Household appliance wire harness aging resistance prediction method

Abstract

The invention provides a household appliance wire harness aging resistance prediction method, and relates to the field of data prediction.The household appliance wire harness aging resistance prediction method comprises a time sequence decomposition preprocessing module, a graph connection strength tensor mapping module, a space-time gravitation extraction module and a structure response predictor module.The method comprises the steps that firstly, multi-dimensional time sequence data of a household appliance wire harness is extracted, trend, seasonal and residual decomposition is conducted on the multi-dimensional time sequence data, and the multi-dimensional time sequence data is obtained; refined input features are provided for subsequent modules; then, a graph connection strength tensor mapping module quantifies the coupling strength and mutual influence between components by constructing a high-order structure response tensor between variables, and generates a dynamic tensor basis for mapping; then, combining the structure curvature with the graph connection strength, accurately extracting the temporal and spatial change characteristics of the household appliance wire harness under different use conditions, and further enhancing the perception ability of the prediction model to the aging process; and finally, through dynamic differential modeling, the future aging trend of the household appliance wire harness is predicted by using the disturbance response and the nonlinear regulation factor, and a final result is output.

Description

Technical Field

[0001] The present invention belongs to the field of data prediction, and in particular relates to a method for predicting the aging resistance of household wiring harnesses. Background Art

[0002] With the popularization of home appliances and the continuous improvement of people's living standards, home appliance harnesses, as key electrical connection components in home appliances, are easily affected by various environmental factors during long-term use, and gradually age, fatigue and fail. The aging problem of home appliance harnesses not only affects the performance and safety of the products, but may also lead to serious accidents. Therefore, it is of great significance to accurately predict the aging resistance of home appliance harnesses.

[0003] In recent years, with the continuous development of sensing technology, the Internet of Things, and big data analysis technology, researchers have begun to try to predict aging trends by collecting real-time data of household wiring harnesses under different working conditions and combining them with machine learning and deep learning technologies. These methods can predict the aging of wiring harnesses to a certain extent by monitoring relevant data in real time and using historical data for trend analysis. However, existing methods still have limitations, lack of accurate modeling of multi-dimensional and spatiotemporal coupling change characteristics, insufficient adaptability to different working environments, low prediction accuracy, poor real-time performance, and difficulty in dealing with sudden aging or failure risks.

[0004] In order to solve these problems, the present invention proposes a method for predicting the aging resistance of household wire harnesses based on multivariable dynamic coupling and structural evolution modeling. By introducing spatiotemporal gravitational flow modeling and high-order differential response mechanism, combined with multi-source monitoring data of household wire harnesses, it can capture the nonlinear coupling relationship between variables in real time and accurately extract the spatiotemporal change characteristics of the aging process. In addition, based on the introduction of graph structure generation and adaptive adjustment algorithm, it can effectively improve the robustness and real-time performance of the prediction model, ensure efficient prediction in complex environments, and thus improve the reliability and safety of household wire harnesses, providing strong support for preventing household appliance product failures and extending their service life. Summary of the Invention

[0005] The present invention provides a method for predicting the aging resistance of household wiring harnesses. A prediction model is proposed for complex and multivariate household wiring harness data. The prediction model consists of a time series decomposition preprocessing module, a graph connection strength tensor mapping module, a spatiotemporal gravity extraction module, and a structural response predictor module.

[0006] The technical solution adopted by the present invention to achieve the above-mentioned purpose specifically includes the following steps: S1. Collect multi-source monitoring data of household wiring harnesses, including electrical performance data, temperature, humidity, and mechanical stress characteristics, and construct an original multidimensional dataset; S2, propose a time series decomposition preprocessing module, use high-order time derivatives to build an adaptive disturbance feedback mechanism, and couple the interaction matrix Dynamically construct the nonlinear synergistic relationship between multiple variables and dynamically adjust the interaction intensity between variables. Finally, introduce the disturbance perception factor to dynamically weight and fuse the trend structure and nonlinear information to obtain the first data set. ; S3, design graph connection strength tensor mapping module, build high-order differential tensor representation mechanism based on structural trajectory changes, introduce covariant response tensor , and then map the structure tensor to the graph structure space to obtain the connection strength matrix , as the second data set; S4. Construct a spatiotemporal gravity extraction module and combine it with the connection strength matrix and the curvature response tensor Define the structural tensor gravitational field response and finally get the comprehensive gravitational field strength As a third data set, the third data set is divided into a training set and a prediction set; S5. Construct a structural response predictor module and use the comprehensive gravitational field strength Calculate the dynamic state change rate of household wiring harnesses , construct an adaptive trend tensor , proposed the structural excitation function and calculated the aging dynamic change rate of household wiring harness components, and finally obtained the aging resistance prediction model; S6. The training set is input into the prediction module, and the structure prediction error loss is defined. Continuously optimize and train the prediction model for the aging resistance of household wiring harnesses; S7: The prediction set is input into the trained home wiring harness aging resistance prediction model, and the predicted value of the wiring harness aging resistance is finally output. .

[0007] Optimally, S1 collects multi-source monitoring data of household wiring harnesses, including electrical performance data, temperature, humidity, and mechanical stress characteristics, to construct an original multidimensional data set, wherein the electrical performance data includes resistance, voltage, current, power factor, conductivity, and contact resistance, and the mechanical stress characteristics include axial tensile force, bending radius, impact strength, and vibration frequency spectrum.

[0008] Preferably, traditional aging modeling methods are difficult to effectively identify complex disturbance signals and multivariable collaborative degradation behaviors. In practical applications, the aging process of household wiring harnesses is usually accompanied by nonlinear changes, high-order fluctuations and dynamic coupling between components. If only static modeling or low-order statistical features are relied upon, key degradation information is easily ignored, resulting in prediction lags. For this reason, the present invention constructs a modeling framework with disturbance response as the core from the bottom layer, introduces high-order derivatives to capture continuous evolution trends, introduces adaptive control factors to cope with the time-varying and heterogeneous nature of the aging process, and quantifies the coupling strength between variables through the interactive coupling matrix to achieve dynamic fusion reconstruction at the structural level, thereby ensuring the dynamics, refinement and engineering interpretability of the modeling process, and providing a more solid feature foundation for subsequent predictions.

[0009] Furthermore, in step S2, the implementation process of the time series decomposition preprocessing module includes: S21. First, an adaptive perturbation feedback mechanism is constructed to effectively characterize the microstructural evolution behavior of wiring harnesses during long-term operation, revealing the nonlinear driving law of wiring harness aging from the bottom layer and dynamically capturing data. The mathematical model of the local perturbation evolution process is: ; Where, It is the high-order derivative of the input state data in the time dimension, which is used to extract the multi-order dynamic response characteristics of the wiring harness aging process. For the The disturbance response weight corresponding to the order derivative term is used to reflect the importance of different order disturbances in aging modeling. is a local adaptive adjustment factor that controls the intensity of the disturbance at each time point. To perturb the nonlinear adjustment index, adjust the response strength of the local state value, and enhance the sensitivity of the model to sudden aging events, The maximum order of higher-order derivatives considered in perturbation modeling; S22. In order to capture the coupling effect between multivariate time series data, the present invention proposes a disturbance coupling interaction matrix modeling method for dynamically constructing the nonlinear synergistic relationship between multiple variables. Based on the interactive influence of data, we can accurately model the cross-variable disturbance dependency in multivariate data and dynamically adjust the interaction strength between different variables. The mathematical model is: ; Where, Describes the variable and At the moment The perturbation coupling strength, is the static coupling strength weight of each pair of variables, controlling the nonlinear coupling of disturbances between the variables, is the attenuation factor, which is used to control the impact of historical disturbances on the current moment. is the length of the historical time window, and then the time series data Decomposition into trend components , seasonal part and the residual , in order to more accurately extract the evolution components and abnormal disturbances in the aging process, the mathematical model is: ; ; ; Where, is the trend component of time series data, is the total number of variables, is the local adaptation factor, is the adaptive exponential adjustment factor, is the seasonal component, which is used to extract the synergistic trend of long-term periodic structural changes among bundle variables. It is the residual component, which is used to represent non-periodic disturbances and mutation behaviors that cannot be modeled by trend and seasonal decomposition, reflecting the unstructured noise in the aging process. Then, the disturbance perception factor is introduced to dynamically weight and fuse the trend structure and nonlinear disturbance information to complete the final data reconstruction. The mathematical model is: ; Where, is the disturbance perception factor, and finally the adaptive fusion reconstruction output is obtained as the first dataset.

[0010] Preferably, the time series decomposition preprocessing module can dynamically analyze the nonlinear aging trend and multivariable coupling characteristics of household wiring harnesses in long-term operation. By introducing high-order derivatives and disturbance adaptive adjustment factors, it can accurately capture weak structural degradation signals, and then use the interactive coupling matrix to characterize the coupled aging correlation between multiple components. Combining trend, cycle and residual decomposition to form a high-quality time series reconstruction representation, compared with traditional single-variable or static processing methods, the time series decomposition preprocessing module not only has high sensitivity to the dynamic evolution of aging, but also can explore collaborative degradation patterns at the component level, greatly improving the prediction model's response ability to early signs of aging and early warning capabilities for complex systemic failures.

[0011] Preferably, in the modeling of the aging resistance process of household wiring harnesses, traditional methods usually assume that the connection relationship between the channel variables is static and homogeneous, which makes it difficult to reflect the evolution of microstructural differences in the wiring harness structure caused by temperature rise, load fluctuation, and insulation degradation during long-term use. Therefore, the present invention introduces a graph connection strength tensor mapping module to define the high-order dynamic response of the structural trajectory between variables from the bottom layer, and use the structural derivative tensor to measure the similarity and divergence between variables, thereby constructing a graph connection strength representation with dynamic variability and physical interpretability. The design of the module can effectively reveal the non-stationary collaborative degradation pattern between multiple variables within the wiring harness, laying a key foundation for constructing a spatiotemporal graph structure for the real aging process.

[0012] Furthermore, in order to effectively extract the response trajectory characteristics of each structural path of the household wiring harness during the aging process, after obtaining the first data set, the present invention constructs a high-order differential tensor representation mechanism driven by the structural trajectory change in step S3. , defined at a point in time The structural trajectory tensor of is, and the mathematical model is: ; Where, For variables At the moment No. The order structural trajectory tensor is used to measure the dynamic response strength of the bundle variables at different microstructural levels. is the differential order of modeling, which is used to represent the observation level of structural response. High-order derivatives characterize the structural activity of variables and are highly sensitive to the mutation behavior in harness aging. Then, in order to measure the degree of coupling and difference between any two variables in the process of structural evolution, the covariant response tensor is introduced. , realizing the control strategy of graph construction with different derivatives, the mathematical model is: ; Where, is the weight coefficient corresponding to each order derivative, which is used to control the contribution of high-order structure tensor in similarity evaluation. is a nonlinear response control factor used to determine the sensitivity of the differential response to extreme changes. Then, the structure tensor is mapped to the graph structure space to further construct the connection strength matrix , a nonlinear shrinkage function is used to compress the difference and enhance stability. The mathematical model is: ; Where, is a stabilizing factor to prevent the denominator from being zero and to provide a lower bound when the difference is minimal. It is the global structural divergence reference scale, which is used to normalize all covariant response tensors to ensure that the overall tensor scale is relatively consistent. is the compression power parameter, which is used to control the nonlinear compression amplitude of the structural response difference. To enhance the power parameter, the compressed connection strength is further enhanced or amplified to further enhance the difference perception, and finally a stable, sparse and structurally interpretable connection strength matrix is obtained. , as the second data set.

[0013] Preferably, the graph connection strength tensor mapping module can comprehensively capture the structural trajectory differences and coupling characteristics between various variables in the aging process of household wiring harnesses by constructing a structural phase-driven graph connection strength matrix, accurately characterize the microscopic change trend of the wiring harness response using high-order derivative tensors, further quantify the nonlinear differences between variables through the covariance-divergence tensor, and combine the nonlinear compression mapping mechanism to generate a sparse, stable and structurally interpretable connection matrix. Compared with the traditional static graph structure, the graph connection strength tensor mapping module can dynamically reflect the evolution law of the internal structure of the wiring harness, provide a more refined and reliable graph structure foundation for subsequent spatiotemporal joint modeling, and significantly enhance the expressiveness and robustness of aging trend modeling.

[0014] Preferably, as household wire harnesses are affected by multiple factors during long-term use, their aging process usually manifests as complex spatiotemporal evolution and nonlinear changes. Traditional methods are difficult to effectively capture the key features of this dynamic evolution, especially the coordinated aging and spatial coupling relationship between variables. In order to overcome this limitation, the present invention introduces a spatiotemporal gravity extraction mechanism. Through the dynamic coupling of structural curvature and graph connection strength, the spatiotemporal change law of household wire harnesses in the aging process is accurately extracted, thereby providing more accurate and high-dimensional feature input for the aging resistance prediction model, ensuring higher sensitivity and prediction accuracy to nonlinear interference and mutation signals in the aging process.

[0015] Preferably, in order to further extract the curvature mutation points and structural disturbance impact paths of each variable structural response of household wiring harnesses during long-term aging, the present invention proposes a spatiotemporal gravity extraction module, which first calculates the curvature response tensor of the wiring harness channel variables in the time series space to construct their structural state changes. , the mathematical model is: ; Where, is the curvature response tensor of the structure, reflecting the local curvature of the variable structure curve of the harness channel. The larger the curvature, the greater the rate of change, reflecting the critical turning point of the structure in the aging process. Then the connection strength matrix The curvature response tensor with variables Combined with the above, we further define the structural tensor gravitational field response, and the mathematical model is: ; Where, is the bending convergence gravity that the variable receives in the structural graph space, indicating that it is guided and disturbed by the surrounding variables in the structural variation trend. Finally, the original variable value, the bending convergence gravity, and the curvature response tensor are combined to construct the final comprehensive gravitational field strength. The mathematical model is: ; Where, 、 and It is a fusion weight hyperparameter used to control the contribution ratio of the original variable value, curved convergence gravity and curvature response tensor to the overall gravitational response. In order to comprehensively analyze the gravitational field strength, it can fully integrate the current state of the bundle variables, the interference of the neighborhood structure, and the sudden change of the evolution trend. It has a strong physical explanation ability for the spatiotemporal reasoning of the aging state, and finally obtains the third data set. At the same time, the third data set is divided into the training set and the prediction set according to the ratio of 7:3.

[0016] Preferably, the design of the spatiotemporal gravity extraction module aims to dynamically capture the spatiotemporal variation characteristics of household wiring harnesses during operation, and accurately identify the key structural evolution points and disturbance sources in their aging process. The module integrates the local variation trend of wiring harness variables and the global spatial coupling effect by constructing a gravitational flow mechanism based on structural curvature response and graph connection strength, and can reveal the coordinated changes and dynamic correlations between different variables in the aging process. Compared with traditional methods, the spatiotemporal gravity extraction module can not only effectively reflect the complex nonlinear coupling relationship between variables, but also provide high-precision spatiotemporal feature representation through the interaction of structural curvature and graph gravity, providing richer and more stable input features for subsequent aging resistance prediction models, significantly improving the accuracy and reliability of predictions.

[0017] Preferably, during the long-term use of household wiring harnesses, the aging process of the wiring harnesses is affected by multiple factors, resulting in complex nonlinear aging characteristics. Traditional prediction methods cannot effectively capture such complex spatiotemporal changes and the coupling effects between multiple variables. In order to solve this problem, the present invention designs a structural response predictor module, which effectively captures the key change trends in the wiring harness aging process by dynamically modeling the spatiotemporal gravitational flow and structural state changes of the wiring harness, combined with high-order differentials, disturbance responses and nonlinear adjustment factors. The module can provide more accurate aging prediction results by adjusting the response to historical data in real time, ensuring accurate modeling of the wiring harness aging process under different working environments, providing more reliable technical support for life assessment and early warning of household wiring harnesses, and ultimately completing the prediction of aging resistance of household wiring harnesses.

[0018] Preferably, the structural response predictor module is implemented as follows: S51. Utilize the comprehensive gravitational field strength Calculate the time of home wiring harness Dynamic state change rate on , the mathematical model is: ; Where, Reflect the aging trend of the harness components and then construct an adaptive trend tensor , used to indicate the The aging trend of the wiring harness in the current state is guided by the mathematical model: ; Where, and Adjustable weights used to control the overall gravitational field strength and The degree of dominance in the forecast trend, It is the outer product operation of the matrix, which is used to interact the dynamic rate of change with the current value of the structural state to enhance the disturbance response; S52. Then a structural excitation function is proposed to adjust the nonlinear excitation term of the trend amplitude to ensure response enhancement in the rapid aging stage. The mathematical model is: ; Where, is the strength of the current structural trend tensor, In order to stimulate the sensitivity factor and ensure that the model does not react too much to small disturbances, the aging dynamic change rate of the home wiring harness component is calculated by combining the influence of the structural activation function and the adaptive trend tensor. The mathematical model is: ; Where, It is the inner product of the adaptive trend tensor and the comprehensive gravitational field strength, which is used to judge whether the current state of the variable is consistent with the evolution direction, and finally obtain the mathematical model for aging resistance prediction: ; Where, is the prediction step length, For the Components in The predicted aging degree at the moment, and the model is trained by defining the structural prediction error loss. The mathematical model is: ; Where, is the number of wiring harness components, is the time series length, is the real measurement value, and finally the trained model is obtained. By predicting the prediction set, the prediction result of the aging resistance of home appliance harness can be obtained. .

[0019] In summary, the present invention proposes a method for predicting the aging resistance of household wiring harnesses, which includes the following modules: a time series decomposition preprocessing module, a graph connection strength tensor mapping module, a spatiotemporal gravity extraction module, and a structural response predictor module. First, the time series decomposition preprocessing module extracts the multi-dimensional time series data of the household wiring harness, and performs trend, seasonality, and residual decomposition on it to provide refined input features for subsequent modules; then, the graph connection strength tensor mapping module constructs a high-order structural response tensor between variables, quantifies the coupling strength and mutual influence between components, and generates a dynamic tensor basis for mapping; then, the spatiotemporal gravity extraction ... The extraction module combines structural curvature and graph connection strength to accurately extract the spatiotemporal variation characteristics of household wire harnesses under different usage conditions, further enhancing the prediction model's perception of the aging process; finally, the structural response predictor module uses dynamic differential modeling, disturbance response and nonlinear adjustment factors to predict the future aging trend of household wire harnesses and output the final results. Compared with traditional methods, the present invention can integrate multi-source data, dynamically capture the aging process of household wire harnesses, and provide more accurate and reliable predictions, significantly improving the prediction accuracy, timeliness and robustness of household wire harnesses, and providing a scientific basis for the safety and long-term reliability of household appliances. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 This is a step-by-step diagram of a method for predicting the aging resistance of household wiring harnesses.

[0021] Figure 2 This is the structural diagram of the aging resistance prediction model for home appliance harnesses.

[0022] Figure 3 This is the structural diagram of the timing decomposition preprocessing module.

[0023] Figure 4 The structure diagram of the graph connection strength tensor mapping module.

[0024] Figure 5 This is a graph showing how the model effect changes with the number of training times.

[0025] Figure 6 The fitting effect diagram of the prediction model to realize the aging resistance prediction of home appliance harnesses. DETAILED DESCRIPTION

[0026] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0027] See Figure 1-Figure 5 The present invention provides a technical solution: a method for predicting the aging resistance of household wiring harnesses, comprising a time series decomposition preprocessing module, a graph connection strength tensor mapping module, a spatiotemporal gravity extraction module and a structural response predictor module. First, the time series decomposition preprocessing module extracts the multi-dimensional time series data of the household wiring harness, and performs trend, seasonality and residual decomposition on it to provide refined input features for subsequent modules; then, the graph connection strength tensor mapping module quantifies the coupling strength and mutual influence between components by constructing a high-order structural response tensor between variables, and generates a dynamic tensor basis for mapping; then, the spatiotemporal gravity extraction module combines structural curvature and graph connection strength to accurately extract the spatiotemporal variation characteristics of the household wiring harness under different usage conditions, further enhancing the prediction model's perception of the aging process; finally, the structural response predictor module predicts the future aging trend of the household wiring harness and outputs the final prediction value through dynamic differential modeling, using disturbance response and nonlinear adjustment factors. The specific steps are as follows: Figure 1 shown.

[0028] Construct a prediction model for the aging resistance of household wiring harnesses, the structure of which is as follows: Figure 2 As shown, the specific steps are:

[0029] S1. Collect multi-source monitoring data of household wiring harnesses, including electrical performance data, temperature, humidity, and mechanical stress characteristics, and construct an original multidimensional dataset.

[0030] Furthermore, the dataset of the present invention includes 1,000 pieces of household wiring harness related data, and the dataset is divided into a training set and a test set in a ratio of 7:3.

[0031] S2, propose a time series decomposition preprocessing module, use high-order time derivatives to build an adaptive disturbance feedback mechanism, and couple the interaction matrix Dynamically construct the nonlinear synergistic relationship between multiple variables and dynamically adjust the interaction intensity between variables. Finally, introduce the disturbance perception factor to dynamically weight and fuse the trend structure and nonlinear information to obtain the first data set. .

[0032] Furthermore, the structure of the time series decomposition preprocessing module is as follows Figure 3As shown, first, the present invention constructs an adaptive disturbance feedback mechanism to dynamically capture the evolution of local disturbances in the wiring harness during aging and reveal the nonlinear driving law of the home wiring harness. The mathematical model is: ; Where, For home appliance harnesses at all times Status data, indicating the multi-dimensional changes in the electrical performance, temperature, humidity, and mechanical stress of the wiring harness components. It is the high-order derivative of the input state data in the time dimension, which is used to extract the multi-order dynamic response characteristics of the wiring harness aging process. For the The perturbation response weight corresponding to the derivative term is initially set to 0.8, and then randomly takes values between (0, 1) to dynamically adjust the model response at each time point. It is a local adaptive adjustment factor that can dynamically adjust the sensitivity of the model according to the changes in the current data state to ensure that the model can respond promptly to rapidly changing aging events. The value is set to 0.5. To perturb the nonlinear adjustment index, adjust the response intensity of the local state value, which can enhance the prediction accuracy of the model in the face of sudden and nonlinear aging phenomena. The value is set to 0.3. The maximum order of high-order derivatives considered in disturbance modeling is set to 3 to ensure that high-order derivatives can adequately model complex aging patterns.

[0033] Furthermore, the dynamic interaction matrix is constructed to describe the mutual influence between the variables of the home appliance harness, thereby achieving a more accurate modeling of the aging process, extracting the high-order dynamic response characteristics of each variable, and combining the interaction influence to construct a coupling interaction matrix. , the mathematical model is: ; Where, Describes the variable and At the moment The perturbation coupling strength can describe their mutual influence in the aging process. is the static coupling strength weight of each pair of variables, which controls the nonlinear coupling of disturbances between variables and is set to 0.6. is the attenuation factor, which is used to control the impact of historical disturbances on the current moment. The value is set to 0.4. The length of the historical time window is 20, and then the time series data Decomposition into trend components , seasonal part and the residual , in order to more accurately extract the evolution components and abnormal disturbances in the aging process, the mathematical model is: ; ; ; Where, is the trend component of time series data, is the total number of variables, the value is 8000, is the local adaptation factor, which controls the degree of response to local state changes at each time point, and its value is set to 0.5. It is an adaptive exponential adjustment factor, with an initial value of 0.9 and subsequent random values between (0, 1.5). is the seasonal component, which is used to extract the synergistic trend of long-term periodic structural changes among bundle variables. is the residual component, which is used to capture sudden failures during the aging process of household wiring harnesses. According to the extracted trend component, seasonal component and residual component, combined with the nonlinear adjustment factor, the final data reconstruction is completed. The mathematical model is: ; Where, is the perturbation perception factor, which is used to reflect the weighted influence of different components at different time points. The value is set to 0.5, and the adaptive fusion reconstruction output is finally obtained. as the first dataset.

[0034] S3, design graph connection strength tensor mapping module, build high-order differential tensor representation mechanism based on structural trajectory changes, introduce covariant response tensor , and then map the structure tensor to the graph structure space to obtain the connection strength matrix , as the second data set.

[0035] Furthermore, the graph connection strength tensor mapping module structure is as follows Figure 4 As shown, we further construct a high-order differential tensor representation mechanism driven by structural trajectory changes, for each bundle channel variable at time point The structural trajectory tensor is expressed by high-order derivatives, and the mathematical model is: ; Where, For variables At the moment No. The order structural trajectory tensor is used to measure the dynamic response strength of the bundle variables at different microstructural levels. is the differential order of modeling, which is used to represent the observation level of structural response. The value is set to 3. Then, in order to measure the degree of coupling and difference between any two variables in the process of structural evolution, the covariant response tensor is introduced. , realizing the control strategy of graph construction with different derivatives, the mathematical model is: ; Where, is the weight coefficient corresponding to each order derivative, which is used to control the contribution of high-order structure tensor in similarity evaluation, and the value is set to 0.7. is the nonlinear response control factor used to determine the sensitivity of the differential response to extreme changes and is set to 2.

[0036] Furthermore, we construct a connection strength matrix , and enhance the focus on important interactions through nonlinear shrinkage mapping. The mathematical model is: ; Where, It is a stabilizing factor, which is used to prevent the denominator from being zero and to provide a lower bound when the difference is minimal. It is set to 1. It is the global structural divergence reference scale, which is used to normalize all covariant response tensors to ensure that the overall tensor scale is relatively consistent. The value is set to 0.5. is the compression power parameter, which is used to control the nonlinear compression amplitude of the structural response difference. Its value is set to 2. To enhance the power parameter and further enhance or amplify the difference perception of the compressed connection strength, the value is set to 2 / 3, and finally a stable, sparse and structurally interpretable connection strength matrix is obtained. , as the second data set.

[0037] S4. Construct a spatiotemporal gravity extraction module and combine it with the connection strength matrix and the curvature response tensor Define the structural tensor gravitational field response and finally get the comprehensive gravitational field strength As the third data set, the third data set is divided into a training set and a prediction set.

[0038] Furthermore, first define the time of each component of the home appliance harness The curvature response tensor of , the mathematical model is: ; Where, is the curvature response tensor of the structure, reflecting the local curvature of the variable structure curve of the harness channel. The larger the curvature, the greater the rate of change, reflecting the critical turning point of the structure in the aging process. is the second-order derivative of the component, which represents the aging rate of the component, is the first-order derivative of the component, which represents the rate of change of the component response, and then the connection strength matrix The curvature response tensor with variables Combined with the above, we further define the structural tensor gravitational field response, and the mathematical model is: ; Where, is the bending convergence gravity that the variable receives in the structural graph space, indicating that it is guided and disturbed by the surrounding variables in the structural variation trend. Finally, the original variable value, the bending convergence gravity, and the curvature response tensor are combined to construct the final comprehensive gravitational field strength. The mathematical model is: ; Where, 、 and The fusion weight hyperparameter is used to control the contribution ratio of the original variable value, curved convergence gravity and curvature response tensor to the overall gravitational response. The values are set to 0.3, 0.4 and 0.3 respectively. In order to comprehensively analyze the gravitational field strength, the third data set is finally obtained, and the third data set is divided into a training set and a prediction set according to a ratio of 7:3.

[0039] S5. Construct a structural response predictor module and use the comprehensive gravitational field strength Calculate the dynamic state change rate of household wiring harnesses , construct an adaptive trend tensor , a structural excitation function is proposed and the aging dynamic change rate of household wiring harness components is calculated, and finally an aging resistance prediction model is obtained.

[0040] Furthermore, using the comprehensive gravitational field strength Calculate the time of home wiring harness Dynamic state change rate on , the mathematical model is: ; Where, Reflect the aging trend of the harness components and then construct an adaptive trend tensor , used to indicate the The aging trend of the wiring harness in the current state is guided by the mathematical model: ; Where, and Adjustable weights used to control the overall gravitational field strength and The dominance in the forecast trend is set to 0.8 and 0.5 respectively. It is the outer product operation of the matrix, which is used to interact the dynamic rate of change with the current value of the structural state to enhance the disturbance response.

[0041] Furthermore, a structural excitation function is proposed to adjust the nonlinear excitation term of the trend amplitude to ensure response enhancement in the rapid aging stage. The mathematical model is: ; Where, is the strength of the current structural trend tensor, To stimulate the sensitivity factor, the value is set to 0.1 to ensure that the model does not react too much to small disturbances. At the same time, the influence of the structural activation function and the adaptive trend tensor are combined to calculate the aging dynamic change rate of the household wiring harness component. The mathematical model is: ; Where, It is the inner product of the adaptive trend tensor and the comprehensive gravitational field strength, which is used to judge whether the current state of the variable is consistent with the evolution direction, and finally obtain the mathematical model for aging resistance prediction: ; Where, For the prediction step, the value is set to 1.5. For the Components in Predicted aging level at the moment.

[0042] S6. The training set is input into the prediction module, and the structure prediction error loss is defined. Continuously optimize and train the aging resistance prediction model for home appliance harnesses.

[0043] Furthermore, the training set is input into the home appliance harness aging resistance prediction model. The model uses the PyTorch deep learning framework and is run in a Linux operating system environment. It is accelerated by using an NVIDIA V100 32GB GPU. During the training process, the batch size is set to 128. During the training phase, the model is trained by defining a structural prediction error loss. The mathematical model is: ; Where, is the real measurement value, and the model effect changes with the number of training times as shown in the figure Figure 5 As shown in the figure, it can be seen that as the model is continuously trained, the difference between the predicted value and the actual value becomes smaller and smaller, and finally a well-trained model is obtained.

[0044] S7: The prediction set is input into the trained home wiring harness aging resistance prediction model, and the predicted value of the wiring harness aging resistance is finally output. .

[0045] Furthermore, the home appliance harness aging resistance prediction model realizes the harness aging resistance prediction fitting effect diagram as shown in the figure below. Figure 6 As shown in the figure, the horizontal axis is the date, the vertical axis is the aging resistance value, the dotted line and the cross represent the predicted value, and the solid line and the dot represent the actual value. It can be seen from the figure that the changing trends of the actual value and the predicted value are roughly similar, and the changing directions of the two are consistent. The experimental results show that the aging resistance prediction model of household wire harnesses can effectively capture the changing trends of household wire harnesses, especially in time periods with large fluctuations, and can better predict the aging resistance of wire harnesses.

Claims

1. A method for predicting the aging resistance of household wiring harnesses, characterized in that: The following steps are involved: S1. Collect multi-source monitoring data of household wiring harnesses, including electrical performance data, temperature, humidity, and mechanical stress characteristics, and construct an original multidimensional dataset; S2, propose a time series decomposition preprocessing module, use high-order time derivatives to build an adaptive disturbance feedback mechanism, and couple the interaction matrix Dynamically construct the nonlinear synergistic relationship between multiple variables and dynamically adjust the interaction intensity between variables. Finally, introduce the disturbance perception factor to dynamically weight and fuse the trend structure and nonlinear information to obtain the first data set. ; S3, design graph connection strength tensor mapping module, build high-order differential tensor representation mechanism based on structural trajectory changes, introduce covariant response tensor , and then map the structure tensor to the graph structure space to obtain the connection strength matrix , as the second data set; S4. Construct a spatiotemporal gravity extraction module and combine it with the connection strength matrix and the curvature response tensor Define the structural tensor gravitational field response and finally get the comprehensive gravitational field strength As a third data set, the third data set is divided into a training set and a prediction set; S5. Construct a structural response predictor module and use the comprehensive gravitational field strength Calculate the dynamic state change rate of household wiring harnesses , construct an adaptive trend tensor , proposed the structural excitation function and calculated the aging dynamic change rate of household wiring harness components, and finally obtained the aging resistance prediction model; S6. The training set is input into the prediction module, and the structure prediction error loss is defined. Continuously optimize and train the prediction model for the aging resistance of household wiring harnesses; S7: The prediction set is input into the trained home wiring harness aging resistance prediction model, and the predicted value of the wiring harness aging resistance is finally output. .

2. The method for predicting the aging resistance of household wiring harnesses according to claim 1, characterized in that: The adaptive disturbance feedback mechanism in S2 dynamically captures data The mathematical model of the local perturbation evolution process is: ； Where, is the high-order derivative of the input state data in the time dimension, For the The disturbance response weight corresponding to the order derivative term, is the local adaptive adjustment factor, is the disturbance nonlinear adjustment index, The maximum order of the higher-order derivatives considered in the perturbation modeling is then expressed by the coupling interaction matrix Modeling cross-variable disturbance dependence in multivariate data, dynamically adjusting the interaction strength between different variables, the mathematical model is: ； Where, For variables and At the moment The perturbation coupling strength, is the static coupling strength weight of each pair of variables, is the attenuation factor, is the length of the historical time window, and then the time series data Decomposition into trend components , seasonal part and the residual Finally, the disturbance perception factor is introduced Dynamically weighted fusion of trend structure and nonlinear disturbance information completes data reconstruction and ultimately obtains adaptive fusion reconstruction output as the first dataset.

3. The method for predicting the aging resistance of household wiring harnesses according to claim 2, characterized in that: For each harness channel variable in the first data set , defined at a point in time The structural trajectory tensor , and then introduce the covariant response tensor , measures the degree of coupling and difference between any two variables in the process of structural evolution. The mathematical model is: ； Where, is the weight coefficient corresponding to each order derivative, is a nonlinear response control factor. Finally, the structure tensor is mapped to the graph structure space, and the nonlinear shrinkage function is used to compress the difference to construct the connection strength matrix. Based on the stabilizing factor and global structural divergence reference scale The denominator is normalized, and then the power compression function is applied to the normalized result. The final overall result is nonlinearly enhanced by the external power function to output the connection strength matrix , as the second data set.

4. The method for predicting the aging resistance of household wiring harnesses according to claim 3, characterized in that: A spatiotemporal gravity extraction module is proposed. First, the curvature response tensor of the bundle channel variable is calculated to construct its structural state change in the time series space. , the mathematical model is: ； Where, is the curvature response tensor value of the structure, and then the connection strength matrix The curvature response tensor with variables Combined, further define the structure tensor gravitational field response , and finally introduce the fusion weight hyperparameter 、 and The original variable values, curved convergence gravity and curvature response tensors are combined to construct the final gravitational flow response term as the third data set. At the same time, the third data set is divided into training set and prediction set according to 7:

3.

5. The method for predicting the aging resistance of household wiring harnesses according to claim 4, characterized in that: Using comprehensive gravitational field strength Calculate the time of home wiring harness Dynamic state change rate on , then construct the adaptive trend tensor , used to indicate the The aging trend of the wiring harness in the current state is guided by the mathematical model: ； Where, and The weight is adjustable.

6. The method for predicting the aging resistance of household wiring harnesses according to claim 5, characterized in that: A structural excitation function is proposed to obtain enhanced response during the rapid aging stage of home wiring harnesses. The mathematical model is: ； Where, is the strength of the current structural trend tensor, In order to stimulate the sensitivity factor, the influence of the structural activation function and the adaptive trend tensor are combined, and the aging dynamic change rate of the household wiring harness component is calculated by using the inner product of the adaptive trend tensor and the gravitational field structure harmonic representation vector. , and finally the mathematical model for aging resistance prediction is obtained: ； Where, is the prediction step length, For the Components at time The predicted aging degree, while the model is defined by the structural prediction error loss After training, a trained model is finally obtained. The prediction results of the aging resistance of home appliance harnesses can be obtained by inputting the prediction set. .

Images