump 7. The controller 9 includes:

[0060] The data receiving module collects in advance the historical training data required for predicting the lubricating oil change of the RV reducer;

[0061] The data analysis module trains a machine learning model for predicting the real-time lubricant shortage based on historical training data, collects the working condition data and status data of the RV reducer during operation at a preset unit time T (preferably T is 4 minutes), and predicts the lubricant shortage based on the trained machine learning model;

[0062] Methods for training a machine learning model to predict real-time lubricant shortage include:

[0063] Convert the collected working condition data and state data of the RV reducer into a corresponding set of feature vectors;

[0064] Each set of feature vectors is used as the input of the machine learning model for predicting the real-time lubricant shortage amount. The machine learning model for predicting the real-time lubricant shortage amount uses the lubricant shortage amount corresponding to the working condition data and status data of each set of RV reducers as the output, uses the lubricant shortage amount actually corresponding to the working condition data and status data of each set of RV reducers as the prediction target, and minimizes the loss function value of the machine learning model as the training target; when the loss function value of the machine learning model is less than or equal to the preset target loss value, the training is stopped;

[0065] The machine learning model loss function value is the mean square error;

[0066] Mean square error is one of the commonly used loss functions. By transforming the loss function formula Minimization is used as the goal to train the model so that the machine learning model better fits the data, thereby improving the performance and accuracy of the model;

[0067] In the loss function is the loss function value of the machine learning model, is the feature vector group number; is the number of eigenvector groups; For the The amount of lubricant shortage corresponding to the group eigenvector, For the The lubricant shortage amount corresponding to the group feature vector in real time;

[0068] Other model parameters of the machine learning model, target loss value, optimization algorithm, training set test set validation set ratio, and loss function optimization are all achieved through actual engineering implementation and continuous experimental tuning.

[0069] The data processing module controls the micro pump 7 to pump the corresponding amount of lubricating oil into the annular oil delivery cavity 8 based on the predicted amount of lubricating oil shortage, and then injects it into the needle steel wheel 4 through the oil injection nozzle 81.

[0070] The historical training data is the training data collected in the experimental environment. The experimental environment is that the test personnel control the changes in the amount of lubricating oil in the RV reducer to continuously collect the working condition data and status data of the RV reducer during operation. The working condition data includes load data and speed data, and the status data includes standard noise data and standard vibration data.

[0071] It should be noted that the load data is obtained by installing a torque sensor on the output shaft, that is, by measuring the torque on the output shaft, thereby indirectly reflecting the load condition, and the speed data can be accurately measured by installing a Hall effect sensor.

[0072] It should be noted that as the amount of lubricating oil in the RV transmission decreases, the noise and vibration of the RV transmission will become louder and louder when it is working.

[0073] Specifically, the RV reducer output disc rack with sound-absorbing and oil storage function is provided with a self-lubricating system. By pre-collecting the historical training data required for the change of the lubricating oil of the RV reducer, and training a machine learning model for predicting the real-time lubricating oil shortage based on the historical data, by collecting the working condition data and status data of the RV reducer during operation, and inputting them into the trained machine learning model, the lack of lubricating oil can be predicted, and based on the predicted lack of lubricating oil, the micro pump 7 is controlled to pump in the corresponding lack of lubricating oil for use, thereby realizing intelligent judgment of whether there is a lack of lubricating grease based on the working state of the RV reducer. When it is lacking, the lubricating grease is automatically and quantitatively added without manual addition.

[0074] Example 2

[0075] Reference Figure 5 and Figure 6The method for collecting standard noise data from the state data of the RV reducer during operation at every preset unit time T includes:

[0076] Step P1: Pre-set noise collection points at the housing, input shaft, and output disk 2 of the RV reducer, install high-sensitivity microphones at the noise collection points, and collect noise data at each noise collection point every T / 4;

[0077] Step P2: extracting the timbre features in each noise data, the timbre features including fundamental frequency, harmonic content and amplitude envelope;

[0078] It should be noted that the fundamental frequency is the lowest frequency component of the sound, which usually represents the pitch. The harmonics are frequency components that are integer multiples of the fundamental frequency. The harmonic content can be determined by analyzing the spectrum. The amplitude envelope represents the overall amplitude change trend of the signal, which can be obtained using peak detection.

[0079] It should be noted that the timbre characteristics of noise data generated by different reasons in the RV reducer will be different. By obtaining the timbre characteristics in the noise data, the type of noise can be obtained, thereby determining whether the noise is caused by lack of lubricating oil.

[0080] Step P3: input the fundamental frequency, harmonic content, amplitude envelope and noise data into the pre-trained noise recognition machine learning model, and output the noise type, which includes the noise type of lack of lubricating oil and the noise type of not lack of lubricating oil;

[0081] The training methods for the noise recognition machine learning model include:

[0082] Collecting a noise data set, the noise data set includes timbre characteristics and noise types of the noise data, wherein the timbre characteristics of the noise data include fundamental frequency, harmonic content and amplitude envelope, and the noise types include a lack of lubricating oil noise type and a non-lack of lubricating oil noise type;

[0083] Convert the timbre features of the collected noise data into a corresponding set of feature vectors;

[0084] Each group of feature vectors is used as the input of the noise recognition machine learning model. The noise recognition machine learning model takes the noise type corresponding to the timbre characteristics of each group of noise data as output, takes the noise type actually corresponding to the timbre characteristics of each group of noise data as the prediction target, and takes minimizing the loss function value of the noise recognition machine learning model as the training goal; training stops when the loss function value of the noise recognition machine learning model is less than or equal to the preset target loss value.

[0085] The noise recognition machine learning model is any one of a deep neural network model or a deep belief network model;

[0086] It should be noted that in the noise recognition machine learning model, the calculation formula for prediction error is: ,in, is the number of the characteristic data, is the prediction error, For the The predicted noise type corresponding to the group feature data, For the The actual noise type corresponding to the group feature data.

[0087] Step P4: marking the noise data output as the type of noise lacking lubricating oil as available noise data, and establishing a noise data set from the available noise data;

[0088] Step P5: preset a lubricating oil missing noise data decibel threshold, obtain the decibel values ​​of all available noise data in the set, and when the decibel value of the available noise data is greater than or equal to the lubricating oil missing noise data decibel threshold, remove the available noise data;

[0089] Step P6: Obtain all available noise data in the noise data set after removal, and calculate the average value as the standard noise data.

[0090] Specifically, based on the above embodiments, this embodiment further discloses a method for obtaining standard noise data. This application first collects noise data from multiple points on the RV transmission to ensure that the measurement points can represent the overall noise level of the transmission, and uses a pre-trained noise recognition machine learning model to determine whether the noise type is a lubricant shortage noise type, thereby obtaining available noise data. Then, the available noise data is secondary screened and eliminated by presetting the decibel threshold of lubricant shortage noise data. Finally, all available noise data in the eliminated noise data set are obtained, and the average value is calculated as the standard noise data, thereby ensuring the accuracy of the standard noise data, overcoming the influence of other external noise data on the collected standard noise data, and improving the accuracy of the machine learning model prediction of the real-time lubricant shortage amount.

[0091] Example 3

[0092] Reference Figure 5 and Figure 7 , based on the above embodiments, this embodiment further discloses a method for obtaining standard vibration data;

[0093] The method for collecting standard vibration data in the state data of the RV reducer during operation at every preset unit time T includes:

[0094] Step K1: Install acceleration sensors at the housing, input shaft, and output disc 2 of the RV transmission, and collect vibration data through the acceleration sensors;

[0095] Step K2: Analyze the frequency spectrum of the vibration signal using fast Fourier transform (FFT) and draw a spectrum diagram;

[0096] Methods for analyzing the frequency spectrum of vibration signals and drawing spectrum graphs using Fast Fourier Transform (FFT) include:

[0097] Convert the time domain signal into the frequency domain signal using Fast Fourier Transform;

[0098] Calculate the frequency axis to determine the frequency corresponding to each frequency component;

[0099] Plot a spectrogram to display the frequency content in a signal.

[0100] Step K3: By analyzing the increase value of the high-frequency component in the spectrum diagram and comparing it with the preset threshold, when the increase value of the high-frequency component is greater than or equal to the preset threshold, the vibration data is marked as available vibration data; when the increase value of the high-frequency component is less than the preset threshold, the vibration data is marked as unavailable vibration data.

[0101] It should be noted that when lubricating oil is lacking, the high-frequency components in the spectrum will increase significantly due to increased friction, that is, the increase in the high-frequency components can be used to determine whether the vibration data is caused by the lack of lubricating oil.

[0102] Step K4: Calculate and obtain the average value of the available vibration data as standard vibration data.

[0103] Specifically, in this implementation, vibration data is first collected at multiple points to ensure that the measurement points can represent the overall vibration level of the transmission, and the frequency spectrum of the vibration signal is analyzed by using Fast Fourier Transform (FFT) and a spectrum diagram is drawn. The increase in the high-frequency component in the spectrum diagram is compared with a preset threshold. When the increase in the high-frequency component is greater than or equal to the preset threshold, the vibration data is marked as available vibration data, that is, the vibration data is identified, and vibration data that is not caused by lack of lubricant is removed. Then, the average value of the available vibration data is calculated as the standard vibration data, thereby ensuring the accuracy of the standard vibration data, overcoming the influence of vibration caused by other reasons on the collected standard vibration data, and improving the accuracy of the machine learning model for predicting the real-time lack of lubricant.

[0104] The basic principles, main features and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and the description in the specification are only to illustrate the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. The RV reducer output disc frame with sound-absorbing and oil-storing function is characterized by: The invention comprises a reducer housing (1), wherein a needle roller steel wheel (4) is arranged in the reducer housing (1), and a self-lubricating system is also arranged on the reducer housing (1) for lubricating the needle roller steel wheel (4), wherein the self-lubricating system comprises: A storage tube (6) disposed on the reducer housing (1) and used for storing lubricating oil, wherein the storage tube (6) is composed of an arc portion (61) and a collecting portion (62); A micro pump (7) mounted and fixed on the collecting portion (62) via a bracket (71); An annular oil delivery cavity (8) is provided in the reducer housing (1); a plurality of oil injection nozzles (81) are provided between the annular oil delivery cavity (8) and the needle roller steel wheel (4); and lubricating oil is injected into a preset arc groove on the needle roller steel wheel (4) through the oil injection nozzles (81); A controller (9) is used to control the operation of the micro pump (7), wherein the controller (9) comprises: The data receiving module collects in advance the historical training data required for predicting the change of the lubricating oil of the RV reducer; The data analysis module trains a machine learning model for predicting the real-time lubricant shortage based on historical training data, collects the working condition data and status data of the RV reducer during operation at preset unit time T, and predicts the lubricant shortage based on the trained machine learning model; The data processing module controls the micro pump (7) to pump the corresponding amount of lubricating oil missing into the annular oil delivery cavity (8) based on the predicted amount of lubricating oil missing, and then injects the lubricating oil into the needle steel wheel (4) through the oil injection nozzle (81); The historical training data is pre-collected training data. By controlling the change of the amount of lubricating oil in the RV reducer, the working condition data and state data of the RV reducer are continuously collected. The working condition data includes load data and speed data, and the state data includes standard noise data and standard vibration data. The method for collecting standard noise data in the state data of the RV reducer during operation at every preset unit time T includes: Step P1: Pre-set noise collection points at the housing, input shaft, and output disk (2) of the RV reducer, install high-sensitivity microphones at the noise collection points, and collect noise data at each noise collection point every T / 4; Step P2: extracting the timbre features in each noise data, wherein the timbre features include fundamental frequency, harmonic content and amplitude envelope; Step P3: inputting the fundamental frequency, harmonic content, amplitude envelope and noise data into a pre-trained noise recognition machine learning model, and outputting a noise type, wherein the noise type includes a lubricating oil lacking noise type and a lubricating oil not lacking noise type; Step P4: marking the noise data output as the type of noise lacking lubricating oil as available noise data, and establishing a noise data set from the available noise data; Step P5: preset a lubricating oil missing noise data decibel threshold, obtain the decibel values ​​of all available noise data in the set, and when the decibel value of the available noise data is greater than or equal to the lubricating oil missing noise data decibel threshold, remove the available noise data; Step P6: Obtain all available noise data in the noise data set after removal, and calculate the average value as the standard noise data; The method for collecting standard vibration data in the state data of the RV reducer during operation at every preset unit time T includes: Step K1: Install acceleration sensors at the housing, input shaft, and output disc (2) of the RV transmission, and collect vibration data through the acceleration sensors; Step K2: Analyze the frequency spectrum of the vibration signal using fast Fourier transform and draw a spectrum diagram; Step K3: By analyzing the increase value of the high-frequency component in the spectrum diagram and comparing it with the preset threshold, when the increase value of the high-frequency component is greater than or equal to the preset threshold, the vibration data is marked as available vibration data; when the increase value of the high-frequency component is less than the preset threshold, the vibration data is marked as unavailable vibration data; Step K4: Calculate and obtain the average value of the available vibration data as standard vibration data.

2. The RV reducer output disc frame with sound-absorbing and oil-storing function according to claim 1 is characterized in that: Methods for using fast Fourier transform to analyze the spectrum of vibration signals and draw spectrum graphs include: Convert the time domain signal into the frequency domain signal using Fast Fourier Transform; Calculate the frequency axis to determine the frequency corresponding to each frequency component; Plot a spectrogram to display the frequency content in a signal.

3. The RV reducer output disc frame with sound-absorbing and oil-storing function according to claim 1 is characterized in that: The training methods for the noise recognition machine learning model include: Collecting a noise data set, the noise data set includes timbre characteristics and noise types of the noise data, wherein the timbre characteristics of the noise data include fundamental frequency, harmonic content and amplitude envelope, and the noise types include a lack of lubricating oil noise type and a non-lack of lubricating oil noise type; Convert the timbre features of the collected noise data into a corresponding set of feature vectors; Each group of feature vectors is used as the input of the noise recognition machine learning model. The noise recognition machine learning model takes the noise type corresponding to the timbre characteristics of each group of noise data as output, takes the noise type actually corresponding to the timbre characteristics of each group of noise data as the prediction target, and takes minimizing the loss function value of the noise recognition machine learning model as the training goal; training stops when the loss function value of the noise recognition machine learning model is less than or equal to the preset target loss value.

4. The RV reducer output disc frame with sound absorption and oil storage function according to claim 1 is characterized in that: A first oil delivery pipe (72) is connected between the micro pump (7) and the collecting portion (62), and a second oil delivery pipe (73) is connected between the micro pump (7) and the annular oil delivery chamber (8).

5. The RV reducer output disc frame with sound absorption and oil storage function according to claim 1 is characterized in that: The arc-shaped portion (61) is welded and fixed to the outer wall of the reducer housing (1); the two ends of the arc-shaped portion (61) are respectively connected to the two ends of the collecting portion (62), so that the internal space of the arc-shaped portion (61) and the collecting portion (62) forms a closed storage space for storing lubricating oil; an oil filling port (611) is provided at the center of the top of the arc-shaped portion (61), and a sealing cover is threadedly connected to the oil filling port (611).

6. The RV reducer output disc frame with sound-absorbing and oil-storing function according to claim 1 is characterized in that: An output disk (2) and a disk rack base (3) are provided at both ends of the reducer housing (1), and the output disk (2) and the disk rack base (3) are connected together by screws (31) to achieve connection and fixation with the reducer housing (1); a protrusion (21) is provided on a surface of the output disk (2) opposite to the disk rack base (3), and a threaded hole (22) for matching with the screw (31) is provided on the protrusion (21).

7. The RV reducer output disc frame with sound-absorbing and oil-storing function according to claim 1 is characterized in that: The reducer housing (1) is provided with a flange (5) on the outside thereof, which is used for installing and fixing the reducer housing (1); the reducer housing (1) is a cylindrical structure, and a sound-absorbing cotton layer (11) is provided on the reducer housing (1).

Images