Air bag controller online test and software burn-in integrated tool and method

Abstract

The invention relates to the technical field of air bag controller testing, in particular to an air bag controller online testing and software burn-in integrated tool and method.The air bag controller online testing and software burn-in integrated tool comprises an equipment body, a fixing frame is installed on the side wall of one side of the equipment body, a supporting plate is installed at the top end of the side wall of one side of the fixing frame, and a hydraulic cylinder is installed at the bottom end of the supporting plate; a hydraulic cylinder is installed at the top end of the equipment body, an upper needle bed assembly is installed at the bottom end of the hydraulic cylinder, a workbench is installed at the top end of the equipment body, a lower needle bed tool is installed at the top end of the workbench, and a safety air bag controller circuit board is arranged at the top end of the lower needle bed tool. According to the invention, firstly, the equipment is electrified, electrical testing and burn-in are carried out through sequential control of the rear load board, and self-inspection is completed after testing is synchronously carried out, so that one-stop testing and software burn-in can be realized, function testing and software burn-in can be completed at the same station, the carrying and repeated positioning time of products among different equipment is reduced, and the production period is shortened.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of airbag controller testing, in particular to an airbag controller online testing and software burning integrated tool and method. BACKGROUND

[0002] The airbag controller is the core control module of the automobile airbag system, which is composed of a collision sensor, a controller and a gas generator. The airbag controller captures the collision signal through the acceleration sensor and the pressure sensor, comprehensively judges the collision strength, direction and vehicle speed, decides whether to trigger the airbag ignition circuit, and sends the ignition instruction to the airbag assembly and the seat belt tensioning device. The system also has the functions of collision data recording, fuel cut-off, unlocking and other safety measures triggering. The airbag controller needs to be tested online during processing.

[0003] In view of the prior art, the following problems exist when the airbag controller is tested online: custom needle bed tool needs to be designed for different PCBs, the design and manufacturing cycle is long; with the change of PCB design, the tool needs to be redesigned or adjusted, increasing the long-term maintenance cost; in the production quality monitoring of the airbag controller, the existing technology mainly relies on the direct detection results output by the artificial intelligence test system. These results can only reflect the apparent qualified rate of the product, and cannot reveal the cognitive ability decline of the artificial intelligence system after the firmware upgrade or the change of the running environment; this state makes it difficult for managers to identify the health status of the monitoring system, affecting the early warning and intervention of potential batch quality risks; in addition, the existing method lacks means for quantitative evaluation of the reliability of the artificial intelligence system, and cannot provide data-driven decision basis for adjusting the production strategy.

[0004] The above status and deficiencies are mainly due to the limitations of monitoring and decision-making logic. The evaluation of system risks only stays at the first-order data analysis level of product quality, and fails to perform second-order analysis on the statistical characteristics of the output data of the monitoring system, resulting in that the cognitive blind area cannot be found; as a result, when the artificial intelligence system fails, managers cannot obtain clear signals that the system itself has become unreliable, delaying the implementation of higher-level verification or stop-line troubleshooting and other risk avoidance measures. SUMMARY

[0005] This section aims to summarize some aspects of the embodiments of the present application and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the abstract and title of the specification to avoid obscuring the purpose of this section, abstract and title. Such simplifications or omissions cannot be used to limit the scope of the present application.

[0006] In view of the above or prior art, when the air bag controller is currently tested online, the customized needle bed tool needs to be designed for different PCB designs, and the design and manufacturing cycle is long; with the change of PCB design, the tool needs to be redesigned or adjusted, which increases the long-term maintenance cost, so an air bag controller online test and software burning integrated tool is needed to meet the use requirements of people.

[0007] Therefore, to solve the above technical problems, the present application provides an air bag controller online test and software burning integrated tool, and the specific technical solutions are as follows:

[0008] An air bag controller online test and software burning integrated tool, comprising a device main body, a fixing frame is installed on the side wall of one side of the device main body, a support plate is installed on the top end of the side wall of one side of the fixing frame, a hydraulic cylinder is installed on the bottom end of the support plate, an upper needle bed assembly is installed on the bottom end of the hydraulic cylinder, a workbench is installed on the top end of the device main body, a lower needle bed tool is installed on the top end of the workbench, a safety air bag controller circuit board is arranged on the top end of the lower needle bed tool, a rear load plate is installed in the device main body, and a conveying assembly is installed on the top end of the workbench.

[0009] As a preferred scheme of the air bag controller online test and software burning integrated tool, the top end of the lower needle bed tool is provided with a lower needle foot test assembly, and the top end of the lower needle foot test assembly is in contact with the bottom end of the safety air bag controller circuit board.

[0010] Preferably, the upper needle bed assembly comprises a connecting plate, the connecting plate is fixedly installed on the bottom end of the hydraulic cylinder, a connecting rod is installed on the bottom end of the connecting plate, an upper needle bed tool is installed on the bottom end of the connecting rod, an upper needle foot test assembly is installed on the bottom end of the upper needle bed tool, and the bottom end of the upper needle foot test assembly is in contact with the top end of the safety air bag controller circuit board.

[0011] Preferably, the conveying assembly comprises a welding plate, the welding plate is fixedly installed on the top end of the workbench, a conveying roller is rotatably installed in the welding plate through a bearing, a conveying belt body is sleeved on the outer wall of the conveying roller, a driving assembly is installed on the side wall of the welding plate, and the output end of the driving assembly is fixedly connected with one end of the conveying roller.

[0012] Preferably, the conveying roller has two groups of the same structure, and the two groups of conveying rollers are connected through a chain wheel and chain.

[0013] Preferably, the height of the conveying belt body is higher than the height of the lower needle foot test assembly.

[0014] Preferably, an air bag controller online test and software burning decision method, the specific steps include:

[0015] Step one, obtaining multi-dimensional test data of the airbag controller online test and software burn-in integrated tooling, and constructing a systematic degradation index based on the multi-dimensional test data; collecting current statistical characteristics of the systematic degradation index and preset historical statistical characteristics, and calculating a cognitive uncertainty index;

[0016] Step two, comparing the cognitive uncertainty index with a preset active detection trigger threshold to determine an active detection rate, and performing controller unit shunting according to the active detection rate; wherein, when the cognitive uncertainty index is greater than the active detection trigger threshold, the active detection rate is calculated based on the cognitive uncertainty index; when the cognitive uncertainty index is less than or equal to the active detection trigger threshold, the active detection rate is zero;

[0017] Step three, obtaining detection results obtained by the shunted controller unit in an offline test tool, and detection results of the same controller unit by the tool; determining a cognitive bias event based on the detection results obtained by the offline test tool and the detection results of the tool; and dynamically calibrating model parameters used to calculate the cognitive uncertainty index according to the cognitive bias event.

[0018] Preferably, the current statistical characteristics include the standard deviation of the systematic degradation index output sequence in the current evaluation period; the current statistical characteristics also include the current probability distribution of the systematic degradation index output sequence; the historical statistical characteristics include the mean of the standard deviation of the systematic degradation index sequence in the historical normal operation data; and the historical statistical characteristics also include the historical normal data probability distribution.

[0019] The step of calculating the cognitive uncertainty index includes:

[0020] calculating volatility bias, wherein the volatility bias is obtained based on the standard deviation of the systematic degradation index output sequence in the current evaluation period and the mean of the standard deviation of the systematic degradation index sequence in the historical normal operation data;

[0021] calculating distribution change, wherein the distribution change is obtained based on the KL divergence between the current probability distribution of the systematic degradation index output sequence and the historical normal data probability distribution;

[0022] weighting and summing the volatility bias and the distribution change to generate the cognitive uncertainty index.

[0023] Preferably, the step of calculating the active detection rate involves using a normalized exponential scaling function, and inputting the cognitive uncertainty index, the active detection trigger threshold, a preset minimum detection rate, a preset maximum detection rate, and a preset detection rate adjustment factor, before outputting the active detection rate.

[0024] Preferably, the step of determining the active detection rate is replaced by:

[0025] Based on the aforementioned cognitive uncertainty index, a potential risk cost is constructed;

[0026] Constructing detection costs related to active detection rate;

[0027] By combining the potential risk costs and the detection costs, a total expected cost function is established;

[0028] The optimal active detection rate is calculated by minimizing the total expected cost function, and the optimal active detection rate is used as the active detection rate.

[0029] The potential risk cost is determined based on the probability of blindness determined by the cognitive uncertainty index, combined with the preset inherent probability of defects and the quantified loss of a single catastrophic failure; the detection cost is determined based on the active detection rate, combined with the number of controllers per batch and the unit output value of a single controller.

[0030] Preferably, the step of determining the cognitive bias event includes:

[0031] When the detection result obtained by the offline testing fixture is a defect and the detection result of the fixture is qualified, a cognitive bias event is determined to have occurred.

[0032] When the detection result obtained by the offline testing fixture is qualified and the detection result of the fixture is defective, a cognitive bias event is determined to have occurred.

[0033] The steps for dynamically calibrating the model parameters include:

[0034] A Bayesian update logic is adopted, based on the prior probability of the AI-induced blindness state, and according to the conditional probability of the detection result mismatch occurring in the AI-induced blindness state and the conditional probability of the detection result mismatch occurring in the normal AI state, the posterior probability representing the AI-induced blindness state is updated.

[0035] The model parameters for dynamic calibration include weighting coefficients for calculating the cognitive uncertainty index and the probability of blindness state for calculating potential risk costs.

[0036] The integrated tooling for online testing and software burning of airbag controllers of the present invention has the following advantages:

[0037] 1. This invention uses a drive assembly to simultaneously rotate the conveyor rollers, which in turn rotate the conveyor belt body. This allows the conveyor belt body to transport the airbag controller circuit board. When the airbag controller circuit board reaches the top of the lower needle bed fixture, a hydraulic cylinder is activated to press down the upper needle bed fixture and the upper needle pin test assembly. This ensures that the test points of the airbag controller circuit board are in close contact with the lower needle pin test assembly at the top of the lower needle bed fixture. Simultaneously, the equipment is powered on, and the load board at the rear controls the electrical testing and software programming in sequence. After the tests are performed synchronously, a self-test is completed. This allows for one-stop testing and software programming, completing functional testing and software programming at the same workstation. This reduces the time spent handling and repositioning products between different devices, shortening the production cycle.

[0038] 2. This invention significantly improves upon the limitations of traditional production monitoring, which relies solely on the direct output of artificial intelligence systems, by constructing a systematic degradation index, calculating a cognitive uncertainty index, implementing controller unit shunting based on the active detection rate, and dynamically calibrating model parameters based on cognitive bias events. Compared to traditional technologies, this method not only quantitatively assesses the cognitive health of the monitoring system itself but also provides more accurate and timely early warnings of potential batch quality risks through dynamic active detection and closed-loop calibration logic. These improvements not only enhance the depth and foresight of quality monitoring but also strengthen the defense capabilities against large-scale quality risks and the level of automation in decision-making.

[0039] 3. This invention calculates the cognitive uncertainty index by introducing analysis of volatility deviations and distribution changes, significantly improving the sensitivity and accuracy of risk identification. The system can dynamically adjust the active detection rate based on the quantified cognitive uncertainty index, promptly identifying and verifying potential cognitive blind spots. Furthermore, by identifying cognitive bias events and using Bayesian update logic for dynamic calibration, the risk model can self-correct and learn based on real offline detection results. These improvements not only enhance the system's risk response capabilities but also endow it with cognitive resilience, strengthening its reliability for long-term operation in complex production environments. Attached Figure Description

[0040] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein:

[0041] Figure 1 This is a schematic diagram of the structure of the present invention;

[0042] Figure 2 This is a schematic diagram of the main structure of the present invention;

[0043] Figure 3 This is a schematic diagram of the conveying component structure of the present invention;

[0044] Figure 4 This is a schematic diagram of the needle point structure of the lower needle bed tooling of the present invention;

[0045] Figure 5 This is a flowchart of the method of the present invention.

[0046] The attached diagram lists the components represented by each number as follows:

[0047] 1. Main body of the equipment; 2. Fixture; 201. Support plate; 202. Hydraulic cylinder; 3. Workbench; 4. Lower needle bed fixture; 401. Lower needle test assembly; 5. Safety airbag controller circuit board; 6. Upper needle bed assembly; 601. Connecting plate; 602. Connecting rod; 603. Upper needle bed fixture; 604. Upper needle test assembly; 7. Rear load plate; 8. Conveying assembly; 801. Welding plate; 802. Conveying roller; 803. Conveyor belt body; 804. Drive assembly. Detailed Implementation

[0048] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0049] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0050] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0051] Example 1:

[0052] Reference Figures 1-4This is the first embodiment of the present invention, which provides an integrated fixture for online testing and software burning of airbag controllers. It aims to solve the following problems currently existing in online testing of airbag controllers: high fixture development costs (customized needle bed fixtures require dedicated fixtures for different PCB designs, resulting in long design and manufacturing cycles); and high maintenance costs (with changes in PCB design, the fixture needs to be redesigned or adjusted, increasing long-term maintenance costs, including...).

[0053] The equipment body 1 has a fixed frame 2 installed on one side wall, and a support plate 201 installed at the top of the side wall of the fixed frame 2. A hydraulic cylinder 202 is installed at the bottom of the support plate 201, and an upper needle bed assembly 6 is installed at the bottom of the hydraulic cylinder 202. A workbench 3 is installed at the top of the equipment body 1, and a lower needle bed fixture 4 is installed at the top of the workbench 3. A safety airbag controller circuit board 5 is installed at the top of the lower needle bed fixture 4. A rear load plate 7 is installed inside the equipment body 1, and a conveying assembly 8 is installed at the top of the workbench 3.

[0054] Combination Figure 1 , Figure 2 and Figure 4 In an embodiment of the present invention, a lower needle test component 401 is installed on the top of the lower needle bed fixture 4, and the top of the lower needle test component 401 contacts the bottom of the airbag controller circuit board 5. By having the top of the lower needle test component 401 contact the bottom of the airbag controller circuit board 5, functional testing and software burning can be completed at the same workstation, reducing the time for handling and repositioning products between different devices and shortening the production cycle.

[0055] Combination Figure 1 , Figure 2 and Figure 3 In an embodiment of the present invention, the upper needle bed assembly 6 includes a connecting plate 601, which is fixedly installed at the bottom end of the hydraulic cylinder 202. A connecting rod 602 is installed at the bottom end of the connecting plate 601, and an upper needle bed fixture 603 is installed at the bottom end of the connecting rod 602. An upper needle test assembly 604 is installed at the bottom end of each upper needle bed fixture 603, and the bottom end of the upper needle test assembly 604 contacts the top end of the airbag controller circuit board 5. By using the hardware resources of the online tester, the software can be directly burned into the circuit board 5 without the need to purchase a dedicated burning device. This reduces the number of operating stations and personnel required, and also reduces the complexity of equipment maintenance.

[0056] Combination Figure 1 , Figure 2 and Figure 3In an embodiment of the present invention, the conveying assembly 8 includes a welding plate 801, which is fixedly installed on the top of the workbench 3. A conveying roller 802 is rotatably mounted inside the welding plate 801 via a bearing, and a conveyor belt body 803 is sleeved on the outer wall of the conveying roller 802. A driving assembly 804 is installed on the side wall of the welding plate 801, and the output end of the driving assembly 804 is fixedly connected to one end of the conveying roller 802. There are two sets of conveying rollers 802 with the same structure, and the two sets of conveying rollers 802 are connected by a sprocket and chain. The height of the conveyor belt body 803 is higher than the height of the lower needle test assembly 401. By starting the driving assembly 804, the conveying roller 802 is simultaneously driven to rotate, so that the conveyor belt body 803 can rotate and thus convey the airbag controller circuit board 5.

[0057] The specific working principle is as follows: When online testing of the airbag controller is required, the airbag controller circuit board 5 is first placed at the top of the conveyor belt body 803. Then, the drive component 804 is activated to simultaneously drive the conveyor roller 802 to rotate, which in turn drives the conveyor belt body 803 to rotate, allowing the conveyor belt body 803 to transport the airbag controller circuit board 5. When the airbag controller circuit board 5 is transported to the top of the lower needle bed fixture 4, the hydraulic cylinder 202 is activated to drive the upper needle bed fixture 603 and the upper needle test component 604 to press down, so that the test points of the airbag controller circuit board 5 are in close contact with the lower needle test component 401 at the top of the lower needle bed fixture 4. At the same time, the equipment is powered on, and the electrical test and software burning are sequentially controlled by the load board 7 at the rear. After the test is performed simultaneously, a self-test is completed, enabling one-stop testing and software burning. Functional testing and software burning are completed at the same station, reducing the time spent on handling and repositioning products between different devices and shortening the production cycle.

[0058] A method for online testing and software burning decision-making of airbag controllers, comprising the following steps:

[0059] Step 1: Obtain multi-dimensional test data from the integrated tooling for online testing of airbag controllers and software burning, and construct a systematic degradation index based on the multi-dimensional test data; collect the current statistical characteristics and preset historical statistical characteristics of the systematic degradation index, and calculate the cognitive uncertainty index.

[0060] Step 2: Compare the cognitive uncertainty index with a preset active detection trigger threshold to determine the active detection rate, and perform controller unit shunting based on the active detection rate; wherein, when the cognitive uncertainty index is greater than the active detection trigger threshold, the active detection rate is calculated based on the cognitive uncertainty index; when the cognitive uncertainty index is less than or equal to the active detection trigger threshold, the active detection rate is zero;

[0061] Step 3: Obtain the detection results obtained by the controller unit of the split-through test fixture on the offline test fixture, and the detection results of the fixture on the same controller unit; based on the detection results obtained by the offline test fixture and the detection results of the fixture, determine the cognitive bias event; and based on the cognitive bias event, dynamically calibrate the model parameters used to calculate the cognitive uncertainty index.

[0062] Example 2:

[0063] Reference Figure 5 A method for online testing and software burning decision-making of airbag controllers, comprising:

[0064] Acquire multi-dimensional test data from an integrated tooling for online testing and software burning of the airbag controller. The multi-dimensional test data includes electrical performance test pass rate, software burning verification success rate, communication protocol handshake duration, and sensor data drift rate.

[0065] A systematic degradation index is constructed based on the multi-dimensional test data to characterize the real-time health status of the integrated tooling production line.

[0066] Specifically, the systemic degradation index The construction steps are as follows:

[0067] Data normalization: Select test data in n dimensions Each dimension of the raw data has a preset ideal target value. The data for each dimension is normalized to obtain normalized values. For indicators where higher values ​​are always better, such as pass rate, the normalization formula is: For metrics where lower values ​​are generally better, such as response time and drift rate, the normalization formula is: ;

[0068] Weighted summation: Normalized data for each dimension Assign a weight coefficient ,in The weighting coefficients are determined based on the importance of the data in that dimension to the overall health of the system.

[0069] Index Calculation: Systemic Deterioration Index It is calculated using the following formula: The larger the index value, the higher the degree of system degradation;

[0070] The current statistical characteristics of the systemic degradation index and the preset historical statistical characteristics are collected to calculate the cognitive uncertainty index.

[0071] The calculation of the cognitive uncertainty index provides a quantitative indicator to assess the health of the artificial intelligence system itself, rather than relying solely on the production quality results output by artificial intelligence, thereby enabling risk management of unknown risks;

[0072] When the cognitive uncertainty index exceeds the preset active detection trigger threshold, the controller unit is shunted. This action can obtain reliable data on the real state of the integrated tooling, providing a data foundation for calibrating the artificial intelligence system and avoiding catastrophic risks.

[0073] Based on the detection results obtained from the offline testing fixture and the detection results from the integrated fixture, the cognitive bias events are determined, and the model parameters used to calculate the cognitive uncertainty index are dynamically calibrated. This closed-loop logic ensures that the system can learn from the detection actions and dynamically and data-drivenally adjust its risk model.

[0074] The current statistical features include the standard deviation of the systematic degradation index output sequence within the current evaluation period; the current statistical features also include the current probability distribution of the systematic degradation index output sequence; the historical statistical features include the mean of the standard deviation of the systematic degradation index sequence in historical normal operation data; the historical statistical features also include the probability distribution of historical normal data;

[0075] The steps for calculating the cognitive uncertainty index include:

[0076] Calculate the volatility deviation, wherein the volatility deviation is obtained based on the mean of the standard deviation of the systematic deterioration index output sequence and the standard deviation of the systematic deterioration index sequence in the historical normal operation data within the current assessment period;

[0077] Calculate the distribution change, wherein the distribution change is obtained based on the KL divergence between the current probability distribution of the systematic degradation index output sequence and the probability distribution of the historical normal data;

[0078] The volatility deviation and the distribution change are weighted and summed to generate the cognitive uncertainty index.

[0079] Example 3:

[0080] This embodiment is a further explanation based on Embodiment 2; the step of calculating the cognitive uncertainty index is aimed at capturing second-order risk signals that appear normal but are actually abnormal.

[0081] The cognitive uncertainty index is generated by calculating the volatility deviation and distribution change and then summing them with weights.

[0082] A normally functioning but blinded AI risk system often exhibits abnormal statistical characteristics in its output systematic degradation index. For example, the output may become abnormally smooth or its data distribution characteristics may completely deviate from historical patterns. The specific calculation formula for the cognitive uncertainty index is constructed as follows:

[0083]

[0084] in, Indicates the cognitive uncertainty index;

[0085] Indicates the systemic degradation index;

[0086] The standard deviation of the systematic degradation index output sequence within the current assessment period is used as the current statistical characteristic.

[0087] The mean of the standard deviation of the systematic degradation index series in the historical normal operation data is used as a historical statistical feature;

[0088] This represents the current probability distribution of the systematic degradation index output sequence. Probability distribution of historical normal data KL divergence between them;

[0089] This represents the weighting coefficients used to adjust for the relative importance of volatility bias and distribution variation; the weighting coefficients and the active detection trigger threshold mentioned in subsequent embodiments Minimum / maximum detection rate Detection rate adjustment factor These preset parameters are not arbitrarily set, but rather calibrated based on historical production data and expert experience; for example, weighting coefficients. The determination of trigger thresholds can be optimized by analyzing the correlation between different statistical characteristics in historical data and known system faults; proactive detection trigger thresholds. The cognitive uncertainty index can be set using statistical process control methods under historical normal operating conditions. Specific quantiles of the distribution, such as the 95th percentile, are used to balance the false alarm rate and the false negative rate.

[0090] The step of calculating the active detection rate involves using a normalized exponential scaling function, and inputting the cognitive uncertainty index, the active detection trigger threshold, the preset minimum detection rate, the preset maximum detection rate, and the preset detection rate adjustment factor, before outputting the active detection rate.

[0091] Example 4:

[0092] This embodiment further explains the step of determining the active detection rate in Embodiment 2. When the cognitive uncertainty just exceeds the threshold, the detection cost should be low, while when the uncertainty approaches the limit, high-frequency detection must be performed regardless of cost to obtain true information. To achieve this, a normalized exponential scaling function is used to calculate the active detection rate. The normalized exponential scaling function is constructed as follows:

[0093]

[0094] in, Indicates the active detection rate; This indicates the preset minimum detection rate; This indicates the preset maximum detection rate; The index representing the cognitive uncertainty of the input; This represents the upper limit of the preset cognitive uncertainty index; Indicates the threshold for active detection; This represents a preset detection rate adjustment factor used to control the nonlinearity of the detection rate as uncertainty increases. The calculation of this active detection rate allows for dynamic adjustment of the detection frequency based on the severity of perceived uncertainty, avoiding the problem that a linearly increasing detection rate may be insufficient to cope with the potential risks of exponential growth.

[0095] The step of determining the active detection rate is replaced by:

[0096] Based on the aforementioned cognitive uncertainty index, a potential risk cost is constructed;

[0097] Constructing detection costs related to active detection rate;

[0098] By combining the potential risk costs and the detection costs, a total expected cost function is established;

[0099] The optimal active detection rate is calculated by minimizing the total expected cost function, and the optimal active detection rate is used as the active detection rate.

[0100] The potential risk cost is determined based on the probability of blindness determined by the cognitive uncertainty index, combined with the preset inherent probability of defects and the quantified loss of a single catastrophic failure; the detection cost is determined based on the active detection rate, combined with the number of controllers per batch and the unit output value of a single controller.

[0101] Example 5:

[0102] This embodiment provides an alternative scheme for determining the active detection rate. This scheme transforms the complex decision-making problem into a computable cost minimization problem; based on the cognitive uncertainty index, a potential risk cost is constructed, and its calculation formula is as follows:

[0103]

[0104] in Indicates potential risk costs; Indicated by the cognitive uncertainty index The probability of being blinded is determined; This represents the pre-defined inherent probability of a defect; This represents the quantified loss from a single catastrophic failure.

[0105] Meanwhile, the detection cost related to the active detection rate is calculated using the following formula:

[0106]

[0107] in Indicates the cost of detection; Indicates the active detection rate; Indicates the number of controllers in a single batch; This represents the unit output value of a single controller;

[0108] By combining potential risk costs and detection costs, a total expected cost function is established, and the optimal active detection rate is calculated by minimizing this total expected cost function. The total expected cost function is constructed as follows:

[0109]

[0110] in, Represents the total expected cost function; The detection effectiveness coefficient represents the detection coverage and accuracy of the offline tooling. This method upgrades decision-making from passive, rule-based triggering to proactive, cost-effective, and dynamic optimization, avoiding both under- and over-detection.

[0111] The steps for determining cognitive bias events include:

[0112] When the detection result obtained by the offline testing fixture is a defect and the detection result of the fixture is qualified, a cognitive bias event is determined to have occurred.

[0113] When the detection result obtained by the offline testing fixture is qualified and the detection result of the fixture is defective, a cognitive bias event is determined to have occurred.

[0114] The steps for dynamically calibrating the model parameters include:

[0115] A Bayesian update logic is adopted, based on the prior probability of the AI-induced blindness state, and according to the conditional probability of the detection result mismatch occurring in the AI-induced blindness state and the conditional probability of the detection result mismatch occurring in the normal AI state, the posterior probability representing the AI-induced blindness state is updated.

[0116] The model parameters for dynamic calibration include weighting coefficients for calculating the cognitive uncertainty index and the probability of blindness state for calculating potential risk costs.

[0117] Example 6:

[0118] This embodiment further illustrates the cognitive resilience closed-loop calibration control logic in Embodiment 2. The step of determining a cognitive bias event is executed; a cognitive bias event is determined to have occurred when the detection results obtained by the offline testing fixture do not match the detection results of the integrated fixture. Each occurrence of a cognitive bias event serves as a falsification of the reliability of the integrated fixture's artificial intelligence system. The step of dynamically calibrating the model parameters employs Bayesian update logic, dynamically correcting its judgment of core risks based on newly obtained evidence. The formula for the Bayesian update logic is constructed as follows:

[0119]

[0120] in, This indicates the observation of cognitive bias events. Then, the posterior probability that the artificial intelligence is in a state of blindness; Represents the prior probability of artificial intelligence-induced blindness; This represents the conditional probability of a mismatch in detection results occurring under conditions of artificial intelligence-induced blindness. This represents the conditional probability of a mismatch in detection results occurring under normal artificial intelligence conditions; the updated posterior probability is used to dynamically calibrate model parameters, specifically including weight coefficients for calculating the cognitive uncertainty index and the probability of blindness state for calculating potential risk costs; this method enables the system to possess cognitive resilience, that is, after encountering uncertainty shocks, it can recover and improve its decision-making ability through learning and adaptation.

[0121] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A fixture integrating online testing and software burning for an airbag controller, characterized in that: include, The equipment body (1) has a fixed frame (2) installed on one side wall, and a support plate (201) is installed on the top of one side wall of the fixed frame (2). A hydraulic cylinder (202) is installed at the bottom of the support plate (201), and an upper needle bed assembly (6) is installed at the bottom of the hydraulic cylinder (202). A workbench (3) is installed at the top of the equipment body (1), and a lower needle bed fixture (4) is installed at the top of the workbench (3). A safety airbag controller circuit board (5) is installed at the top of the lower needle bed fixture (4). A rear load plate (7) is installed inside the equipment body (1), and a conveying assembly (8) is installed at the top of the workbench (3).

2. The integrated tooling for online testing and software burning of an airbag controller as described in claim 1, characterized in that: The top of each of the lower needle bed fixtures (4) is equipped with a lower needle test assembly (401), and the top of the lower needle test assembly (401) is in contact with the bottom of the airbag controller circuit board (5).

3. The integrated fixture for online testing and software burning of an airbag controller as described in claim 1, characterized in that: The upper needle bed assembly (6) includes a connecting plate (601), which is fixedly installed at the bottom end of the hydraulic cylinder (202). A connecting rod (602) is installed at the bottom end of the connecting plate (601), and an upper needle bed fixture (603) is installed at the bottom end of the connecting rod (602). An upper needle test assembly (604) is installed at the bottom end of the upper needle bed fixture (603), and the bottom end of the upper needle test assembly (604) is in contact with the top end of the airbag controller circuit board (5).

4. The integrated tooling for online testing and software burning of an airbag controller as described in claim 1, characterized in that: The conveying assembly (8) includes a welding plate (801), which is fixedly installed on the top of the workbench (3). A conveying roller (802) is rotatably installed inside the welding plate (801) via a bearing, and a conveyor belt body (803) is sleeved on the outer wall of the conveying roller (802). A drive assembly (804) is installed on the side wall of the welding plate (801), and the output end of the drive assembly (804) is fixedly connected to one end of the conveying roller (802).

5. The integrated fixture for online testing and software burning of an airbag controller as described in claim 4, characterized in that: The conveying rollers (802) are of the same structure and there are two sets. The two sets of conveying rollers (802) are connected by sprockets and chains. The height of the conveyor belt body (803) is higher than the height of the lower needle test assembly (401).

6. A method for online testing and software burning decision of an airbag controller is applied to an integrated tooling for online testing and software burning of an airbag controller as described in claims 1-5, characterized in that: The specific steps include: Step 1: Obtain multi-dimensional test data from the integrated tooling for online testing of airbag controllers and software burning, and construct a systematic degradation index based on the multi-dimensional test data; collect the current statistical characteristics and preset historical statistical characteristics of the systematic degradation index, and calculate the cognitive uncertainty index. Step 2: Compare the cognitive uncertainty index with a preset active detection trigger threshold to determine the active detection rate, and perform controller unit shunting based on the active detection rate; wherein, when the cognitive uncertainty index is greater than the active detection trigger threshold, the active detection rate is calculated based on the cognitive uncertainty index; when the cognitive uncertainty index is less than or equal to the active detection trigger threshold, the active detection rate is zero; Step 3: Obtain the detection results obtained by the controller unit of the split-through test fixture on the offline test fixture, and the detection results of the fixture on the same controller unit; based on the detection results obtained by the offline test fixture and the detection results of the fixture, determine the cognitive bias event; and based on the cognitive bias event, dynamically calibrate the model parameters used to calculate the cognitive uncertainty index.

7. The method for online testing and software burning decision of an airbag controller according to claim 6, characterized in that: The current statistical features include the standard deviation of the systematic degradation index output sequence within the current evaluation period; the current statistical features also include the current probability distribution of the systematic degradation index output sequence; the historical statistical features include the mean of the standard deviation of the systematic degradation index sequence in historical normal operation data; The historical statistical characteristics also include the probability distribution of historical normal data; The steps for calculating the cognitive uncertainty index include: Calculate the volatility deviation, wherein the volatility deviation is obtained based on the mean of the standard deviation of the systematic deterioration index output sequence and the standard deviation of the systematic deterioration index sequence in the historical normal operation data within the current assessment period; Calculate the distribution change, wherein the distribution change is obtained based on the KL divergence between the current probability distribution of the systematic degradation index output sequence and the probability distribution of the historical normal data; The volatility deviation and the distribution change are weighted and summed to generate the cognitive uncertainty index.

8. The method for online testing and software burning decision of an airbag controller according to claim 6, characterized in that: The step of calculating the active detection rate involves using a normalized exponential scaling function, and inputting the cognitive uncertainty index, the active detection trigger threshold, the preset minimum detection rate, the preset maximum detection rate, and the preset detection rate adjustment factor, before outputting the active detection rate.

9. The method for online testing and software burning decision of an airbag controller according to claim 6, characterized in that: The step of determining the active detection rate is replaced by: Based on the aforementioned cognitive uncertainty index, a potential risk cost is constructed; Constructing detection costs related to active detection rate; By combining the potential risk costs and the detection costs, a total expected cost function is established; The optimal active detection rate is calculated by minimizing the total expected cost function, and the optimal active detection rate is used as the active detection rate. The potential risk cost is determined based on the probability of blindness determined by the cognitive uncertainty index, combined with the preset inherent probability of defects and the quantified loss of a single catastrophic failure; the detection cost is determined based on the active detection rate, combined with the number of controllers per batch and the unit output value of a single controller.

10. The method for online testing and software burning decision of an airbag controller according to claim 6, characterized in that: The steps for determining cognitive bias events include: When the detection result obtained by the offline testing fixture is a defect and the detection result of the fixture is qualified, a cognitive bias event is determined to have occurred. When the detection result obtained by the offline testing fixture is qualified and the detection result of the fixture is defective, a cognitive bias event is determined to have occurred. The steps for dynamically calibrating the model parameters include: A Bayesian update logic is adopted, based on the prior probability of the AI-induced blindness state, and according to the conditional probability of the detection result mismatch occurring in the AI-induced blindness state and the conditional probability of the detection result mismatch occurring in the normal AI state, the posterior probability representing the AI-induced blindness state is updated. The model parameters for dynamic calibration include weighting coefficients for calculating the cognitive uncertainty index and the probability of blindness state for calculating potential risk costs.

Images