Dynamic meshing force analysis of parallel planetary gear based on RBF neural network

The coupling phenomenon of meshing force often appears in the dynamic response of coupled planetary gears. In previous scholars’ research, Zheng Mingyin and others used Adams to establish the rigid flexible coupling dynamic model of the double planetary dynamic coupling mechanism, and obtained that the dynamic coupling mechanism mainly includes the rotation frequency in the low frequency band and the meshing frequency and its frequency doubling of the two planetary rows under the steady and unsteady conditions; Dou zuocheng et al. Synthesized the meshing force of compound planetary gear under complex excitation and obtained the modulation phenomenon of planetary gear due to time-varying meshing stiffness, various meshing errors and periodic fluctuation of external excitation through time-frequency conversion. A series of meshing frequencies and their frequency doubling are carrier frequencies in the frequency spectrum; Based on the coupling relationship between the force and deformation of the connecting shaft between the stages of two-stage planetary gears, Liu Hui et al. Studied that the vibration displacement of the connected parts with new frequency after coupling will be fed back to the planetary meshing force of the stage in the form of equivalent meshing line deformation, resulting in the mutual coupling between the planetary meshing forces of different stages, which is called coupling meshing frequency; Wei Jing et al. Realized the virtual equivalent shaft element dynamic modeling method on the two-stage planetary gear system, and obtained the conclusion that the coupling characteristics between the stages of the multi-stage planetary gear system are obvious, and the vibration energy is transferred along the opposite direction of the power direction; Wang Jungang et al. Took the two-stage planetary gear train of load split wind power gearbox as the research object, deduced the model and obtained that the coupling vibration frequency between stages can be controlled by changing the coupling stiffness between planetary rows.

For the meshing force of coupling planetary gear, many scholars have shown that the frequency coupling phenomenon of meshing force is obvious, but there is no unified statement about the frequency coupling law of meshing force. Based on the analysis of the relationship between the meshing force frequency of the parallel double planetary gear coupling mechanism, it is concluded that the current meshing force frequency ratio minus the coupling meshing force frequency ratio is proportional to the input speed ratio minus the output side load ratio, but there is a nonlinear relationship between the meshing force frequency ratio and the speed load ratio.

In order to objectively evaluate the howling quality of transmission, the RBF neural network was used to predict several objective evaluation parameters of sound quality, and three main parameters were found out from 11 evaluation parameters by using the connection weights of each network layer; Li Chunxiang applied RBF neural network to improve the traditional harmonic superposition method, and applied it to the calculation of wind load, which not only has high accuracy, but also greatly improves the calculation efficiency; In order to solve the dynamic calculation problem of nonlinear structure, Zhou Chungui and others proposed a new nonlinear dynamic method by combining RBF neural network model with finite element method of linear model. RBF neural network has the characteristics of self-learning, self-organization and self-adaptive. For some nonlinear data, RBF neural network can be used for training, so as to achieve the purpose of numerical prediction. It not only solves the problem of computational efficiency, but also can more completely express the nonlinear relationship.

(1) After clustering some known data and applying RBF neural network model for training, the error range of the prediction results of the training group data is within 5.5%, and that of the test group data is within 6%. It can be seen that RBF neural network model is effective in predicting the influence proportion of the meshing force frequency of parallel planetary gears. The trained RBF neural network model is used to predict the speed ratio and load ratio in a certain range. The results are similar to the previous results, and the curve is smoother than the original results.

(2) The maximum meshing force and dynamic load coefficient are calculated according to the prediction results. The results show that the meshing force increases linearly with the increase of the difference between the speed ratio and the output torque and 1. When the output torque ratio is greater than 1, the dynamic load coefficient is a multiple of its reciprocal. When the output torque ratio is less than 1, the dynamic load coefficient is a multiple of its reciprocal.

(3) When P20 is a high-speed platoon and P10 resistance is less than 0.2 times of P20 resistance, P10 platoon is coupled by P20 platoon, and its meshing force and dynamic load coefficient are large, which may cause large amplitude vibration. The vehicle should avoid working in this section as far as possible. The measures to reduce the phenomenon of large dynamic load coefficient in this section need to be further studied.

Based on the nonlinear relationship that the frequency influence ratio of engagement force is several input parameters, the RBF neural network algorithm is used to train the known data, and make more detailed prediction within the range of known parameters. The relationship between dynamic engagement force and input and output parameters is further analyzed.