A method for optimizing reliability test scheduling under equipment resource constraints

Abstract

The present application relates to the field of scheduling optimization, in particular to a kind of reliability test scheduling optimization method under the detection equipment resource constraint, the present application is by calling the test task collection to be scheduled, obtains the task data of several test projects corresponding thereto, based on the latest start time of each test project reverse scheduling, generate initial scheduling;Extract resource difference feature and the difference amount of each environmental parameter release response duration, analyze the scheduling tightness characteristic value of test task collection, to mark test task collection;According to the marking result, the scheduling optimization of several tasks corresponding to test task collection is carried out;Output initial scheduling and new scheduling, and generate gantt chart containing task time sequence and equipment allocation relationship.The present application realizes the closed-loop optimization of adaptive rearrangement from static scheduling to dynamic exception driving in the typical scene of detection equipment resource constraint.At the same time, the adaptability, resource efficiency and anti-interference ability of reliability test scheduling are improved.

Description

Technical Field

[0001] This invention relates to the field of scheduling optimization, and in particular to a method for optimizing the scheduling of reliability tests under resource constraints of testing equipment. Background Technology

[0002] In reliability testing scenarios, testing equipment is a typical critical and constrained resource. Test tasks typically carry strict time window constraints and involve complex switching requirements for various environmental parameters. Equipment changeover time needs to be reserved between different test items, and the release of various environmental parameters often varies significantly, making it difficult to accurately predict the changeover window. Furthermore, frequent disturbances during test execution not only directly affect the reliability of test results but also cause ineffective long-term occupation of equipment and cavity resources, further exacerbating resource constraints. Therefore, there is an urgent need for a reliability test scheduling optimization method that can adapt to the resource constraints of testing equipment and has dynamic anomaly-driven adaptive rescheduling capabilities.

[0003] Chinese Patent Application Publication No. CN109615115A discloses an integrated production task scheduling method oriented towards task reliability. The specific steps are: 1. Based on the inherent relationship between the manufacturing system, production tasks, and manufactured products, identify the key machines and processes affecting task reliability; 2. Determine the performance status of each machine; 3. Quantify product quality deviation indicators; 4. Establish a QPPN model based on task requirements; 5. Provide heuristic rules based on task reliability; 6. Find the optimal solution for production task scheduling; 7. Analyze the results, comparing the results obtained by this patent's production scheduling method with those of production scheduling methods that do not consider task reliability. This method fully considers the impact of production task scheduling on the quality of manufactured workpieces, fundamentally making up for the shortcomings of traditional manufacturing processes that independently consider production scheduling and product quality inspection, improving the task reliability of the manufacturing system, and has good practical value.

[0004] However, the following problems still exist in the existing technology. They often employ static optimization or simple heuristic rules, making it difficult to fully utilize the device release window for adaptive adjustments, and they lack a real-time response mechanism for abnormal test conditions. Summary of the Invention

[0005] To address this, the present invention provides a reliability test scheduling optimization method under the constraint of testing equipment resources, which overcomes the problems in the prior art, which mostly adopts static optimization or simple heuristic rules, making it difficult to make full use of the equipment release window for adaptive adjustment, and lacks a real-time response mechanism for abnormal test conditions.

[0006] To achieve the above objectives, the present invention provides a method for optimizing reliability test scheduling under resource constraints of testing equipment, comprising: Call the set of test tasks to be scheduled, obtain the task data of several test items in the set of test tasks, and generate an initial schedule by reverse scheduling based on the latest start time of each test item. Extract the resource difference characteristics and the compactness of the latest start time for each of the aforementioned test projects, and analyze the scheduling tightness characterization value of the test task set to label the test task set; Based on the labeling results, the scheduling of several test items corresponding to the test task set is optimized, including, Determine whether an impactful event exists based on the test output data from the primary task; The experimental state data of the influencing event is obtained, the parameter drift curve is constructed, the parameter drift direction is locked, the cumulative drift of the drift parameter in the parameter drift direction and the uniformity of the drift amplitude in a unit time domain segment are identified, and the drift anomaly characterization value of the influencing event is analyzed to determine whether to start the task pre-rearrangement mechanism in advance. The task pre-reordering mechanism includes: extracting the earliest available release time node corresponding to the first task, placing a preset transformation time domain segment into the earliest available release time, filtering tasks that can be inserted into the queue, and determining whether the tasks that can be inserted into the queue meet the task scheduling requirements. Based on the test state characteristics of several interleaved tasks that meet the task scheduling requirements, the interleaved tasks are sorted to generate a new schedule. Output the initial schedule and the new schedule, and generate a Gantt chart containing the task timing and device allocation relationship; The resource difference characteristics include the variation in the release response time of each environmental parameter and the difference in environmental parameters between adjacent tasks. The test state characteristics include the number of sample types and the amount of equipment redundancy occupied.

[0007] Furthermore, the process of analyzing the scheduling tightness representation values ​​of the aforementioned set of experimental tasks includes: The sum of the ratio of the difference in release response time of each environmental parameter to the difference threshold and the ratio of the difference in environmental parameters between adjacent tasks to the difference threshold is used as the first scheduling tight feature value. The ratio of the latest start time compactness to the compactness threshold is used as the second scheduling compactness feature value; The first scheduling tightness feature value and the second scheduling tightness feature value are weighted and summed to determine the scheduling tightness characterization value.

[0008] Furthermore, the set of test tasks is tagged, including: If the scheduling tightness characteristic value of the test task set is greater than or equal to the scheduling tightness characteristic threshold, then the test task set is marked.

[0009] Furthermore, the scheduling of several test items corresponding to the test task set is optimized, including: If any set of test tasks is marked, then the scheduling of several test items corresponding to the set of test tasks is optimized.

[0010] Further, determine whether any influencing events exist, including: If the output uniformity of the experimental output data is less than the output uniformity threshold, and the total duration of the abnormal output data is greater than the total duration threshold, then an influencing event is determined to exist.

[0011] Furthermore, the process of analyzing the drift anomaly characterization values ​​of the influencing events includes: The ratio of the cumulative drift amount of the drift parameter in the drift direction to the drift accumulation threshold is used as the first drift anomaly feature value; The ratio of the drift amplitude uniformity threshold to the drift amplitude uniformity within a unit time domain segment is used as the second drift anomaly feature value. The first drift anomaly feature value and the second drift anomaly feature value are weighted and summed to determine the drift anomaly characterization value.

[0012] Furthermore, determining whether to initiate the task pre-rescheduling mechanism in advance includes: If the drift anomaly representation value affecting the event is greater than or equal to the drift anomaly representation threshold, then the task pre-reordering mechanism is activated in advance.

[0013] Further, filter for tasks that can be skipped, including: If any task meets the scheduling optimization conditions, then the task is identified as the queue-jumping task. The scheduling optimization conditions include that the total required cavity capacity is less than the release cavity capacity, and the number of occupied devices is less than the number of devices that can be released.

[0014] Further, determining whether the queue-jumping task meets the task scheduling requirements includes: If the time leeway between the mandatory start time and the latest start time of the queueable task is greater than the time leeway threshold, and the time interval between the completion release time of the queueable task and the mandatory start time of the original subsequent task is greater than the time interval threshold, then the queueable task is determined to meet the task scheduling requirements.

[0015] Furthermore, the process of sorting the aforementioned interleaved tasks to generate a new schedule includes: The ratio of the number of sample types to the threshold number of sample types is used as the first priority sufficiency feature value; The ratio of the redundancy threshold of occupied equipment to the redundancy of occupied equipment is used as the second priority sufficiency feature value; The first priority sufficiency feature value and the second priority sufficiency feature value are weighted and summed to determine the priority sufficiency representation value of the queueable task. Based on the priority abundance representation value, each eligible task is sorted in ascending order to generate a new schedule.

[0016] Compared with existing technologies, this invention obtains task data for several test items within a set of test tasks to be scheduled. Based on the latest start time of each test item, it performs reverse scheduling to generate an initial schedule. It extracts resource difference characteristics and the variation in response time of various environmental parameters, analyzes the scheduling tightness of the test task set, and marks the test task set accordingly. Based on the marking results, it optimizes the scheduling of several tasks corresponding to the test task set. It outputs the initial schedule and the new schedule, and generates a Gantt chart containing task timing and equipment allocation relationships. This invention addresses typical scenarios with resource constraints on testing equipment, such as environmental reliability testing and aging testing, achieving closed-loop optimization from static scheduling to dynamic anomaly-driven adaptive rescheduling. Simultaneously, it improves the adaptability, resource efficiency, and anti-interference capability of reliability test scheduling.

[0017] In particular, this invention comprehensively evaluates the overall scheduling tightness of experimental tasks under time constraints and environmental influences, adaptively marking the experimental task set to accurately locate high-risk task areas requiring optimization and improve scheduling resource utilization efficiency. From a time perspective, the compactness of the latest start time quantifies the concentration of deadline pressure on the time axis for multiple tasks, i.e., the number of tasks contained within a unit of time. Higher compactness indicates that the latest start times of multiple tasks are very close, the time window is narrower, and the number of tasks is greater. Failure to schedule in a timely manner can easily lead to overdue deadlines or resource preemption, thus reflecting the urgency in the time dimension. From a resource perspective, the difference in environmental parameters between adjacent tasks quantifies the variation range of environmental conditions between consecutive tasks, such as temperature, humidity, and pressure. A larger difference means that the equipment requires a longer adjustment time, such as for heating, cooling, and humidification, thus increasing the waiting overhead for task switching, reflecting the cost of resource switching. Simultaneously, the variation in the release response time of each environmental parameter quantifies the inconsistency, i.e., fluctuation or dispersion, of the time required for different environmental parameters to release from the current state to the target state. The greater the variance, the faster some parameters respond and the slower others respond. The overall switching time is limited by the slowest parameter, leading to decreased scheduling predictability and a higher likelihood of unexpected idle waiting, thus reflecting the uncertainty of resource response. This invention analyzes the scheduling tightness of the experimental task set based on the above three characteristics to characterize the tightness of the experimental task scheduling and the risk of scheduling conflicts. This invention analyzes the coupling effect of time and resources, improves the ability to predict scheduling feasibility, provides a trigger basis for subsequent dynamic rescheduling, adapts to multi-constraint experimental scenarios, and improves the scheduling robustness of tasks.

[0018] In particular, this invention, based on real-time acquisition of the first task's test output data, comprehensively evaluates its output consistency and whether the duration of abnormal states exceeds the normal range, thereby determining whether there are influencing events. This allows for early identification of abnormal fluctuations during the test process, timely triggering of subsequent scheduling adjustments, and preventing invalid tests from occupying equipment resources for extended periods. From a quality consistency perspective, the output consistency of the test output data is considered, i.e., the level of output fluctuation among multiple output data under the same test task, quantifying the dispersion and intensity of the output data. The lower the output consistency, the more dispersed and volatile the data, the more unstable the test state, and the higher the likelihood of being caused by factors such as parameter drift, equipment aging, and environmental disturbances. This indicator characteristic allows for the determination of whether observable performance degradation or abnormal fluctuations have occurred in the current test, which is the primary condition for triggering the determination of influencing events. From an abnormality persistence perspective, the cumulative duration of the abnormal state presented by the output data is quantified by the total duration of the abnormal output data. The longer the duration, the more likely the abnormality is not occasional instantaneous noise, but a persistent drift or failure trend, with a greater impact on the reliability of test results and equipment occupancy time. Furthermore, by combining the two features mentioned above, this invention can effectively distinguish between incidental noise and genuine parameter drift events, improving the accuracy and robustness of anomaly detection. Moreover, after determining the existence of an influencing event, this invention triggers subsequent pre-sorting operations to avoid prolonged invalid experiments due to parameter drift, thereby releasing occupied equipment and cavity resources and improving overall resource utilization. Simultaneously, it improves the accuracy and stability of experimental task determination, ensuring the reliability of experimental results.

[0019] In particular, this invention analyzes the experimental state data affecting events, quantifying the overall deviation of drift parameters in a specific drift direction—that is, the total intensity or total loss of the drift—by the cumulative amount of drift in the drift parameter along the drift direction. For example, the area enclosed between the drift curve and the baseline curve reflects the severity of the accumulated parameter drift. The larger this characteristic value, the more the experimental state deviates from the normal range, and the higher the risk of approaching failure or exceeding tolerance. Furthermore, the stability and consistency of the drift process on the time axis are quantified by the uniformity of the drift amplitude within a unit time domain segment. High uniformity indicates that the drift is uniform and linear, with strong predictability; low uniformity indicates the presence of irregular fluctuations, jumps, or oscillations, and more complex abnormal patterns. Therefore, this invention analyzes the drift anomaly characterization values ​​affecting events by combining the above two dimensions of characteristics, to more accurately characterize the degree of anomaly of the affecting events, thereby determining whether to activate the task pre-rearrangement mechanism. This invention identifies parameter drift trends and proactively triggers a pre-rescheduling mechanism, thereby effectively utilizing equipment waiting or replacement windows to perform other tasks, reducing resource idleness, and improving overall scheduling efficiency.

[0020] In particular, this invention considers that in resource-constrained reliability testing scenarios, fewer types and numbers of test samples generally mean a simpler test object, fewer fixture or tooling changes, a simpler test process, and a relatively shorter execution time. Prioritizing these tasks allows them to complete quickly and release the occupied equipment, thereby shortening the time tasks occupy critical resources and improving equipment turnover. Furthermore, the amount of equipment redundancy reflects the amount of equipment resources currently occupied by a task. A larger redundancy means that once the task is completed, more equipment units can be released, accommodating more new tasks or alleviating resource constraints. Prioritizing these tasks can quickly release a large amount of equipment resources, increasing the system's idle resource pool, which is beneficial for the parallel execution of subsequent tasks or the insertion of urgent tasks, reducing scheduling congestion. Based on this, this invention relies on the priority abundance characteristic value of queueable tasks to characterize the priority scheduling value of queueable tasks in terms of both time efficiency and resource release benefits, and performs dual-dimensional ranking to achieve a balance between resource utilization and task complexity, prioritizing the execution of lightweight tasks and improving equipment turnover. Furthermore, this invention enables the rapid release of cavity capacity and equipment resources, providing a available resource window for subsequent tasks and improving parallel experimental capabilities. By rationally prioritizing tasks, it reduces waiting time, improves overall scheduling efficiency, and shortens the overall experimental cycle. Attached Figure Description

[0021] Figure 1 This is a schematic diagram illustrating the steps of a reliability test scheduling optimization method under testing equipment resource constraints, as described in an embodiment of the invention. Figure 2 A logic diagram for determining whether an influencing event exists in an embodiment of the invention; Figure 3 A logic diagram for determining whether to initiate the task pre-rescheduling mechanism in advance, as shown in an embodiment of the invention. Figure 4 A logic diagram for filtering queue-jumping tasks in an embodiment of the invention. Detailed Implementation

[0022] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0023] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0024] It should be noted that in the description of this invention, the terms "upper", "lower", "left", "right", "inner", "outer", etc., which indicate directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and is not intended to indicate or imply that the device or element must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0025] Furthermore, it should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0026] Please see Figure 1 The diagram illustrates the steps of a reliability test scheduling optimization method under testing equipment resource constraints according to an embodiment of the present invention. The method includes: Step S1: Call the set of test tasks to be scheduled, obtain the task data of several test items in the set of test tasks, and generate an initial schedule by reverse scheduling based on the latest start time of each test item. Step S2: Extract the resource difference characteristics and the compactness of the latest start time for each of the test projects, and analyze the scheduling tightness characterization value of the test task set to mark the test task set; Step S3: Based on the labeling results, optimize the scheduling of several test items corresponding to the test task set, including... Determine whether an impactful event exists based on the test output data from the primary task; The experimental state data of the influencing event is obtained, the parameter drift curve is constructed, the parameter drift direction is locked, the cumulative drift of the drift parameter in the parameter drift direction and the uniformity of the drift amplitude in a unit time domain segment are identified, and the drift anomaly characterization value of the influencing event is analyzed to determine whether to start the task pre-rearrangement mechanism in advance. The task pre-reordering mechanism includes: extracting the earliest available release time node corresponding to the first task, placing a preset transformation time domain segment into the earliest available release time, filtering tasks that can be inserted into the queue, and determining whether the tasks that can be inserted into the queue meet the task scheduling requirements. Based on the test state characteristics of several interleaved tasks that meet the task scheduling requirements, the interleaved tasks are sorted to generate a new schedule. Step S4: Output the initial schedule and the new schedule, and generate a Gantt chart containing the task timing and device allocation relationship; The resource difference characteristics include the variation in the release response time of each environmental parameter and the difference in environmental parameters between adjacent tasks. The test state characteristics include the number of sample types and the amount of equipment redundancy occupied.

[0027] Specifically, the task data includes resource difference characteristics and the compactness of the latest start time, while the test output data includes the cumulative drift of the drift parameter in the drift direction, the uniformity of the drift amplitude within a unit time domain segment, the earliest available release time node, and test state characteristics.

[0028] The relevant feature data involved in the task data and test output data can be obtained by collecting the underlying raw data through existing test management systems and equipment automation systems, such as MES, LIMS, and SCADA, and automatically calculating them according to preset mathematical formulas and statistical algorithms.

[0029] Specifically, the actual changeover time records between different test tasks on the same equipment can be collected, and the 95th percentile of these records can be used as the duration of the preset changeover time range. This approach covers most changeover scenarios while avoiding excessive conservatism. Of course, those skilled in the art can adaptively adjust this duration based on equipment characteristics, test specifications, and real-time status, which will not be elaborated upon here.

[0030] Specifically, this invention uses a reverse scheduling method to determine the initial schedule for several test items in the test task set, thereby improving the scheduling's ability to meet deadlines and ensuring that urgent tasks are not delayed.

[0031] Specifically, the process of analyzing the scheduling tightness representation values ​​of the aforementioned set of experimental tasks includes: The sum of the ratio of the difference in release response time of each environmental parameter to the difference threshold and the ratio of the difference in environmental parameters between adjacent tasks to the difference threshold is used as the first scheduling tight feature value. The ratio of the latest start time compactness to the compactness threshold is used as the second scheduling compactness feature value; The first scheduling tightness feature value and the second scheduling tightness feature value are weighted and summed to determine the scheduling tightness characterization value.

[0032] Specifically, the tightness of the latest start time directly determines whether a task can be completed before the deadline. Under the constraint of equipment resources, if a task misses its latest start time, it will inevitably lead to delays, potentially resulting in hefty penalties or product delivery failures. Therefore, the tightness of the time dimension is a "hard constraint" or "high-priority soft constraint" for scheduling feasibility, reflecting the urgency of the time window constraint. The two features involved in the first scheduling tightness feature value mainly affect equipment switchover time, settling time, or waiting time between tasks. Although both affect the tightness of the schedule and equipment utilization, they can usually be mitigated by adjusting the task order or inserting idle time, and will not directly cause task delays. Therefore, assigning a higher weight coefficient to the second scheduling tightness feature value ensures that the scheduling algorithm prioritizes meeting the time window constraint before optimizing idle time caused by environmental parameters. A value of 0.6-0.7 can be selected; in this embodiment, it is set to 0.65. Correspondingly, the first scheduling tightness feature value is selected between 0.3-0.4; in this embodiment, it is set to 0.35. Of course, the specific values ​​can be dynamically adjusted based on the actual operating data of the testing laboratory, such as the historical task delay rate or the proportion of equipment switching waiting time, which will not be elaborated here.

[0033] In this embodiment, the purpose of setting the variance threshold, environmental parameter difference threshold, and compactness threshold is to characterize situations where the scheduling of test tasks is highly compact and the risk of scheduling conflicts is high. This is achieved by acquiring task data corresponding to several different test projects completed by the same device, and by calling the variance data of the release response time of each environmental parameter, the environmental parameter difference data between adjacent tasks, and the compactness data of the latest start time. The mean variance, mean environmental parameter difference, and mean compactness are then calculated. Based on the purpose of setting the above three thresholds, the variance threshold is determined to be the product of the mean variance and the variance deviation coefficient; the environmental parameter difference threshold is determined to be the product of the mean environmental parameter variance and the difference deviation coefficient; and the compactness threshold is determined to be the product of the mean compactness and the compactness deviation coefficient.

[0034] The variation in the response time of each environmental parameter is significantly affected by factors such as equipment status and environmental disturbances, resulting in noticeable data fluctuations. Therefore, the corresponding variation deviation coefficient is set within a relatively large range [1.2, 1.4], with 1.3 being the preferred value in practice. The difference in environmental parameters between adjacent tasks is determined by the test specifications and fluctuates relatively little. Therefore, the corresponding difference deviation coefficient adopts a small range [1.1, 1.3], and is preferably 1.2 in practice; The compactness of the latest start time is based on precise time data, resulting in the lowest randomness. Therefore, its corresponding compactness deviation coefficient adopts a small range [1.1, 1.2], and is preferably 1.15 in practice.

[0035] The selection of the above-mentioned preferred values ​​can effectively avoid false triggering caused by normal fluctuations while ensuring sufficient sensitivity to actual scheduling tensions. Of course, those skilled in the art can also adjust the specific values ​​within the corresponding range according to the equipment characteristics and the degree of fluctuation in historical data.

[0036] Specifically, the target set values ​​for each environmental parameter of two adjacent tasks are obtained. For each environmental parameter, the absolute difference between the values ​​of the two tasks on the corresponding environmental parameter is calculated. The absolute differences of all environmental parameters are summed to obtain the environmental parameter difference between adjacent tasks.

[0037] Specifically, the variation in the release response time of each environmental parameter refers to the difference in the switching time between the current value and the environmental parameters required for subsequent tasks after the task ends. It is determined by calculating the coefficient of variation of the switching time of each environmental parameter to characterize the dispersion of these release response times. A larger coefficient of variation indicates a greater difference in the recovery speed of each parameter, and thus a greater variation.

[0038] Specifically, the compactness of the latest start time is determined by the following methods: ; Where L represents the compactness of the latest start time, n represents the number of test items in the test task set, Tmax represents the maximum value of the latest start time in the test task set, and Tmin represents the minimum value of the latest start time in the test task set.

[0039] The larger the compactness L of the latest start time, the more concentrated the latest start times of each task are on the timeline, and the higher the scheduling compactness.

[0040] Specifically, this invention comprehensively evaluates the overall scheduling tightness of experimental tasks under time constraints and environmental influences, adaptively marking the experimental task set to accurately locate high-risk task areas requiring optimization and improve scheduling resource utilization efficiency. From a time perspective, the compactness of the latest start time quantifies the concentration of deadline pressure on the time axis for multiple tasks, i.e., the number of tasks contained within a unit of time. Higher compactness indicates that the latest start times of multiple tasks are very close, the time window is narrower, and the number of tasks is greater. Failure to schedule in a timely manner can easily lead to overdue deadlines or resource preemption, thus reflecting the urgency in the time dimension. From a resource perspective, the difference in environmental parameters between adjacent tasks quantifies the variation range of environmental conditions between consecutive tasks, such as temperature, humidity, and pressure. A larger difference means that the equipment needs a longer adjustment time, such as for heating, cooling, and humidification, thus increasing the waiting overhead for task switching, reflecting the cost of resource switching. Simultaneously, the variation in the release response time of each environmental parameter quantifies the inconsistency, i.e., fluctuation or dispersion, of the time required for different environmental parameters to release from the current state to the target state. The greater the variance, the faster some parameters respond and the slower others respond. The overall switching time is limited by the slowest parameter, leading to decreased scheduling predictability and a higher likelihood of unexpected idle waiting, thus reflecting the uncertainty of resource response. This invention analyzes the scheduling tightness of the experimental task set based on the above three characteristics to characterize the tightness of the experimental task scheduling and the risk of scheduling conflicts. This invention analyzes the coupling effect of time and resources, improves the ability to predict scheduling feasibility, provides a trigger basis for subsequent dynamic rescheduling, adapts to multi-constraint experimental scenarios, and improves the scheduling robustness of tasks.

[0041] Specifically, the set of test tasks is labeled, including: If the scheduling tightness characteristic value of the test task set is greater than or equal to the scheduling tightness characteristic threshold, then the test task set is marked.

[0042] To determine the tightness threshold for scheduling, we can statistically analyze the distribution of tightness values ​​corresponding to the optimal feasible scheduling scheme adopted by the testing laboratory in historical batches, such as the approximate optimal solution adjusted by human experts or verified by the solver. The 25th percentile of this distribution is taken as the lower limit reference value. This threshold is usually set between 0.6 and 0.9. In this embodiment, it is set to 0.75, which can ensure that false alarms caused by environmental parameter disturbances or small fluctuations in release time are eliminated while minimizing the number of weighted delayed tasks, and at the same time retain the ability to sensitively distinguish truly urgent tasks.

[0043] The scheduling tightness characterization value calculated is determined as the scheduling tightness characterization threshold when the difference in release response time of each environmental parameter is equal to the difference threshold, the difference in environmental parameters between adjacent tasks is equal to the difference threshold, and the tightness of the latest start time is equal to the tightness threshold.

[0044] Specifically, the scheduling of several test items corresponding to the test task set is optimized, including: If any set of test tasks is marked, then the scheduling of several test items corresponding to the set of test tasks is optimized.

[0045] Specifically, please refer to Figure 2 As shown, this is a logic diagram for determining whether an influencing event exists according to an embodiment of the present invention. Determining whether an influencing event exists includes: If the output uniformity of the experimental output data is less than the output uniformity threshold, and the total duration of the abnormal output data is greater than the total duration threshold, then an influencing event is determined to exist.

[0046] In this embodiment, the purpose of setting the output uniformity threshold is to characterize situations where the output level of the task's output data is relatively volatile and the stability of the test state is poor. The purpose of setting the total duration threshold is to characterize situations where the output data has a significant impact on the reliability of the test results and the equipment occupancy time. By obtaining test state data corresponding to several different test items completed by the same equipment, the output uniformity data of the test output data and the total duration data of abnormal output data are retrieved. The average output uniformity and the average total duration are calculated. Based on the purpose of setting the above two thresholds, the output uniformity threshold is determined as the product of the average output uniformity and the output deviation coefficient, and the total duration threshold is determined as the product of the average total duration and the duration deviation coefficient. The output deviation coefficient is selected within the range [0.85, 0.95], so that the threshold is slightly lower than the normal level, allowing a certain degree of normal fluctuation, and avoiding the accidental triggering of events due to small fluctuations. In practice, 0.9 is preferred. The duration deviation coefficient is selected within the range [1.2, 1.4], such that the threshold is slightly higher than the normal duration to exclude instantaneous or occasional anomalies. The subsequent rearrangement mechanism is only triggered when the anomaly lasts long enough to potentially impair the effectiveness of the test or occupy the equipment for an extended period. In practice, a value of 1.3 is preferred.

[0047] Of course, those skilled in the art can adjust the value range based on equipment stability, test importance, and tolerance for false alarms, or adjust the coefficients within a given range. For example, when high stability is required, a larger output deviation coefficient and a smaller persistence deviation coefficient can be used to improve sensitivity. Conversely, a smaller output deviation coefficient and a larger persistence deviation coefficient can be used to improve fault tolerance.

[0048] Specifically, the continuous output data sequence during the task execution process is obtained according to a fixed sampling frequency, such as temperature, humidity, voltage, stress, etc. The sampling point data and the corresponding data value of each sampling point are recorded, the mean and standard deviation are calculated, and then the coefficient of variation is obtained. The coefficient of variation is normalized, i.e., 1-coefficient of variation, and mapped to the interval [0, 1] to obtain the output uniformity. The closer the output consistency is to 1, the smaller the data fluctuation and the better the consistency; the closer it is to 0, the more dispersed the data and the greater the fluctuation.

[0049] Specifically, in actual reliability testing, each test item has a corresponding normal range for output data, including temperature, humidity, voltage, stress, etc. This normal range is predetermined by the technical requirements of the test task, product specifications, or relevant standards, and serves as part of the input conditions for the test item. Therefore, during the task execution, if the output data collected at any sampling moment exceeds the corresponding task's preset normal range, i.e., the allowable upper or lower limit of fluctuation, then that output data is considered abnormal output data.

[0050] Specifically, this invention, based on real-time acquisition of the first task's test output data, comprehensively evaluates its output consistency and whether the duration of abnormal states exceeds the normal range, thereby determining whether there are influencing events. This allows for early identification of abnormal fluctuations during the test process, timely triggering of subsequent scheduling adjustments, and preventing invalid tests from occupying equipment resources for extended periods. From a quality consistency perspective, the output consistency of the test output data is considered, i.e., the level of output fluctuation among multiple output data under the same test task, quantifying the dispersion and intensity of the output data. The lower the output consistency, the more dispersed and volatile the data, the more unstable the test state, and the higher the likelihood of being caused by factors such as parameter drift, equipment aging, and environmental disturbances. This indicator characteristic allows for the determination of whether observable performance degradation or abnormal fluctuations have occurred in the current test, which is the primary condition for triggering the determination of influencing events. From an abnormality persistence perspective, the cumulative duration of the abnormal output data is quantified by the total duration of the abnormal output data. The longer the duration, the more likely the abnormality is not occasional instantaneous noise but a persistent drift or failure trend, with a greater impact on the reliability of test results and equipment occupancy time. Furthermore, by combining the two features mentioned above, this invention can effectively distinguish between incidental noise and genuine parameter drift events, improving the accuracy and robustness of anomaly detection. Moreover, after determining the existence of an influencing event, this invention triggers subsequent pre-sorting operations to avoid prolonged invalid experiments due to parameter drift, thereby releasing occupied equipment and cavity resources and improving overall resource utilization. Simultaneously, it improves the accuracy and stability of experimental task determination, ensuring the reliability of experimental results.

[0051] Specifically, the process of analyzing the drift anomaly representation values ​​of the influencing events includes: The ratio of the cumulative drift amount of the drift parameter in the drift direction to the drift accumulation threshold is used as the first drift anomaly feature value; The ratio of the drift amplitude uniformity threshold to the drift amplitude uniformity within a unit time domain segment is used as the second drift anomaly feature value. The first drift anomaly feature value and the second drift anomaly feature value are weighted and summed to determine the drift anomaly characterization value.

[0052] Specifically, in electronic component reliability testing, the cumulative amount of parameter drift directly reflects the degree of component degradation and is a direct basis for determining whether it exceeds the allowable tolerance range or whether failure has occurred. Once the cumulative amount exceeds the threshold, it means that the component may fail, thus this feature is of decisive significance. The uniformity of drift amplitude within a unit time domain mainly reflects the stability and consistency of the drift process. Its anomaly may indicate intermittent faults or measurement noise, but an abnormal uniformity does not necessarily lead to failure. This feature is more used to assist in diagnosing drift patterns rather than as a direct basis for determining failure. Therefore, assigning a higher weight coefficient to the first drift anomaly feature value ensures that the scheduling algorithm prioritizes meeting time window constraints before optimizing idle time caused by environmental parameters. This value can be selected between 0.7 and 0.8; in this embodiment, it is set to 0.75. Correspondingly, the second drift anomaly feature value is selected between 0.2 and 0.3; in this embodiment, it is set to 0.25. Of course, the specific values ​​can be dynamically adjusted based on historical failure data or drift monitoring records from the actual testing laboratory, which will not be elaborated further here.

[0053] In this embodiment, the purpose of setting the drift accumulation threshold and the drift amplitude uniformity threshold is to characterize situations where the degree of abnormality of the influencing event is large. This is achieved by acquiring test state data corresponding to several different test items completed by the same device, and calling the drift accumulation data of the drift parameter in the corresponding parameter drift direction, as well as the drift amplitude uniformity data within a unit time domain segment. The mean of the drift accumulation and the mean of the drift amplitude uniformity are calculated. Based on the purpose of setting the above two thresholds, the drift accumulation threshold is determined as the product of the mean drift accumulation and the cumulative deviation coefficient, and the drift amplitude uniformity threshold is determined as the product of the mean drift amplitude uniformity and the amplitude deviation coefficient. The cumulative deviation coefficient is selected within the interval [1.2, 1.4], making the threshold higher than the normal level, allowing a certain degree of random cumulative fluctuation. Subsequent judgment is only triggered when the total drift amount significantly exceeds the standard; in practice, a value of 1.3 is preferred. The amplitude deviation coefficient is selected within the range [0.85, 0.95] to make the threshold lower than the normal level, allowing a certain degree of slight inhomogeneity. It is only judged as abnormal when the drift process shows obvious inhomogeneity, such as intermittent sudden changes. In practice, it is preferred to be 0.9.

[0054] The above-mentioned preferred values ​​balance sensitivity and stability. Of course, those skilled in the art can adjust the value range based on equipment stability, test importance, and tolerance for false alarms, or adjust the coefficients within a given range. To increase sensitivity to drift anomalies, a smaller cumulative deviation coefficient and a larger amplitude deviation coefficient can be used; conversely, to prioritize avoiding false triggers, a larger cumulative deviation coefficient and a smaller amplitude deviation coefficient can be used.

[0055] Specifically, the drift amount of the drift parameter in the drift direction is determined according to the area formula between curves, and the drift amount is summed to obtain the cumulative drift amount.

[0056] The drift curve is divided into several unit time-domain segments of equal length. The area between the drift curve and the reference curve in each time-domain segment is calculated as the drift accumulation of the corresponding time-domain segment. The mean and standard deviation of the drift accumulation of all segments are calculated to obtain the coefficient of variation. Finally, the coefficient of variation is normalized to 1-coefficient of variation and mapped to the interval [0, 1] to obtain the uniformity of the drift amplitude. The closer the value is to 1, the more uniform the drift is in each time-domain segment.

[0057] Specifically, the process of constructing the parameter drift curve includes, Construct a rectangular coordinate system with time as the horizontal axis and environmental parameter values ​​as the vertical axis; The coordinates of the environmental parameter values ​​at each moment are marked in the rectangular coordinate system; The parameter drift curve is obtained by connecting the coordinate points with a smooth curve.

[0058] Specifically, there are no restrictions on the method for constructing the parameter drift curve. For example, the parameter drift curve can be fitted using Matlab correlation fitting software, which will not be elaborated further.

[0059] Based on the overall trend of parameter drift curves, the direction of monotonically increasing or decreasing environmental parameters over time can be identified to pinpoint the direction of parameter drift.

[0060] Specifically, this invention analyzes experimental state data affecting events, quantifying the overall deviation of drift parameters in a specific drift direction—that is, the total intensity or total loss of the drift—by the cumulative amount of drift in the drift parameter's drift direction. For example, the area enclosed between the drift curve and the baseline curve reflects the severity of the accumulated parameter drift. The larger this characteristic value, the more the experimental state deviates from the normal range, and the higher the risk of approaching failure or exceeding tolerance. Furthermore, the stability and consistency of the drift process on the time axis are quantified by the uniformity of the drift amplitude within a unit time domain segment. High uniformity indicates that the drift is uniform and linear, with strong predictability; low uniformity indicates the presence of irregular fluctuations, jumps, or oscillations, with more complex abnormal patterns. Therefore, this invention analyzes the drift anomaly characterization values ​​of events by combining the above two dimensions of characteristics, to more accurately characterize the degree of anomaly of events, thereby determining whether to activate the task pre-rearrangement mechanism. This invention identifies parameter drift trends and proactively triggers a pre-rescheduling mechanism, thereby effectively utilizing equipment waiting or replacement windows to perform other tasks, reducing resource idleness, and improving overall scheduling efficiency.

[0061] Specifically, please refer to Figure 3 As shown, this is a logic diagram for determining whether to start the task pre-rescheduling mechanism in advance according to an embodiment of the present invention. Determining whether to start the task pre-rescheduling mechanism in advance includes: If the drift anomaly representation value affecting the event is greater than or equal to the drift anomaly representation threshold, then the task pre-reordering mechanism is activated in advance.

[0062] The drift anomaly characterization threshold is predetermined. The drift anomaly characterization value calculated is determined by assuming that the drift accumulation amount of the drift parameter in the drift direction is equal to the drift accumulation amount threshold, and the drift amplitude uniformity threshold is equal to the drift amplitude uniformity within a unit time domain segment.

[0063] To determine the threshold for drift anomaly characterization, statistical modeling can be performed on the drift anomaly characterization values ​​of components marked as normal in historical batches of the testing laboratory within a complete test cycle. The 90th percentile of the characterization values ​​of normal samples is taken as the lower limit of the benchmark. Combined with receiver operating characteristic curve analysis, the critical point that balances the accuracy and recall of failure detection is determined. This threshold is usually set between 0.65 and 0.85. In this embodiment, it is set to 0.75, which can control the false alarm rate of normal fluctuations to within 5% while ensuring that the detection rate of real drift failure events is not less than 95%, thereby balancing the sensitivity of failure early warning and engineering reliability.

[0064] Specifically, please refer to Figure 4 As shown, this is a logic decision diagram for filtering queue-jumping tasks according to an embodiment of the present invention. Filtering queue-jumping tasks includes: If any task meets the scheduling optimization conditions, then the task is identified as the queue-jumping task. The scheduling optimization conditions include that the total required cavity capacity is less than the release cavity capacity, and the number of occupied devices is less than the number of devices that can be released.

[0065] Specifically, this invention filters out queueable tasks with matching resource occupancy conditions within the equipment resource release window, making full use of equipment waiting or replacement intervals to avoid resource idleness. It ensures task queueing is performed without compromising the feasibility of the original schedule and fully utilizes the replacement window to improve equipment time utilization and reduce scheduling risks between task sequences. Simultaneously, constraints on resource capacity and equipment occupancy ensure that queued tasks do not exceed the current available resource limit, guaranteeing schedule executability and improving overall equipment utilization efficiency and schedule compactness.

[0066] Specifically, determining whether the queue-jumping task meets the task scheduling requirements includes: If the time leeway between the mandatory start time and the latest start time of the queueable task is greater than the time leeway threshold, and the time interval between the completion release time of the queueable task and the mandatory start time of the original subsequent task is greater than the time interval threshold, then the queueable task is determined to meet the task scheduling requirements.

[0067] In this embodiment, the purpose of setting time leeway thresholds and time interval thresholds is to filter out tasks that have sufficient flexibility and ensure that their release time will not pressure subsequent tasks. This is achieved by acquiring test status data corresponding to several different test items completed on the same device, and by retrieving the time leeway data between the forced start time and the latest start time of the task that can be queued, as well as the time interval data between the completion release time of the task and the forced start time of the original subsequent task. The average time leeway and the average time interval are calculated. Based on the purpose of setting the above two thresholds, the time leeway threshold is determined as the product of the average time leeway and the leeway deviation coefficient, and the time interval threshold is determined as the product of the average time interval and the time deviation coefficient. The leeway deviation coefficient is selected within the interval [1.1, 1.3], making the threshold higher than the average level. Queuing is only allowed when the time flexibility of the task is significantly better than usual, ensuring that the queued task can be completed on time. In practice, a value of 1.2 is preferred. The time deviation coefficient is selected within the range [1.1, 1.3] to make the threshold higher than the average level. Queuing is only allowed when there is sufficient safety buffer between the released queued task and the original subsequent task to avoid chain delays. In practice, a value of 1.2 is preferred.

[0068] Of course, those skilled in the art can adjust the value range according to their tolerance for scheduling risks, or adjust the coefficient within a given range. For example, when the risk tolerance is low, a larger coefficient can be used to raise the screening threshold; when the risk tolerance is high, a smaller coefficient can be used to increase the chance of skipping the queue.

[0069] Specifically, this invention ensures that the insertion of queue-jumping tasks does not disrupt the feasibility and stability of the original schedule by setting dual time constraints. First, the sufficient time leeway between the mandatory start time and the latest start time of a queue-jumping task—that is, the length of the acceptable start time window for the task—means that even if the task starts earlier or later, there is still ample room for maneuver, preventing it from missing the latest start time due to occupying part of the window. This avoids the queue-jumping task itself being forced to delay or fail during execution, thus ensuring its completion. Second, the time interval between the completion and release time of a queue-jumping task and the mandatory start time of the original subsequent task ensures that the queue-jumping task will not delay the normal start of subsequent tasks. Even if there are slight fluctuations in the actual execution of the queue-jumping task, it can act as a buffer, avoiding cascading delays and maintaining the execution plan of subsequent tasks in the original schedule. Furthermore, this allows the pre-rescheduling mechanism to insert new tasks using equipment replacement or waiting windows without disrupting the stability of the original schedule, achieving efficient utilization of equipment idle windows and improving overall scheduling quality.

[0070] Specifically, the process of sorting the aforementioned interleaved tasks and generating a new schedule includes: The ratio of the number of sample types to the threshold number of sample types is used as the first priority sufficiency feature value; The ratio of the redundancy threshold of occupied equipment to the redundancy of occupied equipment is used as the second priority sufficiency feature value; The first priority sufficiency feature value and the second priority sufficiency feature value are weighted and summed to determine the priority sufficiency representation value of the queueable task. Based on the priority abundance representation value, each eligible task is sorted in ascending order to generate a new schedule.

[0071] Specifically, the core of queue-jumping tasks lies in whether they can be inserted for execution without severely interfering with existing tasks. The amount of equipment redundancy reflects the current idle level of equipment resources. The smaller the redundancy, the more strained the equipment, and the greater the risk of conflict and delay caused by queue-jumping. Therefore, this feature plays a decisive role in determining queue-jumping priority. The number of sample types reflects the diversity and complexity of the current batch of tasks, but this feature can be mitigated by task splitting, batch processing, or adjusting the test order. Its urgency is usually lower than the real-time occupancy status of equipment resources. Therefore, assigning a higher weight coefficient to the second priority sufficiency feature value ensures that the priority sufficiency characterization value of queue-jumping tasks more sensitively reflects the real-time idle level of the current testing equipment. This allows for prioritizing times or tasks with larger equipment redundancy in queue-jumping decisions, minimizing the disturbance of queue-jumping operations to the original schedule, and avoiding delays in high-priority tasks or overall schedule failure due to equipment overload. The value can be selected between 0.6 and 0.7; in this embodiment, it is set to 0.65. Correspondingly, the second drift anomaly feature value is selected between 0.3 and 0.4; in this embodiment, it is set to 0.35. Of course, the specific value can be dynamically adjusted based on the actual equipment utilization rate of the testing laboratory, the historical success rate of queue jumping, or the cost of task switching, which will not be elaborated here.

[0072] In this embodiment, the purpose of setting thresholds for the number of sample types and the amount of equipment redundancy is to screen flexible tasks that can balance resource utilization and task complexity. This is achieved by acquiring test status data corresponding to several different test items completed on the same equipment, and calling up data on the number of sample types and the amount of equipment redundancy corresponding to several tasks that can be queued and meet task scheduling requirements. The mean number of sample types and the mean amount of equipment redundancy are calculated. Based on the purpose of setting these two thresholds, the threshold for the number of sample types is determined as the product of the mean number of sample types and the type deviation coefficient, and the threshold for the amount of equipment redundancy is determined as the product of the mean amount of equipment redundancy and the redundancy deviation coefficient. The type deviation coefficient is selected within the interval [1.2, 1.4] to enhance the priority of simple tasks, and is preferably 1.3 in practice. The redundancy deviation coefficient is selected within the interval [0.8, 0.9] to enhance the priority of tasks with high resource release, and is preferably 0.85 in practice.

[0073] The selected optimal values ​​take into account both preferences, ensuring that the sorting results prioritize tasks that are both simple and free up a large number of devices, thereby improving device turnover and scheduling efficiency. Of course, those skilled in the art can adjust the value range according to actual scheduling needs, or adjust the correlation coefficient within a given range. For example, to favor simpler tasks, the category bias coefficient can be increased; to favor highly redundant tasks, the redundancy bias coefficient can be decreased.

[0074] Specifically, this invention considers that in resource-constrained reliability testing scenarios, fewer types and numbers of test samples generally mean a simpler test object, fewer fixture or tooling changes, a simpler test process, and a relatively shorter execution time. Prioritizing these tasks allows them to complete quickly and release occupied equipment, such as environmental test chambers and testing cavities, thereby shortening the time the task occupies critical resources and improving equipment turnover. Furthermore, the amount of equipment redundancy reflects the amount of equipment resources currently occupied by a task, such as multiple test chambers or cavities connected in parallel. A larger redundancy means that once the task is completed, more equipment units can be released, accommodating more new tasks or alleviating resource constraints. Prioritizing these tasks can quickly release a large amount of equipment resources, increase the system's idle resource pool, facilitate the parallel execution of subsequent tasks or the insertion of urgent tasks, and reduce scheduling congestion. Based on this, this invention relies on the priority sufficiency representation value of queueable tasks to characterize their priority scheduling value in terms of both time efficiency and resource release benefits. It then performs a two-dimensional ranking to achieve a balance between resource utilization and task complexity, prioritizing lightweight tasks and improving equipment turnaround time. Furthermore, this invention enables rapid release of cavity capacity and equipment resources, providing available resource windows for subsequent tasks and enhancing parallel experimentation capabilities. By rationally ranking tasks, it reduces task waiting time, improves overall scheduling efficiency, and shortens the overall experimental cycle.

[0075] If the reliability test scheduling optimization method under the resource constraints of the testing equipment of the present invention is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media that can store program code, such as USB flash drives, mobile hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0076] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A method for optimizing reliability test scheduling under resource constraints of testing equipment, characterized in that, include: Call the set of test tasks to be scheduled, obtain the task data of several test items in the set of test tasks, and generate an initial schedule by reverse scheduling based on the latest start time of each test item. Extract the resource difference characteristics and the compactness of the latest start time for each of the aforementioned test projects, and analyze the scheduling tightness characterization value of the test task set to label the test task set; Based on the labeling results, the scheduling of several test items corresponding to the test task set is optimized, including, Determine whether an impactful event exists based on the test output data from the primary task; The experimental state data of the influencing event is obtained, the parameter drift curve is constructed, the parameter drift direction is locked, the cumulative drift of the drift parameter in the parameter drift direction and the uniformity of the drift amplitude in a unit time domain segment are identified, and the drift anomaly characterization value of the influencing event is analyzed to determine whether to start the task pre-rearrangement mechanism in advance. The task pre-reordering mechanism includes: extracting the earliest available release time node corresponding to the first task, placing a preset transformation time domain segment into the earliest available release time, filtering tasks that can be inserted into the queue, and determining whether the tasks that can be inserted into the queue meet the task scheduling requirements. Based on the test state characteristics of several interleaved tasks that meet the task scheduling requirements, the interleaved tasks are sorted to generate a new schedule. Output the initial schedule and the new schedule, and generate a Gantt chart containing the task timing and device allocation relationship; The resource difference characteristics include the variation in the release response time of each environmental parameter and the difference in environmental parameters between adjacent tasks. The test state characteristics include the number of sample types and the amount of equipment redundancy occupied.

2. The reliability test scheduling optimization method under resource constraints of testing equipment according to claim 1, characterized in that, The process of analyzing the scheduling tightness representation values ​​of the aforementioned set of experimental tasks includes: The sum of the ratio of the difference in release response time of each environmental parameter to the difference threshold and the ratio of the difference in environmental parameters between adjacent tasks to the difference threshold is used as the first scheduling tight feature value. The ratio of the latest start time compactness to the compactness threshold is used as the second scheduling compactness feature value; The first scheduling tightness feature value and the second scheduling tightness feature value are weighted and summed to determine the scheduling tightness characterization value.

3. The reliability test scheduling optimization method under resource constraints of testing equipment according to claim 2, characterized in that, The set of test tasks is tagged, including: If the scheduling tightness characteristic value of the test task set is greater than or equal to the scheduling tightness characteristic threshold, then the test task set is marked.

4. The reliability test scheduling optimization method under resource constraints of testing equipment according to claim 3, characterized in that, The scheduling of several test items corresponding to the test task set was optimized, including: If any set of test tasks is marked, then the scheduling of several test items corresponding to the set of test tasks is optimized.

5. The reliability test scheduling optimization method under resource constraints of testing equipment according to claim 1, characterized in that, Determine if any impactful events exist, including: If the output uniformity of the experimental output data is less than the output uniformity threshold, and the total duration of the abnormal output data is greater than the total duration threshold, then an influencing event is determined to exist.

6. The reliability test scheduling optimization method under resource constraints of testing equipment according to claim 1, characterized in that, The process of analyzing the drift anomaly characterization values ​​of the aforementioned influencing events includes: The ratio of the cumulative drift amount of the drift parameter in the drift direction to the drift accumulation threshold is used as the first drift anomaly feature value; The ratio of the drift amplitude uniformity threshold to the drift amplitude uniformity within a unit time domain segment is used as the second drift anomaly feature value. The first drift anomaly feature value and the second drift anomaly feature value are weighted and summed to determine the drift anomaly characterization value.

7. The reliability test scheduling optimization method under resource constraints of testing equipment according to claim 6, characterized in that, Determining whether to initiate the task pre-rescheduling mechanism in advance includes: If the drift anomaly representation value affecting the event is greater than or equal to the drift anomaly representation threshold, then the task pre-reordering mechanism is activated in advance.

8. The reliability test scheduling optimization method under resource constraints of testing equipment according to claim 1, characterized in that, Filter tasks that can be skipped, including: If any task meets the scheduling optimization conditions, then the task is identified as the queue-jumping task. The scheduling optimization conditions include that the total required cavity capacity is less than the release cavity capacity, and the number of occupied devices is less than the number of devices that can be released.

9. The reliability test scheduling optimization method under resource constraints of testing equipment according to claim 1, characterized in that, Determining whether the queue-eligible task meets the task scheduling requirements includes: If the time leeway between the mandatory start time and the latest start time of the queueable task is greater than the time leeway threshold, and the time interval between the completion release time of the queueable task and the mandatory start time of the original subsequent task is greater than the time interval threshold, then the queueable task is determined to meet the task scheduling requirements.

10. The reliability test scheduling optimization method under resource constraints of testing equipment according to claim 1, characterized in that, The process of sorting the aforementioned interleaved tasks and generating a new schedule includes: The ratio of the number of sample types to the threshold number of sample types is used as the first priority sufficiency feature value; The ratio of the redundancy threshold of occupied equipment to the redundancy of occupied equipment is used as the second priority sufficiency feature value; The first priority sufficiency feature value and the second priority sufficiency feature value are weighted and summed to determine the priority sufficiency representation value of the queueable task. Based on the priority abundance representation value, each eligible task is sorted in ascending order to generate a new schedule.

Images