Brake-by-wire assistance function assistance hysteresis software control method, medium and product

Abstract

The application provides a brake-by-wire boost function boost hysteresis software control method, a medium and a product, and belongs to the technical field of automobile chassis electric control. The method comprises the following steps: a pedal state monitoring network is constructed, a push rod displacement is collected, and an original pedal state parameter data set is generated; a go path and a return path reference boost curve is constructed based on the original parameter, and a double-path reference boost mapping data set is generated; a dynamic scaling factor is introduced to process the return path displacement, and a return path preliminary braking force data set is generated; the traditional double-fixed curve simulation mode is broken through, the return path hysteresis is realized through a single go path reference curve and a nonlinear scaling factor, the calibration workload is reduced by more than 50%, and the software reuse rate is improved.

Description

Technical Field

[0001] This invention proposes a software control method, medium, and product for brake-by-wire power assist hysteresis, belonging to the field of automotive chassis electronic control technology. Background Technology

[0002] With the increasing demand for vacuum-free braking systems in new energy vehicles, brake-by-wire systems are gradually replacing traditional vacuum-assisted braking systems. Due to the mechanical structure characteristics of traditional vacuum boosters, there is a boost lag during the braking pedal depressing (outward stroke) and releasing (return stroke). The same pedal displacement corresponds to different output braking forces, and this characteristic has become a core component of the normal pedal feel that drivers are accustomed to.

[0003] Existing brake-by-wire systems mostly use fixed mapping curves to simulate hysteresis, such as setting independent fixed assist gain curves for outward and return strokes. However, this has significant drawbacks: First, static mapping cannot adapt to dynamic conditions such as driver operating habits and vehicle load, resulting in poor consistency in pedal feel; second, when releasing the pedal for return, it tracks in the reverse direction of the outward stroke, which can easily lead to a sudden drop in braking force and a feeling of collapse; third, the residual displacement after the pedal is fully released is handled roughly (such as forced zeroing), causing system vibration or abnormal pedal rebound; fourth, it requires modifying the hardware structure or developing multiple versions of software to meet different customer needs, resulting in long development cycles and high costs.

[0004] Therefore, there is an urgent need for a software-based dynamic hysteresis control scheme to reproduce the traditional braking feel without a vacuum source, while solving the rigidity and adaptability problems of existing technologies. Summary of the Invention

[0005] This invention provides a software control method, medium, and product for brake-by-wire assist function with assist hysteresis, in order to solve the problems mentioned in the background art above:

[0006] The drive-by-wire assist hysteresis software control method proposed in this invention includes:

[0007] S1: Construct a pedal status monitoring network to collect push rod displacement. and direction of motion Generate the original pedal state parameter dataset;

[0008] S2: Construct outbound and inbound baseline assist curves based on the original parameters to generate a dual-path baseline assist mapping dataset;

[0009] S3: Introducing a dynamic scaling factor Process the return displacement to generate a preliminary return braking force dataset;

[0010] S4: Optimize braking force by limiting Xmax by maximum offset, and generate a constrained target braking force dataset;

[0011] S5: Perform gradient decay processing on the residual displacement to generate the final braking force dataset;

[0012] S6: Encapsulate core calibration parameters and combine them with driver operation data for self-learning to generate an adaptive optimization calibration parameter set;

[0013] S7: Full-condition verification and optimization, solidifying the final control strategy for linear control brake assist hysteresis.

[0014] The present invention proposes a non-transitory computer-readable storage medium storing a computer program that is executed by a processor to implement the assist hysteresis software control method for brake assist function as described above.

[0015] The present invention proposes a computer program product, comprising a computer program / instruction, which, when executed by a processor, implements the assist hysteresis software control method for brake assist function as described above.

[0016] The beneficial effects of this invention are as follows:

[0017] The above technical solution breaks through the traditional dual-fixed-curve simulation method. It achieves return hysteresis by using a single outgoing reference curve in conjunction with a nonlinear scaling factor, reducing calibration workload by more than 50% and improving software reusability. Based on braking safety standards, Xmax is calibrated to balance pedal feel reproduction and braking safety, avoiding sluggish braking response caused by excessive hysteresis. By replacing the traditional forced zeroing, residual displacement is handled with a calibrated decay rate, eliminating pedal vibration and abnormal rebound, and improving the smoothness of the release end. By encapsulating core parameters into configurable templates, it supports CAN / OTA dual-interface updates, adapting to different vehicle models without modifying the mechanical structure, shortening the development cycle by 30% to 40%. Based on operating characteristics, parameters are dynamically fine-tuned to achieve a personalized pedal feel that becomes more comfortable with use, breaking the limitations of traditional fixed calibration. Attached Figure Description

[0018] Figure 1 This is a diagram of the method described in this invention. Detailed Implementation

[0019] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0020] One embodiment of the present invention, such as Figure 1 As shown, the software control method for assist hysteresis in brake-by-wire function includes:

[0021] S1: The monitoring dimensions of the brake pedal actuator in the brake-by-wire system of new energy vehicles are divided (core dimensions include push rod displacement and direction of movement). Based on the division results, high-precision displacement sensors and a direction of movement recognition module are deployed to construct a closed-loop pedal status monitoring network. The continuous displacement signal of the brake push rod is collected in real time through this monitoring network. (Unit: mm) and pedal movement direction signal (in Indicates the outward movement of the pedal being pressed. (This represents the return stroke after the pedal is released). The collected discrete signals are aligned according to the timestamps to generate the original pedal state parameter dataset.

[0022] S2: Extract the push rod displacement based on the generated original pedal state parameter dataset. The effective value range (typically 0 to 20 mm) is combined with the hysteresis characteristic curve of a traditional vacuum booster (the rate of increase of braking force with displacement is higher on the outward stroke than on the return stroke). Two independent reference booster curves are constructed through vehicle calibration and fitting: the outward reference booster curve. (Describe the relationship between displacement and target braking force during the outward journey), return journey reference assist curve (Describe the initial correspondence between displacement and target braking force during the return stroke); make the two curves form a closed hysteresis loop in the low to medium stroke range (0 to 10 mm) to reproduce the nonlinear characteristics of the traditional booster and generate a dual-path benchmark booster mapping dataset;

[0023] S3: Based on the generated dual-path baseline-assisted mapping dataset, focusing on the return journey condition ( Design a dynamic scaling mechanism: Define a non-linear dynamic scaling factor. ∈(0.5,1], this factor varies with the displacement of the push rod The increase shows a decreasing trend (small displacement region) Approximately 1 to ensure release sensitivity, large displacement range Reduce to 0.5 to 0.7 to enhance release gentleness); through For the current push rod displacement Nonlinear compression is performed to generate an equivalent return displacement. = × ;Will Substitute the outbound reference assist curve The initial target braking force under the return trip condition is calculated, and the initial braking force dataset for the return trip is generated.

[0024] S4: Based on the generated equivalent backflush displacement Based on the initial braking force dataset during the return stroke, set the maximum permissible hysteresis offset Xmax (ranging from 0.8 to 5.0 mm to avoid excessive hysteresis leading to drag braking); calculate the actual hysteresis offset. - ,like If the value is greater than Xmax, then the equivalent return displacement will be forcibly adjusted to... = -Xmax, and based on Recalculate the target braking force; if If the target braking force is less than or equal to Xmax, then the initial target braking force of S3 is retained, and the constrained target braking force dataset is finally generated.

[0025] S5: Based on the generated constrained target braking force dataset, monitor the push rod displacement in real time. Does it approach 0 (i.e., pedal fully released): If detected →0, but the system still has a small residual displacement. ( The sensor accuracy threshold, typically between 0.1 and 0.3 mm (caused by sensor noise or mechanical backlash), triggers a gradient decay mechanism—at a constant negative rate. The equivalent return displacement is gradually reduced by Kmm / s (K∈0.5-2.0mm / s, which can be calibrated according to the vehicle's inertia). until =0; This process only clears the internal state, does not trigger additional braking actions, and generates the final braking force dataset after the residual displacement is reduced to zero;

[0026] S6: The dynamic scaling factor curve of S3 The Xmax of S4 and the decay rate K of S5 are encapsulated as a set of configurable core calibration parameters, supporting updates via CAN bus or OTA (over-the-air). Simultaneously, based on historical data from S1 to S5, driver operating characteristics (such as pedal speed, common travel range, and release pattern) are recorded, and dynamic fine-tuning is achieved through machine learning algorithms (such as gradient descent). The slope and inflection point of the curve allow the pedal feel to adapt to the driver's habits, generating an adaptive set of optimized calibration parameters;

[0027] S7: Based on the adaptive optimization calibration parameter set generated by S6, real vehicle tests are conducted under different operating conditions (such as no-load / full-load, dry road surface / slippery road surface, low-speed start-stop / high-speed braking) to collect performance indicators such as brake pedal feedback force and braking force response delay; the test results are compared with the pedal feel data of traditional vacuum boosters, and iterative optimization is performed. Parameters such as Xmax and K are adjusted until the preset foot feel consistency standard is met; finally, the optimized control parameters and algorithm logic are solidified to form the final control strategy for linear brake assist hysteresis.

[0028] The working principle and effects of the above technical solution are as follows:

[0029] By replicating the lag feel of traditional vacuum boosters, drivers do not need to readjust to braking operations when switching to new energy vehicles, greatly reducing the unfamiliarity with operation after switching vehicles and improving drivers' adaptability to braking in new energy vehicles.

[0030] By adapting the release requirements of different displacement segments to a dynamic scaling factor, the problem of the collapse feeling of traditional brake-by-wire return stroke is solved, enhancing driving comfort in urban frequent start-stop scenarios and improving the smoothness of return brake release.

[0031] By limiting the maximum hysteresis offset, for example, to 5mm, the system avoids dragging caused by driving against the driver's deceleration intention and excessive hysteresis, reducing the risk of sluggish braking response. At the same time, the residual displacement is gradually decayed, reducing pedal vibration caused by sensor noise and mechanical clearance, enhancing system stability and improving braking system safety.

[0032] The core parameters are all calibrated by software, eliminating the need to modify the mechanical structure for different customer feel requirements, reducing the number of hardware variants, and supporting OTA remote updates, shortening the cycle of later parameter optimization, and reducing development and production costs.

[0033] By fine-tuning parameters through a self-learning algorithm to adapt to different drivers' operating habits, the limitations of fixed calibration on personalized needs are reduced. It can also adapt to multiple working conditions such as no-load / slippery road surfaces, enhance the consistency of braking feel in all scenarios, and improve the system's adaptability.

[0034] In another embodiment of the present invention, step S1 further includes:

[0035] S11: Based on the motion characteristics of the brake pedal actuator in the brake-by-wire system of new energy vehicles (such as the linear reciprocating motion trajectory of the push rod and the correlation between pedal force and displacement), and combined with the core influencing parameters (displacement and direction of motion) of the hysteresis characteristics of traditional vacuum boosters, the functional correlation method is used to divide the monitoring dimensions—the core dimensions include push rod displacement (reflecting pedal travel) and direction of motion (distinguishing between outward and return strokes), and the auxiliary dimensions include pedal depress speed (for subsequent self-learning optimization); output the brake pedal monitoring dimension division scheme, clarifying the definition, unit and monitoring priority of each dimension.

[0036] S12: Based on the division scheme of S11, a high-precision magnetoelectric displacement sensor is selected and deployed on the side of the push rod movement trajectory. The sensor signal is directly collected by the ECU, and the brake pedal is determined to be pressed or released by the increase and decrease of the stroke.

[0037] S13: The ECU acquires the continuous displacement signal of the brake push rod. (Unit: mm); The acquired raw signal is initially filtered (first-order low-pass filter is used to eliminate high-frequency electromagnetic interference) to output the original discrete displacement signal sequence;

[0038] S14: Extract the dual-channel displacement signal from the sensor after first-order low-pass filtering in S13 (denoted as...). , ) and dual-path motion direction signals (denoted as , ); for dual-channel displacement signals and Perform deviation verification and set a displacement deviation threshold. =0.05mm (based on sensor resolution ≥0.01mm, set 5 times the accuracy, balancing sensitivity and anti-interference): If If the value is greater than xmm, the ECU identifies it as a sensor fault and triggers a fault marker. For each sensor signal, the ECU synchronously checks the corresponding alarm status (e.g., single-channel signal loss, signal exceeding range (<0mm or >20mm), abnormal signal transition). If any alarm is detected, the ECU records the alarm type (e.g., "Channel 1 signal lost") and temporarily uses another signal without an alarm as a temporary input (if both channels alarm, the previous valid displacement and direction data is used to avoid data interruption). For displacement and direction signals that pass verification (or enable temporary valid signals), they are matched one by one according to the ECU's unified clock stamp, and discrete data with a timestamp deviation exceeding 5ms are removed (to avoid parameter mismatch caused by timing misalignment). The aligned "timestamp-valid displacement-movement direction-sensor status (normal / fault type)" is stored in a four-tuple format to generate the original pedal status parameter dataset. At the same time, sensor fault information and alarm records are synchronously written to the ECU fault log. The working principle and effect of the above technical solution are as follows:

[0039] The functional correlation method is used to first grasp the core dimensions that affect hysteresis, such as push rod displacement and direction of motion, while leaving the pedal speed for subsequent self-learning. This avoids blindly monitoring irrelevant parameters, reduces the amount of ineffective work in subsequent data processing, and allows the following steps to focus precisely on the key factors of hysteresis control, thus improving the targeting of the monitoring dimensions.

[0040] The selected magnetoelectric sensor can withstand operating conditions from -40℃ to 85℃. Temperature, humidity and vibration interference in the braking system have little impact on it, reducing the deviation of data collected in harsh environments. At the same time, the ECU directly assists in determining the pedal direction by increasing or decreasing the travel distance. In conjunction with the Hall module, the direction recognition is more timely, eliminating the need to wait for communication delays between multiple modules, improving the real-time performance of direction determination and enhancing the environmental adaptability of data acquisition.

[0041] First-order low-pass filtering can filter out high-frequency electromagnetic interference in the braking system, reduce noise mixed into the displacement and direction signals, and make the acquired signals closer to the actual pedal movement state. There is no need to spend time cleaning up signal noise later, laying a clean foundation for the subsequent reference curve and improving the purity of the original signal.

[0042] Dual-channel signal deviation verification (fault report for deviations exceeding 0.05mm) enables timely detection of sensor problems. Even if one channel alarms, the other temporary signal can be used. If both channels fail, historical data is used to avoid data gaps and reduce dataset failures caused by sensor malfunctions. Timestamp alignment removes 5ms abnormal data, improving data temporal consistency and reducing errors when fitting baseline curves. Synchronous fault logging facilitates troubleshooting during later maintenance, reduces the difficulty of fault diagnosis, and enhances the stability and maintainability of data acquisition.

[0043] In another embodiment of the present invention, step S2 further includes:

[0044] S21: During the sensor's production line exit phase, 100% accuracy monitoring has been completed. The ECU reads this exit monitoring data and confirms that the sensor's output accuracy meets the braking hysteresis control requirements. Before using the preset range, the ECU simultaneously verifies the validity of the displacement signal in the original pedal state parameter dataset generated in S14. If the displacement signal exceeds the preset range of 0 to 20 mm, or the sensor provides an accuracy alarm (such as a substandard accuracy marker in the exit monitoring data or abnormal real-time signal fluctuations), the ECU identifies it as a sensor fault and triggers fault recording (recording the fault type as "displacement signal out of range" or "accuracy abnormality"). The system temporarily uses the boundary values ​​of the preset range (20mm when exceeding the upper limit, and 0mm when exceeding the lower limit) as the temporary effective displacement. Combining the sensor's preset range (0 to 20mm) with historical braking operation data (accumulated pedal usage frequency from S1 to S5), it is statistically determined that the common travel range corresponding to more than 95% of braking operations is 0 to 15mm. The information is organized in the format of "preset effective range (0 to 20mm) - common travel range (0 to 15mm) - sensor accuracy status - fault identification logic" to generate a report on the effective value range of push rod displacement, clarifying the boundaries of each range and the rules for ECU access and fault handling.

[0045] S22: Select a traditional vacuum booster from a fuel vehicle of the same class (model matching the braking requirements of new energy vehicles), and collect its hysteresis characteristic data through bench testing—control the push rod to be pressed down (outward stroke) and released (backward stroke) at a constant speed of 0.5 mm / s, record the output braking force corresponding to different displacements, obtain the displacement-braking force forward and back program sequences, and verify that the outward braking force increases with displacement at a rate (average 0.8 N / mm) higher than the backward braking force (average 0.6 N / mm); output the hysteresis characteristic dataset of the traditional vacuum booster;

[0046] S23: Using the effective range of S21 as the horizontal axis and the hysteresis characteristic data of S22 as the vertical axis, a polynomial fitting algorithm (least square method) is used to fit the outbound and return baseline curves respectively: Outbound baseline assist curve = +5x (adapts to the increasing braking force characteristics on the outward trip), return trip baseline assist curve = +4x (adapts to the decreasing braking force characteristics of the return stroke); the accuracy of the curve is verified through residual analysis (fitting residual ≤0.5N), and the reference assist curve equations for the outgoing / return strokes are output;

[0047] S24: The two baseline curves fitted by S23 are verified for closure in the low-to-medium stroke range of 0 to 10 mm (ensuring that the difference between the outgoing braking force and the return braking force forms a stable hysteresis loop under the same displacement); the data is organized in the format of displacement-outgoing braking force-return braking force to generate a dual-path baseline assist mapping dataset, providing a baseline model for subsequent return hysteresis adjustment.

[0048] The working principle and effects of the above technical solution are as follows:

[0049] The sensor is pre-set to have a displacement range of 0 to 20 mm at the factory, which the ECU can directly call up without having to perform fluctuation range statistics on the original dataset, saving redundant data analysis steps; at the same time, the sensor has undergone 100% accuracy monitoring when it leaves the production line, and the ECU can directly confirm that the accuracy is ≥0.01 mm, which lays a reliable accuracy foundation for fitting the reference curve, reduces curve errors caused by insufficient sensor accuracy, and improves the efficiency of determining the effective displacement range.

[0050] The ECU will verify the displacement signal. If it exceeds the range of 0 to 20 mm or the sensor reports an accuracy problem, it will record the fault and temporarily replace it with the boundary value of 0 mm or 20 mm to avoid data gaps and reduce the risk of dataset failure. It can also statistically identify the commonly used range of 0 to 15 mm corresponding to 95% of braking operations, allowing subsequent parameter tuning to focus more on the key range, without wasting computing power in invalid ranges, and enhancing the system's fault tolerance to sensor anomalies.

[0051] During bench testing, the control push rod is pressed and released at a constant speed of 0.5 mm / s, eliminating the interference of speed fluctuations on braking force. It can accurately measure the difference in the rate of increase of braking force during the outward stroke (0.8 N / mm) and the return stroke (0.6 N / mm), reducing the deviation when simulating the braking feel of a fuel vehicle in a new energy vehicle and improving the reliability of the hysteresis data of the traditional vacuum booster.

[0052] The least squares method was used for fitting, and the residual was controlled within 0.5N, so that the outgoing and returning curves could accurately match the change law of braking force with displacement, better restore the nonlinear characteristics of traditional boosters, avoid distortion in subsequent hysteresis simulation, and improve the fitting accuracy of the benchmark boost curve.

[0053] Closed-loop verification was performed on the low-to-medium stroke range of 0 to 10 mm to ensure that the difference between the braking force of the downstroke and return stroke at the same displacement can form a stable hysteresis loop. This provides a precise reference for the return hysteresis adjustment of S3, reduces the trouble of repeatedly modifying parameters due to defects in the reference model during subsequent adjustments, and enhances the reliability of the dual-path reference model.

[0054] In another embodiment of the present invention, step S3 further includes:

[0055] S31: Bench tests and user surveys were conducted to address the return stroke issue of existing brake-by-wire systems. When the brakes were released directly in the reverse direction along the outward curve, the braking force decreased at a rate of 1.2 N / mm in the 10 to 5 mm displacement range, with 70% of users reporting a noticeable collapse sensation. The return stroke control requirements were clarified: in the small displacement range (0 to 5 mm), the braking force needs to decrease slowly (sensitivity), and in the large displacement range (5 to 20 mm), the rate of decrease needs to be further reduced (gentleness). A return stroke brake feel specification was also developed.

[0056] S32: Dynamic scaling factor The control target is the "reduction in return stroke"—that is, the actual displacement of the pedal during the return stroke. After scaling this reduction, the outward braking force is matched according to the scaled equivalent travel reduction (e.g., when i=0.5, an actual return stroke of 1mm corresponds to an equivalent travel reduction of 0.5mm; the braking force corresponding to a 0.5mm displacement in the outward stroke is greater than the braking force directly matched for a 1mm return stroke, thus creating the difference in force required for hysteresis). The return stroke is then divided into three intervals based on the return displacement characteristics (combining the pedal's common travel and safety requirements), clearly defining each interval. Value selection rules:

[0057] When the return displacement x ≤ 1mm: Set to 1 to ensure that when the pedal is nearly fully released (displacement approaches 0), the reduction in travel does not scale, and the braking force drops to 0 synchronously with the displacement, thus meeting the safety requirement of "brake fully released when displacement is 0".

[0058] When the return displacement x ≥ 15mm: The value is fixed at 0.5 and will not be reduced further to avoid excessive scaling during the large displacement return stroke, which would result in too much braking force being retained and to prevent violating the driver's intention to "significantly reduce force when the pedal is released significantly" in deceleration.

[0059] When the return displacement 1 < x < 15 mm: Instead of a linear non-curing curve, on-vehicle point calibration is performed according to the pedal feel preferences of different customers (such as being more sensitive or more gentle). By collecting the braking operation data of customers in this interval (such as the expected braking force feedback corresponding to different displacements), at least 8 characteristic points are selected (such as key displacements like 2 mm, 5 mm, 8 mm, 10 mm, 12 mm, etc.), and a non-linear curve adapted to the customer's needs is generated using polynomial fitting or spline interpolation algorithms;

[0060] Load the preliminarily designed curve (including the fixed head and tail segments and the customer-customized middle segment) into the ECU for on-vehicle road tests (covering scenarios such as urban commuting and releasing the pedal on the highway), collect the driver's subjective evaluations of the pedal feel for each displacement segment (such as "Is the force drop smooth when the return is 5 mm?" "Is there a sudden feeling when the return is 12 mm?"), and fine-tune the coordinates of the fitting points in the middle segment according to the feedback until the pedal feel meets the customer's expectations; Sort out the parameters of the finally determined i(x) curve (including the value-taking rules for each interval, the fitting equation for the middle segment, and the coordinates of the characteristic points), form a dynamic proportional scaling factor curve model, clearly mark the design purposes and customer-customized attributes of each interval, and provide a basis for the calculation of the equivalent return displacement of S33;

[0061] S33: For the return working condition ( ), extract the current push rod displacement from the original dataset of S14, substitute it into the curve model of S32, and calculate the equivalent return displacement according to the formula = + × ; Perform a range check on the calculation result (ensure ≥0), and eliminate abnormal negative displacement values; Output the equivalent return displacement sequence;

[0062] S34: Substitute the equivalent return displacement of S33 into the return baseline boost curve of S23 to calculate the preliminary target braking force at different times; Sort out the data in the format of timestamp-displacement-equivalent displacement-preliminary braking force to generate a return preliminary braking force dataset.

[0063] The working principle and effects of the above technical solutions are as follows:

[0064] Through bench tests, the problem of the excessive decline of the braking force at 1.2 N / mm in the displacement range of 10 to 5 mm is accurately captured. Combining with the pain point of "sinking feeling" feedback from 70% of users, the core requirements of being sensitive in the small displacement segment and gentle in the large displacement segment are clarified. It is not based on experience, reducing rework caused by deviation in the subsequent design direction, giving a clear goal for the subsequent scaling factor design, and improving the accuracy of the return control requirements;

[0065] Designed in three sections based on return displacement. Very reasonable: when x≤1mm, fix it at 1 to ensure that the braking force drops to 0 when the pedal is close to 0, so that there is no braking residue; when x≥15mm, fix it at 0.5 and do not drop it again to avoid too much braking force when releasing the pedal with a large displacement, so as not to violate the driver's intention of "the more you release the pedal, the more force you should reduce", and reduce safety risks; the middle 1 to 15mm segment is a non-linear curve calibrated on the actual vehicle according to customer preferences, unlike the previously fixed linear curve, which can adapt to the different foot feel needs of different customers. For example, customers who prefer more sensitive pedals can adjust it to be steeper, and those who prefer softer pedals can adjust it to be smoother, which improves the personalization and enhances the dual adaptability of the scaling factor to safety and foot feel;

[0066] The calculation uses "reduction in return trip distance". "Using a base value is better than directly using displacement." It is more in line with the scaling logic, and also performs range verification to remove negative displacement outliers, so as to avoid abnormal data from interfering with subsequent braking force calculation, reduce the error of the initial target braking force, provide reliable basic data for return braking force adjustment, and improve the calculation accuracy of equivalent return displacement.

[0067] By reusing the proven accuracy of the outbound reference curve of S23 to calculate the initial braking force, there is no need to build a separate return braking force model. This ensures the consistency of the braking force calculation benchmark, reduces modeling deviation, and makes the pedal feel of the new energy vehicle during the return trip more similar to that of the traditional vacuum booster that the driver is familiar with. When changing vehicles, there is no need to readjust, which reduces adaptation costs and the deviation of the return pedal feel from that of traditional fuel vehicles.

[0068] In another embodiment of the present invention, step S4 further includes:

[0069] S41: Based on the braking safety standard (GB7258-2024 "Technical Conditions for Safe Operation of Motor Vehicles"), bench tests were conducted to verify the impact of different offsets on braking performance. When the offset exceeds 5.0mm, the brake release delay time exceeds 0.3s, posing a risk of dragging the brakes. When the offset is less than 0.8mm, the hysteresis simulation is insufficient, and the difference in pedal feel is significant. The final calibration range for Xmax is 0.8 to 5.0mm, with a specific value of 1.2mm determined for the target vehicle model. The calibration result for the maximum permissible offset of hysteresis is output.

[0070] S42: Dynamic scaling factor The control target is the "return stroke reduction" (i.e., the actual displacement of the pedal from its current position), through... After scaling the reduction amount, match the braking force curve of the forward stroke according to the "scaled equivalent stroke reduction amount". For example, when i = 0.5, the actual return stroke of 1 mm corresponds to an equivalent stroke reduction amount of 0.5 mm. Since the braking force corresponding to a displacement of 0.5 mm in the forward stroke is greater than the braking force directly matched by a return stroke of 1 mm, the force difference formed by the two is hysteresis, ensuring that the hysteresis effect conforms to the characteristics of a traditional vacuum booster;

[0071] According to the requirements of return stroke displacement safety and pedal feel, divide it into three displacement intervals and clarify The value-taking rules to avoid being fixed as a linear curve:

[0072] Small displacement section (return stroke displacement x ≤ 1 mm): Fix it as 1. At this time, there is no scaling of the return stroke reduction amount (the actual reduction of 1 mm corresponds to an equivalent reduction of 1 mm), ensuring that when the pedal displacement approaches 0, the braking force synchronously drops to 0 with the equivalent reduction amount, meeting the safety requirement of "complete release of braking when the displacement is 0", and avoiding the residual braking force affecting the vehicle's gliding;

[0073] Large displacement section (return stroke displacement x ≥ 15 mm): Fix it as 0.5 and no longer decrease. At this time, the actual return stroke of 1 mm corresponds to an equivalent reduction of 0.5 mm, and the degree of braking force retention is moderate, avoiding being too low, resulting in too small equivalent reduction amount and too much retained braking force, violating the driver's deceleration intention of "significantly reducing the braking force when the pedal is released largely", and ensuring the safety of braking operation;

[0074] Middle interval section (return stroke displacement 1 < x < 15 mm): Calibrate the actual vehicle by taking points according to the pedal feel preferences of different customers (such as sensitive type, gentle type). Select at least 10 characteristic displacement points (such as 2 mm, 4 mm, 6 mm, 8 mm, 10 mm, 12 mm, 14 mm, etc.) in this interval, collect the "expected braking force feedback" corresponding to each point through actual vehicle testing, and adjust the value of each point, and then use the cubic spline interpolation or polynomial fitting algorithm to generate a non-linear curve to ensure smooth transition of the pedal feel within the interval and adapt to the customized needs of customers;

[0075] Load the preliminarily calibrated curve into the ECU, conduct actual vehicle road tests in typical scenarios such as urban congestion (frequent small return strokes) and highway cruising (large return strokes and pedal release), collect the driver's subjective evaluations (such as "is it sensitive when the return stroke is 1 mm", "is there a dragging feeling when the return stroke is 15 mm", "is the force reduction smooth in the middle section"), and fine-tune the value of the characteristic points in the middle interval according to the feedback until the pedal feel meets the customer's expectations and there is no safety risk; Organize the finally determined curve parameters (including the division rules of the three intervals, the fitting equation of the middle section, and each characteristic point The values ​​are marked with the design purpose of each interval (such as ensuring release in small displacement segments and avoiding violating the deceleration intention in large displacement segments) and customer-customized attributes, forming a dynamic scaling factor curve model, which provides an accurate basis for S33 to calculate the equivalent return displacement.

[0076] S43: Compared to S42 With S41's Xmax: If If the value is greater than Xmax, then the equivalent return displacement will be forcibly adjusted to... = +Xmax, substitute Recalculate the target braking force; if If Xmax is less than or equal to Xmax, then the initial target braking force of S34 is retained; the adjusted equivalent displacement sequence and the recalculated target braking force sequence are output.

[0077] S44: Merge the initial braking force retained in S43 with the recalculated target braking force according to timestamps, and remove duplicate data and outliers (points where braking force fluctuations exceed 5N); store the data in the format of timestamp-displacement-equivalent displacement-constrained braking force to generate a constrained target braking force dataset.

[0078] The working principle and effects of the above technical solution are as follows:

[0079] Based on the GB7258-2024 standard, bench testing was used to clarify the offset boundaries: an offset exceeding 25.0mm would result in a brake release delay of more than 0.3s (risk of dragging the brake), while an offset below 0.8mm would result in insufficient hysteresis simulation (significant difference in pedal feel). Finally, the Xmax range of 0.8 to 25.0mm was calibrated and the specific value of 1.2mm for the target vehicle model was determined. This avoids the blindness of setting parameters based on experience, reduces safety hazards and pedal feel problems caused by excessive or insufficient hysteresis, and improves the balance accuracy between braking system safety and pedal feel.

[0080] right Using a sliding window average of 5 data points for time-series smoothing effectively eliminates the interference of instantaneous fluctuations on offset judgment, ensuring the calculated... It more closely reflects the actual hysteresis state, reduces the incorrect adjustment of braking force caused by signal fluctuations, provides an accurate basis for subsequent braking force optimization, and improves the reliability of the calculation of actual hysteresis offset.

[0081] according to The system handles different situations depending on the size of Xmax: when the limit is exceeded, the equivalent return displacement is forcibly adjusted and the braking force is recalculated; when the limit is not exceeded, the initial braking force is retained. This avoids residual braking force caused by excessive hysteresis (which violates the intention to decelerate) and also prevents distorted pedal feel caused by insufficient hysteresis. It reduces driving discomfort caused by abnormal braking force output and enhances the rationality of the target braking force output.

[0082] Removing duplicates and outliers with braking force fluctuations exceeding 5N during data merging filters out invalid interference data, allowing the dataset to more accurately reflect the actual braking state. This reduces errors in subsequent residual displacement processing (S5), laying a reliable foundation for generating stable final braking force data and improving the effectiveness of the constrained braking force dataset. In another embodiment of the invention, step S5 further includes:

[0083] S51: Perform threshold judgment on the original pedal state parameter dataset of S14 and set the push rod displacement. When the distance is ≤0.5mm (5 times the minimum resolution threshold of the sensor) and the duration is ≥100ms, the pedal is determined to be fully released; output pedal release status determination signal (1 indicates release, 0 indicates not released);

[0084] S52: Extract pedal motion velocity from the original pedal state parameter dataset of S14 (calculate the displacement change rate during the return phase). Set a rapid release threshold (based on the braking system response characteristics). ≥2mm / s): If the detected pedal release speed meets this threshold, and S51 has determined "actual push rod displacement" If the value is 0, it is determined to be a "rapid release scenario," triggering the displacement offset processing logic. For the rapid release scenario, the ECU reads the "displacement offset corresponding to the braking force output" from the current hysteresis control logic (i.e., the "virtual displacement deviation" retained due to software control lag and hysteresis control, not the mechanical residual displacement): if this displacement offset... (If the braking force is still output with a 1mm offset), it is determined to be "residual braking force offset that needs to be dealt with quickly" - at this time the driver has released the pedal and the residual braking force needs to be eliminated first;

[0085] Based on the requirement of "rapidly releasing braking force", a rapid attenuation slope for the displacement offset is set (with a value ≥5mm / s, much higher than the attenuation rate K of the conventional residual displacement, to ensure rapid elimination of offset). The displacement is then... The braking force rapidly decreases from its current value to 0; during the decay process, the ECU synchronously controls the braking force to decrease as the offset decreases, avoiding shocks caused by a sudden drop in braking force; real-time monitoring. Changes, when detected When =0, the ECU immediately disables the current hysteresis control logic (stops calling the hysteresis calculation modules of S3 to S4) to ensure that the braking force is fully released without any additional hysteresis intervention; records "fast release scenario identifier - initial value of displacement offset - attenuation slope - attenuation time - hysteresis control disabled state", generates displacement offset detection and fast processing results, and provides "the basis for zeroing the braking force in the fast release scenario" for the generation of the final braking force dataset of S54.

[0086] S53: The negative sign of the decay rate K indicates the "direction of decreasing displacement offset". A value of 10mm / s represents a rapid decay rate—this rate is only applicable to the "rapid release scenario" determined by S52 (pedal release speed ≥ 2mm / s and actual displacement is 0). It is used to quickly eliminate residual braking force offset caused by software lag and delayed control, unlike the slow decay in normal scenarios, ensuring that the braking force can be quickly and synchronously cleared to zero after the driver releases the pedal. Combining the braking force response characteristics of the brake-by-wire system (the braking force sensitivity to changes in displacement offset is approximately 5N / mm), calculated at a decay rate of K = -10mm / s: if the maximum residual displacement offset is 2mm during rapid release, it only takes 0.2s to decay the offset from 2mm to 0, corresponding to a synchronous and rapid decrease in braking force from 10N (2mm × 5N / mm) to 0. This avoids the impact caused by a sudden drop in braking force and meets the requirement of "no residual braking force after rapid release". This rate verifies the effectiveness of the method. The balance and adaptability between "rapid zeroing" and "smooth braking" were assessed. A parameter of K=-10mm / s was loaded into the ECU, and 20 repeated real-vehicle tests were conducted in rapid release scenarios (such as rapidly releasing the pedal from 15mm to 0 during high-speed cruising). The decay curve of the displacement offset in each test was collected. If the decay time of all tests was stable between 0.15 and 0.25s (corresponding to a normal residual range of 1.5 to 2.5mm offset), and there were no decay rate fluctuations exceeding ±1mm / s, it was determined that this rate could be stably executed in the system. The calibration data was compiled, clarifying the value of K as -10mm / s, noting its usage limitation of "only applicable to rapid release scenarios," decay direction (negative), suitable displacement offset range (0 to 2.5mm), and the associated logic of "triggering hysteresis control disabling when decaying to 0," generating a gradient decay rate calibration report to provide accurate parameter basis for the displacement offset decay execution of S54.

[0087] S54: For the residual displacement that needs to be processed in S52, gradually decrease the equivalent return displacement at a rate of -K. until This process only updates the internal state and does not send braking force commands to the brake actuator; it records the timestamp-residual displacement-equivalent displacement data during the zeroing process, merges it with the brake force data after the constraint of S44, and generates the final brake force dataset after the residual displacement is zeroed.

[0088] The working principle and effects of the above technical solution are as follows:

[0089] By using the dual conditions of "push rod displacement ≤ 0.5mm + duration ≥ 100ms", we avoid misjudgments caused by instantaneous fluctuations of the sensor (such as the slight rebound when pressing the pedal), reduce subsequent processing errors caused by misjudgment of the release state, make the release judgment more in line with the actual operation of the driver, and improve the accuracy of the pedal release state judgment.

[0090] It is specifically designed to identify rapid scenarios where the release speed is ≥2mm / s and the displacement is 0. For the virtual displacement offset left by software lag, it uses a fast decay slope of ≥5mm / s to clear it to zero. This is much faster than the conventional rate, which can quickly eliminate the residual braking force (avoiding the brake drag after the driver releases the pedal). At the same time, it controls the braking force to decrease with the offset to prevent the impact caused by the sudden drop, reduce driving discomfort during rapid release, and enhance the efficiency and smoothness of clearing the braking force in rapid release scenarios.

[0091] It is clear that K=-10mm / s is only used for rapid release scenarios. Through 20 real vehicle tests, it was verified that under the normal residual offset (1.5 to 2.5mm), the decay time is stable at 0.15 to 0.25s, and the rate fluctuation does not exceed ±1mm / s. This solves the problems of slow zeroing at normal rates (such as 1.0mm / s) and easy jitter at 2.0mm / s, and balances "rapid zeroing" and "smooth braking". It reduces the safety or comfort hazards caused by improper rate and improves the scenario adaptability and stability of decay rate K.

[0092] The residual displacement zeroing process only updates the internal state and does not send braking force commands to the actuator, thus avoiding unnecessary braking actions when processing residual displacement and ensuring driving stability. At the same time, the zeroing data and the constrained braking force data are merged to generate a more complete and cleaner final braking force dataset, providing a reliable foundation for the S6's parameter self-learning and reducing invalid data interference during subsequent optimization.

[0093] In another embodiment of the present invention, step S6 further includes:

[0094] S61: The dynamic scaling factor curve of S3 (Including piecewise function coefficients), Xmax of S4, and K of S5 are classified as core calibration parameters; they are encapsulated into parameter templates according to parameter name-value range-default value-calibration method format and stored in the ECU's Flash memory (supports power-off saving); the core calibration parameter encapsulation template is output;

[0095] S62: Develop dual interfaces for CAN bus and OTA (Remote Online Upgrade). The CAN interface supports offline diagnostic instrument calibration (compatible with ISO15765 protocol), and the OTA interface supports remote cloud updates (using encrypted transmission protocol to prevent parameter tampering); write interface drivers to implement parameter reading, writing, and verification; output remote update interface documentation (including protocol specifications and calling procedures).

[0096] S63: First, read the fixed basic parameters of the current vehicle model (based on the calibration results of S32, S41, and S53, such as...). When x ≤ 1 mm, the value is fixed at 1; when x ≥ 15 mm, the value is fixed at 0.5; Xmax is calibrated at 1.2 mm; K = -10 mm / s. It is clear that feature extraction should focus on "operational behavior adapted to fixed parameters" and does not involve the reconstruction of basic parameters. From the historical datasets of S1 to S5, the braking operation records of the current vehicle model are selected (excluding interference from data of other vehicle models), focusing on "differences in driver operation under fixed parameters of vehicle model" - such as the different pedaling speeds of different drivers within the same vehicle model, the frequency of use of the commonly used return displacement range (the middle section between 1 and 15 mm), and the trigger ratio of rapid release (speed ≥ 2 mm / s).

[0097] For the filtered data of the same vehicle model, three types of key features were extracted:

[0098] Pedaling speed characteristics: Statistical analysis of the average speed of the outward displacement from 0 to 10 mm during a single braking action (distinguishing between preference for rapid and slow braking).

[0099] Common return range characteristics: Statistics on the usage frequency of the three sub-ranges "1 to 5mm, 5 to 10mm, and 10 to 15mm" during return operation (reflecting the driver's sensitivity to the middle range of foot feel);

[0100] Rapid release frequency characteristics: Statistical analysis of "the proportion of operations with a release speed ≥ 2mm / s to the total number of releases" (related to the rapid decay logic adaptation requirements from S52 to S53);

[0101] The extracted feature data is stored in the format of "vehicle model identifier - driver ID - pedal speed feature - common return range feature - rapid release frequency feature - vehicle model fixed parameter version" to ensure that each driver feature is bound to the fixed parameters of the current vehicle model. For example, the features of different drivers under the same vehicle model only reflect the "difference in operating preferences in the 1 to 15 mm range", rather than the difference in basic parameters. This generates a driver operation feature library to provide data support for S64's "self-learning fine-tuning within fixed parameters".

[0102] S64: Employs gradient descent as a self-learning algorithm, using the matching degree between driver operation characteristics and traditional foot feel as the objective function (matching degree ≥90% is considered satisfactory), and dynamically fine-tunes the algorithm. The segmentation factor of the curve, such as for drivers who frequently and rapidly release, increases the displacement segment by 5 to 10 mm. Value (improving release sensitivity); iterative training until the matching degree reaches the target, outputting an adaptively optimized calibration parameter set (including fine-tuned parameters). Xmax, K).

[0103] The working principle and effects of the above technical solution are as follows:

[0104] Will The segmented coefficients, Xmax, and K are packaged into a unified template according to "name-range-default value-calibration method" and stored in the ECU's Flash memory. This template can be saved even after power failure, which avoids the parameter tuning chaos caused by scattered parameter storage and reduces the risk of parameter loss after vehicle power failure. Engineers can quickly locate the required parameters when tuning parameters in the future without having to search for them one by one, saving a lot of redundant operations and improving the management efficiency and stability of core calibration parameters.

[0105] Offline, the CAN interface allows for convenient on-site fine-tuning with diagnostic tools, and it is compatible with the ISO15765 protocol without the need for additional adapters. Online, the OTA interface supports remote cloud upgrades, allowing car owners to update without having to visit a 4S store, reducing upgrade time and labor costs. Moreover, OTA uses encrypted transmission to prevent parameters from being maliciously tampered with, reducing braking safety hazards caused by abnormal parameters.

[0106] First, clarify the fixed parameters of the vehicle model (e.g.) When x≤1mm, fix 1; when x≥15mm, fix 0.5; Xmax=1.2mm). Then, filter the records of the same model from the historical data and extract only the "operational differences within the fixed parameter framework" - such as the frequency of use and the proportion of rapid release of different drivers in the 1 to 15mm range. The generated feature library is bound to the fixed parameters of the model to avoid mixing in data from other models and causing blind judgment. This provides an accurate basis for subsequent fine-tuning and reduces the deviation of the self-learning algorithm in judging driver habits.

[0107] Fine-tuning within a fixed parameter range using gradient descent. For example, for drivers who frequently and rapidly release power, the displacement range should be increased by 5 to 10 mm. The improved sensitivity ensures that the pedal feel matches the traditional experience by ≥90%, eliminating the need for drivers to adapt to fixed parameters and reducing the adaptation time for different drivers after switching vehicles. It is more comfortable to use and breaks the limitations of the traditional fixed calibration "one-size-fits-all" approach, improving the personalized adaptability of the braking pedal feel.

[0108] In another embodiment of the present invention, step S7 further includes:

[0109] S71: Design three typical test conditions to cover the actual use scenarios of new energy vehicles: load condition (no load / half load / full load, corresponding to curb weight 1800kg / 2000kg / 2200kg), road condition (dry asphalt road / wet asphalt road / icy road, corresponding to adhesion coefficient 0.8 / 0.4 / 0.2), and speed condition (low speed start-stop ≤30km / h / medium speed braking 30 to 60km / h / high speed braking ≥60km / h); each condition is set up with 10 repeated tests to output a full-condition test plan;

[0110] S72: Conduct real-vehicle testing according to the S71 scheme, collecting two types of key indicators: objective indicators (brake pedal feedback force fluctuation amplitude, braking force response delay) and subjective indicators (driver's rating of pedal feel as close to that of a traditional fuel vehicle, with a maximum score of 10 points); compare the test results with data from traditional vacuum boosters—for example, under low-speed conditions on a dry, unloaded road surface, the feedback force fluctuation amplitude must be ≤2N, the response delay must be ≤0.2s, and the subjective score must be ≥8 points; output a full-condition performance evaluation report;

[0111] S73: For the operating conditions that did not meet the standards in S72 (such as high-speed braking on icy and snowy roads with a full load, subjective score of 7 points), the parameters were adjusted accordingly, increasing Xmax to 1.5mm (enhancing hysteresis simulation), and fine-tuning was performed. The coefficient for the large displacement segment of the curve (to improve release stability); retest and verify until all operating conditions meet the standards; output the final calibration parameter set after iterative optimization.

[0112] S74: Integrate the final calibration parameter set of S73 and the algorithm logic of S1 to S6 (including data acquisition, curve fitting, scaling calculation, attenuation control, and self-learning) into a complete control strategy; write embedded code (adapted to the ECU's MCU chip, such as Infineon AURIX TC397) and perform functional safety verification (compliant with ISO26262 ASILD level); finally generate the final control strategy document for brake-by-wire assist hysteresis (including code, parameter table, and verification report) for ECU mass production programming.

[0113] The working principle and effects of the above technical solution are as follows:

[0114] The design covers three types of working conditions: load, road surface, and speed, which cover almost all actual use scenarios of new energy vehicles. The 10 repeated tests for each working condition can reduce the interference of accidental factors, avoid the one-sidedness of traditional single working condition tests, and ensure that the control strategy can perform stably in various real scenarios. This reduces the probability of abnormal foot feel in specific scenarios and improves the comprehensiveness of the test.

[0115] It considers both objective data (feedback force fluctuation, response delay) and subjective scores (driver's pedal feel evaluation), which is different from looking only at data and easily overlooking the actual driving experience. For example, it clearly requires that the feedback force fluctuation under no-load and low-speed conditions be ≤2N and the score be ≥8 points, making the evaluation criteria more in line with the user's real feelings, reducing the situation where the data meets the standards but the driving experience is uncomfortable, and enhancing the accuracy of performance evaluation.

[0116] For operating conditions that do not meet the standards (such as high-speed braking on icy and snowy roads with a full load), specific parameter adjustments are made, such as increasing Xmax and making fine adjustments. Instead of blindly changing parameters, the curves solve the problem of general parameters not being suitable for special working conditions, ensuring that the feel underfoot can meet the standards in different scenarios, reducing the local performance shortcomings caused by a one-size-fits-all approach to parameters, and improving the adaptability of parameters to working conditions.

[0117] After integrating the algorithm logic, embedded code was written and passed ISO26262ASILD safety verification to ensure that the strategy not only meets the functional standards but also meets the highest safety level, reducing the braking risk caused by software problems after mass production and vehicle installation; the generated complete documentation also facilitates subsequent mass production burning and maintenance, reduces the probability of errors in the production process, and enhances the reliability of the final control strategy.

[0118] One embodiment of the present invention provides a non-transitory computer-readable storage medium having a computer program stored thereon, which is executed by a processor to implement the brake assist function assist hysteresis software control method as described above.

[0119] Another embodiment of the present invention provides a computer program product, including a computer program / instructions that, when executed by a processor, implement the brake assist function assist hysteresis software control method as described above.

[0120] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A software control method for assist hysteresis in brake-by-wire function, characterized in that, The method includes: S1: Construct a pedal state monitoring network, collect the push rod displacement x(t) and motion direction, and generate the original pedal state parameter dataset; S2: Construct outbound and inbound baseline assist curves based on the original parameters to generate a dual-path baseline assist mapping dataset; S3: Introduce a dynamic scaling factor i(x) to process the return displacement and generate a preliminary return braking force dataset; S4: Optimize braking force by limiting Xmax by maximum offset, and generate a constrained target braking force dataset; S5: Perform gradient decay processing on the residual displacement to generate the final braking force dataset; S6: Encapsulate core calibration parameters and combine them with driver operation data for self-learning to generate an adaptive optimization calibration parameter set; S7: Full-condition verification and optimization, solidifying the final control strategy for linear control brake assist hysteresis; The S2 includes: S21: Based on the sensor's preset range and offline monitoring data, determine the effective range of push rod displacement and output a report; S22: Select a traditional vacuum booster from a fuel vehicle of the same class, collect its hysteresis characteristic data through bench testing, record the output braking force corresponding to different displacements, obtain the displacement-braking force forward and backward program sequences, verify the characteristic that the forward braking force increases with the displacement rate; output the hysteresis characteristic dataset of the traditional vacuum booster. S23: Using the effective range of S21 as the horizontal axis and the hysteresis characteristic data of S22 as the vertical axis, a polynomial fitting algorithm is used to fit the outbound and return reference curves respectively; the accuracy of the curves is verified by residual analysis, and the outbound / return reference assist curve equations are output. S24: Perform closure verification on the two baseline curves fitted in S23 in the low-to-medium stroke range of 0~10mm; organize the data in the format of displacement-outgoing braking force-returning braking force to generate a dual-path baseline assist mapping dataset. The S3 includes: S31: Conduct bench tests and user surveys to address the return stroke issue of existing brake-by-wire systems; clarify return stroke control requirements; and output a return stroke brake feel specification. S32: Design and calibration of dynamic scaling factor i(x) based on return stroke reduction; S33: For the return stroke condition, extract the push rod displacement x(t) from the original dataset of S14, substitute it into the i(x) curve model of S32, and apply the formula... Calculate the equivalent back displacement; perform range verification on the calculation results and remove negative displacement outliers; output the equivalent back displacement sequence. S34: The equivalent return displacement of S33 Substitute the outbound reference assist curve of S23 The initial target braking force at different times is calculated; the data is organized in the format of timestamp-displacement-equivalent displacement-initial braking force to generate the return initial braking force dataset.

2. The software control method for assist hysteresis function of brake-by-wire according to claim 1, characterized in that, S1 includes: S11: Based on the motion characteristics of the brake pedal actuator of the brake-by-wire system of new energy vehicles, and combined with the core influencing parameters of the hysteresis characteristics of traditional vacuum boosters, the functional correlation method is used to divide the monitoring dimensions; output the brake pedal monitoring dimension division scheme. S12: Based on the division scheme of S11, a magnetoelectric displacement sensor is selected and deployed on the side of the push rod's movement trajectory. The sensor signal is directly collected by the ECU, and the brake pedal is determined to be pressed or released by the increase and decrease of the stroke. S13: The ECU collects the displacement x(t) of the brake push rod; performs preliminary filtering on the collected raw signal, and outputs the raw discrete displacement signal sequence; S14: Perform dual-channel sensor signal verification and data alignment processing to generate the original pedal state parameter dataset.

3. The software control method for assist hysteresis function of brake-by-wire according to claim 1, characterized in that, The S4 includes: S41: Based on braking safety standards, bench tests were conducted to verify the impact of different offsets on braking performance. When the offset exceeds 5.0mm, the brake release delay time exceeds 0.3s, posing a risk of dragging the brakes. When the offset is less than 0.8mm, the hysteresis simulation is insufficient, resulting in significant differences in pedal feel. The final calibration range for Xmax is 0.8~5.0mm, with a specific value of 1.2mm determined for the target vehicle model. The calibration result for the maximum permissible offset of hysteresis is output. S42: Design and vehicle calibration of dynamic scaling factor i(x) based on return stroke reduction; S43: Compare ΔX in S42 with Xmax in S41: If ΔX > Xmax, then force an adjustment to the equivalent return displacement. =x(t) + Xmax, substitute into Recalculate the target braking force; if ΔX≤Xmax, retain the initial target braking force of S34; output the adjusted equivalent displacement sequence and the recalculated target braking force sequence; S44: Merge the initial braking force retained in S43 with the recalculated target braking force according to timestamps; store it in the format of timestamp-displacement-equivalent displacement-constrained braking force to generate a constrained target braking force dataset.

4. The software control method for assist hysteresis function of brake-by-wire according to claim 1, characterized in that, The S5 includes: S51: Perform threshold judgment on the original pedal state parameter dataset of S14. Set the push rod displacement x(t)≤0.5mm and the duration≥100ms to determine that the pedal is fully released; output the pedal release state judgment signal. S52: Perform displacement offset detection and rapid decay control in rapid release scenarios; S53: Based on the vehicle's inertia and suspension response characteristics, calibrate and verify the displacement offset decay rate K under rapid release scenarios; S54: For the residual displacement that needs to be processed in S52, gradually decrease the equivalent return displacement at a rate of −K. until Record the timestamp, residual displacement, and equivalent displacement data during the zeroing process, and merge them with the constrained braking force data of S44 to generate the final braking force dataset after the residual displacement is zeroed.

5. The software control method for assist hysteresis function of brake-by-wire according to claim 1, characterized in that, The S6 includes: S61: Classify the dynamic scaling factor i(x) of S3, Xmax of S4, and K of S5 as core calibration parameters; encapsulate them into parameter templates according to parameter name-value range-default value-calibration method format and store them in the ECU's Flash memory; output the core calibration parameter encapsulation template; S62: Develop dual interfaces for CAN bus and OTA; write interface drivers to read, write, and verify parameters; output remote update interface documentation; S63: Extracting driver operation features and constructing a feature library based on fixed vehicle parameters; S64: Gradient descent is used as the self-learning algorithm. The matching degree between the driver's operating characteristics and traditional foot feel is used as the objective function. The piecewise coefficients of the i(x) curve are dynamically fine-tuned. Iterative training is performed until the matching degree reaches the target, and an adaptive optimization calibration parameter set is output.

6. The software control method for assist hysteresis function of brake-by-wire according to claim 1, characterized in that, The S7 includes: S71: Design three typical test conditions to cover the actual use scenarios of new energy vehicles; set 10 repeated tests for each condition and output a full-condition test plan. S72: Conduct real-vehicle testing according to the S71 scheme, collect two types of key indicators; compare the test results with traditional vacuum booster data; and output a full-condition performance evaluation report. S73: For the operating conditions that do not meet the standards in S72, adjust the parameters accordingly; output the final calibration parameter set after iterative optimization; S74: Integrate the final calibration parameter group of S73 and the algorithm logic of S1~S6 into a complete control strategy; write embedded code and perform functional safety verification; finally generate the final control strategy document for brake-by-wire hysteresis for ECU mass production programming.

7. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, The program is executed by the processor to implement the brake assist function assist hysteresis software control method as described in any one of claims 1-6.

8. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instruction is executed by the processor, it implements the brake assist function assist hysteresis software control method as described in any one of claims 1-6.

Images