JP2516466B2 - Measured value judgment method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial application] The present invention relates to an expert system for diagnosing an operating state of a refuse incineration facility, which is based on a sensor value abnormality of a measurement value used for inference of each rule block. The present invention relates to a measurement value determination method for determining presence / absence.

[Conventional technology]

Generally, the furnace of the refuse incinerator is configured as shown in FIG. 1, and the refuse as fuel conveyed by the crane from the refuse pit is thrown into the charging hopper (2) of the incinerator (1).

Then, the thrown-in dust is sent to the lower drying grate (3) in a state where it is accumulated as shown by the diagonal lines in the figure, and is dried by the hot air from the lower wind box (4) by the grate (3). To be done.

Furthermore, the dried refuse of the grate (3) is sent to the combustion grate (5) and burned.

In addition, the grate (5) is the front stage (5a) and the rear stage (5a).
b), and hot air for combustion is sent from the lower air boxes (6a) and (6b) to both grids (5a) and (5b).

Further, since the dust burned in the grate (5) is completely burned, it is sent to the post-combustion grate (7) and further burned.

The hot air for combustion is also sent to the grate (7) from the wind box (8) below.

Further, the ash generated by the combustion of the grate (7) is deposited in the ash pit (9) in the furnace.

On the other hand, a boiler drum (10) is provided above the incinerator (1), and water is supplied to the drum (10) from the water supply pipe line (11).

Then, the heat in the furnace generates steam in the boiler drum (10), and this steam is sent to the outside through the steam pipe (12) for use.

Further, the exhaust gas generated by the combustion passes through the water pipe group (13) and is released to the atmosphere from the exhaust pipe line (14) outside the furnace.

By the way, for example, the exhaust gas temperature is measured by the grate upper exhaust gas thermometer (15), the furnace upper exhaust gas thermometer (16) and the boiler outlet exhaust gas thermometer (17) at different installation locations.

Further, the boiler feed water amount, steam flow rate, and exhaust gas flow rate are measured by the flow meters (18), (19), (20) provided in the water supply pipeline (11), the steam pipeline (12), and the exhaust pipeline (14). Are measured respectively.

And exhaust gas thermometers (15) to (17), flowmeters (18)
Measured values of various sensors inside and outside the incinerator (1) such as measured values of (20) to (20) are read by a control device having a computer configuration.

This control device is usually constructed by a dedicated expert system for driving support, executes the diagnosis of the operating state of the equipment by inferring each rule block of this expert system, adjusts the supply of air and dust, and burns it. Automatically control the state (operating state).

Then, each rule block is described in the "if-then" format in which one or a plurality of measured values are used as antecedent variables, and the reliability of the inference depends on the measured values used as antecedent variables.

Therefore, if the measured value becomes abnormal due to a sensor abnormality (fault or the like), an inference error occurs and erroneous control or the like occurs.

In fact, it is said that one-third to two-thirds of the control troubles of this kind of expert system are abnormal measurement values due to sensor abnormality.

Then, conventionally, the measured value as the antecedent variable of each rule block is unconditionally set to a normal value (true), or the presence or absence (true, False) is determined.

In addition, batch determination before execution is performed to improve the determination accuracy.
For each measured value, for example, complicated inference using the previously measured characteristic data is performed.

[Problems to be Solved by the Invention] When the measured value for inference of each rule block is unconditionally treated as a normal value as in the conventional method, the inference reliability of each rule block is low and the control accuracy based on the inference is low. There is a problem of deterioration.

Further, if the presence / absence of abnormality of each measurement value is determined in advance by the above-described batch determination, the time required for the determination becomes long according to the number of measurement values, and if the measurement values increase, the inference of each block is increased. Takes a long time to be executed, and there is a problem that responsive control cannot be performed.

INDUSTRIAL APPLICABILITY The present invention is capable of quickly executing inference by determining whether or not there is an abnormality in a measurement value used for inference in a short time before inferring each rule block, and inference based on highly accurate determination. It is an object of the present invention to provide a measurement value determination method that can obtain a result.

[Means for solving the problem]

In order to achieve the above-mentioned object, in the measured value determination method of the present invention, before the inference of each rule block of the expert system for diagnosing the operating state of the plant equipment, the measured value used for inference and By inference based on other measured values related to the measured value, the presence / absence of abnormality of the measured value to be determined based on sensor abnormality or the like is simply determined,
After the end of the final rule block or after the end of each rule block, the presence / absence of an abnormality in the measured value of each rule block is re-determined and verified by a method different from that before the inference start of each rule block.

[Action]

In the case of the measurement value determination method of the present invention configured as described above, before the inference of each rule block is started, for each inference measurement value as the antecedent variable, another measurement associated with this measurement value. Whether or not there is an abnormality is quickly determined by simple inference based on the value and based on this determination, the inference of each rule block is executed in order after a short determination of each measured value, Immediately move to the rule block.

Furthermore, after the final rule block or after the end of each rule block, the presence / absence of an abnormality in the measured value is re-determined and verified to improve the determination accuracy.

Therefore, the reliability of inference is higher than in the case where the measured value of each rule block is unconditionally treated as a normal value, and each rule block is swiftly compared to the case where the measured value for inference of each rule block is first determined collectively. Block inference is performed.

Moreover, the verification after the inference can provide the highly reliable inference result based on the highly accurate determination result.

〔Example〕

An embodiment applied to an expert system for supporting the operation of a refuse incineration facility will be described with reference to FIGS. 1 and 2.

First, the measured values of the exhaust gas thermometers (15) to (17) and the measured values of the flow meters (18) to (20) of FIG. 1 will be described.

The measurement value of the furnace upper exhaust gas thermometer (16) is an important indicator of operating conditions.

Then, in the automatic combustion control based on the inference result of the expert system, the measurement value of the furnace upper exhaust gas thermometer (16) is controlled so as to fall within a certain temperature range.

However, in the furnace upper exhaust gas thermometer (16) in which the working atmosphere is high temperature and high dust, disconnection due to high temperature oxidation and the thermometer cover due to the molten clinker are likely to occur, and the reliability of the measured value is low.

This also applies to the grate furnace upper exhaust gas thermometer (15).

On the other hand, the boiler outlet exhaust gas thermometer (17) has a low working atmosphere temperature of 150 to 300 ° C., and has a small amount of scattered dust, so that the exhaust gas thermometers (15) and (16) are The reliability of measured values is extremely high.

And the measured value of each exhaust gas thermometer (15)-(17) is
Correlated and related.

Further, the measurement value of the steam flow meter (19) that faithfully reflects the combustion state of the incinerator (1) is also an important indicator of the operating state.

The measurement value of the steam flow meter (19) is associated with the measurement values of the boiler feed water meter (18) and the exhaust gas flow meter (20).

At this time, the boiler feed water amount is controlled and determined from the amount of steam generated by the steam flow meter (19) and the height of the level meter of the boiler drum (10). 19) Greater than.

Therefore, compared to the value measured by the boiler water supply meter (18),
The reliability of the measured value of the steam flow meter (19) is high.

In addition, the exhaust gas flow meter (20), which is one of the important indicators of the operating state, is apt to cause troubles such as gas drift, blockage of the pitot tube due to dust, etc. Lower than 18) and (19).

Next, the inference of the rule block for diagnosing the operating state from the measurement value of the furnace upper exhaust gas thermometer (16) (hereinafter referred to as the furnace upper measurement value) will be described.

This inference is based on whether or not the measured value at the top of the furnace is 1000 ° C or higher, and refers to the oxygen concentration at the outlet of the boiler to determine whether or not rapid combustion of dust that exceeds the automatic combustion control capability has occurred. , "If-then" format is described as follows.

"If the measured value at the upper part of the furnace is 1000 ° C or higher and the oxygen concentration at the boiler outlet is decreasing at 6% or lower, (there is a possibility that rapid combustion of dust that exceeds the automatic combustion control capability has occurred. Therefore, increase the secondary air so that the secondary combustion in the furnace does not become insufficient, and increase the combustion air from the post-combustion grid. " The oxygen concentration at the boiler outlet does not drop below 6%.

Therefore, for the rule block, the measured value at the upper part of the furnace is used as the measured value to be judged, and whether or not there is an abnormality is judged only for this measured value, and the inference is executed only when the measured value is normal. Control reliability is improved.

When the measured value at the upper part of the furnace is 1000 ° C or higher in a normal state, the measured value of the highly reliable boiler outlet exhaust gas thermometer (17) related to this value (hereinafter referred to as boiler outlet measured value) is It will be in the range of 250-300 ℃.

In addition, the rule block has a furnace top measurement value of 1000 ° C.
It is only necessary to ignite in the above cases.

Therefore, immediately before the rule block, the following reasoning for determining whether or not there is an abnormality in the measurement value of the upper part of the furnace is described.

“If furnace upper measured value ≧ 1000 ° C & boiler outlet measured value ≧ 250 ° C & boiler outlet measured value ≦ 300 ° C then the furnace upper measured value can be regarded as normal value (true)” In this case, the above rule using the furnace upper measured value Before starting the inference of the block, only for the measurement value of the upper part of the furnace used for the inference, the above-mentioned simple inference for determination based on the measurement value of the boiler outlet, which is related to this measurement value, causes an abnormality due to a sensor abnormality, etc. in a short time. The presence or absence of is determined.

Then, only when the measured value in the upper part of the furnace is normal and 1000 ° C. or higher, the rule block is ignited, the combustion state is diagnosed, and necessary control is executed.

If the measured value at the top of the furnace is abnormal or below 1000 ° C,
The rule block does not fire and immediately jumps to the next rule block.

Immediately before each rule block, an inference similar to the above inference for presence / absence of abnormality is described, and before the inference of each rule block, only the required measurement value is measured. Whether or not there is an abnormality in the measured value is quickly determined by simple inference using other measured values related to

Further, based on the result of this determination, each rule block is executed at least only when the measured value is normal.

Then, since the block of abnormal measurement value is skipped and the rule block of normal measurement value is sequentially moved, necessary control based on the normal measurement value can be promptly performed without omission, and control reliability is improved.

By the way, in the case of a simple inference before the inference of each rule block is started, for example, even when the furnace upper part measured value is normal and the boiler outlet measured value is abnormal, the furnace upper part measured value is regarded as abnormal, and this measured value is The rule block used will be skipped and will not be executed.

Even if the measured value in the upper part of the furnace is abnormal, it may be considered as normal depending on the degree.

In view of this, in this embodiment, after the final rule block of the expert system is finished, each measurement value that is simply determined in each rule block is collectively re-determined and verified by reasoning different from the simple determination.

 Next, the reasoning for re-determination will be described.

In the case of the furnace upper part measured values, for example, characteristic data of the boiler outlet gas temperature and the steam flow rate at the furnace upper part temperatures 800 ° C., 900 ° C. and 1000 ° C. shown in FIG. 2 are prepared in advance.

In the characteristic lines (l ₁ ), (l ₂ ), and (l ₃ ) at 800 ° C, 900 ° C, and 1000 ° C in Fig. ₂ , the range of the solid line shows the normal operating region, and the one-dot chain line shows the abnormal operating region. Indicates.

Then, based on the condition of the oxygen concentration at the boiler outlet, the steam flow rate, the furnace upper temperature, the measured value of the boiler outlet gas temperature, that is, the combination of the measured values of the flow meter (19), the exhaust gas thermometer (16), (17). , The exhaust gas thermometer (1
Examine the presence / absence of abnormalities in the measured values in 6) (measured values in the upper part of the furnace) in detail and verify.

"If boiler outlet oxygen concentration is 8-12% And if Steam inflow = 30ton / h (hour) ± 5% & furnace upper temperature = 1000 ℃ ± 1% & boiler outlet gas temperature = 320 ℃ ± 5% then furnace upper The measured value is normal. ”When a determination result different from the simple determination before the inference start is obtained, this determination result is prioritized.

Due to this priority, whether or not to execute the rule block is determined based on the result of the immediately preceding re-determination, for example, not in the result of the simple determination in the next cycle.

Then, each measurement value that is simply determined is verified by the detailed re-determination similar to the above.

Therefore, the judgment accuracy of each measured value is further improved, the situation in which rule blocks are inadvertently skipped or erroneously executed does not occur, and highly reliable inference results based on highly accurate judgment results are obtained and control is performed. The reliability of is significantly improved.

The simple determination of the measurement value of each rule block may be redetermined at the end of each rule block.

And, of course, it can be applied to an expert system for diagnosing the operating states of various plant equipment.

〔The invention's effect〕

Since the present invention is configured as described above,
The following effects are achieved.

Before the inference of each rule block is started, regarding the measured value for inference as each antecedent variable, a simple inference based on this measured value and other measured values makes it possible to quickly detect whether or not there is an abnormality. When each rule block is inferred based on this determination, the rule block is quickly moved from one rule block to the next, and at this time, each measured value is unconditionally treated as a normal value. The reliability of the inference becomes higher, and moreover, the inference of each rule block can be started more quickly than in the case where the measured values of each rule block are first judged collectively.

Furthermore, after the final rule block or after each rule block is finished, the presence or absence of anomaly in the measured value is re-determined and verified by a method different from that before the inference start of each rule block, so the determination accuracy is improved. , It is possible to obtain highly reliable inference results based on highly accurate determination results.

Therefore, it is possible to perform highly accurate diagnosis and the like of the operating state of the refuse incineration facility by quick inference, and it is possible to improve the reliability of the diagnosis and control based on the inference result while suppressing deterioration of its responsiveness. .

[Brief description of drawings]

1 and 2 show one embodiment of the measurement value determination method of the present invention. FIG. 1 is an explanatory view of the construction of a refuse incineration facility, and FIG. 2 is an explanatory view of characteristic data used for re-determination. (15), (16), (17) ... Exhaust gas thermometer, (18), (1
9), (20) ... Flowmeter.

Claims (1)

(57) [Claims] 1. An inference based on a measured value to be judged used for inference and other measured values related to the measured value before starting the inference of each rule block of an expert system for diagnosing the operating state of plant equipment. Therefore, the presence / absence of an abnormality in the determined measurement value based on a sensor abnormality or the like is simply determined, and after the end of the final rule block or after the end of each rule block, the abnormality of the determined measurement value of each rule block is detected. A measurement value determination method characterized by re-determining and validating the presence / absence of each rule block by a method different from that before the inference start.