An adaptive optimization system based on exhaust hook load force and a control method thereof

Abstract

The application discloses a kind of adaptive optimization systems based on exhaust hanger load force and control method thereof, it is related to the field of automotive engineering and intelligent control technology, comprising the following steps: exhaust hanger load force signal and vibration acceleration signal are obtained by sensor arranged at exhaust hanger position, and continuous time series data containing time series features are constructed;Continuous time series data are input into data processing model for feature extraction and dynamic state modeling, and the output results of data processing model are processed using Kalman filtering to perform prediction correction, to form stable dynamic stiffness prediction data.The application realizes the dynamic matching of exhaust lug dynamic stiffness in different frequency bands by real-time acquisition of load and vibration information, and the electromagnetic control auxiliary hanger participates in structural constraint, replacing the traditional thickening reinforcement method, while ensuring the NVH performance, realizing lightweight design, and effectively improving the dynamic stiffness adjustment capability and vehicle vibration control effect under each working condition.

Description

Technical Field

[0001] This invention relates to the fields of automotive engineering and intelligent control technology, specifically to an adaptive optimization system and control method based on the load force of an exhaust hook. Background Technology

[0002] In recent years, with the rapid development of new energy vehicles and traditional fuel vehicles, the NVH (noise, vibration, and harshness) performance of the entire vehicle has gradually become an important indicator affecting user experience. As one of the important excitation sources of vehicle vibration, the exhaust system transmits vibrations to the vehicle body structure through the exhaust hook, directly affecting the noise and vibration levels inside the vehicle. The dynamic stiffness of the exhaust hook mounting point is a key parameter determining the vibration transmission characteristics. In practical engineering, due to the complex and varied operating conditions of vehicles, the excitation frequency of the exhaust system exhibits a wide frequency distribution. Therefore, the design process usually needs to consider the dynamic stiffness matching requirements under different operating conditions. Currently, the industry commonly adopts methods such as strengthening the mounting bracket, increasing the thickness of the base plate, or reinforcing the hook structure to improve the overall stiffness of the exhaust hook to meet the performance requirements under the most unfavorable operating conditions.

[0003] However, existing technologies generally employ fixed structural designs, meaning their dynamic stiffness is not adjustable after design completion. This makes it difficult to adapt to dynamic changes in load and excitation frequency during actual operation, leading to issues such as stiffness redundancy and increased vehicle weight under non-extreme conditions. Furthermore, existing technologies lack the ability to perceive and analyze the load state and vibration characteristics of the exhaust hook in real time, making online prediction and dynamic matching of dynamic stiffness impossible, which can easily cause NVH problems such as abnormal noises and resonance. In addition, even when stiffness is improved through local structural reinforcement, the main hook body often suffers from insufficient stiffness, resulting in limited overall optimization effects.

[0004] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to provide an adaptive optimization system and control method based on the load force of the exhaust hook, so as to solve the problems in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an adaptive optimization control method based on the load force of an exhaust hook, comprising the following steps: By acquiring load force and vibration acceleration signals of the exhaust hook using sensors positioned at the exhaust hook location, continuous time-series data containing time-series characteristics is constructed. Continuous time-series data is input into the data processing model for feature extraction and dynamic state modeling. Kalman filtering is then used to perform prediction correction on the output of the data processing model to form stable dynamic stiffness prediction data. Perform frequency domain analysis and numerical extraction processing on the dynamic stiffness prediction data, and output the dynamic stiffness prediction values ​​and corresponding frequency range information; Read the predicted dynamic stiffness value and the corresponding frequency range information, execute the structural reconfiguration instruction generation rule calculation, and output the structural reconfiguration instruction that matches the current working condition; It receives structural reconfiguration commands and drives the electromagnetic locking device to perform corresponding locking actions, thereby achieving adaptive adjustment of dynamic stiffness in different frequency ranges by changing the structural constraint state of the exhaust hook.

[0007] Preferably, a continuous input data structure is constructed to support subsequent analysis and processing, based on a unified expression process of load signals and vibration information. The steps are as follows: The load force signal and vibration acceleration signal of the exhaust hook are collected, the collected signals are timestamped, and multi-channel synchronous calibration is performed to form raw measurement data; The load force signal and vibration acceleration signal are dimensionally aligned, outlier data are removed, and vectors are concatenated in a unified format to construct a single-time input vector. The input vectors at each time step are arranged continuously according to the sampling order, and missing data are interpolated to complete the sequence, thus generating a continuous input sequence. The continuous input sequence is divided into sliding windows, and normalization and amplitude constraint processing are performed on the data in each window to output a standardized time series data subsequence.

[0008] Preferably, by combining time-series signal processing, feature mapping and information extraction are performed on the original data to form a feature sequence that can be used for modeling. The steps are as follows: The continuous input sequence is segmented, and the convolution window range and stride parameters are set to generate local time window data. Perform one-dimensional convolution operations within a local time window, perform sliding scan calculations on the input signal, and extract local response features; The output of the convolution operation is processed by applying a non-linear activation function, and the feature values ​​are clipped. The feature results corresponding to each time window are concatenated in chronological order, and the concatenated results are subjected to dimensional reconstruction processing to form a continuous feature sequence.

[0009] Preferably, by introducing a memory unit structure, continuous feature data is correlated over time to complete the expression and transmission of dynamic information. The steps are as follows: Input the continuous feature sequence into the temporal network structure to initialize the hidden state variables and memory state variables; The input gate is used to calculate the weight allocation of the feature data at the current time and update the proportion of input information. By using a forget gate to filter historical state data, the degree of involvement of historical information can be controlled. The hidden state data is generated by combining the current memory state with the output gate, and the hidden states at each time point are output in chronological order to form a time-series prediction sequence.

[0010] Preferably, the time series information is centrally processed and transformed into corresponding physical quantity expressions around the output path of the prediction results, and the steps are as follows: Perform time-dimensional statistical processing on the time-series prediction sequence, including maximum value extraction and mean calculation; The statistical processing results are input into a fully connected mapping structure to prepare the input features by linear transformation. Matrix multiplication is performed between the weight vector and the input features, and bias terms are added to form intermediate calculation results; The intermediate calculation results are processed to generate the corresponding predicted dynamic stiffness values.

[0011] Preferably, to address fluctuations in the prediction results, a filtering process is introduced to correct and optimize the data. The steps are as follows: The predicted dynamic stiffness values ​​are input into the filtering unit to construct the current observation data sequence; The current predicted value is obtained by recursively calculating based on the previous dynamic stiffness estimate and state transition relationship; The difference between the predicted estimate and the observed data is calculated, and the error weighting factor is determined based on the difference. The error weighting factor is applied to the predicted estimate to update the current dynamic stiffness and output the correction result.

[0012] Preferably, the calculation process of the error weighting factor is refined. The error weighting factor is determined by the deviation between the predicted estimate and the observed data and the prediction error covariance, and is dynamically adjusted in combination with the observation noise covariance. During the update calculation process, the predicted estimate is corrected by the weight allocation method, thereby outputting the continuous dynamic stiffness estimation result.

[0013] Preferably, the periodic information is extracted and converted into frequency parameters based on the characteristics of vibration signal changes to describe the current vibration state. The steps are as follows: The vibration displacement signal sequence is acquired, and the signal is subjected to discrete sampling processing to form a discrete sequence; Sign change detection is performed on discrete signals to locate the zero-crossing point where the signal changes from a negative value to a positive value; The time difference between adjacent zero crossings is calculated, and statistical processing is performed on data from multiple periods. Perform a reciprocal operation based on the period statistics results, and output the current vibration frequency value.

[0014] Preferably, based on the dynamic stiffness adjustment process, corresponding control commands are generated through frequency range division and stiffness numerical relationship analysis, as follows: Receive dynamic stiffness prediction values ​​and frequency information, and perform data format parsing and unit unification processing; The frequency information is divided into intervals and the frequency is mapped to the corresponding frequency range. The predicted dynamic stiffness value is compared with the target stiffness threshold within an interval, and the current state type is determined. Based on the comparison results, the corresponding auxiliary hook locking control command is output and converted into an execution signal.

[0015] An adaptive optimization system based on exhaust hook load force includes a data acquisition module, a data processing module, a parameter calculation module, an instruction generation module, and an execution control module. The data acquisition module acquires the load force signal and vibration acceleration signal of the exhaust hook through sensors arranged at the position of the exhaust hook, and constructs continuous time series data containing time series characteristics; The data processing module inputs continuous time-series data into the data processing model for feature extraction and dynamic state modeling, and uses Kalman filtering to perform prediction correction processing on the output results of the data processing model to form stable dynamic stiffness prediction data. The parameter calculation module performs frequency domain analysis and numerical extraction processing on the dynamic stiffness prediction data, and outputs the dynamic stiffness prediction values ​​and corresponding frequency range information. The instruction generation module reads the predicted dynamic stiffness value and the corresponding frequency range information, executes the structural reconfiguration instruction generation rule calculation, and outputs a structural reconfiguration instruction that matches the current working condition. The execution control module receives structural reconfiguration commands and drives the electromagnetic locking device to perform corresponding locking actions, thereby achieving adaptive adjustment of dynamic stiffness in different frequency ranges by changing the structural constraint state of the exhaust hook.

[0016] The technical effects and advantages provided by the present invention in the above technical solution are as follows: This invention introduces a real-time load force detection and dynamic adjustment mechanism for exhaust hooks, enabling real-time matching of dynamic stiffness in exhaust lugs across different vibration frequency ranges. Compared to traditional designs relying on fixed structural reinforcement, this technology dynamically adjusts the structural constraint state based on load changes and excitation frequency characteristics during actual operation, ensuring that the dynamic stiffness remains within the appropriate range and effectively avoiding insufficient or excessive stiffness. Furthermore, without requiring significant reinforcement of the main structure, the synergistic effect of the auxiliary hook and electromagnetic control mechanism achieves on-demand improvement in structural performance. This significantly reduces weight increase due to structural redundancy while ensuring optimized NVH performance, demonstrating excellent lightweight advantages.

[0017] This invention constructs a dynamic stiffness adjustment system driven by real-time data, transforming the exhaust hook from a traditional passive response structure into an active adjustment structure. This allows for segmented optimized control of vibration characteristics under different operating conditions. In practical applications, by combining load force signals and vibration frequency information, the locking state of the auxiliary hook is precisely controlled, ensuring that different frequency ranges correspond to different structural stiffness levels, thereby achieving optimal performance matching under multiple operating conditions. This technology not only improves the dynamic stiffness performance of the exhaust system mounting point across various frequency bands but also effectively improves the vehicle's vibration transmission characteristics, reduces the risk of abnormal noises and resonance, and significantly enhances the vehicle's NVH performance and user comfort. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0019] Figure 1 This is a top view of the present invention in the state without locking and without cover plate.

[0020] Figure 2 This is a three-dimensional schematic diagram of the main hook and auxiliary hook one of the present invention.

[0021] Figure 3 This is a cross-sectional view of the locking mechanism structure of the present invention.

[0022] 1. Bracket base; 2. Base mounting bolts; 3. 90-degree auxiliary lifting lug; 4. 45-degree auxiliary lifting lug; 5. 180-degree auxiliary lifting lug; 6. Main hook; 7. Right cover plate; 8. Left cover plate; 9. Locking mechanism one; 10. Locking mechanism two; 11. Locking mechanism three; 12. Pin one; 13. Lever; 14. Pin two; 15. Upper lock body; 16. Lock body mounting bolts; 17. Lower right lock body; 18. Auxiliary hook; 19. Wedge-shaped lock tongue; 20. Solenoid valve; 21. Lower left lock body; 22. Solenoid valve core.

[0023] Figure 4 This is a flowchart of the present invention.

[0024] Figure 5 The flowchart illustrates the process of processing the acquired signal data and performing prediction using Kalman filtering, as described in this invention.

[0025] Figure 6 This represents the dynamic stiffness curve state of the present invention. Detailed Implementation

[0026] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0027] This invention provides, for example Figures 1-6 An adaptive optimization control method based on exhaust hook load force is shown, comprising the following steps: A high-precision triaxial force sensor is integrated into the exhaust hook structure to monitor the load status of the exhaust system in real time. This triaxial force sensor is integrated at the connection point between the exhaust hook and the vehicle body longitudinal beam, as well as at the exhaust pipe lifting lug connection end, enabling direct monitoring of the force path of the exhaust hook through an integrated structural arrangement. The triaxial force sensor can simultaneously acquire load force signals in the X, Y, and Z orthogonal directions, thus constructing complete spatial force information and achieving accurate characterization of the force changes of the exhaust system under different operating conditions. The sampling frequency is set to no less than 1kHz, meeting the high-frequency load change capture requirements of the exhaust system under dynamic conditions such as engine idling, constant speed operation, rapid acceleration, and bumpy road surfaces. Simultaneously, the measurement accuracy is controlled within ±0.5% of the full scale, effectively ensuring the accuracy and stability of the collected data, providing a reliable data foundation for subsequent dynamic stiffness prediction and control. Through this high-frequency, high-precision load data acquisition method, continuous tracking of the dynamic force characteristics of the exhaust hook in complex vibration environments can be achieved.

[0028] In terms of vibration characteristic acquisition, acceleration sensors are placed at the mounting points of the exhaust hook and the vehicle body to synchronously collect the structural vibration response. The acceleration sensors acquire vibration acceleration signals in the connection area between the exhaust hook and the vehicle body during actual operation, reflecting the vibration amplitude and trends of the structure under different excitation conditions. By synchronously processing the acceleration signals with the load force signals acquired by the triaxial force sensors, a complete force-vibration coupled data sequence can be constructed, thus more comprehensively reflecting the dynamic behavior characteristics of the exhaust hook system under actual operating conditions. The introduction of acceleration signals not only improves the ability to identify vibration states but also provides key input data for subsequent frequency analysis and dynamic stiffness calculation, enabling a more accurate description of the dynamic response characteristics of the exhaust hook structure under different operating conditions. Through the coordinated acquisition of load force and vibration signals, effective acquisition of multi-dimensional dynamic information under complex operating conditions of the exhaust system can be achieved, providing data support for the formulation of subsequent control strategies.

[0029] In the dynamic characteristic analysis of the exhaust hook, real-time acquired vibration and load signals are used as the input basis. A multi-stage data processing and prediction mechanism is used to achieve accurate estimation of dynamic stiffness and frequency identification. The overall process starts from the acquisition of raw data, and gradually outputs stable dynamic stiffness results through feature extraction, time series modeling, numerical prediction and filtering correction. It forms a continuous loop processing structure to adapt to changes in dynamic working conditions.

[0030] The input signal is defined as follows: Input = Real-time collected acceleration + Load force; in, Indicates time The input vector, composed of vibration acceleration and triaxial load force signals, describes the dynamic response state of the exhaust hook at the current moment. This input signal has multidimensional characteristics and can simultaneously reflect the coupling relationship between structural forces and vibrations.

[0031] The target for dynamic stiffness prediction is defined as: Dynamic stiffness (the quantity we need to predict in real time); in, Indicates time The corresponding target dynamic stiffness physical quantity is an important basis for subsequent control and structural adjustment.

[0032] The initial output of the model is expressed as follows: The initial predicted value output by the model; in, This represents the intermediate prediction result after processing by the neural network model, reflecting the fitting output of the current input data in the model.

[0033] The final corrected result is expressed as follows: : The final stiffness after correction from the KF output; in, This indicates that the final dynamic stiffness prediction value after Kalman filtering has higher stability and accuracy compared to the initial prediction value.

[0034] In the feature extraction stage, a convolutional neural network is used to perform noise reduction and feature extraction on the input signal, and its expression is as follows: in, Indicates at time The extracted feature vectors; This represents a convolutional neural network structure used for filtering, feature enhancement, and noise reduction of input signals. This represents the window for inputting time series data.

[0035] This processing method can extract more stable and representative feature information from the original collected data, reducing the impact of noise interference on subsequent predictions.

[0036] In the temporal modeling stage, a long short-term memory network is used to dynamically model and predict the feature sequence, and its expression is as follows: in, This represents the structure of a Long Short-Term Memory (LSTM) network, used for processing time-series data. Indicates the feature input at the current time; This indicates the predicted output at a historical moment. This represents the predicted output value at the current moment.

[0037] By incorporating historical state information, the time-dependent characteristics of the exhaust system during dynamic changes can be effectively captured, improving the responsiveness of prediction results to continuous operating condition changes.

[0038] In the initial dynamic stiffness prediction stage, the output of the time series model is mapped through a fully connected layer, and its expression is as follows: in, This represents the preliminary predicted value of dynamic stiffness at the current moment; This represents the temporal prediction results of the input fully connected layer.

[0039] This process maps the feature space to the physical quantity space, converting the model output into dynamic stiffness parameters that have engineering significance.

[0040] In the prediction correction stage, Kalman filtering is introduced to dynamically correct and optimize the preliminary prediction results. The calculation process includes two stages: prediction and update.

[0041] The prediction process is as follows: in, This represents the estimated dynamic stiffness during the prediction phase. This represents the state transition matrix, used to describe a system model where dynamic stiffness changes over time. This represents the predicted value at the previous moment.

[0042] The Kalman gain is calculated as follows: in, Kalman gain is used to measure the weighting relationship between predicted and observed values.

[0043] The update process is as follows: in, This represents the updated final predicted dynamic stiffness value; This indicates a preliminary forecast value; Indicates prediction error; Used to perform weighted correction for errors.

[0044] By using Kalman filtering, the model prediction results and system dynamic characteristics can be integrated to achieve the optimal estimation of dynamic stiffness, thereby improving prediction accuracy and anti-interference capability.

[0045] In terms of frequency calculation, the system vibration frequency is calculated in real time using the zero-crossing detection method, the principle of which is as follows: By monitoring displacement signals The time interval between two adjacent zero-crossing points, from the position where the value changes from negative to positive, is recorded as the oscillation period: Further frequency calculations: in, Indicates the current vibration frequency; Indicates the period of vibration; This represents a displacement signal in discrete time.

[0046] This method is characterized by its simple calculation and strong real-time performance, and can quickly respond to changes in vibration frequency, providing key parameter support for segmented dynamic stiffness control.

[0047] Through the above processing flow, a complete closed-loop process from multi-source sensor data input to dynamic stiffness prediction output is achieved. Combined with frequency identification results, this provides a precise data foundation for subsequent structural adjustment and control strategies. During continuous operation, this process can be executed cyclically to continuously track the dynamic response changes of the exhaust hook under different operating conditions, achieving a balance between real-time performance and stability in dynamic stiffness prediction.

[0048] After the dynamic stiffness prediction results are output, the joint identification stage of dynamic stiffness and frequency begins. By comprehensively analyzing the current dynamic stiffness prediction value and the corresponding vibration frequency, the dynamic judgment of the exhaust hook's operating state is achieved. Frequency parameters, as key indicators of vibration characteristics, reflect the excitation characteristics of the exhaust system under different operating conditions. By dividing the frequency into different intervals, the complex continuous spectrum can be discretized into multiple representative operating intervals, thus facilitating the formulation and execution of control strategies. In specific implementation, the vibration frequency is divided into three typical intervals: low frequency (20-50Hz), mid frequency (50-120Hz), and high frequency (120-250Hz). Each frequency interval corresponds to a different dynamic stiffness target range to achieve structural matching under different vibration characteristic conditions.

[0049] In the low-frequency range, when the vibration frequency is between 20-50Hz, the exhaust system vibration mainly manifests as low-frequency oscillations with relatively large amplitudes. If the estimated dynamic stiffness is less than 600N / mm, it indicates that the current structural stiffness is insufficient to suppress the propagation of low-frequency vibrations, and the stiffness level needs to be improved through structural reinforcement. The control logic generates corresponding instructions based on the estimated dynamic stiffness, triggering the locking action of the lower end of auxiliary lifting lug 1 (45 degrees). As auxiliary lifting lug 1 participates in structural constraint, the overall force path of the exhaust hook changes, and the structural stiffness increases accordingly, adjusting the dynamic stiffness to the range of 600N / m to 1000N / m, thereby effectively suppressing low-frequency vibrations. Within this range, locking the 45-degree auxiliary lifting lug forms a relatively flexible and enhanced structural support, ensuring both increased stiffness and avoiding vibration transmission amplification caused by excessive stiffness.

[0050] Within the mid-frequency range, when the vibration frequency is between 50-120Hz, the exhaust system vibration exhibits mid-frequency characteristics, with a relatively concentrated vibration energy distribution, thus requiring higher structural stiffness. When the estimated dynamic stiffness is below 1000 N / mm, it indicates that the current structure still suffers from insufficient stiffness in this frequency range, necessitating further enhancement of structural constraint capabilities. The control strategy generates corresponding structural adjustment commands, triggering the locking action of the lower end of auxiliary lifting lug 2 (90 degrees). After locking, auxiliary lifting lug 2 forms a more rigid connection with the main hook, further improving the structural constraint stiffness and raising the dynamic stiffness to the 1000 N / m to 1300 N / m range, thereby matching the mid-frequency vibration characteristics. In this state, the structure exhibits stronger resistance to deformation, which is beneficial for reducing the transmission of mid-frequency vibrations.

[0051] Within the high-frequency range, when the vibration frequency is in the 120-250Hz interval, the exhaust system vibration is mainly high-frequency, small-amplitude vibration, requiring the highest structural stiffness. When the estimated dynamic stiffness is below 1300 N / mm, it indicates that the current structural stiffness cannot effectively suppress high-frequency vibration, and maximum structural stiffness enhancement is needed. The control logic generates commands corresponding to the high-frequency operating condition, triggering the locking action of the lower end of the auxiliary hanger (180 degrees). After the auxiliary hanger is locked at 180 degrees, the strongest constraint path is formed, increasing the overall dynamic stiffness to the range of 1300 N / m to 2000 N / m, thereby effectively suppressing high-frequency vibration. This structural state corresponds to the highest stiffness level, which can significantly reduce the amplitude of high-frequency vibration transmitted to the vehicle body.

[0052] Through the above segmented control logic, the mapping relationship between frequency range and dynamic stiffness target can be realized, transforming the dynamic stiffness adjustment process from traditional static design to dynamic adjustment process, enabling the structural stiffness to be matched in real time with the vibration frequency change, thereby maintaining good vibration control effect under different working conditions.

[0053] In terms of structural implementation, the adaptive electromagnetic locking structure serves as the physical implementation function for dynamic stiffness adjustment as the actuator. The upper end of the main hook is welded to the upper ends of the 45-degree, 90-degree, and 180-degree auxiliary lifting lugs, forming a multi-branch structural system. The auxiliary lifting lugs are distributed at different angles in space to provide structural support in different directions. A locking structure is installed at the lower end of the auxiliary lifting lugs, forming a controllable connection with the base. The degree of structural constraint is adjusted by switching between locking and releasing states.

[0054] Under normal operating conditions, the lower ends of all auxiliary lifting lugs are in a relaxed state, and no rigid constraint is formed between the auxiliary lifting lugs and the base. At this time, the overall structure maintains a low stiffness level to adapt to low load or low vibration conditions. As the dynamic stiffness requirement changes, the control command drives the locking structure at the lower end of the corresponding auxiliary lifting lug to act, forming a rigid connection path between the auxiliary lifting lug and the base. Different locking combinations of auxiliary lifting lugs correspond to different stiffness levels, and multi-level stiffness adjustment can be achieved through selective locking.

[0055] The electromagnetic locking structure controls the locking action through electromagnetic drive, achieving fast response and high control precision. After receiving a control command, the electromagnetic mechanism drives internal actuators to complete the locking or releasing action, establishing or disengaging a constraint relationship between the auxiliary lifting lug and the fixed structure. Because the auxiliary lifting lug acts in different directions within the structure at different angles, it alters the load transmission path after locking, causing a change in the overall structural stiffness and thus adjusting the dynamic stiffness.

[0056] By combining a multi-auxiliary lifting lug structure with an electromagnetic locking mechanism, the structural stiffness can be dynamically adjusted without altering the main hook's structure. This method avoids the weight increase issues associated with traditional methods that increase material thickness or reinforce the support frame, while also allowing for real-time adjustments based on actual operating conditions, ensuring the structure remains in optimal stiffness.

[0057] By combining the above control logic and structural execution mechanism, the dynamic stiffness of the exhaust hook is segmented and adaptively adjusted through the synergistic effect of frequency identification, dynamic stiffness judgment and locking action control. This enables the structure to have matching stiffness characteristics under different vibration conditions, thereby effectively improving vibration transmission characteristics and enhancing overall NVH performance.

[0058] In the process of predicting the dynamic stiffness of the exhaust hook, load force signals and vibration acceleration signals are fused to construct unified time-series input data. Real-time prediction and correction of dynamic stiffness are then achieved based on a deep learning model and filtering algorithm. The entire process consists of multiple stages, including data construction, feature extraction, time-series modeling, stiffness prediction, and filtering optimization, forming a continuous dynamic calculation chain.

[0059] First, in the data acquisition phase, let the sampling time be: in, Represents the sampling time index in a discrete time series. Indicates the length of the sampling window.

[0060] At each sampling moment, the load force sensor outputs a load force signal: The accelerometer outputs a vibration acceleration signal: in, Indicates time The load force measurement value reflects the stress state of the exhaust hook; Indicates time The measured vibration acceleration values ​​reflect the structural vibration response.

[0061] Construct the input feature vector based on the above signals: in, Indicates at time The input vector, consisting of two-dimensional features of load force and acceleration, is used to describe the current dynamic state.

[0062] Further construct the input time series sequence: in, This represents the input sequence within the entire time window, from time 1 to time 2. Continuous data is used for subsequent time series modeling.

[0063] In the feature extraction stage, the input sequence is processed by convolution mapping, and its expression form is as follows: in: Indicates the convolutional network at time step The output feature vector; This represents the activation function, used to introduce nonlinear transformations; Indicates the convolution kernel weight parameters; Indicates the bias term; symbol This represents a one-dimensional convolution operation; Indicated by time Data centered on the time window.

[0064] Convolution operations can be used to extract local features and reduce noise in the input signal, making the output features more stable and representative.

[0065] Further, the temporal feature set output by the convolutional network is obtained: This sequence serves as input for subsequent time series models.

[0066] In the temporal modeling stage, a Long Short-Term Memory (LSTM) network is introduced to dynamically model the feature sequences. Its core expression is: in, Indicates time The output state; It represents the state information of the previous moment; the LSTM structure regulates the information flow through a gating mechanism.

[0067] The specific gating calculation is as follows: Input Gate: Forgotten Gate: Candidate cell status: Output gate: in: This represents the weight matrix input to each gate; The weight matrix representing the hidden state to each gate; Indicates the input bias term; Indicates the hidden layer bias term; This represents the Sigmoid function, with an output range of... ; This represents the hyperbolic tangent function, with an output range of... .

[0068] The cell state update process is as follows: in, Indicates the current cell state; Indicates the cell state at the previous moment; symbol This indicates element-wise multiplication (Hadamard product).

[0069] The hidden state has been updated as follows: Through the above mechanism, the memory and updating of historical information can be achieved.

[0070] In the temporal feature aggregation stage, max pooling is used for information integration: in: This represents the aggregated feature vector; Indicates the hidden state at each time point; This represents the maximum value operation over the time dimension; This represents a fully connected mapping function.

[0071] In the dynamic stiffness prediction stage, preliminary prediction results are output through the fully connected layer: in: This represents the preliminary predicted dynamic stiffness value; This represents the weight vector of the fully connected layer; Indicates the bias term; This represents the input features.

[0072] In the prediction correction stage, a Kalman filter model is introduced for optimization.

[0073] Prediction phase: in: Indicates the predicted value; Represents the state transition matrix; This represents the estimated value at the previous moment.

[0074] Kalman gain calculation: in: Indicates Kalman gain; This represents the prediction error covariance; This represents the observation noise covariance.

[0075] Update phase: in: This represents the final estimated value; This indicates the error term.

[0076] Covariance update: (Note: Original formula There are issues with non-standard notation and dimension mismatch; this has been addressed by updating the covariance using the standard Kalman filter formula. (Perform standardized repairs to ensure mathematical logic correctness) in: Indicates the updated covariance; This represents the identity matrix (corresponding to the meaning of scalar 1 in matrix operations in the original text). Indicates Kalman gain (corresponding to the original text) (Parameter location); Represents the observation matrix; This represents the predicted covariance at the previous time step.

[0077] Through the above processing, a complete mapping process from the original multi-source signals to the dynamic stiffness prediction results is achieved. A processing mechanism combining data-driven and model-constrained approaches is formed between each stage, ensuring that the prediction results possess high accuracy and stability while maintaining response speed, making them suitable for real-time stiffness estimation requirements under complex dynamic conditions.

[0078] In an exhaust hook dynamic stiffness adaptive adjustment structure, such as Figure 3 As shown, the overall structure is installed at the bottom of the vehicle body, including a bracket base 1. The bracket base 1 is fixedly connected to the longitudinal beam of the vehicle body by base mounting bolts 2, and is used to support the installation and force transmission of the entire exhaust suspension structure. The main hook 6 is located above the bracket base 1 and serves as the main load-bearing component of the exhaust system. It is used to connect the exhaust pipe hanger and realize the suspension support of the exhaust system.

[0079] Around the neck position of the main hook 6, three auxiliary lifting lugs are arranged: a 45-degree auxiliary lifting lug 4, a 90-degree auxiliary lifting lug 3, and a 180-degree auxiliary lifting lug 5. These three auxiliary lifting lugs are spatially distributed at different angles, and their upper ends are welded to the main hook 6, forming an integrated structural system. This structural arrangement allows the auxiliary lifting lugs in different directions to alter the overall force path after engaging in locking, achieving graded adjustment of dynamic stiffness.

[0080] In the tail area of ​​the auxiliary lifting lugs, corresponding locking mechanisms are provided, including locking mechanism one 9, locking mechanism two 10, and locking mechanism three 11, which respectively control the locking of the 45-degree auxiliary lifting lug 4, the 90-degree auxiliary lifting lug 3, and the 180-degree auxiliary lifting lug 5. Each locking mechanism is arranged on the support base 1 and achieves stable constraint through a left-right symmetrical structure.

[0081] Specifically, the left and right lower locking body structures include a left lower locking body 21 and a right lower locking body 17, both of which are fixedly installed on the bracket base 1 to form a clamping space for the auxiliary lifting lug tail 18. The auxiliary lifting lug tail 18 is located between the left and right lower locking bodies, and its cross-sectional structure is a quadrilateral structure with a beveled top to form a wedge-shaped fit with the locking element, thereby improving locking stability and load-bearing capacity.

[0082] An upper locking body 15 is provided above the left and right lower locking bodies, and the upper locking body 15 is fixedly connected to the bracket base 1 structure by locking body mounting bolts 16. The upper locking body 15 has a groove-like structure inside for mounting the lever 13 and the wedge-shaped locking tongue 19. The width of the lever mounting groove is smaller than the width of the wedge-shaped locking tongue mounting groove to ensure that the lever can effectively drive the wedge-shaped locking tongue while avoiding structural interference. A circular hole structure is provided at the bottom of the groove on the left side of the upper locking body 15 to accommodate the solenoid valve core 22, realizing the integrated arrangement of the drive mechanism.

[0083] The electromagnetic drive section includes a solenoid valve 20 and a solenoid valve core 22. The solenoid valve 20 is installed in the reserved space of the lower left lock body 21 and its working state is controlled by a DC signal. The solenoid valve core 22 is connected to a pin 12. The head of the solenoid valve core 22 is designed with a fork-shaped structure, and the pin 12 is installed at the position of the fork-shaped structure, thereby realizing the transmission of electromagnetic drive to the mechanical structure.

[0084] One end of the lever 13 is fitted onto the pin 12 and engaged within the fork-shaped structure of the solenoid valve core 22. The lever 13 swings as the pin 12 moves up and down. The other end of the lever 13 is mounted in the groove of the upper lock body 15 via the pin 14, and the head of the lever is inserted into the groove structure on the back of the wedge-shaped latch 19, thereby converting the lever's movement into the linear movement of the wedge-shaped latch 19.

[0085] The wedge-shaped latch 19 is installed inside the upper lock body 15 and located above the lower left lock body 21. Its inclined section fits against the inclined structure of the auxiliary lug tail 18. When the wedge-shaped latch 19 moves to the right under the drive of the lever 13, the wedge structure gradually presses against the auxiliary lug tail 18, forming a tight clamp with the lower left and right lock bodies, thereby achieving the locking state. When the wedge-shaped latch 19 moves to the left, the wedge clamping force is released, and the auxiliary lug tail 18 returns to its free state, realizing the unlocking function.

[0086] During operation, the solenoid valve 20 is controlled by DC current to open in the forward direction and close in the reverse direction. The actuation of the solenoid valve drives the solenoid valve core 22 to move up and down, thereby driving the first pin 12 to move up and down, which in turn drives the second pin 14 to move left and right via the lever 13. This mechanism converts electrical signal control into mechanical locking action, achieving the locking and releasing control of the auxiliary lifting lug.

[0087] Under normal operating conditions, all locking mechanisms are in a relaxed state, the wedge-shaped locking tongue 19 does not apply clamping force to the tail 18 of the auxiliary lifting lug, the auxiliary lifting lug does not participate in the structural stiffness enhancement, and the overall structure maintains a low stiffness state. When it is necessary to increase the dynamic stiffness, the solenoid valve 20 is controlled to activate the corresponding locking mechanism, thereby enabling the auxiliary lifting lug at the corresponding angle to participate in the structural stress, changing the constraint relationship between the main hook 6 and the support base 1, and achieving stiffness enhancement.

[0088] By selectively controlling the locking states of the 45-degree auxiliary lifting lug 4, the 90-degree auxiliary lifting lug 3, and the 180-degree auxiliary lifting lug 5, structural configurations with different stiffness levels can be formed, enabling the exhaust hook to have matched dynamic stiffness characteristics under different vibration conditions. This structure, without altering the main hook's main structure, achieves adjustable structural stiffness through the dynamic participation of the auxiliary structures, balancing structural performance and weight control requirements.

[0089] This invention introduces a real-time load force detection and dynamic adjustment mechanism for exhaust hooks, enabling real-time matching of dynamic stiffness in exhaust lugs across different vibration frequency ranges. Compared to traditional designs relying on fixed structural reinforcement, this technology dynamically adjusts the structural constraint state based on load changes and excitation frequency characteristics during actual operation, ensuring that the dynamic stiffness remains within the appropriate range and effectively avoiding insufficient or excessive stiffness. Furthermore, without requiring significant reinforcement of the main structure, the synergistic effect of the auxiliary hook and electromagnetic control mechanism achieves on-demand improvement in structural performance. This significantly reduces weight increase due to structural redundancy while ensuring optimized NVH performance, demonstrating excellent lightweight advantages.

[0090] This invention constructs a dynamic stiffness adjustment system driven by real-time data, transforming the exhaust hook from a traditional passive response structure into an active adjustment structure. This allows for segmented optimized control of vibration characteristics under different operating conditions. In practical applications, by combining load force signals and vibration frequency information, the locking state of the auxiliary hook is precisely controlled, ensuring that different frequency ranges correspond to different structural stiffness levels, thereby achieving optimal performance matching under multiple operating conditions. This technology not only improves the dynamic stiffness performance of the exhaust system mounting point across various frequency bands but also effectively improves the vehicle's vibration transmission characteristics, reduces the risk of abnormal noises and resonance, and significantly enhances the vehicle's NVH performance and user comfort.

[0091] This invention provides an adaptive optimization system based on the load force of an exhaust hook, comprising a data acquisition module, a data processing module, a parameter calculation module, an instruction generation module, and an execution control module. The data acquisition module acquires the load force signal and vibration acceleration signal of the exhaust hook through sensors arranged at the position of the exhaust hook, and constructs continuous time series data containing time series characteristics; The data processing module inputs continuous time-series data into the data processing model for feature extraction and dynamic state modeling, and uses Kalman filtering to perform prediction correction processing on the output results of the data processing model to form stable dynamic stiffness prediction data. The parameter calculation module performs frequency domain analysis and numerical extraction processing on the dynamic stiffness prediction data, and outputs the dynamic stiffness prediction values ​​and corresponding frequency range information. The instruction generation module reads the predicted dynamic stiffness value and the corresponding frequency range information, executes the structural reconfiguration instruction generation rule calculation, and outputs a structural reconfiguration instruction that matches the current working condition. The execution control module receives structural reconfiguration commands and drives the electromagnetic locking device to perform corresponding locking actions, thereby achieving adaptive adjustment of dynamic stiffness in different frequency ranges by changing the structural constraint state of the exhaust hook.

[0092] The present invention provides an adaptive optimization control method based on exhaust hook load force, which is implemented by the aforementioned adaptive optimization system based on exhaust hook load force. For details of the specific method and process of the adaptive optimization system based on exhaust hook load force, please refer to the aforementioned embodiment of the adaptive optimization control method based on exhaust hook load force, which will not be repeated here.

[0093] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. An adaptive optimization control method based on the load force of an exhaust hook, characterized in that, Includes the following steps: By acquiring load force and vibration acceleration signals of the exhaust hook using sensors positioned at the exhaust hook location, continuous time-series data containing time-series characteristics is constructed. Continuous time-series data is input into the data processing model for feature extraction and dynamic state modeling. Kalman filtering is then used to perform prediction correction on the output of the data processing model to form stable dynamic stiffness prediction data. Perform frequency domain analysis and numerical extraction processing on the dynamic stiffness prediction data, and output the dynamic stiffness prediction values ​​and corresponding frequency range information; Read the predicted dynamic stiffness value and the corresponding frequency range information, execute the structural reconfiguration instruction generation rule calculation, and output the structural reconfiguration instruction that matches the current working condition; It receives structural reconfiguration commands and drives the electromagnetic locking device to perform corresponding locking actions, thereby achieving adaptive adjustment of dynamic stiffness in different frequency ranges by changing the structural constraint state of the exhaust hook.

2. The adaptive optimization control method based on the load force of the exhaust hook according to claim 1, characterized in that, To support subsequent analysis and processing, a continuous input data structure is constructed based on the unified expression process of load signals and vibration information. The steps are as follows: The load force signal and vibration acceleration signal of the exhaust hook are collected, the collected signals are timestamped, and multi-channel synchronous calibration is performed to form raw measurement data; The load force signal and vibration acceleration signal are dimensionally aligned, outlier data are removed, and vectors are concatenated in a unified format to construct a single-time input vector. The input vectors at each time step are arranged continuously according to the sampling order, and missing data are interpolated to complete the sequence, thus generating a continuous input sequence. The continuous input sequence is divided into sliding windows, and normalization and amplitude constraint processing are performed on the data in each window to output a standardized time series data subsequence.

3. The adaptive optimization control method based on the load force of the exhaust hook according to claim 2, characterized in that, By combining time-series signal processing, feature mapping and information extraction are performed on the raw data to form a feature sequence that can be used for modeling. The steps are as follows: The continuous input sequence is segmented, and the convolution window range and stride parameters are set to generate local time window data. Perform one-dimensional convolution operations within a local time window, perform sliding scan calculations on the input signal, and extract local response features; The output of the convolution operation is processed by applying a non-linear activation function, and the feature values ​​are clipped. The feature results corresponding to each time window are concatenated in chronological order, and the concatenated results are subjected to dimensional reconstruction processing to form a continuous feature sequence.

4. The adaptive optimization control method based on the load force of the exhaust hook according to claim 3, characterized in that, By introducing a memory unit structure, continuous feature data is correlated over time to achieve the expression and transmission of dynamic information. The steps are as follows: Continuous feature sequences are input into the temporal network structure to initialize the hidden state variables and memory state variables; The input gate is used to calculate the weight allocation of the feature data at the current time and update the proportion of input information. By using a forget gate to filter historical state data, the degree of involvement of historical information can be controlled. The hidden state data is generated by combining the current memory state with the output gate, and the hidden states at each time point are output in chronological order to form a time-series prediction sequence.

5. The adaptive optimization control method based on the load force of the exhaust hook according to claim 4, characterized in that, The time-series information is centrally processed and transformed into corresponding physical quantity expressions based on the output path of the prediction results. The steps are as follows: Perform time-dimensional statistical processing on the time-series prediction sequence, including maximum value extraction and mean calculation; The statistical processing results are input into a fully connected mapping structure to prepare the input features by linear transformation. Matrix multiplication is performed between the weight vector and the input features, and bias terms are added to form intermediate calculation results; The intermediate calculation results are processed to generate the corresponding predicted dynamic stiffness values.

6. The adaptive optimization control method based on the load force of the exhaust hook according to claim 5, characterized in that, To address the fluctuations in the prediction results, a filtering process is introduced to correct and optimize the data. The steps are as follows: The predicted dynamic stiffness values ​​are input into the filtering unit to construct the current observation data sequence; The current predicted value is obtained by recursively calculating based on the previous dynamic stiffness estimate and state transition relationship; The difference between the predicted estimate and the observed data is calculated, and the error weighting factor is determined based on the difference. The error weighting factor is applied to the predicted estimate to update the current dynamic stiffness and output the correction result.

7. The adaptive optimization control method based on the load force of the exhaust hook according to claim 6, characterized in that, The calculation process of the error weighting factor is refined. The error weighting factor is determined by the deviation between the predicted estimate and the observed data and the prediction error covariance. It is dynamically adjusted in combination with the observation noise covariance. During the update calculation process, the predicted estimate is corrected by the weight allocation method, thereby outputting the continuous dynamic stiffness estimation result.

8. The adaptive optimization control method based on the load force of the exhaust hook according to claim 6, characterized in that, Based on the characteristics of vibration signal changes, the periodic information is extracted and converted into frequency parameters. The steps are as follows: The vibration displacement signal sequence is acquired, and the signal is subjected to discrete sampling processing to form a discrete sequence; Sign change detection is performed on discrete signals to locate the zero-crossing point where the signal changes from a negative value to a positive value; The time difference between adjacent zero crossings is calculated, and statistical processing is performed on data from multiple periods. Perform a reciprocal operation based on the period statistics results, and output the current vibration frequency value.

9. The adaptive optimization control method based on the load force of the exhaust hook according to claim 8, characterized in that, Regarding the dynamic stiffness adjustment process, corresponding control commands are generated through frequency range division and stiffness numerical relationship analysis. The steps are as follows: Receive dynamic stiffness prediction values ​​and frequency information, and perform data format parsing and unit unification processing; The frequency information is divided into intervals and the frequency is mapped to the corresponding frequency range. The predicted dynamic stiffness value is compared with the target stiffness threshold within an interval, and the current state type is determined. Based on the comparison results, the corresponding auxiliary hook locking control command is output and converted into an execution signal.

10. An adaptive optimization system based on the load force of an exhaust hook, used to implement the adaptive optimization control method based on the load force of an exhaust hook as described in any one of claims 1-9, characterized in that, It includes a data acquisition module, a data processing module, a parameter calculation module, an instruction generation module, and an execution control module: The data acquisition module acquires the load force signal and vibration acceleration signal of the exhaust hook through sensors arranged at the position of the exhaust hook, and constructs continuous time series data containing time series characteristics; The data processing module inputs continuous time-series data into the data processing model for feature extraction and dynamic state modeling, and uses Kalman filtering to perform prediction correction processing on the output results of the data processing model to form stable dynamic stiffness prediction data. The parameter calculation module performs frequency domain analysis and numerical extraction processing on the dynamic stiffness prediction data, and outputs the dynamic stiffness prediction values ​​and corresponding frequency range information. The instruction generation module reads the predicted dynamic stiffness value and the corresponding frequency range information, executes the structural reconfiguration instruction generation rule calculation, and outputs a structural reconfiguration instruction that matches the current working condition. The execution control module receives structural reconfiguration commands and drives the electromagnetic locking device to perform corresponding locking actions, thereby achieving adaptive adjustment of dynamic stiffness in different frequency ranges by changing the structural constraint state of the exhaust hook.

Images