Vehicle door handle emergency actuator in-place test method and device

Abstract

The invention discloses an in-situ test method for a vehicle door handle emergency actuator, and the method comprises the steps: carrying out the reciprocating motion of an output end of the actuator in an end domain range, collecting a transition relation, and obtaining reference beat data; performing reverse and forward combined actions on the output end to determine single-side stress starting point data; on the basis, executing propulsion, parking, rollback and restart operations in sections to obtain abnormal window position data; executing a detection action in the abnormal window range to obtain an abnormal type label; executing symmetrical micro-motion actions on two sides of the window to obtain an abnormal attribution result; and comprehensively outputting a starting action prediction result. According to the technical scheme, abnormal position locking, abnormal type classification and abnormal attribution judgment can be realized through a rhythmic physical process coupled with an actuator structure and a kinematic chain without an additional sensor, so that the reliability and prediction capability of the test method are improved.

Description

Technical Field

[0001] This invention relates to the field of vehicle door handle emergency actuator testing and functional testing technology, and in particular to an in-situ testing method and apparatus for vehicle door handle emergency actuators. Background Technology

[0002] Emergency door handle actuators are specialized actuation units designed to replace conventional door lock actuation chains and directly apply controllable stroke and force to the door handle / cable / latch mechanism in unconventional scenarios such as vehicle power failure, vehicle network anomalies, collision unlocking failure, external freezing, or water immersion. A typical form is a linear transmission architecture consisting of a motor—reduction gear—lead screw (or worm gear)—slider—push rod, encapsulated in a sealed housing and coupled to the door lock mechanism via a short-stroke interface. Control is triggered by the vehicle's BCM / safety domain controller, requiring the ability to open the door after one or more impacts under adverse conditions such as low temperature, low voltage, vibration, and icing, while meeting long-term durability and protection requirements. The value of this type of actuator lies not in frequent daily operation, but in its deterministic output to complete the door opening task at critical moments; therefore, its reliability and predictability are core indicators.

[0003] In practical applications, actuators are not simply a matter of installation and then being worry-free. Throughout their entire lifecycle, automotive actuators face various challenges, including extreme temperatures (low-temperature icing leading to friction-viscosity-embrittlement coupling), power supply fluctuations (battery degradation and transient voltage drops altering back EMF and torque reserves), lubrication degradation (increased viscosity and oil film rupture expanding the dry friction range), tolerance accumulation and assembly deviations (distorting loads in local stroke segments), localized corrosion and hindrance caused by dust / moisture intrusion, and material aging and wear (evolution of tooth backlash and lead screw / nut pair clearance). These factors combine to create a highly nonlinear and location-dependent change in the "load-stroke-environment" relationship: the same current peak may correspond to end-point impact or mid-stroke jamming; fixed-time thresholds are prone to false alarms at low temperatures / low pressures; and traditional criteria relying solely on current thresholds or position switches are insufficient to distinguish fault modes and their locations. Meanwhile, actuators are housed in miniaturized, sealed installation cavities. Due to considerations of size, cost, wiring harness complexity, and sealing reliability, mass-production platforms generally do not equip them with additional stroke / force sensors. Even if sensors can be added to some models, it will bring limitations on vehicle placement, durability, and consistency risks. How to achieve online status identification, position positioning, and predictable activation success rate of the actuator's drivetrain while maintaining sealing and cost constraints, and addressing the strong nonlinearity and drift caused by environmental factors and aging, thus enabling reliable decisions before and during emergency actions, has become a problem that needs to be solved in this field. Summary of the Invention

[0004] The main objective of this invention is to solve the problem that existing vehicle door handle emergency actuators are difficult to identify and predict due to the strong nonlinearity of load-stroke-environment and drift caused by space constraints and the absence of external sensors.

[0005] To achieve the above objectives, embodiments of this application provide an in-situ testing method for a vehicle door handle emergency actuator, comprising: Within the end domain, the actuator output terminal performs a reciprocating motion within a preset end domain displacement range, and the end domain transition relationship is collected to obtain reference cycle data; Perform a combination of reverse and forward actions on the actuator output to eliminate backlash and determine the starting point data of unilateral force. Based on the single-sided force starting point data, the stroke is divided into multiple segments. In each segment, the actions of advancing, stopping, reversing, and restarting are performed. The rebound characteristics and restart characteristics are compared with the reference beat data to obtain the abnormal window position data. Perform detection actions of different directions or amplitudes within the range corresponding to the abnormal window position data, and obtain the abnormal type label based on the change of restart characteristics; Symmetrical micro-motion actions are performed on both sides of the abnormal window position data, and the abnormality attribution result is obtained based on the comparison results of force-displacement characteristics. Based on the abnormal window location data, the abnormal type label, and the abnormal attribution result, the corresponding opening action prediction result is output.

[0006] To achieve the above objectives, this application also proposes an in-situ testing device for a vehicle door handle emergency actuator, comprising: The end domain acquisition module is used to perform reciprocating motion on the actuator output end within a preset end domain displacement range within the end domain range, acquire the end domain transition relationship, and obtain reference cycle data. The clearance alignment module is used to perform a combination of reverse and forward actions on the actuator output to eliminate clearance and determine the force start point data on one side. The segment discrimination module is used to divide the stroke into multiple segments based on the single-sided force starting point data, perform advance, stop, retraction and restart operations in each segment, and compare the rebound characteristics and restart characteristics according to the reference beat data to obtain abnormal window position data. The anomaly classification module is used to perform detection actions of different directions or amplitudes within the range corresponding to the anomaly window position data, and obtain anomaly type labels based on the changes in restart features; The attribution determination module is used to perform symmetrical micro-motion actions on both sides of the abnormal window position data, and obtain the abnormal attribution result based on the comparison result of force displacement characteristics; The prediction output module is used to output the corresponding start action prediction result based on the abnormal window position data, the abnormal type label, and the abnormal attribution result.

[0007] The technical solution provided in this application uses the small-amplitude movements controllable by the actuator itself and the displacement characteristics and load response available on the electric drive side as the observation carrier. First, it performs reciprocating motion in the end domain to extract the reference rhythm of the "idle stroke - force" transition. Then, it presses the force-bearing surface to the same side through reverse retreat and forward contact, solidifying the force starting point and phase uniformity under the current loading. On the unified baseline, the effective stroke is divided into several segments. Within each segment, the pause rebound and restart performance are compared according to the rhythm of "advance - stop - retreat - restart", and corresponded with the reference rhythm, thereby locking the anomaly into a narrow coordinate window. After obtaining the window, change the detection direction or displacement amplitude within the same coordinate system, observe the differences in restart performance with direction switching and amplitude increase, and give a classification of reversible and irreversible. Then, make symmetrical micro-movements on both sides of the window, compare the consistency of the force displacement sequence or the distortion on one side, and attribute the deviation to the actuation link side or the door lock mechanism side. Finally, perform joint mapping of window coordinates, type classification and attribution results to give a prediction of the success rate of this opening action and strategy selection.

[0008] The key to solving the problem in this invention lies in transforming unmeasurable position and mechanical information into relatively comparable information under controllable timing. End-domain reference allows for the reproduction of baselines under different operating conditions; gap-free contact ensures consistent starting force for all subsequent segments; segmented timing transforms the problem of "global threshold drift" into "relative differences within the same small segment," thus achieving position binding; direction and amplitude detection within the window utilizes the sensitivity of reversible hindrance to unloading and small-amplitude restart, contrasting it with the insensitivity to hard collisions, completing type binding; symmetrical micro-motions on both sides of the window, based on relative order and consistency, distinguish between "overall translation" and "unilateral distortion," forming attribution binding. The superposition of position, type, and attribution makes the judgment no longer dependent on absolute thresholds, but rather derived from the physical response chain in place at that time. Therefore, it can maintain stability under adverse operating conditions such as low temperature, low pressure, and aging, and provide a reliable success rate prediction and strategy selection before action. Attached Figure Description

[0009] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0010] Figure 1 This is a schematic diagram of an embodiment of the in-situ testing method for the emergency actuator of the vehicle door handle in this invention; Figure 2 This is a schematic diagram of an embodiment of the in-situ testing device for the emergency actuator of the vehicle door handle in this invention. Figure 3 This is a structural schematic diagram of one embodiment of an emergency actuator for a vehicle door handle.

[0011] Explanation of icon numbers: 1. Motor; 2. Small helical gear; 3. Lead screw helical gear; 4. Slider; 5. Push rod.

[0012] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0013] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0014] It should be noted that if the embodiments of the present invention involve directional indicators (such as up, down, left, right, front, back, etc.), the directional indicators are only used to explain the relative positional relationship and movement of the components in a specific posture. If the specific posture changes, the directional indicators will also change accordingly.

[0015] Furthermore, the use of terms such as "first" and "second" in this invention is for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the term "and / or" throughout the text includes three solutions; taking A and / or B as an example, it includes technical solution A, technical solution B, and a technical solution that simultaneously satisfies A and B. Furthermore, the technical solutions of various embodiments can be combined with each other, provided that they are feasible for those skilled in the art. When the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by this invention.

[0016] A vehicle door handle emergency actuator is a dedicated electric drive unit that ensures reliable door opening under unconventional conditions. When the vehicle experiences a power outage, a network malfunction, or is exposed to freezing or water, the conventional door lock actuation circuit may fail to respond properly. In such cases, the emergency actuator uses an independent linear drive mechanism to directly act on the door handle, cable, or latch to mechanically unlock the door. During its iteration process, the actuator has gradually evolved towards miniaturization, as shown in the attached figure. Figure 3 The structure shown consists of a motor 1 driving a reduction unit (with a small helical gear 2 meshing with a lead screw helical gear 3), a slider 4, and a final output push rod 5. The speed of motor 1 is transmitted through the meshing of the small helical gear 2 and the lead screw helical gear 3, driving the lead screw to rotate. The lead screw then drives the slider 4 to move axially, pushing the push rod 5 to complete a short-stroke linear motion and rigidly coupling it with the door lock mechanism. This process appears simple, but it requires ensuring instantaneous mechanical output and repeatability under different environments, thus placing extremely high demands on the health of the transmission chain.

[0017] In practical applications, actuators cannot operate stably for extended periods immediately after installation. Due to their enclosed cavity and limited installation space, mass-produced actuators typically lack additional stroke or force sensors to reduce costs and maintain sealing. This means that determining their health status and operational reliability often relies solely on the current threshold or time threshold at motor 1. However, during long-term use, low temperatures can cause lubricant glassization, increased friction between the lead screw and nut, voltage fluctuations may lead to insufficient torque at motor 1, assembly tolerances and wear can cause clearance evolution, and dust or moisture intrusion may cause jamming or obstruction. The combined effect of these factors results in a highly nonlinear and position-dependent relationship between load and stroke, making traditional methods based on a single current threshold prone to false alarms and missed alarms, and difficult to predict whether the door opening task can be successfully completed before the action. Therefore, this solution proposes a novel in-situ testing method that utilizes the actuator's controllable action cycle and the relative changes in displacement-force response to identify abnormal locations, distinguish abnormal types, and attribute causes, thus providing a reliable basis for predicting the success rate of the opening action. Figure 1 A flowchart illustrating an in-situ testing method for a vehicle door handle emergency actuator according to an embodiment of this application. In this embodiment, the method includes: Please see Figure 1 Within the end domain, the actuator output terminal performs reciprocating motion within a preset end domain displacement range, and the end domain transition relationship is collected to obtain reference cycle data; In one embodiment of the present invention, the step of performing a reciprocating motion on the actuator output end within a preset end-domain displacement range within the end-domain range, collecting the end-domain transition relationship, and obtaining reference cycle data includes: Without entering the effective working stroke, the actuator output terminal performs a reciprocating motion within the end domain transition displacement range to trigger the transition process from a low load state to a force-supported state. During the transition process, the correspondence between the displacement characteristics of the action process and the load response is recorded to obtain time-series data characterizing the response characteristics of the end domain. Based on the time series data, the end-domain load features are extracted and correlated with the displacement features to obtain the reference beat data.

[0018] The following is a detailed description of the steps involved in the above embodiments: Specifically, the end-domain transition displacement range refers to the displacement interval relative to the initial zero position of the output end and without entering the effective working stroke, used to trigger the contact transition from a low-load state to a force-supported state; the effective working stroke refers to the displacement interval at the output end that causes the door lock mechanism to perform functional actions (such as the interval for raising the latch or pulling the cable); displacement characteristics refer to the equivalent quantities that can be derived on the controller side and used to represent the displacement of the output end (e.g., equivalent displacement sequences derived from motor commutation counting, time integration at constant angular velocity, or back electromotive force); load response refers to observable electric drive-side quantities that reflect the force state of the transmission chain (e.g., the filtered value of the motor phase current, the equivalent torque change combined with the bus voltage). Without entering the effective working stroke, the controller sends small reciprocating commands to the motor, with the angular velocity using a trapezoidal rhythm of rising-constant speed-falling. The reciprocating displacement amplitude is limited to the end-domain transition displacement range, so that the lead screw after the small helical gear-lead screw helical gear transmission only drives the slider and push rod to reciprocate within a short stroke, gradually compressing the idle stroke until force support is formed without triggering the door lock mechanism to act. The cycle time parameters are determined by two types of constraints: first, the upper limit of displacement does not exceed the upper limit of the transition displacement range of the end domain, to ensure that it does not enter the effective working stroke; second, acceleration and peak angular velocity are limited to the low energy level of the motor, so that temperature rise, noise, vibration, and acoustic roughness are kept within the allowable safe threshold range. Taking the linear chain of motor—small helical gear—lead screw helical gear—slider—follower rod as an example (e.g.) Figure 3 As shown in the diagram, the slider reciprocates within the guide rail. When the push rod approaches the interface surface, the load transitions from a low value to an increasing phase, completing the transition from a low-load state to a supported state. This process, under sealed conditions and without external sensor constraints, can repeatedly cross the contact critical point, forming a comparable interval required for subsequent discrimination, while avoiding accidental triggering of the door lock mechanism. In an equivalent implementation, the reciprocating angular velocity can also employ a sinusoidal beat or a symmetrical triangular beat with symmetrical amplitude and period. As long as the displacement amplitude is limited to the end-domain transition displacement range and can stably cross the contact critical point, the same effect can be achieved.

[0019] During the transition process, a one-to-one correspondence between displacement characteristics and load response needs to be established. The controller uses a unified time base for synchronous sampling. The time base source can be the motor control interrupt cycle or a sampling clock synchronized with PWM (Pulse Width Modulation), enabling the displacement characteristics and load response at the same sampling moment to be paired and stored. Displacement characteristics are obtained in two ways: first, by counting motor commutation phases or accumulating encoder subdivisions to obtain the equivalent angular displacement, which is then converted into the equivalent linear displacement at the output end according to the gear ratio and lead screw; second, under constant angular velocity control, the sampling time is treated as the equivalent displacement independent variable, and equivalent calibration is performed according to angular velocity and transmission parameters when necessary. The load response is preferably the motor phase current after low-pass and notch filtering, combined with the bus voltage to suppress the influence of power supply fluctuations; if necessary, the back electromotive force is collected to estimate the speed change, serving as circumstantial evidence of friction state transition. The data path is as follows: After the cycle command is issued, displacement characteristic samples and load response samples are collected according to a unified time base to form a time-ordered paired sequence. Direction and cycle segment identifiers (ascending, constant speed, descent) are labeled in each reciprocating half-cycle to obtain time-series data characterizing the end-domain response. Figure 3 Taking the structure shown as an example, when the push rod is not in contact, the corresponding curve of current and equivalent displacement is at a low level; after contact, the curve shows an inflection point and rises with displacement. This step solidifies the information of "whether it is under force and at what displacement it is under force" into a searchable time-series trajectory, providing accurate input for subsequent feature extraction. In the equivalent implementation, when commutation counting is unavailable, time can be used as the equivalent independent variable of displacement feature under constant angular velocity; when the resolution of the current channel is limited, the combined change of drive duty cycle and back electromotive force can be used as a substitute for the load response, and the corresponding acquisition can still be completed.

[0020] The reference clock data consists of the correspondence between end-domain load characteristics and displacement characteristics. The time-series data is first preprocessed to remove low-frequency drift and high-frequency ripple. This preprocessing includes DC descaling, notch filtering of the PWM switching frequency, and first- or second-order low-pass filtering to ensure that weak amplitude inflection points are not drowned out by noise. Then, within each reciprocating half-cycle, four types of end-domain load characteristics are identified and extracted: the displacement coordinates and corresponding current inflection point at the contact inflection point, the slope range of the load rise after contact, the rebound amplitude during the pause phase, and the additional overcoming force during the restart phase. The direction identifier and timestamp of each characteristic are retained. These characteristics are then correlated with displacement characteristic samples at the same sampling time, generating a data set consisting of "characteristic name—displacement coordinate—direction—time identifier—characteristic quantity," which is the reference clock data. Taking the linear chain of motor-small helical gear-lead screw helical gear-slider-push rod as an example, after the gap between the small helical gear and the lead screw helical gear is compacted, the slider-push rod contacts the lock interface surface, and the current curve shows a clear inflection point. If there is elasticity or viscosity during parking, the rebound amplitude can be quantified within a short time window, and the additional current during restart reflects the viscous additional load. By binding the above quantities with the displacement coordinates, a reproducible end-domain mechanical signature can be obtained under the current loading posture and environmental conditions. The subsequent advancement, parking, retraction, and restart of each stroke segment are all compared with this as a benchmark, thereby avoiding misjudgment caused by the drift of the absolute threshold with the environment. In the equivalent implementation, in order to reduce the computational load, only the contact inflection point displacement and the restart additional current can be retained as simplified reference cycle data; in order to enhance robustness, the corresponding characteristics of the upward and downward half cycles can also be retained at the same time, and the overlapping interval of the two can be used as the final reference cycle data. Both can support the subsequent positioning and parting process.

[0021] Please continue reading. Figure 1 The actuator output is subjected to a combination of reverse and forward actions to eliminate backlash and determine the starting point data of the force on one side. In one embodiment of the present invention, performing a combined reverse and forward action on the actuator output to eliminate backlash and determine the unilateral force initiation data includes: Under the end domain state determined according to the reference beat data, reverse yielding is performed on the actuator output end within the backlash elimination alignment displacement range so that the idle stroke in the kinematic chain is reduced to no more than a preset threshold. Within a preset continuous time window after the reverse yielding is completed, the actuator output end is subjected to forward contact within the same displacement range to compress the force-bearing surface of the kinematic chain to the same side contact state. Based on the results of the forward contact process, the correspondence between the output displacement characteristics and the load response is recorded to obtain the single-sided force starting point data, which serves as the benchmark for subsequent stroke segment discrimination.

[0022] The following is a detailed description of the steps involved in the above embodiments: Specifically, the backlash elimination alignment displacement range refers to the bidirectional small displacement interval limited to compress the kinematic chain idle time under the end-domain state and before entering the effective working stroke; reverse retraction refers to a small displacement action in the opposite direction to the subsequent working direction, used to push the meshing pair and transmission backlash to the same side; the preset threshold refers to the numerical limit used to determine whether the remaining idle time meets the alignment requirements (which can be calibrated by the contact inflection point difference, reciprocating hysteresis width, or current-displacement slope change amplitude in the end-domain reference cycle data). In implementation, the controller first determines the displacement coordinates of the contact inflection point and the reciprocating hysteresis width based on the reference cycle data, and sets the upper limit of the reverse amplitude of the backlash elimination alignment displacement range accordingly, so that the reverse displacement does not trigger the effective working stroke. Then, a reverse retraction command of trapezoidal angular velocity (low acceleration, low peak angular velocity) is issued, and displacement characteristics and load response are collected simultaneously; when the hysteresis width of the load response-displacement curve drops to no more than the preset threshold, or the current-displacement slope reaches the alignment criterion, it is determined that the idle time compression meets the standard and the retraction stops. Taking a linear chain consisting of a motor, a small helical gear, a lead screw helical gear, a slider, and a push rod as an example, the reverse retraction causes the tooth flank clearance and the lead screw nut pair clearance to converge to the same side, resulting in a slight increase in load response and narrowing hysteresis. This step unifies the meshing contact onto the same force surface, eliminating deviations introduced by phase and backlash differences in subsequent judgments. In an equivalent implementation, the reverse retraction can also employ short-duration reverse torque pulses or small-amplitude symmetrical reciprocating cycles dominated by net reverse displacement, as long as it does not enter the effective working stroke and the hysteresis width decreases to the threshold.

[0023] Forward contact refers to a micro-displacement action of the same amplitude performed along the subsequent working direction within a preset continuous time window after the reverse retraction is completed. This is used to compress the force-bearing surfaces of the kinematic chain to a contact state on the same side. The preset continuous time window refers to the maximum allowable interval (e.g., on the order of tens of milliseconds, calibrated based on the springback time constant in the end-domain reference cycle data) to avoid the influence of material springback and lubrication relaxation. In practice, within this time window after the reverse retraction stops, forward contact is immediately performed within the same displacement range in a trapezoidal cycle of rising-constant speed-falling, maintaining synchronous acquisition of displacement characteristics and load response. When a contact criterion is met (e.g., the current baseline rises further from the reverse end without hysteresis, or a small push no longer results in hysteresis expansion), the contact is terminated and briefly paused to stabilize the force state. The above process ensures that the force on the tooth side, the force on the lead screw and nut pair, and the force on the interface surface are continuously established on the same side, avoiding alignment disruption due to springback reset caused by excessively long intervals. Taking the short-stroke interface of a linear actuator as an example, after positive contact, the load response and displacement characteristics show a continuous monotonic relationship with closed hysteresis, indicating that same-side contact has been formed. In an equivalent implementation, positive contact can also be achieved by superimposing a constant small torque offset with a very small amplitude displacement, or by using short-term micro-vibration superimposed with a positive offset to accelerate contact, as long as the same-side contact criterion is achieved within the time window.

[0024] The single-sided force-starting point data refers to the displacement-force correspondence information set used as the benchmark for subsequent travel segment discrimination under the same-side contact state. It includes the starting coordinates of the displacement characteristics, the corresponding load response baseline, the time marker and direction marker of the contact end. During implementation, the controller synchronously records the displacement characteristics and load response within several sampling cycles after the positive contact ends, and performs detrending and low-pass filtering to eliminate power supply variations and high-frequency ripple. Then, it determines the displacement coordinates of the single-sided force-starting point (e.g., the equivalent displacement at the contact end, or the displacement corresponding to the minimum additional pushing force after contact) and its corresponding load response baseline, storing them in the reference structure as a unified starting point for subsequent segment division and cycle comparison. To enhance robustness, the single-sided force-starting point data can be calculated separately in two consecutive contact cycles. It is only solidified when the coordinate deviation between the two calculations is less than the alignment tolerance; otherwise, the retreat-contact sequence is automatically redone until convergence. This step unifies the discrimination of subsequent advance, stop, retreat, and restart to the same force-starting point, eliminating the influence of idle drift and phase difference. In an equivalent implementation, if the system uses constant angular velocity small-amplitude control, the displacement coordinates can be calculated by converting the timestamp after the contact ends with the angular velocity; or, if a more refined commutation count is available, the cumulative number of commutation pulses can be used directly as the displacement feature. As long as it is synchronized with the load response and fixed as a unified starting point, it can meet the subsequent discrimination requirements.

[0025] Please continue reading. Figure 1 Based on the single-sided force starting point data, the stroke is divided into multiple segments. In each segment, the actions of advancing, parking, reversing, and restarting are performed. The rebound characteristics and restart characteristics are compared with the reference beat data to obtain the abnormal window position data. In one embodiment of the present invention, based on the single-sided force starting point data, the stroke is divided into multiple segments, and within each segment, propulsion, parking, retraction, and restart operations are performed. The rebound characteristics and restart characteristics are compared according to the reference beat data to obtain abnormal window position data, including: Under the state corresponding to the single-sided force starting point data, the starting point and end domain boundary of the effective working stroke are determined according to the reference beat data, and the effective working stroke is divided into multiple stroke segments according to the preset rules to obtain segment index data; Based on the segment index data, advance, stop, retraction and restart operations within the segment's retraction displacement range are executed sequentially according to a preset rhythm within each stroke segment. The synchronous timing correspondence between the displacement characteristics and load response within the corresponding stroke segment is recorded to obtain the segment rhythm timing data. Based on the intra-segment beat timing data, the rebound and restart features after the pause are extracted in each stroke segment, and aligned with the baseline features corresponding to the reference beat data to obtain intra-segment alignment feature data. Based on the intra-segment alignment feature data, a joint comparison is performed among multiple travel segments, and travel segments with continuous deviations are filtered out according to a preset joint deviation threshold to obtain candidate segment data; Around the segment boundary corresponding to the candidate segment data, encrypted advance, stop, retraction and restart operations are performed within the local refined displacement range, and local refined comparison is performed to determine the continuous coordinate range of the deviation interval and obtain the abnormal window position data. The abnormal window location data is checked for consistency with the segment index data, and the abnormal window location data is output when the check passes.

[0026] The following is a detailed description of the steps involved in the above embodiments: Specifically, under the corresponding state of the single-sided force starting point data, the effective working stroke refers to the displacement range at the output end that can trigger the functional action of the door lock mechanism, and the end domain boundary refers to the boundary coordinates of the adjacent ends of the effective working stroke, where force can still be generated but no longer advancing functional action. The controller first reads the contact inflection point, rebound amplitude, and additional overcoming amount of restart in the reference cycle data, and determines the starting point of the effective working stroke by combining the displacement coordinates of the single-sided force starting point data; then, it determines the end domain boundary by the saturation value of the load response, the inflection point where no new functional response appears after further advancement, or the hard contact inflection point at the end. The effective working stroke is divided into multiple stroke segments according to preset rules to obtain segment index data; the rules can be equal displacement division, equal expected load increment division, or adaptive subdivision based on the curvature change of the reference cycle data. The number of segments aims to minimize the load change within a single segment without crossing local features, and is usually taken as 3–8 segments. This processing ensures that subsequent comparisons are performed within homogeneous small intervals, reducing the impact of global drift on the judgment. In an equivalent implementation, equal time division can be used, but it must be used under the condition of small-scale advancement at a constant angular velocity to ensure the consistency of displacement mapping.

[0027] Based on the segment index data, within each stroke segment, the system sequentially performs advance, pause, retraction within the segment's retraction displacement range, and restart operations according to a preset cycle. The retraction displacement range within a segment refers to the retraction amplitude limited within the stroke segment, used to trigger local hysteresis and rebound without entering adjacent segments. The recommended cycle is to advance to a predetermined proportion of the segment's upper limit using a trapezoidal angular velocity, pause briefly to release elastic or viscous stress, then perform a small retraction within the segment's retraction displacement range, and finally restart in the same direction as the advance to overcome residual resistance. The controller synchronously records displacement characteristics and load response under a unified time base, assigning direction and segment identification to each cycle segment, generating segment-level cycle timing data. Taking a linear chain of motor—small helical gear—lead screw helical gear—slider—push rod as an example, the current increases with displacement during the advance phase, rebound occurs during the pause phase if elasticity or viscosity exists, hysteresis closure occurs during the retraction phase, and additional resistance is overcome during the restart phase. This cycle is repeated once or multiple times within each segment to ensure statistical robustness. In an equivalent implementation, the beat can be replaced by a symmetrical triangular velocity or a small-amplitude sinusoidal velocity. As long as the range of retraction displacement within the segment does not exceed the limit and the beat segment is clearly marked, equivalent beat timing data within the segment can be obtained.

[0028] Based on the intra-segment timing data, rebound and restart features after parking are extracted within each stroke segment, and aligned with the baseline features corresponding to the reference timing data to obtain intra-segment aligned feature data. Rebound features can be defined as the displacement retraction or corresponding load release from the end of parking to the point of rest, while restart features can be defined as the additional current peak, brief speed drop, or equivalent overcoming force at the initial restart stage. To eliminate the effects of sampling jitter and gradual power supply variations, the timing data is first detrended, notch-filtered, and low-pass filtered. Then, the time axis is aligned using the timing segment identifier, and the displacement coordinates within each segment are normalized to the coordinate system of the segment's starting point. The alignment strategy employs a dual condition of displacement matching and amplitude correction with features of the same name as the reference timing data: the displacement coordinate difference does not exceed the alignment tolerance, and the amplitude difference is considered successful if it is within the allowable band. Intra-segment aligned feature data (including displacement coordinates, amplitude, direction, and time identifier) ​​is output. This processing transforms the global threshold problem into a relative difference problem, allowing subsequent cross-segment comparisons to focus only on systematic deviations from the baseline. In an equivalent implementation, alignment can also be achieved using the maximum cross-correlation time-shift method, followed by displacement coordinate backfilling, as long as the final output satisfies the consistent reference of the same-named feature.

[0029] Based on the alignment feature data within each segment, a joint comparison is performed across multiple travel segments. Travel segments exhibiting persistent deviations are selected according to a preset joint deviation threshold, yielding candidate segment data. The joint deviation threshold refers to a logical combination threshold for multiple features, such as the displacement coordinate offset of the rebound feature, the additional overcoming amount increment of the restart feature, and the requirements for their unidirectional and continuous nature. During the comparison, the difference vector between each segment and the reference beat data is first calculated. Then, a continuity check is performed on the segment sequence: if at least two types of features continuously exceed the threshold in several adjacent segments and the difference signs are consistent, it is judged as a persistent deviation, and the indices and boundary coordinates of these segments are output as candidate segment data. The threshold setting is based on the statistical dispersion of the end domain and historical convergence data, using a mean-weighted multiple bandwidth to balance robustness and sensitivity. This step compresses anomalies from the entire travel to a small number of adjacent segments, providing a narrow starting point for subsequent localization. In an equivalent implementation, weighted voting or sequence cumulative sum discrimination can be used; as long as the joint deviation is conditional on the continuous unidirectional nature of multiple features, the reliability of the selection results is equivalent.

[0030] Around the segment boundary corresponding to the candidate segment data, encrypted advance, pause, retraction, and restart operations are performed within a locally refined displacement range, and local refinement comparisons are conducted to determine the continuous coordinate range of the deviation interval, thus obtaining the anomaly window position data. The local refined displacement range refers to a small-range refined interval symmetrically set on both sides of the candidate segment boundary, with the step size and cycle period significantly reduced compared to the previous sequence to improve coordinate resolution. During implementation, the refinement center coordinates are first determined based on the gradient extrema of the alignment feature data within the segment, and then encrypted cycle loops are performed on both sides of the center with a fixed step size or an adaptive step size. For each step size point, rebound and restart features are extracted, aligned with the reference cycle data, and then locally compared. When the difference sign on the front and back sides changes and forms a continuous interval on the step size grid, the coordinates at both ends of the sign change are used as the boundary of the anomaly window, and the anomaly window position data is output. This process locks the anomaly into a continuous displacement coordinate range, eliminating the influence of single-point noise. In an equivalent implementation, the step size can be reduced by using bisection scaling or golden ratio scaling. As long as a continuous coordinate interval is given in the end and evidence of alignment on both sides is provided, it has equivalent performance.

[0031] The abnormal window location data is compared with the segment index data for consistency. If the comparison passes, the abnormal window location data is output. The consistency check includes three aspects: First, the window coordinates must fall within a certain travel segment or cross an allowed combination of two adjacent travel segments, and must not cross the boundary of the end domain. Second, the intra-segment alignment features on both sides of the window must have a stable difference in the opposite direction to the reference beat data on the same indicator, avoiding false windows caused by accidental fluctuations. Third, the reproducibility of the window boundary is verified by repeating the refined beat once, and the reproduction error does not exceed the preset coordinate tolerance. If the check fails, the process returns to the previous processing step, adjusts the joint deviation threshold or refines the step size, and retryes. If the check passes, the abnormal window location data is fixed for subsequent classification and attribution. This convergence-check-fixing closed loop improves the reliability of window positioning and avoids false locking under boundary conditions such as low temperature and low voltage. In an equivalent implementation, the consistency check can also include a bypass check based on the combined change of duty cycle and back electromotive force. As long as the criterion is still based on the stable difference between the same features inside and outside the window, the output abnormal window position data has equivalent reliability.

[0032] In one embodiment of the present invention, the step of performing encrypted advance, pause, retraction, and restart operations within a locally refined displacement range around the segment boundary corresponding to the candidate segment data, and performing local refined comparisons to determine the continuous coordinate range of the deviation interval, and obtaining abnormal window position data, includes: Around the segment boundary corresponding to the candidate segment data, the refinement center coordinates are determined according to the changing trend of the alignment feature data within the segment, and the local refinement displacement range is set at its adjacent positions before and after. Within the localized refined displacement range, encrypted advance, stop, retraction and restart operations are executed sequentially according to a preset rhythm, and the corresponding displacement characteristics and load response are recorded to obtain refined time sequence data; The refined time series data is locally compared. When the rebound and restart features of adjacent positions show a change in sign and form a continuous deviation interval, the continuous coordinate range of the deviation interval is determined, and the abnormal window position data is output.

[0033] The following is a detailed description of the steps involved in the above embodiments: It should be noted that the refinement center coordinates refer to the displacement reference point near the segment boundary corresponding to the candidate segment data, used for densification detection; the adjacent positions refer to the adjacent coordinate points on both sides of the refinement center coordinates, separated by one to several refinement step distances; the local refinement displacement range refers to the symmetrical small-range interval extending forward and backward from the refinement center coordinates. In implementation, the controller reads the intra-segment alignment feature data (including displacement coordinates and amplitudes of rebound and restart features) around the segment boundary, performs trend analysis on its differential sequence with displacement, and preferentially selects the position where the differential amplitude gradient reaches an extreme value or exhibits a monotonic abrupt change as the refinement center coordinate. Based on this, local refinement displacement ranges and refinement step distances are set on both sides, constraining it not to cross adjacent travel segments and not to reach the end domain boundary. This setting concentrates subsequent comparisons in the most sensitive local transition areas, improving coordinate resolution and reducing the impact of global drift. In an equivalent implementation, the refinement center coordinates can also be taken as the midpoint of the first stable change of the difference sign at the segment boundary, or as the midpoint of the boundary of the two coarse measurement windows. As long as it is consistent with the change trend of the alignment feature data within the segment and facilitates symmetrical refinement, the same purpose can be achieved.

[0034] Within the localized refined displacement range, the encrypted advance, pause, retraction, and restart operations are executed sequentially according to a preset rhythm, and displacement characteristics and load response are recorded to obtain refined time sequence data. The preset rhythm adopts a trapezoidal angular velocity sequence with low acceleration and low peak angular velocity to control energy and heat load, and ensure that noise, vibration, and acoustic roughness do not exceed safety thresholds; the advance endpoint and retraction amplitude are both limited to the localized refined displacement range to prevent exceeding the limits. The action sequence within each refined step is as follows: advance forward (or backward) from the refined center to the current step coordinate, pause briefly to release elasticity and viscosity, then retract slightly within the retraction displacement range of the segment, and then execute the restart operation from that point; the controller synchronously collects displacement characteristics and load response using a unified time base, and marks the step number, direction, and rhythm segment identifier to form refined time sequence data arranged according to a coordinate grid. Taking the linear chain of motor-small helical gear-lead screw helical gear-slider-push rod as an example, the current increases with displacement during the propulsion stage, rebounds during the pause, closes the hysteresis loop during retraction, and exhibits additional overcoming force upon restarting. In an equivalent implementation, the cycle time can also adopt a symmetrical triangular velocity or a small-amplitude sinusoidal velocity. As long as the step size, amplitude, and time window remain consistent and the local refined displacement range constraint is satisfied, the resulting refined timing data is equivalent.

[0035] Local comparisons are performed on refined time-series data. A sign change is defined as the situation where the difference between the corresponding feature and the reference beat data has opposite signs on both sides of adjacent positions, and the amplitude is not lower than the decision band. A continuous deviation interval refers to a coordinate interval on the refined step grid where the difference signs are consistent and the amplitude continuously exceeds the band. The processing flow is as follows: at each refinement step, the differences between the rebound feature and the restart feature relative to the reference beat data are calculated, and difference sequences are established according to adjacent positions. When both types of features simultaneously satisfy amplitude exceeding the band and sign consistency within several adjacent steps, it is marked as a local consistent deviation. Subsequently, the first sign change pair between the previous and subsequent consistent deviations is searched, and the coordinates of the two ends of this pair are used as the boundary of the deviation interval to obtain the continuous coordinate range and output the abnormal window position data. To avoid accidental flips caused by noise, the boundary needs to be verified by repeating the encrypted beat once, and the verification error must not exceed the coordinate tolerance. This method uses sign changes to lock local deviations into continuous intervals, avoiding single-point anomaly interference and improving positioning reliability. In an equivalent implementation, the boundary can also be obtained by using a binary scaling step method to converge to the sign change point, or by using time-shift alignment with maximum cross-correlation and backfilling displacement coordinates. As long as the final output is a continuous coordinate range that matches the local consistent deviation, it is considered an equivalent implementation.

[0036] Please continue reading. Figure 1 Within the range corresponding to the abnormal window position data, detection actions of different directions or amplitudes are performed, and an abnormal type label is obtained based on the changes in the restart characteristics. In one embodiment of the present invention, the step of performing detection actions of different directions or amplitudes within the range corresponding to the abnormal window position data, and obtaining an abnormal type label based on the change of restart features, includes: Within the range defined by the abnormal window position data, and under the state corresponding to the single-sided force starting point data, a detection action is performed on the actuator output end along the positive direction within the window direction detection displacement range. The restart characteristics after detection are recorded and compared with the baseline restart characteristics corresponding to the reference beat data to obtain the positive detection characteristic data. Within the range defined by the abnormal window position data, and under the state corresponding to the single-sided force starting point data, a detection action is performed on the actuator output end in the reverse direction within the window direction detection displacement range. The restart characteristics after detection are recorded and compared with the baseline restart characteristics corresponding to the reference beat data to obtain the reverse detection characteristic data. Based on the comparison results between the forward detection feature data and the reverse detection feature data, the difference in direction switching of the restart feature is extracted to obtain direction-dependent data; Within the range defined by the abnormal window position data, the detection action is performed step by step within the window amplitude detection displacement range according to the preset amplitude level sequence, and the restart characteristics corresponding to each amplitude level are recorded to obtain amplitude increment data; Based on the joint determination result of the direction-dependent data and the amplitude-increasing data, anomaly type labels of reversible or irreversible anomalies are output.

[0037] The following is a detailed description of the steps involved in the above embodiments: Specifically, the displacement range for window direction detection refers to the continuous coordinate interval given by the abnormal window position data as the upper limit. Within this interval, a sub-interval that does not exceed the boundary is selected for a single detection. The upper limit of its amplitude is set according to the proportional coefficient of the window width to ensure that the detection does not cross the window boundary. The controller starts from the displacement coordinate corresponding to the force starting point data on one side, and advances along the subsequent working direction to the detection endpoint using a trapezoidal beat with low acceleration and low peak angular velocity. After a short pause, it performs a restart operation, collects displacement characteristics and load response on a unified time base, and extracts restart characteristics (such as additional current peak at the initial stage of restart, brief speed drop amplitude, and equivalent overcoming force). Then, this restart characteristic is aligned and compared with the baseline restart characteristic in the reference beat data, and outputs positive detection characteristic data containing displacement coordinates, amplitude, time markers, and direction markers. The reason for using limited displacement and low-energy beats is to avoid introducing external factors beyond the window, while controlling temperature rise, noise, vibration, and acoustic roughness within allowable thresholds, ensuring that the data source is singular and reproducible. In other embodiments, forward detection can also employ equivalent small-amplitude sinusoidal beats or symmetrical triangular beats. As long as the displacement is limited to the window direction detection displacement range and the alignment comparison with the baseline restart feature is completed, equivalent forward detection feature data can be obtained.

[0038] While maintaining the same starting point of force application, the controller performs restricted propulsion in the opposite direction to the subsequent working direction until the endpoint of the reverse detection. After a short pause, it restarts. The amplitude and cycle parameters of this reverse propulsion and restart are symmetrically set with those of the forward detection to ensure comparability of the two detections in terms of energy and displacement constraints. Data acquisition and processing are consistent with the forward detection, obtaining the restart characteristics during reverse detection and aligning them with the baseline restart characteristics. The output includes reverse detection characteristic data containing displacement coordinates, amplitude, time markers, and direction markers. The significance of reverse detection lies in verifying the sensitivity to force direction switching within the same coordinate window: if there is viscosity or local surface contact lubrication instability, direction switching often changes the additional resistance and delay; if it is a hard foreign object or a hard collision with the mechanism, the consistency before and after the direction switching is higher. To prevent reverse detection from disrupting the same-side contact, an alignment check can be set before the reverse propulsion amplitude (e.g., hysteresis width not greater than the alignment threshold). If the requirement is not met, the detection will return to the gap elimination alignment before resuming. In an equivalent implementation, a small-amplitude reverse torque pulse superimposed with a minimal displacement propulsion can be used to replace the complete reverse propulsion. As long as the restart feature is synchronized with the displacement coordinate and aligned with the baseline, equivalent reverse detection feature data can be obtained.

[0039] The controller pairs two sets of detection features within the same window coordinate grid, calculates the difference components after alignment with the reference beat data baseline, and extracts three types of difference quantities: additional overcoming quantity difference, restart delay difference, and initial slope difference. To suppress single-point anomalies, the difference quantities are smoothed within the window by a small step size, and a consistency check is added: if the three types of difference quantities are consistent in direction on several adjacent coordinates and the amplitude exceeds the decision band, then the direction dependency is determined to be valid, and direction dependency data containing the coordinate range, difference quantity vector, and consistency identifier is output. The engineering significance of the direction dependency data lies in revealing the asymmetry of force behavior with direction switching within the window: viscous and lubricated degradation are often sensitive to the unloading-reloading sequence, with significant differences between positive and negative sides; hard collisions or rigid interferences are not sensitive to direction, and the difference quantities are close to zero. The decision band is set based on the reference beat data and the end-domain statistical dispersion to maintain separability under low voltage and low temperature conditions. In an equivalent implementation, the difference can also be characterized using area difference (current-potential accumulation integral difference in the initial restart phase) or energy difference. As long as it is aligned with the displacement coordinates and the consistency test results are output, equivalent direction-dependent data can be formed.

[0040] The window amplitude detection displacement range refers to the maximum detection amplitude upper limit that does not exceed the window boundary; the preset amplitude level sequence consists of several discrete amplitudes, applied step by step in ascending order. The controller executes the same cycle (advance-pause-retreat-restart) at each level, and collects displacement characteristics and load response on a unified time base, extracts the restart characteristics of the corresponding level and aligns them with the baseline, forming a feature sequence sorted by level. Then, a "recovery level" is marked, i.e., the restart characteristic first returns to the baseline band or reaches the minimum level of the preset relative error band; if recovery is not achieved even at the maximum level, it is marked as unrecovered. The rationale for amplitude grading is that for viscous-elastic anomalies, a finite amplitude is sufficient to remove additional resistance and return to the baseline; for hard collisions or rigid interferences, simply increasing the amplitude is insufficient for recovery. To balance efficiency and resolution, geometrical grading or adaptive grading (sparse first, dense later) can be used, with each level's energy and displacement limited to not crossing boundaries and noise, vibration, acoustic roughness, and temperature rise meeting safety thresholds. In an equivalent implementation, the graded variables can also be driving torque or duty cycle levels, while monitoring the displacement to ensure it does not exceed the window amplitude detection displacement range, and the recorded restart feature sequence is regarded as equivalent amplitude increment data.

[0041] Specifically, the joint decision-making process comprises two main rules and a robustness check: Main rule one outputs a reversible anomaly if direction dependency is valid and a finite recovery level exists; main rule two outputs an irreversible anomaly if direction dependency is invalid or recovery is not achieved at the maximum level. The robustness check eliminates false triggers caused by random noise: it selects two or more adjacent coordinates within the window and repeats a low-amplitude and recovery-level probe. If the direction dependency label matches the recovery level, the anomaly type label is fixed; otherwise, it backtracks to the nearest level of the amplitude-increasing data or adjusts the decision band for direction dependency before making a decision. The physical basis of this joint strategy is that viscous-elastic anomalies are sensitive to both direction and amplitude and have finite reversibility, while hard interference is insensitive to direction and amplitude increases are ineffective. The output anomaly type label serves as input for subsequent attribution and strategy decisions, supporting the prior prediction of the success rate of the activation action. In an equivalent implementation, the joint rules can introduce weighted voting (the weights of the direction difference and recovery level can be adaptively adjusted according to environmental conditions), but the core constraint of the direction-amplitude dual-dimensional condition must be maintained, and the output anomaly type label is equivalent to the above rules.

[0042] In one embodiment of the present invention, the step of outputting anomaly type labels of reversible or irreversible anomalies based on the joint determination result of the direction dependence data and the amplitude increment data includes: Under the same conditions of unilateral force starting point data and reference beat data, the directional difference index data are determined based on the deviation of the forward detection feature data and the reverse detection feature data from the baseline restart feature. According to the preset amplitude level sequence, the detection action is performed step by step within the window amplitude detection displacement range to determine the minimum amplitude level that restores the restart feature to the range corresponding to the reference beat data, and the amplitude level data is obtained. Based on the joint judgment condition of the directional difference index data and the span amplitude level data, when the directional difference index data is not less than the preset threshold and there is a limited amplitude level that can restore the restart feature to the range corresponding to the reference beat data, a reversible anomaly type label is output; when the directional difference index data is less than the preset threshold or has not been restored to the range corresponding to the reference beat data at the maximum amplitude level, an irreversible anomaly type label is output.

[0043] The following is a detailed description of the steps involved in the above embodiments: Specifically, the controller performs baseline alignment on the coordinate grid corresponding to the abnormal window position data for both the forward and reverse restart features. The baseline is taken from the corresponding restart feature in the reference cycle data and includes a tolerance band. Three types of restart feature quantities are extracted at each coordinate point: additional overcoming force in the initial restart phase, restart start-up delay, and current-displacement slope in the initial restart phase. The deviations of the forward and reverse directions relative to the baseline are calculated separately and compared pairwise under the same coordinate system to obtain the amplitude difference vector. To suppress single-point noise, the amplitude difference vector is smoothed by a small window on adjacent coordinates and normalized according to the dispersion of the feature quantities to obtain dimensionless directional difference index data. Its data items include coordinates, three types of difference quantities, and a consistency indicator. The engineering implications of directional difference lie in revealing the sensitivity to force direction switching under the same window coordinate system: viscosity or lubrication degradation is highly sensitive to the unloading-reloading sequence, with the three types of difference quantities being significant and consistent in direction; hard collisions or rigid interference are insensitive to changes in direction, with the difference quantities approaching zero. The threshold is set based on the reference clock data and the statistical dispersion of the end domain, taking a multiple of bandwidth that balances sensitivity and robustness. In an equivalent implementation, the directional difference can also be characterized by the current-potential accumulation area difference or the equivalent energy difference in the initial restart phase. As long as it is aligned with the baseline and the difference is output and bound to the coordinates, equivalent directional difference index data can be formed.

[0044] The displacement range detected by the window amplitude is given by the coordinate interval defined by the abnormal window position data. The preset amplitude level sequence consists of several discrete amplitudes, applied sequentially from smallest to largest. At each level, advance, stop, retraction, and restart operations are performed along the window direction. Three types of features are extracted: additional resistance during the initial restart phase, restart delay, and initial slope. Alignment with the baseline band is then determined. Recovery is considered when all three types of features enter the baseline band, and the minimum amplitude level that produces recovery is recorded as the amplitude level data. If the maximum amplitude level still does not enter the baseline band, it is marked as unrecovered. Amplitude grading can use a geometric series to reduce the number of levels, while ensuring that the window boundary is not crossed under energy and displacement constraints, and keeping temperature rise, noise, vibration, and acoustic roughness within allowable thresholds. In an equivalent implementation, the grading variable can also be represented by driving torque or duty cycle levels instead of displacement amplitude. As long as the binding with the displacement coordinate and the unified determination with the baseline band are maintained at each level, the minimum recovery level and the unrecovered determination are considered equivalent amplitude level data.

[0045] Within the same window coordinate range, when the directional difference index data is not lower than the directional difference threshold and there is a minimum recovery level not exceeding the upper limit, a reversible anomaly type label is output; when the directional difference index data is lower than the directional difference threshold, or has not recovered to the baseline band at the maximum amplitude level, an irreversible anomaly type label is output. To improve robustness, the detection of the minimum recovery level and low amplitude level is repeated once by selecting two or more adjacent coordinates within the window. The directional difference index data and recovery level must be consistent for the label to be fixed; if inconsistency occurs, the threshold and grade interval are finely adjusted and then checked within the original window. If inconsistency still exists, an irreversible anomaly type label is output according to the conservative principle. The physical basis for joint judgment is that viscous-elastic anomalies simultaneously satisfy sensitivity to directional switching and recoverability to finite amplitude, while hard collisions or rigid interferences lack directional sensitivity and do not produce recovery to amplitude increases; therefore, the directional-amplitude dual-dimensional constraint can reliably distinguish between the two types of anomalies. In an equivalent implementation, the joint rule can also adopt a weighted voting form, assigning environmentally adaptive weights to directional differences and minimum recovery levels. However, it is necessary to maintain "coexistence of directional sensitivity and limited amplitude recoverability" as a necessary condition for outputting reversible labels, thereby maintaining equivalence with the above joint determination.

[0046] Please continue reading. Figure 1 Symmetrical micro-motion actions are performed on both sides of the abnormal window position data, and the abnormality attribution result is obtained based on the comparison results of force displacement characteristics. In one embodiment of the present invention, the step of performing symmetrical micro-motion on both sides of the abnormal window position data, and obtaining the abnormal attribution result based on the comparison result of force-displacement characteristics, includes: Within the adjacent intervals on the front and rear sides corresponding to the abnormal window position data, symmetrical detection displacement ranges are set respectively. Micro-motion is performed on the actuator output end, and the correspondence between displacement characteristics and load response is recorded to obtain the symmetrical feature data on the front and rear sides. The force-displacement sequences of the front-side symmetrical feature data and the rear-side symmetrical feature data are compared to obtain a symmetrical comparison result. Based on the symmetrical comparison results, if the overall curves of the features on both sides deviate but remain relatively consistent, the deviation is attributed to the actuator drive link side; if the features on both sides deviate only locally on one side, the deviation is attributed to the door lock mechanism side, and the abnormal attribution result is output.

[0047] The following is a detailed description of the steps involved in the above embodiments: Specifically, the front adjacent interval refers to the narrow displacement interval outside the front boundary of the continuous coordinate range shown by the abnormal window position data and connected to that boundary; the rear adjacent interval refers to the narrow displacement interval outside the rear boundary of the continuous coordinate range and connected to that boundary. The symmetrical detection displacement range refers to a small displacement range, with the same amplitude and symmetrical about its respective boundary, set within the front and rear adjacent intervals respectively. Micro-motion refers to a short-range reciprocating cycle executed with low acceleration and low peak angular velocity under the corresponding state of the single-sided force starting point data, without triggering displacement beyond the adjacent interval. During implementation, the controller reads the front and rear boundary coordinates of the abnormal window position data, and sets the width of the adjacent interval and the amplitude of the symmetrical detection displacement range based on the gradient and noise level of the intra-segment alignment feature data, ensuring that the outermost sampling point does not exceed the effective working stroke and end-domain boundary. Within the adjacent front section, a micro-motion is performed from the inner edge of the boundary to the center of the symmetrical detection displacement range, following a "progress-pause-retreat-restart" rhythm. Displacement characteristics and load response are synchronously acquired using a unified time base. This process is repeated several times to obtain stable samples, which are then organized into front-side symmetrical feature data (including displacement coordinates, load response, time, and direction indicators). The same process is repeated in the adjacent rear section with the same amplitude and rhythm to obtain rear-side symmetrical feature data. To maintain alignment with the same side and consistent with the force baseline, an alignment check is performed before entering the micro-motion stage (using hysteresis width or current-displacement slope criteria to confirm no alignment distortion). This setup ensures strict comparability of the data on both sides in terms of energy, step size, and time window, and the output dataset can be directly used for subsequent comparisons between the two sides. In an equivalent implementation, a slow, unidirectional quasi-static advancement superimposed with a very small "jitter" can replace the complete round-trip rhythm. As long as the displacement is limited to the symmetrical detection displacement range and synchronous acquisition is completed, equivalent front-side and rear-side symmetrical feature data can be obtained.

[0048] Force-displacement sequences refer to a set of sequential samples with displacement characteristics as independent variables and load response as dependent variables, distinguished by clock cycle segment identifiers for advance, pause, retraction, and restart sub-segments. Before comparison, detrending and notch and low-pass filtering of the PWM switching frequency are performed on the front and rear symmetrical feature data respectively to eliminate the influence of power supply variation and high-frequency ripple. Then, the displacement coordinates of the two sequences are translated to local coordinate systems with their respective boundaries at zero, and the amplitude is normalized according to the baseline band of the reference clock cycle data to obtain two sets of force-displacement sequences that can be compared in the same domain. The comparison process starts with three types of quantitative indicators: first, the overall offset (the difference between the baseline of the equivalent torque of the two advance segments and the difference between the zero displacement points); second, the shape consistency (the relative difference between the slope, hysteresis loop area, and pause rebound amplitude of the advance segment and the initial restart segment); and third, the local distortion (the density of second-order difference or curvature anomalies within a fixed small window). When the shape consistency is high and the overall offset is basically equal, a symmetrical comparison result of "consistent overall curve translation" is obtained; when only one side exhibits significant local distortion within a small window while the other side remains consistent with the reference beat data, a symmetrical comparison result of "unilateral local deviation" is obtained. To suppress random anomalies, the above indicators must meet the continuity requirement at several adjacent sampling points and be confirmed after a repeated fine-tuning check. In an equivalent implementation, alignment can also be achieved by using the maximum cross-correlation time-shift method to fine-tune the time axis, then backfilling the equivalent displacement coordinates before calculating the above three types of indicators; as long as a clear conclusion of "consistent overall translation" or "unilateral local deviation" is given, it is considered an equivalent symmetrical comparison result.

[0049] The anomaly attribution results are output based on the symmetrical comparison results. When the features on both sides show an overall curve translation and consistent shape, the judgment deviation originates from the actuator drive link side. This is because changes in tooth backlash, lead screw-nut pair friction, or transmission preload will cause an equal-amplitude baseline rise or equivalent zero-point translation within the same adjacent area throughout the entire stroke. If there is no change on the door lock mechanism side, it will not cause synchronous translation on both sides. When the features on both sides show only local deviation on one side, the judgment deviation originates from the door lock mechanism side. This is because the local interference / stickiness of the latch, interface surface, or linkage has positional limitations, showing distortion only in the adjacent area close to that side, while the other side remains consistent with the reference cycle data. In practice, to improve the robustness of the conclusions, a micro-motion can be repeated once in the same adjacent area, requiring the criteria for consistent overall translation or unilateral deviation to remain consistent in repeated measurements. If necessary, the step distance can be fine-tuned for local verification without changing the symmetrical detection displacement range. The final anomaly attribution results include the attribution side (actuator drive link side / lock mechanism side), the corresponding adjacent interval coordinates, and a summary of support indicators, which can be directly used as input for subsequent strategy decisions and success rate prediction. In an equivalent implementation, the attribution rules can introduce weighted voting (the weights of overall offset and shape consistency are adaptively set according to environmental conditions), but the judgment logic of "overall translation consistency → drive link side" and "single-sided local deviation → lock mechanism side" must remain unchanged, thus being equivalent to the above attribution conclusions.

[0050] Please continue reading. Figure 1 Based on the abnormal window location data, the abnormal type label, and the abnormal attribution result, the corresponding opening action prediction result is output.

[0051] In one embodiment of the present invention, the step of outputting a corresponding opening action prediction result based on the abnormal window location data, the abnormal type label, and the abnormal attribution result includes: Based on the correspondence between the abnormal window location data, the abnormal type label and the abnormal attribution result, a joint judgment condition set is constructed, and each input result is associated under the same judgment framework; Under the joint decision condition set, inductive decision is performed on the input results of different combinations, and the prediction result of the opening action corresponding to the combination is output.

[0052] The following is a detailed description of the steps involved in the above embodiments: Specifically, the joint decision condition set refers to grouping three types of inputs—abnormal window location data, abnormal type labels, and abnormal attribution results—along with their respective confidence weights and consistency identifiers, into a set of executable decision entries, specifying the triggering conditions, priorities, and output fields for each entry. The abnormal window location data provides continuous coordinate ranges, window widths, and segment indices to reflect the spatial location and locality of the abnormality; the abnormal type labels distinguish between reversible and irreversible abnormalities and include the generation basis for directional dependence and recovery levels; the deviation in the abnormal attribution result identifier originates from the actuator drive link side or the door lock mechanism side. The confidence weights and consistency identifiers come from the aforementioned repeated measurement and verification steps (e.g., window boundary reproduction error, consistency of type labels in adjacent coordinates, and repeated consistency of attribution in adjacent intervals on both sides), used to control the priority of entries when multiple decision branches compete. During implementation, before entering this step, the controller normalizes the three types of inputs to a unified coordinate and time reference: window coordinates are normalized to the segment's local coordinate system, type labels and attribution results carry the decision band information at the time of their generation, and all inputs carry the most recent verification timestamp. Subsequently, entries are constructed within the decision framework. For example, an entry such as "narrow window located at the end of the stroke, reversible type, attributed to the door lock mechanism side, and recovery level not exceeding the upper limit" is mapped to the output category of "local enhancement and single pass"; similarly, an entry such as "window spanning multiple segments and irreversible type or recovery failure" is mapped to the output category of "conservative degradation and reporting". To avoid sensitivity to a single threshold, the entry conditions employ interval decision and dual conditions (spatial location interval / type / attribution). Simultaneously, upper limit constraints for noise, vibration, acoustic roughness, and temperature rise are recorded in the entries to ensure that the output action strategy is executable within the energy and comfort boundaries. In an equivalent implementation, the joint decision condition set can be implemented using a finite state machine or a multi-entry lookup table. As long as the three types of inputs are aligned and associated within the same decision framework, and priority is controlled by confidence and consistency, the same functional effect and feasibility as described above can be achieved.

[0053] The inductive decision-making process matches and adjudicates different combinations of input results under the joint decision-making condition set, outputting the corresponding activation action prediction result. The controller matches entries in descending order of priority: when the triggering condition of an entry is met and the confidence weight exceeds the activation threshold, a prediction result containing "success rate interval, suggested action category, and key parameters" is generated. The success rate interval is a range-based output that monotonically maps reversibility, spatial locality, and attribution side under the same coordinate and time reference. For example, the combination of "narrow window + reversible + door lock mechanism side" is mapped to a higher interval, and the combination of "span or wide window + irreversible" is mapped to a lower interval. The suggested action category is given according to the type and attribution difference, such as "single enhancement", "segmented multiple impacts", "conservative degradation and reporting", etc. The key parameters limit the upper limit of displacement amplitude, beat shape, energy upper limit, and time window to ensure execution within the constraints of energy and noise, vibration, and acoustic roughness. If the confidence level of the highest priority item is insufficient, the second highest priority item is used according to the preset fallback strategy; if there is a conflict of the same priority, the one with the higher consistency identifier takes precedence, and a light review is required within the original window before the output is fixed. Taking a short-stroke rigidly coupled emergency actuator as an example, when the window is located in the middle of the stroke and the type is reversible, and the attribution is to the door lock mechanism side, the system outputs a prediction of "segmented multiple impacts and amplitude limiting", and gives a medium-to-high success rate range; when the window is close to the end domain and the type is irreversible, and the attribution is to the actuator drive link side, the system outputs a prediction of "conservative downgrading and reporting", and gives a low success rate range. In an equivalent implementation, the inductive judgment can also adopt a hierarchical lookup table weighted decision: first, the basic range is determined by the position and type, and then the attribution and confidence are used for refined weighting, but the synchronous generation of the range output and parameter constraints must be maintained so that the prediction result is both interpretable and can be directly sent to the control execution path.

[0054] The above describes the in-situ testing method for the vehicle door handle emergency actuator in the embodiments of the present invention. The following describes the in-situ testing device for the vehicle door handle emergency actuator in the embodiments of the present invention. Please refer to [link / reference]. Figure 2An embodiment of the in-situ testing device for an emergency actuator of a vehicle door handle according to this invention includes an end-domain acquisition module 101, a clearance alignment module 102, a segment discrimination module 103, an anomaly classification module 104, an attribution determination module 105, and a prediction output module 106. The end-domain acquisition module 101 is used to perform reciprocating motion on the actuator output end within a preset end-domain displacement range within the end-domain, acquiring the end-domain transition relationship and obtaining reference beat data. The clearance alignment module 102 is used to perform a combination of reverse and forward motion on the actuator output end to eliminate clearance and determine the unilateral force starting point data. The segment discrimination module 103 is used to divide the stroke based on the unilateral force starting point data. The system is divided into multiple segments. Within each segment, it performs advance, stop, retraction, and restart operations. Based on the reference beat data, it compares the rebound characteristics and restart characteristics to obtain abnormal window position data. The abnormal classification module 104 performs detection actions of different directions or amplitudes within the range corresponding to the abnormal window position data and obtains an abnormal type label based on the changes in restart characteristics. The attribution determination module 105 performs symmetrical micro-motion actions on both sides of the abnormal window position data and obtains an abnormal attribution result based on the comparison results of force-displacement characteristics. The prediction output module 106 outputs the corresponding start-up action prediction result based on the abnormal window position data, the abnormal type label, and the abnormal attribution result.

[0055] The above description is merely a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural transformations made using the contents of the present invention's specification and drawings under the inventive concept of the present invention, or direct / indirect applications in other related technical fields, are included within the patent protection scope of the present invention.

Claims

1. An in-situ testing method for an emergency actuator of a vehicle door handle, characterized in that, include: Within the end domain, the actuator output terminal performs a reciprocating motion within a preset end domain displacement range, and the end domain transition relationship is collected to obtain reference cycle data; Perform a combination of reverse and forward actions on the actuator output to eliminate backlash and determine the starting point data of unilateral force. Based on the single-sided force starting point data, the stroke is divided into multiple segments. In each segment, the actions of advancing, stopping, reversing, and restarting are performed. The rebound characteristics and restart characteristics are compared with the reference beat data to obtain the abnormal window position data. Perform detection actions of different directions or amplitudes within the range corresponding to the abnormal window position data, and obtain the abnormal type label based on the change of restart characteristics; Symmetrical micro-motion actions are performed on both sides of the abnormal window position data, and the abnormality attribution result is obtained based on the comparison results of force-displacement characteristics. Based on the abnormal window location data, the abnormal type label, and the abnormal attribution result, the corresponding opening action prediction result is output.

2. The in-situ testing method for the vehicle door handle emergency actuator according to claim 1, characterized in that, The process of performing a reciprocating motion on the actuator output within a preset end-domain displacement range, collecting end-domain transition relationships, and obtaining reference cycle time data includes: Without entering the effective working stroke, the actuator output terminal performs a reciprocating motion within the end domain transition displacement range to trigger the transition process from a low load state to a force-supported state. During the transition process, the correspondence between the displacement characteristics of the action process and the load response is recorded to obtain time-series data characterizing the response characteristics of the end domain. Based on the time series data, the end-domain load features are extracted and correlated with the displacement features to obtain the reference beat data.

3. The in-situ testing method for the emergency actuator of the vehicle door handle according to claim 1, characterized in that, The combined reverse and forward actions performed on the actuator output to eliminate backlash and determine the unilateral force initiation data include: Under the end domain state determined according to the reference beat data, reverse yielding is performed on the actuator output end within the backlash elimination alignment displacement range so that the idle stroke in the kinematic chain is reduced to no more than a preset threshold. Within a preset continuous time window after the reverse yielding is completed, the actuator output end is subjected to forward contact within the same displacement range to compress the force-bearing surface of the kinematic chain to the same side contact state. Based on the results of the forward contact process, the correspondence between the output displacement characteristics and the load response is recorded to obtain the single-sided force starting point data, which serves as the benchmark for subsequent stroke segment discrimination.

4. The in-situ testing method for the emergency actuator of the vehicle door handle according to claim 1, characterized in that, Based on the single-sided force starting point data, the stroke is divided into multiple segments. Within each segment, propulsion, parking, retraction, and restart operations are performed. The rebound characteristics and restart characteristics are compared according to the reference beat data to obtain abnormal window position data, including: Under the state corresponding to the single-sided force starting point data, the starting point and end domain boundary of the effective working stroke are determined according to the reference beat data, and the effective working stroke is divided into multiple stroke segments according to the preset rules to obtain segment index data; Based on the segment index data, advance, stop, retraction and restart operations within the segment's retraction displacement range are executed sequentially according to a preset rhythm within each stroke segment. The synchronous timing correspondence between the displacement characteristics and load response within the corresponding stroke segment is recorded to obtain the segment rhythm timing data. Based on the intra-segment beat timing data, the rebound and restart features after the pause are extracted in each stroke segment, and aligned with the baseline features corresponding to the reference beat data to obtain intra-segment alignment feature data. Based on the intra-segment alignment feature data, a joint comparison is performed among multiple travel segments, and travel segments with continuous deviations are filtered out according to a preset joint deviation threshold to obtain candidate segment data; Around the segment boundary corresponding to the candidate segment data, encrypted advance, stop, retraction and restart operations are performed within the local refined displacement range, and local refined comparison is performed to determine the continuous coordinate range of the deviation interval and obtain the abnormal window position data. The abnormal window location data is checked for consistency with the segment index data, and the abnormal window location data is output when the check passes.

5. The in-situ testing method for the emergency actuator of the vehicle door handle according to claim 4, characterized in that, The process involves performing encrypted advance, pause, retraction, and restart operations within a locally refined displacement range around the segment boundary corresponding to the candidate segment data, and conducting local refined comparisons to determine the continuous coordinate range of the deviation interval, thereby obtaining abnormal window position data, including: Around the segment boundary corresponding to the candidate segment data, the refinement center coordinates are determined according to the changing trend of the alignment feature data within the segment, and the local refinement displacement range is set at its adjacent positions before and after. Within the localized refined displacement range, encrypted advance, stop, retraction and restart operations are executed sequentially according to a preset rhythm, and the corresponding displacement characteristics and load response are recorded to obtain refined time sequence data; The refined time series data is locally compared. When the rebound and restart features of adjacent positions show a change in sign and form a continuous deviation interval, the continuous coordinate range of the deviation interval is determined, and the abnormal window position data is output.

6. The in-situ testing method for the emergency actuator of a vehicle door handle according to claim 1, characterized in that, The process involves performing detection actions of different directions or amplitudes within the range corresponding to the abnormal window location data, and obtaining an abnormality type label based on the changes in the restart features, including: Within the range defined by the abnormal window position data, and under the state corresponding to the single-sided force starting point data, a detection action is performed on the actuator output end along the positive direction within the window direction detection displacement range. The restart characteristics after detection are recorded and compared with the baseline restart characteristics corresponding to the reference beat data to obtain the positive detection characteristic data. Within the range defined by the abnormal window position data, and under the state corresponding to the single-sided force starting point data, a detection action is performed on the actuator output end in the reverse direction within the window direction detection displacement range. The restart characteristics after detection are recorded and compared with the baseline restart characteristics corresponding to the reference beat data to obtain the reverse detection characteristic data. Based on the comparison results between the forward detection feature data and the reverse detection feature data, the difference in direction switching of the restart feature is extracted to obtain direction-dependent data; Within the range defined by the abnormal window position data, the detection action is performed step by step within the window amplitude detection displacement range according to the preset amplitude level sequence, and the restart characteristics corresponding to each amplitude level are recorded to obtain amplitude increment data; Based on the joint determination result of the direction-dependent data and the amplitude-increasing data, anomaly type labels of reversible or irreversible anomalies are output.

7. The in-situ testing method for the emergency actuator of a vehicle door handle according to claim 6, characterized in that, The step of outputting anomaly type labels (reversible or irreversible) based on the joint determination result of the direction dependence data and the amplitude increment data includes: Under the same conditions of unilateral force starting point data and reference beat data, the directional difference index data are determined based on the deviation of the forward detection feature data and the reverse detection feature data from the baseline restart feature. According to the preset amplitude level sequence, the detection action is performed step by step within the window amplitude detection displacement range to determine the minimum amplitude level that restores the restart feature to the range corresponding to the reference beat data, and the amplitude level data is obtained. Based on the joint judgment condition of the directional difference index data and the span amplitude level data, when the directional difference index data is not less than the preset threshold and there is a limited amplitude level that can restore the restart feature to the range corresponding to the reference beat data, a reversible anomaly type label is output; when the directional difference index data is less than the preset threshold or has not been restored to the range corresponding to the reference beat data at the maximum amplitude level, an irreversible anomaly type label is output.

8. The in-situ testing method for the emergency actuator of a vehicle door handle according to claim 1, characterized in that, The process involves performing symmetrical micro-motion movements on both sides of the abnormal window position data, and obtaining an anomaly attribution result based on the comparison of force-displacement characteristics, including: Within the adjacent intervals on the front and rear sides corresponding to the abnormal window position data, symmetrical detection displacement ranges are set respectively. Micro-motion is performed on the actuator output end, and the correspondence between displacement characteristics and load response is recorded to obtain the symmetrical feature data on the front and rear sides. The force-displacement sequences of the front-side symmetrical feature data and the rear-side symmetrical feature data are compared to obtain a symmetrical comparison result. Based on the symmetrical comparison results, if the overall curves of the features on both sides deviate but remain relatively consistent, the deviation is attributed to the actuator drive link side; if the features on both sides deviate only locally on one side, the deviation is attributed to the door lock mechanism side, and the abnormal attribution result is output.

9. The in-situ testing method for the emergency actuator of a vehicle door handle according to claim 1, characterized in that, The step of outputting the corresponding activation action prediction result based on the abnormal window location data, the abnormal type label, and the abnormal attribution result includes: Based on the correspondence between the abnormal window location data, the abnormal type label and the abnormal attribution result, a joint judgment condition set is constructed, and each input result is associated under the same judgment framework; Under the joint decision condition set, inductive decision is performed on the input results of different combinations, and the prediction result of the opening action corresponding to the combination is output.

10. An in-situ testing device for an emergency actuator of a vehicle door handle, characterized in that, include: The end domain acquisition module is used to perform reciprocating motion on the actuator output end within a preset end domain displacement range within the end domain range, acquire the end domain transition relationship, and obtain reference cycle data. The clearance alignment module is used to perform a combination of reverse and forward actions on the actuator output to eliminate clearance and determine the force start point data on one side. The segment discrimination module is used to divide the stroke into multiple segments based on the single-sided force starting point data, perform advance, stop, retraction and restart operations in each segment, and compare the rebound characteristics and restart characteristics according to the reference beat data to obtain abnormal window position data. The anomaly classification module is used to perform detection actions of different directions or amplitudes within the range corresponding to the anomaly window position data, and obtain anomaly type labels based on the changes in restart features; The attribution determination module is used to perform symmetrical micro-motion actions on both sides of the abnormal window position data, and obtain the abnormal attribution result based on the comparison result of force displacement characteristics; The prediction output module is used to output the corresponding start action prediction result based on the abnormal window position data, the abnormal type label, and the abnormal attribution result.

Images