A planet roller screw uniform load optimization method based on multi-source error matching

Abstract

This invention discloses a method for optimizing load distribution in planetary roller screws based on multi-source error matching. It relates to the field of model correction technology. The method includes the following steps: S1, based on a preset multi-roller load distribution theoretical model, a complete dataset of multi-roller load distribution coefficients is obtained through simulation calculations and data acquisition; S2, using this dataset as samples, a neural network surrogate model is constructed to establish a precise mapping relationship between multi-source errors and multi-roller load distribution coefficients. This invention can accurately and efficiently determine the optimal solution for load distribution uniformity and the corresponding optimal matching scheme for multi-source errors. It utilizes a neural network to establish a complex nonlinear mapping relationship from multi-source errors to load distribution, replacing time-consuming high-precision physical simulation. Simultaneously, it employs a two-level collaborative optimization strategy: first optimizing load distribution among the rollers, and then optimizing load distribution on each thread tooth of the roller screw and nut, solving the problem of uneven load distribution between multi-roller and threaded pairs in engineering.

Description

Technical Field

[0001] This invention relates to the field of model correction technology, and in particular to a method for correcting interval models that takes into account correlation analysis. Background Technology

[0002] Planetary roller screws possess advantages such as complex structural features, high load-bearing capacity, and long service life, and are widely used in aerospace equipment, weaponry, and humanoid robots. Improving the load distribution among the multiple rollers and the uniformity of the load distribution in the threaded pair is an effective way to enhance their load-bearing performance. However, due to their complex structure, planetary roller screws inevitably generate multiple sources of errors during machining and assembly, resulting in poor load distribution among the multiple rollers and the uniformity of the load distribution in the threaded pair, thereby reducing the load-bearing performance and service reliability of the planetary roller screw. Therefore, constructing an optimization model for the load distribution among the multiple rollers and the uniformity of the load distribution in the threaded pair of planetary roller screws, incorporating multiple sources of errors, is of paramount importance for improving their load-bearing performance.

[0003] Patent application CN202411664812.6, filed on 2024-11-20, entitled "A Load Sharing Optimization Design Method for Reverse Differential Planetary Roller Screws," includes the following steps: Step 1: Based on the deformation coordination relationship of the thread meshing of the reverse differential planetary roller screw, construct deformation coordination formulas for the roller annular groove and nut annular groove, and the roller thread and screw thread; Step 2: Use the random forest algorithm to obtain the importance of thread structure parameters and other parameters to load sharing performance, and extract key influencing parameters; Step 3: Use the GA-BP neural network to establish a prediction model for key parameters and load sharing performance; Step 4: Based on NSGA-II, construct a multi-objective optimization algorithm for mass loss and tolerance sensitivity to achieve multi-objective optimization design of load sharing for reverse differential planetary roller screws under the constraint of processing cost. This method can improve the load sharing effect between the threads of the reverse differential planetary roller screw, thereby increasing its load-bearing capacity and service life.

[0004] The aforementioned patents, while improving the load-sharing effect between thread teeth through multi-objective optimization of structural parameters, neglected the uneven load distribution among multiple rollers caused by multi-source errors. Meanwhile, in the academic journal *Proceedings of the IEEE Transactions on Mechanical Engineering (Series C)*, the uniformity of load distribution in planetary roller screws was improved by correcting the thread pair clearance. Furthermore, based on deformation coordination and force balance, a roller taper correction method was proposed to optimize the load distribution of the thread pair, and the thread pair clearance was controlled by adjusting the roller grinding angle.

[0005] In existing technologies, load sharing optimization methods are all designed for uneven load distribution between thread teeth and neglect the influence of multi-source errors. However, no innovative optimization method exists for optimizing the load distribution of multiple rollers and the uniformity of load distribution in planetary roller screws under multi-source errors. Therefore, there is an urgent need for a load sharing optimization method for planetary roller screws based on multi-source error matching. This method should consider the impact of multi-source errors on the load distribution of multiple rollers and the load distribution of the thread pair, improve the uniformity of load distribution through multi-source error matching, and finally guide the assembly of planetary roller screw components based on the multi-source error matching scheme. Summary of the Invention

[0006] This invention provides a planetary roller screw load sharing optimization method based on multi-source error matching to solve the problems in the prior art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a planetary roller screw load sharing optimization method based on multi-source error matching, comprising the following steps: S1, based on a preset multi-roller load distribution theoretical model, a complete dataset of multi-roller load distribution coefficients is obtained through simulation calculation and data acquisition; S2, using this dataset as samples, a neural network surrogate model is constructed to establish a precise mapping relationship between multi-source errors and multi-roller load distribution coefficients; S3, based on this mapping relationship, a single-objective uniformity optimization model is constructed to solve for the optimized multi-roller load distribution coefficients and corresponding multi-source error matching parameters; S4, using S3... Using the optimized multi-roller load distribution coefficient obtained from the solution as the core input, a theoretical model of the threaded pair load distribution is constructed to obtain the threaded pair load distribution coefficient dataset; S5, using this dataset as a sample, neural network surrogate models are constructed on the screw side and nut side respectively to establish the independent mapping relationship between multi-source errors and the load distribution coefficients on both sides; S6, based on the mapping relationship on both sides, multi-objective optimization models are constructed respectively to obtain the Pareto front solution set and multi-source error matrix; S7, the above results are analyzed using a multi-objective decision method to determine the optimal solution for the uniformity of load distribution on both sides and the optimal matching scheme for multi-source errors, thus completing the overall optimization.

[0008] The aforementioned method for optimizing the load sharing of planetary roller screws based on multi-source error matching further includes, in step one: the expression for the multi-roller load distribution of the planetary roller screw containing multi-source errors can be described as follows:

[0009] ;

[0010] In the formula, The total axial load borne by the planetary roller screw; The overall contact stiffness of a single roller; rollers and Overall deformation difference.

[0011] The aforementioned method for optimizing the load sharing of planetary roller screws based on multi-source error matching further includes, in step two: a surrogate model for multi-roller load distribution coefficients based on a neural network, with 10 hidden layers, and its input and output layers being...

[0012] ;

[0013] In the formula, the input layer includes the lead screw eccentricity error. Nut eccentricity error Error with nominal roller radius The output layer is a multi-roller load distribution coefficient. .

[0014] The aforementioned method for optimizing the load distribution of planetary roller screws based on multi-source error matching further includes, in step three: a multi-roller load distribution uniformity optimization model containing multi-source errors can be expressed as:

[0015] ;

[0016] In the formula, and The lower and upper boundaries of the lead screw eccentricity error are defined separately. and The lower and upper boundaries of the nut eccentricity error are defined separately. and The lower and upper boundaries of the nominal diameter error of the roller are respectively defined. The average value of the multi-roller load distribution factor is [value]. , This represents the number of rollers.

[0017] The aforementioned method for optimizing the load distribution of planetary roller screws based on multi-source error matching further includes, in step four: the load distribution of the threaded pair is closely related to the load borne by each roller. After determining the load borne by each roller, the expression for the load distribution of the threaded pair can be described as follows:

[0018] ;

[0019] In the formula, and These represent the meshing stiffness of the threaded pairs on the roller screw side and the roller nut side, respectively. For the first The threaded pair and the first The difference in deformation between individual threaded pairs.

[0020] The aforementioned method for optimizing the load distribution of planetary roller screws based on multi-source error matching further includes, in step five: a neural network topology for the load distribution coefficient samples of the threaded pairs on the roller screw side and the roller nut side, wherein the input layer and output layer are respectively represented as:

[0021] ;

[0022] In the formula, the input layer includes the lead screw pitch error. Nut pitch error With roller pitch error The output layers are the load distribution coefficients of the roller screw side thread pair. Load distribution coefficient of roller nut side thread pair .

[0023] The aforementioned method for optimizing the load distribution of planetary roller screws based on multi-source error matching further includes, in step six: the optimization model for the uniformity of load distribution in the threaded pair can be expressed as:

[0024] ;

[0025] In the formula, and The lower and upper boundaries of the lead screw eccentricity error are defined separately. and The lower and upper boundaries of the nut eccentricity error are defined separately. and The lower and upper boundaries of the nominal diameter error of the roller are respectively defined. The average value of the multi-roller load distribution factor is [value]. , This represents the number of rollers.

[0026] The aforementioned method for optimizing the load distribution of planetary roller screws based on multi-source error matching further includes, in step seven: the method for determining the optimal solution of the threaded pair load distribution coefficient and the optimal strategy for multi-source errors can be expressed as follows:

[0027] ;

[0028] In the formula, The error matrix corresponding to the Pareto front solution set is optimized to ensure the uniformity of load distribution on the roller screw side and the roller nut side.

[0029] The beneficial effects of this invention are:

[0030] In this invention, the optimal solution for load distribution uniformity and the corresponding optimal matching scheme for multi-source errors can be determined accurately and efficiently. Furthermore, a complex nonlinear mapping relationship from multi-source errors to load distribution is established using a neural network, replacing time-consuming high-precision physical simulation. A two-level collaborative optimization strategy is adopted: first, load distribution is optimized between rollers, and then load distribution is optimized on each thread tooth of the roller screw and nut, solving the problem of uneven load distribution in multi-roller and threaded pairs in engineering. Attached Figure Description

[0031] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0032] Figure 1 It is the optimization result of the multi-roller load distribution coefficient and the multi-source error matching strategy;

[0033] Figure 2 It is the load distribution coefficient of the threaded pair on the roller screw side and nut side based on the optimal strategy;

[0034] Figure 3 It is an optimized matching strategy for the pitch errors of the lead screw, nut, and roller. Detailed Implementation

[0035] To enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings.

[0036] like Figure 1-3As shown, this embodiment provides a planetary roller screw load sharing optimization method based on multi-source error matching, including the following steps: S1, based on a preset multi-roller load distribution theoretical model, obtain a complete dataset of multi-roller load distribution coefficients through simulation calculation and data acquisition; S2, using this dataset as samples, construct a neural network surrogate model to establish a precise mapping relationship between multi-source errors and multi-roller load distribution coefficients; S3, based on this mapping relationship, construct a single-objective uniformity optimization model to solve for the optimized multi-roller load distribution coefficients and corresponding multi-source error matching parameters; S4, using the optimized values ​​obtained in S3... Using the multi-roller load distribution coefficient as the core input, a theoretical model of the threaded pair load distribution is constructed to obtain the threaded pair load distribution coefficient dataset; S5. Using this dataset as a sample, neural network surrogate models are constructed for the screw side and nut side respectively to establish the independent mapping relationship between multi-source errors and the load distribution coefficients on both sides; S6. Based on the mapping relationship on both sides, multi-objective optimization models are constructed respectively to obtain the Pareto front solution set and multi-source error matrix; S7. The above results are analyzed using a multi-objective decision method to determine the optimal solution for the uniformity of load distribution on both sides and the optimal matching scheme for multi-source errors, thus completing the overall optimization.

[0037] In this invention, step one further includes: the multi-roller load distribution expression for a planetary roller screw containing multi-source errors can be described as:

[0038] ;

[0039] In the formula, The total axial load borne by the planetary roller screw; The overall contact stiffness of a single roller; rollers and Overall deformation difference.

[0040] Step two further includes: a surrogate model for multi-roller load distribution coefficients based on a neural network, with 10 hidden layers, and its input and output layers are as follows:

[0041] ;

[0042] In the formula, the input layer includes the lead screw eccentricity error. Nut eccentricity error Error with nominal roller radius The output layer is a multi-roller load distribution coefficient. .

[0043] Step three further includes:

[0044] The optimization model for the uniformity of multi-roller load distribution with multi-source errors can be expressed as:

[0045] ;

[0046] In the formula, and The lower and upper boundaries of the lead screw eccentricity error are defined separately. and The lower and upper boundaries of the nut eccentricity error are defined separately. and The lower and upper boundaries of the nominal diameter error of the roller are respectively defined. The average value of the multi-roller load distribution factor is [value]. , This represents the number of rollers.

[0047] Step four further includes: the load distribution of the threaded pair is closely related to the load borne by each roller. After determining the load borne by each roller, the expression for the load distribution of the threaded pair can be described as:

[0048] ;

[0049] In the formula, and These represent the meshing stiffness of the threaded pairs on the roller screw side and the roller nut side, respectively. For the first The threaded pair and the first The difference in deformation between individual threaded pairs.

[0050] Step five further includes:

[0051] The neural network topology for the load distribution coefficient samples of the threaded pairs on the roller screw side and the roller nut side, with its input and output layers represented as follows:

[0052] ;

[0053] In the formula, the input layer includes the lead screw pitch error. Nut pitch error With roller pitch error The output layers are the load distribution coefficients of the roller screw side thread pair. Load distribution coefficient of roller nut side thread pair .

[0054] In step six, the following is further included: the optimization model for the uniformity of load distribution of the threaded pair can be expressed as:

[0055] ;

[0056] In the formula, and The lower and upper boundaries of the lead screw pitch error are defined separately. and The lower and upper boundaries of the nut pitch error are respectively defined; and The lower and upper boundaries of the roller pitch error are respectively defined; The average value of the load distribution coefficient for the threaded pair on the roller screw side and the roller nut side is [value missing]. .

[0057] Step seven further includes:

[0058] The method for determining the optimal solution of the load distribution coefficient of the threaded pair and the optimal strategy for multi-source error can be expressed as:

[0059] ;

[0060] In the formula, The error matrix corresponding to the Pareto front solution set is optimized to ensure the uniformity of load distribution on the roller screw side and the roller nut side.

[0061] like Figure 1 As shown, the optimized maximum and minimum load distribution coefficients for the multi-roller system are 0.09316 and 0.08936, respectively, and the difference is... The value was 0.0038, indicating that the uniformity of multi-roller load distribution was improved by nearly 13 times after optimization. This optimization result is very close to the ideal multi-roller load distribution coefficient, i.e., the average value (0.0909), demonstrating that optimal matching of multi-source errors can effectively improve the uniformity of multi-roller load distribution in planetary roller screws. Figure 1 This paper describes the optimal matching scheme for multi-source errors corresponding to the optimized load distribution results of multiple rollers, including screw eccentricity error, nut eccentricity error, and nominal diameter error of each roller. It was found that effectively coordinating and controlling the screw eccentricity error, nut eccentricity error, and roller nominal diameter error can reduce the requirement for machining consistency among the nominal diameters of multiple rollers. This is beneficial for improving roller machining efficiency and reducing roller machining costs, and can also provide technical solutions for the selection and assembly of planetary roller screw components.

[0062] like Figure 2As shown, by combining the Pareto front solution set optimization and optimal strategy determination method for the thread pair load distribution coefficient, the optimal solutions for the load distribution of the thread pair on the roller screw side and the roller nut side were determined. Compared with the initial thread pair load distribution, the uniformity of the load distribution on the roller screw side and the roller nut side is significantly improved after optimization. The maximum and minimum load distribution coefficients of the optimized roller screw side thread pair are 0.07714 and 0.07669, respectively, and the maximum and minimum load distribution coefficients of the optimized roller nut side thread pair are 0.07771 and 0.07654, respectively. These results indicate that the optimized load distribution coefficient is very close to its ideal result (0.07692). The difference between the maximum and minimum values ​​of the optimized load distribution coefficients of the roller screw side and the roller nut side thread pair ( B SR and B NR The values ​​were 0.00045 and 0.00117 respectively, indicating that the uniformity of load distribution between the threaded pairs on the roller screw side and the roller nut side improved by nearly 100 times and 20 times. This demonstrates that the difference in load distribution between the threaded pairs on the roller screw side and the roller nut side was significantly reduced after optimization, which is beneficial for improving the load-bearing performance of the planetary roller screw. Figure 3 The paper presents the optimal matching scheme for multi-source errors corresponding to the optimized load distribution of the threaded pairs on the roller screw side and the roller nut side. The table in the figure shows the pitch error values ​​of the screw, nut and roller in the optimal matching scheme. This result shows that effectively coordinating and controlling the pitch errors of the screw, nut and roller can effectively reduce the requirement for consistent pitch of the threaded pairs during the machining process, improve the machining efficiency of parts and reduce the machining cost.

[0063] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for optimizing load sharing in planetary roller screws based on multi-source error matching, characterized in that, Includes the following steps: S1. Based on the preset multi-roller load distribution theoretical model, a complete dataset of multi-roller load distribution coefficients is obtained through simulation calculation and data acquisition. S2. Using this dataset as a sample, construct a neural network surrogate model to establish a precise mapping relationship between multi-source errors and multi-roller load distribution coefficients; S3. Based on this mapping relationship, construct a single-objective uniformity optimization model and solve for the optimized multi-roller load distribution coefficient and the corresponding multi-source error matching parameters. S4. Using the optimized multi-roller load distribution coefficient obtained from S3 as the core input, construct a theoretical model of the threaded pair load distribution and obtain the threaded pair load distribution coefficient dataset. S5. Using this dataset as a sample, construct neural network proxy models for the lead screw side and nut side respectively, and establish the independent mapping relationship between multi-source error and load distribution coefficients on both sides; S6. Based on the mapping relationship between the two sides, construct multi-objective optimization models respectively, and solve them to obtain the Pareto front solution set and multi-source error matrix; S7. Analyze the above results using a multi-objective decision-making method to determine the optimal solution for uniformity of load distribution on both sides and the optimal matching scheme for multi-source errors, and complete the overall optimization.

2. The method for optimizing load sharing of planetary roller screws based on multi-source error matching according to claim 1, characterized in that, In step one, the following is further included: the multi-roller load distribution expression for a planetary roller screw containing multi-source errors can be described as: ； In the formula, The total axial load borne by the planetary roller screw; The overall contact stiffness of a single roller; rollers and Overall deformation difference.

3. The method for optimizing load sharing of planetary roller screws based on multi-source error matching according to claim 1, characterized in that, Step two further includes: a surrogate model for multi-roller load distribution coefficients based on a neural network, with 10 hidden layers, and its input and output layers are as follows: ； In the formula, the input layer includes the lead screw eccentricity error. Nut eccentricity error Error with nominal roller radius The output layer is a multi-roller load distribution coefficient. .

4. The method for optimizing load sharing of planetary roller screws based on multi-source error matching according to claim 1, characterized in that, Step three further includes: The optimization model for the uniformity of multi-roller load distribution with multi-source errors can be expressed as: ； In the formula, and The lower and upper boundaries of the lead screw eccentricity error are defined separately. and The lower and upper boundaries of the nut eccentricity error are defined separately. and The lower and upper boundaries of the nominal diameter error of the roller are respectively defined. The average value of the multi-roller load distribution factor is [value]. , This represents the number of rollers.

5. The method for optimizing load sharing of planetary roller screws based on multi-source error matching according to claim 1, characterized in that, Step four further includes: the load distribution of the threaded pair is closely related to the load borne by each roller. After determining the load borne by each roller, the expression for the load distribution of the threaded pair can be described as: ； In the formula, and These represent the meshing stiffness of the threaded pairs on the roller screw side and the roller nut side, respectively. For the first The threaded pair and the first The difference in deformation between individual threaded pairs.

6. The method for optimizing load sharing of planetary roller screws based on multi-source error matching according to claim 1, characterized in that, Step five further includes: The neural network topology for the load distribution coefficient samples of the threaded pairs on the roller screw side and the roller nut side, with its input and output layers represented as follows: ； In the formula, the input layer includes the lead screw pitch error. Nut pitch error With roller pitch error The output layers are the load distribution coefficients of the roller screw side thread pair. Load distribution coefficient of roller nut side thread pair .

7. The method for optimizing load sharing of planetary roller screws based on multi-source error matching according to claim 1, characterized in that, In step six, the following is further included: the optimization model for the uniformity of load distribution of the threaded pair can be expressed as: ； In the formula, and The lower and upper boundaries of the lead screw pitch error are defined separately. and The lower and upper boundaries of the nut pitch error are respectively defined; and The lower and upper boundaries of the roller pitch error are respectively defined; The average value of the load distribution coefficient for the threaded pair on the roller screw side and the roller nut side is [value missing]. .

8. The method for optimizing load sharing of planetary roller screws based on multi-source error matching according to claim 1, characterized in that, Step seven further includes: the method for determining the optimal solution of the threaded pair load distribution coefficient and the optimal strategy for multi-source errors can be expressed as: ； In the formula, The error matrix corresponding to the Pareto front solution set is optimized to ensure the uniformity of load distribution on the roller screw side and the roller nut side.

Images