CN-121979033-A - Data-driven EMB system brake clearance online estimation method and system

CN121979033ACN 121979033 ACN121979033 ACN 121979033ACN-121979033-A

Abstract

The invention provides a data-driven EMB system braking gap online estimation method and system, which are characterized in that a nonlinear polynomial mapping model based on Legendre orthogonal polynomials is constructed by excavating the relation between a braking motor current and a corner signal in a braking loading stage of an electronic mechanical braking system, an enhanced recursive least square algorithm is designed to realize the coefficient online identification of the current and the corner polynomials, and a braking gap value is analyzed from an orthogonal fitting function through a zero current extrapolation mechanism. The invention does not need force sensor and off-line calibration, can realize low-cost on-line estimation of the braking gap of the electromechanical braking system, provides high-precision braking gap for the wire-controlled electromechanical braking system, and provides theoretical basis for self-adaptive adjustment and control algorithm design of the braking gap of the electromechanical braking system.

Inventors

WEI WENPENG
ZUO TENG
SU JIAN
Yao Huaicheng
YIN GUODONG
Zou Yuqiang
HE DONGLIN

Assignees

东南大学
芜湖伯特利汽车安全系统股份有限公司

Dates

Publication Date: 20260505
Application Date: 20260108

Claims (8)

1. The data-driven EMB system brake clearance online estimation method is characterized by comprising the following steps of: s1, preprocessing abnormal data values, namely collecting motor current in a braking loading stage Motor corner in loading stage And carrying out outlier pretreatment on the obtained product; S2, constructing a Legendre orthogonal basis regression model, namely constructing a current-corner regression model taking a Legendre orthogonal polynomial as a basis function based on the preprocessed current signal ; S3, enhanced RLS parameter identification, namely adopting forgetting factor For the parameter vectors in the regression model by the enhanced recursive least square method of (2) Performing online identification and updating; s4, zero current extrapolation to estimate braking gap, namely, parameter vector to be estimated online Extrapolation calculation of substituted zero current to obtain brake clearance estimation 。
2. The method for online estimation of brake clearance of data-driven EMB system according to claim 1, wherein the step S1 is characterized in that the motor current in the loading phase Motor corner in loading stage Outlier preprocessing based on physical boundary and single-point recursion statistics is performed.
3. The method for online estimating brake clearance of data-driven EMB system according to claim 2, wherein the constructing a regression model in step S2 specifically includes: according to the braking motion rule of the electromechanical braking system, in the whole loading process from the initial position, the current-corner global mapping relation and the current braking gap have definite mathematical relation, and the mapping relation is as follows: ; Wherein, the Indicating the motor current during the loading phase, The motor rotation angle in the loading stage is represented, Representation of And (3) with A, B, C, D is a constant; for motor current in loading stage Performing linear normalization to interval : ; Wherein, the Indicating the maximum current value preset for the actual measurement situation, Representing the auxiliary variable of the linear normalization, Representing motor current during the loading phase; Legendre orthorhombic base interval Has orthogonality and weight function The orthonormal polynomial form is as follows: ; Wherein, the And Represents the Legendre group, An auxiliary variable representing a linear normalization; For the current-corner mapping, the first four Legendre bases are employed, respectively: ; Wherein, the Represents the first Legendre group, Represents a second Legendre group, The third Legendre group is represented, Represents a fourth Legendre group, An auxiliary variable representing a linear normalization; meanwhile, the Legendre group satisfies the following orthogonality: , ; Wherein, the And Represents the Legendre group, An auxiliary variable representing a linear normalization; Constructing regression vectors Defining a parameter vector The current-corner regression model based on Legendre orthometric is obtained as follows: ; Wherein, the Is a Legendre orthogonal-based current-corner regression model, 、、、 Is the Legendre group, the amino acid is the Legendre group, , , , The coefficient corresponds to the Legendre base; Legendre orthometric based current-corner regression model Conversion into a matrix form: ; Wherein, the Is a Legendre orthogonal-based current-corner regression model, As a regression vector, the number of the regression vectors, Is a parameter vector.
4. The method for online brake clearance estimation of a data-driven EMB system according to claim 3, wherein said step S3 uses a method having a forgetting factor And combining Legendre orthogonal basis, establishing an enhanced RLS algorithm for online identification and updating, and carrying out a current-corner regression model based on the Legendre orthogonal basis The predictor and cost function are defined as: ; ; Wherein, the The predicted motor rotation angle value is indicated, As a regression vector, the number of the regression vectors, A parameter vector representing an on-line estimate, Representing a weighted sum of squares of the prediction errors, As a forgetting factor, A current-corner regression model of Legendre orthorhombic basis; establishing an enhanced RLS with forgetting factors: ; Wherein, the Representing a matrix of the gain of the parameter estimation, The covariance matrix is represented by a matrix of covariance, As a forgetting factor, As a regression vector, the number of the regression vectors, A parameter vector representing an on-line estimate, Is a Legendre orthonormal based current-corner regression model.
5. The method for online estimation of brake clearance of data-driven EMB system according to claim 4, wherein the current parameter estimation value is obtained in step S4 After that, normalizing the auxiliary variable Extrapolated to the theoretical zero current point, i.e When the brake clearance estimation value is obtained : ; Wherein, the In order to estimate the brake clearance, , , , Is the estimated Legendre base coefficient.
6. The method for online estimation of brake clearance of a data-driven EMB system according to claim 5, further comprising step S5 after step S4: Updating and storing historical data based on the brake clearance estimation value after each brake cycle is finished Triggering one-time recursion iteration initial value update, and storing a historical result for improving the cold start speed of the electronic mechanical brake system.
7. A data-driven EMB system brake clearance online estimation system, comprising: a data preprocessing module configured to perform step S1 of the method according to claim 2, for outlier preprocessing of the acquired current and rotation angle signals; A model building module configured to perform step S2 of the method according to claim 3 for building a Legendre orthobasal regression model; a parameter identification module configured to perform step S3 of the method according to claim 4 for enabling online adaptive identification of parameters; a clearance calculation module configured to perform step S4 of the method according to claim 5 for calculating a brake clearance estimate; An update storage module configured to perform step S5 of the method according to claim 6 for updating the initial state and managing the history data.
8. The data-driven EMB system brake clearance online estimation system of claim 7, integrated into an electromechanical brake system of a vehicle for providing real-time estimates for adaptive adjustment of brake clearance.

Description

Data-driven EMB system brake clearance online estimation method and system Technical Field The invention belongs to the technical field of automatic driving and intelligent chassis, and particularly relates to a data-driven on-line estimation method and system for braking clearance of an EMB system. Background The brake clearance estimation method of the electromechanical brake (EMB) system in the prior art mainly comprises the following steps: 1. The detection method based on the threshold value is characterized in that the method based on the current and the torque is obviously influenced by temperature, motor efficiency and noise, and has larger signal fluctuation and abnormal jitter under the same working condition, so that detection errors and reliability are reduced; 2. based on an offline fitting model method, the method has high requirements on storage and calculation resources of a real-time control system and high cost depending on fitting data quality and quantity, and meanwhile, on-line updating cannot be realized, and the change of a brake clearance caused by the abrasion of a brake pad is difficult to deal with; 3. Based on the RLS recursive algorithm, the method is extremely sensitive to noise and cannot meet the requirement of stable online estimation, meanwhile, the inside of the regression vector is highly coupled, and various numerical values are excessively different, so that the reliability of an estimated value is greatly reduced. Therefore, the brake clearance estimation method in the prior art has the problems of low precision, dependence on a force sensor and off-line calibration, lack of on-line self-adaptive capacity, poor anti-interference performance, unstable numerical value and the like. 1. Through retrieval, the Chinese patent with publication number CN120277811A discloses a method for identifying braking clearance and estimating self-adaptive clamping force of an EMB system, which is different from the method in the application in that 1, the patent CN120277811A divides braking clearance types based on EMB contact time, establishes a clamping force dynamics model based on a motor moment balance equation by utilizing motor electromagnetic torque, angle and angular speed, realizes clamping force estimation through different types of extended state observers, and focuses on the linkage of clearance classification and clamping force estimation. The technology core is more focused on the accuracy and independence of gap estimation by excavating the association of motor current and corner signals in a brake loading stage, constructing a Legendre orthogonal polynomial nonlinear mapping model, combining an enhanced recursive least square algorithm to realize parameter online identification, and directly analyzing a gap value through a zero current extrapolation mechanism. 2. Although the patent CN120277811a realizes gap identification without a pressure sensor, the accuracy index of gap estimation is not clarified, and the technical logic of the patent CN120277811a needs to match different gap types depending on the targeted design of the observer, and performs identification based on threshold judgment, so that the reliability and the adaptability are poor. The method does not need force sensor and off-line calibration, specifically realizes that the steady state estimation error of the brake clearance is less than 0.6 rad (0.39%), the relative root mean square error under the initial condition of large deviation is less than 3.478 multiplied by 10 < -3 >, and has more advantages in balance of low cost and high precision. 2. Through searching, the Chinese patent with publication number CN121062678A discloses an EMB sensorless clamping force estimation method based on multi-signal fusion and displacement compensation, which is different from the method of the application in that: 1. The patent CN121062678A is based on an EMB system mechanical model, realizes contact point detection through current-position information fusion, and improves the accuracy of clamping force estimation through position error compensation and return backlash compensation, and the technical core is a clamping force optimization strategy of 'multi-signal fusion and multi-link compensation'. The method and the device can directly realize gap estimation by combining a proprietary identification algorithm only through the polynomial mapping relation of the current and the corner signal without depending on a contact point detection and subsequent compensation mechanism, so that a technical path is simpler, and accumulated errors possibly introduced by multiple compensation links are avoided. 2. The core objective of the patent CN121062678A is to promote the effect of the clamp force estimation, the brake clearance related processing is only used as an auxiliary link of the clamp force estimation, the patent uses the brake clearance on-line estimation as the core objective, the technical advantage is constructed