Restraint deployment control method with feed-forward adjustment of deployment threshold

[0002] This invention relates to automotive passenger restraint systems, and more particularly to a control method that differentiates deployment events from non-deployment events.

[0003] In general, automotive passenger restraint systems perform a number of functions including acceleration sensing, signal processing and analysis, and deployment of one or more restraint devices such as frontal or side air bags and seat belt pretensioners in response to a sensed crash event. Typically, the acceleration signal is monitored to detect a potential crash event, and then filtered or integrated over the course of the crash event to produce a velocity change or ΔV signal. If the ΔV signal exceeds a threshold, the crash event is determined to be sufficiently severe to warrant deployment of restraints. The threshold is typically time-dependent, and is calibrated based on data logged for different types of crash events, as well as data logged during rough road driving.

[0004] A problem with the above-described approach is that it is often difficult to synchronize the time progression of the crash (that is, the event clock or timer) with the actual crash event. As a result, it can be difficult to distinguish between deployment events and non-deployment events, particularly in the first portion of the sensed event.

[0005] The present invention is directed to an improved deployment control method for a vehicular supplemental restraint system having an acceleration sensor and a restraint device to be deployed for occupant protection in a crash event, where the deployment threshold is adaptively adjusted based on the magnitude and rate of change of the ΔV signal and the progression level of the sensed event. Since the deployment threshold is adjusted as a function of the input (ΔV) signal, the control is characterized herein as a feed-forward control.

[0006] According to the invention, the acceleration signal is filtered (integrated) to form the ΔV signal, and the deployment threshold is periodically adjusted based on the rate of change (slope) of the ΔV signal, if the magnitude of the ΔV signal is within a ΔV range specified for the progression level of the sensed event. Preferably, the maximum amount of adjustment is also specified based on the progression level of the event. As a result, the deployment threshold tends to follow or track the ΔV signal, particularly for non-deployment events where the magnitude of the ΔV signal is within the specified ΔV range, thereby preventing deployment of the restraints. In deployment events, the magnitude of the ΔV signal is more likely to fall outside the specified ΔV range, which minimizes feed-forward adjustment of the deployment threshold, thereby minimizing any delay in deployment of the restraints.

[0007] FIG. 1 is a schematic diagram of a supplemental restraint system including a programmed microprocessor for carrying out the deployment method of this invention.

[0008] FIG. 2 is a graphical representation of event progression determination and threshold modification according to this invention.

[0009] FIG. 3 graphically depicts the operation of the control method of this invention for a non-deployment event.

[0010] FIG. 4 graphically depicts the operation of the control method of this invention for a deployment event.

[0011] FIGS. 5-6 are flow diagrams representative of computer program instructions executed by the microprocessor of FIG. 1 in carrying out the deployment method of this invention. FIG. 5 is a main flow diagram, and FIG. 6 details a step of the main flow diagram relating to determination of a feed-forward adjustment of the deployment threshold.

[0012] FIG. 1 generally depicts a vehicle supplemental restraint system in which one or more restraints, such as air bags AB, are deployed in a severe crash event to protect the vehicle occupants. The restraints may include without limitation, air bags, belt pretensioners, inflatable tubular structures, side curtains, anti-whiplash devices, etc., and it will be understood that such term AB as used herein does not refer exclusively to a particular type of restraint. A deployment control system, designated generally by the reference numeral 10, may be packaged as a single electronic module and mounted on a frame element in a central portion of the vehicle. Fundamentally, the system 10 includes a longitudinal acceleration sensor LAS (which may comprise a single sensor or a pair of sensors mounted at an offset angle) for sensing longitudinal acceleration of the vehicle, a microprocessor μP for receiving the output signal of longitudinal acceleration sensor LAS, and a firing circuit FC which is triggered by microprocessor μP to deploy the air bags AB in the event of a severe crash. In general, the microprocessor μP filters the longitudinal acceleration signal over a predefined interval, or window, to form a windowed velocity signal, referred to herein as ΔV_WIN. The microprocessor μP then adjusts a deployment threshold, referred to herein as a ΔV Threshold, based on event progression and the magnitude and rate of change of ΔV_WIN, and compares ΔV_WINto the ΔV Threshold. If ΔV_WINcrosses the ΔV Threshold, the microprocessor μP signals the firing circuit FC to deploy the air bags AB.

[0013] According to the present invention, the ΔV Threshold is set at a default level prior to initiation of a crash event and is periodically adjusted during the crash event based on the rate of change (slope) of ΔV_WIN, if the magnitude of ΔV_WINis within a ΔV range specified for the progression level of the event. Additionally, the maximum amount of adjustment is also specified based on the progression level of the event. Thus, for any given progression level of the event, the control of this invention specifies a ΔV range within which ΔV_WINmust fall in order to enable threshold adjustment, and a limit on the amount of the adjustment. Since the slope of ΔV_WINmay be either positive or negative, the threshold may be adjusted either upward or downward, depending on the particular crash signal. Alternatively, threshold adjustment may be limited to positive slope values, in which case only upward adjustment of the threshold can occur. The upper limit of the ΔV range is selected so that, in general, non-deployment events are more subject to adjustment than deployment events, especially in the early progression of the event. This minimizes any deployment delay for deployment events, while permitting significant adjustment of the threshold when ΔV_WINis within the ΔV range, thereby allowing the deployment threshold to follow or track ΔV_WINduring non-deployment events.

[0014] The windowed velocity signal ΔV_WINmay be calculated according to the expression:

[0015] ΔV_WIN(n)=Σ[ACCEL(n−i)], for i=0 to (w−1)  (1)

[0016] where ACCEL is a filtered version of the output of acceleration sensor LAS and w is the window size. In a digital implementation, the window w actually refers to a specified number of successive samples of the acceleration signal. Since the samples are read at a predefined rate, however, the window w also may be viewed as a corresponding time interval.

[0017] The slope or rate of change of ΔV_WIN, referred to herein as ΔV_slope, may be determined according to the expression:

[0018] ΔV_slope(n)=ΔV_WIN(n)−ΔV_WIN(n−1)  (2)

[0019] where (n) refers to a current value, and (n−1) refers to a value determined in the previous calculation.

[0020] In the preferred embodiment, the level of event progression is determined by computing a velocity signal ΔV_biasthat is biased toward zero. The biased velocity signal ΔV_biasmay be determined by computing a filtered acceleration signal ΔV, and then applying a bias “B”. For example, ΔV may be calculated according to the expression:

[0021] ΔV(n)=ΔV(n−1)+ACCEL(n)−ΔV(n−1)/C  (3)

[0022] where C is a constant, with ΔV_biasbeing defined as:

[0023] ΔV_bias=ΔV−B  (4)

[0024] and the bias B being defined as: $\begin{array}{cc}B=\begin{array}{cc}d& \mathrm{if}\Delta V>d,\mathrm{with}d\mathrm{being}a\mathrm{positive}\mathrm{integer}\\ \Delta V& \mathrm{if}\left|\Delta V\right|\le d\\ -d& \mathrm{if}\Delta V<-d\end{array}& \left(5\right)\end{array}$

[0025] Alternatively, the level of event progression can be determined based on windowed velocity ΔV_WINor the time that ΔV_WINis above a specified threshold. For purposes of this description, however, it will be assumed that ΔV_biasis used to determine the level of event progression. To this end, ΔV_biasis compared to a series of predefined velocity values, referred to herein as progression level thresholds a-d, thereby defining four corresponding stages or levels of event progression; obviously, the number of thresholds, and hence progression levels, may vary from one implementation to another.

[0026] The above-described method is graphically depicted in FIG. 2, where Graphs A and B show exemplary values of ΔV_biasand ΔV_WIN, respectively, on a common time scale. The time designations t₁-t₄signify times that coincide with ΔV_biascrossing one of the thresholds a-d, and the event progression level at any given time is indicated at the top of Graph A, and below the time axis of Graph B. For example, progression level of the sensed event is “a” in the time interval t₀-t₁, “b” in the time interval t₁-t₂, “c” in the time interval t₂-t₃, “d” in the time interval t₃-t₄, and so on. The progression level “a” is indicative of no or very low activity. Each of the progression levels a-d are associated with one or more predefined ΔV_WINranges as shown by the vertical columns in Graph B, with an adjustment limit specified for each ΔV_WINrange. Within these limits, the deployment threshold is adjusted based on the value of ΔV_slope. As an example, if the sensed event is in progression level “b”, adjustment of the threshold might be limited to 10 counts if ΔV_WINis within range 1, 20 counts if ΔV_WINis within range 2, and 30 counts if ΔV_WINis within range 3, it being understood that the number of counts is arbitrarily related to the actual units of ΔV_WIN.

[0027] The operation of the above-described feed-forward threshold adjustment technique can be illustrated in reference to FIGS. 3 and 4 which graphically depict the deployment threshold (THR) and ΔV_WINas a function of time for representative non-deployment and deployment events, respectively. In the non-deployment event of FIG. 3, which may be a 9 MPH frontal collision for example, ΔV_WINstays within the upper ΔV limit, and the deployment threshold THR is adjusted upward with ΔV_slope(within limits). As a result, the threshold THR tends to track ΔV_WIN, and deployment never occurs because ΔV_WINnever exceeds THR. On the other hand, in the deployment event of FIG. 4, which may be a 30 MPH frontal collision, adjustment of the threshold THR is somewhat limited, due to one or more of three factors: (1) ΔV_WINis higher than the specified upper ΔV limit (in which case, no adjustment occurs), (2) ΔV_slopeis higher than the specified slope limit, and (3) the adaptation amount exceeds an adaptation limit. As a result, ΔV_WINquickly exceeds the threshold THR, resulting in deployment of the restraints at time t₁. Since the initial upward adjustment of the threshold THR is much less than ΔV_slope, the deployment is only slightly delayed, as compared to the time of deployment if there were no adjustment of the threshold THR. It will also be noted in FIG. 4 that the threshold THR is only adjusted upward, even when ΔV_slopeis negative; this is optional as explained above, and in any event, is irrelevant once deployment occurs.

[0028] The examples described above in reference to FIGS. 3 and 4 illustrate how the feed-forward threshold adjustment of this invention provides immunity to non-deployment events while minimizing any delay in deployment in a deployment event. By contrast, a fixed threshold set high enough to avoid deployment in the collision of FIG. 3, and would significantly delay deployment in the collision of FIG. 4.

[0029] A flow diagram representative of computer program instructions executed by the microprocessor μP of FIG. 1 in carrying out the above-described deployment method is set forth in FIGS. 5-6. FIG. 5 depicts a main loop flow diagram, where the block 100 designates a series of instructions executed at the initiation of vehicle operation for initializing various registers, counters, flags and variables to a predefined state. For example, the ΔV Threshold is initialized to a default value at this point. Thereafter, the block 102 is executed to read the output signal of the longitudinal acceleration sensor LAS, and to filter it to form a filtered acceleration signal ACCEL. The various severity measurements are then computed at block 104; these include, for example, ΔV, ΔV_bias, ΔV_WIN, ΔV_slope, etc. Block 106 is then executed to determine a net threshold adaptation amount T_A, as described in detail below in reference to FIG. 6. If the magnitude of the determined adaptation amount T_Ais greater than a limit value, as determined at block 108, the block 110 is executed to suitably limit adaptation amount T_A. Then, the adaptation amount T_A(which can be either positive or negative) is added to the ΔV Threshold at block 112. If the windowed velocity ΔV_WINexceeds the newly adjusted threshold, as determined at block 114, the block 116 is executed to deploy the restraints AB.

[0030] The flow diagram of FIG. 6 sets forth the main flow diagram step of determining the net threshold adaptation amount T_A(block 106) in further detail. In the illustrated embodiment, the various event progression thresholds a-d, the ΔV ranges, and the associated limits are stored in an adaptation matrix within microprocessor μP, and a series of progression level masks are used to identify corresponding ranges and adjustment limits. Blocks 120-132 comprise a loop for determining the appropriate ΔV range and limit, taking into account the progression level and the magnitude of ΔV_WIN. Thus, for each progression level mask L, the microprocessor μP executes the blocks 122-130 within the progression level loop boundary blocks 120 and 132. At block 122, the current mask L is applied to the matrix, and the microprocessor μP determines if the biased velocity ΔV_biasis within the corresponding progression level thresholds. For example, when the mask for progression level b is selected, block 122 determines if ΔV_biasis between the values a and b, as defined in FIG. 2. If not, the mask L for the next progression level is applied to the matrix, as indicated at block 132. If ΔV_biasis within the corresponding progression level thresholds, block 124 determines if ΔV_WINis within a specified ΔV range corresponding to the progression level of the mask L. If not, no adjustment is made; if so, blocks 126-128 are executed to limit ΔV_slopeto a limit value specified by the mask, and block 130 increases the net threshold adjustment T_Aby ΔV_slope. As indicated above, the threshold adjustment may be limited to positive values of ΔV_slope, if desired; in some cases, this has been found to improve immunity to non-deployment events, such as operating the vehicle on a rough road.

[0031] After the net threshold adaptation amount T_Ais determined for each progression level mask L, the blocks 134-138 are executed to bias the ΔV Threshold toward its default value (initialization threshold) if the event progression level is “a”—i.e., no activity. Block 134 determines if the event progression is at level “a”. If so, block 136 compares the ΔV Threshold to the Initialization Threshold. If the ΔV Threshold has been adjusted to a value less than the Initialization Threshold, block 138 sets the net threshold adaptation amount T_Ato a positive incremental value, referred to in FIG. 6 as+Threshold Recover. Conversely, if the ΔV Threshold has been adjusted to a value greater than the Initialization Threshold, block 138 sets the net threshold adaptation amount T_Ato a negative incremental value, referred to in FIG. 6 as −Threshold Recover.

[0032] In summary, the deployment method of this invention provides a novel, feed-forward control in which the deployment threshold is periodically adjusted based on ΔV_slope, provided that the magnitude of ΔV_WINis within the specified ΔV range. As a result, the deployment threshold tends to follow or track ΔV_WINduring non-deployment events, thereby allowing the threshold to have an initial or default value that is low enough to provide quick deployment of the restraints during deployment events. While described in reference to the illustrated embodiment, it is expected that various modifications in addition to those suggested above will occur to those skilled in the art. In this regard, it will be understood that this invention is not limited to the illustrated embodiment, and that control methods incorporating such modifications may fall within the scope of this invention, which is defined by the appended claims.