JP2008262375A - Field device with function of outputting result of abnormality diagnosis and method of outputting result of abnormality diagnosis - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a field device for performing abnormality diagnosis which performs a pattern recognition process by weighting the time-series data of abnormality diagnosis results in order to reduce the burden on a user by appropriately letting a host device or the like know about an abnormal condition, and which then provides outputs after determining whether or not it possible to output the abnormality diagnosis results based on the result of the pattern recognition process, or an abnormality diagnosis result output module, and a method of outputting abnormality diagnosis results. <P>SOLUTION: The field device for performing abnormality diagnosis is configured to determine whether or not it is possible to output abnormality diagnosis results based on the time-series pattern of the abnormality diagnosis results. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a field device or an abnormality diagnosis result output module for performing an abnormality diagnosis, weighting the time series data of the abnormality diagnosis results to perform pattern recognition processing, and outputting the abnormality diagnosis results based on the pattern recognition processing results The present invention relates to a field device or an abnormality diagnosis result output module that determines whether or not it is output and an abnormality diagnosis result output method thereof.

  The field device calculates a process value and a diagnostic variable value based on a physical quantity detected by the sensor, and performs an abnormality diagnosis process for diagnosing a sensor failure or the like. Then, the field device counts the abnormality diagnosis result, and outputs the abnormality diagnosis result when the counted value is equal to or greater than a predetermined determination value. Such a field device will be described with reference to FIG.

  The field device 13 includes a sensor 1, an input processing unit 2, a drive unit 3, a calculation unit 4, an abnormality diagnosis unit 5, a counting unit 6, an output determination unit 10, an output processing unit 11, a display unit 12, and the like. The output determination unit 10 includes a storage unit 7, a comparison unit 8, an output selection unit 9, and the like. The field device 13 is connected from the output processing unit 11 via a transmission line (not shown) to a host device, a setting device (none of which are shown), and the like, thereby configuring plant equipment.

  The sensor 1 driven by the drive unit 3 detects a measured physical quantity and converts it into an electrical signal. The input processing unit 2 receives an electric signal from the sensor 1 and amplifies it to convert it into a digital value. The drive unit 3 receives the amplified signal and the digital value from the input processing unit 2 and drives the sensor 1 based on the received signal. The calculation unit 4 receives a digital value from the input processing unit 2 and calculates a process value and a diagnostic variable value. The output processing unit 11 receives a process value from the calculation unit 4 and outputs a current or voltage signal corresponding to (proportional to) the process value to a host device or the like. The display unit 12 receives a calculation value such as a process value from the calculation unit 4 and displays it.

  The abnormality diagnosis unit 5 receives a process value and a diagnostic variable value from the calculation unit 4, and performs abnormality diagnosis based on the value. The counting unit 6 counts the abnormality diagnosis result. The comparison unit 8 compares the count value with the determination value stored in the storage unit 7 and sends a selection signal indicating that the count value is less than the determination value or greater than the determination value to the output selection unit 9. .

Based on the selection signal, the output selection unit 9 does not output the abnormality diagnosis result to the output processing unit 11 or the display unit 12 when the count value is less than the determination value, and when the count value is equal to or greater than the determination value. The abnormality diagnosis result is output to the output processing unit 11 and the display unit 12.

When the output processing unit 11 receives an abnormality diagnosis result from the output selection unit 9, the output processing unit 11 outputs the abnormality diagnosis result as a communication signal to a host device or the like. When receiving the abnormality diagnosis result from the output selection unit 9, the display unit 12 displays the abnormality diagnosis result.

Here, operations of the counting unit 6 and the output determining unit 10 will be described with reference to FIG. In FIG. 6, the horizontal axis represents time, and the vertical axis represents the count value obtained by counting the abnormality diagnosis results.

  The abnormality diagnosis unit 5 performs one or more abnormality diagnosis based on the process value and the diagnosis variable value at regular or irregular time intervals. The abnormality diagnosis unit 5 outputs the abnormality diagnosis result to the counting unit 6 as an abnormal state or a normal state for each abnormality diagnosis. The counting unit 6 counts abnormality diagnosis results for each abnormality diagnosis.

The counter 6 increases the count value (for example, adds 1 to the current count value) when the abnormality diagnosis result is an abnormal state, and decreases the count value (for example, from the current count value). Subtract 1). During the time t1, the abnormality diagnosis result is in a normal state, so the count value is maintained at 0. The count value is 0 as the lower limit. During time t2, the abnormality diagnosis result becomes an abnormal state, and the count value increases. Since the count value is less than the determination value between times t1 and t2, the output selection unit 9 does not output an abnormality diagnosis result based on the selection signal from the comparison unit 8. Therefore, the abnormal state during the time t2 is not output to the host device or the like and is not displayed on the display device 12.

During time t3, the abnormality diagnosis result continues to be in an abnormal state, and the count value is equal to or greater than the determination value. Therefore, the output selection unit 9 outputs the abnormality diagnosis result based on the selection signal from the comparison unit 8. Therefore, the abnormal state is output to the host device and displayed on the display device 12. The count value has a determination value as an upper limit value.

During time t4, the abnormality diagnosis result is in a normal state, and the count value decreases. Since the count value is less than the determination value, the output selection unit 9 does not output an abnormality diagnosis result based on the selection signal from the comparison unit 8.

Here, the abnormality diagnosed in each abnormality diagnosis includes a low importance and a high importance. A case of low importance is, for example, a case where electrical noise is temporarily added to a field device. Since this is a temporary phenomenon, the field device 13 does not want to output an abnormal state to the host device or the like or display it on the display device 12. Moreover, as a highly important thing, it is a case where sensor abnormality and an electrical noise are continuously added to a field apparatus, for example. In such a case, it is desirable that the field device 13 outputs an abnormal state to the host device or the like or displays it on the display device 12.

Therefore, the storage unit 7 can receive a determination value as a communication signal from the host device or the like via the output processing unit 11 and set the determination value. When the host device or the like reduces the determination value with respect to the count value of the sensor abnormality diagnosis result, the field device 13 can quickly output the sensor abnormality state to the host device or the like, or display it on the display device 12. Further, when the host device or the like increases the determination value for the count value of the abnormality diagnosis result due to electrical noise mixing, the field device 13 outputs an abnormal state due to temporarily applied electrical noise to the host device or the like or displays On the other hand, an abnormal state due to electrical noise applied continuously instead of temporarily can be output to the host device or the like or displayed on the display unit 12.

In this way, by setting (changing) the judgment value for each abnormality diagnosis result, the user can preferentially know the abnormality having high importance by the host device or the like, while the abnormality having low importance is Users will not be informed. Therefore, the user can reduce the burden on abnormality monitoring, cause investigation, and countermeasures, and can concentrate on the investigation and countermeasures of the highly important abnormality.

Note that abnormalities of low importance include field vibrations such as pressure gauges and flow meters, such as vibrations to the sensor, bubbles and slurry generated in the measurement fluid, and large pressure caused by water hammer generated by opening and closing the valves. May join.

JP-A-7-93682

  The size and frequency of occurrence of the electrical noise, vibration, bubbles, etc. may vary depending on the process state (ambient noise environment, field device installation state, pump operating state for fluid flow, etc.). Therefore, the user needs knowledge about the state of the process in order to set each judgment value to a value suitable for his / her request (priority to know abnormal condition, etc.), and in some cases it is difficult to determine the value. There is a case. In addition, the process state may change over time, and the user needs to change the setting of each determination value again according to the process state periodically or irregularly. As described above, there are cases where the burden of monitoring an appropriate determination value or process state by the user increases.

  An object of the present invention relates to a field device or an abnormality diagnosis result output module for performing an abnormality diagnosis, and weights the time series data of an abnormality diagnosis result in order to appropriately notify an abnormal state to a host device and reduce the burden on the user. To provide a field device or an abnormality diagnosis result output module that performs pattern recognition processing, determines whether or not to output the abnormality diagnosis result based on the pattern recognition processing result, and outputs the abnormality diagnosis result, and a method for outputting the abnormality diagnosis result .

In order to achieve such an object, the invention of claim 1
In a field device including an abnormality diagnosis unit that performs abnormality diagnosis, a counting unit that counts an abnormality diagnosis result detected by the abnormality diagnosis unit, and an output determination unit that determines whether the abnormality diagnosis result can be output based on the count value ,
The output determination unit includes a storage unit that stores the count values in time series, a pattern recognition processing unit that performs pattern recognition processing by weighting the count values stored in the storage unit, and the pattern recognition processing unit. Determining whether the abnormality diagnosis result can be output based on the pattern recognition processing result processed in
It is characterized by that.

The invention of claim 2 is the invention of claim 1,
The pattern recognition process is a hierarchical neural network process or an interconnected neural network process.
It is characterized by that.

The invention of claim 3 is the invention of claim 2,
Based on a signal including the priority of the output abnormality diagnosis result, a combined load used in the pattern recognition process or a combined load determining unit that determines a predetermined threshold;
The pattern recognition process is processed using the determined combined weight or a predetermined threshold value.
It is characterized by that.

The invention of claim 4
An abnormality diagnosis result including an abnormality diagnosis unit that performs abnormality diagnosis, a counting unit that counts an abnormality diagnosis result detected by the abnormality diagnosis unit, and an output determination unit that determines whether the abnormality diagnosis result can be output based on the count value In the output module:
The output determination unit includes a storage unit that stores the count values in time series, a pattern recognition processing unit that performs pattern recognition processing by weighting the count values stored in the storage unit, and the pattern recognition processing unit. Determining whether the abnormality diagnosis result can be output based on the pattern recognition processing result processed in
It is characterized by that.

The invention of claim 5
In the abnormality diagnosis result output method comprising performing abnormality diagnosis, counting the detected abnormality diagnosis result, and determining whether or not the abnormality diagnosis result can be output based on the counted value,
The step of determining whether or not the output is possible includes a step of storing the count value in time series, a step of weighting the stored count value and performing pattern recognition processing, and the abnormality based on the pattern recognition processing result Having a step of judging whether or not to output a diagnosis result;
It is characterized by that.

The invention of claim 6
In field devices that perform abnormality diagnosis,
Based on the time series pattern of the abnormality diagnosis result, configured to determine whether the abnormality diagnosis result can be output,
It is characterized by that.

The present invention relates to a field device or an abnormality diagnosis result output module for performing an abnormality diagnosis, weighting the time series data of the abnormality diagnosis result to perform pattern recognition processing, and based on the pattern recognition processing result, the abnormality diagnosis result By the field device or abnormality diagnosis result output module that outputs whether the output is possible and the abnormality diagnosis result output method and the abnormality diagnosis result output method, it is possible to appropriately notify the host device or the like of the abnormal state and reduce the burden on the user.

[First embodiment]
A first embodiment will be described with reference to FIG. FIG. 1 is a block diagram of a field device to which the present invention is applied. The same components as those in FIG. In this embodiment, the count values are stored in time series, the values are weighted to perform pattern recognition processing, and based on the result, it is determined whether the abnormality diagnosis result can be output.

  The field device 18 includes a sensor 1, an input processing unit 2, a drive unit 3, a calculation unit 4, an abnormality diagnosis unit 5, a counting unit 6, an output determination unit 17, an output processing unit 11, a display unit 12, and the like. The output determination unit 17 includes a storage unit 14, a pattern recognition processing unit 15, an output selection unit 16, and the like. The field device 18 is connected from the output processing unit 11 via a transmission line (not shown) to a host device, a setting device (none of which are shown), and the like, thereby configuring plant equipment.

  The sensor 1 driven by the drive unit 3 detects a measured physical quantity and converts it into an electrical signal. Measurement physical quantities include pressure, flow rate, and temperature. Sensors that detect them include vibration pressure sensors, electromagnetic flowmeters and vortex flowmeter sensors that detect electromotive force and eddy frequency generated in the measurement fluid. There are temperature sensors. The input processing unit 2 receives an electrical signal from the sensor 1, amplifies the signal (including removal of superimposed noise), and converts the amplified signal into a digital value by an AD converter or the like. The drive unit 3 receives the amplified signal and the digital value from the input processing unit 2 and drives the sensor 1 based on the received signal. For example, in a vibration type pressure sensor, the sensor 1, the input processing unit 2, and the driving unit 3 form a resonance circuit to vibrate the sensor 1. In the electromagnetic flow meter sensor, the drive unit 3 generates an electromotive force by applying a magnetic field to the fluid to be measured by causing a current to flow through an exciting coil (not shown) at a predetermined frequency regardless of an amplification signal or a digital value. Let

The calculation unit 4 receives a digital value from the input processing unit 2 and calculates a process value and a diagnostic variable value. The process value includes a pressure value, a flow rate value, a temperature value, and the like. The output processing unit 11 receives the process value from the arithmetic unit 4 and outputs a current corresponding to (proportional to) the process value (for example, a range of 4 to 20 milliamperes) or a voltage signal (for example, a range of 1 to 5 volts) as a host device or the like. Output to. Further, the output processing unit 11 may transmit the process value as a digital signal by fieldbus communication. The display unit 12 receives a calculation value such as a process value from the calculation unit 4 and displays the value. The display part 12 should just be visually recognizable besides a liquid crystal display.

  The abnormality diagnosis unit 5 receives a process value and a diagnostic variable value from the calculation unit 4, and performs abnormality diagnosis based on the value. The abnormality diagnosis includes sensor abnormality, process value range abnormality, vibration abnormality, bubble abnormality, slurry abnormality and water hammer abnormality. The sensor abnormality occurs when the sensor 1 fails, and the process value out-of-process abnormality occurs when the process value exceeds a predetermined range. Vibration abnormality, bubble abnormality and slurry abnormality occur when vibration, bubble amount and slurry noise are detected, respectively. Water hammer abnormality detects pressure greater than the predetermined value when the valve is opened and closed. Occurs when.

The abnormality diagnosis unit 5 performs each abnormality diagnosis periodically or irregularly, detects an abnormal state if it is abnormal, detects a normal state if it is normal, and outputs these as an abnormality diagnosis result. The counting unit 6 counts each abnormality diagnosis result. The memory | storage part 14 receives each count value from the counting part 6, and memorize | stores those values in time series. The pattern recognition processing unit 15 receives each count value stored in time series from the storage unit 14 and performs pattern recognition processing on these values.

The output selection unit 16 receives the pattern recognition processing result from the pattern recognition processing unit 15 as a selection signal, and determines whether to output the abnormality diagnosis result to the output processing unit 11 or the display unit 12 based on the signal. I do. Detailed operations of the counting unit 6 and the output determining unit 17 will be described later.

When the output processing unit 11 receives an abnormality diagnosis result from the output selection unit 16, the output processing unit 11 outputs the abnormality diagnosis result as a communication signal to a host device or the like. When the display unit 12 receives the abnormality diagnosis result from the output selection unit 16, the display unit 12 displays the abnormality diagnosis result.

Here, the abnormality diagnosed in each abnormality diagnosis includes a low importance and a high importance. A case of low importance is, for example, a case where electric noise is temporarily applied to the field device 18. Since this is a temporary phenomenon, the field device 18 does not want the abnormal state to be output to the host device or the like or displayed on the display 12. Moreover, as a highly important thing, it is a case where sensor abnormality and an electrical noise add to the field apparatus 18 continuously, for example. In such a case, it is desirable that the field device 18 outputs an abnormal state to the host device or the like or displays it on the display device 12. Other less significant abnormalities include vibrations to the sensor, bubbles and slurry generated in the measurement fluid, and the large pressure generated by the water hammer generated by opening and closing the valve. May join field equipment.

The operation of the counting unit 6 in the abnormal state will be described with reference to FIG. In FIG. 3, the horizontal axis represents time, and the vertical axis represents the count value obtained by counting the abnormality diagnosis results. The counter 6 increases the count value (for example, adds 1 to the current count value) when the abnormality diagnosis result is an abnormal state, and decreases the count value (for example, from the current count value). Subtract 1).

  A case where electrical noise is temporarily applied to the field device 18 as an abnormality of low importance will be described with reference to FIG. The abnormality diagnosis results are as follows: normal state for D1, abnormal state for D2 and D3, normal state between D4 and Dn-4, abnormal state for Dn-3 and Dn-2, Dn-1 and Dn Sometimes it is in a normal state. In this case, D1 to Dn represent a time series pattern of count values. Further, as a highly important abnormality, for example, when a sensor abnormality or electrical noise is continuously applied to the field device 18, the count value is a time-series pattern from D1 to Dn shown in FIG.

  The storage unit 14 stores the count values shown in FIGS. 3A and 3B in time series (for example, count values from D1 to Dn). The pattern recognition processing unit 15 performs pattern recognition processing by weighting each count value (for example, count values from D1 to Dn) stored in time series, which is one of the time series patterns of the abnormality diagnosis result. .

  As one of the pattern recognition processing, hierarchical neural network processing will be described with reference to FIG. FIG. 4 is a data processing diagram of hierarchical neural network processing.

  Data processing of hierarchical neural network processing includes an input layer 21, an intermediate layer 22, an output layer 23, and the like. Count values stored in time series (for example, count values from D1 to Dn) are given to the input layer 21. Each unit (29, etc.) constituting the intermediate layer 22 performs a weighting operation on each data received from the input layer 21 using a coupling load (28, etc.), and compares the calculated value with a predetermined threshold value. The result is output to the output layer 23. The unit 27 constituting the output layer 23 performs a weighting operation on the data received from the intermediate layer 22 using the coupling loads 24, 25 and 26, and compares the calculated value with a predetermined threshold value (Dout). Is sent to the output selection unit 16 as the output of the pattern recognition processing unit 15.

When Dout is high level data (for example, 1), the output selection unit 16 outputs an abnormal state to the output processing unit 11 and the display unit 12, and Dout is low level data (for example, 0). In this case, the abnormal state is not output to the output processing unit 11 and the display unit 12.

  The output processing unit 11 outputs data indicating an abnormal state received from the output selection unit 16 to a host device or the like as a communication signal. The display unit 12 displays the abnormal state. Here, in the case of the time-series pattern of the count values shown in FIG. 3B, since the count value has many 0, Dout can be low level data, and the output processing unit 11 is in an abnormal state. Do not output certain data. On the other hand, in the case of the time-series pattern of the count value shown in FIG. 3A, since the count value has a value larger than 0, Dout can be high-level data, and the output processing unit 11 is in an abnormal state. Outputs status data.

  Note that hierarchical neural network processing includes back propagation, neocognitron, and multi-module neural network processing. In addition to hierarchical neural network processing, there is also mutual coupling neural network processing, which includes Hopfield network, Boltzmann machine, competitive learning vector quantization neural network, learning vector quantization method (LVQ model) processing.

According to the present embodiment, for the field device or abnormality diagnosis result output module for performing abnormality diagnosis, pattern recognition processing is performed by weighting the time series data of the abnormality diagnosis result, and the abnormality diagnosis result is calculated based on the pattern recognition processing result. A field device or an abnormality diagnosis result output module that determines whether or not output is possible and an abnormality diagnosis result output method and an abnormality diagnosis result output method can appropriately notify an abnormal state to a host device or the like, thereby reducing the burden on the user.

[Second Embodiment]
A second embodiment will be described with reference to FIG. FIG. 2 is a block diagram of a field device to which the present invention is applied. The same components as those in FIG. In this embodiment, the user confirms the output abnormal state and informs the field device whether or not the output of the abnormal state is required or the priority order of necessity, and the field device It has a learning function for changing the coupling weight of the pattern recognition processing based on the information.

The field device 20 includes a combined load determination unit 19 that receives the information from the output processing unit 11 in addition to the configuration of FIG. In the first embodiment, when the user confirms the abnormal state that is output or displayed on the host device or the display device 12, the user can use the abnormal state as information including the priority of the output abnormality diagnosis result. Determine whether you need output or the priority of its necessity. The host device or the like transmits the information determined by the user as a signal to the field device 20, and the output processing unit 11 receives the information (signal) and outputs it to the combined load determination unit 19.

Based on the received information (signal), the joint load determination unit 19 determines a joint load and a predetermined threshold value used for the pattern recognition processing, and sends them to the pattern recognition processing unit 15. The pattern recognition processing unit 15 sets the combined load and a predetermined threshold value to the values sent from the combined load determination unit 19, and performs the pattern recognition process based on the set values. For example, if the output of the abnormal state is not required or the priority of the necessity is low, the connection load determination unit 19 decreases the connection load related in the pattern recognition process or increases the predetermined threshold value. . If the output of the abnormal state is required or the priority of the necessity is high, the connection load determination unit 19 increases the connection load related in the pattern recognition process or decreases the predetermined threshold value. .

According to this embodiment, regarding a field device or an abnormality diagnosis result output module for performing an abnormality diagnosis, a user request for output or display of an abnormal state is accepted and reflected on a value such as a coupling load in pattern recognition processing based on the request. By using a field device having a learning function or an abnormality diagnosis result output module and an abnormality diagnosis result output method thereof, it is possible to appropriately notify the host device or the like of the abnormal state and reduce the burden on the user.

Note that these embodiments are not limited to field devices, and can also be realized in an abnormality diagnosis result output module. In addition, field devices include temperature transmitters and level meters that detect temperature and level as measurement targets (physical quantities), and flow meters include vortex flow meters, ultrasonic flow meters, and Coriolis types in addition to electromagnetic flow meters. There is a flow meter.

It is a block diagram of a field device to which the present invention is applied. It is another example of the block diagram of the field device to which the present invention is applied. It is a figure showing the count value of the time series in the counting part. It is a data processing figure of a hierarchical neural network process. It is a block diagram of the conventional field device. It is a figure showing the comparison with the count value of the conventional time series, and a determination value.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Sensor 2 Input processing part 3 Drive part 4 Operation part 5 Abnormality diagnosis part 6 Counting part 11 Output processing part 12 Display part 14 Storage part 15 Pattern processing recognition part 16 Output selection part 17 Output judgment part 18 Field apparatus

Claims (6)

In a field device including an abnormality diagnosis unit that performs abnormality diagnosis, a counting unit that counts an abnormality diagnosis result detected by the abnormality diagnosis unit, and an output determination unit that determines whether the abnormality diagnosis result can be output based on the count value ,
The output determination unit includes a storage unit that stores the count values in time series, a pattern recognition processing unit that performs pattern recognition processing by weighting the count values stored in the storage unit, and the pattern recognition processing unit. Determining whether the abnormality diagnosis result can be output based on the pattern recognition processing result processed in
Field equipment characterized by that. The pattern recognition process is a hierarchical neural network process or an interconnected neural network process.
The field device according to claim 1, wherein: Based on a signal including the priority of the output abnormality diagnosis result, a combined load used in the pattern recognition process or a combined load determining unit that determines a predetermined threshold;
The pattern recognition process is processed using the determined combined weight or a predetermined threshold value.
The field device according to claim 2, wherein: An abnormality diagnosis result including an abnormality diagnosis unit that performs abnormality diagnosis, a counting unit that counts an abnormality diagnosis result detected by the abnormality diagnosis unit, and an output determination unit that determines whether the abnormality diagnosis result can be output based on the count value In the output module:
The output determination unit includes a storage unit that stores the count values in time series, a pattern recognition processing unit that performs pattern recognition processing by weighting the count values stored in the storage unit, and the pattern recognition processing unit. Determining whether or not to output the abnormality diagnosis result based on the pattern recognition processing result processed in
An abnormality diagnosis result output module. In the abnormality diagnosis result output method comprising performing abnormality diagnosis, counting the detected abnormality diagnosis result, and determining whether or not the abnormality diagnosis result can be output based on the counted value,
The step of determining whether or not the output is possible includes a step of storing the count value in time series, a step of weighting the stored count value and performing pattern recognition processing, and the abnormality based on the pattern recognition processing result Having a step of judging whether or not to output a diagnosis result;
An abnormality diagnosis result output method characterized by the above. In field devices that perform abnormality diagnosis,
Based on the time series pattern of the abnormality diagnosis result, configured to determine whether the abnormality diagnosis result can be output,
Field equipment characterized by that.

Images