Trend Monitoring and Diagnostic Analysis Method for a Vacuum Pump and a Trend Monitoring and Diagnostic Analysis Method for Vacuum Pump and a Trend Monitoring and Diagnostic Analysis System Therefor and Computer-Readable Storage Media Including a Computer Program Which Performs the Method

• US 20080010030A1
• Filed: 12/17/2004
• Published: 01/10/2008
• Est. Priority Date: 12/17/2004
• Status: Active Grant

• Current Assignee: Korea Research Institute of Standards and Science
• Original Assignee: Korea Research Institute of Standards and Science
• Inventors: Chung, Kwang Hwa, Lim, Jong Yeon, Lee, Soo Gab, Cheung, Wan Sup
• Granted Patent: US 7,664,618 B2
• Time in Patent Office: Days
• US Class Current: 702/034
• CPC Class Codes: F04D 19/04 specially adapted to the production of a high vacuum, e.g. molecular pumps
• CPC Class Codes: F04D 27/001 Testing thereof; Determination or simulation of flow characteristics; Stall or surge detection, e.g. condition monitoring
• CPC Class Codes: G05B 23/0232 based on qualitative trend analysis, e.g. system evolution

Abstract

The present invention such as active diagnostic algorithms is developed not only to realize the early detection of degraded vacuum pumps for the protection of pump failure but also to provide their predictive maintenance. According to the present invention, it is possible to find simple and effective ways to deal with technical problems arising from the large variability of the pump-by-pump operation characteristics and the multiple process conditions where pumps run under the idle operation and gas-loaded operation conditions alternately, especially in semiconductor manufacturing process.

Claims

1. A trend monitoring and diagnostic analysis method for the failure protection and predictive maintenance of a vacuum pump running under idle and gas-loaded operation conditions alternately, comprising the steps of:

sampling the time series of the state variables signals at the predetermined rate for the idle and gas-loaded operation conditions;

sorting out the maximum and minimum values of the time series of the state variables from each segmented set of consecutively sampled signals being longer than the period of fluctuating state variable signal components for the idle and gas-loaded operation conditions;

estimating the one set of the best-fitted model parameters of the asymptotic upper bound from the sorted maximum values of each state variable and the other set of the best-fitted model parameters of the asymptotic lower bound from the sorted minimum values of each state variable, by using a linear parametric model-based active diagnostic algorithm, for the idle and gas-loaded operation conditions;

evaluating the pumping speed indicator based on a inlet pressure signal by using the in-situ estimation method whenever the transition state from the gas-loaded operation condition to the idle one is observed;

storing the estimated upper and lower model parameters of the asymptotic upper and lower bounds of all considered state variables for each idle and gas-loaded operation condition and the evaluated pumping speed indicator whenever the transition state from the gas-loaded operation condition to the idle one is observed;

repeating the estimation of model parameters of the upper and lower bounds of each state variable for the idle and gas-loaded operation conditions and the evaluation of the pumping speed indicator; and

monitoring the variation trend from the estimated upper and lower model parameters of all considered state variables collected for the consecutive idle operation conditions and from those of all considered state variables collected for the consecutive gas-loaded operation conditions and from the consecutively collected pumping speed indicators from the repeated transitions from the gas-loaded operation state to the idle state, and diagnosing whether the vacuum pump is abnormal or not on the basis of the variation trend analysis results thereof.

2. The method according to claim 1, wherein the vacuum pump is used in semi-conductor fabrication processes.

3. The method according to claim 1, wherein the state variables comprise the inlet and exhaust pressure signals, the supply currents of the booster and dry pumps, the root-mean-squared mechanical vibration levels on the booster and dry pumps, and the root-mean-squared acoustical noise levels near the mid position between the booster and dry pumps.

4. The method according to claim 3, in the case of the root-mean-squared mechanical vibration levels on the booster and dry pumps, and the root-mean-squared acoustical noise levels near the mid position between the booster and dry pumps, said method further comprises the step of estimating peak values of the mechanical vibration and acoustical noise signals for the idle and gas-loaded operation conditions.

5. The method according to claim 1, wherein the in-situ evaluation method involves measuring the inlet pressure signal in a similar way to the pump down test method where the negative-going transition of the inlet pressure is used to evaluate the pumping speed.

6. The method according to claim 1, wherein the pumping speed indicator is defined as the pumping speed per unit volume, that is

$I_P=\frac{Q}{V}=3.6\times$

10 4 ·

α

where the symbols Ip and alpha denote the pumping speed indicator and the exponential decay constant and the symbols Q and V denote the pumping speed and the volume to be evacuated; and

the exponential decay constant is obtained from the logarithm values of the inlet pressure signals {Pn;

n=1, . . . , N} corresponding to the negative going transition region of the inlet pressure as

y_n=−

α

·

n+β

, y_n=ln(P_n) and β

=ln(P₀)

where the estimation of the best-fitted exponential decay constant alpha and the initial value beta is obtained by using the least squared method.

7. The method according to claim 1, wherein the parametric linear model for describing the upper and lower asymptotic bounds for each sampled state variable is given as

y _k,n =a _k ·

n+b _k

where the subscript k denotes the upper or lower asymptotic models {k=U or L}; and

the two sets of upper and lower model parameters {a_k, b_k;

k=U or L} for each state variable are obtained by using the least square method as follows;

$a_k=\frac{N·<br><br>\underset{n=1}{\overset{N}{\sum <br><br>}}<br><br>n·<br><br>y_{k,n}-\underset{n=1}{\overset{N}{\sum <br><br>}}<br><br>n·<br><br>\underset{n=1}{\overset{N}{\sum <br><br>}}<br><br>y_{k,n}}{N·<br><br>\underset{n=1}{\overset{N}{\sum <br><br>}}<br><br>n^2-{\left(\underset{n=1}{\overset{N}{\sum <br><br>}}<br><br>n\right)}^2},<br><br>b_k=\frac{\underset{n=1}{\overset{N}{\sum <br><br>}}<br><br>n^2·<br><br>\underset{n=1}{\overset{N}{\sum <br><br>}}<br><br>y_{k,n}-\underset{n=1}{\overset{N}{\sum <br><br>}}<br><br>n·<br><br>\underset{n=1}{\overset{N}{\sum <br><br>}}<br><br>n·<br><br>y_{k,n}}{N·<br><br>\underset{n=1}{\overset{N}{\sum <br><br>}}<br><br>n^2-{\left(\underset{n=1}{\overset{N}{\sum <br><br>}}<br><br>n\right)}^2}$

where the first parameters {a_k;

k=U or L} are the slopes of the upper and lower bounds and the second parameters {b_k;

k=U or L} are the initial values of the upper and lower bounds.

8. The method of according to claim 1, wherein determining the transition state from the gas-loaded operation condition to the idle operation condition is implemented by monitoring the threshold level of the inlet pressure level.

9. The method of according to claim 1, wherein determining the transition state from the gas-loaded operation condition to the idle operation condition is implemented by monitoring the threshold level of either the supply current level of the booster pump or the exhaust pressure signal.

10. The method according to claim 1, wherein the steps of the variation trend monitoring and diagnosing of the considered vacuum pump are performed using the pair of two-dimensional structured data matrices which comprise the estimated model parameters and the peak value of each state variable obtained from each idle or gas-loaded operation condition as follows:

{p_idle(n,(j−

1)×

K+k)} and {p_Load(n,(j−

1)×

K+k)}

for n=1, . . . ;

j=1, . . . , J (total number of sampled state variables);

k=1, . . . , K (total number of estimated model parameters)

wherein the subscript symbols “

Idle” and

“

Load”

denote the idle and gas-loaded operation conditions and the raw index n denotes the sequence of fabrication processes and the column indices j and k denote the classification number of the state variables and the order of the estimated model parameters for each state variable.

12. A computer-readable storage media including a computer program which performs the trend monitoring and diagnostic analysis method claimed in any one of claims 1 to 10 for the failure protection and predictive maintenance of vacuum pump running under the idle and gas-loaded operation conditions alternately.

11. A trend monitoring and diagnostic analysis system for the failure protection and predictive maintenance of vacuum pumps running under the multiple processes including idle and gas-loaded operation conditions alternately, comprising:

data acquisition unit for sampling the time series of the state variables signals at the predetermined rate for the idle and gas-loaded stages in a real-time; and

signal processing unit provided with calculation and storage means for the measured signals from the data acquisition unit, and said signal processing unit being characterized in that it provides the steps of;

sorting out the maximum and minimum values of the time series of the state variables from each segmented set of consecutively sampled signals being longer than the period of fluctuating state variable signal components for the idle and gas-loaded operation conditions;

estimating the one set of the best-fitted model parameters of the asymptotic upper bound from the sorted maximum values of each state variable and the other set of the best-fitted model parameters of the asymptotic lower bound from the sorted minimum values of each state variable for each idle and gas-loaded operation condition;

evaluating the pumping speed indicator based on a inlet pressure signal by using the in-situ estimation method whenever the transition state from the gas-loaded operation condition to the idle one is observed;

storing the estimated upper and lower model parameters of the asymptotic upper and lower bounds for each state variable and the evaluated pumping speed indicator;

monitoring the variation trend from the estimated upper and lower model parameters of all considered state variables collected for the consecutive idle operation conditions and from those of all considered state variables collected for the consecutive gas-loaded operation conditions and from the consecutively collected pumping speed indicators from the repeated transitions from the gas-loaded operation state to the idle state, and diagnosing whether the vacuum pump is abnormal or not on the basis of the variation trend analysis results thereof.