JP2007026134A - Abnormality decision device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an abnormality decision device capable of estimating a faulty element apparatus among a plurality of element apparatuses constituting a facility system. <P>SOLUTION: A data analysis device 100 sets a group of measurement values of generated power as an explanation variable x, sets each group of n kinds of measurement values obtained by measuring states or the like of the constituent apparatuses of a cogeneration system 300 as criterion variables y<SB>1</SB>, y<SB>2</SB>, y<SB>3</SB>, etc., y<SB>n</SB>, and performs n pieces of single regression analysis by the following expression: y<SB>1</SB>=β<SB>1</SB>x+α<SB>1</SB>, y<SB>2</SB>=β<SB>2</SB>x+α<SB>2</SB>, etc., y<SB>n</SB>=β<SB>n</SB>x+α<SB>n</SB>. Next, it is decided whether each the obtained regression coefficient β<SB>i</SB>is within a proper range or not. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a technique for monitoring an abnormality that occurs in equipment, and more particularly to a technique that can identify a location where an abnormality occurs.

In power plants and other mechanical / electrical equipment, it is necessary to monitor abnormalities in the components and perform stable operation.
Patent Document 1 discloses a technique for identifying an abnormal state of a gas turbine device used in a power plant or the like.
In general, when determining an abnormal state of an equipment device, a method is used in which various statistical analyzes are performed using data relating to past operating states of the equipment equipment to detect an abnormal value. For example, the following technique is used.

  About the state at the time of the operation | movement of 1 object component which comprises an equipment, the other factor considered to influence the state previously, for example, the other related structure arrange | positioned adjacent to the said object component Performs various analyzes using the information indicating the operation status of each component obtained when the facility equipment is operated normally, by associating it with the operation status of the element, and The normal value of the state of the target component that varies depending on the state is predicted. Next, during actual operation, the presence / absence of an abnormality is detected by comparing the actually obtained value indicating the state of the target component actually obtained with the predicted normal value.

However, according to this method of comparing the measured value with the predicted value, when there is a large variation in the measured data, it may react sensitively to sudden data that is not caused by an abnormality such as equipment failure. There is a problem that abnormalities such as aging deterioration of main parts constituting the device cannot be caught.
Therefore, for the data (objective variable) indicating the state during operation, a multiple regression analysis is performed in the specified time unit from the list of the objective variable and factor data (explanatory variable) that is considered to affect the objective variable, and the regression coefficient A determination method is adopted in which the presence or absence of an abnormality is detected depending on whether the rate of change is within an allowable range.

  When creating a list of objective variables and explanatory variables, it is normal that a specialist engineer decides a plan based on the physical configuration of equipment, etc., and repeatedly evaluates and reviews validity. For each target facility and target data item, an association with an influencing factor must be performed. That is, even if such association is performed for one facility device, there is a problem that versatility to other facility devices is low and the burden on a professional engineer is large.

As a solution to this, prepare explanatory variables x ₁ , x ₂ ,..., X _n for the objective variable y, and use the multiple regression analysis with Equation 1 to calculate the partial correlation coefficient by brute force. A standard method is used, and thus the selection of explanatory variables is automated.
y = β ₁ x ₁ + β ₂ x ₂ +... + β _n x _n + α (Formula 1)
Here, y is an objective variable, x ₁ , x ₂ ,..., X _n are explanatory variables, β ₁ , β ₂ ,... Β _n are regression coefficients, and α is , A constant term. The partial correlation coefficient is a correlation coefficient when the influence of other variables is removed when two specific correlation coefficients are assumed.

  However, according to the above prior art, the selection of the explanatory variable is automated using the partial correlation coefficient. However, since the regression model is focused on predicting the value of the objective variable y, a) There is a possibility that the final regression model itself (multiple regression equation) may be meaningless, and it does not mean that the relevance of many monitoring signals indicating the status of each component of the equipment is extracted. There is a first problem.

For example, consider a case where the equipment includes a cooler that exchanges heat between cooling water passing through the inside and outside air, and the objective variable is the cooling water outlet temperature of the cooler. Among many monitoring signals, it is easy to imagine that the cooling water inlet temperature is selected as an explanatory variable because the partial correlation coefficient shows a very high value. Here, it is assumed that the explanatory variable is one cooling water inlet temperature for easy understanding.
In this case, the regression equation _{y = β 1 x 1 + α} ( Equation 2)
It becomes. It is a clear fact that the inlet temperature and outlet temperature of the cooling water are related to each other. Here, an abnormality occurs somewhere in the equipment, and both the inlet temperature and outlet temperature of the cooling water tend to increase. Suppose. At this time, it is conceivable that the regression coefficient β ₁ does not change if the rate of increase is the same.

Similarly, even if multiple (two or more) explanatory variables are selected using a partial correlation coefficient, if the relation (correlation) is too deep, the possibility that it will change in the same manner when an abnormality occurs is too deep. There is no abnormality in the regression coefficient.
As described above, the method of creating a regression model using the partial correlation coefficient may be appropriate for the purpose of predicting the objective variable y. However, an abnormal value using the regression coefficient β may be used. It is not suitable for the method of making a judgment.

In addition, (b) by having a plurality of explanatory variables in one regression model, even if an abnormality is detected due to fluctuations in the regression coefficient, which component among the plurality of components constituting the equipment There is a second problem that it is difficult to estimate whether the abnormality is detected in (1).
For example, for the objective variable that is the regression model shown in Equation 3,
y = β ₁ x ₁ + β ₂ x ₂ + β ₃ x ₃ + α (Formula 3)
If any one of the three explanatory variables behaves so as to indicate an abnormal tendency, all of β ₁ , β ₂ , and β ₃ are affected and fluctuated. That is, there is a problem in that information on which component component of the facility device has caused a diagnosis result indicating the occurrence of such a defect is concealed, and it is difficult to estimate the component component in which the abnormality has occurred.
JP 2004-204780 A

  In order to solve the above-described problems, the present invention provides an abnormality determination device and an abnormality determination program capable of estimating which element device has an abnormality among a plurality of element devices constituting an equipment system. The purpose is to do.

  In order to achieve the above object, the present invention monitors a facility system that is constructed from a plurality of element devices and continuously obtains one result by continuous operation, and determines the presence or absence of an abnormality in each element device. An abnormality determination device that obtains a plurality of result measurement values indicating the results at a plurality of time points and a plurality of state measurement values related to the operating state of one element device measured at the plurality of time points, respectively. Assigning the obtained plurality of result measurement values to the first variable, assigning the obtained plurality of state measurement values to the second variable, and for the plurality of result measurement values and the plurality of state measurement values Analysis processing means for calculating the regression coefficient by performing a single regression analysis using a regression equation modeled by combining the first variable, the second variable, and the regression coefficient with an operator; Times A determination means for comparing the coefficient and an abnormal threshold value indicating an abnormal range to determine whether or not the regression coefficient is included in the abnormal range, and when it is determined that the regression coefficient is included in the abnormal range, the state measurement Output means for outputting abnormality information indicating the possibility of occurrence of an abnormality in an element device that is a value generation source.

According to this configuration, there is an excellent effect that the element device in which the abnormality has occurred can be specified.
Here, the acquisition unit further acquires a plurality of second state measurement values related to the operating states of the other element devices respectively measured at the plurality of time points, and the analysis processing unit further acquires The plurality of second state measurement values are assigned to a third variable, and the first variable, the third variable, and the second regression coefficient are assigned to the plurality of result measurement values and the plurality of second state measurement values. The second regression coefficient is calculated by performing a single regression analysis using a second regression equation formed by combining with an operator as a model, and the determination means further includes the calculated second regression coefficient, 2 is compared with a second abnormality threshold value indicating an abnormal range to determine whether or not the second regression coefficient is included in the second abnormal range, and the output means is further included in the second abnormal range. Is determined to be a source of the second state measurement value Second abnormality information indicating a possibility of abnormality in the element device may output a.

According to this configuration, similarly to the above, there is an excellent effect that it is possible to identify an element device in which an abnormality has occurred.
Here, the acquisition unit further acquires a plurality of environmental measurement values related to the environmental state of the element device respectively measured at the plurality of time points, and the analysis processing unit further includes the acquired plurality of the plurality of environmental measurement values. An environmental measurement value is assigned to a third variable, and instead of the single regression analysis, the first variable and the second variable for the plurality of result measurement values, the plurality of state measurement values, and the plurality of environment measurement values. The regression coefficient may be calculated by performing a regression analysis using a regression equation formed by combining the variable, the third variable, and the regression coefficient as an operator.

According to this configuration, it is possible to perform the abnormality determination with higher accuracy by reducing the influence of environmental variation factors.
Here, the abnormality determination device further includes a plurality of past state measurement values related to the measured operation state of the element device in a period before the plurality of state measurement values are acquired by the acquisition unit. An abnormality threshold setting means for calculating the abnormality threshold indicating the abnormality range may be included.

According to this configuration, the abnormal threshold value indicating the abnormal range is calculated using the measurement value obtained by actual measurement, so that it is possible to perform abnormality determination with higher accuracy according to the current situation.
Here, when the determination unit determines that it is not included in the abnormal range, the determination unit further determines whether or not the plurality of state measurement values fluctuate so as to satisfy a certain periodicity. If it is determined that the condition is not satisfied, it may be determined as abnormal.

According to this configuration, before the regression coefficient deviates from the abnormality threshold, it is possible to know the periodic disturbance as a sign of abnormality, and it is possible to detect abnormality early.
Here, the equipment system operates by switching a plurality of operation modes, and the abnormality determination device further obtains operation information indicating operation of each element device from the equipment system, and obtained operation information Operating mode determination means for determining an operation mode of the equipment system using the analysis means, the analysis processing means calculates the regression coefficient according to the determined operation mode, and the determination means is the regression coefficient And the output means outputs the abnormality information when it is determined that the output means is included in the abnormality range indicated by the abnormality threshold corresponding to the operation mode. Good.

According to this configuration, in an equipment system that operates by switching a plurality of operation modes, it is possible to make a more appropriate abnormality determination in each operation mode.
Further, the present invention is an abnormality determination device that is constructed from a plurality of element devices, monitors an equipment system that continuously obtains one result by continuous operation, and determines whether there is an abnormality in each element device, A plurality of result measurement values indicating the results at a plurality of time points, a plurality of first state measurement values related to the operating state of the first element device respectively measured at the plurality of time points, and at each of the plurality of time points An acquisition means for acquiring a plurality of second state measurement values related to the measured operating state of the second element device, and assigning the acquired plurality of result measurement values to an explanatory variable x and acquiring the plurality of first values acquired A state measurement value is assigned to the first objective variable y ₁ , the obtained second state measurement values are assigned to the second objective variable y ₂ , and the plurality of result measurement values and the plurality of first state measurement values are assigned The above theory A simple regression model y ₁ = β ₁ × x + α ₁ is formed by combining the bright variable x, the first objective variable y ₁ , the first regression coefficient β ₁ and the constant α ₁ with an operator. The regression analysis is performed to calculate the first regression coefficient β ₁ , and the explanatory variable x, the second objective variable y ₂ and the second objective variable y ₂ are calculated for the plurality of result measurement values and the plurality of second state measurement values. A single regression analysis using a second regression equation y ₂ = β ₂ × x + α ₂ as a model formed by combining two regression coefficients β ₂ and a constant α ₂ by an operator is performed, and the second regression coefficient β _{2 is applied.} and analysis processing means for calculating a first regression coefficient beta ₁ calculated, by comparing the first abnormality threshold value showing a first abnormal range, the first regression coefficient beta ₁ is included in the first abnormal range determines whether a second regression coefficient beta ₂ calculated, by comparing the second abnormality threshold indicating the second abnormality range, the second regression coefficient beta ₂ is, 2 In the element device that is the source of the state measurement value when it is determined that the determination means for determining whether or not any of the regression coefficients is included in any of the abnormal ranges Output means for outputting abnormality information indicating the possibility of occurrence of abnormality.

  According to this configuration, similarly to the above, there is an excellent effect that it is possible to identify an element device in which an abnormality has occurred.

1. Monitoring system 10
A monitoring system 10 as one embodiment according to the present invention will be described.
1.1 Configuration of Monitoring System 10 As shown in FIG. 1, the monitoring system 10 includes a data analysis device 100, a center server device 200, a cogeneration system 300, and a load facility 400.

The data analysis device 100, the center server device 200, and the cogeneration system 300 are connected to each other by a LAN (Local Area Network) 20.
The cogeneration system 300 supplies electric power and heat to the load facility 400. The load facility 400 is, for example, an air conditioner installed in a building, receives supply of electric power and heat, and performs cooling, heating, and ventilation in the building. The data analysis apparatus 100 monitors the operating state of the cogeneration system 300, and identifies an occurrence location when an abnormality occurs in the cogeneration system 300.

The center server device 200 stores information related to past operating conditions in the cogeneration system 300, and transmits the information to the data analysis device 100 as necessary.
1.2 Configuration of Cogeneration System 300 As shown in FIG. 2, the cogeneration system 300 includes a generator 303, an engine (engine) 307, a silencer 308, a cooling tower 309, a heat exchanger 310, a heat exchanger 311, and a fuel tank. Reference numeral 313 denotes a sensor 371, 372, 373, 374 such as a fuel flow meter, a thermometer, a pressure gauge, and a watt hour meter, and other devices.

The engine 307 sucks the fuel / air mixture supplied from the fuel tank 313 into the cylinder, compresses the mixture with the piston, ignites, burns and expands it, and reciprocates the piston. A rotational motion is transmitted to the rotation shaft 314.
A cooling water passage 335 is provided in the engine 307. The passage 335, the passage 336, the passage 337, and the passage 334 are connected in an annular shape, and the cooling water recirculates inside the passage provided in the annular shape. That is, the cooling water discharged from the engine 307 passes through the inside of the heat exchanger 311 through the passage 336, further passes through the inside of the heat exchanger 310 through the passage 337, and passes through the passage 334. Return to the engine 307. The heat exchanger 311 performs heat exchange between the cooling water supplied from the engine 307 and the hot water supplied to the load facility 400. In this way, hot water is supplied from the cogeneration system 300 to the load facility 400.

The rotating shaft 314 is coupled to the rotating shaft included in the generator 303. The generator 303 converts the rotational motion transmitted from the engine 307 via the rotating shaft 314 into electric energy, and uses the obtained electric power. Supply to the load facility 400.
Each sensor is always generated power, supply air pressure, fuel flow rate, supply air temperature, hot water engine outlet temperature, hot water engine inlet temperature, lubricating oil pressure, lubricating oil temperature, first cylinder exhaust temperature, second cylinder exhaust. The temperature, the third cylinder exhaust temperature, the fourth cylinder exhaust temperature, the fifth cylinder exhaust temperature, the sixth cylinder exhaust temperature, and the low temperature water engine outlet temperature are measured, and the measurement values obtained by measuring are measured time and item number. (To be described later) and transmitted to the data analyzing apparatus 100 via the LAN 20.

Other devices that constitute the cogeneration system 300 are well known, and thus description thereof is omitted here.
1.3 Configuration of Center Server Device 200 The center server device 200 includes an information storage unit 201, a control unit 202, an input unit 203, a display unit 204, and a transmission / reception unit 205 (not shown).

  Specifically, the center server device 200 is a computer system including a microprocessor, a ROM, a RAM, a hard disk unit, a communication unit, a display unit, a keyboard, a mouse, and the like. A computer program is stored in the RAM or the hard disk unit. Each component of the center server device 200 achieves its function by the microprocessor operating according to the computer program.

The transmission / reception unit 205 is connected to the data analysis device 100 via the LAN 20, and transmits / receives information between the data analysis device 100 and the control unit 202.
The information storage unit 201 stores a cumulative table having the same data structure as the cumulative analysis data table 135 (described later) shown in FIG. The cumulative analysis data table 135 and the cumulative table are different only in that they include different cumulative analysis information measured on different measurement dates.

  The control unit 202 receives an accumulation table acquisition request from the data analysis apparatus 100 via the LAN 20 and the transmission / reception unit 205. Upon receiving the acquisition request, the control unit 202 reads the accumulation table indicated by the acquisition request from the information storage unit 201 and transmits the read accumulation table to the data analysis apparatus 100 via the transmission / reception unit 205 and the LAN 20. To do.

The input unit 203 receives an instruction from the operator of the center server device 200 and outputs the received instruction to the control unit 202.
The display unit 204 displays various information under the control of the control unit 202.
1.4 Configuration of Data Analysis Device 100 As shown in FIG. 1, the data analysis device 100 includes an analysis processing unit 101, a threshold setting unit 102, an abnormality determination unit 103, a storage unit 104, an input / output control unit 105, and a report generation unit. 106, a mail generation unit 107, a graph generation unit 108, and a communication unit 109. The input / output control unit 105 is connected to the keyboard 110 and the monitor 111.

  Similar to the center server apparatus 200, the data analysis apparatus 100 is a computer system including a microprocessor, a ROM, a RAM, a hard disk unit, a communication unit, and the like. A computer program is stored in the RAM or the hard disk unit. Each component of the data analysis apparatus 100 achieves its function by the microprocessor operating according to the computer program.

(1) Storage unit 104
As shown in FIG. 1, the storage unit 104 includes an item information table 131, a variable correspondence table 132, break time information 133, reference period information 134, a cumulative analysis data table 135, a measurement information table 136, diagnosis period information 137, diagnosis items. An area for storing the information table 138, the current analysis data table 139, and the determination result table 140 is provided.

(Item information table 131)
The item information table 131 is a data table including information on items that are candidates for monitoring by the data analysis device 100 among the states of the devices constituting the cogeneration system 300 and the surrounding environment.
As shown in FIG. 3, the item information table 131 includes a plurality of item information, and each item information corresponds to an item to be monitored by the data analysis apparatus 100. Each item information includes an item number, an item name, and a unit.

Here, the item number is identification information for uniquely identifying an item corresponding to the item information including the item number, and is, for example, “0001”, “0002”,. The item number “0001” is fixedly used for an explanatory variable described later, and the other item numbers other than the item number “0001” are used for an objective variable described later.
The item name is a name given so that an item corresponding to the item information including the item name can be identified by human perception. For example, “generated power”, “supply pressure”, “fuel speed”,・ ・.

The unit represents a unit when the state indicated by the item corresponding to the item information including the unit is measured.
The item information table 131 is generated by the operator of the data analysis apparatus 100 and stored in the storage unit 104.
(Variable correspondence table 132)
The variable correspondence table 132 is a data table that includes an item that becomes an objective variable and an item that becomes an explanatory variable in association with each of a plurality of regression equations (described later) used in the data analysis apparatus 100. Here, as described above, each item is a target to be monitored by the data analysis apparatus 100 among the states of the devices constituting the cogeneration system 300 and the surrounding environment.

As an example, the variable correspondence table 132 includes the same number of pieces of corresponding item information as the plurality of regression equations, and each piece of corresponding item information corresponds to each regression equation. Each corresponding item information includes an objective variable item number and an explanatory variable item number.
Here, the objective variable item number is an item number for identifying an item to be the objective variable in the regression equation corresponding to the corresponding item information including the objective variable item number.

The explanatory variable item number is an item number for identifying an item to be an explanatory variable in the regression equation corresponding to the corresponding item information including the explanatory variable item number.
All the explanatory variable item numbers included in all the corresponding item information of the variable correspondence table 132 have the same contents, specifically “0001”. That is, all explanatory variables indicate the same item.

On the other hand, all target variable item numbers included in all corresponding item information have different contents, specifically, “0002”, “0003”,..., “0014”, “0015”. . That is, all objective variables show different items.
The variable correspondence table 132 is generated by the operator of the data analysis apparatus 100 and stored in the storage unit 104.

(Separation time information 133)
The delimiter time information 133 limits the monitoring and diagnosis by the data analysis device 100 to a specific one delimiter time of the day out of the states of the devices constituting the cogeneration system 300 and the surrounding environment. To do.
As shown in FIG. 5, for example, the break time information 133 includes a break start time 133a and a break end time 133b. The break start time 133a indicates the start time of the break time, and the break end time 133b Indicates the end time of the break time.

The break time information 133 is generated by the operator of the data analysis device 100 and stored in the storage unit 104.
Note that monitoring and diagnosis by the data analysis apparatus 100 may be limited to a specific plurality of break times in one day. That is, at this time, the break time information may include the start time and the end time of each of the plurality of break times. For example, the start time of the first partition time is “8:00”, the end time is “12:00”, the start time of the second partition time is “13:00”, and the end time is It may be “17:00”. The data analysis apparatus 100 may perform monitoring and diagnosis all day. At this time, the partition start time 133a may be set to “0:00” and the partition end time 133b may be set to “24:00”. Further, the break time may be not fixed but may vary depending on the season.

(Reference period information 134)
The reference period information 134 indicates a reference period used as a reference when the data analysis apparatus 100 performs a diagnosis. That is, the data analysis apparatus 100 monitors and diagnoses information included in the diagnosis period with reference to information included in the reference period indicated by the reference period information 134.

As an example, the reference period information 134 includes a reference start date 134a and a reference end date 134b, as shown in FIG. 6. The reference start date 134a indicates the start date of the reference period, and the reference end date 134b Indicates the end date of the reference period.
As an example, the reference start date 134a is “2003.01.01”, and the reference end date 134b is “2004.01.01”. This is because the reference period is
It is shown that it starts on 01/01/2003 and ends on 01/01/2004.

The reference period information 134 is generated by the operator of the data analysis apparatus 100 and stored in the storage unit 104.
(Cumulative analysis data table 135)
The cumulative analysis data table 135 is a data table including analysis results obtained by performing a single regression analysis described later using information acquired from the cogeneration system 300 in the past.

As shown in FIG. 7, the cumulative analysis data table 135 includes a plurality of pieces of cumulative analysis information, for example. Each cumulative analysis information is a single regression for each item to be monitored and diagnosed by the data analysis apparatus 100. Each cumulative analysis information includes an item number, a measurement date, and a parameter β.
Each item number is identification information indicating an item to be monitored and diagnosed by the data analysis device 100 as described above.

Each measurement date is the date on which the single regression analysis is performed for the item identified by the corresponding item number.
Each parameter β is a parameter obtained by performing the single regression analysis on the measurement date corresponding to the parameter β for the item identified by the item number corresponding to the parameter β.

The cumulative analysis data table 135 is stored in the storage unit 104.
(Measurement information table 136)
The measurement information table 136 indicates the states acquired from the cogeneration system 300 for the items to be monitored by the data analysis device 100 among the states of the devices constituting the cogeneration system 300 and the surrounding environment. It is a data table containing measurement information.

As an example, the measurement information table 136 includes a plurality of pieces of measurement information as illustrated in FIG. 8, and each piece of measurement information includes an item number, a measurement time, and a measurement value.
Each item number is identification information indicating an item to be monitored and diagnosed by the data analysis device 100 as described above.
Each measurement time indicates the time when the measurement value corresponding to the measurement time is obtained for each device constituting the cogeneration system 300 and its surrounding environment, by year / month / day / hour / minute / second.

Each measurement value is a value obtained by measuring the state of each device constituting the cogeneration system 300 and the surrounding environment.
Measurement information received as needed from the cogeneration system 300 via the LAN 20 and the communication unit 109 is written in the measurement information table 136.
(Diagnosis period information 137)
The diagnosis period information 137 indicates a diagnosis period to be diagnosed when the data analysis apparatus 100 performs a diagnosis. That is, the data analysis apparatus 100 monitors and diagnoses an event that occurs within the diagnosis period indicated by the diagnosis period information 137.

As an example, the diagnosis period information 137 includes a diagnosis start date 137a and a diagnosis end date 137b, as shown in FIG. 9, the diagnosis start date 137a indicates the start date of the diagnosis period, and the diagnosis end date 137b Indicates the end date of the diagnosis period.
As an example, the diagnosis start date 137a is “2003.01.01”, and the diagnosis end date 137b is “2003.12.31”. This is because the diagnosis period is
It shows that it starts on January 01, 2003 and ends on December 31, 2003.

The diagnosis period information 137 is generated by the operator of the data analysis apparatus 100 and stored in the storage unit 104.
(Diagnostic item information table 138)
The diagnosis item information table 138 is a data table that actually includes items to be monitored and diagnosed by the data analysis apparatus 100.

As an example, the diagnostic item information table 138 is a data table including a plurality of item numbers as shown in FIG. Each item number is identification information for identifying an item to be monitored and diagnosed by the data analysis apparatus 100 described above.
The plurality of item numbers included in the diagnostic item information table 138 are obtained by selecting some item numbers from the plurality of target variable item numbers included in the variable correspondence table 132. As an example, the diagnostic item information table 138 shown in FIG. 10 includes all item numbers selected from a plurality of target variable item numbers included in the variable correspondence table 132 shown in FIG.

The diagnostic item information table 138 is generated by the operator of the data analysis apparatus 100 and stored in the storage unit 104.
(Current analysis data table 139)
The current analysis data table 139 is a data table including analysis results obtained by performing single regression analysis described later using measurement information acquired from the cogeneration system 300 during the diagnosis period.

  Similar to the cumulative analysis data table 135, the current analysis data table 139 includes, as an example, a plurality of current analysis information as shown in FIG. 11, and each current analysis information is a target of monitoring and diagnosis by the data analysis apparatus 100. Is obtained by performing a single regression analysis for each day, and each current analysis information includes an item number, a measurement date, and a current parameter β.

Each item number is identification information indicating an item to be monitored and diagnosed by the data analysis apparatus 100.
Each measurement date is the date on which the single regression analysis is performed for the item identified by the corresponding item number. That is, the date when the measurement value was obtained.
Each current parameter β is a parameter obtained by performing the single regression analysis on the measurement date corresponding to the current parameter β for the item identified by the item number corresponding to the current parameter β.

Each current analysis information is written in the cumulative analysis data table 135 during diagnosis by the data analysis apparatus 100.
(Judgment result table 140)
The determination result table 140 is a data table including determination result information indicating a diagnosis result by the data analysis apparatus 100.

The determination result table 140 includes a plurality of pieces of determination result information as shown as an example in FIG. Each determination result information corresponds to the result of the diagnosis that is made for each item and each measurement date.
Each determination result information includes an item number, a measurement date, and determination information.
Each item number is identification information indicating an item to be monitored and diagnosed by the data analysis apparatus 100.

Each measurement date is the date on which the single regression analysis is performed for the item identified by the corresponding item number. That is, the date when the measurement value was obtained.
Each piece of determination information indicates the result of diagnosis made using single regression analysis. Each determination information takes a value of “0” or “1”, the value “0” indicates that the item is normal, and the value “1” indicates that the item is abnormal. .

(2) Analysis processing unit 101
(2.1) Acceptance of Analysis Instruction The analysis processing unit 101 receives an analysis instruction to start analysis processing from the operator of the data analysis apparatus 100 via the keyboard 110 and the input / output control unit 105.
Upon receiving the analysis instruction, the analysis processing unit 101 reads the diagnosis period information 137, the diagnosis item information table 138, the break time information 133, and the reference period information 134 from the storage unit 104.

(2.2) Acquisition of measurement values related to explanatory variables Next, the analysis processing unit 101 satisfies all of the following conditions 1, 2 and 3 from the measurement information table 136 stored in the storage unit 104. Extract measurement information.
(Condition 1) The item number included in the measurement information is “0001” (Condition 2) The measurement time included in the measurement information is determined by the diagnosis start date 137a and the diagnosis end date 137b included in the read diagnosis period information 137. Within the diagnosis period shown (Condition 3) The hour and minute of the measurement time included in the measurement information is within the break time indicated by the break start time 133a and the break end time 133b included in the read break time information 133 (Condition end)
As described above, the analysis processing unit 101 extracts all the measurement information including the measurement time within the diagnosis period and within the delimiter time and including the item number “0001”.

Next, the analysis processing unit 101 rearranges all the extracted measurement information in ascending order of measurement times included in the measurement information.
Here, for each day in the diagnosis period, a period formed by combining each day and the separation time is defined as an analysis period. Then, the same number of analysis periods as the number of days included in the diagnosis period can be assumed.

For example, if the diagnosis period is from January 1, 2003 to December 31, 2003, and the delimiter time is from 8:00 to 17:00, the same number of analyzes as the number of days included in the diagnosis period As an example, the period
“January 1, 2003 from 8:00 to 17:00”
“January 2, 2003 from 8:00 to 17:00”,
“January 3, 2003 from 8:00 to 17:00”
...
“December 31, 2003 from 8:00 to 17:00”.

Next, the analysis processing unit 101 divides the rearranged measurement information into a plurality of explanatory variable groups so that measurement information including measurement times included in the same analysis period is classified into the same explanatory variable group. Classify.
Here, the maximum number of explanatory variable groups to be generated is equal to the number of days included in the diagnosis period. Each explanatory variable group corresponds to one analysis period.

(2.3) Repetition for Each Objective Variable Next, the analysis processing unit 101 performs step 1 shown below for each item number included in the read diagnostic item information table 138, that is, for each item that becomes the objective variable. Repeat step 5. Here, each item number is referred to as a target item number.
(Step 1)
The analysis processing unit 101 extracts measurement information that satisfies all of the following conditions 4, 5 and 6 from the measurement information table 136 stored in the storage unit 104.

(Condition 4) The item number included in the measurement information matches the target item number (Condition 5) The measurement time included in the measurement information includes the diagnosis start date 137a and the diagnosis end included in the read diagnosis period information 137 Those within the diagnosis period indicated by the date 137b (Condition 6) The hour and minute of the measurement time included in the measurement information is indicated by the break start time 133a and the break end time 133b included in the read break time information 133 Within the delimiter time (condition end)
As described above, the analysis processing unit 101 extracts all measurement information including the measurement time within the diagnosis period and within the delimiter time and including the target item number. (End of Step 1)
(Step 2)
Next, the analysis processing unit 101 rearranges the plurality of extracted pieces of measurement information in ascending order of measurement times included in the measurement information, and further performs the same analysis on the rearranged measurement information in the same manner as described above. The measurement information including the measurement time included in the period is classified into a plurality of objective variable groups so that the measurement information is classified into the same objective variable group. (End of Step 2)
(Step 3)
Next, the analysis processing unit 101 repeats the following processing 1 to processing 2 for each analysis period for the same number of analysis periods as the number of days included in the diagnosis period.

(Process 1)
The analysis processing unit 101 performs robust regression analysis using the measurement information included in the objective variable group corresponding to the analysis period as an objective variable and the measurement information included in the explanatory variable group corresponding to the analysis period as an explanatory variable. One regression coefficient β _ij and one constant α _ij are obtained.
y _i = β _ij x + α _ij
Here, y _i is an objective variable corresponding to the target item number, x is an explanatory variable corresponding to the item number “0001”, β _ij is a regression coefficient, and α _ij is a constant term. It is. Further, i is the target item number, and j is identification information that uniquely indicates the analysis period among the same number of analysis periods as the number of days included in the diagnosis period. (End of Process 1)
Robust regression analysis is a regression analysis that eliminates the influence of outliers and provides an analysis result close to the true distribution even when there are a small number of outliers in the observed values to be analyzed. Is the law.

On the other hand, the least square method is sensitive to deviations from standard assumptions of linear regression, and the regression coefficient is greatly affected by one outlier.
For example, as shown in FIG. 13, in a coordinate system 741 configured such that an explanatory variable x is assigned to the horizontal axis 742 and an objective variable y is assigned to the vertical axis 743, a plurality of measurements composed of explanatory variables and objective variables. A value group 745 and a straight line 744 obtained by applying a normal regression analysis, for example, a least square method to a measured value group 745 including explanatory variables and objective variables are plotted.

  For example, as shown in FIG. 14, in a coordinate system 751 configured by assigning an explanatory variable x to the horizontal axis 752 and assigning an objective variable y to the vertical axis 753, a plurality of explanatory variables and objective variables are included. Measurement value group 755, one sudden abnormal value 756 composed of explanatory variables and objective variables, and a straight line 754 obtained by subjecting the measurement value group 745 and abnormal value 756 to normal regression analysis. And are plotted. In this figure, due to the influence of the abnormal value 756, the slope of the straight line 754 is significantly different from the slope of the straight line 744.

  The robust regression analysis described above solves this problem. The robust regression analysis is a known technique and will not be described here. For robust regression analysis, see “Finite sample breakdown of M- and P-estimators” Huber, PJ, The Annals of Statistics, Vol. 12, No1, 119-126 (1984) and “Robust Linear Regression” by Motoki Kaneko. It is stated in.

(Process 2)
The analysis processing unit 101 associates the target item number, the measurement date constituting the analysis period, and the obtained regression coefficient β _ij to the current analysis data table 139 as the item number, measurement date, and current parameter β. Write. (End of Process 2)
(Step 3 end)
(Step 4)
Next, the analysis processing unit 101 instructs the threshold setting unit 102 to set an upper limit value and a lower limit value of the parameter β. (Step 4 end)
(Step 5)
Next, the analysis processing unit 101 instructs the abnormality determination unit 103 to determine whether the parameter β is abnormal. (Step 5 end)
(2.4)
Prior to the analysis processing, the analysis processing unit 101 outputs a cumulative table acquisition request to the center server device 200 via the communication unit 109.

(3) Threshold setting unit 102
The threshold setting unit 102 sets an upper limit value and a lower limit value of the parameter β in accordance with an instruction from the analysis processing unit 101 as follows.
The threshold value setting unit 102 extracts all measurement information that satisfies the following conditions 7 and 8 from the cumulative analysis data table 135.

(Condition 7) Item number included in cumulative analysis information matches target item number (Condition 8) Measurement start date included in cumulative analysis information is reference start date included in read reference period information 134 Within the reference period indicated by 134a and reference end date 134b (condition end)
Next, the threshold value setting unit 102 uses the parameters β included in all the extracted measurement information as described below, and 99% of the parameters β exist below the values. The upper limit value and the lower limit value of the parameter β are calculated by setting the value as the upper limit value and setting 99% of the parameter β as the lower limit value.

Specifically, the upper limit value is calculated as follows.
(I) The threshold setting unit 102 counts the total number of parameters β included in all the extracted measurement information, and extracts the maximum value of the parameter β and the minimum value of the parameter β.
(Ii) A section from the minimum value to the maximum value is divided into a plurality of continuous small sections. The width of each small section is the same. An appropriate number of small sections is about 50 to 100. A small section including the minimum value is referred to as a minimum value small section, and a small section including the maximum value is referred to as a maximum value small section.

(Iii) For each small section, the number of small sections of the parameter β included in each small section is counted.
(Iv) For each small section, the cumulative number of parameters β included from the smallest small section to the small section is counted by adding the number of small sections of all the small sections from the smallest small section to the small section. .

(V) A small section that satisfies the following condition 9 is selected.
(Condition 9) Cumulative number / total number of small sections / small section corresponding to cumulative number satisfying 100 = 99 (%) (End of condition)
(Vi) A value representative of the selected small section, for example, a median value in the small section is set as the upper limit value.

The same applies to the lower limit value.
(4) Abnormality determination unit 103
The abnormality determination unit 103 determines whether or not there is an abnormality in the parameter β in accordance with an instruction from the analysis processing unit 101 as follows.
The abnormality determination unit 103 repeats the following processes (i) to (ii) for each current analysis information included in the current analysis data table 139, in other words, for each current parameter β.

(I) The abnormality determination unit 103 sequentially reads one piece of current analysis information from the current analysis data table 139, and extracts a current parameter β from the read current analysis information. (I) End (ii) The extracted current parameter β is compared with the upper limit value and the lower limit value.
When the lower limit value ≦ current parameter β ≦ upper limit value is satisfied, the target item number, the measurement date, and normal information indicating normality are written to the determination result table 140.

When the lower limit value ≦ the current parameter β ≦ the upper limit value is not satisfied, the target item number, the measurement date, and the abnormality information indicating the abnormality are written in the determination result table 140. (Ii) End (5) Communication unit 109
The communication unit 109 is connected to each sensor included in the center server device 200 and the cogeneration system 300 via the LAN 20.

The communication unit 109 acquires measurement information from each sensor and writes the acquired measurement information in the measurement information table 136.
Further, the communication unit 109 receives the accumulation table from the center server device 200 via the LAN 20, and adds the received accumulation table to the accumulation analysis data table 135 and writes it.

(6) Report generation unit 106, mail generation unit 107, and graph generation unit 108
The report generation unit 106 reads the determination result information from the determination result table 140, reads the item information from the item information table 131, and generates a report including the read determination result information and the read item information. The report includes abnormality information indicating a possibility of occurrence of abnormality in each device constituting the cogeneration system 300.

  The mail generation unit 107 reads the determination result information from the determination result table 140, reads the item information from the item information table 131, generates an e-mail including the read determination result information and the read item information, and generates the generated e-mail. Then, the data is transmitted to a predetermined destination via the communication unit 109. The e-mail includes abnormality information indicating a possibility of occurrence of abnormality in each device constituting the cogeneration system 300.

The graph generation unit 108 reads the determination result information from the determination result table 140, reads the item information from the item information table 131, and generates a graph including the read determination result information and the read item information. The graph includes abnormality information indicating a possibility of occurrence of abnormality in each device configuring the cogeneration system 300.
(7) Input / output control unit 105, keyboard 110, and monitor 111
The keyboard 110 and the input / output control unit 105 receive an operation instruction and information input from an operator of the data analysis apparatus 100, and analyze the received operation instruction and information as an analysis processing unit 101, a report generation unit 106, a mail generation unit 107, a graph The data is output to the generation unit 108.

The input / output control unit 105 and the monitor 111 display the report and graph generated by the report generation unit 106 and the graph generation unit 108. Various information is displayed according to instructions from the analysis processing unit 101 and the abnormality determination unit 103.
1.5 Operation of Monitoring System 10 The operation of the monitoring system 10 will be described with reference to the flowcharts shown in FIGS.

When the analysis processing unit 101 receives an analysis instruction to start the analysis processing, the analysis processing unit 101 reads the diagnosis period information 137 and the diagnosis item information table 138 from the storage unit 104 (step S101), and reads the delimiter time information 133 (step S102). ), The reference period information 134 is read (step S103).
Next, the analysis processing unit 101 determines from the measurement information table 136 stored in the storage unit 104 measurement information (corresponding to the explanatory variable) that is within the diagnosis period and within the delimiter period and whose item number is “0001”. Are extracted (step S104), all the extracted measurement information is rearranged in ascending order of the measurement time included in the measurement information (step S105), and the rearranged measurement information is included in the same analysis period. The measurement information including the measured time is classified into a plurality of explanatory variable groups so as to be classified into the same explanatory variable group (step S106).

Next, in step S107 to step S123, the analysis processing unit 101 repeats step S108 to step S122 for each item number included in the read diagnostic item information table 138.
The analysis processing unit 101 extracts all the measurement information including the measurement time within the diagnosis period and within the delimiter time and including the target item number (step S108). Next, the plurality of extracted measurement information is measured. The measurement information included in the information is rearranged in ascending order (step S109), and the measurement information including the measurement times included in the same analysis period is classified into the same objective variable group. Then, it is classified into a plurality of objective variable groups (step S110).

Next, in step S111 to step S114, the analysis processing unit 101 repeats step S112 to step S113 for each analysis period for the same number of analysis periods as the number of days included in the diagnosis period.
The analysis processing unit 101 performs robust regression analysis using the measurement information included in the objective variable group corresponding to the analysis period as an objective variable and the measurement information included in the explanatory variable group corresponding to the analysis period as an explanatory variable. One regression coefficient β _ij and one constant α _ij are obtained (step S112), and then the target item number, the measurement date constituting the analysis period, and the obtained regression coefficient β _ij are associated with each other. The item number, measurement date, and current parameter β are written to the current analysis data table 139 (step S113).

Next, the threshold setting unit 102, from the cumulative analysis data table 135, the item number included in the cumulative analysis information matches the target item number, and all the measurement dates included in the cumulative analysis information are within the reference period. (Step S115), and using all the extracted measurement information, the upper limit value and the lower limit value of the parameter β are calculated (step S116).
Next, in step S117 to step S122, the abnormality determination unit 103 repeats steps S118 to S121 for each current analysis information included in the current analysis data table 139, that is, for each current parameter β.

  The abnormality determination unit 103 sequentially reads one piece of current analysis information from the current analysis data table 139, extracts the current parameter β from the read current analysis information (step S118), extracts the extracted current parameter β and the upper limit value. If the lower limit value ≦ current parameter β ≦ upper limit value is satisfied (YES in step S119), the target item number, the measurement date, and normal information indicating normality are determined. When the result table 140 is written (step S121) and the lower limit value ≦ current parameter β ≦ upper limit value is not satisfied (NO in step S119), the target item number, the measurement date, and abnormality information indicating an abnormality Is written in the determination result table 140 (step S120).

1.6 Summary In the monitoring system 10 described above, a group of measurement values obtained by measuring the generated power that is the output of the cogeneration system 300 is set as an explanatory variable x within a predetermined diagnosis period, and the cogeneration system 300 is The objective variables y ₁ , y ₂ , y ₃ ,... For each group of n types of measurement values obtained by measuring the state of the component equipment to be configured or the state of the surrounding environment (n types of states). As _n 1, n one-to-one relationships (n) of the explanatory variable x and the objective variable y _i (i = 1, 2, 3,..., n) are modeled on the following regression equations. Perform a single regression analysis.

y ₁ = β ₁ x + α ₁
y ₂ = β ₂ x + α ₂
y ₃ = β ₃ x + α ₃
...
y _n = β _n x + α _n
Here, β ₁ , β ₂ , β ₃ ,..., Β _n are regression coefficients, respectively, and α ₁ , α ₂ , α ₃ ,..., Α _n are constant terms, respectively. .

Thus, in the above embodiment, in each of the n simple regression analyses, the final output of the system is selected as the explanatory variable x, and other measurement values obtained from the system are selected as the objective variable. Yes.
The correspondence between the explanatory variable x and the objective variable y _i measured and used in the cogeneration system 300 is “supply air pressure” 601 and “generated power” 602 as shown in FIG. “Fuel flow rate” 603 and “generated power” 604, “supply air temperature” 605 and “generated power” 606, as shown in FIG.

Next, the monitoring system 10 determines whether or not the regression coefficient β ₁ obtained in this way is within an appropriate range obtained from the results of single regression analysis obtained in a past predetermined reference period. If it is within the appropriate range, it is determined that there is no abnormality in the component device that is the measurement target of the “supply pressure” 601. If it is outside the appropriate range, it is determined that there is an abnormality in the component device that is the measurement target of the “supply pressure” 601.

Monitoring system 10 obtained regression coefficient beta _2, beta _3, · · ·, for beta _n even, in the same manner, and determines whether within the appropriate range corresponding to each of the regression coefficients, the appropriate range If it is within the range, it is determined that there is no abnormality in the corresponding component device.
Thus, in the monitoring system 10, when an abnormality occurs in the system to be monitored and diagnosed, there is an excellent effect that a component that causes the abnormality can be specified.

  However, in the conventional method using the regression coefficient by multiple regression analysis, the explanatory variable and the objective variable are selected closely, so there is a possibility that they will affect each other and the abnormal value cannot be detected. There is. Even when fluctuations appear in the regression coefficient, it is conceivable that the extraction of explanatory variables is automated and multiple regression analysis is used, so that it is difficult to know which part of the regression coefficient fluctuation indicates an abnormality.

  In addition, the monitoring system 10 is very good at capturing an aged deterioration as shown below. This is because the final output of the system is adopted as an explanatory variable. For this reason, when a trouble that does not affect the final output occurs in the system, an abnormal regression coefficient can be detected at the first stage when the abnormality starts to occur.

Further, as described above, since robust regression analysis is employed, it is excellent in that false alarms due to sudden abnormalities can be reduced.
2. Modification (1)
A monitoring system 10a (not shown) as a modified example of the monitoring system 10 will be described. The monitoring system 10a has the same configuration as the monitoring system 10, and includes a data analysis device 100a, a center server device 200, a cogeneration system 300a, and a load facility 400. Below, it demonstrates centering on the difference with the monitoring system 10. FIG.

  In the cogeneration system 300a, a plurality of thermometers are installed in the room and outside in order to measure the temperature inside the room where the components constituting the cogeneration system 300a are installed and the outside air temperature. Each thermometer is connected to the LAN 20. Each thermometer is assigned an item number, and each thermometer always measures the ambient temperature, and the item number corresponding to each thermometer and the time when the temperature was measured (year / month / day / hour / minute / second) ) And the measured temperature are transmitted to the data analysis apparatus 100 a via the LAN 20.

The data analysis device 100a receives the item number, the time when the temperature was measured, and the measured temperature from the cogeneration system 300a via the LAN 20, and the received item number, time, and temperature are measured every time the information is received in the measurement information table. Write to 136.
In the data analysis apparatus 100a, when the objective variable is data not related to temperature, single regression analysis is used as in the data analysis apparatus 100. On the other hand, if the objective variable is data related to temperature, add environmental parameters to the temperature data that is not related to equipment operation, such as room temperature and outside temperature, as the second explanatory variable. Use multiple regression analysis.

The data analysis apparatus 100a stores a variable correspondence table 610 in the storage unit 104 in place of the variable correspondence table 132 as shown in FIG. 19 as an example.
The variable correspondence table 610 includes the same number of corresponding item information as a plurality of regression equations, and each corresponding item information corresponds to each regression equation. Each corresponding item information includes an objective variable item number and an explanatory variable item number, or includes an objective variable item number, an explanatory variable item number, and a second explanatory variable purpose number.

Corresponding item information including the objective variable item number and the explanatory variable item number relates to a component in which the objective variable is not related to room temperature or outside temperature, and the objective variable item number, the explanatory variable item number, and the second explanatory variable purpose Corresponding item information including a number relates to a component in which the objective variable relates to room temperature or outside temperature.
Here, since the objective variable item number and the explanatory variable item number are as described above, description thereof will be omitted.

The second explanatory variable item number is an item number for identifying a thermometer that measures room temperature or outside air temperature.
The analysis processing unit 101 of the data analysis apparatus 100a uses a group of measurement values obtained by measuring the output of the cogeneration system 300a within the predetermined diagnosis period for the corresponding item information including the second explanatory variable as an explanatory variable. and x _1, the group of measured temperature as explanatory variables x _2, a group of measurement values obtained by measuring the state of a constituent device which constitutes the cogeneration system 300 as an objective variable y, the explanatory variable x ₁ A multiple regression analysis is performed on the one-to-one relationship between the explanatory variable x ₂ and the objective variable y _i by the following equation.

y = β ₁ x ₁ + β ₂ x ₂ + α
Here, β ₁ and β ₂ are regression coefficients, and α is a constant term.
Here, it is assumed that only one outside air temperature measured by one thermometer among the plurality of thermometers of the cogeneration system 300 is used.
Of the regression coefficients thus obtained, the analysis processing unit 101 uses β ₁ to determine abnormality. The abnormality determination method is the same as that of the data analysis apparatus 100.

Note that the analysis processing unit 101 performs the same single regression analysis as the data analysis apparatus 100 on the corresponding item information that does not include the second explanatory variable.
In the monitoring system 10a, the influence of seasonal variation is smoothed by adding the temperature, which is an environmental parameter, to the regression equation including the objective variable related to temperature.
When the objective variable is related to temperature, in the monitoring system 10, for example, as shown in FIG. 20, a coordinate system 701 is configured such that the passage of time is assigned to the horizontal axis 702 and the regression coefficient is assigned to the vertical axis 703. , A plurality of regression coefficients 706 over time are plotted. As shown in this figure, the plurality of regression coefficients 706 are fluctuating between an upper limit value 704 and a lower limit value 705 while substantially drawing a sine curve as shown in this figure. That is, it fluctuates so as to be close to the values 704 and 705 respectively in the time corresponding to summer and winter, and fluctuates so as to be close to the median of the values 704 and 705 in the time corresponding to spring and autumn. is doing. This is because the objective variable is related to temperature as described above.

On the other hand, in the monitoring system 10a, for example, as shown in FIG. 21, in a coordinate system 721 configured by assigning the passage of time to the horizontal axis 722 and assigning the regression coefficient β ₁ to the vertical axis 723, A plurality of associated regression coefficients 726 are plotted. As shown in this figure, a plurality of regression coefficients 726 exist between an upper limit value 724 and a lower limit value 725, but a sine curve as shown in FIG. 20 is not drawn. β ₂ x ₂ is because they absorb the impact of temperature.

As described above, in the monitoring system 10a, the influence of the seasonal variation can be smoothed by adding a parameter related to the environment to the objective variable related to temperature in the regression equation, and is set by the threshold setting unit 102. Since the range of the upper limit value and the lower limit value can be kept small, it is possible to increase the detection accuracy of the component in which an abnormality occurs.
3. Modification (2)
Even if the regression coefficient of the above-mentioned single regression analysis result does not deviate from the upper limit value or the lower limit value, it may be detected as abnormal when the periodic fluctuation collapses as in the following modification. Good.

  A monitoring system 10b (not shown) as a modification of the monitoring system 10 will be described. The monitoring system 10b has a configuration similar to that of the monitoring system 10, and includes a data analysis device 100b, a center server device 200, a cogeneration system 300, and a load facility 400. Below, it demonstrates centering on the difference with the monitoring system 10. FIG.

The data analysis apparatus 100b makes the following determination using, for example, FFT (Fast Fourier Transform).
The data analysis apparatus 100b sets the period of the periodic operation indicated by each component device included in the cogeneration system 300 as a time domain for FFT analysis.
Each time a single regression analysis is performed and a regression coefficient is obtained, an FFT is applied to a plurality of regression coefficients obtained so far including the regression coefficient in the same manner as described above.

As a result, the result obtained by the current FFT is compared with the result obtained by the previous FFT, and if the change in the frequency at which the maximum value is obtained is equal to or greater than a predetermined value, the data analysis apparatus 100b has a periodicity. Anomaly is reported as collapsed.
Next, the operation of the data analysis apparatus 100b will be described using a flowchart. The operation by the data analysis apparatus 100b is substantially the same as the operation by the data analysis apparatus 100 shown in the flowcharts of FIGS. 15 to 17, and the differences are shown in FIG. Of the steps of FIGS. 15 to 17, steps S115 to S122 may be replaced with the steps shown in FIG.

  The threshold value setting unit 102 of the data analysis device 100b reads from the cumulative analysis data table 135 that the item number included in the cumulative analysis information matches the target item number, and the measurement date included in the cumulative analysis information is within the reference period. All of the measurement information is extracted (step S115), and the analysis processing unit 101 performs FFT on the past analysis result and the current analysis result (step S302), and each obtained frequency component (maximum value is obtained). It is determined whether or not there is a significant difference in the frequency) (step S303). If there is a significant difference (step S303), the target item number, the measurement date, and the abnormality are indicated. The abnormality information is written into the determination result table 140 (step S120). If there is no significant difference (step S303), the target item number, measurement date, and normal information indicating normality are written to the determination result table 140 (step S121).

  For example, as shown in FIG. 23, in a coordinate system 731 configured by assigning the passage of time to the horizontal axis 732 and assigning the regression coefficient β to the vertical axis 733, a plurality of regression coefficients 736 and 737 with the passage of time. Is plotted. As shown in this figure, the plurality of regression coefficients 736 and 737 exist between an upper limit value 734 and a lower limit value 735. Of these, a sine curve is drawn for the plurality of regression coefficients 736, but the fluctuations of the plurality of regression coefficients 737 are not sine curves. In such a case, the data analysis apparatus 100b can detect the abnormal fluctuation of the plurality of regression coefficients 737.

As described above, the data analysis apparatus 100b regards the fluctuations per unit time of a plurality of time-series regression coefficients as being constant, and determines that the fluctuations are not normal.
As described above, it is possible to detect a phenomenon leading to an abnormality that appears in the deviation from the upper limit and the lower limit without overlooking, and the accuracy of abnormality detection can be improved.
In addition, wavelet analysis of time-frequency analysis, which is a well-known analysis method, or autoregressive AR (autoregressive) model or ARMA (autoregressive moving average) model, etc. A similar analysis can be performed using the captured analysis.

4). Modification (3)
A monitoring system 10c (not shown) as a modification of the monitoring system 10 will be described. The monitoring system 10c has the same configuration as the monitoring system 10, and includes a data analysis device 100c instead of the data analysis device 100, and includes an air conditioning system 500 instead of the cogeneration system 300 and the load facility 400. ing.

The air conditioning system 500 is an air conditioner that employs a heat pump heat source. As shown in FIG. 24, the air conditioning system 500 includes an air conditioner 507, a heat source 570, and a cooling tower 506. The air conditioner 507 includes a fan 508 and a coil 509. Coil 509 serves as both a cooling coil and a heating coil.
The pumps 571, 573, and 572 send water (secondary refrigerant) from the cooling tower 506 to the heat source 570 or from the heat source 570 to the air conditioner 507.

  The heat source 570 includes a four-way switching valve 505 inside, and switches the flow of water among the cooling tower 506, the heat source 570, and the air conditioner 507. The arrows in the figure indicate the direction in which water (secondary refrigerant) or refrigerant flows. Specifically, at the time of cooling, water sent from the cooling tower 506 is sent to the condenser 501, the water warmed by the condenser 501 is returned to the cooling tower 506, and the water cooled by the evaporator 503 is sent to the air conditioner. The valve is opened and closed so that the water warmed by the air conditioner 507 is returned to the evaporator 503. During heating, the water sent from the cooling tower 506 is sent to the evaporator 503, the water cooled by the evaporator 503 is returned to the cooling tower 506, and the water warmed by the condenser 501 is sent to the air conditioner 507. In the harmony machine 507, the cooled water is returned to the condenser 501.

The air conditioning system 500 includes a plurality of thermometers that measure the temperature of each component and the surrounding temperature, a measuring instrument that measures the wind speed from the air conditioner 507 to the room, and power that measures power consumed in the air conditioning system 500. Sensors 530, 531, 532, 535, 536, 537, 539, 540, and 541 are provided.
Each sensor transmits an item number for identifying each sensor, a measurement time (year / month / day / hour / minute / second), and a measurement value to the data analysis apparatus 100 c via the LAN 20.

The data analysis device 100c has the same configuration as the data analysis device 100. Here, the difference from the data analysis apparatus 100 will be mainly described.
The storage unit 104 stores an item information table 131 c instead of the item information table 131. The item information table 131c has a data structure similar to that of the item information table 131. For example, as shown in FIG. 25, the item information table 131c is composed of a plurality of item information. Name and unit, or item number, item name, unit and associated mode.

The item number, item name, and unit are as described above.
The item names are, for example, “indoor temperature”, “indoor unit inlet temperature”, “indoor unit outlet temperature”, “indoor unit wind speed”, “evaporator inlet temperature”, “evaporator outlet temperature”, “condenser inlet”. These are “temperature”, “condenser outlet temperature”, “main body electric energy”, and “set temperature”.
The related mode indicates whether the corresponding item is related to a mode during operation of the air conditioning system 500, specifically, a cooling operation mode or a heating operation mode. The related modes “1” and “2” correspond to the cooling operation mode and the heating operation mode, respectively.

As shown in this figure, item information 591, 592, 593, and 594 include item numbers, item names, units, and related modes. Other item information includes an item number, an item name, and a unit.
Since the item information 591 and 592 each include the related mode “1”, the item information 591 and 592 are items related to the cooling operation mode. Since the item information 593 and 594 each include the related mode “2”, the item information 593 and 594 are items related to the heating operation mode.

The analysis processing unit 101 included in the data analysis apparatus 100c operates as follows.
The operation of the data analysis apparatus 100c is almost the same as the operation of the data analysis apparatus 100 shown in the flowcharts of FIGS. 15 to 17, and the differences are shown in FIGS. Of steps in FIGS. 15 to 17, steps S117 to S122 may be replaced with the steps shown in FIGS. 26 and 27.

In steps S117 to S122, the abnormality determination unit 103 of the data analysis device 100c repeats steps S118 to S121 for each current analysis information included in the current analysis data table 139, that is, for each current parameter β.
The abnormality determination unit 103 sequentially reads one piece of current analysis information from the current analysis data table 139, extracts the current parameter β from the read current analysis information (step S118), extracts the extracted current parameter β and the upper limit value. If the lower limit value ≦ current parameter β ≦ upper limit value is satisfied (YES in step S119), the target item number, the measurement date, and normal information indicating normality are determined. Write to the result table 140 (step S121).

On the other hand, when the lower limit value ≦ the current parameter β ≦ the upper limit value is not satisfied (NO in step S119), the abnormality determination unit 103 sets the set temperature set by the user of the air conditioning system 500 and the power consumption of the air conditioning system 500. Is acquired (step S201).
Next, the abnormality determination unit 103 determines that 18.0 degrees (Celsius) ≦ set temperature <26.0 degrees and 0.1 (kWh) ≦ main body power amount ≦ 9.0 (kWh). (Step S202), the mode is set to “1” on the assumption that it is a cooling operation (Step S205).

In addition, when 26.0 degrees ≤ set temperature ≤ 30.0 degrees and 0.1 (kWh) ≤ body power ≤ 9.0 (kWh) (step S203), the heating operation is performed. Accordingly, the mode is set to “2” (step S206).
In other cases, the mode is set to “0” (step S204).
Next, when the mode is “1” (step S207), the abnormality determination unit 103 determines that the target item number is “0005” or “0006” (step S208), and the mode is “2”. ”(Step S207), if the target item number is“ 0007 ”or“ 0008 ”(step S209), the upper limit of the parameter β is extracted from the extracted cumulative analysis information in the same manner as the data analysis apparatus 100. A value and a lower limit value are calculated (step S210).

Next, if the upper limit value ≦ the current parameter β ≦ the lower limit value (step S211), the target item number, the measurement date, and normal information indicating normality are written to the determination result table 140 (step S121). .
On the other hand, if the upper limit value ≦ current parameter β ≦ lower limit value is not satisfied (step S211), the target item number, the measurement date, and the abnormality information indicating abnormality are written to the determination result table 140 (step S120).

  Further, when the mode is “1” (step S207), the abnormality determining unit 103 does not set the target item numbers to “0005” and “0006” (step S208), and the mode is “2”. (Step S207), if the target item number is not “0007” or “0008” (step S209), the target item number, the measurement date, and normal information indicating normality are obtained. It writes in the determination result table 140 (step S121).

In addition, when the mode is “0” (step S207), the abnormality determination unit 103 writes the target item number, the measurement date, and abnormality information indicating abnormality in the determination result table 140 (step S207). S120).
5. Other Modifications Although the present invention has been described based on the above-described embodiment, it is needless to say that the present invention is not limited to the above-described embodiment. The following cases are also included in the present invention.

  (1) Specifically, each of the above devices is a computer system including a microprocessor, a ROM, a RAM, a hard disk unit, a display unit, a keyboard, a mouse, and the like. A computer program is stored in the RAM or the hard disk unit. Here, the computer program is configured by combining a plurality of instruction codes indicating instructions for the computer in order to achieve a predetermined function. Each device achieves its function by the microprocessor operating according to the computer program. That is, the microprocessor reads each instruction included in the computer program one by one, decodes the read instruction, and operates according to the decoding result.

  (2) A part or all of the constituent elements constituting each of the above devices may be constituted by one system LSI (Large Scale Integration). The system LSI is a super multifunctional LSI manufactured by integrating a plurality of components on one chip, and specifically, a computer system including a microprocessor, a ROM, a RAM, and the like. . A computer program is stored in the RAM. The system LSI achieves its functions by the microprocessor operating according to the computer program.

In addition, each part of the constituent elements constituting each of the above devices may be individually made into one chip, or may be made into one chip so as to include a part or all of them. Further, although it is referred to as LSI here, it may be referred to as IC, system LSI, super LSI, or ultra LSI depending on the degree of integration.
Further, the method of circuit integration is not limited to LSI's, and implementation using dedicated circuitry or general purpose processors is also possible. An FPGA (Field Programmable Gate Array) that can be programmed after manufacturing the LSI or a reconfigurable processor that can reconfigure the connection and setting of circuit cells inside the LSI may be used.

  (3) Part or all of the constituent elements constituting each of the above devices may be configured from an IC card that can be attached to and detached from each device or a single module. The IC card or the module is a computer system including a microprocessor, a ROM, a RAM, and the like. The IC card or the module may include the super multifunctional LSI described above. The IC card or the module achieves its function by the microprocessor operating according to the computer program. This IC card or this module may have tamper resistance.

(4) The present invention may be the method described above. Further, the present invention may be a computer program that realizes these methods by a computer, or may be a digital signal composed of the computer program.
The present invention also provides a computer-readable recording medium such as a flexible disk, hard disk, CD-ROM, MO, DVD, DVD-ROM, DVD-RAM, BD (Blu-ray Disc). ), Recorded in a semiconductor memory or the like. Further, the present invention may be the computer program or the digital signal recorded on these recording media.

Further, the present invention may transmit the computer program or the digital signal via an electric communication line, a wireless or wired communication line, a network represented by the Internet, a data broadcast, or the like.
The present invention may be a computer system including a microprocessor and a memory, wherein the memory stores the computer program, and the microprocessor operates according to the computer program.

In addition, the program or the digital signal is recorded on the recording medium and transferred, or the program or the digital signal is transferred via the network or the like, and is executed by another independent computer system. It is good.
(5) The above embodiment and the above modifications may be combined.

1 is a system configuration diagram showing a configuration of a monitoring system 10. FIG. 2 is a configuration diagram showing a configuration of a cogeneration system 300. FIG. It is a data block diagram which shows the data structure of the item information table 131. 4 is a data configuration diagram showing a data structure of a variable correspondence table 132. FIG. It is a data block diagram which shows the data structure of the division | segmentation time information 133. FIG. 3 is a data configuration diagram showing a data structure of reference period information 134. FIG. 4 is a data configuration diagram showing a data structure of a cumulative analysis data table 135. FIG. 4 is a data configuration diagram showing a data structure of a measurement information table 136. FIG. It is a data block diagram which shows the data structure of the diagnostic period information 137. It is a data block diagram which shows the data structure of the diagnostic item information table 138. It is a data block diagram which shows the data structure of the current analysis data table 139. 4 is a data configuration diagram showing a data structure of a determination result table 140. FIG. It is a graph showing the result of the regression analysis by the conventional least squares method. On the graph, a measured value group and a straight line obtained by regression analysis are plotted. It is a graph showing the result of the regression analysis by the conventional least squares method. On the graph, a measurement value group including sudden abnormal values and a straight line obtained by regression analysis are plotted. 3 is a flowchart showing the operation of the monitoring system 10. Continuing to FIG. 3 is a flowchart showing the operation of the monitoring system 10. Continued to FIG. 3 is a flowchart showing the operation of the monitoring system 10. Continuing from FIG. The correspondence between the objective variable used in the monitoring system 10 and the explanatory variable is shown. The correspondence of the objective variable, explanatory variable, and 2nd explanatory variable used in the monitoring system 10a as a modification is shown. 6 is a graph showing a change with time of a regression coefficient obtained by the monitoring system 10 when the objective variable is related to temperature. When an objective variable is related to temperature, it is a graph showing the change with progress of time of the regression coefficient obtained by the monitoring system 10a. It is a flowchart which shows the operation | movement by the data-analysis apparatus 100b as a modification. It is a graph showing the change with progress of time of the regression coefficient obtained by the monitoring system 10b. It is a block diagram which shows the structure of the air conditioning system 500 which comprises the monitoring system c as a modification. It is a data block diagram which shows the data structure of the item information table 131c. It is a flowchart which shows the operation | movement by the data analyzer 100c. Continued to FIG. It is a flowchart which shows the operation | movement by the data analyzer 100c. Continuing from FIG.

Explanation of symbols

10 Monitoring system 20 LAN
DESCRIPTION OF SYMBOLS 100 Data analysis device 101 Analysis processing part 102 Threshold setting part 103 Abnormality judgment part 104 Storage part 105 Input / output control part 106 Report generation part 107 Mail generation part 108 Graph generation part 109 Communication part 200 Center server apparatus 300 Cogeneration system 400 Load equipment

Claims (5)

An abnormality determination device that is constructed from a plurality of element devices, monitors an equipment system that continuously obtains one result by continuous operation, and determines the presence or absence of an abnormality in each element device,
Obtaining means for obtaining a plurality of result measurement values indicating the results at a plurality of time points, and a plurality of state measurement values relating to the operating state of one element device measured at each of the plurality of time points;
The obtained plurality of result measurement values are assigned to a first variable, the obtained plurality of state measurement values are assigned to a second variable, and the first result is obtained with respect to the plurality of result measurement values and the plurality of state measurement values. Analysis processing means for performing a single regression analysis using a regression equation formed by combining a variable, the second variable, and a regression coefficient by an operator as a model, and calculating the regression coefficient;
A determination means for comparing the calculated regression coefficient with an abnormal threshold value indicating an abnormal range to determine whether the regression coefficient is included in the abnormal range;
An abnormality determination apparatus comprising: output means for outputting abnormality information indicating a possibility of occurrence of abnormality in an element device that is a generation source of the state measurement value when it is determined to be included in an abnormality range. The acquisition means further acquires a plurality of second state measurement values related to the operating states of the other component devices respectively measured at the plurality of times.
The analysis processing means further assigns the plurality of second state measurement values acquired to a third variable, and the first variable and the plurality of second state measurement values with respect to the plurality of result measurement values and the plurality of second state measurement values. Performing a single regression analysis modeled on a second regression equation configured by combining a third variable and a second regression coefficient by an operator, and calculating the second regression coefficient;
The determination means further compares the calculated second regression coefficient with a second abnormality threshold value indicating the second abnormality range, and determines whether or not the second regression coefficient is included in the second abnormality range. Judgment
The output means further outputs second abnormality information indicating a possibility of occurrence of an abnormality in another element device that is a generation source of the second state measurement value when it is determined that the output means is included in the second abnormality range. The abnormality determination device according to claim 1, wherein: The acquisition means further acquires a plurality of environmental measurement values related to the environmental state of the element device measured at each of the plurality of time points,
The analysis processing unit further assigns the acquired plurality of environment measurement values to a third variable, and instead of the single regression analysis, the plurality of result measurement values, the plurality of state measurement values, and the plurality of environment measurements. The regression coefficient is calculated by performing regression analysis on the value using a regression equation that is a combination of the first variable, the second variable, the third variable, and the regression coefficient as an operator. The abnormality determination device according to claim 1, wherein: The abnormality determination device further includes:
The abnormality indicating the abnormality range using a plurality of past state measurement values related to the measured operation state of the element device in a period before the plurality of state measurement values are acquired by the acquisition unit. The abnormality determination device according to claim 1, further comprising an abnormality threshold setting unit that calculates a threshold. It is an abnormality determination program used in an abnormality determination device that is constructed from a plurality of element devices, monitors an equipment system that continuously obtains one result by continuous operation, and determines the presence or absence of an abnormality in each element device. ,
An acquisition step of acquiring a plurality of result measurement values indicating the results at a plurality of time points, and a plurality of state measurement values related to the operating state of one element device measured at each of the plurality of time points;
The obtained plurality of result measurement values are assigned to a first variable, the obtained plurality of state measurement values are assigned to a second variable, and the first result is obtained with respect to the plurality of result measurement values and the plurality of state measurement values. An analysis processing step of calculating a regression coefficient by performing a single regression analysis using a regression equation formed by combining a variable, the second variable, and a regression coefficient as an operator;
A determination step of comparing the calculated regression coefficient with an abnormal threshold value indicating an abnormal range to determine whether the regression coefficient is included in the abnormal range;
And an output step of outputting abnormality information indicating the possibility of occurrence of abnormality in the element device that is the generation source of the state measurement value when it is determined to be included in the abnormality range.

Images