WO2011132900A2 - 산업설비의 공정 여유도 감시 시스템용 데이터 수집 방법 및 그 저장매체 - Google Patents
산업설비의 공정 여유도 감시 시스템용 데이터 수집 방법 및 그 저장매체 Download PDFInfo
- Publication number
- WO2011132900A2 WO2011132900A2 PCT/KR2011/002758 KR2011002758W WO2011132900A2 WO 2011132900 A2 WO2011132900 A2 WO 2011132900A2 KR 2011002758 W KR2011002758 W KR 2011002758W WO 2011132900 A2 WO2011132900 A2 WO 2011132900A2
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- data
- variables
- monitoring system
- process margin
- industrial
- Prior art date
Links
- 238000000034 method Methods 0.000 title claims abstract description 122
- 230000008569 process Effects 0.000 title claims abstract description 57
- 238000012544 monitoring process Methods 0.000 title claims abstract description 46
- 238000013480 data collection Methods 0.000 title claims abstract description 25
- 238000005070 sampling Methods 0.000 claims abstract description 17
- 238000012545 processing Methods 0.000 claims abstract description 7
- 238000012549 training Methods 0.000 claims description 54
- 238000009499 grossing Methods 0.000 claims description 7
- 238000012795 verification Methods 0.000 claims description 7
- 230000007423 decrease Effects 0.000 claims description 3
- 238000010200 validation analysis Methods 0.000 claims description 3
- 238000004590 computer program Methods 0.000 claims 1
- 230000006870 function Effects 0.000 abstract description 10
- 238000009434 installation Methods 0.000 description 7
- 230000008859 change Effects 0.000 description 6
- 239000000446 fuel Substances 0.000 description 5
- 238000010248 power generation Methods 0.000 description 5
- 230000008901 benefit Effects 0.000 description 4
- 230000006866 deterioration Effects 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 230000009467 reduction Effects 0.000 description 4
- 238000010187 selection method Methods 0.000 description 4
- 238000013144 data compression Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000001556 precipitation Methods 0.000 description 3
- 238000000611 regression analysis Methods 0.000 description 3
- 239000013535 sea water Substances 0.000 description 3
- 239000000498 cooling water Substances 0.000 description 2
- 239000000203 mixture Substances 0.000 description 2
- 238000013179 statistical model Methods 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000005856 abnormality Effects 0.000 description 1
- 238000007792 addition Methods 0.000 description 1
- 230000032683 aging Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000008570 general process Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B23/00—Testing or monitoring of control systems or parts thereof
- G05B23/02—Electric testing or monitoring
- G05B23/0205—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
- G05B23/0208—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterized by the configuration of the monitoring system
- G05B23/021—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterized by the configuration of the monitoring system adopting a different treatment of each operating region or a different mode of the monitored system, e.g. transient modes; different operating configurations of monitored system
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
Definitions
- the present invention relates to a data collection method for a process margin monitoring system of an industrial facility and a storage medium thereof, and more specifically, to a monitoring system for analyzing a process margin of an industrial facility based on a statistical learning technique,
- the present invention relates to a data collection method for a process margin monitoring system of an industrial facility and a storage medium thereof, for collecting learning data from a computer database and converting the learning data into a form that is easy to learn.
- Industrial equipment is a device consisting of a plurality of systems and devices to achieve a specific purpose, generally one or more instruments are installed for checking the operation and safety status, it consists of a configuration that can be measured offline or online.
- the efficiency and safety of the equipment will vary depending on external conditions (air temperature, pressure, humidity, seawater or precipitation temperature if cooling water is required), characteristics of the injected fuel, deterioration of the equipment, operating range, etc.
- process margin the extent of change that can tolerate the efficiency and safety of the installation.
- Most industrial facilities have a stop / protection function for a specific system or device in order to prevent operation beyond this process margin, and in order to implement such a stop / protection function, the value of a specific operating variable is A control device is in place to forcibly stop the facility when the stop / protection signal is exceeded.
- the process margin and the stop / protection set point are mutually dependent variables.
- the stop / protection set point is set too high, the process margin becomes relatively large, and the cost benefit of operating the industrial equipment increases. There is a problem that can cause a serious accident and cause a long period of equipment shutdown.
- the above stop / protection set value is set too low, the probability of an accident is lowered, but the process margin is relatively low, and thus the industrial equipment is frequently shut down, thereby reducing the cost benefit of operating the industrial equipment. You lose.
- the overall process margin is determined by reflecting these two-sided properties.
- the process margin is set to a conservative value inclusive of all external conditions, input fuel, deterioration of equipment, and operating range. It is common.
- Such a preliminary stop / protection set point is usually a static value, and once set, the set point is determined as a function of two conditions, as long as the value does not change, or the characteristic of the installation even when the value changes.
- An object of the present invention is to solve the conventional problems as described above, in implementing a monitoring system for analyzing the process margin of the industrial equipment based on statistical learning techniques, collecting the training data from the database of the power plant computer In order to convert this into a form that is easy to learn, it is to provide a data collection method and a storage medium for the process margin monitoring system of industrial facilities.
- the configuration of the present invention comprises the steps of: providing a training data set based on data determined to be in a normal state in a driving history of an industrial facility, and classifying the training data set for each operation mode; When the industrial facilities are provided with a plurality of facilities performing the same function, receiving data for each of the plurality of facilities and processing the data for the plurality of facilities; Selecting and grouping related data among data included in the training data set; And sampling the collected data to reduce the number of data.
- the learning data set is composed of the first data set to the N-th data set (N is a natural number of two or more) according to the size of the data to be collected or the collection point of the data, in which case, the first data set is one of the industrial facilities.
- the second data set includes signals included in the entire industrial facility for monitoring the process margin of the entire industrial facility for monitoring the process margin of a specific facility, and the third data.
- the set may consist of signals relating to all or part of the plant immediately after a particular event occurs in all or part of the plant.
- the present invention may further include collecting analog signals that can replace the digital signals and converting the digital signals into the analog signals when there is data represented by the digital signals in the training data set. have.
- a variable having a correlation coefficient greater than or equal to a set value is regarded as the same group, and a smoothing parameter is calculated using a 4-fold validation method for the variables considered to be the same group.
- variables other than the variables considered as the same group may be used by selecting only variables related to the characteristics of the facility in consideration of the characteristics of the facility.
- the correlation coefficient is preferably analyzed by the following equation.
- ⁇ XY correlation coefficient between variables X and Y
- X i i-th value based on sampling interval of training data
- Y i i-th value based on sampling interval of training data
- ⁇ X average
- ⁇ X of the variable Y : the mean
- ⁇ Y standard deviation of variable Y
- N the number of times of the data collection interval in the sampling interval of the training data .
- the number of data for the variable in the grid is reduced or the standard deviation ( ⁇ X ) for the specific variable is calculated based on the variance of the value of the specific variable as a reference of the grid size.
- the number of data remaining in the grid is determined by the product of a set ratio of the number of data for the variable in the grid, and at least one data is set in each grid.
- the present invention has the effect of collecting the training data from the database of the power plant computer and converting it into a form that is easy to learn in implementing a monitoring system for analyzing the process margin of the industrial facility based on a statistical learning technique. .
- FIG. 1 is a schematic diagram of a general power generation system as an industrial facility.
- FIG. 2 is a view showing an example of the configuration of the multi-learning data set of the data collection method for the process margin monitoring system of the industrial equipment according to an embodiment of the present invention.
- FIG. 3 is a view illustrating a user interface for selecting a training data set of a data collection method for a process margin monitoring system of an industrial facility according to an embodiment of the present invention.
- FIG. 4 is a view illustrating an example of collecting analog data or digital data of a data collection method for a process margin monitoring system of an industrial facility according to an exemplary embodiment of the present invention.
- FIG. 5 is a view illustrating a virtual tag generation of a data collection method for a process margin monitoring system of an industrial facility according to an embodiment of the present invention.
- FIG 6 and 7 are views for showing the step-by-step selection of the data collection method for the process margin monitoring system of the industrial equipment according to an embodiment of the present invention.
- FIG. 8 is a view for showing the results of the step-by-step variable selection and the cross grouping of the variables of the data collection method for the process margin monitoring system of the industrial facilities according to an embodiment of the present invention.
- 9 and 10 are diagrams for explaining the principle of data compression of the data collection method for the process margin monitoring system of the industrial equipment according to an embodiment of the present invention.
- the process margin monitoring system uses statistical data obtained from the operational history of the facility (hereinafter referred to as "learning data"). To distinguish between errors in the measuring instrument and abnormalities in the actual equipment.
- the accuracy of the process margin monitoring system is to determine how reliable the training data is collected from the operation history of the facility, and how to group the collected training data and use it in constructing a predictive model.
- the conditions required to increase the accuracy of the process margin monitoring system can be further classified as follows.
- the method of selecting start and end points for collecting training data from a database installed on a power plant computer is a method of selecting start and end points for collecting training data from a database installed on a power plant computer.
- Steady state means that the operating conditions of the equipment remain unchanged and stable, and the collected data are generally easy to construct statistical models.
- digital data which mainly informs the operation status of the equipment such as valve open / closed and pump run / stop, plays an important role in the statistical learning model. Problems in reflecting the statistical learning model arise. Therefore, there is a need for a method of receiving digital data from a database installed on a power plant computer and inputting it into a process margin monitoring system.
- Industrial facilities that perform critical functions often have more than one backup facility capable of performing the same function. For example, if several pumps are running and the other is standing still, and if one of the running pumps is stopped for some reason, the pump that is standing still is running and It will replace the role. In this case, the operating conditions will not change as the number of operating facilities does not change as a whole, but there is a part that needs to be changed in providing monitoring results to the user since the operating facilities have changed. In other words, there is a need for a method of inputting data of the same characteristics from a plurality of facilities from a database installed in a power plant computer, processing it, and then inputting a process margin to a monitoring system.
- the list of signals monitoring power plants is generally very large, some of which are important in determining the process margin of the plant, but also include many unnecessary signals.
- the simplest way to group is to look at the correlation coefficients between the signals and make the ones with the highest correlation the same group.
- the grouping result may be inconsistent depending on the policy of collecting the training data. Therefore, there is a need for a statistical method and a method of grouping the data to reflect the engineer's knowledge of the equipment and inputting it to the process margin monitoring system.
- a general power generation system includes a steam generator 1 such as a boiler of an energy power plant or a steam generator of a nuclear power plant, and a steam turbine 2 connected to the steam generator 1. And a condenser 3 connected to the steam turbine 2 and a pump 4 connected between the condenser 3 and the steam generator 1.
- a to G are signals obtained from sensors installed in respective equipments, A is an outlet pressure signal of the steam generator 1, B is a pressure signal of the condenser 3, and C is a plurality of temperatures. Is a signal, D is an outlet pressure signal of the pump 4, E is a feed water flow rate signal, F is a pressure signal in the steam generator 1, and G is a temperature signal in the steam generator 1.
- Ideal training data should be prepared only from the operating conditions of a normal installation without ageing or deterioration, and should be prepared for all external conditions (such as ambient temperature, pressure, or humidity, seawater or precipitation temperature if cooling water is required) and internal conditions ( Operating data in combination with the characteristics of input fuel, operating range). However, in reality, it is impossible to collect such data perfectly, so prepare the training data in the following way.
- the training data set can be configured in multiple numbers accordingly. Therefore, the training data set may include the first data set, the second data set, the third data set,... N-th data set (N is a natural number).
- the first data set includes signals C, D, for monitoring the process margin of a specific facility (eg, the pump 4 of the power generation system).
- E has a learning database consisting of only, and periodically collects and stores three months data collected immediately after the replacement or maintenance of the facility (see Fig. 2 (a)).
- the second data set all the signals (A, B, C, D, E, F, G) are included in the training database for monitoring the process margin of the entire plant, which is one year after the plant is first installed. Operation history data is included.
- the second data set is used to confirm how much the state of the current power generation facility differs from the design value (see FIG. 2B).
- the third data set contains signals for the entire installation (A, B, C, D, E, F, G), but for specific events, such as three months after each planned outage, each summer or winter, Signals are periodically updated, such as three months after installation.
- the third set can be utilized to observe the state compared to it, based on the plant conditions immediately after the particular event occurred (see FIG. 2C).
- Statistical learning methods are divided into learning mode and execution mode. Multiple training data sets are modeled in training mode for each set, and provide an appropriate interface for the user to select when entering run mode.
- 3 illustrates an example of a user interface for selecting the training data set configured in FIG. 2.
- the system is stopped for the first time and then started, and the operating conditions are kept constant. After a certain period of time, the process is stopped. Therefore, it can be divided into start mode, normal operation mode and stop mode. In some cases, the operation mode can be subdivided and operated.
- the operation mode can be subdivided and operated.
- a model suitable for the operation mode is used.
- it is executed only when the data obtained under the operating condition does not exceed the range of data prepared in the learning mode, and when the system state is not entered, the user is not sure of the reliability of the output result. Either generate a or allow the calculation to be bypassed automatically.
- the training data is collected using an analog signal that can replace the digital signal.
- the digital signal indicating the opening and closing of the valve
- the flow rate, pressure, temperature, etc. in the pipe located downstream of the valve must be included in the learning data to indirectly know the opening / closing state of the valve.
- 4 shows an example of collecting analog data or digital data.
- A1 is an analog signal relating to the discharge part pressure of the pump 4
- A2 is an analog signal relating to the discharge part temperature of the pump 4
- D1 is an on / off state of the pump 4.
- 4 (b) illustrates a data set for a case where digital data is not available
- FIG. 4 (c) illustrates a data set for a case where digital data is available.
- kernel regression is used as a model for training data, there is no problem using a mixture of analog and digital data.
- important digital data must be assigned to the same group of training data.
- important digital data can be lost in the grouping process. Therefore, a method of finding an optimal combination of groupings, which will be described later, should be utilized.
- the result of the digital signal is not only 0 or 1, but may be a median value or a value out of it. In this case, it is determined that there is a possibility that an indicator such as open / close or stop / operation of a digital signal is wrong.
- the concept of virtual tag is a necessary signal, but it can be used to represent a location where no instrument is actually installed, a location where the instrument cannot be installed, or a physical quantity that is not of a measurable nature.
- the enthalpy other than the thermometer and the pressure gauge is to be used as a signal at the positions H1 to H4 at the discharge port side of the pumps 4a, 4b, 4c, and 4d at the points H1 to H4 in FIG. It is possible to make and use enthalpy gazig as a function of temperature and pressure.
- singularities included in the learning data should be removed.
- Representative examples of singularity include the case where no data is input at all, such as 'Bad Input', and the case where the data is input at the same time as 'Out of Range', but is temporarily larger or smaller than well beyond the normal range.
- the reliability of the training data is improved by simultaneously removing the data of all variables acquired at that time. All variables that do not change during the sampling period of the training data are treated as 'Bad Inputs' so that they do not become noise in modeling.
- Training data contains a mixture of useful and useful information to inform the status of a particular facility. Also, even signals that contain useful information, not all signals provide status for every facility in the system. Therefore, it is necessary to group signals containing information useful for checking the status of each target equipment. By performing such grouping, signals containing information that are not useful can be excluded from the training data, and the number of signals required for monitoring a specific facility can be reduced to an appropriate level.
- the correlation coefficient used as a criterion for grouping in the statistical learning method is analyzed for all pairs of variables constituting the training data, and is calculated as shown in Equation 1 below. If the calculated correlation coefficient is more than the set value, it is regarded as training data, otherwise it is dropped from the training data.
- the setting value is input by the user.
- ⁇ XY correlation coefficient between variables X and Y
- X i i-th value based on sampling interval of training data
- Y i i-th value based on sampling interval of training data
- ⁇ X average
- ⁇ X of the variable Y : the mean
- ⁇ Y standard deviation of variable Y
- N the number of times of the data collection interval in the sampling interval of the training data .
- the first problem is that the correlation coefficient between variables that need to be physically correlated is very low, so it is not likely to belong to the same group.
- the correlation coefficient represents the linear relationship of two variables.
- the linearity of any two variables can be analyzed differently according to the duration of the training data sampled. For example, variables that change much more slowly than process changes in the plant, such as outside conditions, seawater or precipitation conditions, and fuel conditions, affect the overall performance of the plant, but are slow and not sufficiently reflected in the correlation coefficient. . You can think of these variables as independent variables of the whole system. In other words, changes in the system do not affect these variables, but these variables affect the changes in the system.
- the second problem is that if these variables belong to one group, they cannot belong to another group. Since independent variables in the system affect all groups, they need to be shared across multiple groups.
- step variable selection method is proposed.
- variables that represent preset values or user-specified values such as 0.8 or more, are considered to be the same group by using the correlation coefficient.
- the quadruple verification method is to divide the training data into quadrants, make an autocorrelation regression model using the data of the third quadrant, and then repeat the method of verifying the model using the remaining data in different combinations. This results in a total of four verifications.
- the third data used to create the autocorrelation regression model is called learning data
- the first data used to verify the created regression model is called testing data.
- Each verification step is called a run. Therefore, the quad verification method performs four runs. For each run, we use the Square Sum of Residuals (SSR) as an indicator of the superiority of the regression model. In this case, the calculated sum of squares of the residuals (SSR) is defined as SSR 1 .
- the set value is a ratio of the reduction rate of the sum of squares of the residuals in Case 5 to the sum of the squares of the residuals in Case 4 to the reduction ratio of the sum of squares of the residuals in Case 4 to the sum of squares of the residuals in Case 3 shown in FIG. Can be determined.
- such a set value can be understood as a numerical value for selecting a state in which the decrease in the sum of squares of the residuals is sharply slowed or no longer decreases. Therefore, in the case of FIGS. 6 and 7, the variables A, B, C, and F are determined to be the same group.
- step 3 It is very likely that step 3 will take a very long time, since you have to think about the combination of many variables.
- the variables related to the characteristics of the facility are determined as independent variables in consideration of the characteristics of the facility, and step 3 is performed only for the independent variables.
- the second problem is automatically solved by using the stepwise variable selection method described above.
- the result of the stepwise variable selection and the cross grouping of the variables will be as shown in FIG. 8.
- Three variables A0001, A0002, and A0003 shown in FIG. 8 belong to groups 1, 2, and 3, respectively, and in particular, A0002 shows that they belong to group 1 as well.
- step (5) The amount of training data actually collected is often so large that even modern computers are difficult to analyze. In this case, it can be very time-consuming to select and step group in step (5).
- a method of reducing the number of data in the grid based on the signal distribution as a criterion size is proposed as follows. First, the variance of the value of a particular variable is calculated and set as the reference grid size. The reference grid size allows the user to set larger or narrower. Next, set the grid for each variable, and hit the actual data into each grid.
- FIG. 9 shows original data.
- the grid drawn on the horizontal and vertical axes was determined by the magnitude of the variance of the variable on the horizontal axis and the variable on the vertical axis.
- Resolution means how many times the variable is to be truncated to cut the grid. That is, the larger the resolution is set, the smaller the grid is set, the greater the amount of training data will be.
- the grid size (GridSize x ) according to the resolution may be calculated as shown in Equation 2 based on the standard deviation ⁇ X of the variable.
- each variable is cut out by dividing -5 ⁇ to + 5 ⁇ by the resolution.
- the reason for using -5 ⁇ to + 5 ⁇ rather than dividing the minimum value to the maximum value of the variable by resolution is that the learning data sometimes contains abnormally large or small values. This is because the distribution of the grid can be abnormally divided. And since the variables follow the natural distribution, most of the data is distributed between -5 ⁇ and + 5 ⁇ . For example, if you set the resolution to 4, the grid will be cut into 4 grids from -5 ⁇ to -2.5 ⁇ , -2.5 ⁇ to average, + 2.5 ⁇ to average, + 2.5 ⁇ to + 5 ⁇ , and resolution is set to 2. Will be reduced to two lattice, average at -5 ⁇ and + 5 ⁇ at mean.
- the number of data in all grids is reduced according to this ratio by using a preset ratio or a constant ratio input by the user. If the data is reduced according to this ratio, if at least one is left, at least one should be left. 10 is a view of the data remaining after being removed by this principle.
- the distance from the total data is reflected. Most process variables follow the form of a normal distribution. Therefore, when looking at the entire interval, the learning data is concentrated at an intermediate point. This affects the signal prediction, which results in the prediction being totally centered. However, it is difficult to completely rule out the importance of occasional external data. Using this method reduces the number by considering the distribution of the data, which helps to effectively reduce the number without losing important data, which is another advantage of this method.
- the data compression method can be used in various ways in the statistical learning method, but in order to have the best effect, the data compression method should be performed in the same group after the grouping of the variables is performed. This is because the compression effect can be reduced when applied to a signal that has not undergone any signal processing.
Abstract
Description
Claims (11)
- 산업설비의 운전이력에서 정상적인 상태로 판정되는 데이터를 토대로 학습용 데이터 세트를 마련하되, 상기 학습용 데이터 세트를 운전모드별로 구분하는 단계;상기 산업설비에서 동일한 기능을 수행하는 복수의 설비를 구비하는 경우, 상기 복수의 설비 중 각 설비에 대한 데이터를 입력받아 상기 복수의 설비에 대한 데이터로 가공하는 단계;상기 학습용 데이터 세트에 포함된 데이터 중 서로 관련된 데이터를 선별하여 그룹핑하는 단계; 및수집된 데이터를 샘플링하여 데이터 개수를 감소시키는 단계;를 포함하는 것을 특징으로 하는 산업설비의 공정 여유도 감시 시스템용 데이터 수집방법.
- 제1항에 있어서,상기 학습용 데이터 세트는 수집 대상 데이터의 규모 또는 데이터의 수집 시점에 따라 제1데이터 세트~제N데이터 세트(N은 2이상의 자연수)로 구성되는 것을 특징으로 하는 산업설비의 공정 여유도 감시 시스템용 데이터 수집방법.
- 제2항에 있어서,제1데이터 세트는 상기 산업설비 중 특정 설비의 공정 여유도 감시를 위하여 상기 특정 설비에 관련된 신호로 구성되고,제2데이터 세트는 상기 산업설비 전체의 공정 여유도 감시를 위하여 상기 산업설비 전체에 포함된 신호로 구성되며,제3데이터 세트는 상기 산업설비 전체 또는 일부에서 특정 이벤트가 발생된 직후의 상기 산업설비 전체 또는 일부에 관한 신호로 구성되는 것을 특징으로 하는 산업설비의 공정 여유도 감시 시스템용 데이터 수집방법.
- 제1항에 있어서,상기 학습용 데이터 세트에서 디지털 신호로 표시되는 데이터가 있는 경우, 상기 디지털 신호를 대신할 수 있는 아날로그 신호를 수집하여 상기 디지털 신호를 상기 아날로그 신호로 변환하는 단계;를 더 포함하는 것을 특징으로 하는 산업설비의 공정 여유도 감시 시스템용 데이터 수집방법.
- 제1항에 있어서,상기 그룹핑 단계는,변수간의 상관계수가 설정값 이상인 변수를 동일 그룹으로 간주하는 단계와,동일 그룹으로 간주된 변수를 대상으로 4중 검증(4-fold validation) 방법을 이용하여 평활 모수를 계산하는 단계와,동일 그룹으로 간주된 변수 이외의 모든 변수의 조합을 상기 그룹에 포함시켜 4중 검증 방법을 이용하여 평활 모수를 계산하면서 잔차의 제곱합(SSR)을 산출하는 단계와,특정 잔차의 제곱합에 대한 직후의 잔차의 제곱합의 감소율이 설정치 이하인 경우, 상기 특정 잔차의 제곱합을 산출한 시점에서 그룹핑을 종료하는 단계를 포함하는 것을 특징으로 하는 산업설비의 공정 여유도 감시 시스템용 데이터 수집방법.
- 제5항에 있어서,상기 잔차의 제곱합을 산출하는 단계에서, 상기 동일 그룹으로 간주된 변수 이외의 변수는, 설비의 특성을 고려하여 상기 설비의 특성에 관련된 변수만을 선별하여 이용되는 것을 특징으로 하는 산업설비의 공정 여유도 감시 시스템용 데이터 수집방법.
- 제1항에 있어서,상기 데이터의 샘플링 단계에서는,특정 변수가 갖는 값의 분산을 격자크기의 기준으로 삼아 해당 격자 내에서 상기 변수에 대한 데이터의 개수를 줄이는 것을 특징으로 하는 산업설비의 공정 여유도 감시 시스템용 데이터 수집방법.
- 제8항 또는 제9항에 있어서,상기 격자 내에 잔류하는 데이터 개수는 해당 격자 내에서 상기 변수에 대한 데이터 개수에 대한 설정된 비율의 곱으로 결정되되, 각 격자에는 최소한 1개의 데이터는 잔류하는 것을 특징으로 하는 산업설비의 공정 여유도 감시 시스템용 데이터 수집방법.
- 제1항 내지 제9항 중 어느 한 항에 따른 산업설비의 공정 여유도 감시 시스템용 데이터 수집방법이 컴퓨터 프로그램화되어 저장되어 있는 것을 특징으로 하는 저장매체.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US13/519,070 US20120271826A1 (en) | 2010-04-19 | 2011-04-18 | Data collecting method for detection and on-time warning system of industrial process |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
KR1020100035755A KR101096793B1 (ko) | 2010-04-19 | 2010-04-19 | 산업설비의 공정 여유도 감시 시스템용 데이터 수집 방법 및 그 저장매체 |
KR10-2010-0035755 | 2010-04-19 |
Publications (2)
Publication Number | Publication Date |
---|---|
WO2011132900A2 true WO2011132900A2 (ko) | 2011-10-27 |
WO2011132900A3 WO2011132900A3 (ko) | 2012-02-02 |
Family
ID=44834615
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/KR2011/002758 WO2011132900A2 (ko) | 2010-04-19 | 2011-04-18 | 산업설비의 공정 여유도 감시 시스템용 데이터 수집 방법 및 그 저장매체 |
Country Status (3)
Country | Link |
---|---|
US (1) | US20120271826A1 (ko) |
KR (1) | KR101096793B1 (ko) |
WO (1) | WO2011132900A2 (ko) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113325797A (zh) * | 2021-06-11 | 2021-08-31 | 中山凯旋真空科技股份有限公司 | 控制设备的数据采集方法、装置、存储介质和电子设备 |
Families Citing this family (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
AU2012233018B2 (en) | 2011-10-06 | 2013-10-24 | Schlumberger Technology B.V. | High-frequency data capture for diagnostics |
CN105657367B (zh) * | 2016-01-05 | 2018-07-17 | 山东康威通信技术股份有限公司 | 一种监狱多子系统接入的监管中心综合平台及接入方法 |
JP6386488B2 (ja) * | 2016-03-17 | 2018-09-05 | ファナック株式会社 | 工作機械の稼働管理方法およびプログラム |
US11200790B2 (en) | 2016-08-29 | 2021-12-14 | Korea Hydro & Nuclear Power Co., Ltd. | Method for pre-detecting abnormality sign of nuclear power plant device including processor for determining device importance and warning validity, and system therefor |
KR101876185B1 (ko) * | 2016-08-29 | 2018-07-09 | 한국수력원자력 주식회사 | 조기경보시스템에서 외부 영향을 고려한 최신 데이터 학습 방법 및 그 시스템 |
KR101967524B1 (ko) | 2017-04-13 | 2019-04-09 | 경희대학교 산학협력단 | 커널 회귀 모델을 통하여 시변 신호를 검증하는 온라인 검증 장치 및 방법 |
KR102020247B1 (ko) | 2018-02-07 | 2019-09-10 | 두산중공업 주식회사 | 플랜트 이상 예측 장치 및 방법 |
KR102047815B1 (ko) | 2018-03-23 | 2019-11-22 | 두산중공업 주식회사 | 플랜트 이상 예측 장치 및 방법 |
TR201806963A2 (tr) * | 2018-05-17 | 2018-06-21 | Siskon Enduestriyel Otomasyon Sistemleri Sanayi Ve Ticaret Anonim Sirketi | Veri̇ toplama si̇stemi̇ |
JP7031502B2 (ja) * | 2018-06-07 | 2022-03-08 | オムロン株式会社 | 制御システム、制御方法、学習装置、制御装置、学習方法及び学習プログラム |
JP7162560B2 (ja) * | 2019-03-28 | 2022-10-28 | 日立造船株式会社 | 情報処理装置、情報処理方法、情報処理プログラム、およびゴミ焼却プラント |
KR102235808B1 (ko) * | 2019-06-05 | 2021-04-02 | 조만영 | 산업 플랜트 설비의 신호 그룹 기반 학습 모델을 이용한 운전 예측 시스템, 방법 및 컴퓨터 프로그램 |
CN111581191B (zh) * | 2020-04-10 | 2023-10-13 | 岭东核电有限公司 | 核安全数据校验方法、装置、计算机设备及存储介质 |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2003295937A (ja) * | 2002-03-29 | 2003-10-17 | Hitachi Ltd | プラント監視装置とプラント監視方法 |
JP2005283269A (ja) * | 2004-03-29 | 2005-10-13 | Toshiba Corp | 沸騰水型原子炉の過渡沸騰遷移監視システムおよび監視方法 |
KR20090021697A (ko) * | 2007-08-28 | 2009-03-04 | 비앤에프테크놀로지 주식회사 | 산업플랜트 설비의 종합 정지 여유도 감시방법 및 그 장치 |
KR20090024055A (ko) * | 2008-05-07 | 2009-03-06 | 비앤에프테크놀로지 주식회사 | 산업플랜트의 트립원인 추적 방법 및 그 시스템 |
Family Cites Families (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP3918697B2 (ja) * | 2001-12-27 | 2007-05-23 | 日本電気株式会社 | サーバ構築支援システム及びサーバ構築支援方法並びにプログラム |
US7143098B2 (en) * | 2002-05-10 | 2006-11-28 | International Business Machines Corporation | Systems, methods, and computer program products to reduce computer processing in grid cell size determination for indexing of multidimensional databases |
US7751955B2 (en) * | 2006-06-30 | 2010-07-06 | Spx Corporation | Diagnostics data collection and analysis method and apparatus to diagnose vehicle component failures |
US20090155784A1 (en) * | 2007-01-22 | 2009-06-18 | Wyeth | Assessment of asthma and allergen-dependent gene expression |
KR100928205B1 (ko) * | 2007-06-05 | 2009-11-25 | 삼성전자주식회사 | 반도체 제조설비 관리시스템 및 그의 통계적 공정 관리방법 |
-
2010
- 2010-04-19 KR KR1020100035755A patent/KR101096793B1/ko active IP Right Grant
-
2011
- 2011-04-18 US US13/519,070 patent/US20120271826A1/en not_active Abandoned
- 2011-04-18 WO PCT/KR2011/002758 patent/WO2011132900A2/ko active Application Filing
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2003295937A (ja) * | 2002-03-29 | 2003-10-17 | Hitachi Ltd | プラント監視装置とプラント監視方法 |
JP2005283269A (ja) * | 2004-03-29 | 2005-10-13 | Toshiba Corp | 沸騰水型原子炉の過渡沸騰遷移監視システムおよび監視方法 |
KR20090021697A (ko) * | 2007-08-28 | 2009-03-04 | 비앤에프테크놀로지 주식회사 | 산업플랜트 설비의 종합 정지 여유도 감시방법 및 그 장치 |
KR20090024055A (ko) * | 2008-05-07 | 2009-03-06 | 비앤에프테크놀로지 주식회사 | 산업플랜트의 트립원인 추적 방법 및 그 시스템 |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113325797A (zh) * | 2021-06-11 | 2021-08-31 | 中山凯旋真空科技股份有限公司 | 控制设备的数据采集方法、装置、存储介质和电子设备 |
CN113325797B (zh) * | 2021-06-11 | 2022-07-12 | 中山凯旋真空科技股份有限公司 | 控制设备的数据采集方法、装置、存储介质和电子设备 |
Also Published As
Publication number | Publication date |
---|---|
KR101096793B1 (ko) | 2011-12-21 |
WO2011132900A3 (ko) | 2012-02-02 |
KR20110116378A (ko) | 2011-10-26 |
US20120271826A1 (en) | 2012-10-25 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2011132900A2 (ko) | 산업설비의 공정 여유도 감시 시스템용 데이터 수집 방법 및 그 저장매체 | |
CN108596229B (zh) | 在线异常的监测诊断方法和系统 | |
CN109146093B (zh) | 一种基于学习的电力设备现场勘查方法 | |
CN112098915B (zh) | 双母线分段接线下多台电压互感器继发性误差的评估方法 | |
WO2015026166A1 (ko) | 비정상 데이터 예측 방법 및 이 방법을 수행하기 위한 프로그램이 저장된 컴퓨터 판독가능한 저장매체 | |
CN104821789A (zh) | 一种光伏发电系统可靠性的检测方法 | |
CN115689524A (zh) | 一种数据中心机房电气设备预测性维护系统及方法 | |
CN114065875B (zh) | 基于大数据的电力电网故障识别系统 | |
CN115453267A (zh) | 一种电力信息系统故障诊断系统 | |
CN107607342A (zh) | 空调机房设备群的健康能效检测方法 | |
CN104317778A (zh) | 基于海量监测数据的变电设备故障诊断方法 | |
CN111563685B (zh) | 一种基于自联想核回归算法的发电设备状态预警方法 | |
CN112258013A (zh) | 一种换热器关键易结垢群组划分及能效状态监测方法 | |
CN108765192A (zh) | 一种基于大数据的变压器上层油温预测方法、系统及设备 | |
CN114970311A (zh) | 一种远端模块寿命预测模型的建立方法与寿命预测方法 | |
CN110321527B (zh) | 一种基于多元基础信息融合的数据有效性判别方法 | |
CN113672658A (zh) | 基于复相关系数的电力设备在线监测错误数据识别方法 | |
CN113431800A (zh) | 一种实时判别风机运行状态的方法 | |
WO2021238013A1 (zh) | 一种模拟量测值智能预测算法的自适应测试方法 | |
CN110516960A (zh) | 一种变电站继电保护设备的可靠性指标定量计算方法 | |
CN113159411B (zh) | 一种电网气象风险预警模型的检验方法及系统 | |
CN112417363B (zh) | 一种变电站的负载分析方法及系统 | |
Traxler | Fault detection of large amounts of photovoltaic systems | |
Hongda et al. | Research on Remote Diagnosis Model of Metering Point Anomaly of 10kV Distribution Line Based on AMI Data | |
Hu et al. | An Artificial Neural Network Model for Monitoring Real-Time Parameters and Detecting Early Warnings in Induced Draft Fan |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 11772188 Country of ref document: EP Kind code of ref document: A2 |
|
WWE | Wipo information: entry into national phase |
Ref document number: 13519070 Country of ref document: US |
|
WWE | Wipo information: entry into national phase |
Ref document number: 8428/DELNP/2012 Country of ref document: IN |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 11772188 Country of ref document: EP Kind code of ref document: A2 |