A method for optimizing processing parameters for reducing residual stress of a transmission system case

Abstract

This application relates to the field of stress optimization technology, specifically to a method for optimizing machining parameters to reduce residual stress in transmission system housings. The method includes: during high-speed cutting of the transmission system housing, performing dimensionality reduction processing on the multidimensional frequency domain features of cutting force signals, vibration signals, and temperature signals in each local time period; extracting independent potential fluctuation modes characterizing the dynamic changes of the thermal field; quantitatively characterizing the thermo-mechanical coupling response of the dynamic fluctuations of the thermal field; and quantifying the residual stress sensitivity value, which characterizes the degree of influence of the current machining state on residual stress. Using the cutting parameters as optimization variables and minimizing the predicted residual stress value, the machining parameters are optimized. During the solution process, the inertia weight of the optimization algorithm is dynamically adjusted based on the residual stress sensitivity value, and the optimal machining parameters for the transmission system housing are output. This application improves the optimization accuracy of the machining parameters for transmission system housings and effectively reduces the residual stress in the transmission system housing.

Description

Technical Field

[0001] This application relates to the field of stress optimization technology, specifically to a method for optimizing processing parameters to reduce residual stress in the housing of a transmission system. Background Technology

[0002] As the manufacturing positioning benchmark for precision transmission systems, the transmission system housing undergoes a cutting process accompanied by intense thermo-mechanical coupling effects. The residual stress distribution on the material surface directly affects the dimensional stability and fatigue life of the parts. In actual production, due to disturbances such as tool wear and the evolution of the stiffness of thin-walled workpiece structures, the thermal field during machining exhibits significant unsteady dynamic fluctuation characteristics.

[0003] Existing methods for optimizing machining parameters are mostly based on steady-state assumptions, making it difficult to capture and quantify the nonlinear mapping relationship between instantaneous fluctuations in the thermal field and the formation of residual stress in real time. Due to the lack of quantitative characterization of the sensitivity to machining conditions, the residual stress is distributed extremely unevenly along the machining trajectory on the surface of the housing, which easily induces local stress concentration. This causes uncontrollable deformation of the workpiece due to stress release during subsequent service, seriously affecting the dimensional consistency and accuracy stability of batch machining of precision housings. Summary of the Invention

[0004] To address the aforementioned technical problems, this application provides a method for optimizing processing parameters to reduce residual stress in the transmission system housing, thereby resolving the existing issues.

[0005] The method for optimizing machining parameters to reduce residual stress in the transmission system housing, as proposed in this application, adopts the following technical solution: One embodiment of this application provides a method for optimizing machining parameters to reduce residual stress in a transmission system housing. The method includes the following steps: During the high-speed cutting process of the transmission system housing, cutting force signals and vibration signals are collected in real time, and temperature signals in the machining area of ​​the housing and residual stress on key feature surfaces of the housing are monitored. The multidimensional frequency domain features of cutting force signals, vibration signals, and temperature signals in each local time period are reduced in dimension to extract each independent potential fluctuation mode that characterizes the dynamic changes of the thermodynamic field. The thermodynamic coupling response of the dynamic fluctuations of the thermal field is quantitatively characterized by weighted aggregation of the contribution of each potential fluctuation mode and its intensity component in the corresponding local time period. The correlation between the thermo-coupling responsivity and the measured residual stress value is analyzed, and the residual stress level under the current processing state is predicted. Combining the correlation and the residual stress level, a residual stress sensitivity value characterizing the degree of influence of the current processing state on the residual stress is quantified. Using cutting parameters as optimization variables and minimizing the predicted residual stress value as the objective, the machining parameters are optimized. During the solution process, the inertial weight of the optimization algorithm is dynamically adjusted based on the residual stress sensitivity value, and the optimal machining parameters of the transmission system housing are output.

[0006] In one embodiment, the extraction of each independent potential fluctuation pattern characterizing the dynamic changes of the thermal field includes: The cutting force signal, vibration signal, and temperature signal in each local time period are decomposed to obtain the energy distribution in each frequency band. This energy distribution is then used as the input to the factor analysis algorithm to output independent potential fluctuation modes that characterize the dynamic changes of the thermodynamic field.

[0007] In one embodiment, the quantitative characterization of the thermodynamic coupling responsivity to dynamic fluctuations in the thermal field includes: Obtain the variance contribution rate and factor score of each factor output by the factor analysis algorithm, and perform positive fusion of the variance contribution rate and factor score to obtain the thermodynamic coupling response.

[0008] In one embodiment, the calculation process for the thermo-coupling responsivity is as follows: Calculate the product of the variance contribution rate of each factor and the factor score, and sum the products of all factors obtained in each local time period as the thermodynamic coupling responsivity.

[0009] In one embodiment, predicting the residual stress level under the current processing state includes: Using cutting force, vibration, and temperature signals at the current and historical moments as multidimensional feature data, and the measured residual stress at the current and historical moments as the target value, the residual stress level at the next moment is predicted using the random forest regression algorithm.

[0010] In one embodiment, the quantification process of the residual stress sensitivity value is as follows: The residual stress sensitivity value is determined based on the correlation between the residual stress sensitivity value and the aforementioned correlation, as well as the positive correlation between the residual stress level and the residual stress sensitivity value.

[0011] In one embodiment, the expression for the residual stress sensitivity value is: In the formula, denoted as the current residual stress sensitivity value, I as the correlation between the thermo-coupling response and the measured residual stress value, H as the predicted residual stress level under the current processing condition, and e as the natural constant.

[0012] In one embodiment, the inertia weights of the optimization algorithm dynamically adjusted based on the residual stress sensitivity value include: Calculate the normalized result of the residual stress sensitivity value to determine the adjustment span between the minimum and maximum values ​​of the inertia weight of the optimization algorithm; use the normalized result to weight the adjustment span, and obtain the adjusted inertia weight based on the weighted result and the maximum value.

[0013] In one embodiment, the difference between the maximum value and the weighted result is calculated as the adjusted inertia weight.

[0014] In one embodiment, the optimal machining parameters for the output transmission system housing include: By setting the iteration cutoff number for the particle swarm optimization algorithm, after all iterations are completed, the combination of cutting parameters that minimizes the predicted residual stress is obtained as the optimal machining parameters.

[0015] This application has at least the following beneficial effects: This application overcomes the limitation of single time-domain features in fully quantifying the fluctuation characteristics of complex cutting processes by performing multi-dimensional frequency domain feature reduction and fusion of cutting force, vibration, and temperature signals. It extracts independent potential fluctuation modes and calculates the thermo-mechanical coupling responsivity, effectively identifying instantaneous energy mutations caused by unsteady disturbances such as tool wear and part stiffness evolution. This directly corresponds to the physical causes of residual stress inhomogeneity during box machining, providing scientific feedback indicators for subsequent precise control and achieving refined quantification and real-time characterization of unsteady fluctuations in the machining thermal field. Secondly, through mutual information analysis, the quantitative dependence strength between thermo-mechanical fluctuations and residual stress is established, and the residual stress level is predicted in real time. This not only considers static machining parameter settings but also fully integrates the fluctuation information of the dynamic cutting process, establishing a multi-dimensional, high-precision mapping relationship. The introduction of a residual stress sensitive value greatly enhances the ability to detect high-stress-risk conditions. The algorithm improves the sensitivity and prediction accuracy of residual stress state identification, enhancing the generalization ability and relevance interpretability of residual stress prediction. Furthermore, by adaptively adjusting the inertia weight of the optimization algorithm based on the residual stress sensitivity value, it overcomes the problem of traditional optimization strategies being prone to blindness and getting trapped in local optima when dealing with high-dimensional nonlinear parameter optimization. The algorithm enhances the local optimization accuracy in the stress-sensitive region and accelerates the global search efficiency in the stable region. This dynamic feedback adjustment mechanism enables the processing parameters of the transmission system housing to be optimized in real time according to the working conditions, effectively reducing the peak fluctuation of residual stress and its distribution range on complex paths, and significantly optimizing the global search efficiency and convergence accuracy of the parameter solving algorithm. This application effectively reduces the late-stage aging deformation of parts caused by local concentration of residual stress, ensuring the fatigue resistance and precision life of key components of high-end equipment under complex service conditions, and improving the overall yield of mass production of precision housings. Attached Figure Description

[0016] Figure 1 A flowchart illustrating the steps of a method for optimizing processing parameters to reduce residual stress in a transmission system housing, as provided in this application. Detailed Implementation

[0017] The following description, in conjunction with the accompanying drawings, details a specific scheme for optimizing processing parameters to reduce residual stress in the transmission system housing, as provided in this application.

[0018] This application provides an embodiment of a method for optimizing machining parameters to reduce residual stress in a transmission system housing. Specifically, it provides the following method for optimizing machining parameters to reduce residual stress in a transmission system housing. Please refer to [link to relevant documentation]. Figure 1 The method includes the following steps: Step S001: During the high-speed cutting process of the transmission system housing, the cutting force signal and vibration signal are collected in real time, and the temperature signal of the housing machining area and the residual stress on the key feature surfaces of the housing are monitored.

[0019] During the high-speed cutting process of the transmission system housing, a triaxial cutting force signal is collected in real time by a triaxial piezoelectric force gauge arranged on the positioning reference surface of the housing, and the amplitudes are arranged in time sequence to form a cutting force sequence; a triaxial vibration signal is collected in real time by an accelerometer arranged in the key thin-walled feature area of ​​the housing, and the amplitudes are arranged in time sequence to form a vibration sequence. Infrared thermal imagers are used to monitor the temperature field evolution in the processing area of ​​the enclosure, and the highest temperature value in each frame is extracted to form a one-dimensional transient temperature sequence.

[0020] The residual stress of key characteristic surfaces such as bearing holes and mounting end faces of the housing was measured offline using an X-ray diffractometer. Based on the spatial physical coordinates of the measurement points on the housing and the tool path recorded by the CNC system, the residual stress amplitude was spatiotemporally correlated and aligned with the cutting force, temperature, and vibration signal sequences at the corresponding machining moments to establish a mapping dataset between the structural features of the housing and the machining state.

[0021] In this embodiment, the data acquisition frequency for the four signals—cutting force, temperature, vibration signal, and residual stress—was set to 10kHz. Min-Max normalization was performed on all data acquired for each signal. The processed data at each acquisition moment and its historical moments were sorted in ascending order of time to obtain the cutting force sequence, temperature sequence, vibration sequence, and residual stress amplitude sequence for each acquisition moment. To ensure data validity, the cutting parameters were optimized 10 seconds after the high-speed cutting of the transmission system housing began.

[0022] Step S002: Dimensionality reduction processing is performed on the multidimensional frequency domain features of cutting force signal, vibration signal, and temperature signal in each local time period to extract each independent potential fluctuation mode that characterizes the dynamic changes of the thermal field.

[0023] During high-speed cutting, the transmission system housing will experience unsteady dynamic changes in the cutting thermal field due to tool wear and time-varying workpiece stiffness. These unsteady dynamic changes will cause the surface material of the workpiece to bear different thermal loads at different times, which will further lead to uneven application of residual stress and the formation of local stress concentration areas.

[0024] Based on the above analysis, using the cutting force sequence, temperature sequence, and vibration sequence at each acquisition moment as input, a wavelet packet decomposition algorithm is used to construct a sliding time window. Each sliding time window corresponds to a local time period, and the continuous sequence within the sliding time window is used as the basic input unit for wavelet packet decomposition. In this embodiment, the length of the sliding time window is set to 100 milliseconds, and the overlap rate between adjacent windows is 50%, thereby ensuring the continuity and stability of frequency domain feature extraction. Implementers can set this according to actual conditions. The number of wavelet packet decomposition layers is set to 3 to cover the main frequency band range during the cutting process, while also considering the need for computational efficiency and frequency band subdivision. Finally, the db4 wavelet basis function is selected to obtain the energy distribution of each frequency band within each time window in the cutting force sequence, temperature sequence, and vibration sequence. The energy of each frequency band directly represents the fluctuation intensity of different frequency bands of the cutting force, temperature, and vibration signals, naturally quantifying the frequency domain characteristics of the dynamic changes in the thermal field. The wavelet packet decomposition algorithm is a well-known existing technology, and the specific process will not be elaborated.

[0025] However, the number of multi-band energy features obtained by the above wavelet packet decomposition is large, and the information contained in the multi-band features is redundant, making it difficult to intuitively reflect the overall law of thermal field fluctuations.

[0026] To address the aforementioned issues, a vector composed of energy characteristics from each frequency band within each time window of the cutting force, temperature, and vibration sequences is used as input. Factor analysis is employed, with principal component analysis (PCA) retained to preserve common factors with eigenvalues ​​greater than 1. Variance maximization rotation is used to enhance factor interpretability. The final output includes a factor loading matrix, the variance contribution rate of each factor, and the factor score for each factor. The output factor loading matrix reflects the correlation strength between the energy in each frequency band and the common factors; the variance contribution rate of each factor represents the proportion of the total variance of the original data for that factor; and the factor score characterizes the relative deviation of the sample from each factor. By reducing the dimensionality of the high-dimensional cutting frequency domain energy characteristics through factor analysis, the complex frequency domain fluctuations of the thermodynamic field can be quantified into a few independent potential fluctuation modes. Both factor analysis and principal component analysis are existing, well-known techniques, and their specific processes are not detailed here.

[0027] Step S003: Based on the contribution of each potential fluctuation mode and its intensity component in the corresponding local time period, a weighted aggregation is performed to quantitatively characterize the thermodynamic coupling response of the dynamic fluctuation of the thermal field.

[0028] The variance contribution rate and factor score of each factor output by the factor analysis algorithm are obtained. These variance contribution rates and factor scores are then positively fused to obtain the thermodynamic coupling responsivity within each sliding time window. Positive fusion combines multiple variables, which can be achieved through methods such as addition, multiplication, or averaging.

[0029] In this embodiment, the expression for the thermal coupling responsivity within each sliding time window is: In the formula, The thermal coupling responsivity within each sliding time window, The variance contribution rate of the j-th factor obtained within each sliding time window. The factor score of the j-th factor obtained within each sliding time window. The number of factors acquired within each sliding time window.

[0030] This embodiment quantifies the overall intensity of dynamic fluctuations in the thermal field at the current moment by integrating the contributions of each independent potential fluctuation mode through the calculated variance contribution rate and factor score. The larger the obtained thermo-coupling responsibility A value, the more severe the influence of disturbances such as tool wear and workpiece stiffness changes on the machining process, the more unstable the thermo-coupling effect, and the higher the risk of uneven residual stress distribution and local stress concentration.

[0031] Step S004: Analyze the correlation between the thermo-mechanical coupling response and the measured value of residual stress, and predict the residual stress level under the current processing state; combine the correlation and the residual stress level to quantify the residual stress sensitivity value that characterizes the degree of influence of the current processing state on the residual stress.

[0032] During high-speed cutting, the cutting thermodynamic field exhibits significant time-varying characteristics due to real-time fluctuations in cutting parameters and dynamic changes in the tool-workpiece system. This unsteady fluctuation in the thermodynamic field leads to inconsistent thermal loads on the surface material of the transmission system housing at different processing times, resulting in discrete and non-uniform residual stress distribution within the housing. This is the fundamental reason affecting the stability and dimensional consistency of precision batch machining of housings.

[0033] Based on this, this embodiment obtains the thermo-coupling responsivity A within each time window at each acquisition moment, sorts the thermo-coupling responsivity within all time windows in chronological order, thereby obtaining the thermo-coupling sequence at each acquisition moment, and then uses a k-nearest neighbor-based mutual information analysis algorithm, taking the thermo-coupling sequence and the residual stress amplitude sequence of the box at the same moment as input, taking the nearest neighbor number k=5, and outputting the mutual information value I, which is the dependence strength between the thermo-coupling responsivity A and the residual stress, expressing a quantitative explanation of the formation of residual stress by thermal field fluctuations. The k-nearest neighbor-based mutual information analysis algorithm is a well-known existing technology, and this embodiment will not elaborate on it.

[0034] However, the mutual information value I obtained by the mutual information analysis algorithm can only measure the correlation between thermal field fluctuations and residual stress, and it is difficult to directly give the risk level of residual stress under specific processing conditions. Therefore, this embodiment uses the cutting force sequence, temperature sequence, and vibration sequence at the current acquisition time as multidimensional feature data, and the residual stress amplitude sequence of the box at the current acquisition time as the target value. Both the multidimensional feature data and the target value are used as inputs to the random forest regression algorithm. The number of decision trees is set to 100 to ensure the generalization ability of the random forest regression, and the maximum depth is set to 10 to prevent overfitting. The output is the predicted residual stress value H at the next time step. The output predicted residual stress value represents the residual stress level that may occur under the current processing parameters and thermal field conditions.

[0035] Based on the above analysis, the residual stress sensitivity value at the current acquisition time is calculated, and the specific calculation expression is as follows: In the formula, denoted as the current residual stress sensitivity value, I as the correlation between the thermo-coupling response and the measured residual stress value, H as the predicted residual stress level under the current processing condition, and e as the natural constant.

[0036] This embodiment quantitatively analyzes the dependence between thermo-coupling responsivity and measured residual stress using mutual information. A higher value indicates a stronger correlation between current thermal field fluctuations and residual stress formation. Simultaneously, a random forest regression model is used to predict the residual stress level corresponding to the current processing state. Based on this, a residual stress sensitivity value B is calculated by combining the correlation and the predicted residual stress value. This comprehensively quantifies the sensitivity of current processing state fluctuations to residual stress from two dimensions: correlation weight and residual stress level. The introduction of the residual stress sensitivity value B provides a dynamic evaluation index for the adaptive adjustment of inertia weights in the subsequent particle swarm optimization algorithm, enabling real-time feedback on high-risk residual stress states during processing.

[0037] Step S005: Using the cutting parameters as optimization variables and minimizing the predicted residual stress value as the objective, the machining parameters are optimized. During the solution process, the inertia weight of the optimization algorithm is dynamically adjusted based on the residual stress sensitivity value, and the optimal machining parameters of the transmission system housing are output.

[0038] When dealing with high-dimensional nonlinear machining parameter optimization problems, the standard particle swarm optimization algorithm is prone to getting trapped in local optima and cannot meet the requirements of residual stress control for the transmission system housing. Therefore, this embodiment adaptively adjusts the inertia weight in the particle swarm optimization algorithm based on the residual stress sensitivity value B calculated above. Specifically, the optimization variables of the particle swarm optimization algorithm are defined as the cutting speed, feed rate, and depth of cut during the high-speed cutting process of the transmission system housing; the fitness function is defined as minimizing the residual stress prediction value output by the random forest model.

[0039] In particle swarm optimization (PSO) algorithms, the inertia weight determines the degree to which particles inherit the velocity from the previous moment, directly balancing the algorithm's global search capability and local optimization accuracy. However, traditional linear decreasing weight strategies rely solely on the iterative process and fail to respond in real-time to the dynamic changes in the current processing state. To address the complex residual stress fluctuations in the transmission system housing processing, this embodiment introduces a residual stress-sensitive value B to adaptively and dynamically adjust the inertia weight. Specifically, the expression for the inertia weight adjusted in each iteration of the PSO algorithm is: ;in, Here are the inertia weights adjusted by the particle swarm optimization algorithm in the t-th iteration. and These represent the maximum and minimum values ​​of the inertia weight, respectively. The maximum value ranges from 0.8 to 0.9, and the minimum value ranges from 0.3 to 0.4. In this embodiment, the maximum inertia weight is set. The minimum inertia weight is 0.9. It is 0.4. This is the sigmoid normalization function.

[0040] It should be understood that by providing real-time feedback on the residual stress sensitivity value B, the inertia weight adaptively switches according to the sensitivity of the processing state. When the residual stress sensitivity is low, the inertia weight is increased to improve the global search efficiency and find a better combination of processing parameters. When the residual stress sensitivity is high and there is a risk of stress concentration, the inertia weight is decreased to strengthen the local fine search, thereby achieving rapid convergence of the optimal control parameters. This improves the robustness and convergence accuracy of the optimization process, overcomes the problem of insufficient convergence performance of traditional fixed weight strategies under unsteady conditions, and significantly improves the real-time performance and stability of parameter optimization.

[0041] It should be noted that in the particle swarm optimization algorithm optimization process, this embodiment sets the population size to 50 and the maximum number of iterations to 200. The position vector of each particle has a dimension of 3, corresponding to the cutting speed, feed rate, and depth of cut, respectively. The individual learning factor c1 and the group's social learning factor c2 are both set to 2.0. Based on the constraints of the specific machining parameters, the boundaries of the search space are limited: the cutting speed range is 100~300 m / min, the feed rate range is 0.05~0.2 mm / r, and the depth of cut range is 0.1~0.5 mm. The initial particle positions are randomly generated within the constraints, and the initial velocity is set to 0. After the particle swarm optimization algorithm iteration ends, the cutting speed, feed rate, and depth of cut that minimize the residual stress prediction value output by the random forest model are obtained as the optimal combination of machining parameters. Furthermore, the machining trajectory of the transmission system housing is divided into several machining segments according to machining characteristics. Each machining segment corresponds to an independent structural feature area on the housing. The division of machining segments is based on tool path reversal points, depth of cut change points, or machining feature switching points in the CNC program. The machining time of each segment is typically tens of seconds to several minutes, much longer than the signal analysis time. At the beginning of each machining segment, the cutting parameters are adjusted once, and the cutting parameters within the machining segment remain stable. The cutting parameters determined at the beginning of the machining segment are the optimal combination of machining parameters that appears most frequently in the previous machining segment. If the optimal combination of machining parameters does not appear most frequently in the previous machining segment, the optimal combination of machining parameters at the last moment of the previous machining segment is used as the cutting parameters for the next machining segment, effectively reducing the residual stress in the transmission system housing.

Claims

1. A method for optimizing machining parameters to reduce residual stress in a transmission system housing, characterized in that, The method includes the following steps: During the high-speed cutting process of the transmission system housing, cutting force signals and vibration signals are collected in real time, and temperature signals in the machining area of ​​the housing and residual stress on key feature surfaces of the housing are monitored. The multidimensional frequency domain features of cutting force signals, vibration signals, and temperature signals in each local time period are reduced in dimension to extract each independent potential fluctuation mode that characterizes the dynamic changes of the thermodynamic field. The thermodynamic coupling response of the dynamic fluctuations of the thermal field is quantitatively characterized by weighted aggregation of the contribution of each potential fluctuation mode and its intensity component in the corresponding local time period. The correlation between the thermo-coupling responsivity and the measured residual stress value is analyzed, and the residual stress level under the current processing state is predicted. Combining the correlation and the residual stress level, a residual stress sensitivity value characterizing the degree of influence of the current processing state on the residual stress is quantified. Using cutting parameters as optimization variables and minimizing the predicted residual stress value as the objective, the machining parameters are optimized. During the solution process, the inertial weight of the optimization algorithm is dynamically adjusted based on the residual stress sensitivity value, and the optimal machining parameters of the transmission system housing are output.

2. The method for optimizing machining parameters to reduce residual stress in a transmission system housing as described in claim 1, characterized in that, The extraction of independent potential fluctuation patterns characterizing the dynamic changes of the thermal field includes: The cutting force signal, vibration signal, and temperature signal in each local time period are decomposed to obtain the energy distribution in each frequency band. This energy distribution is then used as the input to the factor analysis algorithm to output independent potential fluctuation modes that characterize the dynamic changes of the thermodynamic field.

3. The method for optimizing machining parameters to reduce residual stress in a transmission system housing as described in claim 2, characterized in that, The quantitative characterization of the thermodynamic coupling responsivity of the dynamic fluctuations of the thermal field includes: Obtain the variance contribution rate and factor score of each factor output by the factor analysis algorithm, and perform positive fusion of the variance contribution rate and factor score to obtain the thermodynamic coupling response.

4. The method for optimizing machining parameters to reduce residual stress in a transmission system housing as described in claim 3, characterized in that, The calculation process for the thermo-coupling responsivity is as follows: Calculate the product of the variance contribution rate of each factor and the factor score, and sum the products of all factors obtained in each local time period as the thermodynamic coupling responsivity.

5. The method for optimizing machining parameters to reduce residual stress in a transmission system housing as described in claim 1, characterized in that, The prediction of the residual stress level under the current processing state includes: Using cutting force, vibration, and temperature signals at the current and historical moments as multidimensional feature data, and the measured residual stress at the current and historical moments as the target value, the residual stress level at the next moment is predicted using the random forest regression algorithm.

6. The method for optimizing machining parameters to reduce residual stress in a transmission system housing as described in claim 1, characterized in that, The quantification process of the residual stress sensitivity value is as follows: The residual stress sensitivity value is determined based on the correlation between the residual stress sensitivity value and the aforementioned correlation, as well as the positive correlation between the residual stress level and the residual stress sensitivity value.

7. The method for optimizing machining parameters to reduce residual stress in a transmission system housing as described in claim 6, characterized in that, The expression for the residual stress sensitivity value is: In the formula, denoted as the current residual stress sensitivity value, I as the correlation between the thermo-coupling response and the measured residual stress value, H as the predicted residual stress level under the current processing condition, and e as the natural constant.

8. The method for optimizing machining parameters to reduce residual stress in a transmission system housing as described in claim 1, characterized in that, The inertia weights of the optimization algorithm based on the residual stress sensitivity value include: Calculate the normalized result of the residual stress sensitivity value to determine the adjustment span between the minimum and maximum values ​​of the inertia weight of the optimization algorithm; use the normalized result to weight the adjustment span, and obtain the adjusted inertia weight based on the weighted result and the maximum value.

9. The method for optimizing machining parameters to reduce residual stress in a transmission system housing as described in claim 8, characterized in that, The difference between the maximum value and the weighted result is calculated and used as the adjusted inertia weight.

10. The method for optimizing machining parameters to reduce residual stress in a transmission system housing as described in claim 1, characterized in that, The optimal machining parameters for the output transmission system housing include: By setting the iteration cutoff number for the particle swarm optimization algorithm, after all iterations are completed, the combination of cutting parameters that minimizes the predicted residual stress is obtained as the optimal machining parameters.

Images