Method and medium for path authorization discrimination based on post-propagation consequence residual and forward propagation support

Abstract

This invention relates to a path authorization discrimination method and storage medium based on propagation consequence residuals and forward propagation support. The method includes: acquiring candidate paths; determining the path time within an event based on the path triggering time, acquiring subsequent propagation consequence information, updating the component residuals of each expected consequence in the propagation consequence set, and calculating the total path residual of the path; determining the effective time of early support, generating and using early positive support within the effective time of early support, and generating global positive support; acquiring enhanced confirmation conditions, and determining the authorization variable and path level variable of the candidate path based on the total path residual, early positive support, global positive support, and enhanced confirmation conditions. Compared with the prior art, this invention can reduce the risk of false authorization caused by relying solely on binary confirmation, improve the stability of locally induced path authorization discrimination, suppress the risk of false authorization in non-deployment conditions, and is compatible with scenarios with late central propagation and sensor degradation.

Description

Technical Field

[0001] This invention relates to the field of vehicle data processing technology, and in particular to a path authorization discrimination method and medium based on propagation consequence residuals and forward propagation support. Background Technology

[0002] Modern vehicle frontal collision detection algorithms generally use front satellite sensors as distributed sensing units external to the electronic control unit (ECU). These sensors exhibit early signal response, a characteristic determined by their installation location and the vehicle's impact propagation structure. In a severe frontal collision, the impact energy first acts on the front end of the vehicle and then gradually propagates towards the center through energy-absorbing boxes, longitudinal beams, and the vehicle frame. Because the front satellite sensors are located at the very front of the vehicle, they can capture local responses in the initial stages of impact propagation, with signal changes significantly earlier than those of the central sensor built into the ECU. Therefore, in critical scenarios such as severe frontal collisions, offset collisions, and small overlap collisions, the front satellite sensors can provide early judgment criteria for collision identification candidate paths, effectively improving the algorithm's collision response timeliness.

[0003] However, the front satellite sensor can only sense the local motion state of its surrounding structure and mounting point, resulting in a significant local sensitivity defect in its output signal. Conditions such as localized hard contact, low-speed impacts, bracket vibrations, maintenance knocks, and isolated sensor malfunctions during vehicle operation and maintenance can generate high-amplitude short pulse signals in the front local channel. The waveform characteristics of these signals in the local sensing channel are highly similar to real collision pulses, but they lack the central vehicle response and velocity accumulation characteristics essential for a severe frontal collision. Simultaneously, various disturbances such as curb impacts, transportation disturbances, and mounting bracket vibrations can also cause significant responses in the front local channel. These responses are fundamentally different from the propagation process of a severe frontal collision at the vehicle level, easily interfering with the front collision algorithm's candidate path determination logic, thus leading to localized false triggering problems.

[0004] To mitigate the risk of local false triggering caused by front-end satellite sensors, the existing engineering system has developed a mature correction and processing scheme, which mainly includes central safing verification (central security verification), secondary confirmation, multi-sensor consistency verification, dynamic threshold adjustment, and fixed no-fire window (fixed no-trigger window) filtering. The above methods have clear application value and practicality in engineering implementation.

[0005] Central safing verification is the core constraint mechanism. It relies on the ECU's built-in central sensor to build an independent confirmation channel, performing double-layer constraint verification on the triggering logic of the front satellite sensor path. Its core judgment rule is: the algorithm only outputs a collision request when the front sensor path meets the triggering conditions and the central sensor channel completes effective confirmation within the calibration time constraint; otherwise, it directly suppresses the trigger command. This judgment logic is simple, clear, and intuitive, significantly reducing the difficulty of algorithm development and calibration, and is currently the mainstream engineering implementation method in the industry. However, this traditional binary confirmation logic has significant technical shortcomings, the core problem being insufficient ability to distinguish operating conditions. When the binary confirmation conditions are not met, existing algorithms often struggle to distinguish different reasons for non-confirmation, including scenarios such as the collision impact not yet reaching the center of the vehicle, the vehicle speed change not yet effectively accumulating, the overall collision mode not yet stabilizing, the central confirmation channel signal being temporarily unobservable, and local impacts being non-collision-related abnormal interference. At the same time, a single confirmation point cannot explicitly depict the dynamic evolution of the collision signal from early feature loss to later feature completion, resulting in insufficient recognition accuracy for complex operating conditions.

[0006] In addition, other conventional optimization methods also have limitations: the dynamic threshold method only focuses on adjusting the threshold parameters at different stages of the collision event, and the optimization dimension is relatively simple; the fixed no-fire window relies on a fixed duration delay mechanism to filter short-term pulse interference, and can only achieve simple static filtering.

[0007] In summary, existing forward collision algorithms and their supporting verification mechanisms suffer from numerous problems, such as a single discrimination dimension, weak ability to distinguish working conditions, and insufficient accuracy in candidate path determination. Therefore, it is urgent to systematically discriminate and optimize the accuracy of candidate paths in existing forward collision algorithms. Summary of the Invention

[0008] The purpose of this invention is to overcome the shortcomings of the existing technology by providing a path authorization discrimination method and medium based on propagation consequence residuals and positive propagation support. This method updates the path status based on observable propagation consequences and positive evidence within the event time, and corrects the authorization variables and grade variables of the path accordingly. This reduces the risk of false authorization caused by relying solely on binary confirmation and improves the stability of locally induced path authorization discrimination.

[0009] The objective of this invention can be achieved through the following technical solutions: A path authorization determination method based on propagation consequence residuals and forward propagation support includes the following steps: After the front collision algorithm is activated, a candidate path is obtained that is induced by a local impact at the front. The local impact corresponding to the candidate path meets the set requirements and the corresponding front satellite sensor is in a healthy state. The path time within the event is determined based on the path triggering time. Subsequent propagation consequence information is obtained based on the path time within the event. Based on a pre-set propagation consequence set, the component residuals of each expected consequence in the propagation consequence set are updated according to the subsequent propagation consequence information, and the total path residual of the path is calculated. The effective time of early support is determined based on the path triggering time. Early positive support is generated and used within the effective time of early support, and global positive support is generated. After the effective time of early support expires, the global positive support is used as the basis for positive support for path authorization determination. Obtain enhanced confirmation conditions, and determine the authorization variables and path level variables of the candidate path based on the total path residual, early positive support, global positive support, and enhanced confirmation conditions.

[0010] Furthermore, the local impact meeting the set requirements includes: The local impact intensity reaches the path threshold and continues for a set duration.

[0011] Furthermore, the subsequent propagation consequences information includes central response information, severity information, pattern continuity information, and confirmation and support information.

[0012] Furthermore, the anticipated consequences of the propagation consequences set include central vehicle response, severity accumulation, path pattern continuity, and authorization confirmation support.

[0013] Furthermore, the component residuals of each expected consequence are updated using the following formula: N j,p (k) = sat[N j,p (k-1)+ΔN miss,j,p (k)+ΔN late,j,p (k)+ΔN brk,j,p (k)-ΔN fill,j,p (k)-ΔN rec,j,p (k)] Where j represents the consequence type, N j,p (k) represents the component residual corresponding to consequence j on path p; ΔN miss,j,p (k), ΔN late,j,p (k) and ΔN brk,j,p (k) represents the weighted contribution of in-window missing, overdue missing, and mode interruption, respectively; ΔN fill,j,p (k) and ΔN rec,j,p (k) represents the initial compensation offset and the stable recovery contribution, respectively; sat[·] represents the saturation limit.

[0014] Furthermore, when calculating the total residual of the path, each expected consequence is grouped according to its source, the component residuals of each expected consequence within the group are limited, and the residuals of different groups are weighted and synthesized into the total residual of the path. The residual weights of different groups are determined based on the independence of evidence sources, propagation reliability, and calibration of real vehicle collision / non-deployment condition data. The contribution within the same source group is constrained by the group limit.

[0015] Furthermore, the values ​​of the early positive support and the global positive support are both 0 or 1; wherein, the early positive support is formed based on the multi-point consistency of the front satellite sensors, local persistence and local pattern stability within the early window, and the global positive support is formed based on the central vehicle response, severity accumulation, pattern continuity and confirmation support; the corresponding evidence group is set to 1 after meeting the threshold and passing the same source amplitude limiting, otherwise it is set to 0.

[0016] Furthermore, the authorization variables and path ranking variables used to determine candidate paths specifically include: The residual state is determined based on the total path residual and the state threshold. The residual state includes a pass state, an observe state, a degraded state, and a suppressed state. A corrected authorization variable is obtained based on the path latch state, residual state, observable coverage, positive support state, and enhanced confirmation conditions. : When the path is not latched or the residual state is in a suppressed state =0; When the residual state is either passable or observable, the observable coverage is not less than the set value, and K G p When (k)=1, = A p 0 (k), K G p (k) represents global positive support, A p 0 (k) represents the original authorized variable; When the residual state is either passable or observable, the observable coverage is not less than the set value, and K G p (k)=0, K E p (k)=1、d p (k)≤T E hour, = A p 0 (k), K E p (k) indicates early positive support, d p (k) represents the path time within the event; When the residual status is in a degraded state, or the residual status is in a pass or observation state and the observable coverage is less than a set value, = A p 0 (k)∧E p (k), E p (k) indicates an enhanced confirmation condition, and ∧ indicates a logical AND; When the residual state is either passable or observable, the observable coverage is not less than the set value, and K G p (k)=0, and does not satisfy K E p (k)=1、d p (k)≤T E When the early support conditions are met, =0; The revised path-level variables are obtained based on the revised authorization variables, residual status, observable coverage, positive support status, and enhanced validation conditions. : when When =0, =0; When the residual state is either passable or observable, the observable coverage is not less than the set value, and K G p When (k)=1, = l p 0 (k), l p 0 (k) represents the original path ranking variable; When the residual state is either passable or observable, the observable coverage is not less than the set value, and K G p (k)=0, K E p (k)=1、d p (k)≤T E hour, =min(l p 0 (k), L E,max ), L E,max Indicates the initial level cap; When the residual state is in a degraded state, or the residual state is in a pass state or an observation state and the observable coverage is less than the set value and E p When (k)=1, =min(l p 0 (k), L C,max ), L C,max Indicates the upper limit of the conservative level; The observable coverage is the ratio between the sum of the weights of the currently observable and applicable key consequences of the path and the sum of the weights of the key consequences that should be observed and applicable to the path.

[0017] Furthermore, after a candidate path enters the suppression state, the recovery conditions include the total path residual decreasing to below the recovery threshold, and the global positive support and enhancement confirmation conditions being met continuously for a set period.

[0018] The present invention also provides a computer-readable storage medium including one or more programs executable by one or more processors of an electronic device, the one or more programs including instructions for performing the path authorization discrimination method based on propagation consequence residuals and forward propagation support as described above.

[0019] Compared with the prior art, the present invention has the following beneficial effects: This invention describes the process of a candidate path from local triggering to authorization as a temporal consistency judgment between early local evidence and the consequences of propagation throughout the vehicle. It considers residual state, early positive support, global positive support, observable coverage, and enhanced confirmation conditions to jointly discriminate and correct the authorization variables and path ranking variables of candidate paths, resulting in high reliability. The joint discrimination and correction of the authorization variables and path ranking variables of candidate paths reduces the risk of false authorization and improves the stability of the judgment.

[0020] Furthermore, by recording the processes of missing, overdue, supplemented, interrupted, and recovered, this invention extends traditional binary confirmation to path-level state discrimination, which can distinguish different working conditions such as late propagation of real collisions, local hard contact, abnormal spikes of single sensors, and sensor degradation.

[0021] Furthermore, this invention employs common-origin consequence limiting, observable coverage, and enhanced confirmation conditions to jointly constrain authorized output, which can reduce the risk of duplicate penalties or misauthorization caused by a single link anomaly, and maintain conservative output when critical consequences are unobservable.

[0022] Furthermore, the present invention can be implemented using fixed-point integers, lookup windows, and finite state machines, which facilitates the verification of worst-case execution time, storage usage, and diagnostic degradation strategies in cyclic tasks such as airbag ECUs. Attached Figure Description

[0023] Figure 1 This is a schematic diagram of the process of the present invention; Figure 2 This is a schematic diagram of the single-path process of the present invention, from strong local impact to residual update, positive support calculation and authorization correction. Detailed Implementation

[0024] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are based on the technical solution of the present invention and provide detailed implementation methods and specific operating procedures. However, the scope of protection of the present invention is not limited to the following embodiments.

[0025] Terminology Explanation In this embodiment, "consequence" refers to observable evidence that should appear in the whole vehicle or confirmation link according to the candidate path propagation mechanism after the occurrence of a frontal local impact, including central vehicle response C, severity accumulation V, path pattern continuity M, and authorization confirmation support Q.

[0026] "Confirmation support" refers to confirmatory evidence used to authorize or restrict path authorization, including central safing, redundant sensors, independent confirmation channels, and system severity confirmation; "positive support" is a state of positive evidence relative to the propagation consequence residual, and is divided into early positive support K. E p (k) and global positive support K G p (k), whose values ​​are all 0 or 1.

[0027] "Velocity accumulation" refers to the ΔV, cumulative acceleration, or severity integral obtained within the event time based on the central inertial measurement link or equivalent vehicle motion link, used to characterize the degree of accumulation of collision energy after it propagates to the vehicle level.

[0028] "Latching" refers to maintaining a record of an event or state's validity after it first meets certain conditions, until the event is reset. For example, path latching is used to retain the triggering time of candidate paths, while completion latching is used to record that a certain propagation consequence has consistently appeared within the completion window.

[0029] In this embodiment, "not true" means that the corresponding binary variable is 0; "true" means that the corresponding binary variable is 1. Unless otherwise specified, all support states, health states, and mask states are interpreted according to this binary semantics.

[0030] Symbol Explanation: In this embodiment, the variables for path p are uniformly written using subscript / superscript notation, for example, K E p (k), K G p (k), K p (k), A p 0 (k) l p 0 (k) d p (k), E p (k), O p (k), N j,p(k), N p (k), Z p (k), R j,p and M j,p (k); T E Indicates an early support limit on the effective time, O min L represents the observable coverage threshold. E,max and L C,max S represents the early level cap and the conservative level cap, respectively. E p (k), S G p (k) represents the early / global evidence score, respectively, and θ E θ G Σ represents the threshold; ∈ represents summation; min(x,y) represents taking the smaller of the two values; ECU represents electronic control unit; abuse represents non-deployment or abuse conditions.

[0031] This embodiment provides a path authorization discrimination method based on propagation consequence residuals and forward propagation support. It studies the authorization discrimination problem of candidate paths induced by strong local impacts in the early stage of the event. Based on the local candidate paths formed in the early stage of the event, and according to the occurrence status of subsequent propagation consequences and the status of forward propagation support, a modified path authorization variable and path level variable are constructed.

[0032] refer to Figure 1 As shown, the path authorization discrimination method based on propagation consequence residuals and forward propagation support provided in this embodiment includes: Data Input: Acquire input data, including: front satellite sensor signals, central vehicle response, ΔV / severity index, mode / side identification information, confirmation support, sensor health status, consequence observability mask, and path applicability flag R. j,p Original Authorization A p 0 (k) Original level l p 0 (k); Candidate path formation and screening: Obtain the original candidate path set P, and extract the front local induction path set P from it. F Determine whether a strong local impact on the leading satellite occurs on each path. If a strong local impact occurs for the first time, latch the path and record the trigger time. ; Calculation of propagation consequence residuals and positive propagation support: The path time within an event is determined based on the path trigger time. Subsequent propagation consequence information is obtained based on this time. Based on a pre-set propagation consequence set, the component residuals of each expected consequence in the propagation consequence set are updated according to the subsequent propagation consequence information, and the total path residual for that path is calculated. The effective time for early support is determined based on the path trigger time. Early positive support is generated and used within this effective time, and global positive support is generated based on central vehicle response, severity accumulation, pattern continuity, and confirmation support. After the effective time for early support expires, the global positive support is used as the basis for positive support in path authorization determination. Authorization discrimination and state machine: Obtain enhanced confirmation conditions, and determine the authorization variables and path level variables of the candidate path based on the total path residual, early positive support, global positive support and enhanced confirmation conditions.

[0033] In other embodiments, refer to Figure 1 As shown, the above method also includes system-level arbitration. The system-level arbitration logic receives the modified local path authorization and path level output by this method, and combines them with other path requests, independent severe collision path requests, system severity indicators and confirmation conditions to generate a system-level protection request.

[0034] This embodiment describes the method of the present invention from the aspects of problem localization, candidate path and state variable definition, propagation consequence residual model, forward propagation support and authorization discrimination, periodic implementation process, and state evolution analysis under typical working conditions.

[0035] 1. Problem Identification Existing forward collision algorithms and their corresponding verification mechanisms suffer from numerous problems, including a single discrimination dimension, weak ability to distinguish working conditions, and insufficient accuracy in candidate path determination. The applicant has discovered that, from the perspective of temporal propagation consistency, the timing of the subsequent propagation consequences, missing states, and completed states after a localized strong impact can still be further described as path-level state variables. For the temporal evolution process after a localized impact, the timing and completed states of the central vehicle response, speed accumulation, mode continuity, and confirmation support during the waiting period are additional discriminative information of interest to the method of this invention.

[0036] Based on existing technology, the discrimination problem addressed by the method of this invention is: after a local strong impact on a satellite sensor triggers a certain path, the path state is updated within the event time based on observable propagation consequences and positive evidence, and the authorization variables and grade variables of the path are corrected accordingly.

[0037] The differences between the method of this invention and the prior art are shown in Table 1.

[0038] Table 1. Description of typical engineering methods and the method of this invention in key modeling dimensions. Binary safing Forward satellite path and center confirmation position Only the status of central confirmation is recorded; missing, overdue, and supplementary records are not documented. Authorization / Suppression Binary Results Fixed no-fire window Forward satellite path and latency Suppress local paths within the waiting window; do not retain the state of consequence components after the window ends. Path request results after the window ends Dynamic threshold Event phase and sensor amplitude / integration The threshold switches according to the event stage; no residual propagation consequences are established. Threshold determination result Method of the present invention Evidence of front satellite path and vehicle propagation Establish missing, overdue, interrupted, completed, and restored states using C, V, M, and Q. Path authorization, path level, residual status

[0039] 2. Research Subjects and Overall Framework After a frontal collision event is triggered by the system, the method of this invention (vehicle frontal collision algorithm) runs at a fixed sampling period, denoted as k. The algorithm maintains several candidate paths, denoted as P. Candidate paths are defined according to sensor region, collision side, and collision mode, such as left frontal path, right frontal path, center path, full-width frontal collision path, offset collision path, and small overlap path. This method studies the set of candidate paths induced by a localized strong impact from a frontal satellite sensor, denoted as... P F ⊆ P set P F The paths within the path enter the residual update, forward support calculation, and authorization correction process. For each path p∈P F A strong frontal localized impact indicates a strong response in a specific localized area at the front of the vehicle; path determination requires observing the expected propagation consequences after this response. This embodiment categorizes the expected propagation consequences into four types, as shown in Table 2.

[0040] Table 2 Expected propagation consequences after a strong local impact at the front C Central vehicle response The ECU's central sensor generates a vehicle deceleration response within the propagation window that aligns with the candidate front path. Central longitudinal acceleration, central inertial measurement link V Severity accumulation The central corridor's integral and severity indicators have reached the calibration threshold for that path. ΔV, cumulative acceleration, severity integral M Path pattern continuity If candidate paths are consecutively valid within the valid window, the path competition results do not reverse. Pattern bit consecutive counting, majority voting, path race results Q Authorization confirmed support The verification conditions for the central safing, second sensor, and independent link, as per the calibrated configuration, are met. Saving position, redundant sensors, independent confirmation channel

[0041] The path pattern continuity consequence is supported by short-window counting and majority voting to generate stable patterns, rather than directly using single-cycle pattern bits. This process reduces the impact of path contention, noise flicker, and single-cycle pattern transitions on the path state.

[0042] Furthermore, the short window is a local sliding window used for continuous counting or majority voting, typically taking 3-8 algorithm sampling periods; for example, a 1ms sampling period corresponds to approximately 3-8ms, and a 0.5ms sampling period corresponds to approximately 1.5-4ms. The specific window length is calibrated based on the noise level of the front satellite sensors, the duration of local impacts, and the stability of the path pattern.

[0043] The propagation consequence residuals are calculated based on the absence of observable and applicable consequences. When a consequence becomes unobservable due to sensor failure, communication anomaly, or diagnostic degradation, no new missing, overdue, or interrupted penalties are imposed on that consequence in the current period. Existing component residuals are frozen or maintained according to the diagnostic strategy, and no positive support is provided. When a consequence is not applicable to path p itself, the applicable flag R is used to... j,p =0 is removed from the set of observable consequences, coverage numerator, and coverage denominator for that path; this case is not considered an unobservable channel. Therefore, an observable mask M is set for each type of consequence. j,p(k), where 1 represents an observable state and 0 represents an unobservable state. When the observable mask is 1, the consequence is involved in missing item detection, completion cancellation, stability recovery, and positive support calculation; when the observable mask is 0, the existing component residuals of the consequence are frozen or preserved, and no missing items or positive support items are added.

[0044] The above 1 / 0 is defined as the observable mask semantics used in this embodiment: 1 indicates that the consequence can currently be effectively observed by the corresponding sensor or link, and 0 indicates that the consequence is unobservable due to fault, communication anomaly, or diagnostic degradation; Path not applicable, not passing M. j,p (k)=0 indicates that, through R j,p =0 indicates that it is used to distinguish between two states: "should be observed but cannot be observed" and "this path does not need to be observed".

[0045] Observable coverage O p (k) can be represented as: J p M is the set of key consequences of path p. j,p (k) is the observable mask for consequence j, R j,p w is an applicable flag for whether path p needs to observe consequence j. j For the observable coverage weight of consequence j; when all consequences have equal weight, O p (k) degenerates into the ratio of the number of observable and applicable consequences to the number of observable and applicable consequences; when When it is 0, O p (k) Process as 0 and enter the enhanced confirmation branch for insufficient coverage.

[0046] Propagation Consequence Residual N p (k) indicates the degree of lack of expected propagation consequences of path p.

[0047] Positive propagation supports K p (k) represents the state of positive evidence obtained by path p, and is divided into early positive support K. E p (k) and global positive support K G p (k): K p (k)=K E p (k)∨K G p (k) Among them, K E p (k) is used to characterize the spatial consistency, local persistence, and local pattern stability shortly after a local impact; K G p(k) is used to characterize the central vehicle response, severity accumulation, pattern continuity, and confirmation support after a localized impact propagates to the vehicle level; ∨ represents logical OR, when K E p (k)=1 or K G p When (k)=1, K p (k)=1, when both are 0, K p (k)=0. Early positive support participates in path authorization determination within the early window after a local impact; global positive support participates in authorization determination after the whole vehicle propagation evidence has stabilized.

[0048] Let the original local candidate path authorization be A. p 0 (k), the path-level authorization variable after propagation consistency judgment is A′. p (k). When the original algorithm outputs a multi-level path hierarchy, the original path hierarchy is denoted as l. p 0 (k), the corrected path level is denoted as l′ p (k). Path-level correction results are derived from A′ p (k) and l′ p (k) represents the path and is input to the system-level arbitration function as a local path. The system-level arbitration function then combines other path requests, system severity indicators, and confirmation conditions to generate a system-level protection request. The relationship between path-level state updates and system-level arbitration is as follows: Figure 1 As shown.

[0049] 3. Residual Model of Propagation Consequences 1) Path triggering and latching.

[0050] For path p, the algorithm continuously monitors the local impact intensity, short-window duration, and health status of the corresponding forward satellite sensor. When the local impact intensity reaches the path threshold, the duration meets the requirements, and the corresponding forward satellite sensor is healthy, path p enters a strong local impact state. After a strong local impact occurs, the algorithm latches the path and records the path trigger time. This moment serves as the origin for subsequent consistency judgments, and the intra-event path time is defined as... d p (k)=k-

[0051] from Initially, the algorithm observes the subsequent propagation consequences of the local impact, such as the central vehicle response, severity accumulation, pattern continuity, and confirmation support, based on the path time within the event. When the event ends, the system enters a stable state, and the preset reset conditions are met, the path latch state, completion state, residual state, and authorization state are cleared.

[0052] 2) Observe the window and complete the window.

[0053] Different propagation consequences have different calibration occurrence times. Central vehicle response requires structural impact energy to be transferred to the central measuring point; velocity accumulation requires integration time; pattern continuity requires path development over multiple cycles; and confirmation support has inherent link delays. Missing observation windows and completion windows are set for each type of consequence, and overdue stages and event resets are added as additional states for residual updates. Their state relationships are defined by the window parameters in Table 3 and below.

[0054] The relationship between the observation window, the completion window, and the effective window is as follows: The observation window is used to determine whether a certain consequence is missing within the expected propagation time; the completion window covers the observation window and can be extended backward by an allowable delay interval, used to handle samples with late central propagation but still conforming to the real collision propagation mechanism; the effective window is used to continue to check whether its stable support is maintained after the consequence is completed.

[0055] In one embodiment, the observation window for consequence j can be denoted as [t]. j,start ,t j,obs The completion of the window can be recorded as [t]. j,start ,t j,fill The effective window can be denoted as [t]. j,valid,start ,t j,valid,end ], and satisfy t j,start ≤t j,obs ≤t j,fill When the consequences that have been addressed disappear within the effective window, trigger mode interruption or stability degradation related residuals.

[0056] Table 3 State variables and update directions of component residuals <![CDATA[Missing item ΔN within the window miss > The consequences are under observation, the corresponding channel is observable, and the latch replenishment has not been established. component residual increase <![CDATA[Missing item after due date ΔN late > After the observation window ends, the consequences are still not fully resolved and the corresponding channel remains observable. Component residuals increase by overdue option weighting <![CDATA[Mode interruption item ΔN brk > The M-consequence has been established stably; the corresponding channel is observable within the subsequent effective window, and the stable mode supports loss. The residual of the M component increases. <![CDATA[First fill in the offset term ΔN fill > The consequences are observable within the completion window and the stability condition is met for the first time. The corresponding component residuals decreased and were latched in one go, and have been replenished. <![CDATA[Stable recovery term ΔN rec > The consequences have been rectified and are observable and persistently stable within the current recovery window. The corresponding component residual decreases with the recovery rate Event Reset Event ended, static stability condition met, diagnostic repositioning condition met. Clear path latches, complete latches, residuals, and authorization status.

[0057] 3) Complete the latch.

[0058] Original consequence evidence is generated periodically; when not latched, it only indicates the current period's status. Once a consequence briefly meets the stability condition within a window, it indicates that the consequence has occurred; subsequent signal fluctuations should not re-enter the "never occurred" state. This method sets a completion latch amount for each type of consequence. After a consequence first meets the stability condition within the completion window, it is latched as completed; after latching, the accumulation of "uncompleted" residuals for that consequence ceases.

[0059] For central vehicle response, speed accumulation, and confirmation support, the completion evidence comes from the corresponding consequence evidence. For pattern continuity, the completion evidence comes from stable pattern support generated by window counting, rather than directly using single-cycle pattern bits. The first completion corresponds to a one-time offset term, used to offset part of the residual caused by the previous absence of the consequence. When the same consequence continues to hold, the recovery process is described by the stable support recovery term.

[0060] 4) Missing items, overdue items, interrupted items, completed items, and restored items.

[0061] The propagation consequence residuals consist of three types of negative terms and two types of positive offsetting terms. All residual increment terms are in R0. j,p Calculate under the condition that R = 1 and the corresponding consequences are currently observable; when R j,p When M = 0, consequence j does not participate in the missing penalty, positive support contribution, total residual synthesis, and coverage denominator accumulation for that path; when M j,p When (k)=0, N j,p (k)=N j,p (k-1) or remain as per the diagnostic strategy, without adding ΔN. miss ΔN late ΔN brk ΔN is not triggered. fill ΔN rec Missing items within a window occur when the current period is within the missing observation window of a certain consequence, the consequence is observable, and it has not yet been filled in; overdue missing items occur when a consequence exceeds the regular observation window, is still not filled in, and the corresponding channel is observable; the mode interruption item is used for mode continuity consequences, describing the state where the mode was once stable and the corresponding channel was observable, but the stable mode support disappeared within the subsequent effective window; the first filling cancellation item acts on the corresponding component residual once when the consequence is observable and the filling condition is met for the first time; the stability recovery item reduces the corresponding component residual at the calibration rate when the filled consequence is observable and remains stable.

[0062] 5) The decline in residuals is triggered by subsequent evidence completion and stabilization support.

[0063] For each type of consequence, the algorithm generates a stable support state through continuous counting and window majority voting. When a consequence has been completed and remains stable, the corresponding component residual is recovered at the calibrated rate. For samples with small overlaps and late central propagation, after the central vehicle response and velocity accumulation stabilize within subsequent windows, the corresponding component residual decreases with the recovery term.

[0064] 6) Component residuals and total path residuals.

[0065] To handle independent missing and independent recovery among different consequences, this method establishes component residuals for each type of propagation consequence. Each component residual is updated jointly by in-window missing, overdue missing, pattern interruption, initial completion cancellation, and stable recovery. Let j∈{C,V,M,Q}, the corresponding update formula can be written as: N j,p (k)=sat[N j,p (k-1)+ΔN miss,j,p (k)+ΔN late,j,p (k)+ΔN brk,j,p (k)-ΔN fill,j,p (k)-ΔN rec,j,p (k)] Where j represents the consequence type, N j,p (k) represents the component residual corresponding to consequence j on path p; ΔN miss,j,p (k), ΔN late,j,p (k) and ΔN brk,j,p (k) represents the weighted contribution of in-window missing, overdue missing, and mode interruption, respectively; ΔN fill,j,p (k) and ΔN rec,j,p (k) represents the initial compensation offset and the steady recovery contribution, respectively; sat[·] represents the saturation limit. The residual decrease is triggered only by the initial compensation offset and the steady support recovery; the waiting time itself does not generate a recovery term.

[0066] In practical systems, when the central vehicle response, speed accumulation, and basic verification are derived from the same central inertial measurement link, they are considered to have the same source of consequences. Directly adding the residuals of these three would result in duplicate penalties for delays and anomalies from the same observation link across multiple consequence items. This paper introduces a limiting mechanism for similar-source consequences: first, consequences with similar origins are grouped together, and the residuals within each group are limited; then, the residuals from different groups are weighted and synthesized into a total path residual N. p (k). When each consequence comes from an independent observation link, each consequence is grouped separately; when the central vehicle response, speed accumulation and basic confirmation mainly come from the same central observation link, the three are placed in the same group, and the impact of repeated penalties for related consequences is reduced by limiting the amplitude within the group.

[0067] In one embodiment, the total path residual can be expressed as N. p (k)=Σ g λ g ·min(Σ j∈Gg N j,p (k),C g ) calculate, where G g For the g-th evidence source group, C g For this set of limiting values, λ g λ represents the group weight. gThe weighting is determined based on the independence of evidence, the credibility of the link, the path type, and the sample calibration results; independent measurement links can be configured with higher weights, while shared evidence groups from the same source are limited and configured with lower or total limited weights.

[0068] Optionally, λ g Dimensionless integers or equivalent fixed-point calibration values ​​of 1-4 can be used; the total contribution of the homologous group after group amplitude limiting should not be higher than that of the independent key evidence group, and the specific value should be jointly calibrated with the path state threshold.

[0069] 4. Forward propagation support and path authorization judgment The propagation consequence residual is used to characterize the degree of accumulation of missing, overdue, or interrupted evidence in the applicable and observable expected consequences; the lower the residual, the less negative evidence there is that is inconsistent with the current path propagation mechanism; when the early consequences of an event have not yet entered the observation window and some consequences are unobservable, path authorization requires the joint participation of positive propagation support and enhanced confirmation conditions in the judgment.

[0070] Early positive support for K E p (k) is used to characterize the spatial consistency and persistence of front-end satellite sensor signals in the early stages of an event; global positive support K G p (k) is used to characterize the evidentiary state after a local impact propagates to the whole vehicle level. The evidentiary composition of both is shown in Table 4.

[0071] Table 4. Evidence supporting positive propagation Early positive support Multiple points of consistency in the front satellite Same-side and adjacent forward satellite sensors form a compatible response within the early window, reducing the impact of single-point spikes. Early positive support Local persistence and local timing The local energy at the front persists within a short window, and the response sequence of the front measurement points conforms to the candidate path hypothesis. Global positive support Central vehicle response and severity accumulation The central sensor establishes stable vehicle response, and the ΔV / severity integral reaches the path threshold. Global positive support Pattern Continuity and Confirmation of License The candidate paths are consecutively true within the event window, and the safing and independent confirmation conditions meet the authorization requirements.

[0072] Early positive support is generated within a short window after a localized impact and participates in path authorization determination. Global positive support is generated based on the central vehicle response, severity accumulation, pattern continuity, and confirmation support, and participates in path authorization determination after the vehicle propagation evidence stabilizes. After the effective time of early support has expired, path authorization no longer depends on early positive support, but is constrained by global positive support and enhanced confirmation conditions. For multiple frontal evidences from the same source, the algorithm employs grouped amplitude limiting. Amplitudes, integrals, and short-window energy derived from the same frontal satellite sensor are grouped into the same evidence group; evidence from different installation points, different sides, and independent measurement links are set as independent evidence groups.

[0073] Positive support for K E p (k) can be determined by the early evidence score S E p (k) Determine: Assign integer scores to the front-end multi-point consistency, local persistence, and local pattern stability, and limit the amplitude of the homologous evidence group; when d p (k)≤T E And SE p (k)≥θ E At that time, K E p (k)=1, otherwise K E p (k)=0.

[0074] Global positive support K G p (k) can be determined by the global evidence score S G p (k) Determine: Assign integer scores to the central vehicle response, severity accumulation, pattern continuity, and confirmation support, and require key global evidence to meet stability conditions; when S G p (k)≥θ G And when the key evidence set was not blocked by health gates, K G p (k)=1, otherwise K G p (k)=0.

[0075] Global positive support primarily comes from central vehicle response, speed accumulation, pattern continuity, and confirmation support. For central vehicle response and speed accumulation, the algorithm checks both "previously completed" and "currently stable": the consequence occurred within the completion window and remains stable within the current window. For pattern continuity and confirmation support, positive support checks the current stable state. When central vehicle response, speed changes, and basic confirmations primarily originate from the same central inertial link, they are grouped into the same positive evidence group; confirmation support from independent sensor links is grouped separately.

[0076] Total residual N along the path p (k) is divided into four states: pass, observe, degrade, and inhibit. The path state is denoted as Z. p (k)∈{Z0,Z1,Z2,Z3}. The residual state employs a recovery rule with hysteresis. After a path enters a suppressed state, the recovery conditions include the residual decreasing below the recovery threshold, and the global positive support and enhanced confirmation conditions being met consecutively for a set period. The enhanced confirmation condition, denoted as Ep(k), is used for degraded states or branches with insufficient observable coverage. It is generated based on confirmation rules that are more stringent or independent than global positive support, including high-level central vehicle response, higher speed changes or severity indicators, enhanced safing conditions, dual front sensor consistency, and independent confirmation channels or redundant link confirmations. The specific combination is determined by the path type and safety policy; when the corresponding enhanced confirmation rule is met, Ep(k)=1, otherwise Ep(k)=0.

[0077] Table 5 Path Status and Authorization Output Logic

[0078] The conditions in Table 5 are executed sequentially according to path latch, healthy state, suppressed state, degraded or insufficient coverage branch, positive support branch in pass / observe state, and enhanced confirmation condition. When O p (k) <O min When this happens, prioritize entering the enhanced confirmation branch, and do not apply K. G p (k) or K E p (k) Directly use the branch with the original authorization; when the residual is in the Z0 / Z1 state and O p (k)≥O min When global positive support is established, the corrected authorization follows the original path authorization; when the residual is in the Z0 / Z1 state and O p (k)≥O min When only early positive support is established, authorization is limited to the early window and subject to the level cap constraint; when the path is in a degraded state or has insufficient observable coverage, authorization must be granted by enhanced confirmation conditions.

[0079] The "residual state" in Table 5 refers to the path state Z. p (k), where Z0 represents the pass state, Z1 represents the observe state, Z2 represents the degraded state, and Z3 represents the suppressed state; this state is determined by the total path residual N. p (k) is determined together with the state threshold with hysteresis.

[0080] In Table 5, "positive support is not valid" refers to K. G p (k)=0 and does not satisfy K E p (k)=1、d p (k)≤T E Early support conditions; "Enhanced confirmation not established" refers to E p (k)=0. Insufficient observable coverage means O p (k) <O min O min Coverage thresholds configured by path type and security policy; in Z0 / Z1 and O p (k)≥O min In the branch, E p (k) is not used as a separate release condition when positive support is lacking.

[0081] When the original algorithm outputs multi-level path grades, Table 6 provides grade corrections based on the authorization results in Table 5. The corrected grade is capped at the original path grade; the original grade retention in Z0 / Z1 states and early weighted outputs are both contingent on observable coverage not being less than a set value; early positive support corresponds to short-term evidence for the candidate grade, and the grade is subject to the early grade cap L. E,max Constraints; In downgraded and insufficient coverage states, the level is subject to a conservative level cap L. C,max constraint.

[0082] Table 6 Multi-level Path Hierarchy Correction Logic

[0083] 5. Algorithm Implementation and Engineering Computation This method employs fixed-point logic and is applicable to periodic tasks in airbag ECUs and vehicle restraint controllers. Each front-end local induced path maintains the strong local impact latch state, trigger time, consequence-complete latch, component residuals, total residuals, forward propagation support, and enhanced confirmation condition E. p (k) Path status and necessary health gating information. Window, weight, threshold, and recovery rate are given by a calibration table.

[0084] The pseudocode description of this method is as follows: Input: Set P of anterior local evoked pathways F Front satellite sensors: local impact intensity, persistence, and health status; ECU central sensors: vehicle response, speed accumulation, pattern continuity, and supporting evidence; consequence observability mask, path applicability marker R. j,p Early positive evidence; original path authorization and original path rank.

[0085] Output: Corrected path authorization A′ p (k) Corrected path level l′ p (k) Path propagation consequence residual N p (k) Early positive support K E p (k) Global positive support K G p (k) and path status.

[0086] for each path p in P F do if the event reset is true, then Clear path latches, trigger times, complete latches, residuals, positive support, and path status; Set path authorization and path level to zero; continue; end if Determine whether path p will experience a strong local impact on the leading satellite; If a strong local impact occurs for the first time, then latch path p and record the path trigger time; initialize and complete the latch, component residuals, and path state; end if path p is already latched, then calculate the path time d within the event. p (k); Generate stable pattern support based on the original pattern evidence; for each consequence j in {C,V,M,Q} do determine the missing observation window, completion window, and overdue stage of consequence j; update the observable mask M based on health status. j,p (k), and update the applicable flag R according to the path applicability. j,p If R j,p If M = 0, then skip the missing penalty, positive support contribution, path total residual synthesis, and coverage denominator accumulation for that consequence, and proceed to the next consequence; if M j,p If (k)=0, the corresponding component residual is maintained, no new missing, overdue, or interruption penalties are added, no completion offsetting or stability recovery is triggered, and the corresponding positive support is set to zero, and the next consequence is entered; when consequence j meets the stability condition within the completion window, the completion latch is updated; when consequence j is observable but not yet completed, the missing residual is accumulated according to the window state; when consequence j is observable but overdue and still not completed, the overdue missing residual is accumulated; when the mode consequence has been completed, is currently observable, and the mode stability support has disappeared, the mode interruption residual is accumulated; when consequence j has been completed, is currently observable, and the stability support is established, residual recovery is performed; the component residuals of consequence j are saturated and limited; end for grouping and limiting the total path residual by same-origin consequence; calculate observable coverage O p (k) Early positive support, global positive support, and enhanced confirmation conditions E p (k); Update the path status based on the total path residual and hysteresis recovery rules; correct the path authorization and path grade based on the path status, coverage priority, positive support, and enhanced acknowledgment conditions; else set the path authorization and path grade to zero; end if end for The system-level arbitration function takes the corrected local path authorization and path level output by this method as local path input, and combines them with existing independent severe collision path requests, other path requests, system severity indicators and confirmation conditions to generate a system-level protection request.

[0087] The propagation consistency judgment and authorization correction process for a single local induced path is as follows: Figure 2 As shown, the process begins with the establishment of a strong forward local impact and sequentially completes path latching, propagation consequence residual update, forward support calculation, residual state judgment, and authorization correction.

[0088] This algorithm does not require floating-point operations; residuals are represented using unsigned integers and constrained within a fixed range through saturation addition and subtraction. The consequence window is determined by the event time difference and lookup table parameters. Path states are implemented using a finite state machine. The computation for each path mainly includes Boolean judgments, window judgments, weighted summation, amplitude limiting, and saturation addition and subtraction. The propagation consequence types and the number of early evidence are both small constants; the computational cost per path in a periodic task is linearly related to the number of paths.

[0089] The advantages of the above implementation are: fixed-point integers and saturated addition and subtraction can avoid floating-point unit dependence and reduce overflow risk; lookup window facilitates calibration reuse; finite state machine facilitates state coverage testing and security analysis; small constant evidence set makes single-cycle computation predictable, suitable for worst-case execution time verification of ECU cycle tasks.

[0090] The health status is entered into the propagation consistency model through front satellite sensor health gating, consequence observability masking, and path applicability flags. When the front satellite sensor itself malfunctions, strong local impulse conditions are not allowed, and no new path latching is permitted for unlatched paths; latched paths enter a degraded branch and proceed according to E. p (k) and the upper limit of the conservative level L C,max Restrict front-end path authorization. When a consequence path becomes unobservable, no new missing, overdue, or interrupted penalties are applied to that consequence in the current period; existing component residuals are frozen or maintained according to the diagnostic strategy, and the consequence does not provide positive support. When a consequence is not applicable to path p, it is handled through R. j,p =0 removes it from the set of observables and the denominator of the coverage. The observable coverage and residual status are entered into the authorization function together to distinguish between three cases: "the consequence is observable but has not occurred", "should be observed but is not observable", and "this path does not need to be observed".

[0091] The propagation consistency model in this method is the model part of the path authorization discrimination algorithm of this invention used to update the propagation consequence residual, forward propagation support, observable coverage, path state, and authorization output.

[0092] Health gating restrictions include: when the current satellite sensor health indicator is 0, strong local shock conditions are not allowed, and no new paths can be latched for unlatched paths; latched paths enter degraded branches, in E... p When (k)=0, A′ p (k)=0, in E p When (k)=1, only the upper limit of the conservative level L is allowed. C,max Output: When the central or confirmation consequence channel is unobservable, the corresponding component residual is frozen or held, no new penalties are added for missing, overdue or interrupted items, the corresponding positive support is set to 0, and enhanced confirmation constraints are triggered by observable coverage.

[0093] 6. Mechanism Analysis and Discussion 1) State evolution in real severe collisions.

[0094] In a real severe frontal collision, the front satellite sensors initially generate a localized strong impact. Subsequently, the central vehicle response, velocity change accumulation, collision mode continuity, and confirmation support gradually emerge within their respective windows. A small number of missing residuals are formed in the early stages; as subsequent evidence is gathered, the initial compensation and stabilization support restore the corresponding component residuals. Forward propagation support is updated as evidence emerges: early spatial consistency and local persistence form early positive support within a short time window, while global positive support is formed after the central vehicle response, velocity accumulation, mode stabilization, and confirmation support stabilize.

[0095] 2) Localized hard contact and abnormal spikes in single sensors.

[0096] Localized hard contact generates strong short pulses in the front satellite sensors; due to insufficient collision energy and limited effective range, the central vehicle response and velocity accumulation are not valid. The path is latched when the local triggering condition is met; within subsequent windows, the central vehicle response, velocity accumulation, pattern continuity, and confirmation support are continuously missing, and the corresponding component residuals gradually increase. Single-sensor abnormal spikes lack persistence and subsequent vehicle propagation. When the spike does not meet the persistence threshold, strong local impact latching is not valid; when the spike meets the local triggering condition, subsequent propagation consequences are still missing, the residuals increase with the window state, and forward propagation support is unstable.

[0097] 3) Samples from later central transmission.

[0098] For real-world collision samples where there is a significant delay between the front local signal and the central propagation, the missing term in the central propagation increases the residual within the early window. When the central vehicle response, speed accumulation, and confirmation support appear within the completion window, the initial completion cancels out the corresponding component residuals. As subsequent support remains stable, the stabilization recovery term continues to reduce the residuals.

[0099] 4) Sensor degraded state.

[0100] Propagation residuals are calculated based on the absence of observable and applicable consequences. When a consequence becomes unobservable due to sensor failure, communication anomalies, or channel degradation, no new missing, overdue, or interrupted penalties are imposed on that consequence in the current period. Existing component residuals are frozen or maintained according to the diagnostic strategy, providing no positive support, and enhanced acknowledgment constraints are triggered through observable coverage. When a consequence is not applicable to the current path, the applicable flag R is used to... j,p=0 is simultaneously removed from both the numerator and denominator of the coverage. When the front satellite sensors are abnormal, path latching and authorization are restricted by the front health gating; when the central channel and confirmation channel are unobservable, the residuals of the corresponding consequence components are frozen or maintained, and are constrained by observable coverage before entering the authorization function. When multiple key consequences are unobservable, observable coverage will affect the path authorization function, causing the authorization decision to switch to the enhanced confirmation branch.

[0101] 5) The impact of window, weight, and threshold on path state discrimination.

[0102] When the central propagation window is set too short, reasonable propagation delays in real collisions are counted as overdue missing data; when the window is set too long, local anomalies remain under observation for an extended period. Missing data weights, offsetting data weights, and recovery rates can be calibrated using real-vehicle collision data as well as data from non-deployment / abuse scenarios. The threshold for early positive support also needs calibration: if the early positive support threshold is set too low, non-deployment local disturbances may meet the early support conditions; if the threshold is set too high, local evidence from the early stages of real collisions is difficult to participate in authorization determination, and path authorization will rely more on global propagation evidence.

[0103] 6) There is a correlation between different consequences.

[0104] When the central vehicle response, velocity accumulation, and confirmation support originate from the same central inertial channel, component residuals and grouped amplitude limiting structures are used to control the cumulative contribution of related consequences; positive support terms with highly correlated sources are amplitude-limited according to the same source evidence group. For paths where the central longitudinal response may be delayed or have weak amplitude, such as offset collisions and small overlap collisions, front multi-point consistency, lateral consistency, mode continuity, and independent confirmation links can be selected as supplementary evidence according to path type, and calibrated through corresponding windows and weights.

[0105] In one specific implementation, the above method is integrated in parallel with the system-level arbitration as a front-end local induced path correction module.

[0106] The aforementioned method studies the authorization determination problem of candidate paths induced by strong local impacts from frontal satellite sensors in vehicle frontal collision algorithms, and constructs a path-level correction method based on propagation consequence residuals and forward propagation support. This method describes the process from local triggering to authorization of a candidate path as a temporal consistency judgment between early local evidence and the overall vehicle propagation consequence. The strong frontal local impact serves as the path time origin, while the central vehicle response, velocity accumulation, collision mode continuity, and confirmation support are considered as the expected propagation consequences.

[0107] In this method, when the expected consequence is observable within a reasonable window but has not yet occurred, the consequence residual is propagated cumulatively according to the path weight. When subsequent evidence is supplemented, the residual is updated through initial supplementation offsetting and stabilizing support recovery. Path authorization is jointly determined by residual status, early positive support, global positive support, observable coverage, and enhanced confirmation conditions.

[0108] Forward propagation support is divided into early forward support and global forward support. Early forward support is based on multi-point consistency of front satellite sensors, local impact persistence, and local pattern stability, and is used to describe short-term local evidence. Global forward support is based on central vehicle response, velocity accumulation, pattern continuity, and confirmation support, and is used to describe the state of evidence after the local impact propagates to the vehicle level.

[0109] The path state variables in this method include consequence completion latch, component residuals, total path residuals, observable coverage, and path state. Residual update terms include in-window missing, overdue missing, mode interruption, first completion cancellation, and stable recovery. Authorization constraints include path applicability flags, consequence observability masks, same-source consequence limiting, positive support same-source limiting, and hysteresis recovery rules. The above state evolution illustrates that, after the propagation window, weights, and positive support thresholds are calibrated, this framework can characterize path state changes under local hard contact, single-sensor anomalous spikes, late-propagating samples in the center, and sensor degradation states. Real-vehicle collision data and non-deployment condition data are used to determine the propagation window, missing weights, cancellation weights, positive support thresholds, and path state gating thresholds. Target ECU periodic testing is used to check worst-case execution time and resource consumption. The diagnostic degradation strategy can be validated through use cases such as sensor failure, communication anomalies, and unobservable channels.

[0110] In an optional calibration embodiment implemented with a 1ms algorithm cycle, the propagation window can be set according to the consequence type: central vehicle response C approximately 3-25ms, cumulative severity V approximately 8-40ms, path pattern continuity M approximately 3-20ms, and authorization confirmation support Q approximately 5-30ms; the upper limit of the window can be extended by approximately 5-30ms after the corresponding observation window to cover samples with later central propagation but still conforming to the real collision mechanism.

[0111] Weights can be quantified using dimensionless integers: 1-4 for missing values ​​within the window, 2-8 for overdue missing values, 1-6 for mode interruption, 2-12 for initial completion and offset, and 1-4 per cycle for stable recovery rate; Homologous group amplitude limiting C g It can be set to 1-2 times the upper limit of a single item within the group to avoid duplicate penalties for links from the same source. Group weight λ g 1-4 or equivalent fixed-point values ​​can be selected and calibrated in conjunction with the thresholds for pass, observation, degradation, and inhibition states.

[0112] The positive support threshold can be an integer score threshold: early positive support threshold θ E A score of 2-4 is acceptable, with a global positive support threshold θ. G A score of 3-6 is acceptable; however, the above range can be adjusted proportionally based on real vehicle collision samples, non-deployment condition samples, and target false triggering constraints, depending on the vehicle model, sensor layout, or ECU cycle.

[0113] If the above methods are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause an electronic device or vehicle controller to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0114] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present invention can be implemented using programming languages, model code, or embedded software instructions suitable for a target controller.

[0115] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0116] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0117] The preferred embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make numerous modifications and variations based on the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of existing technology should be within the scope of protection defined by the claims.

Claims

1. A path authorization discrimination method based on propagation consequence residuals and forward propagation support, characterized in that, Includes the following steps: After the front collision algorithm is activated, a candidate path is obtained that is induced by a local impact at the front. The local impact corresponding to the candidate path meets the set requirements and the corresponding front satellite sensor is in a healthy state. The path time within the event is determined based on the path triggering time. Subsequent propagation consequence information is obtained based on the path time within the event. Based on a pre-set propagation consequence set, the component residuals of each expected consequence in the propagation consequence set are updated according to the subsequent propagation consequence information, and the total path residual of the path is calculated. The effective time of early support is determined based on the path triggering time. Early positive support is generated and used within the effective time of early support, and global positive support is generated. After the effective period of the early support has expired, the global positive support will be used as the positive support basis for path authorization determination; Obtain enhanced confirmation conditions, and determine the authorization variables and path level variables of the candidate path based on the total path residual, early positive support, global positive support, and enhanced confirmation conditions.

2. The path authorization determination method based on propagation consequence residuals and forward propagation support according to claim 1, characterized in that, The localized impact meeting the set requirements includes: The local impact intensity reaches the path threshold and continues for a set duration.

3. The path authorization determination method based on propagation consequence residuals and forward propagation support according to claim 1, characterized in that, The information regarding the consequences of subsequent propagation includes information on central response, severity, pattern continuity, and confirmation / support.

4. The path authorization determination method based on propagation consequence residuals and forward propagation support according to claim 1, characterized in that, The anticipated consequences of the propagation consequences include central vehicle response, cumulative severity, path pattern continuity, and authorization confirmation support.

5. The path authorization determination method based on propagation consequence residuals and forward propagation support according to claim 1, characterized in that, The component residuals for each expected consequence are updated using the following formula: N j,p (k) = sat[N j,p (k-1)+ΔN miss,j,p (k)+ΔN late,j,p (k)+ΔN brk,j,p (k)-ΔN fill,j,p (k)-ΔN rec,j,p (k)] Where j represents the consequence type, N j,p (k) represents the component residual corresponding to consequence j on path p; ΔN miss,j,p (k), ΔN late,j,p (k) and ΔN brk,j,p (k) represents the weighted contribution of in-window missing, overdue missing, and mode interruption, respectively; ΔN fill,j,p (k) and ΔN rec,j,p (k) represents the initial compensation offset and the stable recovery contribution, respectively; sat[·] represents the saturation limit.

6. The path authorization determination method based on propagation consequence residuals and forward propagation support according to claim 1, characterized in that, When calculating the total residual of the path, the expected consequences are grouped according to their sources, the component residuals of each expected consequence within a group are limited, and the residuals of different groups are weighted and synthesized into the total residual of the path. The residual weights of different groups are determined based on the independence of evidence sources, propagation reliability, and calibration of real vehicle collision / non-deployment condition data. The contribution within the same source group is constrained by the group limit.

7. The path authorization discrimination method based on propagation consequence residuals and forward propagation support according to claim 1, characterized in that, The values ​​of early positive support and global positive support are both 0 or 1; wherein, early positive support is formed based on the multi-point consistency of front satellite sensors, local persistence and local pattern stability within the early window, and global positive support is formed based on the central vehicle response, severity accumulation, pattern continuity and confirmation support; the corresponding evidence group is set to 1 after meeting the threshold and passing the same source amplitude limiting, otherwise it is set to 0.

8. The path authorization determination method based on propagation consequence residuals and forward propagation support according to claim 1, characterized in that, The authorization variables and path ranking variables used to determine candidate paths specifically include: The residual state is determined based on the total path residual and the state threshold. The residual state includes the pass state, the observe state, the degraded state, and the suppressed state. The corrected authorization variable is obtained based on the path latch state, residual state, observable coverage, positive support state, and enhanced confirmation conditions. : When the path is not latched or the residual state is in a suppressed state =0; When the residual state is either passable or observable, the observable coverage is not less than the set value, and K G p When (k)=1, = A p 0 (k), K G p (k) represents global positive support, A p 0 (k) represents the original authorized variable; When the residual state is either passable or observable, the observable coverage is not less than the set value, and K G p (k)=0, K E p (k)=1、d p (k)≤T E hour, = A p 0 (k), K E p (k) indicates early positive support, d p (k) represents the path time within the event; When the residual status is in a degraded state, or the residual status is in a pass or observation state and the observable coverage is less than a set value, = A p 0 (k)∧E p (k), E p (k) indicates an enhanced confirmation condition, and ∧ indicates a logical AND; When the residual state is either passable or observable, the observable coverage is not less than the set value, and K G p (k)=0, and does not satisfy K E p (k)=1、d p (k)≤T E When the early support conditions are met, =0; The revised path-level variables are obtained based on the revised authorization variables, residual status, observable coverage, positive support status, and enhanced validation conditions. : when When =0, =0; When the residual state is either passable or observable, the observable coverage is not less than the set value, and K G p When (k)=1, = l p 0 (k), l p 0 (k) represents the original path ranking variable; When the residual state is either passable or observable, the observable coverage is not less than the set value, and K G p (k)=0, K E p (k)=1、d p (k)≤T E hour, =min(l p 0 (k), L E,max ), L E,max Indicates the initial level cap; When the residual state is in a degraded state, or the residual state is in a pass state or an observation state and the observable coverage is less than the set value and E p When (k)=1, =min(l p 0 (k), L C,max ), L C,max Indicates the upper limit of the conservative level; The observable coverage is the ratio between the sum of the weights of the currently observable and applicable key consequences of the path and the sum of the weights of the key consequences that should be observed and applicable to the path.

9. The path authorization determination method based on propagation consequence residuals and forward propagation support according to claim 8, characterized in that, After a candidate path enters the suppression state, the recovery conditions include the total path residual dropping below the recovery threshold, and the global positive support and enhancement confirmation conditions being met continuously for a set period.

10. A computer-readable storage medium, characterized in that, Includes one or more programs executable by one or more processors of an electronic device, said one or more programs including instructions for performing the path authorization discrimination method based on propagation consequence residuals and forward propagation support as described in any one of claims 1-9.

Images