Active intake grille adaptive control method and system under dynamic perception

Abstract

This invention discloses an active air intake grille adaptive control method and system under dynamic perception, relating to the field of vehicle ventilation control. The method includes: acquiring basic perception data by collecting vehicle state and environmental parameters at a first sampling granularity; analyzing the time-varying characteristics of each parameter in the data to determine the strength of the control relationship between each parameter and the air intake grille control; selecting key perception parameters based on the strength and configuring a second sampling granularity; determining the control tone and control rhythm parameters of the air intake grille based on the secondary perception data; and, based on the tone and parameters, searching for a target opening sequence of the air intake grille with heat dissipation, energy consumption, and mechanical loss as multiple objectives to generate a control strategy. This application solves the problem of mismatch between the time-varying characteristics of physical quantity acquisition behavior and control requirements caused by the lack of perception resource allocation capability in existing air intake grille control systems, achieving real-time matching between perception resource allocation and parameter control importance, and eliminating the time-varying mismatch between acquisition behavior and control requirements.

Description

Technical Field

[0001] This application relates to the field of vehicle ventilation control, and in particular to a method and system for adaptive control of active air intake grilles under dynamic perception. Background Technology

[0002] In vehicle thermal and energy management systems, the control precision and response speed of the air intake grille directly affect the powertrain's heat dissipation efficiency and the vehicle's aerodynamic energy consumption, making it a crucial factor in ensuring vehicle performance, economy, and reliability. Currently, the industry commonly employs control methods based on preset rule tables or fixed thresholds. These methods use several temperature or vehicle speed sensors to collect signals at a constant frequency, combining this with pre-calibrated lookup logic to output grille opening commands. However, when faced with complex and varied real-world driving conditions, the fixed sampling frequency and quantization precision of the sensors fail to adapt to the real-time changes in the importance of different parameters to the control task. This results in delayed updates of critical parameters and redundant data for non-critical parameters under non-steady-state conditions, leading to sluggish control system response or wasted computational resources.

[0003] At present, the control of vehicle air intake grilles suffers from a lack of perception resource allocation capabilities to dynamically identify the importance of parameters, resulting in a mismatch between the acquisition behavior of multi-source physical quantities and the time-varying characteristics of control requirements. Summary of the Invention

[0004] This application provides an active intake grille adaptive control method and system under dynamic perception. It solves the technical problem of existing vehicle intake grille control lacking the ability to dynamically identify the importance of parameters, leading to a mismatch between the time-varying characteristics of multi-source physical quantity acquisition and control requirements. This achieves real-time matching between the allocation of perception resources and the importance of parameter control, thereby eliminating the time-varying mismatch between acquisition behavior and control requirements. The method involves collecting basic perception data at a first sampling granularity, analyzing the time-varying characteristics of each parameter to determine the strength of the control relationship, selecting key perception parameters based on this first sampling granularity, determining the control tone and rhythm parameters based on the second sampling granularity, and then using heat dissipation, energy consumption, and mechanical loss as multi-objective search targets to generate control strategies.

[0005] This application provides an active air intake grille adaptive control method under dynamic perception, comprising: acquiring vehicle state and environmental parameters at a first sampling granularity to obtain basic perception data; analyzing the time change characteristics of each parameter in the basic perception data to determine the control relationship strength between each parameter and the air intake grille control; screening key perception parameters according to the control relationship strength, configuring a second sampling granularity for the key perception parameters, wherein the second sampling granularity is finer than the first sampling granularity; determining the control tone and control rhythm parameters of the air intake grille based on the secondary perception data of the key perception parameters, wherein the control tone reflects the tendency of prioritizing heat dissipation or energy saving, and the control rhythm parameters reflect the control cycle, step size change rate, and response tolerance; and, based on the control tone and the control rhythm parameters, searching for a target opening sequence of the air intake grille with heat dissipation, energy consumption, and mechanical loss as multiple objectives to generate a control strategy for executing adaptive control.

[0006] In a possible implementation, the temporal variation characteristics of each parameter in the basic sensing data are analyzed to determine the control relationship strength between each parameter and the air intake grille control. The following processing is performed: Based on the basic sensing data, a sliding window is used to traverse the time series of each parameter, and dynamic behavior indicators of the corresponding parameter are extracted in each window. The dynamic behavior indicators include the rate of change, fluctuation amplitude, inertia strength, and mutation frequency. According to the dynamic behavior indicators, a correlation analysis strategy is dynamically configured for each parameter: when the rate of change is higher than a first threshold, a short-window time-domain cross-correlation analysis is used; when the inertia strength is higher than a second threshold, the maximum delay range of the Granger causality test is increased; when the fluctuation amplitude is lower than a third threshold and the rate of change is higher than a fourth threshold, frequency-domain cross-correlation analysis is introduced. According to the configured correlation analysis strategy, the temporal correlation strength, information dependence strength, and causal driving strength between each parameter and the air intake grille opening are calculated respectively. The temporal correlation strength, information dependence strength, and causal driving strength are fused and analyzed to obtain the control relationship strength of each parameter.

[0007] In a possible implementation, the control relationship strength between each parameter and the intake grille control is determined, and the following processing is performed: if the rate of change of the parameter is higher than a first threshold and the mutual information exhibits a non-monotonic characteristic of first increasing and then decreasing with the rate of change, then a nonlinear correction coefficient is applied to the control relationship strength to correct it according to the degree to which the current rate of change deviates from the peak range; if the phase difference of the cross power spectrum between the parameter and the grille opening is positive in the frequency domain cross-correlation analysis, then a compensation gain is applied to the control relationship strength according to the lead time length; if the control relationship strength of the parameter is lower than a fifth threshold, a binary combination operation is performed to generate derived features. When the mutual information of the derived features exceeds a set multiple of the source parameter, the mutual information of the derived features and the strength of the source parameter are weighted and fused to obtain the updated control relationship strength.

[0008] In a possible implementation, a second sampling granularity is configured for the key sensing parameters, and the following processing is performed: Based on the control relationship strength value of each key sensing parameter, a sampling frequency multiplier factor is determined proportionally. For every 0.1 increase in control relationship strength, the multiplier factor increases by 0.5-1.0, and the multiplier factor ranges from 2 to 10. The multiplier factor is then multiplied by the reference sampling frequency of the first sampling granularity to obtain the secondary sensing sampling frequency configuration for the parameters. Based on the change rate and noise level of each key sensing parameter, quantization accuracy is configured. When the change rate is higher than a change rate threshold or the signal-to-noise ratio is lower than a noise threshold... When setting the value, the quantization precision is adjusted, increasing the number of bits from the default by 1-4 bits; based on the phase lead time reference of each key sensing parameter, the sampling time offset of the secondary sensing is configured to align the sampling data with the future time domain requirements of the grille control, and the offset value is equal to the phase lead time reference; based on the current control tone, an incremental coefficient is added to the sampling frequency multiplier factor of the key sensing parameters directly related to the tone, and the incremental coefficient ranges from 1.2 to 2.0, wherein the upper limit of the sampling frequency of the final second sampling granularity is limited by the sampling frequency corresponding to the minimum response time of the intake grille drive motor.

[0009] In a possible implementation, the control tone of the air intake grille is determined, and the following processing is performed: based on secondary sensing data of key sensing parameters, the current heat dissipation demand index and energy saving demand index are obtained; when the heat dissipation demand index exceeds a first preset threshold and the ambient temperature is higher than a second preset threshold, the control tone is determined to be a priority heat dissipation tone; when the vehicle speed is higher than a third preset threshold and the battery state of charge is lower than a fourth preset threshold, the control tone is determined to be a priority energy saving tone; when neither the heat dissipation demand index nor the energy saving demand index exceeds its respective threshold, the current control tone remains unchanged; when both the heat dissipation demand index and the energy saving demand index exceed their respective thresholds, the control tone is determined to be a balanced tone, and multi-objective optimization is initiated to determine the target opening degree.

[0010] In a possible implementation, the following processing is performed: the control rhythm parameters include: control period, control step size, and response delay tolerance; wherein, the control period is inversely proportional to the time constant of the fastest changing key sensing parameter; the step size change rate is constrained by the control tone and the second derivative of the corresponding parameter, and the upper limit of the step size is linearly reduced when the second derivative exceeds the threshold; the response delay tolerance is dynamically adjusted according to the prediction confidence interval width of the secondary sensing data, and the step size is reduced and the frequency is increased when the uncertainty is high.

[0011] In a possible implementation, the target opening sequence of the air intake grille is searched using heat dissipation, energy consumption, and mechanical loss as multiple objectives. The following processing is performed: Optimization indicators are determined for heat dissipation, energy consumption, and mechanical loss as multiple objectives. Heat dissipation energy consumption is quantified by the integral of the square of the difference between the coolant temperature and the target temperature; air resistance energy consumption is quantified by the integral of the product of the drag coefficient and the square of the vehicle speed, where the drag coefficient is a function of the grille opening, and a smaller quantification indicator indicates lower air resistance energy consumption; mechanical loss is quantified by the integral of the square of the grille opening change rate, and a smaller quantification indicator indicates smaller actuator movement and less wear; optimization constraints are determined based on the safe upper limit of coolant temperature and the actuator physical limit of the grille opening change rate; weight coefficients for each optimization objective are determined based on the current control tone; and an optimization space is constructed based on each optimization indicator and weight coefficients, combined with the optimization constraints, for searching the target opening sequence of the air intake grille.

[0012] In possible implementations, the weight coefficients of each optimization objective are determined based on the current control tone, and the following processing is performed: if the current control tone is a priority heat dissipation tone, the weight of heat dissipation energy consumption is increased; if the current control tone is a priority energy saving tone, the weight of air resistance energy consumption is increased; if the current control tone is a balance tone, the weights of each objective are evenly distributed.

[0013] In possible implementations, a target opening sequence of the air intake grille is searched to generate a control strategy. A staged coupled control strategy is adopted, and the following processing is performed: The first stage is coarse adjustment approach, which adopts proportional-derivative control, allowing large step size and fast cycle; the second stage is fine adjustment convergence, which switches to proportional-integral-derivative control after the temperature deviation enters the first error band, limits the step size, and introduces a mechanical loss optimization term to suppress opening jitter; the third stage is steady state maintenance, which extends the control cycle and reduces the step size after the temperature deviation enters the second error band and stabilizes. When the deviation in the third stage exceeds the limit, it automatically reverts to the first or second stage, and the revert target is dynamically selected according to the overshoot.

[0014] This application also provides an active air intake grille adaptive control system under dynamic perception, comprising: a basic perception data acquisition module, used to acquire vehicle state and environmental parameters at a first sampling granularity to obtain basic perception data; a control relationship strength determination module, used to analyze the time change characteristics of each parameter in the basic perception data and determine the control relationship strength between each parameter and the air intake grille control; a second sampling granularity configuration module, used to filter key perception parameters according to the control relationship strength and configure a second sampling granularity for the key perception parameters, wherein the second sampling granularity is finer than the first sampling granularity; a control information determination module, used to determine the control tone and control rhythm parameters of the air intake grille based on the secondary perception data of the key perception parameters, wherein the control tone reflects the tendency of prioritizing heat dissipation or energy saving, and the control rhythm parameters reflect the control cycle, step size change rate, and response tolerance; and a control strategy generation module, used to search for the target opening sequence of the air intake grille with heat dissipation, energy consumption, and mechanical loss as multiple objectives according to the control tone and the control rhythm parameters, generate a control strategy for executing adaptive control.

[0015] The proposed method and system for adaptive control of the active air intake grille under dynamic perception, as described in this application, firstly collects vehicle state and environmental parameters at a first sampling granularity to obtain basic perception data. Next, it analyzes the time-varying characteristics of each parameter in the basic perception data to determine the strength of the control relationship between each parameter and the air intake grille control. Then, based on the strength of the control relationship, it selects key perception parameters and configures a second sampling granularity for these key perception parameters. The second sampling granularity is finer than the first sampling granularity. Based on the secondary perception data of the key perception parameters, it determines the control tone and control rhythm parameters for the air intake grille. The control tone reflects a preference for prioritizing heat dissipation or energy saving, and the control rhythm parameters reflect the control cycle, step size change rate, and response tolerance. Finally, based on the control tone and the control rhythm parameters, it searches for a target opening sequence of the air intake grille with heat dissipation, energy consumption, and mechanical loss as multiple objectives, generating a control strategy for adaptive control. Through this process, the method and system proposed in this application achieve the technical effect of real-time matching of the allocation of perception resources with the importance of parameter control, thereby eliminating the time-varying mismatch between data acquisition behavior and control requirements. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings of the embodiments of the present invention will be briefly described below. Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed precisely in sequence. Instead, various steps can be processed in reverse order or simultaneously as needed. Furthermore, other operations can be added to these processes, or one or more steps can be removed from these processes.

[0017] Figure 1 This is a flowchart illustrating the active intake grille adaptive control method under dynamic sensing provided in an embodiment of this application.

[0018] Figure 2 A schematic diagram of the structure of the active intake grille adaptive control system under dynamic sensing provided in the embodiments of this application.

[0019] Explanation of reference numerals in the attached figures: 10 for basic sensing data acquisition module, 20 for control relationship strength determination module, 30 for second sampling granularity configuration module, 40 for regulation information determination module, and 50 for regulation strategy generation module. Detailed Implementation

[0020] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0021] This application provides an active intake grille adaptive control method under dynamic sensing, such as... Figure 1 As shown, the method includes: Step S100: Collect vehicle status and environmental parameters at the first sampling granularity to obtain basic perception data.

[0022] Specifically, the first sampling granularity refers to the sampling period or frequency used when collecting vehicle status and environmental parameters during the routine sensing phase. A timed data acquisition task is configured in the vehicle control unit, setting the baseline sampling period for the first sampling granularity to, for example, 100ms. This value is determined based on the vehicle's standard control cycle and is not the only limitation. The collected vehicle status parameters include at least: grille opening, coolant temperature, ambient temperature, vehicle speed, battery state of charge, engine speed, motor torque, and air conditioning compressor power. The collected environmental parameters include at least: ambient humidity and atmospheric pressure. These parameters are read from the digital signals of the corresponding sensors via the vehicle controller's local area network (LAN) bus. These sensors include, but are not limited to: grille opening position sensor, coolant temperature sensor, ambient temperature sensor, vehicle speed sensor, battery management system, engine control unit, motor control unit, and air conditioning control unit. The control unit stores the collected parameters in a time-series data structure according to timestamps. Each parameter corresponds to an independent one-dimensional array, with the array length being a preset sliding window length, for example, 500 sampling points. The data within the array is arranged in ascending order of sampling time.

[0023] Step S200: Analyze the time variation characteristics of each parameter in the basic sensing data to determine the strength of the control relationship between each parameter and the air intake grille control.

[0024] Specifically, for each parameter in the basic sensing data, the time series array of the parameter and the time series array of the intake grille opening recorded simultaneously are read, and the processing flow from step S210 to step S270 is executed in sequence. Finally, the control relationship strength corresponding to each parameter is output. This strength is a comprehensive quantitative value of the relationship between each parameter and the intake grille opening in three dimensions: time domain correlation, information dependence, and causal driving. It is used to characterize the importance of the parameter to the grille control and is a dimensionless value between 0 and 1.

[0025] In one possible implementation, the temporal variation characteristics of each parameter in the basic sensing data are analyzed to determine the strength of the control relationship between each parameter and the air intake grille control. Step S200 further includes step S210, which, based on the basic sensing data, uses a sliding window to traverse the time series of each parameter and extracts the dynamic behavior indicators of the corresponding parameter in each window. The dynamic behavior indicators include the rate of change, fluctuation amplitude, inertia strength, and frequency of abrupt changes. Specifically, a sliding window is used to traverse the time series array of each parameter. The window width of the sliding window is set to, for example, 20 sampling points, and the sliding step size is 5 sampling points. The window width and sliding step size are determined according to the signal sampling frequency and the typical change period of the analyzed parameter, and are not the only limitations. Within each window, dynamic behavior indicators for the parameter within that window are extracted to describe the local dynamic characteristics of the parameter's time series. These indicators include: rate of change (the absolute value of the slope of the least-squares fitted line of the parameter value within the window is divided by the parameter's rated range of change to obtain the normalized rate of change); fluctuation amplitude (the difference between the maximum and minimum values ​​of the parameter within the window is divided by the parameter's rated range of change to obtain the normalized fluctuation amplitude); inertia strength (the value of the autocorrelation coefficient of the parameter within the window at the first-order delay is calculated; a higher autocorrelation coefficient indicates stronger inertia); and mutation frequency (the number of events where the rate of change between adjacent sampling points of the parameter within the window exceeds a preset mutation threshold, such as a 10% change per second, is counted and divided by the window duration to obtain the mutation frequency per unit time). After the sliding window traversal is completed, each parameter obtains a set of dynamic behavior indicators at each window position.

[0026] Step S220: Based on the dynamic behavior index, dynamically configure the correlation analysis strategy for each parameter: when the rate of change is higher than the first threshold, use short-window time-domain cross-correlation analysis; when the inertia intensity is higher than the second threshold, increase the maximum delay range of the Granger causality test; when the fluctuation amplitude is lower than the third threshold and the rate of change is higher than the fourth threshold, introduce frequency-domain cross-correlation analysis. Specifically, based on the dynamic behavior index calculated in step S210, dynamically configure the correlation analysis strategy for each parameter at the current window position. The configuration rules are as follows: when the rate of change is higher than the first threshold, for example, 0.3, configure short-window time-domain cross-correlation analysis, with the short window length set to 0.5 times the sliding window width; when the inertia intensity is higher than the second threshold, for example, 0.7, configure increasing the maximum delay range of the Granger causality test, increasing the maximum delay range from the default 5 sampling points to 15 sampling points; when the fluctuation amplitude is lower than the third threshold, for example, 0.1 and the rate of change is higher than the fourth threshold, for example, 0.5, configure introducing frequency-domain cross-correlation analysis, which uses fast Fourier transform to calculate the cross-power spectrum. The threshold values ​​mentioned above are for illustrative purposes only and should be calibrated according to the physical characteristics of the parameters during actual deployment. After configuration, each parameter will receive a set of correlation analysis strategy configuration items for each window position.

[0027] Step S230: According to the configured correlation analysis strategy, calculate the temporal correlation strength, information dependency strength, and causal driving strength between each parameter and the air intake grille opening. Specifically, according to the correlation analysis strategy configured in step S220, calculate the temporal correlation strength, information dependency strength, and causal driving strength between each parameter and the air intake grille opening. The temporal correlation strength is calculated as follows: if short-window temporal cross-correlation analysis is configured, the parameter sequence and grille opening sequence within the current window and the forward delay window are extracted, and the absolute value of the Pearson correlation coefficient is calculated as the temporal correlation strength; if short-window temporal cross-correlation analysis is not configured, the temporal correlation strength is calculated using the entire sequence. The information dependency strength is calculated as follows: the parameter sequence and grille opening sequence are discretized with equal width, and the number of discretization intervals is set to, for example, 10. After constructing a joint histogram, the mutual information value is calculated, and the mutual information value is divided by the joint entropy of the parameter and the grille opening to obtain the normalized information dependency strength. The causal driving strength is calculated as follows: using the Granger causality test, an autoregressive model is constructed based on the configured maximum delay range, and the reduction ratio of the prediction error of the grid opening sequence after introducing the parameter sequence is calculated as the causal driving strength.

[0028] Step S240 involves fusing the temporal correlation strength, information dependency strength, and causal driving strength to obtain the control relationship strength for each parameter. Specifically, the temporal correlation strength, information dependency strength, and causal driving strength calculated in step S230 are weighted and fused. The weight coefficients for temporal correlation strength, information dependency strength, and causal driving strength are set to, for example, 0.3, 0.4, and 0.3, respectively, with the sum of the three weight coefficients equal to 1. The fusion calculation method is: Initial value of control relationship strength = Temporal correlation strength × Temporal correlation strength weight coefficient + Information dependency strength × Information dependency strength weight coefficient + Causal driving strength × Causal driving strength weight coefficient. This initial value serves as the preliminary control relationship strength and is output to steps S250, S260, and S270 for correction processing.

[0029] In one possible implementation, determining the control relationship strength between each parameter and the intake grille control, step S200 further includes step S250: if the parameter's rate of change is higher than a first threshold and the mutual information exhibits a non-monotonic characteristic of first increasing and then decreasing with the rate of change, then a nonlinear correction coefficient is applied to the control relationship strength to correct it based on the degree to which the current rate of change deviates from the peak interval. Specifically, for each parameter, it is determined whether its rate of change is higher than a first threshold, for example, 0.3. If it is higher than the first threshold, the characteristics of the mutual information between the parameter and the grille opening changing with the rate of change are further analyzed. Specifically, the rate of change of the parameter is divided into multiple continuous intervals, with interval boundaries for example, 0.3-0.4, 0.4-0.5, 0.5-0.6, and 0.6-0.7, and the width of each interval is 0.1. The mutual information value between the parameter and the grille opening is calculated in each interval, and the mutual information calculation method is the same as the information dependence strength calculation method in step S230. If the mutual information value exhibits a non-monotonic characteristic of first increasing and then decreasing with the rate of change, i.e., there exists a peak interval where the mutual information reaches its maximum value, while the mutual information values ​​in the intervals on both sides are lower than this maximum value, then the non-monotonic correction condition is satisfied. In this case, the position of the current rate of change within the non-monotonic characteristic is obtained: if the current rate of change is within the peak interval of the mutual information, an increased correction coefficient is applied to the initial control relationship strength, for example, multiplied by 1.2; if the current rate of change deviates from the peak interval, the correction coefficient is linearly reduced according to the degree of deviation. The linear reduction rule is: correction coefficient = 1 - deviation distance × 0.1, where the deviation distance is the absolute value of the difference between the current rate of change and the nearest boundary of the peak interval divided by the interval width, and the lower limit of the correction coefficient is set to 0.7. The value obtained by multiplying the initial control relationship strength by the correction coefficient is used as the updated control relationship strength, and this parameter is marked as a high-gain sensitive parameter. For example, if the initial control relationship strength of a certain parameter is set to 0.6 and the rate of change is 0.55, the pre-calculated mutual information change data with the rate of change is shown in Table 1. The current rate of change of 0.55 is within the peak range of 0.5 to 0.6, with a deviation distance of 0. Therefore, the correction factor is taken as 1.2, and the updated control relationship strength is equal to 0.6 × 1.2 = 0.72. This parameter is marked as a high-gain sensitive parameter.

[0030] Table 1. Examples of mutual information changing with rate of change

[0031] In step S260, if the phase difference between the cross-power spectrum of the parameter and the grid opening in the frequency domain cross-correlation analysis is positive, then a compensation gain is applied to the control relationship strength based on the lead time length. Specifically, it is determined whether a frequency domain cross-correlation analysis was performed on the parameter in step S230. If a frequency domain cross-correlation analysis was performed, the cross-power spectrum between the parameter and the grid opening is extracted from the frequency domain cross-correlation analysis results, and the phase angle spectrum of the cross-power spectrum is calculated. The phase difference value is read at the dominant frequency component, which is the phase angle of the parameter relative to the grid opening. If the phase difference value is positive, it indicates that the change of the parameter leads the change of the grid opening, and time alignment compensation is applied to the initial control relationship strength based on the lead time length. The lead time length is calculated as: lead time length = phase difference value ÷ (2π × dominant frequency). The compensation coefficient is calculated as: compensation coefficient = 1 + lead time length / reference time length, where the reference time length is set to 2 seconds for example, and the upper limit of the compensation coefficient is set to 2.0. The updated control relationship strength is equal to the initial control relationship strength multiplied by the compensation coefficient. If the phase difference is negative, it indicates that the parameter change lags behind the grid opening change. In this case, no compensation is performed, or the intensity is appropriately reduced according to the system configuration by multiplying by 0.9. For example, if the initial control relationship intensity for a certain parameter is set to 0.5, and the frequency domain cross-correlation analysis yields a dominant frequency of 0.2Hz and a phase difference of +30°, then the lead time length = 30 / 360 ÷ 0.2 ≈ 0.417 seconds. The reference time length is 2 seconds, the compensation coefficient = 1 + 0.417 / 2 = 1.2085, and the updated control relationship intensity = 0.5 × 1.2085 = 0.60425.

[0032] In step S270, if the control relationship strength of a parameter is lower than the fifth threshold, a binary combination operation is performed to generate derived features. When the mutual information of the derived features exceeds a set multiple of the source parameters, the mutual information of the derived features is weighted and fused with the strength of the source parameters to obtain the updated control relationship strength. Specifically, for parameters whose initial control relationship strength calculated according to step S240 is lower than the fifth threshold, for example, 0.3, a derived feature fusion correction is performed. Derived features refer to the new parameter sequence generated after performing addition, subtraction, multiplication, division, or linear weighting operations on two original parameters. First, a binary combination search is performed on the parameter: all other parameters are traversed, the current parameter is denoted as source parameter A, and the other parameter is denoted as source parameter B. Addition, subtraction, multiplication, division, and linear weighting operations are performed on source parameter A and source parameter B. For example, the weight coefficient combination of the linear weighting operation is source parameter A weight 0.3, source parameter B weight 0.7, source parameter A weight 0.5, source parameter B weight 0.5, and source parameter A weight 0.7, source parameter B weight 0.3. Each operation generates a derived feature sequence. The mutual information between each derived feature and the grid opening is calculated, using the same method as the information dependency strength calculation in step S230. If the mutual information of a derived feature exceeds a set multiple (e.g., 1.5 times) corresponding to the initial control relationship strength of any of its source parameters, then the mutual information of that derived feature is used as an additional strength component. The weighted fusion method is as follows: the updated control relationship strength of the source parameters equals the initial control relationship strength of the source parameters multiplied by its weight (e.g., 0.7) plus the mutual information of the derived features multiplied by an additional weight (e.g., 0.3). Simultaneously, the derived feature and its combination relationship are recorded as candidate key combinations for reference during secondary perception configuration.

[0033] For example, the initial control relationship strength of a source parameter A is set to 0.25, which is lower than the fifth threshold of 0.3, and its mutual information with the grid opening is 0.2. The initial control relationship strength of source parameter B is 0.6, and its mutual information with the grid opening is 0.45. A multiplication operation is performed on source parameters A and B, generating a derived feature sequence equal to source parameter A multiplied by source parameter B. The mutual information between this derived feature and the grid opening is calculated to be 0.42, which is 2.1 times the mutual information of source parameter A (0.2), satisfying the condition of being greater than 1.5 times. Therefore, the mutual information of the derived feature (0.42) is used as an additional strength component. After weighted fusion, the updated control relationship strength of source parameter A is equal to 0.25 × 0.7 + 0.42 × 0.3 = 0.301, and this multiplication combination is recorded as a candidate key combination.

[0034] Steps S250, S260, and S270 can be executed individually or in combination. When multiple correction conditions are met simultaneously, corrections are applied sequentially in the order of steps S250, S260, and S270. Each correction is calculated based on the updated control relationship strength obtained from the previous correction. The final output control relationship strength is the updated value after at least one correction.

[0035] Step S300: Select key sensing parameters based on the strength of the control relationship, and configure a second sampling granularity for the key sensing parameters, wherein the second sampling granularity is finer than the first sampling granularity.

[0036] Specifically, the control relationship strength of each parameter obtained in step S200 is sorted from largest to smallest, and parameters with control relationship strength higher than a sixth threshold, such as 0.5, are selected as key sensing parameters. If no parameter is higher than the sixth threshold, the parameters with the top 30% control relationship strength are selected as key sensing parameters. For each selected key sensing parameter, the sampling granularity configuration in steps S310 to S340 is executed sequentially to configure a sampling period or sampling frequency that is finer than the first sampling granularity for the key sensing parameter.

[0037] In one possible implementation, a second sampling granularity is configured for the key sensing parameters. Step S300 further includes step S310, which determines a sampling frequency multiplier factor proportionally based on the control relationship strength value of each key sensing parameter. For every 0.1 increase in control relationship strength, the multiplier factor increases by 0.5-1.0, and the multiplier factor ranges from 2 to 10. The multiplier factor is multiplied by the reference sampling frequency of the first sampling granularity to obtain the secondary sensing sampling frequency configuration for the parameter. Specifically, the sampling frequency multiplier factor is determined proportionally based on the control relationship strength value of each key sensing parameter. The calculation rule for the multiplier factor is: for every 0.1 increase in control relationship strength, the multiplier factor increases by a value between 0.5 and 1.0. The specific increase is preset according to the parameter type; for example, 0.5 for temperature parameters, 0.8 for speed parameters, and 1.0 for voltage parameters. The range of the multiplier factor is limited to 2 to 10, i.e., the minimum multiplier factor is 2, and the maximum multiplier factor is 10. Once the scaling factor is determined, it is multiplied by the reference sampling frequency of the first sampling granularity to obtain the secondary sensing sampling frequency configuration for that parameter.

[0038] For example, if the reference sampling frequency for the first sampling granularity is 10Hz, and the control relationship strength for a certain key sensing parameter is 0.7, and the control relationship strength increases by a total of 0.7 from 0 to 0.7, with each 0.1 increment representing a 0.8-fold increase, then the multiplier factor is equal to 0.7 ÷ 0.1 × 0.8 = 5.6, rounded down to 6. The sampling frequency for the second sensing is equal to 10 × 6 = 60Hz.

[0039] Step S320: Based on the change rate and noise level of each key sensing parameter, configure the quantization precision. When the change rate is higher than the change rate threshold or the signal-to-noise ratio (SNR) is lower than the noise threshold, adjust the quantization precision by increasing the default bit depth by 1-4 bits. Specifically, the change rate is the current window change rate value calculated in step S210. The noise level is represented by the SNR, which is calculated as follows: perform power spectrum estimation on the parameter sequence. The signal power is the integral of the power spectrum from zero to the cutoff frequency, and the noise power is the integral of the power spectrum in the high-frequency band above the cutoff frequency. Divide the signal power by the noise power, take the logarithm to base 10, and multiply by 10 to obtain the SNR value in dB. Set the change rate threshold, for example, to a change rate of 5% per second, and set the noise threshold, for example, to a SNR below 23 dB. When the change rate is higher than the change rate threshold or the SNR is lower than the noise threshold, adjust the quantization precision from the default bit depth. The default quantization bit depth is, for example, 10 bits. The adjustment method is to increase the bit depth by 1 to 4 bits. The specific bit depth increase is determined by the multiple by which the change rate exceeds the threshold or the magnitude by which the signal-to-noise ratio falls below the threshold: increase by 1 bit for every 1x increase in the change rate above the threshold, and increase by 1 bit for every 3dB decrease in the signal-to-noise ratio below the threshold, up to a maximum of 4 bits.

[0040] For example, the default quantization precision of a key sensing parameter is 10 bits. The current rate of change is 7% per second, the threshold for the rate of change is 5% per second, and any change exceeding the threshold by a factor of 1.4 is rounded down to 1, resulting in a 1-bit increase. The current signal-to-noise ratio is 18 dB, the noise threshold is 23 dB, and any change below the threshold by 5 dB is increased by 1 bit for every 3 dB difference. Overall, this results in a 2-bit increase, configuring the secondary sensing quantization precision to 12 bits.

[0041] Step S330: Based on the phase lead time reference of each key sensing parameter, configure the sampling time offset of the secondary sensing to align the sampled data with the future demand time domain of the grid control. The offset value is equal to the phase lead time reference. Specifically, the phase lead time reference is obtained from step S260. If the phase lead time of this parameter is not calculated in step S260, the default phase lead time reference is 0. The configuration rule for the sampling time offset is: the offset value is equal to the phase lead time reference. If the offset is positive, it means that the sampling time is advanced by the offset time relative to the original sampling time; if the offset is negative or zero, no offset adjustment is performed. The offset is implemented by setting a sampling timer in the control unit, subtracting the offset time from the timer's trigger time as the new trigger time, so that the sampled data is aligned with the future demand time domain of the grid control. For example, if the phase lead time reference of a certain key sensing parameter is 0.417 seconds, then the sampling time offset of the secondary sensing is configured to be 0.417 seconds earlier, that is, each sampling is executed 0.417 seconds before the original timing trigger time.

[0042] Step S340: Based on the current control tone, an incremental coefficient is added to the sampling frequency multiplier factor of key sensing parameters directly related to the tone. The incremental coefficient ranges from 1.2 to 2.0. The upper limit of the sampling frequency for the final second sampling granularity is limited by the sampling frequency corresponding to the minimum response time of the air intake grille drive motor. Specifically, the control tone refers to the tendency of the air intake grille control system to prioritize either heat dissipation or energy saving under the current operating conditions. The control tone is determined by step S400. If step S400 has not yet been executed when step S300 is performed, a default control tone, such as a balanced tone, is used. Key sensing parameters directly related to the priority heat dissipation tone include coolant temperature, ambient temperature, and engine or motor heating-related parameters; key sensing parameters directly related to the priority energy saving tone include vehicle speed, battery state of charge, and grille opening. The incremental coefficient ranges from 1.2 to 2.0, and the specific value is determined based on the intensity of the control tone's tendency; the higher the tendency, the larger the incremental coefficient. The upper limit of the sampling frequency for the final second sampling granularity is limited by the sampling frequency corresponding to the minimum response time of the air intake grille drive motor. This minimum response time is determined by the mechanical time constant of the motor and drive mechanism. For example, if the minimum response time of the drive motor is 50ms, then the maximum sampling frequency is 20Hz. If the sampling frequency calculated in step S310 plus the increment coefficient exceeds this upper limit, then the upper limit value shall prevail.

[0043] For example, if the current control mode prioritizes heat dissipation, then coolant temperature is determined to be a key sensing parameter directly related to this mode, with a scaling factor of 1.8. Multiplying this by the first sampling granularity reference sampling frequency of 10Hz yields 18Hz. The increment coefficient is set to 1.5, resulting in an adjusted sampling frequency of 10 × 1.8 × 1.5 = 27Hz. However, the minimum response time of the drive motor is 50ms, corresponding to an upper limit of 20Hz. 27Hz exceeds this limit, therefore the actual configured sampling frequency is 20Hz.

[0044] Step S400: Based on the secondary sensing data of the key sensing parameters, determine the control tone and control rhythm parameters of the air intake grille. The control tone reflects the tendency of prioritizing heat dissipation or energy saving, and the control rhythm parameters reflect the control cycle, step size change rate, and response tolerance. The control rhythm parameters include: control cycle, control step size, and response delay tolerance. The control cycle is inversely proportional to the time constant of the fastest changing key sensing parameter. The step size change rate is constrained by the control tone and the second derivative of the corresponding parameter; when the second derivative exceeds a threshold, the upper limit of the step size decreases linearly. The response delay tolerance is dynamically adjusted based on the predicted confidence interval width of the secondary sensing data; when uncertainty is high, the step size is reduced and the frequency is increased.

[0045] Specifically, the key sensing parameter data collected at the second sampling granularity configured in step S300 is read and sent to the processing buffer of the control unit. The control rhythm parameters include three components: control period, control step size, and response delay tolerance, used to determine the temporal behavior of the grille opening control. They are determined according to the following rules: the control period is inversely proportional to the time constant of the fastest changing key sensing parameter. The time constant is obtained by fitting a first-order inertial model of the parameter sequence, with a proportionality coefficient set to, for example, 0.5, meaning the control period is equal to 0.5 times the minimum time constant. The step size change rate is constrained by the control tone and the second derivative of the corresponding parameter. First, a basic step size change rate range is set according to the control tone, for example, a basic step size change rate of 5% to 10% per second under a priority heat dissipation tone. Then, the absolute value of the second derivative of the corresponding parameter sequence is calculated. If the absolute value of the second derivative exceeds a preset second derivative threshold, for example, 3% / s... 2 The upper limit of the step size is linearly reduced, and the reduction method is: the corrected upper limit of the step size = the original upper limit of the step size × the second derivative threshold ÷ the absolute value of the second derivative of the parameter sequence. The response delay tolerance is dynamically adjusted according to the predicted confidence interval width of the secondary sensing data. The predicted confidence interval uses an autoregressive moving average model to predict the parameter values ​​of the next 5 sampling points, calculates the standard deviation of the predicted values, and takes the confidence interval width as 1.96 times the standard deviation. The response delay tolerance is negatively correlated with the confidence interval width. For every unit increase in width, the tolerance decreases by 10%, while the control step size decreases by 5% and the control frequency increases by 10%. The specific execution method for determining the control tone of the air intake grille is shown in steps S410 to S450.

[0046] In one possible implementation, after determining the control tone of the air intake grille, step S400 further includes step S410, which involves obtaining the current heat dissipation demand index and energy-saving demand index based on secondary sensing data of key sensing parameters. Specifically, the heat dissipation demand index is a quantified value obtained by weighting the coolant temperature, ambient temperature, and engine or battery heat generation; the energy-saving demand index is a quantified value obtained by weighting the vehicle speed, battery state of charge, and air resistance coefficient. The heat dissipation demand index is calculated as follows: Heat dissipation demand index = |coolant temperature - target coolant temperature| × first weighting coefficient + |ambient temperature - reference ambient temperature| × second weighting coefficient + current heat generation power / rated heat generation power × third weighting coefficient, where the sum of the three weighting coefficients is 1, for example, the first weighting coefficient is 0.5, the second weighting coefficient is 0.3, and the third weighting coefficient is 0.2. The heat generation power is calculated based on engine speed and torque or motor current and voltage. The energy-saving demand index is calculated as follows: Energy-saving demand index = Current vehicle speed / Reference vehicle speed × Fourth weighting coefficient + |Battery state of charge - Ideal state of charge| / Ideal state of charge × Fifth weighting coefficient + |Current air drag coefficient - Minimum air drag coefficient| × Sixth weighting coefficient. The sum of the three weighting coefficients is 1. For example, the fourth weighting coefficient is 0.4, the fifth weighting coefficient is 0.4, and the sixth weighting coefficient is 0.2. The air drag coefficient is a function of the grille opening; the smaller the grille opening, the smaller the air drag coefficient.

[0047] In step S420, when the heat dissipation demand index exceeds a first preset threshold and the ambient temperature is higher than a second preset threshold, the control tone is determined to be a priority heat dissipation tone. Specifically, it is determined whether the heat dissipation demand index exceeds the first preset threshold, for example, 0.7, and simultaneously whether the ambient temperature is higher than the second preset threshold, for example, 35°C. If both conditions are met, the control tone is determined to be a priority heat dissipation tone. After determination, a control tone flag is output as priority heat dissipation, and this flag is used for weight configuration in steps S340 and S530.

[0048] Step S430: When the vehicle speed is higher than the third preset threshold and the battery state of charge is lower than the fourth preset threshold, the control mode is determined to be the priority energy-saving mode. Specifically, it is determined whether the vehicle speed is higher than the third preset threshold, for example, 80 kilometers per hour, and simultaneously whether the battery state of charge is lower than the fourth preset threshold, for example, 30%. If both conditions are met, the control mode is determined to be the priority energy-saving mode. After determination, the control mode flag is output as priority energy saving.

[0049] Step S440: When neither the heat dissipation demand index nor the energy saving demand index exceeds their respective thresholds, the current control tone remains unchanged. Specifically, it is determined whether neither the heat dissipation demand index nor the energy saving demand index exceeds their respective thresholds. The threshold corresponding to the heat dissipation demand index is a first preset threshold, and the threshold corresponding to the energy saving demand index is a fifth preset threshold, for example, 0.6. If the heat dissipation demand index is lower than or equal to the first preset threshold and the energy saving demand index is lower than or equal to the fifth preset threshold, the current control tone remains unchanged. If there is no effective control tone at present, it is set to the default balanced tone.

[0050] In step S450, when both the heat dissipation demand index and the energy saving demand index exceed their respective thresholds, the control tone is determined to be balanced, and multi-objective optimization is initiated to determine the target opening degree. Specifically, it is determined whether the heat dissipation demand index exceeds the first preset threshold and whether the energy saving demand index simultaneously exceeds the fifth preset threshold. If both conditions are met, the control tone is determined to be balanced, and multi-objective optimization is initiated to determine the target opening degree. Under the balanced tone, the multi-objective optimization is initiated by calling the optimization process in step S500 and setting the weights of each optimization objective to a balanced distribution, for example, a weight of 0.34 for heat dissipation energy consumption, a weight of 0.33 for air resistance energy consumption, and a weight of 0.33 for mechanical loss.

[0051] Step S500: Based on the control tone and the control rhythm parameters, a control strategy is generated by searching for the target opening sequence of the air intake grille with heat dissipation, energy consumption, and mechanical loss as multiple objectives, and then used to perform adaptive control.

[0052] Specifically, the control tone flag and control rhythm parameters determined in step S400 are used as inputs. Combined with the secondary sensing data of the current key sensing parameters, an optimization problem is constructed and the target opening sequence is solved. The solution is a set of grid opening setpoints over a time series, which is output to the actuator drive module. The specific optimization indicators, constraints, and search methods are set in steps S510 to S540.

[0053] In one possible implementation, the target opening sequence of the air intake grille is searched using heat dissipation, energy consumption, and mechanical loss as multiple objectives. Step S500 further includes step S510, which determines optimization indicators for heat dissipation, energy consumption, and mechanical loss as multiple objectives. Specifically, heat dissipation energy consumption is quantified by the integral of the square of the difference between the coolant temperature and the target temperature; air resistance energy consumption is quantified by the integral of the product of the drag coefficient and the square of the vehicle speed, where the drag coefficient is a function of the grille opening, and a smaller quantification indicator indicates lower air resistance energy consumption; mechanical loss is quantified by the integral of the square of the rate of change of the grille opening, and a smaller quantification indicator indicates smaller actuator movement and less wear. Specifically, the quantification indicators for each objective are determined using heat dissipation, energy consumption, and mechanical loss as multiple objectives. The quantification indicator for heat dissipation energy consumption is the integral of the square of the difference between the coolant temperature and the target temperature in the prediction time domain, with the integration method being the accumulation of discrete sampling points, i.e., heat dissipation energy consumption indicator = ∑(coolant temperature - target coolant temperature). 2 × Sampling time interval. The quantitative index of air resistance energy consumption is the integral of the product of drag coefficient and vehicle speed squared over the prediction time domain, that is, air resistance energy consumption index = ∑ (drag coefficient × vehicle speed) 2 The coefficient of motion is calculated as () × sampling time interval, where the drag coefficient is a function of the grille opening. This function is pre-calibrated into an interpolation table through wind tunnel experiments. The input is the grille opening, and the output is the drag coefficient. A smaller quantification index indicates lower air resistance and energy consumption. The mechanical loss index is calculated as the integral of the square of the rate of change of the grille opening over the prediction time domain, i.e., Mechanical Loss Index = ∑[(Current Opening - Previous Opening) / Sampling Time Interval] 2 × Sampling time interval, the smaller this quantitative indicator, the smaller the movement range of the actuator and the less wear.

[0054] Step S520: Determine optimization constraints based on the safe upper limit of coolant temperature and the actuator physical limit of the grille opening change rate. Specifically, the safe upper limit of coolant temperature is set to, for example, 110°C, and the predicted coolant temperature value generated by any candidate opening sequence during the optimization process must not exceed this upper limit. The actuator physical limit of the grille opening change rate includes a single-step maximum change amplitude limit and a unit-time maximum change rate limit, for example, a single-step maximum change amplitude of 10% of the full stroke and a unit-time maximum change rate of 30% of the full stroke. During the optimization process, the absolute value of the opening change between adjacent sampling points in the candidate opening sequence must not exceed the single-step maximum change amplitude, and the cumulative change rate of multiple consecutive steps must not exceed the unit-time maximum change rate.

[0055] Step S530: Determine the weight coefficients of each optimization objective based on the current control tone. Based on each optimization index and its weight coefficients, and in conjunction with the optimization constraints, construct an optimization space for searching the target opening sequence of the air intake grille. Specifically, if the current control tone is a priority heat dissipation tone, increase the weight of heat dissipation energy consumption; if the current control tone is a priority energy saving tone, increase the weight of air resistance energy consumption; if the current control tone is a balanced tone, the weights of each objective are evenly distributed. Specifically, if the current control tone is a priority heat dissipation tone, set the weight of heat dissipation energy consumption to, for example, 0.6, the weight of air resistance energy consumption to, for example, 0.2, and the weight of mechanical loss to, for example, 0.2, with the sum of the three weights being 1. If the current control tone is a priority energy saving tone, set the weight of heat dissipation energy consumption to, for example, 0.2, the weight of air resistance energy consumption to, for example, 0.6, and the weight of mechanical loss to, for example, 0.2. If the current control tone is a balanced tone, set the weight of heat dissipation energy consumption to, for example, 0.34, the weight of air resistance energy consumption to, for example, 0.33, and the weight of mechanical loss to, for example, 0.33. After determining the weighting coefficients, a weighted sum-form optimization objective function is constructed, namely, Optimization Objective = Heat Dissipation Energy Consumption Weight × Heat Dissipation Energy Consumption Index + Air Resistance Energy Consumption Weight × Air Resistance Energy Consumption Index + Mechanical Loss Weight × Mechanical Loss Index. Combining the optimization constraints determined in step S520, a complete optimization space is constructed. The rolling time-domain optimization method in dynamic programming or model predictive control is then used to search for the grid opening sequence in the optimization space that minimizes the optimization objective.

[0056] In one possible implementation, the target opening sequence of the air intake grille is searched to generate a control strategy. Step S500 further includes step S540, which adopts a phased coupled control strategy, including: the first stage is coarse adjustment approach, which adopts proportional-derivative control and allows large step size and fast cycle; the second stage is fine adjustment convergence, which switches to proportional-integral-derivative control after the temperature deviation enters the first error band, limits the step size and introduces a mechanical loss optimization term to suppress opening jitter; the third stage is steady state maintenance, which extends the control cycle and reduces the step size after the temperature deviation enters the second error band and stabilizes. When the deviation in the third stage exceeds the limit, it automatically reverts to the first stage or the second stage, and the revert target is dynamically selected according to the overshoot.

[0057] Specifically, a phased coupled control strategy is employed to search for and generate the target opening sequence of the air intake grille. The first phase is the coarse-adjustment approach phase: proportional-derivative control is used, with the proportional coefficient set to, for example, 0.8 and the derivative coefficient set to, for example, 0.2. In this phase, the step size change rate is allowed to reach 80% to 100% of the maximum set value, and the control period is 0.5 to 0.8 times the reference period, for example, 100ms, prioritizing response speed. The second phase is the fine-tuning convergence phase: when the temperature deviation enters the first error band, for example, ±3℃ and the deviation change rate is lower than the preset threshold, for example, 0.5℃ per second, it switches to proportional-integral-derivative control, with the proportional coefficient set to, for example, 0.3, the integral coefficient set to, for example, 0.1, and the derivative coefficient set to, for example, 0.15. Simultaneously, the step size change rate is limited to 30% to 50% of the maximum set value, the control cycle is restored to the reference cycle of 100ms, and a mechanical loss optimization term is introduced in this stage. The mechanical loss optimization term is the square of the opening change rate multiplied by a penalty coefficient, which is set to, for example, 0.05, to actively suppress the reciprocating jitter of the opening and prioritize control accuracy and actuator life. The third stage is the steady-state maintenance and monitoring stage: when the temperature deviation enters the second error band, for example ±1℃ and the duration exceeds the preset stabilization time, for example 5 seconds, the control cycle is extended to 1.5-2.0 times the reference cycle, i.e., 150ms to 200ms, the step size change rate is limited to less than 10% of the maximum set value, and the deviation is continuously monitored to see if it exceeds the third error band, for example ±2.5℃.

[0058] The switching conditions between stages include: switching from stage one to stage two requires the absolute value of the deviation change rate to be lower than a preset threshold (e.g., 0.2°C per second) for three consecutive cycles and the absolute value of the deviation to be lower than the upper limit of the first error band by 3°C; switching from stage two to stage three requires the absolute value of the deviation to be lower than the upper limit of the second error band by 1°C for five consecutive cycles and the change in control output to be lower than a preset threshold (e.g., 1%) for three consecutive cycles throughout the entire stroke. When stage three detects a deviation exceeding the third error band by 2.5°C, it automatically reverts to stage one or stage two. The revert target is dynamically selected based on the overshoot magnitude: overshoot = (current deviation - third error band limit) ÷ third error band limit × 100%. If the overshoot exceeds a set value (e.g., 50%), it reverts to stage one; otherwise, it reverts to stage two.

[0059] For example, with the initial coolant temperature set at 95℃ and the target temperature at 90℃, the deviation is 5℃, exceeding the first error band by 3℃, and the system is in the first stage. Proportional-derivative (PD) control is executed, with the step size change rate set to 90% of the maximum setpoint and the control period set to 60ms. After 10 cycles, the coolant temperature drops to 92.5℃, with a deviation of 2.5℃ and a deviation change rate of 0.15℃ per second. For three consecutive cycles, the deviation change rate is below 0.2℃ per second and the absolute value of the deviation is below 3℃, at which point the system switches to the second stage. The second stage uses proportional-integral-derivative (PD) control, with the step size change rate limited to 40% of the maximum setpoint, and the control period returns to 100ms. After further adjustment, the coolant temperature stabilizes within the range of 90.5℃ ± 0.5℃ for 6 seconds, satisfying the condition that the absolute value of the deviation is below 1℃ for five consecutive cycles and the change in control output is below 1% for three consecutive cycles, at which point the system switches to the third stage. The third stage extends the control period to 180ms, with the step size change rate limited to 8% of the maximum setpoint. If the coolant temperature rises to 93℃ due to external disturbances, and the deviation exceeds the third error band by 2.5℃, the overshoot is 20%, which is less than the 50% judgment threshold. Therefore, it will automatically revert to the second stage for further adjustment.

[0060] This application's embodiments solve the technical problem of existing vehicle air intake grille control lacking the ability to dynamically identify the importance of parameters in the allocation of sensing resources, leading to a mismatch between the time-varying characteristics of the acquisition behavior of multi-source physical quantities and the control requirements. This is achieved by collecting basic sensing data at a first sampling granularity, analyzing the time-varying characteristics of each parameter to determine the strength of the control relationship, screening key sensing parameters and configuring a finer second sampling granularity based on the second sensing data, determining the control tone and control rhythm parameters based on the second sensing data, and generating control strategies by using heat dissipation, energy consumption, and mechanical loss as multi-target search target opening sequences.

[0061] In the above text, refer to Figure 1 The active air intake grille adaptive control method under dynamic sensing according to embodiments of the present invention is described in detail. Next, reference will be made to... Figure 2 This invention describes an active intake grille adaptive control system under dynamic sensing according to an embodiment of the present invention.

[0062] The adaptive control system for active air intake grilles under dynamic perception, according to an embodiment of the present invention, addresses the technical problem of existing vehicle air intake grille control systems lacking the ability to dynamically identify the importance of parameters in terms of perception resource allocation. This leads to a mismatch between the time-varying characteristics of multi-source physical quantity acquisition and control requirements. The system achieves real-time matching between the allocation of perception resources and the importance of parameter control, thereby eliminating the time-varying mismatch between acquisition behavior and control requirements. The adaptive control system for active air intake grilles under dynamic perception includes: a basic perception data acquisition module 10, a control relationship strength determination module 20, a second sampling granularity configuration module 30, a control information determination module 40, and a control strategy generation module 50.

[0063] The system comprises: a basic perception data acquisition module 10, used to acquire vehicle state and environmental parameters at a first sampling granularity to obtain basic perception data; a control relationship strength determination module 20, used to analyze the time change characteristics of each parameter in the basic perception data to determine the control relationship strength between each parameter and the air intake grille control; a second sampling granularity configuration module 30, used to filter key perception parameters according to the control relationship strength and configure a second sampling granularity for the key perception parameters, wherein the second sampling granularity is finer than the first sampling granularity; a regulation information determination module 40, used to determine the regulation tone and regulation rhythm parameters of the air intake grille based on the secondary perception data of the key perception parameters, wherein the regulation tone reflects the tendency of prioritizing heat dissipation or energy saving, and the regulation rhythm parameters reflect the regulation cycle, step size change rate, and response tolerance; and a regulation strategy generation module 50, used to search for the target opening sequence of the air intake grille with heat dissipation, energy consumption, and mechanical loss as multiple objectives according to the regulation tone and the regulation rhythm parameters, generate a regulation strategy for executing adaptive regulation.

[0064] The detailed description of the specific configuration of the control relationship strength determination module 20 is explained as follows: As mentioned above, the module analyzes the time change characteristics of each parameter in the basic sensing data to determine the control relationship strength between each parameter and the air intake grille control. The control relationship strength determination module 20 may further include: a dynamic behavior index extraction unit, which uses a sliding window to traverse the time series of each parameter based on the basic sensing data and extracts the dynamic behavior index of the corresponding parameter in each window. The dynamic behavior index includes the rate of change, fluctuation amplitude, inertia strength, and mutation frequency; an association analysis strategy configuration unit, which dynamically configures an association analysis strategy for each parameter according to the dynamic behavior index: when the rate of change is higher than a first threshold, a short-window time-domain cross-correlation analysis is used; when the inertia strength is higher than a second threshold, the maximum delay range of the Granger causality test is increased; when the fluctuation amplitude is lower than a third threshold and the rate of change is higher than a fourth threshold, frequency-domain cross-correlation analysis is introduced; an intensity calculation unit, which calculates the time-domain association strength, information dependence strength, and causal driving strength between each parameter and the air intake grille opening according to the configured association analysis strategy; and a fusion analysis unit, which performs fusion analysis on the time-domain association strength, information dependence strength, and causal driving strength to obtain the control relationship strength of each parameter.

[0065] The control relationship strength determination module 20, which determines the control relationship strength between each parameter and the air intake grille control, may further include: a control relationship strength correction unit, which, if the rate of change of a parameter is higher than a first threshold and the mutual information exhibits a non-monotonic characteristic of first increasing and then decreasing with the rate of change, applies a nonlinear correction coefficient to the control relationship strength based on the degree to which the current rate of change deviates from the peak range; a control relationship strength compensation gain unit, which, if the phase difference between the cross power spectrum of the parameter and the grille opening in the frequency domain cross-correlation analysis is positive, applies a compensation gain to the control relationship strength based on the lead time length; and a control relationship strength update unit, which, if the control relationship strength of a parameter is lower than a fifth threshold, performs a binary combination operation to generate derived features, and when the mutual information of the derived features exceeds a set multiple of the source parameter, weightedly fuses the mutual information of the derived features with the strength of the source parameter to obtain the updated control relationship strength.

[0066] The detailed description of the specific configuration of the second sampling granularity configuration module 30 is explained as follows: As mentioned above, the second sampling granularity is configured for the key sensing parameters. The second sampling granularity configuration module 30 may further include: a secondary sensing sampling frequency configuration unit, used to determine a sampling frequency multiplier factor proportionally based on the control relationship strength value of each key sensing parameter. For every 0.1 increase in control relationship strength, the multiplier factor increases by 0.5-1.0, and the multiplier factor ranges from 2 to 10. The multiplier factor is multiplied by the reference sampling frequency of the first sampling granularity to obtain the secondary sensing sampling frequency configuration of the parameter; and a quantization accuracy configuration unit, used to configure the quantization accuracy based on the change rate and noise level of each key sensing parameter. When the quantization rate is higher than the rate of change threshold or the signal-to-noise ratio is lower than the noise threshold, the quantization accuracy is adjusted, increasing the default number of bits by 1-4 bits. The sampling time offset configuration unit is used to configure the sampling time offset of the secondary sensing based on the phase lead time reference of each key sensing parameter, so that the sampling data is aligned with the future time domain requirements of the grille control. The offset value is equal to the phase lead time reference. The incremental coefficient addition unit is used to add an incremental coefficient to the sampling frequency multiplier factor of the key sensing parameters directly related to the current control tone, based on the current control tone. The incremental coefficient ranges from 1.2 to 2.0. The upper limit of the sampling frequency of the final second sampling granularity is limited by the sampling frequency corresponding to the minimum response time of the intake grille drive motor.

[0067] The detailed description of the specific configuration of the control information determination module 40 is explained as follows: As mentioned above, to determine the control tone of the air intake grille, the control information determination module 40 may further include: a demand index acquisition unit for acquiring current heat dissipation demand index and energy saving demand index based on secondary perception data of key perception parameters; a control tone determination unit for determining the control tone as a priority heat dissipation tone when the heat dissipation demand index exceeds a first preset threshold and the ambient temperature is higher than a second preset threshold; determining the control tone as a priority energy saving tone when the vehicle speed is higher than a third preset threshold and the battery state of charge is lower than a fourth preset threshold; maintaining the current control tone unchanged when neither the heat dissipation demand index nor the energy saving demand index exceeds their respective thresholds; and determining the control tone as a balanced tone and initiating multi-objective optimization to determine the target opening degree when both the heat dissipation demand index and the energy saving demand index exceed their respective thresholds.

[0068] The control information determination module 40 may further include: the control rhythm parameters include: control period, control step size, and response delay tolerance; wherein, the control period is inversely proportional to the time constant of the fastest changing key sensing parameter; the step size change rate is constrained by the control tone and the second derivative of the corresponding parameter, and the upper limit of the step size is linearly reduced when the second derivative exceeds the threshold; the response delay tolerance is dynamically adjusted according to the prediction confidence interval width of the secondary sensing data, and the step size is reduced and the frequency is increased when the uncertainty is high.

[0069] The detailed description of the specific configuration of the control strategy generation module 50 is explained below: As mentioned above, the control strategy generation module 50 further includes an optimization index determination unit for determining optimization indexes based on multiple objectives, namely heat dissipation, energy consumption, and mechanical loss. Specifically, heat dissipation energy consumption is quantified by the integral of the square of the difference between the coolant temperature and the target temperature; air resistance energy consumption is quantified by the integral of the product of the drag coefficient and the square of the vehicle speed, where the drag coefficient is a function of the grille opening. A higher quantification index results in a higher energy consumption. The smaller the value, the lower the air resistance energy consumption; mechanical loss is quantified by the integral of the square of the grille opening change rate. The smaller the quantitative index of mechanical loss, the smaller the action range of the actuator and the less wear. The constraint condition determination unit is used to determine the optimization constraint conditions based on the safe upper limit of coolant temperature and the actuator physical limit of grille opening change rate; the optimization space construction unit is used to determine the weight coefficient of each optimization objective based on the current control tone, and construct the optimization space based on each optimization index and weight coefficient, combined with the optimization constraint conditions, for searching the target opening sequence of the air intake grille.

[0070] Among them, the weight coefficients of each optimization objective are determined according to the current control tone, and the optimization space construction unit may further include: increasing the weight of heat dissipation energy consumption when the current control tone is a priority heat dissipation tone; increasing the weight of air resistance energy consumption when the current control tone is a priority energy saving tone; and balancing the weights of each objective when the current control tone is a balance tone.

[0071] The search for the target opening sequence of the air intake grille and the generation of the control strategy can be further included in the control strategy generation module 50, which adopts a phased coupled control strategy, including: the first stage is coarse adjustment approach, which adopts proportional-derivative control and allows large step size and fast cycle; the second stage is fine adjustment convergence, which switches to proportional-integral-derivative control after the temperature deviation enters the first error band, limits the step size and introduces a mechanical loss optimization term to suppress opening jitter; the third stage is steady state maintenance, which extends the control cycle and reduces the step size after the temperature deviation enters the second error band and stabilizes. When the deviation in the third stage exceeds the limit, it automatically reverts to the first stage or the second stage, and the revert target is dynamically selected according to the overshoot.

[0072] The adaptive control system for the active air intake grille under dynamic sensing provided in the embodiments of the present invention can execute the adaptive control method for the active air intake grille under dynamic sensing provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the method.

[0073] Although this application makes various references to certain modules in the system according to the embodiments of this application, any number of different modules can be used and run on user terminals and / or servers. The various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy distinction between each other and are not used to limit the scope of protection of this invention.

[0074] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for adaptive control of an active intake grille under dynamic sensing, characterized in that, include: Vehicle status and environmental parameters are collected at the first sampling granularity to obtain basic perception data; Analyze the time variation characteristics of each parameter in the basic sensing data to determine the strength of the control relationship between each parameter and the air intake grille control; Key sensing parameters are selected based on the strength of the control relationship, and a second sampling granularity is configured for the key sensing parameters, wherein the second sampling granularity is finer than the first sampling granularity; Based on the secondary sensing data of the key sensing parameters, the control tone and control rhythm parameters of the air intake grille are determined. The control tone reflects the tendency of prioritizing heat dissipation or energy saving, and the control rhythm parameters reflect the control cycle, step size change rate and response tolerance. Based on the control tone and the control rhythm parameters, a control strategy is generated by searching for the target opening sequence of the air intake grille with heat dissipation, energy consumption, and mechanical loss as multiple objectives, and then performing adaptive control.

2. The active intake grille adaptive control method under dynamic sensing according to claim 1, characterized in that, Analyze the time variation characteristics of each parameter in the basic sensing data to determine the strength of the control relationship between each parameter and the air intake grille control, including: Based on the aforementioned basic sensing data, a sliding window is used to traverse the time series of each parameter, and dynamic behavior indicators of the corresponding parameter are extracted in each window. The dynamic behavior indicators include rate of change, fluctuation amplitude, inertial strength, and frequency of sudden changes. Based on the dynamic behavior indicators, a correlation analysis strategy is dynamically configured for each parameter: when the rate of change is higher than the first threshold, a short-window time-domain cross-correlation analysis is used; when the inertia intensity is higher than the second threshold, the maximum delay range of the Granger causality test is increased; and when the fluctuation amplitude is lower than the third threshold and the rate of change is higher than the fourth threshold, frequency-domain cross-correlation analysis is introduced. According to the configured correlation analysis strategy, the temporal correlation strength, information dependence strength and causal driving strength between each parameter and the air intake grille opening are calculated respectively. The control relationship strength of each parameter is obtained by fusing and analyzing the temporal correlation strength, information dependence strength, and causal driving strength.

3. The active intake grille adaptive control method under dynamic sensing according to claim 2, characterized in that, Determining the strength of the control relationship between each parameter and the air intake grille control also includes: If the rate of change of the parameter is higher than the first threshold and the mutual information exhibits a non-monotonic characteristic of first increasing and then decreasing with the rate of change, then a nonlinear correction coefficient is applied to the strength of the control relationship to correct it according to the degree to which the current rate of change deviates from the peak range. If the phase difference between the cross power spectrum of the parameters and the grid opening in the frequency domain cross-correlation analysis is positive, then a compensation gain is applied to the strength of the control relationship based on the lead time length. If the control relationship strength of the parameter is lower than the fifth threshold, a binary combination operation is performed to generate derived features. When the mutual information of the derived features exceeds the set multiple of the source parameter, the mutual information of the derived features and the strength of the source parameter are weighted and fused to obtain the updated control relationship strength.

4. The active intake grille adaptive control method under dynamic sensing according to claim 1, characterized in that, Configure a second sampling granularity for the key sensing parameters, including: Based on the control relationship strength value of each key sensing parameter, the sampling frequency multiplier factor is determined proportionally. For every 0.1 increase in control relationship strength, the multiplier factor increases by 0.5-1.0, and the multiplier factor ranges from 2 to 10. The multiplier factor is multiplied by the reference sampling frequency of the first sampling granularity to obtain the secondary sensing sampling frequency configuration of the parameter. Based on the rate of change and noise level of each key sensing parameter, the quantization accuracy is configured. When the rate of change is higher than the rate of change threshold or the signal-to-noise ratio is lower than the noise threshold, the quantization accuracy is adjusted, which is increased by 1-4 bits from the default bit depth. Based on the phase lead time reference of each key sensing parameter, the sampling time offset of the secondary sensing is configured so that the sampling data is aligned with the future time domain requirements of the grid control. The offset value is equal to the phase lead time reference. Based on the current control tone, an incremental coefficient is added to the sampling frequency multiplier factor of key sensing parameters directly related to the tone. The incremental coefficient ranges from 1.2 to 2.

0. The upper limit of the sampling frequency of the final second sampling granularity is limited by the sampling frequency corresponding to the minimum response time of the air intake grille drive motor.

5. The method for adaptive control of the active intake grille under dynamic sensing according to claim 1, characterized in that, Determine the overall tone for the grille's design, including: Based on secondary sensing data of key sensing parameters, current heat dissipation demand indicators and energy saving demand indicators are obtained. When the heat dissipation demand index exceeds the first preset threshold and the ambient temperature is higher than the second preset threshold, the control tone will be determined as the priority heat dissipation tone. When the vehicle speed is higher than the third preset threshold and the battery state of charge is lower than the fourth preset threshold, the control mode will be determined to be the priority energy-saving mode. When neither the heat dissipation demand index nor the energy saving demand index exceeds its respective threshold, the current regulatory tone will remain unchanged. When both the heat dissipation demand index and the energy saving demand index exceed their respective thresholds, the control tone will be determined to be balanced, and multi-objective optimization will be initiated to determine the target opening degree.

6. The method for adaptive control of the active intake grille under dynamic sensing according to claim 1, characterized in that, The control rhythm parameters include: control period, control step size, and response delay tolerance. The control period is inversely proportional to the time constant of the fastest changing key sensing parameter; the step size change rate is constrained by the control tone and the second derivative of the corresponding parameter, and the upper limit of the step size is linearly reduced when the second derivative exceeds the threshold; the response delay tolerance is dynamically adjusted according to the width of the prediction confidence interval of the secondary sensing data, and the step size is reduced and the frequency is increased when the uncertainty is high.

7. The method for adaptive control of the active intake grille under dynamic sensing according to claim 1, characterized in that, The target opening sequence of the air intake grille is searched based on multiple objectives, including heat dissipation, energy consumption, and mechanical loss, and includes: With heat dissipation, energy consumption, and mechanical loss as multiple objectives, optimization indicators were determined. Among them, heat dissipation energy consumption was quantified by the integral of the square of the difference between the coolant temperature and the target temperature; air resistance energy consumption was quantified by the integral of the product of the drag coefficient and the square of the vehicle speed, where the drag coefficient is a function of the grille opening, and the smaller the quantification indicator, the lower the air resistance energy consumption; mechanical loss was quantified by the integral of the square of the rate of change of the grille opening, and the smaller the quantification indicator of mechanical loss, the smaller the movement range of the actuator and the less wear. The optimal constraint conditions are determined based on the safe upper limit of coolant temperature and the physical limit of actuator for the rate of change of grille opening. The weight coefficients of each optimization objective are determined based on the current regulatory tone. Based on each optimization index and the weight coefficients, and in conjunction with the optimization constraints, an optimization space is constructed for searching the target opening sequence of the air intake grille.

8. The active intake grille adaptive control method under dynamic sensing according to claim 7, characterized in that, The weighting coefficients for each optimization objective are determined based on the current regulatory tone, including: If the current regulatory tone is to prioritize heat dissipation, the weight of heat dissipation energy consumption will be increased; if the current regulatory tone is to prioritize energy conservation, the weight of air resistance energy consumption will be increased; if the current regulatory tone is to maintain balance, the weights of each target will be evenly distributed.

9. The method for adaptive control of the active intake grille under dynamic sensing according to claim 1, characterized in that, The system searches for the target opening sequence of the air intake grille, generates a control strategy, and employs a staged coupled control strategy, including: The first stage is coarse-scale approach, which uses proportional-derivative control and allows for large step sizes and fast cycles. The second stage is fine-tuning convergence. After the temperature deviation enters the first error band, it switches to proportional-integral-derivative control, limits the step size and introduces a mechanical loss optimization term to suppress opening jitter. The third stage is steady-state maintenance. After the temperature deviation enters the second error band and stabilizes, the control cycle is extended and the step size is reduced. When the deviation in the third stage exceeds the limit, it automatically reverts to the first or second stage. The revert target is dynamically selected according to the overshoot.

10. An active intake grille adaptive control system under dynamic sensing, characterized in that, The system is used to implement the active intake grille adaptive control method under dynamic sensing as described in any one of claims 1-9, the system comprising: The basic perception data acquisition module is used to collect vehicle status and environmental parameters at the first sampling granularity to obtain basic perception data. The control relationship strength determination module is used to analyze the time change characteristics of each parameter in the basic sensing data and determine the control relationship strength between each parameter and the air intake grille control. The second sampling granularity configuration module is used to filter key sensing parameters according to the control relationship strength and configure a second sampling granularity for the key sensing parameters, wherein the second sampling granularity is finer than the first sampling granularity. The control information determination module is used to determine the control tone and control rhythm parameters of the air intake grille based on the secondary sensing data of the key sensing parameters. The control tone reflects the tendency of prioritizing heat dissipation or energy saving, and the control rhythm parameters reflect the control cycle, step size change rate and response tolerance. The control strategy generation module is used to generate a control strategy based on the control tone and the control rhythm parameters, using heat dissipation, energy consumption and mechanical loss as multiple objectives to search for the target opening sequence of the air intake grille, and to perform adaptive control.

Images