KR20080109572A - Interlock system and the appreciation method - Google Patents

Abstract

An interlock system and evaluation method is provided to accurately determine the status of the normal operation of a facility by comparing reference data considering to the point of time when sensor data showing the status information of facility are transmitted. An interlock evaluation method of the interlock system comprises: similarity between patterns is determined by dynamic time warping data pattern and the reference date pattern transmitted from sensor(540). The manufacturing process is determined by the decision result.

Description

Interlock system and the appreciation method

1 is a block diagram of an interlock system according to an embodiment of the present invention.

FIG. 2 is a diagram illustrating the interlock evaluation unit of FIG. 1. FIG.

3 is a graph of matching data patterns to the DTW algorithm.

4 is a graph in which a sensor data pattern and a reference data pattern are matched with a DTW algorithm according to an embodiment of the present invention.

5 is a flowchart illustrating an interlock evaluation method of an interlock system according to an embodiment of the present invention.

* Description of symbols on the main parts of the drawings *

100: sensor 200: plant control device

300: FDC device 310: data processing unit

320: database 330: interlock setting unit

340: interlock evaluation unit 350: alarm unit

360: display unit 400: MES controller

The present invention relates to an interlock system and a method for evaluating the same. More particularly, the present invention relates to a sensor data pattern and a reference data pattern by a dynamic time warping (DTW) algorithm to determine similarity between the patterns. The present invention relates to an interlock system and an evaluation method for determining an abnormality in a manufacturing process based on the determination result.

In general, semiconductor and flat panel display manufacturing process has a large production cost in each unit process, quality and productivity are greatly affected by the state of the manufacturing process. Therefore, in order to suppress the production of defective products and the waste of unnecessary production resources, there is a need for early detection of abnormalities in each process. Due to this necessity, an interlock system has been introduced to automatically determine whether a semiconductor process facility is abnormal and to stop the process when it is determined that the facility is malfunctioning.

As described in Korean Patent Laid-Open Publication No. 2005-0066703 (Interlock System and Method of Semiconductor Process Equipment), the system receives data corresponding to each process state at predetermined time intervals through sensors mounted on the equipment. The data were separated into a predetermined length, and one-to-one comparison was made at the same time point as the preset data, and when it was determined that the input data did not satisfy the preset criteria, it was determined that the manufacturing process was abnormal.

However, in the semiconductor manufacturing process, the progress time of each process may vary. For example, in the process of etching a wafer, if the processing of the wafer in the etching facility is delayed, the processing of the wafer to be etched later becomes slow. This resulted in a time difference between the data output from the sensor and the preset data, making accurate matching impossible and causing a false alarm.

SUMMARY OF THE INVENTION The present invention has been made to solve the above problems, and an object of the present invention is to provide an interlock system and an evaluation method for stopping an installation by accurately detecting a process of a facility having an abnormality in a manufacturing facility.

According to the present invention for achieving the above object, the similarity between the patterns is determined by dynamic time warping (DTW) between the sensor data pattern and the reference data pattern, and the abnormality of the manufacturing process is determined based on the determination result.

In addition, information is extracted by performing first dynamic time warping (DTW) between the plurality of reference data patterns, and the plurality of reference data patterns and the second dynamic time warping are performed on the sensor data pattern transmitted from the sensor. Information is extracted, the similarity is determined using the extracted information, and the abnormality of the manufacturing process is determined based on the determination result.

Further, the similarity is similarity between the DTW values extracted by the first dynamic time warping and the second dynamic time warping.

Also, the similarity is similarity between the warping path and the best warping path extracted by the first dynamic time warping and the second dynamic time warping.

The similarity is similarity of individual distance data for each warping coordinate extracted by the first dynamic time warping and the second dynamic time warping.

In addition, when the sensor data pattern is a flow type, the lengths of the sensor data pattern and the reference data pattern are adjusted at a predetermined ratio.

In addition, if there is no abnormality in the manufacturing process as a result of the determination, the reference data pattern is changed to the sensor data pattern.

In addition, the reference data is characterized in that a plurality.

In addition, dynamic time warping is performed between a plurality of reference data patterns to obtain a first DTW average value and the standard deviation between the reference data patterns, and the sensor data pattern transmitted from the sensor is subjected to the dynamic time warping. Calculate a second DTW value between the sensor data pattern and the plurality of reference data patterns, calculate a similarity percentage between the sensor data pattern and the plurality of reference data patterns using the obtained values, and calculate the similarity. If the percentage is less than the reference value, the abnormality of the manufacturing process is judged.

In addition, the similarity percentage is calculated using the following equation.

P = | 100 × (| SDTW-M |) / 3S-100 |

(Where, SDTW = second DTW value, M = first DTW mean value, S = standard deviation of first DTW mean value, P = similarity percentage)

In addition, a warping pass, a best warping pass, and a warping constraint path are obtained by dynamically time warping the sensor data pattern and the reference data pattern transmitted from the sensor, and obtaining a mean value of the maximum distance difference that the warping constraint path can allow for the warping path. And calculating a similarity percentage between the warping path and the best warping path, and determining the abnormality of the manufacturing process when the calculated similarity percentage is less than a reference value.

In addition, the similarity percentage is calculated using the following equation.

P = (Max_avg_dist²-Avg_dist²) / (Max_avg_dist²) × 100

(However, Max_avg_dist = warping path constraint path is the maximum allowable distance difference warping path, Avg_dist = warping path average distance value, P = similarity percentage)

In addition, dynamic time warping is performed between a plurality of reference data patterns to obtain an average value and standard deviation of individual distance data for each warping coordinate matched between the reference data patterns, and the allowable range is calculated using the average value and the standard deviation thereof. Obtaining individual distance data of warping coordinates in which the sensor data pattern is matched with the plurality of reference data patterns by dynamically time warping the sensor data pattern transmitted from the sensor with a plurality of reference data patterns, and obtaining the individual distance for each warping coordinate. The number of data existing within the allowable range is obtained, and the similarity percentage between the sensor data pattern and the reference data pattern is obtained. When the calculated similarity percentage is less than the reference value, abnormality of the manufacturing process is determined.

In addition, the similarity percentage is calculated using the following equation.

P = 100 × (Total_data-Total_out) / Total_data

 (However, Total_data = number of distance data for each warping coordinate whose sensor data pattern is matched with a plurality of reference data patterns, and Total_out = number out of allowable range among Total_data.)

 In addition, a sensor mounted on a manufacturing facility for outputting sensor data, a database for storing a reference data pattern, a data processing unit for making the sensor data into a sensor data pattern, the sensor data pattern generated in the data processing unit the reference data pattern And an interlock system for determining the similarity between the patterns by dynamic time warping.

The data processing unit adjusts the length of the sensor data pattern and the reference data pattern at a predetermined ratio when the sensor data pattern is a flow type.

In addition, a plurality of reference data patterns stored in the database.

The interlock evaluation unit updates the reference data pattern of the database with the sensor data pattern when there is no abnormality in the manufacturing process as a result of the similarity determination.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the drawings. As shown in FIG. 1, an interlock system according to an exemplary embodiment of the present invention is installed in a semiconductor facility in a semiconductor manufacturing process and outputs data relating to its identification number and the state of the facility at a sampling interval of a predetermined time. Sensor data, referred to as 'sensor data', controls the driving of semiconductor devices and receives sensor data from facility control device 200 and facility control device 200 that receive sensor data transmitted by sensor 100. Fault Detection and Classification (FDC) device 300 that receives inputs to analyze and determine the normal progress of the facility, and MES (Manufacturing Execution System) control device that controls the execution and interruption of semiconductor equipment ( 400).

The FDC apparatus 300 includes a data processing unit 310 for forming sensor data into a pattern, a database 320 for storing a reference data pattern representing a data pattern when a wafer is normally processed in a unit process of each facility, and an interface to a user. The interlock setting unit 330 to receive the set value of the lock system, the interlock evaluation unit 340 to evaluate the interlock, the alarm unit 350 to inform the MES of the interlock information, and monitor the interlock information to the user. It includes a display unit 360 to display through.

The data processing unit 310 receives sensor data provided from the facility control apparatus 200 and receives a data pattern of a format usable in the FDC device 300, in particular, the interlock evaluation unit 340 (hereinafter, 'sensor data pattern'). And converts the data to the interlock evaluation unit 340.

Semiconductor equipment is largely divided into two types: chamber type equipment for inserting and processing a wafer into a chamber space, and processing and inserting and processing the next wafer; and flow type for continuous processing if a wafer flows from a conveyor belt. There is facility of. The data sent in the chamber type process (hereinafter referred to as 'chamber type data') has a constant wafer processing time and the start and end signals are clearly identified. However, data sent from a flow type facility (hereinafter referred to as 'flow type data') is a combination of continuous data that reduces or increases the time that the next wafer is processed depending on the time the previous wafer is processed because of the continuous operation of the facility. Consists of

In order to match the input sensor data with the reference data pattern, the data processing unit 310 processes a wafer when a data value exceeding an allowable range of an idle value, which is a data value when the wafer is not processed in a facility, is input. If the data value within the allowable range of the initial value is inputted thereafter, it is judged to be the end of wafer processing, and the data inputted within this time is made into a sensor data pattern representing the unit process of the facility.

However, since the length of the flow type data is variable, it is necessary to adjust the length of the flow type data to a predetermined range of the reference data pattern length in order to accurately compare the reference data pattern. Therefore, the data processing unit 310 confirms the process type through the database 320 using the sensor identification number transmitted together with the sensor data in order to determine the type of the created sensor data pattern. By loading the reference data pattern corresponding to the sensor identification number to make the ratio of the pattern length within a certain range, the accuracy of the interlock determination of the interlock evaluation unit 340 is improved.

In this case, in addition to the method of classifying the process type by the identification number of the sensor, a method of classifying the process type by using the change in the length of the sensor data pattern is applicable.

The interlock setting unit 330 receives various setting data for the sensor 100 to be managed from a user. For example, the process type and reference data pattern of the facility corresponding to the number of the sensor are received. In this case, it is preferable to receive a plurality of reference data patterns for accurate interlock determination. The setting data input through the interlock setting unit 330 is stored in the setting database 320.

The interlock evaluation unit 340 compares the sensor data pattern transmitted from the data processing unit 310 with the reference data pattern stored in the setting database 320 to determine whether the interlock is performed. The interlock evaluation unit 340 notifies the alarm unit 350 and the display unit 360 if it is determined that an interlock has occurred in the semiconductor facility. If the sensor data pattern is normal, the interlock evaluation unit 340 updates the reference data pattern of the database 320 with the sensor data pattern so as to take a quick response to the change in the process.

 The database 320 stores the process type corresponding to the identification number of the sensor and its reference data patterns, and the reference data pattern is updated by the interlock evaluation unit 340. At this time, the stored reference data patterns are updated in the oldest order.

The alarm unit 350 receives the evaluation information on the occurrence of the interlock from the interlock evaluator 340, analyzes the corresponding information, and determines that the operation of the corresponding facility should be stopped, the MES control apparatus 400. By notifying the MES control device 400 to stop the operation of the facility.

The display unit 360 is a device for informing the user of the evaluation result from the interlock evaluation unit 340 in a form that can be recognized by the user, and may include a display device such as a monitor.

2 is a block diagram illustrating an interlock evaluator 340 according to an embodiment of the present invention. The interlock evaluator 340 includes a data pattern matching unit 341, a DTW similarity analyzer 342, and a war. The ping path similarity analyzer 343, the matching distance similarity analyzer 344, and the interlock determiner 345 are included.

The DTW algorithm was introduced to the IEEE by H. Sakoe and S. Chiba as "Dynamic Programming Algorithm Optimization for Spoken Word Recognition". This algorithm eliminates the time difference between patterns by warping the time axis until one pattern matches as closely as possible with the other. 3 is a graph in which patterns are matched with a DTW algorithm. As shown in FIG. 3, data for each coordinate of a test data pattern is matched to coordinates closest to the distance between the reference data pattern through pattern matching. By this matching, a value obtained by dividing the sum of distance values for each coordinate whose warping path and test data pattern are matched with the reference data pattern by the number of warping paths (hereinafter referred to as a DTW value) can be obtained.

The data pattern matching unit 341 receives a sensor data pattern and an identification number of the corresponding sensor from the data processing unit 310, and receives a reference data pattern corresponding to the identification number through the database 320. Then, DTW algorithm matching between the reference data patterns and DTW algorithm matching between the sensor data pattern and the reference data pattern are performed. 4 is a graph in which a sensor data pattern and a reference data pattern are matched with a DTW algorithm according to an embodiment of the present invention. As shown in FIG. 4, the DTW algorithm requires that a warping path constraint is set. In the present invention, the warping path constraint path is set using Equation 1 below.

[Equation 1]

j X (n / m) + n X 0.1 ≤ i ≤ j X (n / m) + n X 0.1

(N = length of sensor data pattern, m = length of reference data pattern, i = coordinate of sensor data pattern, j = coordinate of reference data pattern)

At this time, the best warping path becomes (length of reference data pattern / length of sensor data pattern).

The data pattern matching unit 341 may apply the DTW algorithm of the patterns, and the DTW value average and standard deviation of the reference data patterns, the DTW values of which the sensor data pattern is matched with the reference data patterns, and the average and sensor data pattern Individual distance data for warping coordinates, warping path data matched with these reference data patterns, The best warping pass data, the mean of the maximum distance difference that the warping path restriction pass can allow, and the mean distance value of the warping path are extracted. The data pattern matching unit 341 selects and transmits data required by the DTW similarity analyzer 342, the warping path similarity analyzer 343, and the matching distance similarity analyzer 344 among the extracted data.

The DTW similarity analysis unit 342 receives a DTW value for each pattern obtained by the data pattern matching unit 341 to obtain a similarity percentage of the sensor data pattern. The DTW value average (M) and standard deviation (S) of the reference data patterns are used to set the similarity allowance of the sensor data patterns within the M ± 3S range. In general, in the industrial field, it is determined that there is an error when the equipment deviates from 3S in the average data, and the present invention also applies this range. The percentage of the DTW value of the sensor data pattern compared to the allowable range of similarity is calculated using Equation 2 below, and transmitted to the interlock determination unit 345.

[Equation 2]

P = | 100 × (| SDTW-M |) / 3S-100 |

(However, SDTW = DTW value of sensor data pattern, P = DTW value similarity percentage)

For example, assuming that the DTW value average (M) of the reference data pattern is 100, the standard deviation (S) is 1, and the DTW value of the sensor data pattern is 100, and the data are substituted into the above equation, the similarity is 100. % Comes out. If the sensor data pattern exceeds the allowance for similarity, it is determined that there is a problem with the equipment and the similarity percentage is set to 0%.

The warping path similarity analysis unit 343 may determine a maximum distance difference average of the warping path that the warping path restriction path transmitted by matching the sensor data pattern and the reference data pattern in the data pattern matching unit 341 with dynamic time warping. Using the average distance value, the warping path, and the best warping path data, the similarity percentage of the warping path is obtained using Equation 3 below, and transmitted to the interlock determination unit 345.

[Equation 3]

 P = (Max_avg_dist²-Avg_dist²) / (Max_avg_dist²) × 100

(Where Max_avg_dist = average of the maximum distance difference allowed by the warping path constraint pass, Avg_dist = average distance value of the warping path, and P = the similarity percentage of the warping path)

The use of the square size in Equation 3 is to subdivide and monitor the variation of the difference value of the individual similarity. For example, suppose you have three pieces of data, and all three things differ by two and one or two and three things have the same difference mean, but the squared size makes the latter more different. In the present invention, square size is used because it is important to determine whether or not there is an extreme difference value. The similarity percentage of the warping path calculated by Equation 2 is obtained as the distance difference between the average of the maximum distance difference and the average distance value of the warping path that the warping path constraint pass allows. That is, as the warping pass approaches the best warping path, the similarity percentage of the warping path increases.

The matching distance similarity analysis unit 344 calculates an average M and standard deviations of individual distance data for each warping coordinate matched to the reference data patterns transmitted from the data pattern matching unit 341.

Obtain The individual distance value of coordinates matched by the reference data pattern and the DTW algorithm is matched between the reference data pattern (total_out) and the sensor data pattern matched with the reference data pattern. The matching distance similarity percentage is obtained by using the total number of distance data (Total_data) for each warping coordinate. At this time, the allowable range is set to M ± 3S similarly to the DTW similarity analyzer 342. Then, using Equation 4 below, the percentage obtained without departing from the allowable range value compared to Total_data is obtained, and transmitted to the interlock determination unit 345.

[Equation 4]

P = 100 × (Total_data-Total_out) / Total_data

(However, Total_out = Number out of the allowable range among Total_data, Total_data = Number of distance data for each warping coordinate whose sensor data pattern matches the reference data pattern.)

 The interlock determination unit 345 analyzes each percentage calculated by the similarity measurement unit 342, the warping path analysis unit 343, and the mapping distance analysis unit 344 to determine the interlock state of the facility and analyze the analyzed information. Send to alarm and display. For example, if the percentage is greater than 0% and less than 25%, the status of the equipment is evaluated as 'warning'. If the percentage is greater than 25% and less than 50%, the status of the equipment is evaluated as 'caution' and sent to the alarm and display. do.

Hereinafter, an interlock evaluation method of an interlock system according to an embodiment of the present invention will be described with reference to FIG. 5.

The data processing unit 310 receives sensor data measured by the sensor 100 at sampling intervals of a predetermined time, and in order to match the sensor data with a reference data pattern stored in the database 320, one wafer is installed in a facility. Only the data during the processing time is separated to form a pattern (hereinafter referred to as 'sensor data pattern'). At this time, the sensor data pattern is composed of data input within the time using the process start point of the wafer and the end time point at which the process ends (500).

The data processing unit 310 receives a corresponding process type through the database 320 using the identification number transmitted from the sensor together with the sensor data to determine whether the sensor data pattern created in step 500 is a chamber type or a flow type. Check the process type (510).

If the process type identified in step 510 is a flow type, the data processing unit 310 makes the ratio of the length of the sensor data pattern to the length of the reference data pattern stored in the database 320 within a predetermined range, and the data pattern matching unit In operation 341, the similarity percentage of the values obtained by applying the sensor data pattern and the reference data pattern to the DTW algorithm is improved (520).

The data pattern matching unit 341 retrieves a plurality of reference data patterns to be matched with the sensor data patterns received in operation 510 and 520 from the database 320 through the identification numbers input together with the sensor data patterns. A DTW value between the reference data patterns is first obtained by matching the plurality of reference data patterns with the DTW algorithm, and the DTW value of the reference data pattern matched with the sensor data pattern is obtained by matching the sensor data pattern and the reference data patterns with the DTW algorithm. For example, if the reference data pattern is 5, DTW matching between the reference data patterns requires 10 times according to 5C2 (where C is a combination mathematical symbol).

According to the matching, the data pattern matching unit 341 determines that the average DTW values and standard deviations of the reference data patterns, the sensor data patterns match the reference data patterns, and the matched DTW values and the average, the sensor data patterns correspond to the reference data patterns. Then, individual distance data for each warping coordinate, DTW, and warping path data are extracted. The data pattern matching unit 341 selects and transmits data required by the DTW similarity analyzer 342, the warping path similarity analyzer 343, and the matching distance similarity analyzer 344 among the extracted data (530). ).

In operation 530, the DTW similarity analyzer 342 receives DTW averages and standard deviations of the reference data patterns and DTW values between the sensor data pattern and the reference data pattern. In order to determine how similar the DTW value extracted by matching the sensor data pattern with the reference data patterns is identical to the average of the extracted DTW values, Equation 2 is used to obtain a DTW similarity percentage and the inter The lock determination unit 345 transmits the data to the lock determination unit 345.

In step 530, the warping path similarity analyzer 343 receives the maximum distance difference average that the warping path restriction path can allow, the average distance value of the warping path, the warping path, and the best warping path data. Subsequently, the data is substituted into Equation 3 to obtain a similarity percentage of the warping path, and is transmitted to the interlock determination unit 345 (550).

 In operation 530, the matching distance similarity analysis unit 344 may determine the average distance M of the individual distance data for each warping coordinate of the reference data patterns, the standard deviation S, and the individual distance for each warping coordinate for which the sensor data pattern matches the reference data pattern. Receive data (Total_data). Then, M ± 3S is determined as an allowable range, the number of individual distance data (Total_out) for each warping coordinate exceeding the allowable range is determined, and the matching distance similarity percentage of Total_out to Total_data is calculated using Equation 4. The percentage value is transmitted to the interlock determination unit 345 (560).

When the three similarity percentages calculated in steps 540, 550, and 560 are input to the interlock determination unit 345, the interlock determination unit 345 evaluates the interlock state using the percentage range for the facility state, If any value is less than 50%, the percentage value and the evaluation information are transmitted to the alarm unit 350 and the display unit 360 (570).

  In operation 570, the alarm unit 350 receives the interlock state information, analyzes the information, and if it is determined that the operation of the facility should be stopped, the alarm unit 350 notifies the MES control apparatus 400 to notify the MES control apparatus 400. In operation 580, the display unit 360 notifies the user of the interlock state information through the display apparatus.

As described above in detail, the present invention compares the reference data indicating the normal state of the facility by considering the time until the sensor data indicating the state information of the facility is transmitted, thereby accurately determining whether or not the normal operation of the facility.

Claims (18)

Dynamic time warping (DTW) between the sensor data pattern and the reference data pattern transmitted from the sensor determines the similarity between the patterns, The interlock evaluation method of the interlock system which judges the abnormality of a manufacturing process based on the said determination result. Extracts information by executing first dynamic time warping (DTW) between a plurality of reference data patterns, Extracting information by performing the plurality of reference data patterns and the second dynamic time warping on the sensor data pattern transmitted from the sensor; Determine similarity using the extracted information, The interlock evaluation method of the interlock system which judges the abnormality of a manufacturing process based on the said determination result.  The method of claim 2, wherein the similarity is And a similarity between the DTW values extracted by the first dynamic time warping and the second dynamic time warping.  The method of claim 2, wherein the similarity is And a similarity degree between the warping path and the best warping path extracted by the first dynamic time warping and the second dynamic time warping.  The method of claim 2, wherein the similarity is And a similarity degree of individual distance data for each warping coordinate extracted by the first dynamic time warping and the second dynamic time warping.  The method according to claim 1 or 2, If the sensor data pattern is a flow type, The interlock evaluation method of the interlock system of matching the length of the sensor data pattern and the reference data pattern in a predetermined ratio. The method according to claim 1 or 2, If there is no abnormality in the manufacturing process, The interlock evaluation method of the interlock system of changing the reference data pattern to the sensor data pattern. The method of claim 1, And said reference data is a plurality of interlock systems. Dynamic time warping is performed between a plurality of reference data patterns to obtain a first DTW average value and its standard deviation between the reference data patterns, Performing a dynamic time warping with the plurality of reference data patterns and the sensor data pattern transmitted from a sensor to obtain a second DTW value between the sensor data pattern and the plurality of reference data patterns, Obtaining a similarity percentage between the sensor data pattern and the plurality of reference data patterns using the obtained values, The interlock evaluation method of the interlock system to determine the abnormality of the manufacturing process if the calculated similarity percentage is less than the reference value. The method of claim 9, The similarity percentage is an interlock evaluation method of an interlock system calculated using the following equation.  P = | 100 × (| SDTW-M |) / 3S-100 | (Where, SDTW = second DTW value, M = first DTW mean value, S = standard deviation of first DTW mean value, P = similarity percentage) Dynamic time warping of the sensor data pattern and the reference data pattern transmitted from the sensor to obtain the warping pass, the best warping pass, and the warping constraint pass, Obtaining a mean value of the maximum distance difference that the warping constraint path can allow for the warping path, obtaining a similarity percentage between the warping path and the best warping path, The interlock evaluation method of the interlock system to determine the abnormality of the manufacturing process if the calculated similarity percentage is less than the reference value. The method of claim 11, The similarity percentage is an interlock evaluation method of an interlock system calculated using the following equation.  P = (Max_avg_dist²-Avg_dist²) / (Max_avg_dist²) × 100 (However, Max_avg_dist = warping path constraint path is the maximum allowable distance difference warping path, Avg_dist = warping path average distance value, P = similarity percentage) Dynamic time warping is performed between a plurality of reference data patterns to obtain an average value and standard deviation of individual distance data for each warping coordinate between the reference data patterns. The allowable range is calculated using the average value and its standard deviation, Dynamic time warping of a sensor data pattern transmitted from a sensor with a plurality of reference data patterns to obtain individual distance data for each warping coordinate where the sensor data pattern matches the plurality of reference data patterns, Obtaining the number of individual distance data for each warping coordinate exists within the allowable range, Obtaining a similarity percentage between the sensor data pattern and the reference data pattern, The interlock evaluation method of the interlock system to determine the abnormality of the manufacturing process if the calculated similarity percentage is less than the reference value. The method of claim 13, The similarity percentage is an interlock evaluation method of an interlock system calculated using the following equation. P = 100 × (Total_data-Total_out) / Total_data (However, Total_data = number of distance data for each warping coordinate whose sensor data pattern is matched with a plurality of reference data patterns, and Total_out = number out of allowable range among Total_data.) A sensor mounted on the manufacturing facility and outputting sensor data; A database storing a reference data pattern; A data processing unit for making the sensor data into a sensor data pattern; And an interlock evaluation unit configured to determine the similarity between the patterns by dynamically time warping the sensor data pattern generated by the data processing unit with the reference data pattern. The method of claim 15, wherein the data processing unit If the sensor data pattern is a flow type, The interlock system of matching the length of the sensor data pattern and the reference data pattern in a predetermined ratio.  The method of claim 15, And a plurality of reference data patterns stored in the database. The method of claim 15 or 17, wherein the interlock evaluation unit If there is no abnormality in the manufacturing process as a result of the similarity determination, An interlock system for updating a reference data pattern of the database with the sensor data pattern.

Images