Systems and methods for dynamic signal processing for vacuum decay leak detection

Abstract

The leak detection apparatus includes a test chamber and a vacuum source connected to the test chamber via a pneumatic line. A first pressure transducer is in communication with the test chamber. A controller is configured to collect and analyze data generated by the first pressure transducer at a predetermined frequency throughout a test cycle. The controller is also configured to convert the designated collected data to a logarithmic scale to accumulate a first set of data, apply a linear regression to the first set of data, and determine regression parameters.

Description

[Technical Field]

[0001] TECHNICAL FIELD The present disclosure relates to systems and methods for leak detection. [Background technology]

[0002] Package leak testing often uses a vacuum decay test method in which a vacuum is drawn on the package within a test chamber. The vacuum level within the test chamber is measured to determine the presence or absence of a leak in the package. A pressure increase at a specified time above a predetermined pass / fail limit, established using, for example, a negative control, indicates a leak. Accurate measurement and data interpretation are critical to achieving accurate test results. However, improving the signal-to-noise ratio of pressure measurements is a challenge. Summary of the Invention [Means for solving the problem]

[0003] In a given example, the leak detection apparatus includes a test chamber and a vacuum source connected to the test chamber via a pneumatic line. The leak detection apparatus includes a first pressure transducer in communication with the test chamber. The leak detection apparatus also includes a controller. The controller is configured to collect and analyze data generated by the first pressure transducer at a predetermined frequency throughout a test cycle. The controller is also configured to convert the designated collected data to a logarithmic scale to accumulate a first set of data, apply a linear regression to the first set of data, and determine regression parameters.

[0004] The controller may be configured to perform a vacuum decay leak test using a leak detection system including the sensor. The controller includes a data acquisition module configured to collect data generated by the sensor. The controller includes a statistical signal processing module configured to convert the collected data designated to a logarithmic scale to accumulate a first set of data and to determine linear regression parameters for the data based on the first set of data.

[0005] A method for performing a vacuum decay leak test using a leak detection system including a pressure transducer. The method includes initiating a vacuum decay leak test and setting a predetermined evacuation period and a predetermined steady-state period. The method includes collecting data from the pressure transducer at a predetermined frequency throughout the test cycle. The method further includes performing statistical signal processing on the collected data and dynamically updating test conditions. [Brief explanation of the drawings]

[0006] [Figure 1] 1 is a diagram of an exemplary vacuum decay leak detection system. [Figure 2] FIG. 2 is a diagram of an example controller of the vacuum decay leak detection of FIG. 1. [Figure 3] 2 illustrates an exemplary statistical analysis performed by the vacuum decay leak detection system of FIG. 1. [Figure 4] 2 illustrates an exemplary method for vacuum decay leak detection using the vacuum decay leak detection system of FIG. 1. DETAILED DESCRIPTION OF THE INVENTION

[0007] The devices shown schematically in the drawings may have components that are examples of elements recited in the system claims and may operate in steps that are examples of elements recited in the method claims. Functionally equivalent components and steps within the scope of the present disclosure, in addition to those described herein, will be apparent to those skilled in the art from the description that follows. Such modifications and variations are intended to fall within the scope of the claims.

[0008] The present disclosure is directed to systems and methods for testing packages or containers, such as non-porous vials, ampoules, injection cartridges, ophthalmic product packages, syringes, pouches, blister packages, and other packages containing pharmaceutical and / or chemical products. More particularly, the present disclosure is directed to systems and methods for vacuum or pressure decay leak testing, intended to reduce test cycle time and improve signal-to-noise ratios and determination confidence levels. The entire leak test cycle includes both evacuation and steady-state periods. During both the evacuation and steady-state periods, system variables are continuously monitored to observe dynamic changes occurring within the test chamber. The monitored dynamic variables, including at least the test chamber pressure, are measured as quickly and accurately as possible and recorded along with the time when each measurement is performed. Other dynamically changing variables may also be monitored, including, but not limited to, gas temperature, concentrations of different components in the gas mixture, humidity, test chamber volume, and others. During both the evacuation and steady-state periods, the leak detection test system described herein is configured to perform calculations to determine the degree to which the experimental data matches one or several expected, suitable analytical functions. The expected functions for either the evacuation or steady-state periods correspond to the corresponding pressure curves observed during testing of the negative control sample. The analytical functions include at least static equation parameters and dynamic equation parameters. The leak detection system is also capable of storing the measured parameters and identifying a particular set of stored parameters as a calibrated vacuum decay leak detection response curve (e.g., pressure profile as a function of time). At least one known non-leaking sample (e.g., a negative control sample) is used to collect the calibration curve; however, a calibration curve may also be obtained from statistical calculations spanning many calibration tests of negative samples. For example, a single calibration test on a negative control sample or a set of calibration tests on one or more negative control samples may be performed to generate a calibration curve.An estimate of the statistical variance or statistical standard deviation of each measurement and / or an estimate of the statistical variance of each equation parameter for the analytical function may be recorded and used to dynamically determine the performance and / or results of the leak detection test. Additionally, the standard deviation and average values ​​for the parameters of the set of calibration values ​​may be used to determine the results of the leak detection test.

[0009] Once the calibrated response curves (from both the evacuation and steady-state periods) and associated parameters are established, testing of unknown samples is performed using the same calculations. Deviation of a new measured quantity from the calibrated response curve can be evaluated based on the standard deviation of the calibrated statistical residuals or based on the standard deviation of the parameter values ​​for a set of calibration readings. Statistical hypotheses are used to determine whether to terminate or continue the leak detection test. For example, based on the statistical hypotheses, if the deviation between the measured quantity and the corresponding calibrated response curve is greater than a predetermined value, the test system may determine to terminate the test and fail the sample. Statistical hypotheses can also be used to dynamically determine and update test conditions.

[0010] Once the leak detection test is completed, the measured response curve parameters are calculated and compared to the calibrated response curve parameters (from both the evacuation and steady-state periods). Differences are assessed based on the standard deviation of the calibrated parameters to determine the pass / fail status of the test sample. Significant differences between the static equation parameters can be used to screen for non-standard test conditions. The presence of non-standard test conditions can affect the judgment confidence and can be used as a warning of an increased probability of a false negative or false positive result.

[0011] The systems and methods disclosed herein for vacuum or pressure decay leak testing have the ability to characterize the pressure curve over time and interpret pressure measurements in the logarithmic domain, thereby enabling accurate processing of dynamic pressure changes (as opposed to measured pressure at a point in time). The systems and methods disclosed herein use a mathematical model derived from the ideal gas law and the relationship between mass flow from a venting chamber to analyze dynamic pressure behavior. Equations similar to those derived in compressible flow texts (e.g., isothermal and adiabatic gas expansion) are used to develop an expected function for the pressure over time model. The systems and methods disclosed herein rapidly measure pressure during the decay or pressurization of the test chamber and map the behavior to the expected model.

[0012] Because the ideal gas law defines pressure as proportional to volume, temperature, and the amount (mass) of a substance, all variables are therefore distributed log-normally rather than normally. The systems and methods disclosed herein therefore use log-normal domain variables to perform the analysis. This allows for a true normal distribution analysis for linear regression of test data. The systems and methods disclosed herein provide a new model for the dynamic behavior of pressure decay curves, capable of achieving increased statistical accuracy in distinguishing test results.

[0013] FIG. 1 shows an example testing system (e.g., a vacuum decay leak detection system) 100 configured to inspect a package or container 102. In the illustrated example, the package 102 is a non-porous container, such as a vial, pouch, blister pack, ampoule, syringe, injection cartridge, or ophthalmic package, containing a chemical and / or pharmaceutical product. The testing system 100 includes a vacuum supply or source 104 and a test chamber 106 connected by an air pressure line 108. A selector valve 110 is disposed in communication with the air pressure line 108 between the vacuum supply 104 and a chamber valve 116. A compressed air supply 112 (e.g., clean, dry compressed air) for venting and an ambient air exhaust 114 communicate with the air pressure line 108 at the selector valve 110. The chamber valve 116 is disposed in communication with the pneumatic line 108 between the selector valve 110 and the test chamber 106. The selector valve 110 is configured to vary the source supplied to the pneumatic line 108 connected to the chamber valve 116. For example, the selector valve 110 can be used to selectively supply vacuum (from the vacuum supply 104), compressed air (from the compressed air supply 112), or ambient air (from the ambient air 114) to the pneumatic line 108 to the chamber valve 116. In embodiments in which the compressed air supply 112 is omitted, the ambient air 114 is used for venting.

[0014] A bypass or bypass pneumatic line 118 communicates with the pneumatic line 108 via a bypass valve 120 that is placed in communication with the bypass 118. A precision restriction orifice 121 is placed in communication with the pneumatic line 108 between the chamber valve 116 and the test chamber 106 or between the chamber valve 116 and the bypass pneumatic line 118. The precision restriction orifice 121 is configured to provide more consistent flow characteristics, which may increase the accuracy or ability of the test system to more strongly correspond to the results of another similarly constructed test system.

[0015] The test system 100 includes sensors 122 including a first pressure transducer 124, a differential pressure transducer 126, a second pressure transducer 128, and a weather sensor 130. The first pressure transducer 124 is configured to measure a pressure corresponding to the pressure inside the test chamber 106. For example, the first pressure transducer 124 may be an absolute pressure transducer or a gauge pressure transducer. The first pressure transducer 124 is disposed in communication with the test chamber 106. The first pressure transducer 124 may be in communication with the air pressure line 108 between the chamber valve 116 and the test chamber 106. Alternatively, the first pressure transducer 124 may be directly connected to the test chamber 106 instead of connecting to the test chamber 106 through the air pressure line 108. A differential pressure transducer 126 is disposed in communication with the bypass 118, and the bypass valve 120 is intermediate the chamber valve 116 and the differential pressure transducer 126. A weather sensor 130 is disposed in communication with the ambient air outlet 114 to measure the pressure, temperature, humidity, and / or gas composition at the ambient air outlet 114. The weather sensor 130 may include a pressure sensor (e.g., a barometric pressure sensor), a temperature sensor, a humidity sensor, a gas sensor (e.g., a water vapor sensor, an oxygen sensor, a carbon dioxide sensor, a helium or other traceable gas analyzer), a mass flow meter, or a combination thereof.

[0016] The incorporation of sensors 122 within test system 100 includes elements essential to leak detection, such as first pressure transducer 124, and may include other elements that increase the accuracy, repeatability, and / or reproducibility of test results. As one example, differential pressure transducer 126 has the ability to handle pressure spans 10 to 100 times smaller, thereby increasing measurement resolution. This effectively reduces electromagnetic interference noise and digitization errors.

[0017] As another example, if the vacuum supply is not steady, which can occur when several unsynchronized elements of the test system compete for the same vacuum source or compress air, obtaining good repeatability for successive leak test results can be a challenge. The incorporation of the second pressure transducer 128 allows for direct measurement of the pressure at the vacuum supply 104. This information is valuable for improving the repeatability of test results. For example, the second pressure transducer 128 can be an absolute pressure transducer or a gauge pressure transducer.

[0018] As another example, leak tests performed in different ambient environments can affect the repeatability of test results. The incorporation of weather sensors 130 provides the opportunity to obtain direct measurements of pressure, temperature, humidity, and other quantities at the ambient air outlet 114. This information is valuable for improving the repeatability of test results.

[0019] Test system 100 includes a controller or control system 132 configured to operate test system 100 and coordinate the operation of test system 100. Controller 132 may be a computer or may include any suitable processor, microprocessor, transceiver, memory, timer, analog-to-digital converter (ADC), programmable logic controller (PLC), human machine interface (HMI), etc. to enable its functions. Controller 132 may include any suitable user interface and / or display to enable output of test results and to allow a user to program or control the operation of test system 100. Controller 132 may perform leak tests according to pre-programmed procedures and / or may perform dynamic analysis to update test procedures in situ. In particular, controller 132 includes specialized modules to perform dynamic signal processing and updating the operation of test system 100.

[0020] In alternative embodiments, one or more components of test system 100 may be omitted. For example, precision restrictive orifice 121 may be omitted. For example, test system 100 may omit bypass 118, bypass valve 120, differential pressure transducer 126, and second pressure transducer 128. In this case, test system 100 may perform a leak test in fundamental mode based on the pressure signal from first pressure transducer 124.

[0021] For example, the test system 100 may omit the second pressure transducer 128. In this case, the test system 100 may perform a leak test in a dual sensor mode based on the pressure signals from the first pressure transducer 124 and the differential pressure transducer 126.

[0022] For example, test system 100 may omit bypass 118, bypass valve 120, and differential pressure transducer 126, and test system 100 may perform leak testing in a repeatability-improved mode based on pressure signals from first pressure transducer 124 and second pressure transducer 128. In one embodiment, test system 100 may omit bypass 118, bypass valve 120, and differential pressure transducer 126, and test system 100 may perform leak testing in a repeatability-improved mode based on pressure signals from first pressure transducer 124 and second pressure transducer 128, as well as signals from weather sensor 130 (e.g., signals corresponding to pressure, temperature, humidity, and / or gas composition at ambient air outlet 114). In one embodiment, the test system 100 may perform leak testing under a combination of two or more of the test modes described above (e.g., basic mode, dual sensor mode, improved repeatability mode, improved reproducibility mode).

[0023] 2 shows an exemplary controller 132 configured to perform dynamic signal processing for a vacuum decay leak detection test. In the illustrated embodiment, the controller 132 includes an automatic control module 200, a data acquisition module 202, a statistical signal processing module 204, a statistical hypothesis evaluation module or null hypothesis module 206, and a memory 208. The automatic control module 200 is capable of sending signals to control the operation of various components of the test system 100.

[0024] The data acquisition module 202 is capable of receiving data or signals from the sensor 122. Specifically, the data acquisition module 202 collects data during the entire test cycle (e.g., both the vent period and the steady-state period), as opposed to collecting data only at certain points in time (e.g., at the beginning of a test cycle, at the end of a test cycle, at venting, etc.). For example, the data acquisition module 202 collects the entire signal from the first pressure transducer 124, the differential pressure transducer 126, or both, throughout the entire test cycle, including the vent period and the steady-state period. The first pressure transducer 124 and the differential pressure transducer 126 are designed to sample data at a certain frequency. All data sampled at the designed frequency throughout the entire test cycle, including the vent period and the steady-state period, is collected by the data acquisition module 202. The collected signals are transmitted to a statistical signal processing module 204, and the processed signals are transmitted to a null hypothesis module 206, which determines whether the test should continue or be terminated, the test result (e.g., pass / fail, confidence level, etc.), and / or suitable test conditions (e.g., basic mode, dual sensor mode, repeatability-enhanced mode, reproducibility-enhanced mode, adjusted test conditions, or a combination thereof). Preferably, the response times of the valves (selector valve 110, chamber valve 116, and bypass valve 120), the transducers (e.g., first pressure transducer 124, differential pressure transducer 126, and second pressure transducer 128), and the data acquisition sampling are each at least two orders of magnitude faster than the evacuation time.

[0025] The controller 132 is capable of performing dynamic signal processing and dynamically updating the operation of the test system 100. In a typical vacuum decay leak detection test, pressure exhibits exponential changes that indicate gas flow, while fluctuations in pressure readings are noise. These noises may be caused by variables unrelated to the leak detection test and may reduce the accuracy of the test results. The statistical processing module 204 uses statistical techniques to filter noise from pressure readings that are proportional to gas flow. Taking advantage of the fact that pressure readings can be approximated according to a power law or equation, or an exponential model or equation, taking the natural logarithm of the power or exponential equation that approximates the pressure reading allows for differentiation between noise, semi-static variables, and variables that cause proportional dynamic pressure changes. The statistical processing module 204 may even map dynamic variables to more than one expected pressure curve. In such cases, the natural logarithm of any given model curve should result in a normally distributed system.

[0026] In particular, the pressure reading is expected to follow a known step response function, which can be approximated as an exponential function or other similar function. Equation (1) shows the exponential function. P(t)=Ae tb ξ Equation (1) where p(t) is the pressure as a function of time, A is a coefficient reflecting a semi-static variable, b is a coefficient reflecting a variable that causes a proportional dynamic pressure change, and ξ is the noise.

[0027] The natural logarithm of equation (1) gives equation (2). ln(P(t))=ln(A)+bt+ln(ξ) Equation (2) Substituting y=ln(P), α=ln(A), x=t, β=b, and ε=ln(ξ) gives equation (3). y=α+βx+ε i Formula (3)

[0028] Equation (3) is therefore a regression of normally distributed pressure readings. Parameter α is related to A and is associated with static or quasi-static components, such as the starting pressure conditions in the vacuum supply 104 and the air pressure in the chamber being vented (e.g., compressed air supply 112). Parameter β is equivalent to b and is associated with components that affect the dynamic behavior of the system, such as the air volume in the test chamber 106 and orifices (e.g., leaks) in the package 102. Parameter ε i is a noise or error term. A single parameter or a mathematical combination of parameters may be used for leak test cycle evaluation.

[0029] As another example, a step response function can be approximated by a power function: Equation (4) shows the power function: P(t)=At b ξ Equation (4)

[0030] n data pairs {(x i ,y i ), i=1,...,n)} is generated during the leak test cycle, the noise or error ε i Contains the term, y i and x i The underlying relationship between x and x can be estimated using a linear regression model. i ,y i , n)}. The solutions for the crossover, slope, standard deviation of the crossover, standard deviation of the slope, and standard deviation of the residual are listed in equations (5)-(9), respectively. a=(Σy-bΣx) / n Equation (5) b=(nΣxy-ΣxΣy) / (nΣx 2 -(Σx) 2 ) Formula (6)

number

number

number

[0031] The data, calibrated regression parameters a and b, and the calibrated response function or curve (y = a + b) from the negative control sample are stored in memory 208. In some embodiments, the calibration data (from the negative control sample) may exclude pressure data during the initial stages of evacuation. In one example, modeling or calculation of the calibrated response curve only occurs after the pressure reaches a predetermined vacuum threshold, e.g., about 800 mbar, and pressure data before this point is ignored. The calibration data (e.g., the calibrated regression parameters a and b, and S a , S b Once the statistical values ​​(such as the mean value, mean time, or other meaningful statistical values) are established, the controller 132 uses statistical methods to determine whether to continue or terminate the test, to determine the test period results, and / or to determine suitable test conditions.

[0032] 3 illustrates an exemplary statistical analysis 300 performed by the statistical signal processing module 204 and the null hypothesis module 206. The data discussed herein may be any data from the sensor 122. For purposes of discussion, pressure data collected over time (e.g., pressure data as a function of time) is used to illustrate the statistical analysis 300. The statistical analysis 300 enables the determination of a test result (e.g., pass / fail) under a basic mode based on a null hypothesis evaluation, and further the selection of a suitable test mode (e.g., basic mode, dual sensor mode, improved repeatability mode, adjusted test conditions, and improved reproducibility mode). The statistical analysis 300 may be performed on the exhaust and steady-state periods of a test cycle.

[0033] In cases where statistical analysis 300 is performed to determine a test result (e.g., pass / fail) under basic mode, statistical analysis 300 includes receiving basic mode test data (step 302) and converting designated collected data to a logarithmic scale to accumulate a first set of data (step 304). The designated collected data may include all or a portion of the collected data. For example, if the response curve (pressure readings as a function of time) is to be approximated by an exponential function (e.g., Equation (1)), then only the pressure values ​​are converted to a logarithmic scale in step 304, while the time values ​​remain on a normal scale. Alternatively, if the response curve is to be approximated by a power function (e.g., Equation (4)), then both the pressure and time values ​​are converted to a logarithmic scale in step 304.

[0034] The statistical analysis 300 includes applying a linear regression to a first set of data (step 306) to determine regression parameters a1 and b1. The first set of data is expected to generally follow the linear relationship of equation (3). In the event that a different step response function is used, the data would be expected to follow the appropriate relationship associated with a log-domain analysis of this alternative step function.

[0035] Statistical analysis 300 includes storing calibration data (step 308). If the sample being tested is a negative control sample, the response curve, the first set of data accumulated in step 304, and / or data including regression parameters a1 and b1 are stored as calibration data (for at least one of the exhaust period and the stead state period). For example, regression parameters a1 and b1 from the exhaust curve response of the negative control sample are stored in memory 208 as calibrated regression parameters or baseline parameters a0 and b0 for the exhaust period. The exhaust response curve of the negative control sample is further stored as calibration data for determining a suitable exhaust time or exhaust time cutoff for the test sample. The steady-state response curve of the negative control sample may also be stored as calibration data for determining a suitable steady-state period for the test sample. The calibration data may include data from a single test run, for example, the standard deviation of the residuals is used, or the calibration data may include a set of data, for example, a set of calibration tests is performed and the standard deviation and averaged a0 and b0 are determined.

[0036] The statistical analysis 300 includes performing a null hypothesis evaluation (step 310). The null hypothesis evaluation is set up or configured to determine whether a significant statistical difference exists between a leak test performed on a test sample and a leak test performed on a reference sample, such as a negative control sample or a known sample, e.g., a sample tested under basic mode. The null hypothesis evaluation is performed by the null hypothesis module 206 of the controller 132. The null hypothesis module 206 may use a two-sample t-test to test whether the mean of the test data for the reference sample and the mean of the test data for the test sample are equal. The null hypothesis evaluation may be based on a comparison of individual measurands, static parameters, or dynamic responses, or any suitable combination or transformation of these parameters, of the test and reference samples.

[0037] In step 310, a null hypothesis evaluation can be set up to decide to continue or terminate the test based on a comparison between the analyzed test data (data obtained at the end of step 306) and the calibration data (stored in step 308 during the calibration test). If there is no significant difference between the measured value y1(x) of the test sample and the measured value y0=a0+b0x of the calibrated response function, the hypothesis is accepted. The null hypothesis module 206 determines the t-value T based on equation (10): e Calculate and T e may be compared to the T-statistic for a given number n of measurands on the response curve and n-2 degrees of freedom. T e =(y1(x)-a0-b0x) / S e Formula (10)

[0038] If the null hypothesis is accepted based on equation (10), the leak test continues and updates the data in step 312. As the leak test continues, test data is received and processed in the same manner as discussed in steps 302 through 310, so that the data updated in step 312 includes data from the entire test cycle.

[0039] If the null hypothesis is rejected, the test is terminated in step 314, which further indicates that the test sample fails (eg, a leak is present in the sample).

[0040] The null hypothesis in step 310 can be performed at any time during the test cycle before its completion. For example, the null hypothesis in step 310 to determine whether to continue or terminate the test can be performed when the evacuation period expires. In other examples, the null hypothesis in step 310 can be performed early (e.g., the first third) of the evacuation period, mid- (e.g., the middle third) of the evacuation period, late (e.g., the late third) of the evacuation period, early (e.g., the first third) of the steady-state period, mid- (e.g., the middle third) of the steady-state period, or late (e.g., the late third) of the steady-state period, and before the end of the test cycle. The ability to determine early termination of a leak test when the sample response deviates significantly from the response of the negative control sample can save time and reduce the possibility of chamber contamination from leaks.

[0041] The statistical analysis 300 includes performing a null hypothesis evaluation (step 316) and determining test results under basic mode for a specified test period (step 318). The specified test period may include a vent period, a steady-state period, or both. In steps 316 and 318, the null hypothesis module 206 may determine a pass or fail status for the test sample based on a null hypothesis evaluation of a comparison between the data collected and analyzed at the end of the test cycle and the calibration data from the negative control sample. If a significant statistical difference exists between the dynamic responses of the test sample and the negative control sample, the null hypothesis module 206 may determine a fail status for the test sample; otherwise, the test sample passes.

[0042] In an embodiment where the response curve is approximated by an exponential function (e.g., Equation (1)), the null hypothesis module 206 calculates the t-value T based on Equation (11): b Calculate and T b may be compared to the T-statistic for a given number n of measurands on the response curve and n-2 degrees of freedom. T b =(b1-b0) / S b Formula (11) In an embodiment in which the response curve is approximated by a power law function (e.g., Equation (4)), the null hypothesis module 206 may calculate a t-value T based on Equation (12) and compare T to the T-statistic for a given number n of measurands on the response curve and n-2 degrees of freedom. T=[(a1-a0)+(b1-b0)ln(t)] / S b Formula (12) If the hypothesis is rejected, the test sample fails (e.g., the presence of a leak). If the null hypothesis is accepted, the test sample passes (e.g., the absence of a leak).

[0043] The test results from step 318 and the statistical values ​​or parameters determined by statistical analysis 300 (e.g., a0 and b0, a1 and b1, t values, and S a , Sb , etc.) may be stored in memory 208 as calibration data (eg, calibration data for tests performed under basic mode).

[0044] Alternatively, steps 310, 312, and 314 may be omitted. In this case, statistical analysis 300 includes performing a leak test under basic mode in step 302 and receiving test data when a test cycle is completed. The specified collected data is processed / analyzed according to steps 304 and 306. Statistical comparisons between a1 and a0, and between b1 and b0 for both the steady-state and evacuation periods are performed to determine whether significant statistical differences exist to determine the test outcome (e.g., pass / fail) of the sample tested under basic mode (e.g., steps 316 and 318).

[0045] In embodiments in which statistical analysis 300 is performed to determine a suitable test mode, statistical analysis 300 includes receiving non-fundamental mode test data (e.g., adjusted test conditions, dual sensor mode, improved repeatability mode, and improved reproducibility mode) in step 303. The analysis then proceeds in the same manner as discussed in steps 304 through 316.

[0046] In step 319, the null hypothesis module 206 determines whether the test sample tested under the non-basic mode passes or fails based on a null hypothesis evaluation of the comparison between the data collected and analyzed at the end of the test cycle and the calibration data from the negative control sample. This evaluation may be performed in the same manner as the evaluations discussed in steps 316 and 318. If there is a significant statistical difference between the dynamic responses of the test sample and the negative control sample, the null hypothesis module 206 determines a fail status for the test sample; otherwise, the test sample passes.

[0047] Steps 303, 304-316, and 319 may be repeated for tests performed under each mode (e.g., adjusted test conditions, dual sensor mode, improved repeatability mode, improved reproducibility mode). Results from each test mode or condition may be stored as calibration data in memory 208.

[0048] In step 320, the null hypothesis module 206 compares the test results of different modes (e.g., basic mode, adjusted test conditions, dual sensor mode, repeatability-improved mode, and reproducibility-improved mode) to determine a suitable test mode for the test sample based on the null hypothesis evaluation. The null hypothesis module 206 may calculate and compare t-values ​​to determine a more suitable test mode.

[0049] For example, the null hypothesis module 206 may calculate a t-value T for a test performed under the adjusted test conditions based on equation (13): a Calculate and T a may be compared to the T-statistic tested under the fundamental mode for a given number n of measurands on the response curve and n-2 degrees of freedom. T a =(a1-a0) / S a Formula (13) If the null hypothesis is accepted based on equation (13), the null hypothesis module 206 determines the base mode as the preferred mode. If the null hypothesis is rejected, the null hypothesis module 206 determines that adjusting the test conditions is more suitable.

[0050] For example, the null hypothesis module 206 may determine the t-value T based on equation (11) for tests performed under dual sensor mode. b The null hypothesis module 206 may calculate T bmay be compared to the T-statistic obtained under the fundamental mode for a given number n of measurands on the response curve and n-2 degrees of freedom. If the null hypothesis is accepted, the null hypothesis module 206 determines the fundamental mode as the preferred mode. If the null hypothesis is rejected, the null hypothesis module 206 determines the dual-sensor mode as the preferred mode.

[0051] For example, the null hypothesis module 206 may calculate T for tests performed under the improved repeatability mode. a Calculate the T a may be compared with the T-statistic obtained under the basic mode. If the null hypothesis is accepted, the null hypothesis module 206 determines the basic mode as the preferred mode. If the null hypothesis is rejected, the null hypothesis module 206 determines the mode with improved repeatability as the preferred mode.

[0052] For example, the null hypothesis module 206 may calculate T for tests performed under the improved reproducibility mode. a Calculate the T a may be compared with the T-statistic obtained under the basic mode. If the null hypothesis is accepted, the null hypothesis module 206 determines the basic mode as the preferred mode. If the null hypothesis is rejected, the null hypothesis module 206 determines the mode with improved reproducibility as the preferred mode.

[0053] For example, the null hypothesis module 206 may use a T (obtained for a selected test mode, e.g., adjusted test conditions, dual sensor mode, repeatability-improved mode, reproducibility-improved mode) to determine a suitable test mode for the test sample. a , T b , or both may be compared with the T-statistic obtained from the fundamental mode.

[0054] Furthermore, the null hypothesis module 206 may determine to upgrade the basic mode, the dual sensor mode, or the repeatability-improved mode to the reproducibility-improved mode with data from the weather sensors 130. Based on the null hypothesis evaluation, the null hypothesis module 206 may determine to upgrade the basic mode, the dual sensor mode, the repeatability-improved mode, the repeatability-improved mode, or any combination thereof with data from sensors 122 including mass flow sensors, displacement sensors, deformation or volume change sensors (e.g., using laser triangulation, machine vision, etc.), stress sensors, or combinations thereof.

[0055] 4 illustrates a method 400 for vacuum decay leak detection using test system 100. The steps discussed herein are controlled and executed by controller 132. Method 400 includes preparing the test system (step 402). In step 402, test chamber 106 is vented by ambient air exhaust 114, a test sample (e.g., package 102) is placed inside test chamber 106, and test chamber 106 is closed.

[0056] Method 400 includes starting a vacuum decay leak test and setting a predetermined pump-down period and a predetermined steady-state period (step 404). This predetermined time may be set by an automated process performed during statistical evaluation calibration and stored as calibration data (step 308 in FIG. 3). In step 404, automatic control module 200 sends a control signal to connect selector valve 110 to vacuum supply 104 and open chamber valve 116, thereby connecting vacuum supply 104 to test chamber 106 and beginning to pull a vacuum on test chamber 106. As soon as vacuum is being pulled on test chamber 106, a timer in test system 100 is triggered, marking the beginning of the pump-down period.

[0057] Method 400 includes performing data acquisition and statistical signal processing (step 406). In step 406, data acquisition module 202 receives readings or data from sensors 122. For example, data acquisition module 202 receives readings from first pressure transducer 124 and / or second pressure transducer 128 along with a time measurement. Statistical signal processing module 204 then converts some or all of this data to a logarithmic scale and stores the data for regression calculations, as previously discussed in step 304 of FIG. 3.

[0058] Method 400 includes determining whether to continue testing when a predetermined evacuation period expires (step 408). In step 408, automatic control module 200 sends control signals to close chamber valve 116 and to connect selector valve 110 to ambient air exhaust 114. In some embodiments, automatic control module 200 sends control signals to close bypass valve 120 (if testing is to proceed under dual sensor mode). Statistical signal processing module 204 completes statistical signal processing of data received from first pressure transducer 124 during the evacuation period, and null hypothesis module 206 determines whether to proceed to a steady state period. 3, if the null hypothesis is rejected, the null hypothesis module 206 makes a determination that the package 102 will fail the evacuation test period (e.g., a large leak is detected), and the method 400 proceeds to step 416 to terminate the test. In step 416, the automatic control module 200 sends a control signal to abort the leak test and initiate venting. If the null hypothesis is adopted, the null hypothesis module 206 determines to continue to the steady-state period. Step 408 allows for a shortened test cycle, substantially reducing the variability or spread of chamber contamination from leaking packages.

[0059] Alternatively, step 408 may be performed at any time before the entire test cycle is completed. For example, the decision to continue or abort the test may occur early in the pump-down period (e.g., the first third), in the middle of the pump-down period (e.g., the middle third), in the later part of the pump-down period (e.g., the later third), at the beginning of the steady-state period (e.g., the first third), in the middle of the steady-state period (e.g., the middle third), or in the later part of the steady-state period (e.g., the later third), and before the end of the test cycle.

[0060] Method 400 includes dynamically updating test conditions (step 410) for a predetermined steady-state period. As previously discussed in steps 316, 319, and 320 of FIG. 3 , null hypothesis module 206 may determine to continue the leak test under basic mode (with standard or adjusted conditions), dual-sensor mode, improved repeatability mode, improved reproducibility mode, or a combination thereof. Based on the determination, controller 132 updates the test conditions and collects data from the corresponding sensors 122. Under basic mode operation, step 410 may be omitted since the test conditions remain the same (standard conditions), or the test conditions are updated without updating in step 410.

[0061] The method includes performing data acquisition and statistical signal processing (step 412). In step 412, the data acquisition module 202 receives data from the sensor 122. During the entire test cycle, the data acquisition module 202 always receives data from the first pressure transducer 124 along with a time measurement. If the test continues under dual sensor mode, in addition to data from the first pressure transducer 124, the data acquisition module 202 also receives data from the differential pressure transducer 126 along with a time measurement. If the test continues under improved repeatability mode, in addition to data from the first pressure transducer 124, the data acquisition module 202 also receives data from the second pressure transducer 128 along with a time measurement. If testing continues under the repeatability-improved mode, in addition to data from the first pressure transducer 124, the data acquisition module 202 also receives data from the second pressure transducer 128, along with time measurements, and data from weather sensors 130 and / or other sensors in sensors 122, such as mass flow sensors, displacement sensors, deformation or volume change sensors (e.g., using laser triangulation, machine vision, etc.), stress sensors, or combinations thereof. In step 412, the statistical signal processing module 204 converts the specified data collected during the predetermined test period to a logarithmic scale and stores these data for regression calculations, as previously discussed in steps 304 and 306 of FIG. 3 .

[0062] Method 400 includes determining a test result, e.g., pass / fail, from the steady-state period when the predetermined steady-state period expires (step 414). In step 414, null hypothesis module 206 determines whether the test sample (e.g., package 102) passes or fails based on the regression response curve calculated from step 412 and the null hypothesis evaluation, as previously discussed in steps 316 and 318 or steps 316, 319, and 320 of FIG. 3. In addition to the pass / fail status, the test result may include a confidence level associated with the pass / fail status.

[0063] Finally, the method 400 includes terminating the test cycle and venting (step 416), where the automatic control module 200 sends a control signal to open the chamber valve 116 and connect the selector valve 110 to the compressed air supply 112 for venting.

[0064] The method 400 may include steps 402, 404, 406, 412, 414, and 416 to test a sample under fundamental mode (steps 408 and 410 are omitted), with the test chamber 106 closed between steps 406 and 412 to enable steady-state testing.

[0065] Those skilled in the art will appreciate that, for this and other processes and methods disclosed herein, the functions performed in the processes and methods may be implemented in differing order. Furthermore, the outlined steps and operations are provided merely as examples, and some of the steps and operations may optionally be combined into fewer steps and operations or expanded into additional steps and operations without detracting from the essence of the disclosed embodiments.

Claims

1. a test chamber; a vacuum source connected to the test chamber via a pneumatic line; at least one sensor including a first pressure transducer and a second pressure transducer; a controller, initiating a leak test of a test sample in a basic mode, wherein the leak test is performed based on a pressure signal from the first pressure transducer under the basic mode; collecting and analyzing data generated by the at least one sensor at a predetermined frequency throughout a test cycle; designating a portion of the collected data; converting the designated portion of the collected data to a logarithmic scale to accumulate a first set of data; applying a linear regression to said first set of data to determine a first set of regression parameters; determining whether to terminate or continue the leak test at or before a predetermined evacuation period has elapsed based on a statistical hypothesis evaluation comparing the first set of regression parameters to calibrated regression parameters determined based on at least one control sample; and responsive to a decision to continue the leak test, selecting a mode from options including the basic mode and an improved repeatability mode, collecting data under the selected mode, and determining a second set of regression parameters for the test sample; a controller configured to: A leak detection device comprising: The leak detection apparatus performs the leak test based on pressure signals from the first pressure transducer and the second pressure transducer under the repeatability-enhanced mode.

2. A leak detection device as described in claim 1, wherein the at least one sensor further includes a differential pressure transducer, the options further include a dual sensor mode, and under the dual sensor mode, the leak test is performed based on pressure signals from the first pressure transducer and the differential pressure transducer.

3. A leak detection device as described in claim 2, wherein the at least one sensor further includes a meteorological sensor, and the options further include an improved repeatability mode, and under the improved repeatability mode, the leak test is performed based on pressure signals from the first pressure transducer, the second pressure transducer and the meteorological sensor.

4. a selector valve connected to the pneumatic line for selectively isolating the vacuum source from the test chamber; a chamber valve connected to the pneumatic line between the selector valve and the test chamber; an ambient air exhaust connected to the air pressure line; The leak detection device of claim 1 further comprising:

5. 5. The leak detection apparatus of claim 4, wherein the selector valve has a first response time, the chamber valve has a second response time, the first pressure transducer has a third response time, and the controller has a fourth response time of data acquisition sampling, and each of the first, second, third, and fourth response times is at least two orders of magnitude faster than an evacuation time of the leak detection system.

6. The leak detection device of claim 2 further comprising a precision restrictive orifice connected to the pneumatic line.

7. a bypass air pressure line connected to the air pressure line between the chamber valve and the test chamber; a bypass valve connected to the bypass air pressure line; Furthermore, 3. The leak detection apparatus of claim 2, wherein the differential pressure transducer is connected to the bypass air pressure line, and the bypass valve is intermediate the differential pressure transducer and the test chamber.

8. The leak detection apparatus of claim 1 , wherein the second pressure transducer is configured to measure a pressure corresponding to the vacuum source.

9. The leak detection apparatus of claim 1 , wherein the statistical hypothesis evaluation comprises a null hypothesis evaluation based on a comparison of the dynamic responses of the test sample and the at least one control sample.

10. The leak detection device of claim 1 , wherein the controller is configured to determine a pass or fail result of the leak test based on a null hypothesis evaluation based on a comparison of the dynamic responses of the test sample and the at least one control sample.

11. 10. The leak detection apparatus of claim 1, wherein the selection of the mode is based on a null hypothesis evaluation based on a comparison of the dynamic responses of the test sample and the at least one control sample.

12. 2. The leak detection device of claim 1, wherein the collected data comprises pressure data and time data, and the controller is further configured to select either the pressure data alone or both the pressure data and the time data based on a selected function to form the designated portion of the collected data, the designated portion of the collected data consisting of the pressure data alone if the selected function is an exponential function, and consisting of both the pressure data and the time data if the selected function is a power function.

13. The leak detection apparatus of claim 1 , wherein the at least one sensor further comprises a pressure sensor, a temperature sensor, a humidity sensor, a gas sensor, a mass flow meter, a displacement sensor, a deformation or volume change sensor, a stress sensor, or a combination thereof.

Images