JP7035842B2 - Monitoring system - Google Patents

Description

The present invention relates to a system for monitoring equipment operating in a plant or the like.

The outline of the monitoring system of the steam use equipment described in Patent Document 1 will be described. This monitoring system monitors the state of the equipment attached to the steam-using equipment, that is, the steam controller (steam trap), in order to detect the failure (abnormality) of the steam-using equipment at an early stage.

Information such as temperature, pressure, and flow rate is acquired by the sensor attached to this steam controller, and the state of the steam controller is detected based on the acquired information. At this time, the sensor may show an abnormal value due to a failure of the steam controller or the like before the failure of the steam-using equipment occurs. By capturing the abnormal value of this sensor, we are trying to prevent the failure of the steam-using equipment.

Japanese Patent No. 5818337

As described above, the invention of Patent Document 1 estimates the failure of the steam-using equipment by detecting the failure of the steam controller that occurs earlier, but has the following problems.

That is, the invention of Patent Document 1 cannot detect a failure of steam-using equipment unless the steam controller fails, and is not effective as a system for predicting the occurrence of failure of steam-using equipment.

In addition, since it is necessary to attach various instruments such as temperature sensors, vibration sensors, pressure gauges, and flow meters to the steam controller as sensors to monitor the condition, it is difficult to apply to existing equipment and the cost may increase. There is.

The present invention has been made to solve such a conventional problem, and it is an object of the present invention to provide a monitoring system that catches a sign of equipment failure, predicts the failure, and is easily applicable to existing equipment.

(1) The present invention is a system for monitoring equipment to be monitored.
A measuring unit that measures the input given to the equipment and the equipment state of the equipment when the input is given.
A recording unit that records the measurement data of the measurement unit at regular intervals and stores the measurement data for the cycle as time-series data.
An analysis processing unit that calculates the tendency of the temporal change of the equipment state with respect to the input based on the time series data of the recording unit, and the analysis processing unit.
If the tendency of the temporal change calculated by the analysis processing unit deviates from the normal range determined in advance based on the tendency of the temporal change during normal operation, it is determined that a failure sign has occurred for the equipment. It has an evaluation unit.

(2) One aspect of the present invention is a heat exchange device in which the equipment cools the heat of the exhaust gas of the engine.
The measuring unit measures the amount of heat input to the heat exchange device and the amount of heat generated by the cooling water of the heat exchange device.
The analysis processing unit calculates a linear function that approximates the tendency of the temporal change between the calorific value and the calorific value for a predetermined period from the stop of the engine.
If the tendency of the linear function deviates from the normal range, the state evaluation unit determines that a failure sign has occurred in the heat exchange device.

(3) Another aspect of the present invention is a generator in which the equipment is driven by a drive source.
The measuring unit measures the temperature that fluctuates in correlation with the driving force applied to the generator and the calorific value of the lubricating oil caused by the driving force of the generator.
The analysis processing unit calculates a cubic function that approximates the tendency of the temporal change between the temperature and the calorific value in a predetermined time from the start of operation of the generator.
If the third-order coefficient deviates from the normal range, the state evaluation unit determines that a failure sign has occurred for the generator.

(4) Still another aspect of the present invention is a power generation facility including a rotary machine in which the facility generates driving force according to a flow rate and a generator for converting the driving force of the rotary machine into generated power. There,
The measuring unit measures the cumulative flow rate given to the rotating machine and the cumulative generated power amount of the generator.
The analysis processing unit calculates a linear function that approximates the tendency of the temporal change between the cumulative flow rate and the cumulative power generation amount from the start to the stop of the operation of the power generation facility.
If the tendency of the linear function deviates from the normal range, the state evaluation unit determines that a failure sign has occurred in the power generation facility.

(5) One aspect of the rotary machine includes an internal combustion engine that consumes fuel to generate driving force as the rotary machine, and the cumulative flow rate is the cumulative fuel flow rate that has flowed into the internal combustion engine.

(6) If the tendency of the temporal change calculated by the analysis processing unit belongs to a predetermined failure range, a failure alarm can be issued for the equipment. Further, the measuring unit can measure the input and the equipment state by using the sensor.

According to the present invention, it is possible to provide a monitoring system that catches a sign of equipment failure, predicts the failure, and is easily applicable to existing equipment.

The schematic diagram of the monitoring system which concerns on embodiment of this invention. Configuration diagram of the monitoring system. Recorded data configuration diagram of the same data recording unit. Explanatory drawing which shows the preprocessing of the analysis processing part. Explanatory drawing which shows the analysis processing of the analysis processing part. Explanatory drawing which shows the processing process of the same state evaluation part. Explanatory drawing which shows the structure and the measurement item of the exhaust gas boiler which concerns on Example 1. The graph of the tendency of the time state change of the exhaust gas boiler is shown, (a) is a graph at the time of normal operation, (b) is a graph at the time of a failure sign, and (c) is a graph at the time immediately before a failure. Graph showing changes in the same coefficient value Explanatory drawing which shows the structure and measurement item of the generator which concerns on Example 2. A graph showing the tendency of the temporal state change of the generator is shown, (a) is a graph at the normal time, and (b) is a graph at the time of abnormality detection. The graph which shows the change of the 3rd order coefficient. FIG. 6 is a device configuration diagram of the power generation facility of the third embodiment. The graph of the tendency of the temporal state change of the power generation facility is shown, (a) is a graph at the normal time, and (b) is a graph at the time of abnormality (maintenance recommended). Same bar graph.

Hereinafter, the monitoring system according to the embodiment of the present invention will be described. This monitoring system is mainly used for monitoring the equipment operating in the plant, and as shown in FIG. 1, the input X given to the monitored equipment 11 and the monitored equipment 11 when the input X is given. The change of the equipment condition Y is measured and recorded, and the tendency of the temporal change of the equipment condition is monitored.

The tendency of the temporal change monitored here is compared with the tendency at the normal time, and if the tendency is significantly different, it is determined that the monitored equipment 11 has a sign of failure. That is, since the state Y of the monitored equipment 11 changes when the input X is given, both X and Y are measured by the sensor.

Focusing on the temporal change of the measurement data, a difference is likely to appear between the state in which the monitored equipment 11 is operating normally and the state in which it is in a malfunctioning state (a state in which a sign of failure appears). By capturing this difference, the monitoring system detects a sign of failure of the equipment to be monitored 11.

≪Overall system configuration example≫
A configuration example of the monitoring system will be described with reference to FIG. The monitoring system 12 uses the measurement information of the sensors S1 and S2 attached to the input side of the monitored equipment (system) 11 as the information of the input X, while the sensors S3 and S3 attached to the output side of the monitored equipment 11. The measurement information of S4 is used as the information of the equipment state Y.

The information of the input X and the equipment state Y is selected from the temperature, pressure, flow rate, current, voltage, etc. according to the equipment to be monitored 11. By this selection, the types of sensors S1 to S4 (thermometer, pressure gauge, flow meter, ammeter, voltmeter, etc.) are determined.

Specifically, the monitoring system 12 is configured by a computer, and as a result of cooperation between hardware resources (CPU, RAM.ROM, SSD, HDD, etc.) and software resources (OS, application, etc.), the measurement unit 13, data. The recording unit 14, the analysis processing unit 15, and the state evaluation unit 16 are mounted.

The measurement unit 13 measures the information of the input X and the information of the equipment state Y by the sensors S1 to S4, and A / D converts the measured values of the sensors S1 to S4 into digital data. Further, the data recording unit 14 is built in a computer storage device (main storage device such as RAM, ROM, auxiliary storage device such as HDD, SSD, etc.), and records the measurement data of the measurement unit 13 at regular intervals. Accumulate as series data.

The analysis processing unit 15 calculates the tendency of the temporal change of the equipment state with reference to the time-series data accumulated in the data recording unit 14. Here, the approximate function of the temporal change of the equipment state with respect to the input is calculated with the input as the X-axis and the equipment state as the Y-axis.

The state evaluation unit 16 stores the approximate function calculated in advance during normal operation of the monitored equipment 11 in the storage device. The characteristics of the approximate function (inclination, etc.) stored here are compared with the characteristics of the approximate function calculated by the analysis processing unit 15 when the equipment is in operation. Is determined to have occurred.

Therefore, according to the monitoring system 12, by using the determination result of the state evaluation unit 16 as the failure prediction, it is possible to catch the sign of the failure of the monitored equipment 11 and give an early warning. At this time, the sensors S1 to S4 can be implemented by existing sensors or sensors of a type that can be easily retrofitted so that the monitoring system 12 can be easily applied to existing equipment. Hereinafter, the details of each part 14 to 16 will be described.

≪Data recording unit 14≫
The details of the recorded data of the data recording unit 14 will be described with reference to FIG. Here, the data recording unit 14 collects the measured values for each of the sensors S1 to S4 and records them as a vector. Hereinafter, specific recorded data will be described.

Here, the real number is the equation (1) and the positive integer is the equation (2). At this time, the equation (3) is used as the sensor number, and the jth sensor is identified as “S _j ”. Further, the measured value of the sensor “S _j ” at the measurement time “k ≧ 1” is set to the equation (4), and the following processing (STEP01 to STEP03) is performed on the measured values obtained by the sensors S1 to S4. To.

STEP01: The measured values of the sensors S1 to S4 at the measurement start time "k", that is, the measured values of the equation (4) are sent to the data recording unit 14 via the measuring unit 13.
Where k is a positive integer.

STEP02: As shown in FIG. 3, the data recording unit 14 receives the measured value of the formula (4), organizes it as the time series data of the formula (5), and stores it. Here, "(・) ^T " means transpose of the vector. Further, although the measured values are shown in the vector format, if the sensors S1 to S4 have a plurality of numerical values, another format such as a matrix format may be used.

At this time, if a new measured value is added according to the passage of the measurement time "k", the time series data of the equation (5) is sequentially updated. For the time-series data in Eq. (5), if the past measured values are no longer needed due to a prolonged measurement period, the data will be selected from the viewpoint of suppressing the data capacity and excluding outliers. Discard.

Since the processing of STEP 02 is executed for each of the sensors S1 to S4, the time series data "V _k ^(S1) " to "V _k ^(S4) " are generated as shown in FIG.

STEP03: In STEP03, the time series data of the equation (5) is converted into the data set of the equation (7) used for the analysis. To explain the details, the past N measured values are extracted from t with the extraction time “t” as a reference for the time series data of the equation (5), and the extracted data of the equation (6) is generated. The extraction time “t” is an observation time of a signal used for analysis separately from the measurement time “k”. Here, the extraction time "t" is a positive integer. The sample length "N" can be changed according to the transition speed and accuracy of the time change of the monitored equipment 11.

The extracted data of the formula (6) is added as a column of a data set (matrix of N rows × t columns). That is, the extraction data is generated while reducing the extraction time "t" by one with respect to the time series data of the equation (5), and is added to the column of the matrix. In this way, the data set equation (7) for use in the analysis is obtained.

Since the processing of STEP03 is executed for each of the sensors S1 to S4, the data set "d _{(S1) to} d _(S4) " is generated and stored in the data recording unit 14 as shown in FIG. .. The data sets generated and saved here are integrated and sent to the analysis processing unit 15.

<< Analysis processing unit 15 >>
After receiving a plurality of data sets "d _{(S1) to} d _(S4) " from the data recording unit 14, the analysis processing unit 15 performs the preprocessing (STEP04 to STEP07) of FIG. 4 and the analysis processing (STEP08, STEP08) of FIG. 09) is executed.

(1) Preprocessing STEP04: The analysis processing unit 15 selects a pair of data sets from the received data set group. Here, as an example, the data set of the sensor number “i, j”, that is, the pair of the data set of the equation (8) will be selected.

STEP05: For the pair of data sets selected in STEP04, make a pair with the extracted data at the same extraction time. Here, the data group shown in the equation (9) is created.

STEP06: The feature value is calculated for each set of the data group created in STEP05. At this time, it is mapped to the feature space by the function of the equation (10) that returns one variable from the set of extracted data. Here, the feature value is given by the equation (11).

However, the set of extracted data does not have to be limited to two variables, and it is possible to create a set with a plurality of extracted data as variables. In this case, the function is designed according to the number of variables.

STEP07: The feature values calculated in STEP06 are summarized as the feature data of the equation (12).

Such processing of STEP05 to S07 is executed for all pairs selected based on the data set "Equation (7)" received from the data recording unit 14 in STEP04, and feature data is calculated.

To explain based on FIG. 4, first, the data set "d _(S1) , d _(S2) " of the sensor number "i = 1, j = 2" is selected in STEP 04, and the feature data "d (S1), d (S2)" is processed through STEP 05, 06. _rt ^(1,2) ”is calculated. This calculation result is referred to as feature data 1.

Next, returning to STEP 04, the data set "d _(S3) , d _(S4) " of the sensor number "i = 3, j = 4" is selected, and similarly, the feature data "rt (r _t ⁽ " is processed through STEP 05, 06. ^3,4) ”is calculated. This calculation result is referred to as feature data 2.

(2) Analysis processing For analysis processing, a pair of feature data calculated in the preprocessing is used. Here, as shown in FIG. 5, the feature data 1 “ _rt ^(1,2) ” and the feature data 2 “ _rt ^(3,4) ” are used.

In this analysis process, the input / output of the monitored equipment 11 is defined as " _rt ^(1,2) , _rt ^(3,4) ", and the system representing the input / output relationship is calculated as a function. Here, the feature data 1 is used as the input X, while the feature data 2 is used as the output (equipment state) Y.

However, since the purpose of this analysis process is to understand the equipment status by applying the relationship between the data to the input / output system model, it is called the data that is the input side and the data that is the output side in the actual equipment configuration. There is no need to consider physical relationships. Hereinafter, specific processing contents (STEP08, STEP09) will be described.

_STEP08 : First, for the input / output "rt ^(1,2) , rt ^(3,4) ", a pair of elements having a common extraction time _t is created. Here, it is assumed that the set of the equation (13) is prepared.

STEP09: Next, the input / output system (function) "gt (・)" is estimated based on the input / output data of the equation (13) prepared in STEP08. That is, while the noise of the equation (14) is present at an arbitrary extraction time t, the function of the equation (16) expressing the relational expression of the equation (15) with sufficient accuracy is obtained from the input / output data of the equation (13). Calculated using an estimation algorithm.

Regression analysis, adaptive filters, parameter estimation methods, etc. are used for this estimation algorithm, and they are arbitrarily selected according to the required processing speed and estimation accuracy. The relationship between the input and output of the estimation target, that is, the monitoring target equipment 11, changes with the passage of the measurement time k.

Therefore, the input / output system “gt (・)” that models it also changes as the extraction time t increases. That is, the estimation function "Equation (16)" of the input / output system "gt (.)" Is updated as the extraction time t increases, and is sent to the state evaluation unit 16 in real time or after a certain period of time has elapsed.

≪Condition evaluation unit 16≫
The operation processing of the state evaluation unit 16 will be described with reference to FIG. The state evaluation unit 16 evaluates the estimation function "Equation (16)" obtained by the analysis processing unit 15 and captures the temporal change. Here, attention is paid to the evaluation value (coefficient, etc.) of the estimation function “Equation (16)”. This evaluation value is shown as "α _t " in FIGS. 5 and 6.

That is, since the fluctuation information measured by the sensors S1 to S4 is included in the feature data 1 and 2, the input / output relationship of the feature data 1 and 2 is changed, and the estimation function "Equation (16)" is changed. .. By quantifying the change in the estimation function "Equation (16)", it is possible to detect a failure or abnormality of the monitored equipment 11.

Here, as the operation processing of the state evaluation unit 16, the temporal transition of the estimation function “Equation (16)” is monitored. At this time, the monitoring operation of comparing the pre-learned equipment abnormality data with the measurement data is not executed. Therefore, even if the failure state is unknown, it is possible to detect an abnormality or failure by capturing the change in the evaluation value (α _t ). In addition, since the precursor of an abnormality in peripheral equipment is captured from the fluctuation of the evaluation value (α _t ), the precursor of failure can be detected and the prediction before the occurrence of an abnormality becomes possible. Hereinafter, details will be described based on Examples 1 and 2.

<< Example 1 >>
The first embodiment of the monitoring system 12 will be described with reference to FIGS. 7 to 9. Here, the monitoring system 12 is applied to monitor the failure sign of the exhaust gas boiler 21.

(1) Overall configuration The exhaust gas boiler 21 of the monitored equipment 11 is a heat exchange device that cools the heat of the exhaust gas of the engine 23 with the cooling water of the water tank 24 in the boiler equipment 22 of the diesel power plant. And are discharged. In the first embodiment, the heat exchange state of the exhaust gas boiler 21 is monitored to detect a failure at an early stage.

Here, in the analysis of the equipment state, it is assumed that the transition of the temperature change appearing after the operation of the engine 23 is stopped is monitored. Therefore, the sensor S1 measures the boiler inlet temperature, the sensor S2 measures the boiler outlet temperature, the sensor S3 measures the cooling water supply water temperature, and the sensor S4 measures the outside air temperature.

Specifically, the boiler inlet temperature of the sensor S1 indicates the temperature before the exhaust gas generated by the combustion of the engine 23 flows into the exhaust gas boiler 21. Further, the boiler outlet temperature of the sensor S2 indicates the temperature of the cooled exhaust gas discharged through the exhaust gas boiler 21.

The cooling water supply temperature of the sensor S3 indicates the temperature of the cooling water measured on the inlet side of the exhaust gas boiler 21. Further, the outside air temperature of S4 is measured outside the exhaust gas boiler 21.

Here, the amount of heat input to the exhaust gas boiler 21 is used for the sensor data of the input X, while the amount of heat generated by the cooling water is used for the sensor data of the equipment state (system state) Y. The input heat amount of the input X can be calculated by "S1-S2", and the cooling water calorific value of the equipment state Y can be calculated by "S3-S4" (however, the absolute value is used for the input X).

At this time, the monitoring system 12 of the first embodiment adopts an operation of recording the measurement data during operation of the exhaust gas boiler 21 in the data recording unit 14 and performing batch processing after the operation is stopped to periodically monitor the data. , Real-time processing as described above may be used.

Then, the start / stop of the operation of the engine 23 is determined by the measured value obtained by the measuring unit 13 by the rotation sensor (not shown) attached to the engine 23. Further, the data recording unit 14 refers to the measurement data of the measurement unit 13, and from the measured value (rotational speed) of the rotation sensor, "operating (rotational speed> 0)" and "operating stopped (rotational speed = 0)". ".

Here, when the state changes from "stopped" to "running", it is determined to be "started", while when the state changes from "running" to "stopped", it is determined to be "stopped". , Record the occurrence history of operation start / stop together with the time.

(2) Processing content The data recording unit 14 can store the measurement data measured by the measurement unit 13 through the sensors S1 to S4 in a vector format every minute, and can store the measurement data for one day or more for each of the sensors S1 to S4. Capacity.

Since the sensors S1 to S4 are data per minute, 1440 samples are acquired per day. At this time, the data recording unit 14 records the measurement data of the sensors S1 to S4 every minute, and the sensor S1 becomes the time series data “V ^(S1) ” to “V ^(S4) ” of the equation (17). Prepare four vectors for each of ~ S4.

Here, since the measurement data is recorded at 1-minute intervals, the sample length of the extracted data is set to "N = 8" when the data set of the equation (18) is generated. This is to reduce the influence of noise in the measurement data, and since it is handled in the preprocessing stage of the analysis processing unit 15, it is converted as shown in the equation (18).

Here, the data set d _(S1) of the sensor S1 is illustrated, but it is assumed that the sensors S2 to S4 also generate the data set "d _(S2) to d _(S4) " in the same format, and the generated four data. The set is sent to the analysis processing unit 15.

The analysis processing unit 15 first divides the four data sets into two pairs in the preprocessing, and then generates each feature data. That is, the pair of the data set "d _(S1) , d _(S2) " and the pair of the data set "d _(S3) , d _(S4) " are selected.

Here, an example of generating feature data 1 from a set of "d _(S1) , d _(S2) " will be described, but feature data 2 is also generated from a set of "d _(S3) , d _(S4) ". It shall be.
For the pair of data set "d _(S1) , d _(S2) ", make a set with the extracted data at the same extraction time, and create the data group of the formula (9). The feature value of the formula (11) is calculated using the formula (19) for each set of extraction time of the created data group.

As a result, the feature data of the equation (12) is obtained.
Let "P" be the feature data 1 calculated by the pair of the data set "d _(S1) , d _(S2) ". Similarly, the feature data 2 calculated by the pair of “d _(S3) , d _(S4) ” is set as “Q” and is used in the equation (20).

This vector "P" shows the fluctuation value of the input X, and the vector Q shows the fluctuation value of the equipment state Y.

Next, the analysis process based on the generated feature data 1 and 2 will be described. Here, for the vectors "P" and "Q", a sample pair for each sample at the same time shown in the equation (21) is generated.

After that, the calculation of the input / output system, that is, the linear regression is executed with the input as "P" and the output as "Q", and the estimation function is calculated. At this time, it is assumed that a two-dimensional graph is drawn with each sample pair as a point, with the X-axis as "P" and the Y-axis as "Q".

(3) Explanation of two-dimensional graph Figures 8 (a) to 8 (c) show the two-dimensional graph, and the point sequence in each figure is the fluctuation transition (time) drawn by the measurement data (1440 samples) for one day. The target change) is shown, and 60 samples for 60 minutes from the stop time of the engine 23 are approximated by a linear function by regression analysis. Therefore, each time the engine 23 is stopped, the coefficient value (slope) of the approximate function is obtained as a feature from the data for 60 minutes thereafter.

First, FIG. 8A will be described. FIG. 8A shows the state of the exhaust gas boiler 21 during normal operation, and shows large fluctuations in both the X-axis and the Y-axis. That is, it is shown that the cooling water calorific value "Q" of the Y-axis increases with the increase of the input heat amount "P" of the X-axis, and the cooling function of the exhaust gas boiler 21 is operating normally. The temperature change in FIG. 8A, that is, “the coefficient value (slope) = 0.30” is stored in the state evaluation unit 16.

Next, FIG. 8B will be described. FIG. 8B shows the state of the exhaust gas boiler 21 at the time of failure sign. Here, the amount of fluctuation on the X-axis and the Y-axis is attenuated as compared with FIG. 8A, and is reduced to "the coefficient value (slope) = 0.005". Therefore, it is presumed that the equipment state of the exhaust gas boiler 21 has changed so that the cooling function is lowered.

Finally, FIG. 8 (c) will be described. FIG. 8C shows the state of the exhaust gas boiler 21 immediately before the failure on the failure date. Here, there is almost no fluctuation in the X-axis and the Y-axis, and "the coefficient value (slope) = -0.002" is shown, indicating that the cooling function is not functioning.

(4) Condition evaluation unit 16
The state evaluation unit 16 records the coefficient value (slope) calculated by the analysis processing unit 15 every day, and monitors the transition. At this time, the state evaluation unit 16 identifies the state of the exhaust gas boiler 21 by comparing with the stored normal time coefficient value (slope).

That is, the analysis processing unit 15 calculates the coefficient value (slope) for the data for 60 minutes after the engine 23 is stopped, and the calculated value is set as the evaluation value “α _t ”, and the coefficient value (slope) in the normal state is used. Compare with. At this time, the coefficient value in the normal state belongs to the range of "about 0.1 to about 0.5", while if the exhaust gas boiler 21 becomes malfunctioning, it approaches "0" as the failure occurrence approaches.

Therefore, a threshold value in the normal range (for example, "0.5> slope ≥ 0.1") is determined based on the stored normal coefficient value (slope). At this time, if the coefficient value (slope) recorded every day exceeds the threshold value, it is determined that a failure sign has occurred in the exhaust gas boiler 21, and a warning is given. As a result, it is possible to detect a failure or abnormality of the exhaust gas boiler 21 at an early stage.

Further, if a failure range (for example, less than 0.0) is provided for the coefficient value (slope), it is possible to detect immediately before the failure of the exhaust gas boiler 21 and issue a failure alarm to encourage the implementation of maintenance work.

Table 1 shows the state transition of the exhaust gas boiler 21 from “November 23 to December 7”, and the state transition of the exhaust gas boiler 21 is graphed in FIG. Here, "November 23 to November 28" fluctuates in the range of "the coefficient value (slope) = 0.213 to 0.315". Therefore, the coefficient value (slope) belongs to the threshold value, and the exhaust gas boiler 21 is determined to be normal.

On the other hand, "November 30th to December 4th" fluctuates in the range of "the coefficient value (slope) = 0.058 to 0.089". Therefore, the coefficient value (slope) exceeds the threshold value, and the exhaust gas boiler 21 is determined to be a sign of failure. Further, since the coefficient value (slope) is less than "0.0" in "December 5th to December 7th", a failure alarm is issued.

As described above, according to the monitoring system 12, it is possible to detect a failure sign or failure of the exhaust gas boiler 21 by capturing the numerical fluctuations of the sensors S1 to S4. Therefore, it is possible to determine an abnormality (failure sign / failure) of the exhaust gas boiler 21 without detecting a failure of peripheral equipment. In this respect, the abnormality can be detected earlier and more reliably than in Patent Document 1, and the occurrence of failure can be predicted effectively.

Further, since temperature sensors may be used for the sensors S1 to S4, it is not necessary to use various instruments such as a vibration sensor, a pressure gauge, and a flow meter. In this respect, it is easy to apply to existing equipment and the cost is low. It can also contribute to suppression.

<< Example 2 >>
The second embodiment of the monitoring system 12 will be described with reference to FIGS. 10 to 12. Here, the monitoring system 12 is applied to monitor the failure sign of the generator 31.

(1) Overall configuration The generator 31 of the equipment to be monitored 11 rotates the rotor 32 by the driving force of an engine or the like to generate electric power by excitation. That is, when the rotor 32 is rotated by the driving force, an exciting current is generated in the stator 33. The stator 33 is composed of an iron core, windings, and the like, and is installed in the frame of the generator 31.

The rotor 32 is composed of a spindle, an iron core, windings, and the like, and is rotatably supported by a bearing 34 inside the stator 33. In order to support the rotor 32, pressure and friction are applied to the bearing 34. Lubricating oil is circulated in the bearing 34 in order to suppress this pressure imbalance and friction and smoothly rotate the rotor 32.

When such a generator 31 starts operation, a current flows through the stator 33 and the rotor 32, and heat is generated due to the loss due to the resistance component. At this time, a failure may occur due to wear or scratches on the bearing 34, poor circulation of lubricating oil, or the like. Therefore, in the second embodiment, the monitoring system 12 monitors the temperature changes of the stator 33, the bearing 34, and the lubricating oil, and detects a sign of failure of the generator 31 at an early stage.

The sensor S1 is installed in the stator 33 in order to measure the temperature of the stator 33 of the generator 31. Further, the sensor S2 is installed on the bearing 34 in order to measure the bearing temperature of the generator 31. Further, the sensor S3 measures the temperature of the lubricating oil that has passed through the bearing 34 in order to measure the lubricating oil temperature. The sensor S4 measures the outside air temperature outside the generator 31.

Here, the generator driving force given to the rotor 32 is used for the sensor data of the input X. That is, since the amount of heat generated in the stator 33 is considered to correlate with the generator driving force, "S1-S2" excluding the amount of heat loss generated in the bearing 34 is measured as the generator driving force of the input X.

On the other hand, the calorific value of the lubricating oil is used for the sensor data of the equipment state Y. The amount of heat generated by the lubricating oil is a temperature change of the lubricating oil caused by the driving force of the generator 31, and is measured by "S3-S4". Here, as in the first embodiment, the operation of recording the measurement data during operation of the generator 31 in the data recording unit 14 and performing batch processing after the operation is stopped to periodically monitor the data is adopted, but real-time processing is also adopted. Make it good.

Then, the start / stop of operation of the generator 31 is determined by the measured value obtained by the measuring unit 13 by the rotation sensor (not shown) attached to the generator 31. Further, the data recording unit 14 refers to the measurement data of the measurement unit 13, and is "in operation (rotational speed> 0)" and "in operation stopped (rotational speed = 0)" from the measured value (rotational speed) of the rotation sensor. And are determined.

Here, when the state changes from "stopped" to "running", it is determined to be "started", while when the state changes from "running" to "stopped", it is determined to be "stopped". , Record the occurrence history of operation start / stop together with the time.

(2) Processing content The data recording unit 14 records the measurement data measured by the measurement unit 13 through the sensors S1 to S4 every minute and accumulates it as time-series data in vector format, as in the first embodiment, and the sensor S1 The capacity is set so that data for one day or more can be stored for each S4.
Similar to the first embodiment, the data set “d _{(S1) to} d _(S4) ” of the equation (18) is generated.
The four generated data sets are sent to the analysis processing unit 15.

The analysis processing unit 15 first selects a pair of the data set "d _(S1) , d _(S2) " and a pair of the data set "d _(S3) , d _(S4) ".

Feature data is generated for each pair, but in Example 2, equation (22) is used for conversion to feature values.

Here, the sign function "sign (・)" is given by the equation (23).

In this way, the feature data 1 is calculated from the pair of "d _(S1) , d _(S2) ". Similarly, the feature data 2 is calculated from the pair of "d _(S3) , d _(S4) ".

Based on the calculated feature data 1 and 2, the analysis processing unit 15 uses the X-axis as the generator driving force (P) and the Y-axis as the lubricating oil calorific value (Q), as shown in FIGS. 11 (a) and 11 (b). Draw a two-dimensional graph. The sequence of points in this graph shows the fluctuation transition (temporal change) depicted by the measurement data for one day. Here, the data (data of 30 samples) for 30 minutes from the operation start time of the generator 31 in the two-dimensional graph is approximated by a cubic function. Therefore, a cubic function during normal operation is calculated in advance, and the cubic coefficient is stored in the state evaluation unit 16.

Similarly, each time the operation of the generator 31 is started, a cubic function is calculated from the measurement data for 30 minutes thereafter, and the calculated cubic coefficient is recorded as the evaluation value “α _t ”. The cubic coefficient recorded here is compared with the cubic coefficient during normal operation.

At this time, since the change in temperature relation becomes close to a quadratic function at the start of operation, the weight of the cubic coefficient becomes small, and the value of the cubic coefficient in normal operation is calculated as "± 0.01 or less". The coefficient. For example, in the normal operation of FIG. 11A, it is shown as “the third-order coefficient = −0.0023”.

On the other hand, when poor circulation of lubricating oil or wear / scratch of bearings occurs, the change in the calorific value of lubricating oil with respect to the driving force of the generator 31 tends to be complicated, and the weight of the third-order coefficient increases. For example, in the abnormality detection of FIG. 11B, it is shown as "the third-order coefficient = −5.991".

Therefore, a threshold value in the normal range (for example, -0.05 <the third-order coefficient <0.05) is set based on the stored third-order coefficient during normal operation, and the third-order coefficient recorded every day is recorded. If the coefficient deviates from the threshold value, a warning is given by regarding the occurrence of a failure sign of the generator 31.

Table 2 shows the state transition of the generator 31 from “November 23 to December 7”, and the state transition of the generator 31 is graphed in FIG. Here, "November 23 to December 3" fluctuates in the range of "the cubic coefficient = −0.00727 to 0.007818". Therefore, the third-order coefficient belongs to the threshold value, and the generator 31 is determined to be normal. On the other hand, in "December 4th to December 7th", since the third-order coefficient deviates from the threshold value, it is determined that a failure sign has occurred in the generator 31.

In this way, in order to determine the occurrence of a failure sign of the generator 31 by capturing the numerical fluctuations of the sensors S1 to S4, the abnormality of the generator 31 is detected without detecting the failure of the peripheral equipment as in the first embodiment. be able to. In this respect, as in the first embodiment, the abnormality can be detected earlier and more reliably than in Patent Document 1, and the occurrence of a failure can be effectively predicted.

Further, since temperature sensors may be used for the sensors S1 to S4, it is not necessary to use various instruments such as a vibration sensor, a pressure gauge, and a flow meter. In this respect, it is easy to apply to existing equipment and the cost is low. It can also contribute to suppression.

<< Example 3 >>
The third embodiment of the monitoring system 12 will be described with reference to FIGS. 13 to 15. In this embodiment, the monitoring system 12 is applied to monitor the failure sign of the power generation facility of the thermal power plant. Here, instead of monitoring the temperature change of the lubricating oil as in Example 2, the power generation facility is evaluated from the viewpoint of power generation efficiency, and the maintenance time thereof is proposed.

(1) Overall configuration The power generation equipment to be monitored will be described with reference to FIG. Here, the power generation facility 41 using an internal combustion engine is monitored. The power generation facility 41 is composed of a large number of devices such as a rotary machine 40 on the drive side, a generator 44 on the power generation side, a cooling device (not shown), and a variable reception facility, and is centered on the rotary machine 40 and the generator 44. ..

The rotary machine 40 is mainly composed of an engine 43, and rotates a spindle 49 by consuming fluid fuel supplied from a fuel tank 42 to generate a driving force. The rotation of the spindle 49 rotates the rotor 46 of the generator 44 to generate an exciting current in the stator 47, which is converted into generated power. The generated power converted here is output to the premises 48 on the demand side.

The power generation facility 41 is required to efficiently convert fuel into electric power. If the devices (equipment) constituting the power generation facility 41 continue to deteriorate or malfunction (for example, a failure of the cooling device or wear of the bearing of the spindle 49), the power generation operation is hindered and the conversion of fuel to the amount of power generated is performed. Efficiency (power generation efficiency) drops.

Therefore, the monitoring system 12 detects the decrease in power generation efficiency to detect the maintenance time of the power generation facility 41. Here, the input X and the equipment state Y are not calculated by the difference between the two sensors as in the first and second embodiments, but are constructed only by the information (sensor data) of each sensor. This can be regarded as when the elements of the time series data “V ^(S2) ” and “V ^(S4) ” for correction in Examples 1 and 2 take zero values.

Specifically, "cumulative fuel flow rate" is used for the input X, while "cumulative power generation amount" is used for the equipment state Y. This "cumulative fuel flow rate" indicates a cumulative value obtained by measuring (counting) the amount of oil flowing from the fuel tank 42 into the engine 43 with a sensor S1 such as a flow meter.

Further, the "cumulative power generation amount" indicates a cumulative value obtained by measuring (counting) the amount of power generated by the generator 44 with a sensor S2 such as a power meter. Here, as in the first and second embodiments, periodic batch processing or real-time processing after the operation is stopped is performed based on the sensor data of the sensors S1 and S2 during the operation of the generator 44. The operation start / stop of the generator 44 is the same as in the second embodiment. The details of the processing contents will be described below.

≪Processing content≫
The data recording unit 14 records the measurement data measured by the measurement unit 13 through the sensors S1 and S2 every minute and records it as time-series data, and can accumulate data for one day or more for each of the sensors S1 and S2. The capacity. Here, the time-series data V ^(S2) of the sensor S2 is read as the time-series data V ^(S3) of Examples 1 and 2, and the time-series data "V ^(S2) " and "V ^(S4 ) of Examples 1 and 2 are read. ⁾ ”Is set as a zero value, and the data set“ d _(S1) to d _(S4) ”is created.

The analysis processing unit 15 performs analysis processing in the same manner as in Examples 1 and 2 based on the recorded data of the data recording unit 14. As a result of this analysis processing, a two-dimensional graph is drawn with the X-axis as the "cumulative fuel flow rate" and the Y-axis as the "cumulative power generation amount". Here, the data during the operating period from the start to the stop of the generator 44 is approximated by a linear function. The first-order coefficient of this approximate function indicates the power generation efficiency (power generated with respect to the unit fuel), and the coefficient value (slope) is stored in the state evaluation unit 16.

If any part of the power generation facility 41 deteriorates or malfunctions, the power generation efficiency of the facility as a whole decreases. For example, if the heat exchanger / cooling device of the power generation facility 41 is deteriorated or malfunctions, it is necessary to reduce the movable load of the entire facility.

Therefore, the analysis processing unit 15 calculates the linear function each time the generator 44 is operated, and the state evaluation unit 16 stores the calculated linear coefficient value (coefficient value) in the state evaluation unit 16 in advance. The occurrence of a failure sign is determined by comparing with the coefficient value at the normal time. At this time, the normal range as the threshold value is determined based on the coefficient value in the normal state stored in the state evaluation unit 16. Further, as a result of the comparison, when the coefficient value stored for each operation deviates from the normal range, a warning is issued by regarding it as the maintenance time of the power generation facility 41.

An example will be described with reference to FIGS. 14 (a) and 14 (b). FIG. 14A is a two-dimensional graph showing the state of the power generation facility 41 on a normal operating day. According to this two-dimensional graph, it is shown that the cumulative generated power on the Y-axis increases in proportion to the cumulative fuel flow rate on the X-axis, and the power generation facility 41 is operating normally.

The change in power generation efficiency in FIG. 14A, that is, “the coefficient value (slope) = 4.00” is stored in the state evaluation unit 16 as the coefficient value in the normal state. Here, as an example, it is assumed that the threshold value “the coefficient value> 3.9” in the normal range is set.

On the other hand, FIG. 14B is a two-dimensional graph showing the state of the power generation facility 41 on the day before the failure. Here, it is presumed that the fluctuation amount of the X-axis and the Y-axis is attenuated and decreased to "the coefficient value (slope) = 3.83", and the equipment condition of the power generation facility 41 is changed to reduce the power generation efficiency. Will be done.

In this case, the state evaluation unit 16 compares the coefficient value “3.83” in FIG. 14 (b) with the coefficient value “4.00” in the normal state. As a result, since the coefficient value "3.83" in FIG. 14B deviates from the normal range "the coefficient value> 3.9", it is determined that a failure sign has occurred, and a warning is given. Is done. As a result, it is possible to detect a failure or abnormality of the power generation facility 41 at an early stage.

Table 3 shows the state transition of the power generation facility 41 from "February to May 2013". This state transition is graphed in FIG.

Here, "February 2nd to March 15th" fluctuates within the range of "the coefficient value = 3.9 to 4.10". Since this is a fluctuation within the normal range "the coefficient value> 3.9", the state evaluation unit 16 does not determine that a failure sign has occurred, and is regarded as a sound power generation efficiency.

On the other hand, since "March 19th to March 29th" deviates from the normal range "the coefficient value> 3.9", it is determined that a failure sign has occurred and a warning is issued. In addition, although the power generation efficiency recovered and returned to the normal range from "April 1st to April 8th", it deviated from the normal range again on "April 11th", so a re-warning was issued.

Here, since the failure range "the coefficient value <3.8" was set, "the coefficient value = 3.48" was recorded on "April 16", so that a failure alarm was issued and emergency maintenance was urged. rice field.

In this way, the third embodiment predicts the failure of the power generation facility 41 by combining the cumulative value data of the sensors S1 and S2 and capturing the increasing tendency (slope of the two-dimensional graph). Therefore, as in the first and second embodiments, the abnormality of the power generation equipment 41 can be detected without detecting the failure of the peripheral equipment.

Further, since the sensors S1 and S2 for predicting failure may be a flow meter or a wattmeter, it is possible to detect a malfunction of the entire power generation facility 41 without conducting a sensor installation survey for each device / device. In this respect, it is easy to apply to existing equipment and can contribute to cost reduction.

The present invention is not limited to the above embodiment, and can be modified and implemented within the scope described in each claim. For example, the monitored equipment 11 of the monitoring system 12 is not limited to the exhaust gas boiler 21, the generator 31, and the power generation equipment 41, and the failure sign of other equipment (for example, the power generation equipment of the hydroelectric power plant) is generated. It can also be used for monitoring. For the application of this hydroelectric power plant to the power generation facility, the input X of the third embodiment may be replaced with the “cumulative water flow rate”.

The present invention can also be configured as a program that causes the computer to function as the monitoring system 12. According to this program, the computer functions as each of the parts 12 to 16 and can monitor the occurrence of a failure sign of the equipment to be monitored 11.

11 ... Equipment to be monitored (equipment to be monitored)
12 ... Monitoring system 13 ... Measurement unit 14 ... Data recording unit 15 ... Analysis processing unit 16 ... State evaluation unit 21 ... Exhaust gas boiler (heat exchange device)
31, 44 ... Generator 40 ... Rotating machine S1 to S4 ... Sensor

Claims (5)

A system that monitors the equipment to be monitored,
A measuring unit that measures the input given to the equipment and the equipment state of the equipment when the input is given.
A recording unit that records the measurement data of the measurement unit at regular intervals and stores the measurement data for the cycle as time-series data.
An analysis processing unit that calculates the coefficient value (slope) of the approximate function indicating the temporal change of the equipment state with respect to the input based on the time series data of the recording unit.
A state evaluation unit that determines the equipment state by comparing the coefficient value calculated by the analysis processing unit with the coefficient value of the approximate function calculated in advance based on the temporal change during normal operation .
Equipped with
The state evaluation unit determines in advance a normal range, a failure sign range, and a failure alarm range based on the coefficient value during normal operation.
The equipment state is determined according to the range to which the coefficient value calculated by the analysis processing unit belongs.
A monitoring system characterized by that. The equipment is a heat exchange device that cools the heat of the exhaust gas of the engine.
The measuring unit measures the amount of heat input to the heat exchange device and the amount of heat generated by the cooling water of the heat exchange device.
The analysis processing unit calculates a linear function that approximates the tendency of the temporal change between the calorific value and the calorific value for a predetermined period from the stop of the engine.
The monitoring system according to claim 1, wherein the state evaluation unit determines that a failure sign has occurred in the heat exchange device if the tendency of the linear function deviates from the normal range. The equipment is a generator driven by a drive source.
The measuring unit measures the temperature that fluctuates in correlation with the driving force applied to the generator and the calorific value of the lubricating oil caused by the driving force of the generator.
The analysis processing unit calculates a cubic function that approximates the tendency of the temporal change between the temperature and the calorific value in a predetermined time from the start of operation of the generator.
The monitoring system according to claim 1, wherein the state evaluation unit determines that a failure sign has occurred for the generator if the third-order coefficient deviates from the normal range. The equipment is a power generation facility including a rotary machine that generates a driving force according to a flow rate and a generator that converts the driving force of the rotary machine into generated power.
The measuring unit measures the cumulative flow rate given to the rotating machine and the cumulative generated power amount of the generator.
The analysis processing unit calculates a linear function that approximates the tendency of the temporal change between the cumulative flow rate and the cumulative power generation amount from the start to the stop of the operation of the power generation facility.
The monitoring system according to claim 1, wherein the state evaluation unit determines that a failure sign has occurred in the power generation facility if the tendency of the linear function deviates from the normal range. .. The rotary machine comprises an internal combustion engine that consumes fuel to generate driving force.
The monitoring system according to claim 4, wherein the cumulative flow rate is the cumulative fuel flow rate that has flowed into the internal combustion engine.

Images