A tractor brake heat fade prediction method based on multi-source sensor fusion

Abstract

The application discloses a tractor brake thermal recession prediction method based on multi-source sensor fusion, and belongs to the technical field of agricultural machinery brake control. It comprises the following steps: identifying the brake working condition of the current tractor based on multi-source state sensing data, and dividing the current brake process into multiple brake working condition segments according to the brake working condition; determining the fusion weight of each state sensing data according to the brake working condition segment, and carrying out fusion processing on each state sensing data according to the fusion weight to obtain fused state data; extracting thermal characteristic parameters affecting the brake thermal state change based on the fused state data, and estimating the current thermal state of the brake based on the thermal characteristic parameters to obtain brake thermal state representation results; and predicting the temperature change trend and the friction coefficient attenuation trend of the brake within a preset time window based on the brake thermal state representation results, and determining the thermal recession risk level corresponding to the brake according to the predicted friction coefficient attenuation trend.

Description

Technical Field

[0001] This application relates to the field of agricultural machinery braking control technology, and more specifically, to a method for predicting the thermal fade of tractor brakes based on multi-source sensor fusion. Background Technology

[0002] Tractors, as the main power machinery in agricultural production, are widely used in hilly and mountainous areas, orchards on slopes, and long-distance transportation operations. In these scenarios, tractor brakes need to be in a braking state frequently or for extended periods. Especially under continuous braking conditions on long downhill slopes, the brake disc and friction pads generate heat due to friction, causing the brake temperature to rise rapidly. When the temperature exceeds the thermal stability range of the friction material, the coefficient of friction decreases significantly with increasing temperature, resulting in thermal fade. This leads to insufficient braking torque, increased braking distance, and in severe cases, even brake failure, threatening operational safety. Therefore, accurately sensing the thermal fade state of tractor brakes and predicting its development trend in time before it occurs, and taking corresponding control measures, is a technical problem that urgently needs to be solved to ensure tractor braking safety.

[0003] Chinese Invention Patent Publication No. CN115771497A discloses a brake thermal fade compensation control method based on brake pressure control. This method obtains real-time state parameters of the vehicle during the braking process through a three-axis inertial unit, wheel speed sensor, brake pressure sensor and pedal displacement sensor. Based on these parameters, the brake temperature and friction coefficient are calculated. A logic threshold method is used to determine whether the brake has experienced thermal fade. After determining that thermal fade has occurred, the wheel braking pressure is actively increased through feedforward and PI feedback to maintain a constant deceleration during the braking process. While this method can perform active compensation control after brake fade occurs, it only indirectly calculates the brake temperature and friction coefficient estimates for a single path based on vehicle operating parameters. It does not perform time synchronization and dynamic weighting fusion processing of multi-source sensor data such as temperature, speed, and pressure according to braking conditions, resulting in limited accuracy and anti-interference capabilities of the estimation results. Furthermore, this method can only trigger compensation control after the brake temperature reaches a preset threshold and brake fade has already occurred, which is a post-event compensation. It cannot predict and provide graded warnings of the brake's temperature change trend and friction coefficient decay trend over a future period before brake fade occurs, making it difficult to meet the need for early prevention of brake fade risk under continuous heavy-load braking conditions such as long downhill driving of tractors.

[0004] Chinese invention patent publication number CN116872901A discloses a self-compensation system and control method for thermal fade of a vehicle brake-by-wire system. This method calculates the brake wear coefficient and thermal fade friction coefficient based on real-time hydraulic braking force obtained from a hydraulic braking force detection module and real-time wheel rotation speed obtained from a wheel speed detection module, respectively, and then calculates the self-compensating hydraulic braking force to compensate for the target braking force of the hydraulic braking system. However, this method does not use a dedicated temperature sensor to directly measure the brake temperature. Instead, it indirectly estimates the brake temperature change and thermal fade friction coefficient based on hydraulic braking force and wheel speed using models such as heat flux density and Newton's law of cooling. It lacks direct acquisition and fusion of multi-source state perception data such as brake temperature, wheel speed, and slope, making the estimation results susceptible to model simplification errors and actual operating conditions (such as slope changes and wheel slippage), leading to deviations. Furthermore, this method also performs real-time self-compensation control after calculating the thermal fade friction coefficient, without involving advance prediction of the trend of brake thermal state changes, nor considering the differential braking force distribution adjustment when the thermal load between the brakes of each wheel is unbalanced. Its proactiveness and precision in risk response need to be improved.

[0005] Furthermore, Chinese invention patent publication number CN102663186A discloses a parameterized analysis method for brake thermal fade. This method parameterizes the brake's structural dimensions, material properties, air parameters, vehicle parameters, and braking conditions, and uses the finite difference method to perform offline calculations of the brake's temperature field, obtaining the brake's maximum temperature and temperature change curves under different operating conditions, which can be used for brake thermal fade analysis during the design phase. This method is an offline parameterized simulation analysis method, requiring prior manual determination of the brake's structural parameters, material properties, and specific braking conditions (such as emergency braking, constant speed braking on long slopes, and multiple cycle braking) before calculations can be performed. It cannot perform online monitoring of thermal fade status, trend prediction, and risk level assessment based on real-time changing multi-source sensor data during actual tractor field operations, nor does it have the function of automatically generating brake control adjustment commands based on prediction results. Therefore, it is difficult to directly apply to onboard real-time prediction and active control scenarios for tractor brake thermal fade.

[0006] In summary, existing methods for assessing brake thermal fade either rely on single or limited state parameters to indirectly estimate the brake's thermal state and can only provide post-fade compensation control, lacking the technical means to dynamically weight and fuse multi-source sensor data according to braking conditions to improve the accuracy of thermal state perception; or they rely on offline parametric simulation analysis, which cannot be used for real-time online prediction and control in vehicles. Furthermore, most existing methods are designed for passenger cars and commercial vehicles, and some active compensation schemes require assistance from specific power forms such as regenerative braking, making it difficult to fully meet the universal needs for early prediction and graded active control of thermal fade trends under typical operating conditions such as continuous heavy-load braking of tractors on long downhill slopes in hilly and mountainous terrain. Therefore, there is an urgent need to propose a method that can perform time synchronization and dynamic weighted fusion of multi-source state perception data during tractor braking, and predict the trends of brake temperature change and friction coefficient decay before thermal fade occurs, thereby achieving thermal fade risk classification and active adjustment of brake control to improve the braking safety of tractors under complex operating conditions. Summary of the Invention

[0007] To overcome a series of shortcomings in the existing technology, the purpose of this application is to provide a method for predicting the thermal fade of tractor brakes based on multi-source sensor fusion, comprising the following steps: Acquire multi-source state perception data during the tractor braking process. The multi-source state perception data includes brake temperature data, wheel speed data, brake pressure data, vehicle speed data, and slope data. Based on multi-source state perception data, the current braking condition of the tractor is identified, and the braking process is divided into multiple braking condition segments according to the braking condition. The fusion weights of each state perception data are determined based on the braking condition segment, and the fusion data of each state perception data are fused according to the fusion weights to obtain fused state data. Based on the fused state data, thermal characteristic parameters that affect the change of the brake's thermal state are extracted, and the current thermal state of the brake is estimated based on the thermal characteristic parameters to obtain the brake's thermal state characterization results. Based on the thermal state characterization results of the brake, the temperature change trend and friction coefficient decay trend of the brake within a preset time window are predicted, and the thermal fade risk level corresponding to the brake is determined according to the predicted friction coefficient decay trend.

[0008] In some embodiments, the method for identifying braking conditions is as follows: Acquire vehicle speed data, brake pressure change rate data, gradient data, and braking duration data during the braking process, and construct braking condition characteristic data; Based on historical braking data, the braking condition feature data is labeled with the condition category to form a training sample set. The braking condition classification model is then trained based on the training sample set to obtain the braking condition classification model parameters. The performance of the trained braking condition classification model is evaluated based on the validation samples, and the target braking condition classification model that meets the preset classification accuracy requirements is determined based on the model evaluation results. The real-time acquired braking condition feature data is input into the target braking condition classification model to obtain the braking condition category corresponding to the current sampling time. The braking condition categories corresponding to multiple consecutive sampling times are subjected to time-series smoothing to eliminate category jumps caused by instantaneous fluctuations and obtain stable braking condition categories. The current braking condition of the tractor is determined based on the stable braking condition category.

[0009] In some embodiments, the method for obtaining fusion state data is as follows: Multi-source state sensing data is processed for time synchronization to obtain multi-source synchronized state data. State estimation and noise suppression are performed on multi-source synchronous state data to obtain state estimation data; The state estimation data is normalized to obtain normalized state data; Based on the fusion weights corresponding to each state perception data, the normalized state data is weighted and fused to obtain the fused state data.

[0010] In some embodiments, the method for extracting thermal feature parameters is as follows: Based on the fused state data corresponding to each braking condition segment, the comprehensive temperature index, comprehensive braking intensity index, and comprehensive driving condition index are determined. Based on the time-varying relationship of comprehensive temperature indicators, the rate of temperature rise and the rate of temperature fall of the brake are determined. The temperature gradient of the brake is determined based on the temperature difference at different temperature measurement points within the same braking condition segment. Based on comprehensive braking intensity index and comprehensive driving condition index, the heat accumulation characteristics during the braking process are determined, and the heat accumulation rate per unit time is calculated. Based on the rate of temperature rise, rate of temperature fall, temperature gradient, and rate of heat accumulation per unit time, the thermal characteristic parameters of the brake are generated.

[0011] In some embodiments, the method for obtaining the brake thermal state characterization results is as follows: A thermal state estimation model is constructed based on the heat exchange process of the brake, and the thermal state estimation model includes heat generation, heat transfer and heat dissipation terms. The model parameters corresponding to the thermal state estimation model are determined based on the fused state data and thermal characteristic parameters. Based on the fused state data, thermal characteristic parameters, and model parameters, the thermal state estimation model is solved to obtain the temperature distribution state corresponding to the brake. Extract thermal state characterization parameters based on temperature distribution; The thermal state characterization results of the brake are generated based on the thermal state characterization parameters.

[0012] In some embodiments, the method for predicting the temperature change trend of the brake within a preset time window is as follows: Based on the thermal state characterization results, thermal characteristic parameters and braking condition categories corresponding to historical braking condition segments, a temperature trend prediction training sample set is constructed. Based on the temperature trend prediction training sample set, the temperature trend prediction model corresponding to different braking conditions is trained to obtain the temperature trend prediction model corresponding to each braking condition category. Based on the braking condition category corresponding to the current braking condition segment, determine the target temperature trend prediction model; Input the thermal state characterization results and thermal characteristic parameters corresponding to the current braking condition segment into the target temperature trend prediction model to obtain the brake temperature change trend prediction results within the preset time window. When a change in braking condition category is detected, the corresponding target temperature trend prediction model is redefined, and the temperature change trend prediction results are updated.

[0013] In some embodiments, it is determined whether the thermal degradation risk level has reached the preset safety control conditions; If so, a brake control adjustment command is generated, and the brake force distribution mode and / or braking mode are adjusted according to the brake control adjustment command to reduce the risk of brake fade. If not, the current braking force distribution and braking method will be maintained, and the multi-source state perception data during the tractor braking process will continue to be collected and the thermal fade risk level will be updated and judged.

[0014] In some embodiments, the method for generating braking control adjustment commands is as follows: Obtain the operating condition category and thermal fade risk level corresponding to the current braking operating condition segment; The corresponding brake control adjustment level is determined based on the risk level of heat fade. The braking control adjustment strategy is determined based on the braking control adjustment level and the operating condition category. The corresponding braking control adjustment command is generated based on the braking control adjustment strategy.

[0015] In some embodiments, the method for adjusting the braking force distribution mode according to the braking control adjustment command is as follows: In response to the brake control adjustment command, the thermal fade risk status of each wheel brake is obtained; The braking force distribution adjustment parameters are determined based on the heat fade risk status of each wheel brake. Based on the braking force distribution adjustment parameters, the braking force distribution ratio corresponding to each wheel brake is adjusted. The vehicle stability constraint parameters are determined based on wheel speed data, vehicle speed data, and slope data, and the adjusted braking force distribution ratio is then constrained and corrected based on these parameters. The target braking force distribution method is determined based on the revised braking force distribution ratio.

[0016] In some embodiments, the method for adjusting the braking mode according to the braking control adjustment command is as follows: When the safety control condition determination result indicates that the preset safety control conditions are met, the current auxiliary braking capability status of the tractor is obtained; The target braking mode is determined based on the auxiliary braking capability status and the vehicle operating status, wherein the target braking mode includes a compound braking mode and an intermittent braking mode; A braking control adjustment command is generated based on the target braking mode, and the current braking mode is adjusted according to the braking control adjustment command; Monitor the actual braking status of the vehicle corresponding to the adjusted braking method and determine whether it meets the safety braking requirements; When the safety braking requirements are not met, the braking mode is restored to the state before adjustment to obtain the final braking control result.

[0017] Compared with the prior art, this application has the following beneficial effects: By performing time synchronization, consistency correction, and dynamic fusion of multi-source state sensing data such as brake temperature, wheel speed, brake pressure, vehicle speed, and gradient during tractor braking, the accuracy and robustness of brake thermal state sensing are improved. This overcomes the problems of existing technologies that rely on single-parameter estimation of thermal state and are susceptible to changes in operating conditions and sensor noise. By constructing thermal state estimation models, temperature trend prediction models, and thermal fade relationship models, the early prediction of brake temperature change trends and friction coefficient decay trends is achieved, completing the transformation from post-event detection to pre-event prediction and early warning. By fusing friction coefficient decay degree, decay rate, and brake temperature state, thermal fade risk assessment parameters are constructed and multiple risk levels are classified. This invention enables continuous and refined assessment of heat fade risk, providing a basis for the graded triggering of braking force distribution and braking control strategies. After reaching the preset safety control conditions, the braking force distribution is dynamically adjusted according to the heat fade risk of each wheel, and combined with vehicle stability constraints and auxiliary braking capabilities, the invention switches between compound braking and intermittent braking modes, reducing the risk of heat fade while ensuring the braking stability and safety of the entire vehicle. In addition, this invention does not rely on specific power forms such as regenerative braking, and is applicable to various vehicle types such as traditional fuel tractors. It adopts a closed-loop feedback mechanism based on risk level, ensuring the real-time performance of predictive control while taking into account the efficiency of onboard computing resource utilization, and can meet the real-time monitoring, prediction and active control requirements of tractor brake heat fade. Attached Figure Description

[0018] Figure 1 A schematic diagram of the overall process of a tractor brake thermal fade prediction method based on multi-source sensor fusion provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the process for determining the risk level of thermal degradation, adjusting braking control, and updating feedback in an embodiment of the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the scope of protection of the invention.

[0020] This embodiment uses a certain model of wheeled tractor as an example for explanation. The tractor has four wheel brakes (all disc brakes) on the front and rear axles. Non-contact infrared temperature measuring elements and contact temperature sensors are installed in the area of ​​each brake. Speed ​​sensors are installed at the wheel hubs of the four wheels, and pressure sensors are installed at the master cylinder and wheel cylinders of each wheel. The vehicle body is also equipped with a satellite positioning module (GNSS) and an inertial measurement unit (IMU). The data collected by each sensor is read, synchronized and fused by the data acquisition and processing unit in the vehicle controller (VCU). The processing results are used to predict the risk of brake fade and output brake control adjustment commands. Figure 1 The overall flow of the method in this embodiment is shown.

[0021] Non-contact infrared temperature sensing elements are installed on the outer friction track surface of the brake disc or in the corresponding area on the outer wall of the brake drum to collect infrared radiation temperature signals; simultaneously, contact temperature sensors are installed on brake structural components (such as brake calipers and brackets) to collect conducted temperature signals. Both types of signals have independent timestamps and are denoted as infrared radiation temperature signal and conducted temperature signal, respectively.

[0022] First, based on the timestamps corresponding to the two types of temperature signals, and taking the sampling period as a reference, linear interpolation is used to align the heterogeneous signals to a unified time axis, thereby obtaining time-synchronized infrared radiation temperature and conduction temperature.

[0023] Subsequently, outlier detection and filtering were performed on the time-synchronized temperature data: the three-standard-deviation (3σ) criterion within a sliding window was used to identify abnormal sampling values ​​caused by vibration during braking, dust obstruction, oil contamination, and sensor disturbance. Specifically, when the difference between the temperature value of a sampling point and the window mean exceeded three times the window standard deviation, it was determined to be an outlier and replaced with the weighted average of the valid sampling points within the window. The data was then smoothed using a first-order low-pass filter to obtain the preprocessed temperature data. .

[0024] Based on this, consistency correction is performed on the preprocessed temperature data according to the deviation between the infrared radiation temperature signal and the conduction temperature signal. The deviation amount is defined. for: Equation (1) In the formula, The infrared radiation temperature after time alignment is expressed in °C. It reflects the instantaneous radiation temperature of the brake disc surface. It has a fast response speed but is greatly affected by surface emissivity and dust blockage. The conduction temperature after time alignment is expressed in °C. It reflects the conduction temperature of the brake structural components. It has a slow response speed but is less affected by environmental interference and has good stability. This represents the deviation between two types of temperature signals, expressed in °C.

[0025] According to the deviation and preset consistency correction coefficient ( Based on the heat conduction path between the brake disc and the structural components, k is taken in this embodiment. c =0.35), and consistency correction is performed on the preprocessed temperature data to obtain the corrected temperature data. As shown in equation (2): Equation (2) In the formula, Temperature data is preprocessed and the unit is °C. The consistency correction coefficient is dimensionless and ranges from [0,1]. It reflects the correction weight of the conduction temperature signal relative to the correction result. The larger the value, the greater the contribution of the conduction temperature signal to the correction result. Temperature data is presented in °C for calibration purposes.

[0026] Finally, based on the heat transfer relationship between the brake disc and the friction pads, heat transfer compensation is performed on the corrected temperature data to obtain the final output brake temperature data. As shown in equation (3): Equation (3) In the formula, This is the heat transfer time constant between the brake disc and the friction lining, expressed in seconds. Its value is pre-calibrated based on the thermal diffusivity of the brake disc material and the lining thickness. In this embodiment, it is taken as... =1.2s; To correct the time rate of change of temperature data, the unit is ℃ / s. Equation (3) introduces a temperature rate of change term to compensate for the lag in the conduction measurement, so that... It is closer to the actual temperature of the brake disc friction interface.

[0027] For example, if at a certain sampling moment, the infrared radiation temperature is 186℃, the conduction temperature is 171℃, the preprocessed temperature after 3σ filtering is 172℃, and the temperature change rate is 8℃ / s, then the deviation can be obtained from equation (1) as 15℃; the corrected temperature data can be obtained from equation (2) as 177.25℃; and the brake temperature data can be obtained from equation (3) as 186.85℃. This value is the brake temperature data output at that moment, which is used for subsequent thermal feature extraction and thermal state estimation.

[0028] The wheel hub speed sensor outputs pulse signals, which are then analyzed to obtain the wheel hub speed of each wheel. (Unit: r / min). Combined with the actual rolling radius of the corresponding wheel. (Unit: m, factory calibration value or real-time correction based on tire pressure and load), convert the wheel hub speed to the wheel conversion speed v according to formula (4). w : Equation (4) In the formula, The wheel speed is expressed in m / s. This represents the hub rotation speed, in r / min. This represents the actual rolling radius of the wheel, in meters (m).

[0029] At the same time, the positioning speed data output by the satellite positioning module is acquired. (Unit: m / s), and inertial velocity data v obtained by integrating the acceleration data collected by the inertial measurement unit. i (Unit: m / s) The three speed data are aligned with timestamps to obtain multi-source synchronous speed data.

[0030] Speed ​​consistency analysis was performed on multi-source synchronized speed data to extract the deviation characteristics between the wheel-calculated speed and the positioning speed. As shown in equation (5), the wheel slip ratio is calculated accordingly. As shown in equation (6): Equation (5) Equation (6) In the formula, The deviation characteristic between the wheel conversion speed and the positioning speed is expressed in m / s; The slip ratio of the corresponding wheel is dimensionless; A very small positive number is set to prevent the denominator from being zero (in this embodiment, it is taken as...). =0.1m / s). When the slip ratio When the slip threshold is exceeded (0.15 in this embodiment), the wheel is determined to be in a slipping state, and the wheel's converted speed v is then determined. w The credibility of this information has decreased.

[0031] Based on the wheel slip state and the reliability of the positioning speed data and inertial speed data (the reliability is determined by the positioning accuracy factor DOP value of the satellite positioning module and the relationship between the cumulative error of the inertial measurement unit and the time), the wheel conversion speed is fused and corrected to obtain the final output wheel speed data, as shown in Equation (7): Equation (7) In the formula, The wheel speed data (represented as velocity) is integrated and corrected, with units of m / s; , These are the confidence weighting coefficients for positioning velocity data and inertial velocity data, respectively, and c p +c i =1, its value is updated online based on the current DOP value and the accumulated error of the inertial measurement unit. When the wheel is not in a slipping state (s i When it is smaller, Speed ​​converted by wheel Mainly; when the wheels are in a slipping state ( When it is relatively large, The system will gradually transition to a system that primarily uses a weighted average of the positioning speed and inertial speed to avoid wheel speed distortion during skidding conditions affecting subsequent braking condition identification.

[0032] The vehicle pitch angle data collected by the inertial measurement unit is acquired, and the inertial slope estimation data is determined accordingly. (Unit: %, expressed as a percentage slope); simultaneously based on altitude change data output by the satellite positioning module. With displacement change data Calculate the estimated positioning slope data according to formula (8). : Equation (8) In the formula, This represents the height change between two adjacent sampling points, in meters. This represents the change in horizontal displacement between two adjacent sampling points, in meters (m). The slope estimation data is presented as a percentage.

[0033] For inertial slope estimation data Low-pass filtering is performed to obtain filtered inertial slope estimation data. This filters out high-frequency noise introduced by vehicle body vibration and short-term attitude disturbances. The speed is determined based on current wheel speed data. and Corresponding fusion weights When the vehicle speed is low or the wheels slip, the calculation error of the positioning slope estimation increases, so the weight of the inertial slope estimation should be increased. When the vehicle speed is high and the wheels do not slip, the positioning slope estimation accuracy is high, so its weight should be increased appropriately. The calculation of the fusion weight and slope data are shown in Equations (9) and (10), respectively: Equation (9) Equation (10) In the formula, The fusion weights for the inertial slope estimation data range from [0,1]; v is the vehicle speed corresponding to the current wheel rotation speed data, in m / s. The preset vehicle speed threshold (taken in this embodiment) =3m / s), The shape factor used to control the steepness of the weight curve (in this embodiment, it is taken as...) =0.8); This is the final output slope data, expressed as a percentage. When v is much smaller than... hour, When v approaches 1, slope data is mainly estimated using inertia; when v is much greater than 1... hour, Approaching 0, the slope data is mainly estimated by location, thus achieving a smooth switch between the two types of slope estimation methods under different vehicle speed ranges.

[0034] Acquire brake master cylinder pressure signal and the corresponding brake caliper pressure signal for each wheel. Where r = 1, 2, 3, 4 represent the left front, right front, left rear, and right rear wheels, respectively. This is based on the brake master cylinder pressure signal and preset braking force distribution relationships (such as the front and rear axle braking force distribution coefficient k). b1 k b2 , satisfying k b1 +k b2 =1), determine the theoretical braking pressure data corresponding to each wheel brake. As shown in equation (11): Equation (11) In the formula, This is the brake master cylinder pressure signal, in MPa. is the preset braking force distribution coefficient corresponding to the r-th wheel brake, which is dimensionless; This represents the theoretical braking pressure data for the r-th wheel brake, in MPa.

[0035] Based on the pump pressure signal Determine the actual braking pressure data corresponding to each wheel brake (i.e. (It itself), and based on the difference between theoretical braking pressure and actual braking pressure, determine the braking pressure deviation. As shown in equation (12): Equation (12) According to the brake pressure deviation Deviation from preset conditions (In this embodiment, the deviation threshold is taken as 0.3MPa) The matching results between these values ​​determine the braking load state corresponding to each wheel brake: When When the wheel brake is under normal load, it is determined that the wheel brake is in a normal load condition; when At this point, it is determined that the wheel brake is under abnormal load (possibly due to brake line pressure loss or brake caliper sticking). The actual brake pressure data will then be used. Together with the corresponding braking load status flag, it serves as the output of braking pressure data for subsequent braking condition identification and thermal feature extraction.

[0036] Based on the acquisition of the above multi-source state perception data, the braking conditions are further identified, and the braking condition segments are divided accordingly, and the fusion weight of each state perception data is determined.

[0037] Extract vehicle speed v and rate of change of braking pressure during the braking process. ,slope and braking duration The braking condition feature vector X is constructed as shown in equation (13): Equation (13) Based on historical braking data, the braking condition feature vectors are labeled with braking condition categories (in this embodiment, they are divided into 4 categories: C1 light braking, C2 moderate braking, C3 long downhill continuous braking, and C4 emergency braking) to form a training sample set. The braking condition classification model is then trained based on the training sample set. In this embodiment, the braking condition classification model adopts a multilayer perceptual classification model with hidden layers, and its output is shown in equation (14): Equation (14) In the formula, This is the weight matrix of the classification model. Both are bias vectors, and are parameters obtained from model training. This represents the probability distribution vector corresponding to each operating condition category. The performance of the trained model is evaluated based on validation samples (in this embodiment, classification accuracy is used as the evaluation metric, with a preset classification accuracy requirement of 95%), to determine the target braking condition classification model that meets the accuracy requirement.

[0038] The real-time acquired braking condition feature vector is input into the target braking condition classification model, and the category with the highest probability is taken as the braking condition category at the current sampling time. As shown in equation (15): Equation (15) A sliding majority voting time-series smoothing process is performed on the braking condition categories corresponding to multiple consecutive sampling times (five consecutive sampling periods in this embodiment) to eliminate category jumps caused by instantaneous fluctuations, obtain stable braking condition categories, and determine the current braking condition of the tractor accordingly.

[0039] Determining the start time t of the braking process based on braking pressure data s(The brake master cylinder pressure signal exceeds the initial threshold p for the first time) s0 In this embodiment, p is taken as s0 (Time when =0.2MPa) and termination time t e (The brake master cylinder pressure signal drops to p) s0 The following time intervals are used to obtain the initial braking condition segment [t]. s ,t e Based on the changes in the stable braking condition category within the segment, the condition category change information is determined; when the duration for which two adjacent stable braking condition categories remain different exceeds the preset stability determination duration (1.5s in this embodiment), this location is determined as the segment segmentation position, and the initial braking condition segment is divided into multiple candidate braking condition sub-segments accordingly.

[0040] Further, based on the duration of each candidate braking condition sub-segment (if it is shorter than the preset minimum segment duration of 0.5s, it is determined to be an invalid fragment) and the continuity relationship of the condition category between adjacent candidate sub-segments (adjacent sub-segments with the same condition category or adjacent condition intensity levels are merged), the candidate braking condition sub-segments are merged to finally obtain multiple braking condition segments corresponding to this braking process.

[0041] According to the braking condition category C corresponding to the current braking condition segment k And a pre-defined weight mapping relationship (the reliability of state-sensing data varies under different operating conditions, such as the significant increase in the importance of temperature data under long downhill continuous braking conditions), and the initial weight coefficients corresponding to each state-sensing data are determined by looking up a table. (g=1,2,3,4,5 correspond to five types of data: temperature, rotational speed, pressure, vehicle speed, and gradient, respectively).

[0042] Obtain the data quality parameter q corresponding to the state sensing data within each braking condition segment. g As shown in equation (16): Equation (16) In the formula, Let g be the signal power of the g-th type of state-sensing data within this segment. q represents the corresponding noise power. g The unit is dB, and a higher value indicates higher data quality within the current segment. This is based on the data quality parameter q. g The weight adjustment coefficient is determined according to formula (17). : Equation (17) Based on weighting adjustment coefficient For the initial weight coefficients After correction, the target weight coefficient is obtained. As shown in equation (18), and after normalizing the target weight coefficients, the fusion weight w corresponding to each state perception data is obtained. g As shown in equation (19): Equation (18) Equation (19) For example, in a long downhill continuous braking condition segment (C3 type), the initial weight coefficients of the five types of data, namely temperature, speed, pressure, vehicle speed and gradient, are found to be 0.35, 0.15, 0.20, 0.15 and 0.15 respectively according to the weight mapping relationship; after data quality assessment, the corresponding weight adjustment coefficients are 0.24, 0.19, 0.21, 0.18 and 0.18 respectively; calculated by equations (18) and (19), the final fusion weights of the five types of data are approximately 0.31, 0.11, 0.16, 0.10 and 0.10 (the sum of the five after normalization is 1). Temperature data obtains the highest fusion weight under this long downhill condition, which is consistent with its dominant role in predicting thermal fading under this condition.

[0043] After time synchronization processing of the multi-source state sensing data, state estimation and noise suppression processing are performed (in this embodiment, Kalman filtering is used to estimate the state of various types of data) to obtain state estimation data; the state estimation data is then normalized (by using range normalization to map to the [0,1] interval) to obtain normalized state data. Based on the fusion weight w corresponding to each state-aware data g The normalized state data is weighted and fused to obtain the fused state data. As shown in equation (20): Equation (20) Fusion state data It comprehensively characterizes the overall state level of five types of data within the current braking condition segment: brake temperature, wheel speed, brake pressure, vehicle speed, and gradient. It serves as the basic input for subsequent extraction of thermal characteristic parameters and estimation of the brake thermal state.

[0044] Based on the obtained fusion state data, thermal characteristic parameters affecting the change of the brake's thermal state are further extracted, and the current thermal state of the brake is estimated accordingly.

[0045] Based on the fused state data corresponding to each braking condition segment, the comprehensive temperature index, comprehensive braking intensity index, and comprehensive driving condition index are determined (all three are derived from the fused state data). The corresponding component is obtained by averaging the values ​​of different operating conditions. Based on the time-varying relationship of the comprehensive temperature index, the brake temperature rise rate is determined. With the rate of temperature decrease As shown in equations (21) and (22): (Heating section) Equation (21) (Cooling section) Equation (22) The temperature gradient of the brake is determined based on the temperature difference between different temperature measuring points (such as measuring points corresponding to the four wheel brakes) within the same braking condition segment. As shown in equation (23): Equation (23) In the formula, T b,max T b,min These represent the maximum and minimum brake temperature data at each measuring point within the same segment, in °C; L d This represents the spatial distance between corresponding measuring points, in meters (m). The unit is ℃ / m, and it is used to characterize the degree of uneven heating between the brakes of each wheel.

[0046] Based on the comprehensive braking intensity index and comprehensive driving condition index, the heat accumulation characteristics during the braking process are determined, and the heat accumulation rate Q per unit time is calculated according to formula (24). h : Equation (24) In the formula, μ is the current friction coefficient (initially taken as the calibration value μ0); F n This represents the braking force, measured in N (N), calculated from braking pressure data combined with brake caliper geometric parameters; v r Q represents the relative sliding speed between the brake disc and the friction lining, in m / s, calculated from wheel speed data; m is the effective heat capacity mass of the brake disc, in kg; c is the specific heat capacity of the brake disc material, in J / (kg·℃); h The unit is ℃ / s, representing the rate of temperature rise of the brake disc due to frictional heat generation per unit time. Based on the temperature rise rate k... up , rate of temperature decrease k down Temperature gradient ΔT g and the rate of heat accumulation per unit time Q h Generate the brake thermal feature parameter vector [k] up ,k down ,ΔT g Q h ].

[0047] A thermal state estimation model is constructed based on the heat exchange process of the brake. This model treats the brake disc as a lumped-parameter heat container, and its thermal balance relationship is composed of heat generation, heat transfer, and heat dissipation terms, as shown in equation (25): Equation (25) In the formula, The term for heat generation (i.e., frictional heat generation power) is shown in equation (26); For heat transfer (the heat dissipation power of the brake disc conducted through the bracket, wheel hub to the axle and other structural components), as shown in equation (27); The heat dissipation term (convective heat dissipation power between the brake disc and the surrounding air) is shown in equation (28): Equation (26) Equation (27) Equation (28) In the formula, The ambient temperature is expressed in °C and is obtained from the vehicle's ambient temperature sensor. The equivalent thermal resistance from the brake disc to the surrounding structural components is expressed in °C / W and is calibrated based on the material and connection method of the structural components; h is the convective heat transfer coefficient, expressed in W / (m²). 2 ·℃), which is related to vehicle speed (the higher the vehicle speed, the stronger the convection cooling); A is the effective heat dissipation area of ​​the brake disc, in m². 2 .

[0048] Based on the fused state data and thermal characteristic parameters, the parameters corresponding to the above model are determined, and the first-order forward difference method is used to recursively solve equation (25), as shown in equation (29): Equation (29) The temperature distribution state T(x,t) corresponding to the brake is constructed by recursively solving the results at each measuring point, where x represents the spatial location of each measuring point. Thermal state characterization parameters, including the peak temperature at the current moment, are extracted based on the temperature distribution state. Thermal equilibrium temperature (The temperature value corresponding to the rate of temperature change approaching 0 for multiple consecutive sampling periods) and the aforementioned thermal characteristic parameters Together, they constitute the results of the brake thermal state characterization. As shown in equation (30): Equation (30) For example, in a continuous braking scenario on a long downhill slope, the initial brake temperature is 80℃, the ambient temperature is 25℃, the thermal resistance is 1.5℃ / W, and the convective heat transfer coefficient is 25W / (m²). 2 (℃), heat dissipation area is 0.12m² 2 The effective heat capacity of the brake disc is m = 6 kg, and its specific heat capacity is 460 J / (kg·℃). Assume the coefficient of friction at this moment is 0.40, the braking force is 4500 N, and the relative sliding speed is 6 m / s. If it is 1s, then from equation (26) we can obtain =10800W; From equation (27), we can obtain ≈36.7W; From equation (28), we can obtain =165W; Substituting into equation (29), the brake temperature after this sampling period can be obtained. The value is approximately 83.84℃, meaning that the brake temperature rises continuously at a rate of approximately 3.84℃ per second under this braking condition. This value will be used as one of the initial inputs for subsequent temperature trend prediction.

[0049] Based on the results of the brake thermal state characterization, the temperature change trend and friction coefficient decay trend of the brake within a preset time window are predicted.

[0050] Based on the thermal state characterization results, thermal characteristic parameters, and braking condition categories corresponding to historical braking condition segments, temperature trend prediction training sample sets are constructed for different braking condition categories, and temperature trend prediction models corresponding to each braking condition category are trained separately. In this embodiment, the temperature trend prediction model... (k is the operating condition category number) The input is as follows: The output is a preset time window for the future. (In this embodiment, we take) The time-series prediction model for the temperature sequence within a period of 60s (sampling period 1s) is shown in equation (31). Equation (31) Based on the braking condition category corresponding to the current braking condition segment, determine the corresponding target temperature trend prediction model. The model is then input with the current thermal state characterization results and thermal characteristic parameters to obtain the predicted trend of brake temperature change within a preset time window. When a change in braking condition category is detected (such as switching from moderate braking to continuous braking on a long downhill slope), the corresponding target temperature trend prediction model is redefined, and the temperature change trend prediction result is recalculated based on the latest data of the changed condition segment to avoid prediction deviations caused by using a model that does not match the condition category.

[0051] Continuing with the aforementioned numerical example, if the brake temperature rises continuously at a rate of approximately 3.84℃ / s during a long downhill continuous braking segment, and based on the target temperature trend prediction model's prediction of the temperature decay pattern (the rate of temperature increase gradually slows down over time) from historical similar conditions, the temperature change trend over the next 60 seconds is roughly as follows: within the first 20 seconds, the temperature rises from approximately 84℃ to approximately 156℃ (average rate of approximately 3.6℃ / s); from 20 to 45 seconds, the temperature rises to approximately 230℃ (average rate of approximately 2.96℃ / s); and from 45 to 60 seconds, the temperature increase slows down to approximately 262℃ (average rate of approximately 2.13℃ / s). The rate of temperature increase gradually decreases, which is consistent with the physical law that the temperature tends to reach thermal equilibrium under long downhill continuous braking conditions.

[0052] Based on the temperature-friction coefficient variation relationship corresponding to different friction material types, a friction coefficient decay relationship model is established. In this embodiment, the relationship is described in a piecewise form: when the brake temperature is lower than the initial decay temperature T1, the friction coefficient remains basically stable; when the brake temperature is between T1 and the thermal decay critical temperature T2, the friction coefficient decays with increasing temperature in an approximately exponential manner, as shown in equation (32): Equation (32) In the formula, The initial coefficient of friction of the friction material at the reference temperature (factory calibration value) is taken in this embodiment. =0.42; The initial decay temperature is given in °C. In this embodiment, the temperature is taken as the corresponding friction material type. =180℃; The thermal decay critical temperature is expressed in °C. In this embodiment, it corresponds to the type of friction material. =320℃; when the brake temperature is higher than At that point, the coefficient of friction no longer decreases further with temperature and remains at a constant value. Corresponding minimum value The condition remains unchanged, meaning that thermal decay reaches a saturation point and no longer decreases indefinitely. The attenuation coefficient is expressed in units of 1 / ℃ and is related to the type of friction material. In this embodiment, it is taken as... =0.0065 (this value is obtained by fitting friction material bench thermal decay test data); T is the predicted brake temperature in °C. Based on the type of friction material corresponding to the current brake, a matching target friction coefficient decay relationship model is determined (different friction materials correspond to different...). , , The value is obtained by looking up a table.

[0053] Predicted results of brake temperature change trend within a preset time window Substituting the values ​​into the target friction coefficient decay model, we obtain the friction coefficient variation trend within the corresponding time window. Based on this trend, the evaluation parameters for the degree of friction coefficient attenuation are calculated. As shown in equation (33), the friction coefficient decay rate is calculated. As shown in equation (34): Equation (33) Equation (34) In the formula, The parameter is used to evaluate the degree of friction coefficient decay, expressed as a percentage. The larger the value, the greater the decrease in friction coefficient within the predicted time window. This represents the rate of decrease in friction coefficient, expressed in units of 1 / s. According to... and The numerical value determines the predicted friction coefficient decay trend.

[0054] Continuing with the previous numerical example, substituting the predicted temperature sequence into equation (32): at t+20s, T pred ≈156℃ (below T1=180℃, the friction coefficient μ≈0.42 remains stable); at t+45s, T pred ≈230℃, from equation (32) we can get μ≈0.303; at t+60s T pred ≈262℃, therefore μ≈0.246. From equation (33), D can be obtained. μ ≈41.4%, from equation (34) we can obtain v μ The value ≈0.0029 / s indicates that the friction coefficient of this braking condition segment shows a significant decreasing trend within the predicted time window.

[0055] Based on the evaluation parameters and decay rate of the friction coefficient attenuation, the thermal degradation risk level is further determined, and it is judged whether the preset safety control conditions have been met. The specific process is as follows: Figure 2 As shown.

[0056] Evaluation parameters based on the degree of friction coefficient decay Friction coefficient decay rate and brake temperature status parameters Construct the thermal fade risk assessment parameter R. Among them, the brake temperature state parameter... The current brake temperature relative to the safe temperature With limiting temperature The normalized result is shown in Equation (35); the thermal degradation risk assessment parameter R is the weighted sum of the three indicators after normalization, as shown in Equation (36): Equation (35) Equation (36) In the formula, The safe temperature threshold for the brake is expressed in °C. In this embodiment, it is taken as... =150℃; The limit temperature threshold of the brake is expressed in °C. In this embodiment, it is taken as... =350℃; , These are the evaluation parameters for the degree of friction coefficient decay. With decay rate The result after normalization to the [0,1] interval by the preset range; , , For the corresponding weight coefficients, and + + =1, in this embodiment, we take... =0.45、 =0.30、 =0.25, the weight value is determined according to the degree of influence of friction coefficient decay on braking performance, among which the degree of friction coefficient decay has the greatest contribution to risk assessment; R is the thermal fade risk assessment parameter, and the value range is [0,1].

[0057] Based on the matching relationship between the heat fade risk assessment parameter R and the corresponding judgment thresholds for each risk level, the risk level range corresponding to the current brake is determined, and then the corresponding heat fade risk level is generated. In this embodiment, the risk level determination rule is as follows: when R < 0.30, the risk level of thermal fade is determined to be Level I (normal), at which point the brake friction coefficient remains stable and the thermal state is within the normal range; when 0.30 ≤ R < 0.55, the risk level of thermal fade is determined to be Level II (mild warning), indicating that the friction coefficient begins to show a slight attenuation trend during braking, and continuous monitoring of the brake's thermal state changes is required; when 0.55 ≤ R < 0.75, the risk level of thermal fade is determined to be Level III (moderate warning), indicating that the friction coefficient attenuation is relatively obvious, and the risk of decreased braking performance gradually increases; when R ≥ 0.75, the risk level of thermal fade is determined to be Level IV (severe fade), indicating that the friction coefficient has significantly attenuated, and the brake has a significant risk of decreased braking performance.

[0058] Continuing with the previous numerical examples, =41.4%, after normalization to the preset range (0%~60%) ≈0.690; ≈0.0029 / s, after normalization to the preset range (0~0.006 / s) ≈0.483; Current brake temperature T bThe predicted peak temperature is approximately 262℃. From equation (35), T can be obtained. s =0.56. Substituting into equation (36), we get R≈0.595. R=0.595 falls within the interval [0.55,0.75), so the current heat fade risk level of the brake is determined to be Level III (moderate warning).

[0059] Obtain the current brake's thermal fade risk level L. r Braking condition category C k Based on the slope data θ, the safety control judgment parameter S is constructed as shown in equation (37): Equation (37) In the formula, Risk level The normalized results after mapping from level I to level IV to 0, 0.33, 0.67, and 1; This is a condition category indicator. It is set to 1 for continuous braking or emergency braking on long downhill slopes and 0 for other conditions, in order to reflect the faster heat fade under continuous braking conditions. For slope data The result is normalized according to the preset slope range (0-25%). , , For the corresponding weight coefficients, and their sum is 1, in this embodiment, we take... =0.5、 =0.3、 =0.2. The safety control judgment parameter S is compared with the preset safety control threshold. (In this embodiment, we take) =0.5) for comparison: when When the preset safety control conditions are met, it is determined that the conditions have been met; when If the preset safety control conditions are not met, it is determined that the conditions are not met.

[0060] Continuing with the previous example, Level III, after normalization =0.67; The current operating condition is C3 long downhill continuous braking. ;slope =12%, after normalization =12 / 25=0.48. From equation (37), we can get S≈0.731, S≈0.731>S th =0.5, indicating that the preset safety control conditions have been met, and the braking control adjustment process begins.

[0061] When the preset safety control conditions are met, a braking control adjustment command is generated, and the braking force distribution mode and / or braking mode are adjusted accordingly. The specific process is as follows: Figure 2 The lower half is shown.

[0062] Obtain the operating condition category C corresponding to the current braking operating condition segment. k and thermal degradation risk level L r The corresponding brake control adjustment level G is determined based on the risk level of heat fade. c In this embodiment, levels I through IV are mapped sequentially to G. c =0 (No adjustment required), 1 (Minor adjustment), 2 (Medium adjustment), 3 (Major adjustment). Based on braking control adjustment level G c and operating condition category C k The system determines the braking control adjustment strategy (such as whether to adjust the brake force distribution, whether to switch the braking mode, and the adjustment level) according to a preset strategy table, and generates corresponding braking control adjustment commands accordingly. In the aforementioned example, L... r Level III, corresponding to G c =2, triggering a moderate adjustment strategy, which means simultaneously adjusting the braking force distribution method and assessing whether it is necessary to switch the braking mode.

[0063] In response to the braking control adjustment command, the heat fade risk status of each wheel brake is obtained (based on the risk assessment parameter R calculated separately for each wheel brake). i (And risk level). Based on the heat fade risk status of each wheel brake, determine the braking force distribution adjustment parameter Δk. i As shown in equation (38): Equation (38) In the formula, R i R is the thermal fade risk assessment parameter corresponding to the i-th wheel brake; avg γ is the average value of the risk assessment parameters of each wheel brake; γ is the adjustment gain coefficient, which is taken as 0.15 in this embodiment. Equation (38) shows that the braking force distribution ratio of wheel brakes with risk assessment parameters higher than the average level will be appropriately reduced, and the braking force distribution ratio of wheel brakes with risk assessment parameters lower than the average level will be appropriately increased, so as to achieve a balanced distribution of thermal load of each wheel brake and delay the aggravation of local thermal decay.

[0064] Based on braking force distribution adjustment parameters The braking force distribution ratio of each wheel brake was adjusted to obtain preliminary adjustment results. Further, based on wheel speed data, vehicle speed data, and gradient data, the vehicle stability constraint parameter C is determined. stab (In this embodiment, the upper limit of the yaw rate deviation and the difference in braking force distribution ratio between each wheel is used to characterize the difference, such as the difference in braking force distribution ratio between adjacent wheels not exceeding 0.1), and based on C stabThe adjusted braking force distribution ratio is constrained and corrected as shown in equation (39) to obtain the corrected target braking force distribution method. : Equation (39) In the formula, To constrain the truncation function, , To meet the upper and lower limits of the braking force distribution ratio of the i-th wheel under vehicle stability constraints, a target braking force distribution method is determined to ensure overall vehicle braking stability while reducing the brake fade rate of high-risk wheels.

[0065] When the safety control condition determination result indicates that the preset safety control conditions are met, the current auxiliary braking capability status of the tractor is obtained (such as whether it has auxiliary braking capabilities such as hydraulic retarder and engine braking, and their current available capacity). Based on the auxiliary braking capability status and the vehicle's operating status (vehicle speed, gradient, load, etc.), the target braking mode is determined: when the auxiliary braking capability is sufficient, a compound braking mode (the service brake and auxiliary brake work together to reduce the load proportion of a single brake) is preferred; when the auxiliary braking capability is insufficient or the auxiliary braking conditions are not met, an intermittent braking mode is adopted (the brake is intermittently applied and released according to a preset duty cycle to provide a heat dissipation interval for the brake).

[0066] Based on the determined target braking mode, a braking control adjustment command is generated, and the current braking mode is adjusted accordingly. The actual braking state of the vehicle corresponding to the adjusted braking mode (such as braking deceleration and braking distance margin) is monitored to determine whether the safety braking requirements are met (in this embodiment, the actual braking deceleration is required to be no less than a preset safety braking deceleration threshold). When the safety braking requirements are not met, the braking mode is restored to the one before adjustment, and the final braking control result is obtained by combining the braking force distribution adjustment result to ensure that the basic braking safety performance is not sacrificed while reducing the risk of heat fade.

[0067] Continuing with the previous example, since the tractor has engine braking assist capability and currently has sufficient available capacity, the target braking mode is determined to be a composite braking mode, that is, while maintaining the modified braking force distribution mode k. i target Based on this, engine braking is simultaneously engaged to share part of the braking load, thereby reducing the frictional heat generation power Q of the service brake. gen This slows down the further decay of the friction coefficient.

[0068] When the risk level of heat fade does not meet the preset safety control conditions, the current braking force distribution and braking mode are maintained, and multi-source state sensing data during the tractor braking process continues to be collected. The heat fade risk level is updated according to the preset cycle, forming a closed-loop feedback optimization mechanism. The specific process is as follows: Figure 2 As shown.

[0069] Specifically, the current braking force distribution and braking method will remain unchanged; the risk level update cycle will be determined based on the current braking condition category. As shown in equation (40): Equation (40) In the formula, To preset the baseline update cycle, this embodiment takes... =10s; As the risk sensitivity coefficient, in this embodiment, we take... =2; The result is the normalized result of the current risk level. Equation (40) shows that the higher the risk level, the shorter the update cycle. In this embodiment, the update cycles corresponding to levels I to IV are approximately 10s, 6s, 4.3s, and 3.3s, respectively, in order to achieve a dynamic adjustment mechanism where the higher the risk, the more intensive the monitoring.

[0070] According to the risk level update cycle, multi-source state perception data during the braking process is obtained. The process of updating the fused state data, extracting thermal characteristic parameters, obtaining the results of the brake thermal state characterization, predicting the temperature change trend and the friction coefficient decay trend is repeated to obtain the updated results. According to the updated friction coefficient decay trend, the thermal fade risk evaluation parameters are recalculated according to Equations (33) to (36) and the thermal fade risk level corresponding to the current brake is determined. Then, according to the updated thermal fade risk level and the current braking condition category, the next risk level update cycle is determined according to Equation (40). This cycle is repeated to form a continuous closed-loop monitoring and prediction of the thermal fade state of the tractor brake.

[0071] In summary, this embodiment overcomes the problems of high noise and susceptibility to environmental interference in single-sensor measurements by performing time synchronization, anomaly removal, consistency correction, and dynamic weighted fusion based on operating condition segments on multi-source state sensing data such as brake temperature, wheel speed, brake pressure, vehicle speed, and gradient. It achieves early prediction of brake thermal fade trends by constructing a thermal state estimation model that includes heat generation, heat transfer, and heat dissipation terms, combined with temperature trend prediction models trained according to operating condition categories and a decay model based on the material temperature-friction coefficient relationship. Furthermore, it provides a quantitative decision-making basis for brake control adjustment by constructing thermal fade risk assessment parameters that integrate friction coefficient decay degree, decay rate, and temperature state, and by determining the risk level using tiered thresholds. Finally, it reduces the risk of brake thermal fade while ensuring the overall vehicle braking stability and safety by dynamically adjusting the braking force distribution ratio according to the risk state of each wheel's brakes and flexibly switching between compound braking and intermittent braking modes based on auxiliary braking capabilities.

[0072] It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for predicting thermal fade of tractor brakes based on multi-source sensor fusion, characterized in that, Includes the following steps: Acquire multi-source state perception data during the tractor braking process. The multi-source state perception data includes brake temperature data, wheel speed data, brake pressure data, vehicle speed data, and slope data. Based on multi-source state perception data, the current braking condition of the tractor is identified, and the braking process is divided into multiple braking condition segments according to the braking condition. The fusion weights of each state perception data are determined based on the braking condition segment, and the fusion data of each state perception data are fused according to the fusion weights to obtain fused state data. Based on the fused state data, thermal characteristic parameters that affect the change of the brake's thermal state are extracted, and the current thermal state of the brake is estimated based on the thermal characteristic parameters to obtain the brake's thermal state characterization results. Based on the thermal state characterization results of the brake, the temperature change trend and friction coefficient decay trend of the brake within a preset time window are predicted, and the thermal fade risk level corresponding to the brake is determined based on the predicted friction coefficient decay trend.

2. The method for predicting the thermal fade of tractor brakes according to claim 1, characterized in that, The method for identifying braking conditions is as follows: Acquire vehicle speed data, brake pressure change rate data, gradient data, and braking duration data during the braking process, and construct braking condition characteristic data; Based on historical braking data, the braking condition feature data is labeled with the condition category to form a training sample set. The braking condition classification model is then trained based on the training sample set to obtain the braking condition classification model parameters. The performance of the trained braking condition classification model is evaluated based on the validation samples, and the target braking condition classification model that meets the preset classification accuracy requirements is determined based on the model evaluation results. The real-time acquired braking condition feature data is input into the target braking condition classification model to obtain the braking condition category corresponding to the current sampling time. The braking condition categories corresponding to multiple consecutive sampling times are subjected to time-series smoothing to eliminate category jumps caused by instantaneous fluctuations and obtain stable braking condition categories. The current braking condition of the tractor is determined based on the stable braking condition category.

3. The method for predicting the thermal fade of tractor brakes according to claim 1, characterized in that, The method for obtaining fusion state data is as follows: Multi-source state sensing data is processed for time synchronization to obtain multi-source synchronized state data. State estimation and noise suppression are performed on multi-source synchronous state data to obtain state estimation data; The state estimation data is normalized to obtain normalized state data; Based on the fusion weights corresponding to each state perception data, the normalized state data is weighted and fused to obtain the fused state data.

4. The method for predicting the thermal fade of tractor brakes according to claim 1, characterized in that, The method for extracting thermal characteristic parameters is as follows: Based on the fused state data corresponding to each braking condition segment, the comprehensive temperature index, comprehensive braking intensity index, and comprehensive driving condition index are determined. Based on the time-varying relationship of comprehensive temperature indicators, the rate of temperature rise and the rate of temperature fall of the brake are determined. The temperature gradient of the brake is determined based on the temperature difference at different temperature measurement points within the same braking condition segment. Based on comprehensive braking intensity index and comprehensive driving condition index, the heat accumulation characteristics during the braking process are determined, and the heat accumulation rate per unit time is calculated. Based on the rate of temperature rise, rate of temperature fall, temperature gradient, and rate of heat accumulation per unit time, the thermal characteristic parameters of the brake are generated.

5. The method for predicting the thermal fade of tractor brakes according to claim 1, characterized in that, The method for obtaining the thermal state characterization results of the brake is as follows: A thermal state estimation model is constructed based on the heat exchange process of the brake, and the thermal state estimation model includes heat generation, heat transfer and heat dissipation terms. The model parameters corresponding to the thermal state estimation model are determined based on the fused state data and thermal characteristic parameters. Based on the fused state data, thermal characteristic parameters, and model parameters, the thermal state estimation model is solved to obtain the temperature distribution state corresponding to the brake. Extract thermal state characterization parameters based on temperature distribution; The thermal state characterization results of the brake are generated based on the thermal state characterization parameters.

6. The method for predicting the thermal fade of tractor brakes according to claim 1, characterized in that, The method for predicting the temperature change trend of the brake within a preset time window is as follows: Based on the thermal state characterization results, thermal characteristic parameters and braking condition categories corresponding to historical braking condition segments, a temperature trend prediction training sample set is constructed. Based on the temperature trend prediction training sample set, the temperature trend prediction model corresponding to different braking conditions is trained to obtain the temperature trend prediction model corresponding to each braking condition category. Based on the braking condition category corresponding to the current braking condition segment, determine the target temperature trend prediction model; Input the thermal state characterization results and thermal characteristic parameters corresponding to the current braking condition segment into the target temperature trend prediction model to obtain the brake temperature change trend prediction results within the preset time window. When a change in braking condition category is detected, the corresponding target temperature trend prediction model is redefined, and the temperature change trend prediction results are updated.

7. The method for predicting thermal fade of tractor brakes according to any one of claims 1-6, characterized in that, Determine whether the risk level of thermal degradation has reached the preset safety control conditions; If so, a brake control adjustment command is generated, and the brake force distribution mode and / or braking mode are adjusted according to the brake control adjustment command to reduce the risk of brake fade. If not, the current braking force distribution and braking method will be maintained, and the multi-source state perception data during the tractor braking process will continue to be collected and the thermal fade risk level will be updated and judged.

8. The method for predicting the thermal fade of tractor brakes according to claim 7, characterized in that, The method for generating braking control adjustment commands is as follows: Obtain the operating condition category and thermal fade risk level corresponding to the current braking operating condition segment; The corresponding brake control adjustment level is determined based on the risk level of heat fade. The braking control adjustment strategy is determined based on the braking control adjustment level and the operating condition category. The corresponding braking control adjustment command is generated based on the braking control adjustment strategy.

9. The method for predicting the thermal fade of tractor brakes according to claim 8, characterized in that, The method for adjusting the braking force distribution mode based on the braking control adjustment command is as follows: In response to the brake control adjustment command, the thermal fade risk status of each wheel brake is obtained; The braking force distribution adjustment parameters are determined based on the heat fade risk status of each wheel brake. Based on the braking force distribution adjustment parameters, the braking force distribution ratio corresponding to each wheel brake is adjusted. The vehicle stability constraint parameters are determined based on wheel speed data, vehicle speed data, and slope data, and the adjusted braking force distribution ratio is then constrained and corrected based on these parameters. The target braking force distribution method is determined based on the revised braking force distribution ratio.

10. The method for predicting the thermal fade of tractor brakes according to claim 8, characterized in that, The method for adjusting the braking mode based on the braking control adjustment command is as follows: When the safety control condition determination result indicates that the preset safety control conditions are met, the current auxiliary braking capability status of the tractor is obtained; The target braking mode is determined based on the auxiliary braking capability status and the vehicle operating status, wherein the target braking mode includes a compound braking mode and an intermittent braking mode; A braking control adjustment command is generated based on the target braking mode, and the current braking mode is adjusted according to the braking control adjustment command; Monitor the actual braking status of the vehicle corresponding to the adjusted braking method and determine whether it meets the safety braking requirements; When the safety braking requirements are not met, the braking mode is restored to the state before adjustment to obtain the final braking control result.

Images