An adaptive viscosity intelligent lubrication method and system applied to an RV reducer and a structure thereof

Abstract

The present application relates to the technical field of RV reducer, and particularly relates to an adaptive viscosity intelligent lubrication method and system applied to an RV reducer and a structure thereof, the method comprising: according to geometric structure characteristics and friction pair distribution inside the RV reducer, dividing a lubrication space into P functional partitions; coating an oil film partition micro-grid layer on an inner wall of a cam follower cavity, the micro-grid layer being composed of a magnetic guiding microstructure and a fluid microchannel, the fluid microchannel being internally filled with a lubricating liquid containing magnetic response particles; and arranging a local magnetic field modulation unit under each functional partition, and adjusting the viscosity of the lubricating liquid in the fluid microchannel by controlling the local magnetic field intensity. According to the friction pair distribution inside the reducer, the lubrication space is divided into P functional partitions, so that each region in different functional partitions and under different working conditions can obtain a matched lubrication state, thereby taking into account low friction, strong load bearing and temperature stability.

Description

Technical Field

[0001] This invention relates to the field of RV reducer technology, specifically to an adaptive viscosity intelligent lubrication method, system, and structure applied to RV reducers. Background Technology

[0002] RV reducers, as high-precision, high-rigidity, and high-torque-density transmission mechanisms, are widely used in industrial robots, precision machine tools, and semiconductor manufacturing equipment. Internally, they contain multiple meshing and rolling pairs, including an eccentric shaft, cycloidal wheel, pin gear housing, and output shaft. Operating continuously under high load, high-frequency reciprocating, and variable-speed conditions, they place extremely high demands on the stability of lubrication and the quality of oil film formation. Lubricating oil viscosity, as a key parameter affecting oil film thickness, frictional characteristics, and transmission efficiency, directly determines the reducer's energy consumption, temperature rise, accuracy maintenance, and service life.

[0003] Existing RV reducers generally use fixed-grade lubricating oils or greases, supplied through splash lubrication or oil bath lubrication. This traditional lubrication method has significant limitations: its viscosity is a fixed value and cannot be adjusted according to operating conditions, load changes, and ambient temperature. Lubrication requirements vary significantly depending on the reducer's operating stage. For example, during high-speed, no-load operation, low-viscosity oils can reduce stirring resistance and friction loss; while during low-speed, high-load or prolonged stationary operation, higher-viscosity oils are required to maintain sufficient oil film thickness and load-bearing capacity. Fixed-viscosity oils cannot simultaneously meet both of these conditions, resulting in poor lubricant flow and high starting resistance at low temperatures, and insufficient oil film strength and susceptibility to boundary friction and pitting at high temperatures.

[0004] Furthermore, the internal temperature of the RV reducer is not constant during operation, but rather exhibits cyclical fluctuations with time and load. These temperature changes cause a dynamic decrease in oil viscosity, triggering a feedback effect of "temperature rise—viscosity decrease—increased friction—further temperature rise," creating a thermally unstable cycle. Fixed-viscosity oil cannot achieve balanced lubrication across all parts, easily leading to localized over-lubrication or dry friction. Summary of the Invention

[0005] To address the aforementioned problems, this invention provides an adaptive viscosity intelligent lubrication method, system, and structure for use in RV reducers.

[0006] This invention employs the following technical solution: an adaptive viscosity intelligent lubrication method applied to RV reducers, comprising: Based on the internal geometric features and friction pair distribution of the RV reducer, the lubrication space is divided into P functional zones. An oil film partitioned microgrid layer is coated on the inner wall of the cycloidal wheel cavity. The microgrid layer is composed of magnetically conductive microstructures and fluid microchannels. The fluid microchannels are filled with lubricating fluid containing magnetically responsive particles. Each functional zone is equipped with a local magnetic field modulation unit, which adjusts the viscosity of the lubricant in the fluid microchannel by controlling the local magnetic field strength Bᵢ(t). Collect the operating status data of the RV reducer, and determine the operating condition type of the RV reducer based on the operating status data; Collect the current thermal and lubrication data of the RV reducer, input the acquired operating condition type, thermal data, lubrication data and functional partition type into the pre-built lubrication viscosity demand prediction model, and output the predicted lubricating oil viscosity. By installing miniature vibration viscosity sensors at the oil return channels or micro-grid outlets of functional zones, the real-time lubricating oil viscosity of each functional zone is collected. The real-time lubricating oil viscosity is compared and analyzed with the predicted lubricating oil viscosity to generate lubricating oil viscosity adjustment instructions.

[0007] As a further description of the above technical solution: P=4, the four functional zones are the tooth surface meshing zone, the pin rolling pair zone, the eccentric shaft bearing zone, and the oil sump buffer zone; The tooth meshing area corresponds to the meshing area between the cycloidal wheel and the pin tooth housing; the pin rolling pair area corresponds to the rolling contact area between the pin and the cycloidal wheel hole; the eccentric shaft bearing area corresponds to the eccentric shaft and the supporting bearing part; and the oil sump buffer zone is the area used for buffering transitions to balance heat and viscosity.

[0008] As a further description of the above technical solution: the operating status data includes input speed, output speed, and output torque; The operating conditions include high-speed light-load conditions, low-speed heavy-load conditions, and periodic start-stop conditions.

[0009] As a further description of the above technical solution: the method for obtaining the operating condition type of the RV reducer based on operating status data includes: The acquired operating status data is input into the pre-built operating condition type prediction model, and the operating condition type label of the RV reducer is output.

[0010] The training method for the working condition type prediction model includes: Under experimental conditions, different operating conditions of the RV reducer were simulated, including high-speed light-load condition, low-speed heavy-load condition, and periodic start-stop condition. K sets of training data were collected in advance, where K is a positive integer greater than 1. The K sets of training data include operating status data and operating condition type labels corresponding to the operating status data. The operating condition type labels are set as follows: label 0: high-speed light-load condition, label 1: low-speed heavy-load condition, label 2: periodic start-stop condition. A random forest classification model was adopted as the working condition type prediction model, and initial hyperparameters were set. The collected training data is divided into training and validation sets according to a preset ratio. The training set is used to train the model. Multiple training subsets are randomly generated using a bootstrap sampling method. Each decision tree is assigned a different training subset for independent training. During the node splitting process of each decision tree, features are selected from the candidate features corresponding to the running state data. The splitting features and splitting thresholds are determined based on the Gini coefficient. Finally, the prediction results of all decision trees are integrated through majority voting. At the same time, the hyperparameters are tuned using a Bayesian optimization method to select the optimal hyperparameter combination. The trained model is evaluated using the validation set and out-of-bag data. Core evaluation metrics are calculated to measure the correct prediction rate for all operating conditions, the recall rate for various fault types, the correct identification rate for a certain operating condition type, the confusion matrix, and the misclassification of operating conditions. Evaluation standards are set. Once the model performance evaluation meets the standards, it is used.

[0011] As a further description of the above technical solution: the thermal data includes the lubricating oil temperature; The lubrication data includes real-time oil pressure, oil flow rate, and the initial viscosity of the lubricant under non-magnetic field conditions.

[0012] As a further description of the above technical solution: the training method of the lubrication viscosity demand prediction model includes: Q sets of training data are collected in advance, where Q is a positive integer greater than 0. The training data includes operating condition type, thermal data, lubrication data and functional partition type, as well as the corresponding lubricating oil viscosity. The lubrication viscosity demand prediction model is trained using training data. Operating condition type, thermal data, lubrication data, and functional zone type are used as inputs to the model, while lubricating oil viscosity is used as the output. The stochastic gradient descent method is employed, and the weights and biases of the model are adjusted via backpropagation to minimize the error between the predicted and actual results. A loss function, the mean squared error, is set. Training stops when the loss function converges, and the model corresponding to the convergence of the loss function is taken as the trained lubrication viscosity demand prediction model.

[0013] As a further description of the above technical solution: the lubricating oil viscosity adjustment command includes a first adjustment command and a second adjustment command, wherein the first adjustment command is an instruction to increase the lubricating oil viscosity; and the second adjustment command is an instruction to decrease the lubricating oil viscosity. When the first adjustment command or the second adjustment command is generated, the local magnetic field strength is controlled. The viscosity of the lubricant is adjusted within the fluid microchannel.

[0014] As a further description of the above technical solution: the method for generating lubricating oil viscosity adjustment instructions includes: Obtain the viscosity difference between the real-time lubricating oil viscosity and the predicted lubricating oil viscosity. ; Preset gradient viscosity difference threshold and ,in <0< ; when < At that time, the first adjustment instruction is generated; when ≤ ≤ At this time, no lubricating oil viscosity adjustment command is generated; when > At that time, a second adjustment instruction is generated.

[0015] An adaptive viscosity intelligent lubrication system for RV reducers, used to implement the aforementioned adaptive viscosity intelligent lubrication method for RV reducers, the system comprising: The partitioning module divides the lubrication space into P functional partitions based on the internal geometric features and friction pair distribution of the RV reducer. The grid layer setting module coats the inner wall of the cycloidal wheel cavity with an oil film partitioned microgrid layer. The microgrid layer is composed of magnetically conductive microstructures and fluid microchannels. The fluid microchannels are filled with lubricating fluid containing magnetically responsive particles. The magnetic field modulation module has a local magnetic field modulation unit arranged under each functional area. By controlling the local magnetic field strength Bᵢ(t), the viscosity of the lubricant is adjusted in the fluid microchannel. The operating condition identification module collects the operating status data of the RV reducer and obtains the operating condition type of the RV reducer based on the operating status data; The viscosity prediction module collects the current thermal and lubrication data of the RV reducer, inputs the acquired operating condition type, thermal data, lubrication data and functional partition type into the pre-built lubrication viscosity demand prediction model, and outputs the predicted lubricating oil viscosity. The viscosity adjustment module uses miniature vibration viscosity sensors installed in the oil return channels or micro-grid outlets of functional zones to collect the real-time lubricating oil viscosity of each functional zone. The module compares and analyzes the real-time lubricating oil viscosity with the predicted lubricating oil viscosity to generate lubricating oil viscosity adjustment commands.

[0016] An adaptive viscosity intelligent lubrication structure for RV reducers is provided to implement the aforementioned adaptive viscosity intelligent lubrication method for RV reducers. The structure includes a cycloidal wheel cavity and an oil film partitioned microgrid layer. The inner wall of the cycloidal wheel cavity is coated with an oil film partitioned micro-mesh layer, which is composed of magnetically conductive microstructures and fluid microchannels, wherein the fluid microchannels are filled with lubricating fluid containing magnetically responsive particles.

[0017] The width of the fluid microchannel ranges from 50 to 300 μm; The magnetically conductive microstructure is a magnetically conductive sheet, which is plated on the wall of the fluid microchannel.

[0018] Compared with the prior art, the beneficial effects of this invention are as follows: This invention provides an adaptive viscosity intelligent lubrication method for RV reducers. Based on the distribution of friction pairs inside the reducer, the lubrication space is divided into multiple functional zones. In each functional zone, an oil film micro-mesh layer composed of magnetically conductive microstructures and fluid microchannels is constructed and filled with magnetically responsive lubricating fluid. The magnetic field strength Bᵢ(t) is controlled in real time by a local magnetic field modulation unit, which can realize the spatial distribution and dynamic adjustment of lubricating oil viscosity. This allows each area to obtain a matching lubrication state under different working conditions, thereby taking into account low friction, high load-bearing capacity and temperature stability.

[0019] This invention utilizes a pre-built multi-input deep learning model based on operating condition type, thermal data, lubrication data, and functional partition type to predict the optimal viscosity of each functional partition. Real-time feedback is generated through distributed vibration viscosity sensors. The difference between predicted and measured values ​​is analyzed to generate viscosity adjustment commands. This effectively overcomes the positive feedback problem in traditional lubrication where temperature rise causes viscosity decrease and boundary friction to intensify. It achieves adaptive optimization and thermal stability control of lubrication state, significantly improving the energy efficiency, accuracy retention, and service life of RV reducers. Attached Figure Description

[0020] The present invention will be further explained below with reference to the accompanying drawings and embodiments: Figure 1 A flowchart of an adaptive viscosity intelligent lubrication method for RV reducers provided in Embodiment 1 of the present invention; Figure 2 This is a module connection diagram of an adaptive viscosity intelligent lubrication system for RV reducers provided in Embodiment 2 of the present invention; Figure 3 This is a cross-sectional schematic diagram of an adaptive viscosity intelligent lubrication structure applied to an RV reducer, provided in Embodiment 3 of the present invention; Figure 4 Provided for Embodiment 3 of the present invention Figure 3 Enlarged view of area A in the image.

[0021] Figure descriptions: 1. Cycloidal wheel cavity; 2. Oil film partitioned microgrid layer; 21. Magnetic microstructure; 22. Fluid microchannel. Detailed Implementation

[0022] To make the technical means, creative features, objectives, and effects of this invention readily understandable, the invention is further described below with reference to specific illustrations. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0023] Example 1 Please see Figure 1 This invention provides a technical solution: an adaptive viscosity intelligent lubrication method for RV reducers, the method comprising: Collect the operating status data of the RV reducer, and determine the operating condition type of the RV reducer based on the operating status data; The operating conditions include high-speed light-load conditions, low-speed heavy-load conditions, and periodic start-stop conditions.

[0024] It should be noted that different operating conditions require different lubrication requirements. For high-speed, light-load conditions, low-viscosity oil is needed to reduce rotational resistance. For low-speed, heavy-load conditions, high-viscosity oil is needed to maintain oil film thickness and prevent boundary friction. For periodic start-stop conditions, the oil needs to have high viscosity recovery and shear resistance. The operating status data includes input speed, output speed, and output torque; Methods for obtaining the operating condition type of an RV reducer based on operational status data include: The acquired operating status data is input into the pre-built operating condition type prediction model, and the operating condition type label of the RV reducer is output. The training method for the working condition type prediction model includes: Under experimental conditions, different operating conditions of the RV reducer were simulated, including high-speed light-load condition, low-speed heavy-load condition, and periodic start-stop condition. K sets of training data were collected in advance, where K is a positive integer greater than 1. The K sets of training data include operating status data and operating condition type labels corresponding to the operating status data. The operating condition type labels are set as follows: label 0: high-speed light-load condition, label 1: low-speed heavy-load condition, label 2: periodic start-stop condition. A random forest classification model was adopted as the working condition type prediction model. The initial hyperparameters were set as follows: the number of decision trees was 100, the maximum depth of a single tree was 8, the maximum number of features considered when splitting a node was 3, and the minimum number of samples for splitting a node was 4. The collected training data is divided into a training set of 80% for model parameter learning and a validation set of 20% for hyperparameter tuning according to a preset ratio. The model is trained using a training set. Multiple training subsets are randomly generated using a bootstrap sampling method, and each decision tree is assigned a different training subset for independent training. During the node splitting process of each decision tree, features are selected from candidate features corresponding to the running state data, and the splitting features and splitting thresholds are determined based on the Gini coefficient. Finally, the prediction results of all decision trees are integrated through majority voting. At the same time, the hyperparameters are tuned using a Bayesian optimization method. The hyperparameter search range is: 80-120 decision trees, 6-10 layers for the maximum depth of a single tree, and 2-6 minimum number of samples for node splitting. The optimal combination of hyperparameters is selected with the validation set accuracy as the optimization objective. The trained model is evaluated using a validation set. When the accuracy of the operating condition type prediction and the recall of each operating condition type meet the preset evaluation criteria, the trained operating condition type prediction model is used for the operating condition type prediction of the RV reducer.

[0025] In this embodiment, by collecting the operating status data of the RV reducer and using a random forest classification model to automatically identify the operating condition type of the RV reducer, the lubrication requirements under different operating conditions vary significantly. For example, low viscosity oil is required under high-speed light load conditions to reduce frictional resistance, while high viscosity oil is required under low-speed heavy load conditions to maintain oil film thickness. Through model identification, the system can quickly and accurately determine the type of operating condition, thereby providing a basis for subsequent prediction and adjustment of lubrication viscosity, significantly improving the pertinence and response speed of lubrication control.

[0026] Based on the internal geometric features and friction pair distribution of the RV reducer, the lubrication space is divided into P functional zones. It should be noted that a friction pair refers to two parts surfaces that come into contact with each other and generate relative motion in a mechanical system. It is the direct object of phenomena such as friction, wear, and lubrication.

[0027] Optionally, P=4, and the four functional zones are the tooth surface meshing zone, the pin rolling pair zone, the eccentric shaft bearing zone, and the oil sump buffer zone; It should be noted that the tooth meshing area corresponds to the meshing area between the cycloidal wheel and the pin tooth housing; the pin rolling pair area corresponds to the rolling contact area between the pin and the cycloidal wheel hole; the eccentric shaft bearing area corresponds to the eccentric shaft and the supporting bearing part; and the oil sump buffer zone is the area used for buffering transitions to balance heat and viscosity. The four zones have different requirements for lubricating oil viscosity: the lubricating oil viscosity needs to be reduced sequentially in the tooth meshing zone, the pin rolling pair zone, the eccentric shaft bearing zone, and the oil sump buffer zone.

[0028] An oil film partitioned microgrid layer is coated on the inner wall of the cycloidal wheel cavity. The microgrid layer is composed of magnetically conductive microstructures and fluid microchannels. The channel width ranges from 50 to 300 μm, and the interior is filled with lubricating fluid containing magnetically responsive particles. Each functional zone is equipped with a local magnetic field modulation unit. By controlling the local magnetic field strength Bᵢ(t), the viscosity of the lubricant can be adjusted within the fluid microchannel. It should be noted that when As the viscosity increases, the internal particle chain structure of the lubricant strengthens, and the shear stress increases, thus forming a high-viscosity zone; when When the magnetic field is reduced or removed, the liquid regains its fluidity and forms a low-viscosity zone. By adjusting the magnetic field intensity distribution in different zones, spatial differentiation of local lubrication characteristics can be achieved.

[0029] In this embodiment, the lubrication space inside the RV reducer is divided into multiple functional zones, including the tooth meshing zone, the pin rolling pair zone, the eccentric shaft bearing zone, and the oil sump buffer zone, making lubrication control more precise. Each functional zone has different viscosity requirements based on the characteristics of the friction pair. This method can achieve differentiated viscosity control in different zones, avoiding the problem of uniform oil performance in traditional centralized lubrication. This ensures sufficient lubrication while reducing energy loss and improving mechanical efficiency and lifespan.

[0030] Collect the current thermal and lubrication data of the RV reducer, input the acquired operating condition type, thermal data, lubrication data and functional partition type into the pre-built lubrication viscosity demand prediction model, and output the predicted lubricating oil viscosity. The thermal data includes the lubricating oil temperature; The lubrication data includes real-time oil pressure, oil flow rate, and initial viscosity of the lubricant under non-magnetic field conditions; It should be noted that the lubricating oil temperature is acquired in real time by a temperature sensor, the oil pressure and flow rate are acquired by a pressure sensor and a flow sensor, and the initial viscosity of the lubricating fluid under non-magnetic field conditions is obtained from the lubricating oil's product manual.

[0031] The training method for the lubrication viscosity demand prediction model includes: Q sets of training data are collected in advance, where Q is a positive integer greater than 0. The training data includes operating condition type, thermal data, lubrication data and functional partition type, as well as the corresponding lubricating oil viscosity. The lubrication viscosity demand prediction model is trained using training data. Operating condition type, thermal data, lubrication data, and functional zone type are used as inputs to the model, while lubricating oil viscosity is used as the output. The stochastic gradient descent method is employed, and the weights and biases of the model are adjusted via backpropagation to minimize the error between the predicted and actual results. A loss function, the mean squared error, is set. Training stops when the loss function converges, and the model corresponding to the convergence of the loss function is taken as the trained lubrication viscosity demand prediction model.

[0032] The lubrication viscosity demand prediction model is a multi-input regression deep neural network model.

[0033] By installing miniature vibration-type viscosity sensors at the oil return channels or micro-mesh outlets of functional zones, the real-time lubricating oil viscosity of each functional zone is collected, and the real-time lubricating oil viscosity is recorded. Compare and analyze the predicted lubricating oil viscosity to generate lubricating oil viscosity adjustment instructions; The method for generating lubricating oil viscosity adjustment instructions includes: Obtain the viscosity difference between the real-time lubricating oil viscosity and the predicted lubricating oil viscosity. ; Preset gradient viscosity difference threshold and ,in <0< ; when < At that time, the first adjustment instruction is generated; when ≤ ≤ At this time, no lubricating oil viscosity adjustment command is generated; when > At that time, a second adjustment instruction is generated; The first adjustment command is to increase the viscosity of the lubricating oil; the second adjustment command is to decrease the viscosity of the lubricating oil. The first adjustment instruction or the second adjustment instruction adjusts the viscosity value to... .

[0034] When the first adjustment command or the second adjustment command is generated, the local magnetic field strength is controlled. The viscosity of the lubricant is adjusted within the fluid microchannel.

[0035] It should be noted that the gradient viscosity difference threshold and The empirical threshold method can be used for setting. It is based on the difference range between the actual lubricating oil viscosity and the predicted lubricating oil viscosity during the historical operation of the RV reducer, and combined with the preset allowable viscosity deviation, which is preset by the technicians.

[0036] In one embodiment, the target magnetic field strength can be determined through a zoned experimental calibration mapping. Specifically, at multiple calibration temperatures and multiple calibration flow rates, a magnetic field is applied to the i-th functional zone in an increasing sequence of magnetic field strengths. After each magnetic field stabilizes, the corresponding apparent viscosity is collected to establish a zoned calibration mapping between temperature, flow rate, magnetic field strength, and apparent viscosity. During the control phase, the target magnetic field strength is determined through the zoned calibration mapping based on the current temperature, current flow rate, and target apparent viscosity. When the apparent viscosity increment corresponding to consecutive adjacent magnetic field levels is not greater than three times the resolution of the viscosity sensor, the magnetic field strength that first meets this condition is determined as the magnetic saturation initiation strength. The target magnetic field strength must not exceed the magnetic saturation initiation strength.

[0037] In this embodiment, based on the distribution of friction pairs inside the reducer, the lubrication space is divided into multiple functional zones. Within each zone, an oil film microgrid layer composed of magnetically conductive microstructures and fluid microchannels is constructed, filled with magnetically responsive lubricating fluid. By controlling the magnetic field strength Bᵢ(t) in real time through a local magnetic field modulation unit, the spatial distribution and dynamic adjustment of the lubricating fluid viscosity can be achieved, ensuring that each region obtains a matching lubrication state under different operating conditions, thus balancing low friction, high load-bearing capacity, and temperature stability.

[0038] By inputting operating status, thermal data, and lubrication data into a pre-built multi-input deep learning model, the optimal viscosity of each functional zone is predicted. Real-time feedback is generated through distributed vibration viscosity sensors. The difference between the predicted and measured values ​​is analyzed to generate viscosity adjustment commands. This effectively overcomes the positive feedback problem in traditional lubrication where temperature rise causes viscosity decrease and boundary friction to intensify. It achieves adaptive optimization and thermal stability control of lubrication status, significantly improving the energy efficiency, accuracy retention, and service life of RV reducers.

[0039] Example 2 Please see Figure 2 This invention provides a technical solution: an adaptive viscosity intelligent lubrication system for RV reducers, which is used to implement the aforementioned adaptive viscosity intelligent lubrication method for RV reducers. The system includes: The partitioning module divides the lubrication space into P functional partitions based on the internal geometric features and friction pair distribution of the RV reducer. The grid layer setting module coats the inner wall of the cycloidal wheel cavity with an oil film partitioned microgrid layer. The microgrid layer is composed of magnetically conductive microstructures and fluid microchannels. The fluid microchannels are filled with lubricating fluid containing magnetically responsive particles. The magnetic field modulation module has a local magnetic field modulation unit arranged under each functional zone. By controlling the local magnetic field strength Bᵢ(t), the viscosity of the lubricant is adjusted in the fluid microchannel. The local magnetic field modulation unit is set in the stationary shell position adjacent to the corresponding functional zone. The power supply module and control line are set outside the moving cavity of the reducer. The magnetic field generated by the local magnetic field modulation unit acts on the magnetic response lubricant in the corresponding functional zone. The operating condition identification module collects the operating status data of the RV reducer and obtains the operating condition type of the RV reducer based on the operating status data; The viscosity prediction module collects the current thermal and lubrication data of the RV reducer, inputs the acquired operating condition type, thermal data, lubrication data and functional partition type into the pre-built lubrication viscosity demand prediction model, and outputs the predicted lubricating oil viscosity. The viscosity adjustment module uses miniature vibration viscosity sensors installed in the oil return channels or micro-grid outlets of functional zones to collect the real-time lubricating oil viscosity of each functional zone. The module compares and analyzes the real-time lubricating oil viscosity with the predicted lubricating oil viscosity to generate lubricating oil viscosity adjustment commands.

[0040] In this embodiment, adaptive lubrication control of the RV reducer can be achieved under multiple operating conditions and temperature zones, significantly reducing friction loss and tooth surface wear, suppressing thermal instability, and extending the service life of the equipment. At the same time, by realizing the coordinated adjustment of zoned magnetic control and intelligent prediction, this system has the characteristics of high energy efficiency, long life and intelligence, providing a universal intelligent lubrication solution for high-precision transmission equipment.

[0041] Example 3 Please see Figures 3-4 This invention provides a technical solution: an adaptive viscosity intelligent lubrication structure for RV reducers, which implements an adaptive viscosity intelligent lubrication method for RV reducers. The structure includes a cycloidal wheel cavity 1 and an oil film partitioned microgrid layer 2; wherein region A is an arbitrarily selected area in the microgrid layer. Figure 4 This is a magnified view of area A; The inner wall of the cycloidal wheel cavity 1 is coated with an oil film partitioned micro-mesh layer 2. The micro-mesh layer 2 is composed of a magnetically conductive microstructure 21 and a fluid microchannel 22, wherein the fluid microchannel 22 is filled with a lubricant containing magnetically responsive particles.

[0042] The width of the fluid microchannel 22 ranges from 50 to 300 μm; The magnetically conductive microstructure 21 is a magnetically conductive sheet, which is plated on the wall of the fluid microchannel 22.

[0043] In this embodiment, an oil film partitioned microgrid layer 2 structure is introduced into the inner wall of the cycloidal wheel cavity 1. A magnetically responsive oil film layer is formed using a magnetically conductive microstructure 21 and a fluid microchannel 22. Local magnetic field modulation units are arranged under each functional partition. By adjusting the magnetic field strength Bᵢ(t) in real time, the internal particle chain structure of the lubricant containing magnetically responsive particles can be directly controlled, changing its viscosity state and achieving instantaneous adjustment of the lubricant viscosity. Compared to the traditional method of using a fixed grade of lubricating oil, this structure can dynamically generate high-viscosity or low-viscosity zones according to the load changes and temperature fluctuations of the reducer. This ensures reduced flow resistance at high speeds and light loads, and maintains sufficient oil film thickness at low speeds and heavy loads, thus simultaneously achieving energy consumption control and wear protection.

[0044] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. An adaptive viscosity intelligent lubrication method applied to RV reducers, characterized in that, include: Based on the internal geometric features and friction pair distribution of the RV reducer, the lubrication space is divided into P functional zones. An oil film partitioned microgrid layer is coated on the inner wall of the cycloidal wheel cavity. The microgrid layer is composed of magnetically conductive microstructures and fluid microchannels. The fluid microchannels are filled with lubricating fluid containing magnetically responsive particles. Each functional zone is equipped with a local magnetic field modulation unit, which adjusts the viscosity of the lubricant in the fluid microchannel by controlling the local magnetic field strength Bᵢ(t). Collect the operating status data of the RV reducer, and determine the operating condition type of the RV reducer based on the operating status data; Collect the current thermal and lubrication data of the RV reducer, input the acquired operating condition type, thermal data, lubrication data and functional partition type into the pre-built lubrication viscosity demand prediction model, and output the predicted lubricating oil viscosity. By installing miniature vibration viscosity sensors at the oil return channels or micro-grid outlets of functional zones, the real-time lubricating oil viscosity of each functional zone is collected. The real-time lubricating oil viscosity is compared and analyzed with the predicted lubricating oil viscosity to generate lubricating oil viscosity adjustment instructions.

2. The adaptive viscosity intelligent lubrication method for RV reducers according to claim 1, characterized in that, The functional zones are four, namely, the tooth surface meshing zone, the pin rolling pair zone, the eccentric shaft bearing zone, and the oil sump buffer zone. The tooth meshing area corresponds to the meshing area between the cycloidal wheel and the pin tooth housing; the pin rolling pair area corresponds to the rolling contact area between the pin and the cycloidal wheel hole; the eccentric shaft bearing area corresponds to the eccentric shaft and the supporting bearing part; and the oil sump buffer zone is the area used for buffering transitions to balance heat and viscosity.

3. The adaptive viscosity intelligent lubrication method for RV reducers according to claim 1, characterized in that, The operating status data includes input speed, output speed, and output torque; The operating conditions include high-speed light-load conditions, low-speed heavy-load conditions, and periodic start-stop conditions.

4. The adaptive viscosity intelligent lubrication method for RV reducers according to claim 3, characterized in that, Methods for obtaining the operating condition type of an RV reducer based on operational status data include: The acquired operating status data is input into the pre-built operating condition type prediction model, and the operating condition type label of the RV reducer is output. The training method for the working condition type prediction model includes: Under experimental conditions, different operating conditions of the RV reducer were simulated, including high-speed light-load condition, low-speed heavy-load condition, and periodic start-stop condition. K sets of training data were collected in advance, where K is a positive integer greater than 1. The K sets of training data include operating status data and operating condition type labels corresponding to the operating status data. The operating condition type labels are set as follows: label 0: high-speed light-load condition, label 1: low-speed heavy-load condition, label 2: periodic start-stop condition. A random forest classification model was adopted as the working condition type prediction model, and initial hyperparameters were set. The collected training data is divided into training and validation sets according to a preset ratio. The training set is used to train the model. Multiple training subsets are randomly generated using a bootstrap sampling method. Each decision tree is assigned a different training subset for independent training. During the node splitting process of each decision tree, features are selected from the candidate features corresponding to the running state data. The splitting features and splitting thresholds are determined based on the Gini coefficient. Finally, the prediction results of all decision trees are integrated through majority voting. At the same time, the hyperparameters are tuned using a Bayesian optimization method to select the optimal hyperparameter combination. The trained model is evaluated using the validation set and out-of-bag data. Core evaluation metrics are calculated to measure the correct prediction rate for all operating conditions, the recall rate for various fault types, the correct identification rate for a certain operating condition type, the confusion matrix, and the misclassification of operating conditions. Evaluation standards are set. Once the model performance evaluation meets the standards, it is used.

5. The adaptive viscosity intelligent lubrication method for RV reducers according to claim 1, characterized in that, The thermal data includes the lubricating oil temperature; The lubrication data includes real-time oil pressure, oil flow rate, and the initial viscosity of the lubricant under non-magnetic field conditions.

6. The adaptive viscosity intelligent lubrication method for RV reducers according to claim 5, characterized in that, The training method for the lubrication viscosity demand prediction model includes: Q sets of training data are collected in advance, where Q is a positive integer greater than 0. The training data includes operating condition type, thermal data, lubrication data and functional partition type, as well as the corresponding lubricating oil viscosity. The lubrication viscosity demand prediction model is trained using training data. The operating condition type, thermal data, lubrication data, and functional partition type are used as inputs to the lubrication viscosity demand prediction model, and the lubricating oil viscosity is used as the output. The stochastic gradient descent method is used, and the weights and biases of the lubrication viscosity demand prediction model are adjusted through the backpropagation algorithm to minimize the error between the prediction results and the actual results. A loss function is set, which is the mean square error. When the loss function value reaches convergence, the training of the lubrication viscosity demand prediction model is stopped, and the lubrication viscosity demand prediction model corresponding to the convergence of the loss function value is used as the trained lubrication viscosity demand prediction model. The lubrication viscosity demand prediction model is a multi-input regression deep neural network model.

7. The adaptive viscosity intelligent lubrication method for RV reducers according to claim 1, characterized in that, The lubricating oil viscosity adjustment command includes a first adjustment command and a second adjustment command. The first adjustment command is to increase the lubricating oil viscosity; the second adjustment command is to decrease the lubricating oil viscosity. When the first or second adjustment command is generated, the local magnetic field strength is controlled. The viscosity of the lubricant is adjusted within the fluid microchannel.

8. The adaptive viscosity intelligent lubrication method for RV reducers according to claim 7, characterized in that, The method for generating lubricating oil viscosity adjustment instructions includes: Obtain the viscosity difference between the real-time lubricating oil viscosity and the predicted lubricating oil viscosity. ; Preset gradient viscosity difference threshold and ,in <0< ; when < At that time, the first adjustment instruction is generated; when ≤ ≤ At this time, no lubricating oil viscosity adjustment command is generated; When > At that time, a second adjustment instruction is generated.

9. An adaptive viscosity intelligent lubrication system for RV reducers, used to implement the adaptive viscosity intelligent lubrication method for RV reducers as described in any one of claims 1-8, characterized in that, The system includes: The partitioning module divides the lubrication space into P functional partitions based on the internal geometric features and friction pair distribution of the RV reducer. The grid layer setting module coats the inner wall of the cycloidal wheel cavity with an oil film partitioned microgrid layer. The microgrid layer is composed of magnetically conductive microstructures and fluid microchannels. The fluid microchannels are filled with lubricating fluid containing magnetically responsive particles. The magnetic field modulation module has a local magnetic field modulation unit arranged under each functional area. By controlling the local magnetic field strength Bᵢ(t), the viscosity of the lubricant is adjusted in the fluid microchannel. The operating condition identification module collects the operating status data of the RV reducer and obtains the operating condition type of the RV reducer based on the operating status data; The viscosity prediction module collects the current thermal and lubrication data of the RV reducer, inputs the acquired operating condition type, thermal data, lubrication data and functional partition type into the pre-built lubrication viscosity demand prediction model, and outputs the predicted lubricating oil viscosity. The viscosity adjustment module uses miniature vibration viscosity sensors installed in the oil return channels or micro-grid outlets of functional zones to collect the real-time lubricating oil viscosity of each functional zone. The module compares and analyzes the real-time lubricating oil viscosity with the predicted lubricating oil viscosity to generate lubricating oil viscosity adjustment commands.

10. An adaptive viscosity intelligent lubrication structure for use in RV reducers, used to implement the adaptive viscosity intelligent lubrication method for RV reducers as described in any one of claims 1-8, characterized in that, The structure includes a cycloidal wheel cavity (1) and an oil film partitioned microgrid layer (2). The inner wall of the cycloidal wheel cavity (1) is coated with an oil film partitioned micro-mesh layer (2), which is composed of a magnetically conductive microstructure (21) and a fluid microchannel (22), wherein the fluid microchannel (22) is filled with a lubricant containing magnetically responsive particles; The width of the fluid microchannel (22) ranges from 50 to 300 μm; The magnetically conductive microstructure (21) is a magnetically conductive sheet, which is plated on the wall of the fluid microchannel (22).

Images