JPH0763586A - Mutual diagnostic method for sensor - Google Patents

Abstract

PURPOSE:To detect abnormality in an early stage by computing deviation or the degree of abnormality from the integrated quantity or variation per unit time and the integrated quantity or variation of the output of plural other related sensors excluding a concerned sensor, and making a fuzzy inference. CONSTITUTION:The inflow water quantity to a distributing reservoir 1 is detected by an inflow meter 2 and inputted to an inflow integrating part 5 so as to compute the integrated quantity DELTAF1 of inflow water per unit time. The water level of the distributing reservoir 1 is detected by a distributing reservoir level gage 3 and inputted to a water level variation computing part 6 so as to compute the water level variation DELTAL per unit time. The outflow water quantity from the distributing reservoir 1 is further detected by an outflow meter 4 and outputted to an outflow integrating part 7 so as to compute the integrated quantity DELTAF2 of outflow water per unit time. The inflow DELTAF1, variation DELTAL, outflow DELTAF2 and the cross-sectional area of the distributing reservoir 1 are inputted to a water level-flow deviation computing part 8 so as to compute the water level deviation quantity HL, and inflow and outflow deviation quantity HF1, HF2 respectively. The inflow and outflow DELTAF1, DELTAF2 are inputted to a reserved quantity computing part 9 so as to compute the reserved quantity DELTAF0.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a sensor mutual diagnosis system for early detection of abnormalities in a plurality of sensors used for monitoring and controlling a plant or process by fuzzy inference based on inter-sensor related information.

[0002]

2. Description of the Related Art For example, in a water supply system, there are various processes from water intake to drainage, and there are facilities corresponding thereto. The distribution reservoir is one of them, and it is an important facility for the quantitative operation of the water supply. Its operation and management is based on the water level of the distribution reservoir, the inflow rate to the distribution reservoir, and the outflow rate from the distribution reservoir. There is.

Therefore, the sensor for detecting such information is important for operation management, and it is necessary to detect the abnormality early and take necessary measures immediately.

Generally, the abnormality of the sensor is such that a single judgment is made by a level check whether the sensor output deviates from a unified signal range (for example, DC 4 to 20 mA), and further. It is necessary for the operator to comprehensively judge the information of the plurality of related sensors.

[0005]

Therefore, there are the following problems.

(1) Deviation from the unified signal range of the sensor is a level check at the final stage when the sensor becomes abnormal.
Preventive maintenance cannot be performed by early detection of abnormalities.

(2) Since the judgment is made only by the single information of the sensor itself, an abnormality in the unified signal range cannot be detected in principle.

(3) Since it is not a mutual diagnostic method that utilizes a plurality of related sensor information, an operator can judge a sensor abnormality while making a comprehensive judgment on the process status and the organic relationship between the sensors. As a result, the burden on the operator is increased.

The present invention has been made in view of the above problems, and an object of the present invention is to early detect sensor abnormality within a unified signal range, simultaneously detect abnormality of a plurality of sensors, and reduce operator's judgment load. To do.

[0010]

[Means and Actions for Solving the Problems] In a method of diagnosing a sensor for detecting an abnormality of a plurality of sensors used for monitoring and controlling a plant or a process, an integrated amount or change amount of sensor output per unit time and each sensor The degree of abnormality of each sensor is defined by using the deviation amount estimated from the accumulated amount or change amount of the output from the accumulated amount or change amount of the outputs of other sensors related to the sensor output other than the relevant sensor as a phenomenon item. Fuzzy inference is performed as a cause item to detect anomalies in multiple sensors.

[0011]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will now be described with reference to the block diagram of the sensor mutual diagnosis system of FIG. 1, the membership function of FIG. 2 and the rule matrix of Table 1.

Reference numeral 1 is a distribution reservoir, 2 is an inflow flow meter, 3 is a distribution reservoir water level meter, and 4 is an outflow flow meter. An inflow flow rate integrating unit 5 integrates the output of the inflow flowmeter 2 per unit time and outputs an integrated flow rate ΔF ₁ . A water level change amount calculation unit 6 calculates a water level change amount ΔL per unit time (about 60 minutes) of the distribution reservoir 1. Reference numeral 7 denotes an outflow flow rate integrating unit, which integrates the output of the outflow flow rate meter 4 per unit time to obtain an integrated flow rate ΔF.
Output ₂ Reference numeral 8 denotes a water level / flow rate deviation calculating unit, which inputs the integrated inflow amount ΔF ₁ , the water level change amount ΔL, the integrated outflow amount ΔF ₂ and the cross-sectional area A of the drainage basin, and changes in the integrated inflow / outflow flow rate and the distribution water level. The absolute value of the deviation amount between the actual measurement value of each sensor and the expected value from other sensor information is calculated by the following equation: Water level deviation amount _HL = | ΔL− (ΔF ₁ −ΔF ₂ ) / A | ... (1) Inflow deviation amount H _F1 = | ΔF ₁ − (ΔL · A + ΔF ₂ ) | …… (2) Outflow deviation amount H _F2 = | ΔF ₂ − (ΔF ₁ −ΔF ₂ ) | …… (3) calculate. Reference numeral 9 denotes a storage amount calculation unit, which is an integrated inflow amount Δ
F ₁ and enter the accumulated outflow [Delta] F ₂ storage amount _{ΔF 0 = | ΔF 1 -ΔF 2} | calculates a ... (4).

Reference numeral 10 denotes a fuzzy inference unit, which inputs the water level deviation amount _HL , the inflow deviation amount HF ₁ , the outflow deviation amount HF ₂ , the storage amount ΔF ₀ and the water level change amount ΔL, and the rule matrix of Table 1 and Based on the membership function of FIG. 2, fuzzy inference is performed with the input items as phenomenon items, and the inflow flowmeter abnormality degree FM ₁ , the water level gauge abnormality degree LM,
Output the outflow meter abnormality degree FM ₂ as the cause item.

Next, the operation of this system will be described. The amount of inflow water to the distribution reservoir 1 is detected by the inflow flow meter 2, and is input to the inflow flow rate integration unit 5, and the integrated amount of inflow water per unit time Δ
Calculate F ₁ . Further, the water level of the distribution reservoir 1 is detected by the distribution reservoir water level meter 3, and is input to the water level variation calculation unit 6 to calculate the variation ΔL of the water level per unit time. Further, the amount of outflow water from the distribution reservoir 1 is detected by the outflow flow meter 4, and is input to the outflow flow rate integration unit 7 to be an integrated amount ΔF of outflow water per unit time.
Calculate ₂ .

The integrated inflow amount ΔF ₁ and the water level change amount Δ
L, the cumulative outflow amount ΔF ₂ and the reservoir cross-sectional area A are input to the water level / flow rate deviation calculation unit 8, and the water level deviation amount _HL , the inflow deviation amount HF ₁ and the outflow deviation amount HF ₂ are respectively expressed by the equation (1), ( 2)
It is calculated by the equation (3). At the same time, the accumulated inflow amount Δ
F ₁ and the cumulative outflow amount ΔF ₂ are input to the storage amount calculation unit 9,
The storage amount ΔF ₀ is calculated by the equation (4).

In each of the above equations, when there is no error in each sensor such as the water level meter 3, the flow rate meters 2 and 4, and the true water level and flow rate are detected, the water level deviation amount H _L , The values of the inflow deviation amount HF ₁ and the outflow deviation amount HF ₂ should be zero.

However, if there is a slight abnormality or error in each sensor, the water level deviation amount H _L and the inflow error amount H
F ₁ and outflow deviation amount HF ₂ are generated.

The water level deviation amount H _L , the inflow deviation amount HF ₁ ,
The outflow deviation amount HF ₂ , the storage amount ΔF _0, and the water level change amount ΔL are input to the fuzzy inference unit 10.

In the fuzzy inference unit 10, the water level deviation amount H _L , the inflow deviation amount HF ₁ , the outflow deviation amount HF ₂ , and the storage amount Δ.
F ₀ and the water level change amount ΔL are the phenomenon items, and the inflow flowmeter abnormality degree FM ₁ , the water level gauge abnormality degree LM, and the outflow flowmeter abnormality degree FM ₂ are the cause items, and membership functions corresponding to the respective items in FIG. Fuzzy inference is performed according to the following control rules based on the rule matrix of Table 1.

[0020]

[Table 1]

IF HL = B and ΔF ₀ = S THEN LM = B IF HF ₁ = B and ΔL = S THEN FM ₁ = B IF HF ₂ = B and ΔL = S THEN FM ₂ = B IF HL = S and HF ₁ = S and HF ₂ = S THEN FM ₁ = S and LM = S and FM ₂ = S IF HL = M and HF ₁ = M THEN FM ₂ = M IF HL ₁ = M and HF ₂ = M THEN LM = M IF HL = M and HF ₂ = M THEN FM ₁ = M IF ΔF ₀ = B and ΔL = B THEN LM = S According to the inference result, the inflow flowmeter abnormality degree FM ₁ ,
Check the abnormal level LM of the water level meter and the abnormal level FM ₂ of the outflow flow meter to perform preventive maintenance.

[0022]

According to the present invention, in a system monitored and controlled by a plurality of sensors, the integrated amount or change amount of each sensor output is estimated from the integrated amount or change amount of a sensor other than the sensor, and its deviation is estimated. Since it is a mutual diagnostic abnormality detection method based on related multiple sensor information, that is, the degree of abnormality of the sensor is diagnosed by performing fuzzy inference based on (1) It becomes possible to detect the sensor abnormality at an early stage within the unified signal range. Preventive maintenance will be facilitated.

(2) Abnormality detection of all related sensors is possible at the same time.

(3) This is a method of detecting an abnormality in which the organic relationship between the sensors is taken into consideration to some extent, and the burden is reduced by supporting the judgment of the operator who performs operation management.

(4) Since fuzzy inference is used, a flexible algorithm can be constructed and rules can be easily changed or modified.

It has excellent effects such as

[Brief description of drawings]

FIG. 1 is a configuration diagram of an embodiment of a sensor mutual diagnosis system.

[Figure 2] Membership function

[Explanation of symbols] 1: Distribution reservoir 2: Inflow flow meter 3: Distribution reservoir water level meter 4: Outflow flow meter 5: Inflow flow rate integration section 6: Water level change calculation section 7: Outflow flow rate integration section 8: Water level / flow rate deviation Calculation unit 9: Storage amount calculation unit 10: Fuzzy inference unit

Claims (1)

[Claims] 1. A method of diagnosing a sensor for detecting an abnormality of a plurality of sensors used for monitoring or controlling a plant or a process, comprising: an integrated amount or change amount of sensor output per unit time; and an integrated amount of each sensor output. Alternatively, fuzzy inference is performed by using the amount of change as the phenomenon item with each deviation amount calculated from the integrated amount or the amount of change of the output of multiple other sensors related to the sensor output other than that sensor as the cause item. A sensor mutual diagnosis method characterized by detecting an abnormality of each sensor.