An engine support load real-time monitoring method and system

Abstract

The application discloses an engine support load real-time monitoring method and system, and belongs to the field of automobile engineering state monitoring. The method comprises the following steps: acquiring original strain signals, real-time temperature signals and three-axis vibration acceleration signals synchronously collected by a composite sensing unit; performing elastic modulus correction and temperature drift deduction on the original strain signals according to the real-time temperature signals to obtain corrected stress values; inputting the corrected stress values into low-pass and high-pass filters arranged in parallel to extract low-frequency quasi-static load components and high-frequency dynamic alternating load components, and performing phase compensation; constructing a Kalman filter state space model with the three-axis vibration acceleration signals as control input and the corrected stress values as observation values, fusing multi-source information through filtering recursion, and outputting real-time load vectors. The application has the advantages of improving the precision and robustness of load monitoring, and providing reliable data support for fatigue evaluation and active vibration control of the engine support.

Description

Technical Field

[0001] This application belongs to the field of automotive engineering condition monitoring, specifically relating to a method and system for real-time monitoring of engine bracket load. Background Technology

[0002] With the rapid development of the automotive industry, the importance of powertrain support systems in improving vehicle vibration and noise performance and structural safety is becoming increasingly prominent. Among these, the engine mount, as a core component connecting the powertrain to the vehicle body, requires real-time monitoring of its load status for assessing component fatigue life and optimizing active vibration damping control. Especially under complex and variable driving conditions, the mount must withstand high-frequency excitation from the engine and random impacts from the road surface, placing higher demands on the monitoring system's data analysis accuracy, signal interference resistance, and operational stability in extreme environments.

[0003] However, traditional load monitoring schemes have significant limitations when facing the complex physical environment inside the engine compartment. The drastic temperature fluctuations generated by engine operation alter the mechanical properties of the support material, leading to severe nonlinear temperature drift errors when converting a single strain signal into load values. Simultaneously, the high-frequency alternating vibrations and electromagnetic noise of the engine are intertwined, and traditional static calibration methods lack effective decoupling capabilities for dynamic and static loads, making it difficult to extract true dynamic impact characteristics from the highly aliased raw signals. Furthermore, the monitoring process relies excessively on feedback from a single sensing channel, lacking redundant verification and fusion processing mechanisms for multi-source information. If sensor signal deviation or environmental interference occurs, the system faces the risk of monitoring results becoming invalid. Therefore, a real-time engine support load monitoring scheme based on multi-source information fusion is desired. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for real-time monitoring of engine bracket load, which can effectively solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] A method for real-time monitoring of engine mount load includes the following specific steps:

[0007] The original strain signal, real-time temperature signal, and triaxial vibration acceleration signal are synchronously acquired by the composite sensing unit installed at the preset stress concentration position of the engine bracket.

[0008] Based on the real-time temperature signal, the original strain signal is corrected for elastic modulus and temperature drift is subtracted to obtain the corrected stress value.

[0009] The corrected stress value is input to the parallel low-pass filter and high-pass filter respectively. The low-frequency quasi-static load component is extracted through the low-pass filter, and the high-frequency dynamic alternating load component is extracted through the high-pass filter. The filtered signal is then phase-compensated to obtain the time-aligned low-frequency and high-frequency components.

[0010] A Kalman filter state-space model is constructed with triaxial vibration acceleration signals as control inputs and corrected stress values ​​as observations. The corrected stress values ​​and triaxial vibration acceleration signals are recursively fused through Kalman filtering to output a real-time load vector.

[0011] Furthermore, the original strain signal is corrected for elastic modulus and temperature drift is subtracted based on the real-time temperature signal to obtain the corrected stress value, including:

[0012] Based on a pre-calibrated stiffness-temperature mapping table, the elastic modulus at the current real-time temperature is calculated by querying or interpolation.

[0013] Multiply the original strain signal by the elastic modulus at the current real-time temperature to obtain the stress value based on the real-time stiffness;

[0014] Based on a pre-calibrated temperature-stress drift mapping table, the stress drift baseline value at the current real-time temperature is calculated by querying or interpolation.

[0015] The corrected stress value is obtained by subtracting the stress drift reference value from the stress value based on real-time stiffness.

[0016] Furthermore, phase compensation is performed on the filtered signal. Specifically, a bidirectional filtering technique is used, where the input signal is passed through the filter in the forward direction, and then the filtering result is passed through the same filter in the reverse direction. The reverse order of the reversed result is taken as the output to eliminate the phase lag introduced by the filter, so that the low-frequency component and the high-frequency component of the output are strictly aligned with the original excitation in time.

[0017] Furthermore, a Kalman filter state-space model is constructed using triaxial vibration acceleration signals as control inputs and corrected stress values ​​as observations, including:

[0018] Construct a state vector, which includes at least the load forces and their rates of change of the engine mount in three directions in the vehicle body coordinate system;

[0019] A state transition matrix is ​​constructed based on the discrete-time kinematic equations, and an input control matrix is ​​constructed based on the equivalent mass matrix of the support to convert the triaxial vibration acceleration signal into an input control matrix that affects the load change rate.

[0020] An observation matrix is ​​constructed based on the stress-load conversion coefficients obtained from the static calibration experiment. The observation matrix is ​​used to map the state vector to an observation vector consisting of corrected stress values.

[0021] Furthermore, after outputting the real-time load vector, the process also includes: according to the generalized Hooke's law, using the real-time corrected elastic modulus, the preset Poisson's ratio, and the triaxial principal strain obtained by the composite sensing unit, performing multiaxial strain synthesis correction on the real-time load vector to obtain the corrected load vector.

[0022] Furthermore, the composite sensing unit is installed at a preset stress concentration location on the engine mount. The preset stress concentration location is determined by performing finite element simulation analysis on the engine mount to identify the region with the highest strain energy density under different load combinations.

[0023] Furthermore, the composite sensing unit includes a strain sensor, a temperature sensor, and an acceleration sensor, all of which are integrated and packaged in a housing. The housing is an engineering plastic shell filled with thermally conductive and insulating material, and a vibration damping structure is provided between the acceleration sensor mounting base and the shell.

[0024] Furthermore, the composite sensing unit also integrates a self-diagnostic module, which is used to monitor the bridge balance of the strain sensor and the resistance range of the temperature sensor in real time. When a sensor failure is diagnosed, the system automatically switches to an open-loop estimation mode based on acceleration signals, using historical stiffness data and real-time acceleration signals to maintain basic load estimation functions.

[0025] Furthermore, the real-time load vector is transmitted to the vehicle electronic control unit via the controller area network bus for fatigue life assessment or as a control reference input for the active suspension system; the real-time load vector is encrypted before transmission.

[0026] An engine mount load real-time monitoring system, comprising:

[0027] The composite sensing unit is integrated and installed at the preset stress concentration position of the engine bracket to synchronously acquire the original strain signal, real-time temperature signal and triaxial vibration acceleration signal.

[0028] The signal preprocessing module is used to perform zero-point offset compensation on the triaxial vibration acceleration signal and to synchronously acquire and convert all signals to digital.

[0029] The temperature compensation module is used to correct the elastic modulus and reduce the temperature drift of the original strain signal based on the real-time temperature signal, and output the corrected stress value.

[0030] The filtering and decoupling module has a built-in low-pass filter and a high-pass filter connected in parallel to filter and separate the corrected stress value, and perform phase compensation on the filtered signal to output time-aligned low-frequency quasi-static load components and high-frequency dynamic alternating load components.

[0031] The Kalman filter incorporates a state-space model that uses triaxial vibration acceleration signals as control inputs and corrected stress values ​​as observations. It is used to recursively fuse corrected stress values ​​and triaxial vibration acceleration signals through Kalman filtering to output a real-time load vector.

[0032] In summary, this application includes at least one of the following beneficial technical effects:

[0033] 1. By introducing a real-time temperature compensation mechanism, this invention effectively eliminates the impact of drastic temperature fluctuations in the engine compartment on the mechanical properties of the support material. Utilizing preset material property curves and stiffness-temperature mapping tables, the system can correct the thermal decay of the elastic modulus in real time, reducing temperature drift error by a predetermined proportion. This dynamic correction capability ensures that load monitoring results maintain a high degree of linearity and accuracy across the entire temperature range, from cold starts to prolonged high-speed driving.

[0034] 2. The dual-channel filtering algorithm employed in this invention, combined with dynamic state equations, successfully achieves effective separation of low-frequency quasi-static loads and high-frequency dynamic alternating loads. The low-pass filter branch accurately captures changes in engine weight and installation preload, providing a basis for the static strength assessment of the bracket; while the high-pass filter branch, combined with acceleration feedback, can sensitively capture impact or transient peak loads. This decoupling capability provides crucial data support for the refined tuning of vehicle vibration and noise performance and the accurate assessment of bracket fatigue damage.

[0035] 3. By integrating multi-source sensors for strain, temperature, and acceleration, this invention constructs a monitoring system with redundancy verification capabilities. The introduction of a Kalman filter allows the system to move beyond relying solely on a single strain signal, incorporating acceleration signals as kinematic constraints into the load estimation process. This multi-source information fusion strategy effectively overcomes the shortcomings of single sensors, which are susceptible to temperature drift, signal offset, or random noise interference. Even with the degradation of some sensor performance, the system can still maintain basic monitoring functions through the remaining sensing channels and dynamic model, significantly improving the robustness of the engine mount load monitoring system under extreme conditions and its long-term operational stability.

[0036] 4. The load vector output in real time by this invention is seamlessly integrated with the vehicle's electronic control unit via the controller local area network bus. This not only provides real-time feedback for the adaptive control of the active suspension system, enabling precise suppression of powertrain vibration, but also provides detailed data accumulation for component life prediction based on load spectrum. Through statistical analysis of long-term monitoring data, maintenance plans can be formulated more scientifically, preventing safety accidents caused by bracket structure failure, thereby comprehensively improving the structural safety and ride comfort of the entire vehicle. Attached Figure Description

[0037] Figure 1This is an overall schematic diagram of the real-time monitoring method for engine mount load;

[0038] Figure 2 This is a schematic diagram illustrating the core principle of multi-source information fusion load estimation based on Kalman filtering;

[0039] Figure 3 It is a logic flowchart of the original signal temperature compensation and the decoupling of dynamic and static load components;

[0040] Figure 4 This is a schematic diagram of the multi-level interaction and data flow between the real-time load vector and the vehicle control system. Detailed Implementation

[0041] To make the objectives, technical solutions, and advantages of this invention clearer, the following description is provided in conjunction with the appendix. Figure 1-4 The present invention will be further described in detail with reference to specific embodiments.

[0042] The real-time monitoring method for engine mount load is implemented according to the following specific steps:

[0043] First, in step S1, a composite sensing unit is installed. This unit integrates a strain sensor, a temperature sensor, and an acceleration sensor, enabling it to simultaneously sense the mechanical response of the support, ambient temperature, and dynamic excitation. To ensure the accuracy of the installation location and the representativeness of the signal, a full-condition simulation model of the engine support is performed using finite element analysis. The region with the highest strain energy density under different load combinations is identified, i.e., the preset stress concentration location. The composite sensing unit is then installed at the main stress-bearing ribs of the support to maximize the capture of minute deformations during the load-bearing process. This includes the following specific implementation sub-steps:

[0044] Step S101: Determine the installation location and prepare the bracket surface. First, perform a full-condition simulation model of the engine bracket using finite element analysis software such as Abaqus or ANSYS. When modeling, input the bracket's geometric model, material properties including elastic modulus and Poisson's ratio, and boundary conditions including constraints on the engine connection point and the chassis connection point.

[0045] In the simulation, various typical load conditions, such as engine maximum torque, road impact, and emergency braking, are applied to calculate the strain energy density distribution of the support under different load combinations. Post-processing analysis identifies the nodes or element regions with the highest strain energy density under all load conditions, and these regions are defined as preset stress concentration locations at the main load-bearing ribs.

[0046] Use sandpaper to sand the surface of the bracket at this location to remove the oxide layer and oil stains, then clean it with acetone or alcohol to ensure that the surface is dry and dust-free, providing a good adhesion base for the sensor to be pasted.

[0047] Step S102: Install a triaxial strain rose. To meet the requirements of subsequent multiaxial stress calculations, this step involves installing a triaxial strain rose (i.e., three mutually orthogonal strain gauges). Based on the principal stress directions obtained from the finite element analysis in step S101, a local coordinate system is established on the support surface, defining the X-axis along the direction of the maximum principal stress and the Y-axis along the direction of the minimum principal stress. The three strain gauges of the triaxial strain rose are attached along the X-axis (0°), Y-axis (90°), and at a 45° angle to the X-axis, ensuring that the deviation of the strain gauge grid length direction from the corresponding marking line does not exceed ±5 degrees. High-strength adhesive is used during attachment to ensure complete adhesion between the substrate and the support surface without air bubbles. After curing, signal lines are led out and connected to a Wheatstone bridge circuit. The output signals of the three channels... It will be used for subsequent calculations of triaxial principal strain.

[0048] After bonding, heat or allow to stand to cure according to the adhesive's curing requirements. The strain gauge's output leads are soldered to the signal transmission line. The strain gauge is connected to a Wheatstone bridge circuit, with either a full-bridge or half-bridge configuration selected based on measurement requirements. If a half-bridge configuration is used, the working strain gauge is bonded to the support surface, while the temperature-compensated strain gauge is bonded to a compensation block made of the same material as the support but not subjected to stress, and placed in the same temperature environment.

[0049] The bridge circuit converts resistance changes into voltage signal output. The symmetry of the bridge cancels out deformation interference and temperature effects in non-target directions, thereby improving the signal-to-noise ratio.

[0050] Step S103: Install the temperature sensor. A negative temperature coefficient thermistor is selected as the temperature sensor; its resistance decreases exponentially with increasing temperature. The distance between the thermistor and the center of the strain sensor must be less than a preset distance threshold.

[0051] This threshold is determined experimentally based on the thermal response time of the thermistor and the thermal diffusivity of the support material. Specifically, the support is heated in the laboratory and the temperature change in the strain gauge area is measured while the thermistor response is monitored. The maximum spacing corresponding to the temperature difference between the two is less than 0.5℃, which is the preset spacing threshold, for example, 5 mm.

[0052] The thermistor is attached to the bracket surface with thermally conductive adhesive to ensure good thermal contact with the bracket, and the leads are soldered out. The thermistor and the fixed resistor form a voltage divider circuit. By measuring the voltage divider voltage, the temperature value is calculated, achieving spatial synchronization of temperature sensing and providing an accurate basis for subsequent temperature compensation.

[0053] Step S104: Install the accelerometer sensor. The accelerometer sensor is a microelectromechanical system (MEMS) triaxial accelerometer, such as a capacitive or piezoresistive MEMS accelerometer. The measurement range is selected according to the engine vibration amplitude as ±50g or ±100g. The accelerometer is installed at the geometric center of the composite sensing unit, and during installation, it must be ensured that its sensitive axis is aligned with the vehicle coordinate system.

[0054] To achieve axis alignment, a positioning reference surface can be designed on the bracket, or a clamp can be used to fit the accelerometer housing against a pre-machined plane on the bracket. This plane is precision-machined to ensure parallelism with the vehicle coordinate system. The accelerometer is fixed with screws or high-strength adhesive, and its output signal line is shielded and led out.

[0055] Accelerometers are used to monitor the vibration acceleration of the engine mount in real time along the mutually perpendicular longitudinal (X), lateral (Y), and vertical (Z) axes. This signal serves as a kinematic reference for dynamic load prediction and provides input for subsequent Kalman filter fusion.

[0056] Step S105: Package the composite sensing unit and configure the signal conditioning circuit. The composite sensing unit is entirely encapsulated in an engineering plastic housing, which is made of polyetheretherketone (PEEK) material.

[0057] The interior of the housing is filled with an insulating material with a preset thermal conductivity, which is selected to be close to that of the support material, such as 30 W / m·K, to improve the response speed of the temperature sensor. At the same time, the filling material, such as silicone rubber filled with alumina ceramic powder, can both conduct heat and provide insulation.

[0058] The encapsulation process ensures unobstructed heat conduction between the sensor and the mounting bracket, and the filler material makes close contact with each sensor while avoiding stress. A vibration-damping structure is incorporated between the accelerometer mounting base and the housing, such as a 1 mm thick silicone gasket placed under the base, to attenuate interference from non-structural high-frequency noise in the acceleration measurement. The strain sensor's signal conditioning circuitry includes a differential amplifier and an anti-aliasing filter.

[0059] The differential amplifier uses an instrumentation amplifier such as AD620, with a gain adjustable from 100 to 1000 times, to amplify the millivolt-level voltage output of the bridge to the input range of the analog-to-digital converter.

[0060] The anti-aliasing filter employs a second-order active low-pass filter with a cutoff frequency set at least twice the engine's highest vibration frequency, for example, 1 kHz, to filter out high-frequency electromagnetic interference and avoid sampling aliasing. All conditioned signals are transmitted to the multi-channel synchronous acquisition module via shielded cables.

[0061] In summary, through step S1, the installation and debugging of the composite sensing unit are completed, enabling multi-source synchronous sensing of the support strain, temperature, and vibration acceleration. This sensing unit provides a reliable data foundation for subsequent temperature compensation, dynamic and static load decoupling, and multi-source information fusion, thus laying the hardware support for real-time monitoring of engine support load. Next, step S2 will be executed to acquire these sensor signals in real time and perform preprocessing.

[0062] Then, step S2 is executed to acquire the original strain signal, real-time temperature signal, and triaxial vibration acceleration signal of the support in real time, and to complete data preprocessing. This process is implemented in detail according to the following sub-steps:

[0063] Step S201, multi-source signal synchronous acquisition; the system uses a multi-channel synchronous acquisition module to perform analog-to-digital conversion on the strain, temperature and acceleration electrical signals output by the composite sensing unit at a preset sampling frequency.

[0064] The sampling frequency is set based on the main vibration frequency component at the engine's highest operating speed, and is taken to be more than 2.56 times that frequency component to ensure complete coverage of the signal spectrum and satisfy the Nyquist sampling theorem, thus avoiding spectral aliasing.

[0065] Sampling accuracy of 16 bits or higher is selected to reduce the impact of quantization noise on weak strain signals. During the acquisition process, a high-precision clock synchronization mechanism is used internally within the module, such as through the same hardware trigger signal or by sharing the same high-stability crystal oscillator clock source, to control the deviation of sampling time of each channel within the microsecond level. This ensures that strain, temperature, and acceleration data are strictly aligned on the time scale, providing a time-consistent basic data stream for subsequent fusion calculations.

[0066] Step S202, zero-point offset compensation of acceleration signal; since the accelerometer is sensitive to gravitational acceleration in a static installation state, zero-point offset compensation is required before the signal enters the processor.

[0067] The specific operation is as follows: when the vehicle is stationary and the engine is off, continuously collect several frames of the three-axis output values ​​of the accelerometer, and take the arithmetic mean as the static bias vector.

[0068] Then, during vehicle operation, the static bias vector is subtracted from the original acceleration signal in real time to eliminate the influence of gravity component and obtain dynamic acceleration generated only by vibration.

[0069] This compensation process can effectively avoid gravitational interference and improve the accuracy of acceleration signals as kinematic references.

[0070] Step S203, Signal quality monitoring and anomaly marking: While acquiring data, the system monitors the amplitude range and rate of change of the original signal in real time.

[0071] The reasonable range of amplitude is preset according to the range of each sensor and the signal characteristics under normal operating conditions. For example, the strain signal is -2000 με to +2000 με, the temperature signal is -40℃ to +150℃, and the acceleration signal is -50g to +50g.

[0072] The threshold for the rate of change is set according to the maximum possible rate of change of the physical process, such as the strain rate of change not exceeding 1000 με / s, the temperature rate of change not exceeding 10℃ / s, and the acceleration rate of change not exceeding 500g / s.

[0073] The monitoring process is achieved by comparing the amplitude of each sampling point with the rate of change between adjacent sampling points to see if they exceed the corresponding threshold. Once a channel signal is detected to exceed the preset range or a sudden change occurs, the system immediately marks the abnormal state flag in the data frame and reduces the weight of the data in subsequent processing by adjusting the corresponding elements of the measurement noise covariance matrix of the Kalman filter, or directly triggers the sensor self-diagnosis process.

[0074] The self-diagnostic process includes checking whether the sensors are disconnected, short-circuited, or exceed their electrical characteristic range, and reporting the diagnostic results to the vehicle's electronic control unit via the controller area network bus so that the system can take timely degrade or alarm measures.

[0075] In summary, step S2 achieves high-fidelity synchronous acquisition and preliminary processing of the support strain, temperature, and vibration acceleration, providing an accurate and time-consistent data foundation for subsequent temperature compensation, dynamic and static load decoupling, and multi-source information fusion. Next, step S3 will be executed to correct the elastic modulus of the original strain using real-time temperature.

[0076] Next, step S3 is executed to perform temperature compensation on the original strain signal, eliminating the deviation in material mechanical properties caused by temperature changes and obtaining a preliminary corrected stress value. This process is implemented in detail according to the following sub-steps:

[0077] Step S301: Pre-calibrate the material property curve and temperature drift reference value; under laboratory conditions, prepare standard tensile specimens from the metal material used for the engine bracket and place them in a material testing machine. At different temperature points, for example, every 10°C from -40°C to 150°C as a test point, apply a constant load to the specimen and measure its strain, and calculate the elastic modulus at each temperature point through the stress-strain relationship.

[0078] Record the corresponding data of temperature and elastic modulus, and use the least squares method to fit the functional relationship between elastic modulus and temperature, or directly construct a stiffness-temperature mapping table and store it in the system memory. At the same time, under no-load conditions, install the strain sensor on a test block of the same material as the support, and place it in a temperature-controlled chamber for cyclical changes from low temperature to high temperature. Record the output strain value of the strain sensor at each temperature, and this output value is the strain drift value.

[0079] The strain drift values ​​measured at each temperature point The elastic modulus of the material measured at the same temperature Multiplying them together yields the stress drift reference value. A temperature-stress drift mapping table is constructed and stored in memory. By fitting the drift values ​​to the temperature data, a curve or table showing the stress drift as a function of temperature can also be obtained for subsequent subtraction.

[0080] Step S302: Obtain the real-time temperature and query the correction coefficient; the system reads the temperature sensor signal acquired in step S2 in real time. .according to Look up the corresponding elastic modulus correction factor in the pre-stored stiffness-temperature mapping table.

[0081] like If it is exactly equal to a certain temperature node in the table, then the elastic modulus value of that node is directly used; if If the temperature is located between two temperature nodes, a linear interpolation algorithm is used to calculate the elastic modulus at the current temperature. The interpolation formula is:

[0082]

[0083] in and These are the elastic moduli of adjacent low-temperature and high-temperature nodes, respectively. and This corresponds to the temperature value. This interpolation method ensures the continuous smoothness of the correction coefficient across the entire temperature domain.

[0084] Step S303: Calculate the elastic modulus at the current temperature; based on the elastic modulus correction coefficient obtained in step S302, combine it with the elastic modulus at the standard temperature. Calculate the actual elastic modulus at the current temperature. If the stiffness-temperature mapping table directly stores the elastic modulus value, then This is the query result; if the stored value is the attenuation ratio relative to the standard temperature, it is calculated using the following formula:

[0085]

[0086] in The thermoelastic coefficient of the material is obtained by performing linear regression fitting on the elastic modulus data at different temperatures in step S301. For reference standard temperature, 20℃ is usually taken.

[0087] Step S304: Initially correct the stress value and subtract temperature drift from the original strain signal obtained in step S2. Multiply by the current temperature elastic modulus obtained in step S303 This yields stress values ​​based on real-time stiffness.

[0088] Then based on real-time temperature Query the pre-calibrated temperature-stress drift mapping table and obtain the stress drift reference value at the current temperature through linear interpolation. Subtracting this drift from the above stress value yields the final, preliminarily corrected stress value. The calculation formula is uniformly expressed as:

[0089]

[0090] in The stress was obtained from the temperature-stress drift table using interpolation. This completed the joint compensation for the effect of temperature on material stiffness and sensor temperature drift, providing accurate stress input for subsequent decoupling of dynamic and static loads.

[0091] Through step S3 above, the system achieves real-time temperature compensation for the original strain signal, effectively eliminating measurement errors caused by temperature fluctuations in the engine compartment. The compensated stress value more accurately reflects the mechanical load-bearing state of the support. Next, step S4 will be executed to decouple the dynamic and static load components of the corrected stress signal.

[0092] After obtaining the preliminary corrected stress value, step S4 is executed, which uses a dual-channel filtering algorithm to decouple the dynamic and static loads of the signal. Specifically, the following sub-steps are implemented:

[0093] Step S401: Design a dual-channel filter combination architecture; the system constructs a dual-channel filter structure consisting of a low-pass filter and a high-pass filter connected in parallel, which processes the corrected stress signal obtained in step S3. Two filter channels are input simultaneously.

[0094] The low-pass filter extracts the low-frequency quasi-static load components reflecting the engine's weight and installation preload, while the high-pass filter extracts the high-frequency dynamic alternating load components reflecting engine operating excitation and road impact. The two filters are designed independently to ensure that the decoupled components can be used separately for static strength assessment and dynamic fatigue analysis.

[0095] Step S402: Set the low-pass filter parameters and extract the low-frequency quasi-static load component; the low-pass filter is a Butterworth filter with a preset order of 4 to obtain maximum flatness in the passband and ensure sufficient stopband attenuation.

[0096] Cutoff frequency of low-pass filter The first preset cutoff frequency is set, which is determined based on the frequency range of the quasi-static load actually borne by the engine mount. By analyzing the natural frequency of the powertrain mounting system and the spectral characteristics of road excitation, the cutoff frequency is set to 1 Hz to ensure that all load components below 1 Hz (such as engine gravity changes and installation preload fluctuations) can pass through without attenuation, while dynamic components above 1 Hz are effectively suppressed.

[0097] The filter is implemented digitally, filtering the input signal through iterative difference equations, and the output is the low-frequency quasi-static load component. .

[0098] Step S403: Set the high-pass filter parameters and extract the high-frequency dynamic alternating load components; the high-pass filter is also a Butterworth filter, preset to order 4, to ensure a steep transition band and suppress low-frequency interference. The cutoff frequency of the high-pass filter... The second preset cutoff frequency is set based on the fundamental frequency corresponding to the engine's highest operating speed and the main frequency range of road impact.

[0099] For example, for an engine with a maximum speed of 6000 rpm, its ignition fundamental frequency is approximately 200 Hz. To fully capture dynamic impact characteristics, the cutoff frequency is set to 10 Hz, ensuring that high-frequency components above 10 Hz (such as engine reciprocating inertial force and road impact force) pass through, while quasi-static components below 10 Hz are filtered out. The filter is also implemented digitally, outputting high-frequency dynamic alternating load components. .

[0100] Step S404: Compensating for filter phase lag ensures time alignment. Because digital filters introduce frequency-dependent phase delays, the filtered components may misalign with the original excitation on the time axis, affecting the accuracy of subsequent analysis. Therefore, the system employs a piecewise bidirectional filtering technique to achieve zero phase distortion while meeting real-time requirements.

[0101] In practice, the continuously input signal stream is divided into fixed lengths. Divide into data blocks, and maintain the distance between adjacent blocks. The overlap of sampling points (usually taken) (i.e., 50% overlap). Bidirectional filtering is performed independently on each data block: first, the signal within the block is passed through the filter in the forward direction; then, the filtered result is passed through the same filter in the reverse direction, and the reverse order of the reversed result is taken as the zero-phase output of that block. Since bidirectional filtering requires a complete data block, the output will have a fixed delay relative to the input, approximately half the block length (i.e., ...). (Sampling period).

[0102] Buffer length The choice needs to balance the phase compensation effect with the real-time response requirements: Too large an amount will increase latency and affect the real-time performance of control; If the value is too small, it may reduce the filtering accuracy of low-frequency components. The sampling frequency is determined based on the frequency range of the engine mount's dynamic response (the lowest frequency of interest is typically 1 Hz). Hz, block length The delay is approximately 0.5 seconds. This delay is perfectly acceptable for offline analysis tasks such as fatigue assessment. For applications with high real-time requirements, such as active suspension control (control cycle is typically 10–100 ms), the load component after zero-phase filtering can be used as the reference signal, and the actual closed-loop control still uses causal filtering output, or piecewise bidirectional filtering can be applied to data post-processing.

[0103] Compensated low-frequency quasi-static load components and high-frequency dynamic alternating load components Strictly aligned with the original incentive in time, providing a reliable basis for subsequent independent monitoring and evaluation.

[0104] In summary, through step S4, the system achieves precise decoupling of dynamic and static loads in the corrected stress signal. The low-frequency component accurately reflects the foundation bearing state of the support under static or quasi-static conditions, while the high-frequency component captures dynamic impact characteristics, laying the foundation for subsequent fatigue life assessment and active vibration control. Next, step S5 will be executed, using a Kalman filter to fuse multi-source information and output the final real-time load vector.

[0105] Finally, step S5 is executed, where a Kalman filter is used to fuse multi-source information, outputting the final real-time payload vector. This process is implemented in detail according to the following sub-steps:

[0106] Step S501: Construct the state-space model of the Kalman filter; the system uses the triaxial vibration acceleration signal as the control input and the corrected stress value obtained in step S3 as the observation value to establish a state-space model describing the dynamic characteristics of the support.

[0107] The state vector X is defined as containing the load forces and their rates of change of the bracket in the three directions of the vehicle body coordinate system, i.e. , where Fx, Fy, and Fz are the longitudinal force, transverse force, and vertical force, respectively, and dFx, dFy, and dFz are the corresponding rates of change of force.

[0108] The state transition matrix A is determined based on the discrete-time kinematic equations. If the sampling period is Δt, then the specific form of matrix A is:

[0109]

[0110] This matrix indicates that the load force at the current moment is equal to the load force at the previous moment plus the product of the rate of change of force and the sampling period, and the rate of change of force remains constant at the current moment.

[0111] The input control matrix B converts the acceleration signal into its effect on the rate of change of the load. Its construction is based on a discretized form of Newton's second law: the rate of change of the dynamic load on the support satisfies the following relationship: ,in The equivalent mass matrix of the stent (a 3×3 diagonal matrix, with diagonal elements of 1) , representing the equivalent mass in the longitudinal, transverse, and vertical directions, respectively. This represents the triaxial acceleration increment.

[0112] In a discrete-time system, let the sampling period be... At present The acceleration is The previous moment was Then the acceleration increment Increment of the rate of change of load Substituting the above relationships into the discretized equations of the state-space model From this, it can be deduced that The specific form of the matrix.

[0113] Due to the state vector Control input Triaxial vibration acceleration (Zero-point offset compensation has been performed). The matrix is The matrix has the following element arrangement:

[0114]

[0115] The physical meaning of this matrix is: the acceleration input at the current moment will directly change the rate of change of the load at the next moment (the last three components of the state vector), while the load force itself (the first three components of the state vector) will be affected by the subsequent state transition matrix. Obtained by accumulating the rate of change. Equivalent mass matrix. The sampling period can be obtained through finite element analysis or experimental modal analysis. It is a known constant.

[0116] In this embodiment, the composite sensing unit can simultaneously obtain the corrected stress values ​​in three directions, therefore the observation vector is... The observation matrix H is a 3×6 matrix, with the first three columns corresponding to stress-load conversion coefficients and the last three columns being zero.

[0117] The elements of H were obtained through a static calibration experiment: known triaxial loads were applied to the support, the stress observations were recorded, and the transformation coefficient matrix H was obtained by fitting using the least squares method.

[0118] Step S502: Initialize filter parameters; at system startup, initialize the state vector based on the static load observations under specific engine conditions. For example, when the engine is idling and the vehicle is stationary, the stress value measured by the strain sensor is converted to obtain the initial load force, while assuming the initial force change rate is zero.

[0119] The initial error covariance matrix P0 is set as a diagonal matrix, and the diagonal elements are set according to the degree of uncertainty of the initial state. For example, the initial covariance of the force direction is 10 N², and the force rate of change direction is 1 (N / s)².

[0120] The process noise covariance matrix Q reflects the influence of model error and acceleration noise. It is determined as follows: with the vehicle stationary and the engine off, acceleration signals are collected for a period of time, and the covariance matrix is ​​calculated as the baseline value for the acceleration-related part of the Q matrix. For the process noise corresponding to the rate of change of load force, the model error is analyzed through finite element simulation, and an isotropic empirical value is taken. Finally, Q is set as a diagonal matrix, with the diagonal elements being the sum of the above two items.

[0121] The noise covariance matrix R reflects the noise level of stress observation. It is determined by acquiring the output signal of the strain sensor under constant temperature and load conditions and calculating its variance as the value of matrix R. In this embodiment, a triaxial observation is performed, so R is a 3×3 diagonal matrix, with the diagonal elements representing the variance of each channel.

[0122] Step S503: Perform Kalman filtering recursion; at each sampling time, the system sequentially executes a prediction step and an update step. The prediction step uses the state equation and acceleration input to calculate the prior state estimate and prior error covariance.

[0123] The update step uses the corrected stress observations at the current moment to calculate the Kalman gain, and then updates the posterior state estimate and posterior error covariance. The above recursive process is implemented using the standard Kalman filter algorithm, and the filtered state vector contains the real-time load forces Fx, Fy, and Fz at the current moment.

[0124] Step S504: Consider the influence of Poisson's ratio to correct the load vector; since the support material has a lateral shrinkage effect under multiaxial stress, it needs to be corrected according to Poisson's ratio ν.

[0125] This system employs a multiaxial strain synthesis algorithm, utilizing strain gauges mounted in orthogonal directions to obtain the triaxial principal strain. , , The triaxial principal stresses were calculated using the generalized Hooke's law:

[0126]

[0127]

[0128]

[0129] Where E is the corrected elastic modulus at the current temperature. The calculated triaxial principal stresses are converted into load forces using the transformation matrix calibrated in step S501, resulting in the corrected load vector. , , .

[0130] Step S505: Output the real-time load vector; the corrected triaxial load force is output as the real-time load vector, which comprehensively represents the force distribution of the engine mount in the vehicle coordinate system. The system transmits this vector to the vehicle electronic control unit via the controller local area network bus for subsequent fatigue life assessment and active vibration control.

[0131] In summary, through step S5 above, the system achieves multi-source information fusion based on Kalman filtering, combines acceleration signals with corrected stress values, and outputs a high-precision, highly robust real-time load vector.

[0132] This embodiment also includes data transmission and application steps: the calculated real-time load vector is transmitted to the vehicle electronic control unit via the controller local area network bus. The communication cycle of the controller local area network bus is set to a preset communication cycle, such as 10 milliseconds, to ensure that the real-time performance of the load monitoring results meets the requirements of the vehicle control system. To ensure data security, the method also includes local encryption of the monitoring data: the AES-128 symmetric encryption algorithm is used, and the encryption key is pre-stored in the vehicle's security chip. The load vector data is encrypted before transmission, and the receiver uses the same key to decrypt it, preventing the load data from being illegally tampered with during bus transmission.

[0133] After receiving the real-time load vector, the on-board electronic control unit (ECU) performs fatigue life assessment based on its fluctuation characteristics. Specifically, the system uses the rainflow counting method to extract stress cycles from the load time history, combines this with the SN curve of the engine mount material (obtained through material fatigue testing), and applies the Miner linear cumulative damage criterion to calculate the cumulative damage value D. It is assumed that the total design damage tolerance of the mount is... The current remaining useful life percentage is When the accumulated damage exceeds a preset threshold or the peak load exceeds a safety threshold, the system triggers an alarm logic and issues a warning to the driver or maintenance personnel via the vehicle's instrument panel or remote diagnostic system.

[0134] Furthermore, the real-time load vector is also used as a control reference input for the active suspension system. The system decomposes the load vector into low-frequency quasi-static components and high-frequency dynamic components: the low-frequency component is used to adjust the dynamic stiffness of the active suspension, and the high-frequency component is used to adjust the damping. The specific adjustment strategy is as follows: when the amplitude of the high-frequency dynamic load increases, the damping coefficient of the suspension is linearly increased to absorb vibration energy; when the low-frequency quasi-static load increases, the dynamic stiffness is appropriately increased to maintain support stability. The adjustment amount is determined by looking up a table from a pre-calibrated MAP chart, thereby achieving real-time compensation and suppression of powertrain vibration and optimizing the NVH performance of the entire vehicle.

[0135] The real-time engine mount load monitoring method operates continuously during vehicle operation and dynamically adjusts the process noise parameters of the Kalman filter based on the engine's operating status. For example, the process noise weight is lowered during high-speed cruising and increased during aggressive driving or rough road conditions to adapt to changes in vibration intensity under different operating conditions. The composite sensing unit also integrates a self-diagnostic module, which monitors the bridge balance of the strain gauges and the resistance range of the temperature sensor to determine in real time whether the sensors are experiencing performance abnormalities. When the self-diagnostic module detects a strain sensor failure, the system automatically switches to an open-loop estimation mode based on acceleration signals, using historical stiffness data and real-time acceleration to maintain basic load estimation capabilities and improve the system's fault tolerance and robustness.

[0136] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention. Therefore, the embodiments should be regarded as exemplary and non-limiting in all respects.

[0137] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A method for real-time monitoring of engine bracket load, characterized in that, Includes the following steps: The original strain signal, real-time temperature signal, and triaxial vibration acceleration signal are synchronously acquired by the composite sensing unit installed at the preset stress concentration position of the engine bracket. Based on the real-time temperature signal, the original strain signal is corrected for elastic modulus and temperature drift is subtracted to obtain the corrected stress value. The corrected stress value is input to the parallel low-pass filter and high-pass filter respectively. The low-frequency quasi-static load component is extracted through the low-pass filter and the high-frequency dynamic alternating load component is extracted through the high-pass filter. The filtered signal is then phase-compensated to obtain the time-aligned low-frequency quasi-static load component and high-frequency dynamic alternating load component. A Kalman filter state-space model is constructed with triaxial vibration acceleration signals as control inputs and corrected stress values ​​as observations. The corrected stress values ​​and triaxial vibration acceleration signals are recursively fused through Kalman filtering to output a real-time load vector containing forces in three translational directions.

2. The method for real-time monitoring of engine bracket load according to claim 1, characterized in that, Based on the real-time temperature signal, the original strain signal is corrected for elastic modulus and temperature drift is subtracted to obtain the corrected stress value, including: Based on a pre-calibrated stiffness-temperature mapping table, the elastic modulus at the current real-time temperature is calculated by querying or interpolation. Multiply the original strain signal by the elastic modulus at the current real-time temperature to obtain the stress value based on the real-time stiffness; Based on a pre-calibrated temperature-stress drift mapping table, the stress drift baseline value at the current real-time temperature is calculated by querying or interpolation. The corrected stress value is obtained by subtracting the stress drift reference value from the stress value based on real-time stiffness.

3. The method for real-time monitoring of engine bracket load according to claim 1, characterized in that, Phase compensation is performed on the filtered signal. Specifically, a bidirectional filtering technique is used. The input signal is passed through the filter in the forward direction, and then the filtered result is passed through the same filter in the reverse direction. The reverse order of the reversed result is taken as the output to eliminate the phase lag introduced by the filter, so that the low-frequency quasi-static load component and the high-frequency dynamic alternating load component of the output are strictly aligned with the original excitation in time.

4. The method for real-time monitoring of engine bracket load according to claim 1, characterized in that, A Kalman filter state-space model is constructed using triaxial vibration acceleration signals as control inputs and corrected stress values ​​as observations, including: Construct a state vector, which includes at least the load forces and their rates of change of the engine mount in three directions in the vehicle body coordinate system; A state transition matrix is ​​constructed based on the discrete-time kinematic equations, and an input control matrix is ​​constructed based on the equivalent mass matrix of the support to convert the triaxial vibration acceleration signal into an input control matrix that affects the load change rate. An observation matrix is ​​constructed based on the stress-load conversion coefficients obtained from the static calibration experiment. The observation matrix is ​​used to map the state vector to an observation vector consisting of corrected stress values.

5. The method for real-time monitoring of engine bracket load according to claim 4, characterized in that, After outputting the real-time load vector, the method further includes: according to the generalized Hooke's law, using the real-time corrected elastic modulus, the preset Poisson's ratio, and the triaxial principal strain obtained by the composite sensing unit, performing multiaxial strain synthesis correction on the real-time load vector to obtain the corrected load vector.

6. The method for real-time monitoring of engine bracket load according to claim 1, characterized in that, The composite sensing unit is installed at a preset stress concentration location on the engine mount. The preset stress concentration location is determined by performing finite element simulation analysis on the engine mount to identify the region with the highest strain energy density under different load combinations.

7. The method for real-time monitoring of engine bracket load according to claim 1, characterized in that, The composite sensing unit includes a strain sensor, a temperature sensor, and an acceleration sensor, all of which are integrated and packaged in a single housing. The housing is made of engineering plastic and filled with thermally conductive and insulating material. A vibration damping structure is also provided between the acceleration sensor mounting base and the housing.

8. The method for real-time monitoring of engine bracket load according to claim 7, characterized in that, The composite sensing unit also integrates a self-diagnostic module, which is used to monitor the bridge balance of the strain sensor and the resistance range of the temperature sensor in real time. When a sensor failure is diagnosed, the system automatically switches to an open-loop estimation mode based on acceleration signals, using historical stiffness data and real-time acceleration signals to maintain basic load estimation functions.

9. The method for real-time monitoring of engine bracket load according to claim 1, characterized in that, The real-time load vector is transmitted to the vehicle electronic control unit via the controller local area network bus for fatigue life assessment or as a control reference input for the active suspension system; the real-time load vector is encrypted before transmission.

10. A real-time engine mount load monitoring system, used to implement the method as described in any one of claims 1 to 9, characterized in that, include: The composite sensing unit is integrated and installed at the preset stress concentration position of the engine bracket to synchronously acquire the original strain signal, real-time temperature signal and triaxial vibration acceleration signal. The signal preprocessing module is used to perform zero-point offset compensation on the triaxial vibration acceleration signal and to synchronously acquire and convert all signals to digital. The temperature compensation module is used to correct the elastic modulus and reduce the temperature drift of the original strain signal based on the real-time temperature signal, and output the corrected stress value. The filtering and decoupling module has a built-in low-pass filter and a high-pass filter connected in parallel to filter and separate the corrected stress value, and perform phase compensation on the filtered signal to output time-aligned low-frequency quasi-static load components and high-frequency dynamic alternating load components. The Kalman filter incorporates a state-space model that uses triaxial vibration acceleration signals as control inputs and corrected stress values ​​as observations. It is used to recursively fuse corrected stress values ​​and triaxial vibration acceleration signals through Kalman filtering to output a real-time load vector.

Images