JP2000298512A - Control of operation of plant by near infrared analysis - Google Patents

Abstract

PROBLEM TO BE SOLVED: To omit useless interpolating work in the case of controlling the operation of a plant on the basis of values measured by a near infrared analytical method, to execute routine analysis while efficiently interpolating/evaluating an analytical curve when the interpolation of the analytical curve is required and to control the plant on the basis of the measured values. SOLUTION: In the method for controlling the operation of the plant on the basis of measured values obtained by the near infrared analysis of the plant, the near infrared analysis of the sample is executed on the basis of a previously prepared analytical curve, and when a measured value other than an allowable range is obtained, a measured value obtained by a general analytical method is compared with the allowable range. When the measured value by the general analytical method is included within the allowable range, past near infrared analytical data are inputted to obtain a predicted value, and when the predicted value exceeds the allowable range, the analytical curve is interpolated and the operation of the plant is controlled by near infrared analysis.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for controlling plant operation by near-infrared analysis (near-infrared spectroscopy), and more particularly to a method for controlling plant operation while evaluating and complementing a calibration curve.

[0002]

2. Description of the Related Art In the field of the chemical industry, when controlling the production of chemical products, raw materials, solvents, moisture, intermediate products, products, by-products, etc. are analyzed by near-infrared analysis, and production is performed based on the measured values. It has been proposed to control the operation. In this method, the operation of the plant is controlled so that the measured value does not exceed a certain standard value so that the obtained product maintains a certain quality. In the near-infrared analysis, a near-infrared spectrum of a specific region is measured, and a target component value is calculated based on a calibration curve created in advance from a combination of absorbances of a specific wavelength included in the spectrum, and the measured value ( Predicted value).

[0003] In plant operation control based on measured values of near-infrared analysis, a control value having a certain width is usually provided inside the above-mentioned standard value, and when the measured value obtained by near-infrared analysis deviates from the control value, the plant is controlled. Is changed, and the control device issues a control signal to control the plant so that the measured value returns to the control value, thereby preventing the measured value from exceeding the specified value. In this case, in the near-infrared analysis, a sample is intermittently or continuously sampled and analyzed, and a control action such as a condition change is taken when an abnormal value outside the control value is obtained.

The near-infrared spectrum contains information on a plurality of components, and a plurality of factors such as other components, water, concentration, temperature, and particle size contained in the sample are complexly combined to form a spectrum. When these factors change, the position and height of the peak change. Therefore, calibration is performed by a statistical method such as a linear multiple regression (MLR) or a partial least squares (PLS). A line has been created.

[0005] The above-mentioned calibration curve is prepared by analyzing a plurality of samples by both a general analysis method such as chromatography and a near-infrared analysis, and determining each target component from a spectrum obtained by the near-infrared analysis. The data of the absorbances of a plurality of peaks to be obtained are selected, and the calibration is performed by the above-described statistical method using these data and the measured values of the general analysis method, and a calibration curve is created. In this case, a separate calibration curve is created by performing the same operation for each component.

However, since the number of samples used for calibration is limited in the calibration curve created in this way, there are cases where the selection of explanatory variables or objective variables in the calibration process is not appropriate. For some reasons, general formulas are not always obtained. For this reason, the created calibration curve is evaluated and confirmed, and a more general calibration curve is created and used for routine analysis.

Nevertheless, due to the complexity of the near-infrared spectrum, with changes in the conditions affecting the seasons, the environment, and other manufacturing conditions, the measured values obtained by the calibration curve already prepared are measured by the general analysis method. In some cases, the values may not match, and the calibration curve needs to be complemented. However, complementation of the calibration curve requires a considerable amount of time and effort, and therefore, it is necessary to eliminate useless complementation work and to carry out complementation efficiently when necessary.

[0008]

SUMMARY OF THE INVENTION It is an object of the present invention to provide a method for controlling a production operation based on a measurement value obtained by a near-infrared analysis method. An object of the present invention is to propose a plant operation control method based on near-infrared analysis that can perform routine analysis while efficiently complementing a calibration curve at a point in time and control the plant based on the measured values.

[0009]

SUMMARY OF THE INVENTION The present invention is a plant operation control method based on the following near-infrared analysis. (1) A method for controlling the operation of a plant based on measured values obtained by analyzing a sample by near-infrared analysis, wherein the sample is subjected to near-infrared analysis based on a calibration curve created in advance, and a near-infrared analysis method is provided. The measured value of the general analysis method is compared with the allowable value.If the measured value outside the allowable value is obtained, the analysis is performed by the general analysis method. If the predicted value is out of the allowable value, check the near-infrared analyzer if the predicted value is outside the allowable value. A plant operation control method using near-infrared analysis, which performs complementation and evaluation of a calibration curve. (2) The method according to (1) above, wherein the plant inspection is performed when the measured value of the general analysis method is out of the allowable value. (3) When a measured value outside the permissible value is obtained by the near-infrared analysis of the sample, the near-infrared analysis is performed again. The method according to the above (1) or (2), wherein the measured value is compared with an allowable value. (4) The method according to any one of (1) to (3), wherein after the calibration curve is complemented, the calibration curve is evaluated and confirmed by comparing the measured value at the time of occurrence of an abnormality. (5) The method according to any one of the above (1) to (4), wherein after the calibration curve is complemented, the past near-infrared analysis data is input and the calibration curve is evaluated and confirmed. (6) One of the above (1) to (5), in which the value measured by the general analysis method is periodically compared with the value measured by the near-infrared analysis method, and when a difference is found, the calibration curve is complemented / evaluated. The method described in. (7) Regularly compare the measured value of the general analysis method with the measured value of the near-infrared analysis method, qualitatively compare and confirm the product pattern when there is a difference, and if there is no difference, a calibration curve The method according to the above (6), wherein the confirmation is made. (8) When the measured value within the allowable value is obtained by the near-infrared analysis of the sample, the measured value is compared with the control value. The method according to any one of (1) to (7), wherein no control action is performed in the case of (1).

The near-infrared analysis generally has a wavelength of 400 to 2500.
nm, preferably 800 to 2500 nm, more preferably 1000 to 2000 nm, the sample is irradiated with near-infrared rays to detect transmitted light or reflected light, and a calibration curve prepared in advance from the absorption spectrum is used to determine the properties, components, etc. of the sample. It is a method of performing analysis. In the control method of the present invention, the sample collected in the manufacturing process is analyzed as it is without pre-processing the sample, and control is performed based on the measured value. The control is performed using a control device such as a computer so that the measured value by the near-infrared analysis falls within the range of the standard value.

The near-infrared spectrometer used in this method is
The noise level is 30 × 10 ⁻⁶ Abs or less, preferably 2 × 10 ⁻⁶ Abs.
A high-precision material having a wavelength of 0 × 10 ⁻⁶ Abs or less and a wavelength reproducibility of ± 0.3 nm or less, preferably ± 0.01 nm or less can be used. The measuring method of the noise level and the wavelength reproducibility is as follows.

Measurement method of noise level If the measurement method is a reflection type, the ceramic plate is measured twice, and if it is a transmission type, the measurement is made twice in the air.
Measure 0 sets. The standard deviation of the difference (effective value) between the first measurement value and the second measurement value is defined as the noise level.

Measurement method of wavelength reproducibility A standard polystyrene film is put in the optical path and measured according to JIS K0117-1979 rules of infrared spectroscopy.
At this time, the reference near-infrared absorption wavelength is 1143.6330 n.
m, 1684.2700 nm, 2166.4000 nm
And 2305.9300 nm. Ten standard deviations are the values of the wavelength reproducibility.

The objects to be measured by the near-infrared analysis include everything used for controlling the plant, such as raw materials, solvents, moisture, intermediate products, products, and by-products used in the production process. Near-infrared rays have less energy than ultraviolet rays and do not change sample components. Also, unlike the case of visible light, the analysis is based on the absorption spectrum, and is not affected by the transparency or other form of the sample, so that adjustment of the film thickness or the like is not required.

A near-infrared spectrum is obtained by subjecting such a sample to spectroscopic analysis using a near-infrared analyzer.
This spectrum contains a plurality of pieces of information as described above, and a measured value (predicted value) of a target measured component is calculated from a combination of data of specific peaks using a calibration curve created in advance. In this case, measurement values of a plurality of measurement components can be obtained from the same spectrum.

The calibration equation is a mathematical relationship between the spectrum data and the analysis value of the measured component. As described above, the analysis of a plurality of samples by a general analysis method such as chromatography and the near infrared Analyze both methods of analysis, select the data of the absorbance of multiple peaks determined for each target component from the spectrum obtained in the near-infrared analysis, using these data and the measured values of the general analysis method described above It is created by a statistical method such as the MLR method or the PLS method.

In this case, a separate calibration curve can be prepared by performing the same operation for each measurement component. The statistical method used for such calibration is arbitrarily determined depending on the measurement components, the accuracy thereof, and the like.
The LS method is preferred. In addition, colorimetric analysis,
A conventionally used method for analyzing a measurement component such as gas chromatography is included.

The method for controlling operation of a plant according to the present invention efficiently analyzes a measurement component used for control by near-infrared analysis in a routine analysis while manufacturing a target product, and efficiently detects abnormal values. The calibration curve is complemented / evaluated, measurement is performed using the complemented calibration curve, and operation control is performed based on the measured value. In the present invention, the object to be controlled is a production operation of a chemical product, a food product, or the like that can be measured by near-infrared analysis, but is suitable for a production operation of a chemical product, particularly, a polyester or a phenol.

In the present invention, the standard value is a value required in terms of product quality, and a value outside the standard value is rejected. In the present invention, the management value is a value at which control is started and is set inside the specification value so as not to exceed the standard value. Perform control actions. In the present invention, the permissible value is a value at which the calibration curve is permissible. If the permissible value is outside the permissible value, the value is set outside the control value in order to complement the calibration curve. A value in the vicinity is preferable, but another value may be used. Assuming that the center value (average value) of the standard values is the target value m, it is preferable that the management value is set to m ± σ (σ is a standard deviation) and the allowable value is set to m ± 3σ.

In the present invention, in a routine analysis, a sample is subjected to near-infrared analysis based on a calibration curve prepared in advance, and when a measured value outside a permissible value is obtained, analysis is performed by a general analysis method to obtain a measured value. If the measured value of the general analysis method is out of the permissible value as compared with the permissible value, the plant inspection is performed, and the equipment can be adjusted or the setting can be changed. If the measured value of the general analysis method is within the allowable value, the data of the past near-infrared analysis is input to obtain the predicted value, and if the predicted value is out of the allowable value, the near-infrared analyzer is inspected. When the predicted value is within the allowable value, the calibration curve is supplemented. The complementation of the calibration curve is performed by adding the abnormal data to the data for preparing the calibration curve, and again preparing and evaluating the calibration curve by a statistical method such as the MLR method or the PLS method.

In the above case, when a measured value outside the permissible value is obtained by the near infrared analysis of the sample, the near infrared analysis is performed again,
When a measured value outside the permissible value is obtained by the re-analysis, the measurement value obtained by the general analysis method is compared with the permissible value, thereby avoiding the generation of a useless calibration curve due to a sudden data abnormality.

In the routine analysis, when the measured value of the near-infrared analysis is within the allowable value, the measured value is compared with the control value. If no control action is taken.

After the calibration curve is complemented, an accurate complementation calibration curve can be obtained by evaluating and confirming the calibration curve in comparison with the measured value of the near-infrared analysis method when an abnormality occurs. In addition, after the calibration curve is complemented, inappropriate complementation can be avoided by inputting past near-infrared analysis data and evaluating and confirming the calibration curve.

In addition to the calibration curve by the routine analysis, the measured value of the general analysis method is periodically compared with the measured value of the near-infrared analysis method, and when a difference is found, the calibration curve is complemented. By doing so, it is possible to quickly recognize the situation where the calibration curve becomes inappropriate, to complement the calibration curve in response to this, and to perform the control properly. In this case, the measured value of the general analysis method is regularly compared with the measured value of the near-infrared analysis method, and when there is a difference, the product pattern is qualitatively compared and confirmed. It is possible to avoid creating a calibration curve due to a sudden abnormality by confirming the above.

[0025]

According to the present invention, when an abnormal value appears in a measured value obtained by near-infrared analysis, the measured value obtained by the general analysis method is compared with an allowable value set outside the control value, and the result is determined. Is used to supplement the calibration curve, so when performing plant operation control based on the measured values obtained by near-infrared analysis, wasteful supplementary work is omitted, and the efficiency is improved when the calibration curve needs to be supplemented. Routine analysis can be performed while supplementing the calibration curve well, and the plant can be controlled based on the measured values.

Furthermore, the measured value obtained by the general analysis method is periodically compared with the measured value obtained by the near-infrared analysis method, and when a difference is found, the calibration curve is complemented, thereby making the calibration curve inappropriate. Is quickly recognized, and the calibration curve is complemented accordingly,
Control can be performed properly.

[0027]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a flow chart showing the manufacturing apparatus of the embodiment, FIG. 2 is a flow chart showing a control method, and FIG. 3 is a distribution chart showing a relationship between respective values.

In FIG. 1, reference numeral 1 denotes a plant, which is configured to supply a raw material 2 to produce a product 3. 4 is a control device, 5 is a near-infrared analyzer, and 6 is a general analyzer. The control device 4 detects the detection signal 1 from the plant 1
a, the control signal 4a is sent to the plant to control the reaction conditions and the like. The near-infrared analyzer 5 is configured to perform near-infrared analysis based on a control signal 4 b from the controller 4 and send a measurement signal 5 a to the controller 4. The general analyzer 6 is constituted by chromatography or the like, performs analysis by a control signal 4c from the controller 4, and sends a measurement signal 6a to the controller 4.

In the manufacturing operation method using the above apparatus, the raw material 2 is supplied to the plant 1 by the control signal 4a of the control device 4 to manufacture the product 3. During this time, the detection signal 1
a is sent to the control device 4. At a certain interval, the near-infrared analyzer 5 samples the plant 1 to perform near-infrared analysis based on the control signal 4 b from the controller 4, and sends a measurement signal 5 a to the controller 4. The control device 4 calculates a measured value from a calibration curve created in advance, and sends a control signal 4a to the plant 1 based on the result to control the manufacturing conditions of the plant 1. Also, the general analyzer 6 is controlled by the control signal 4c.
And sends the result to the control device 4. The controller 4 complements the calibration curve based on the result.

The above control method will be described in detail with reference to FIG. 2, and the relationship between the values is shown in FIG. That is, a is a standard value, and assuming that the center value (average value) is the target value m, the allowable value b is set at the position of m ± 3σ, and the management value c is set at the position of m ± σ. The allowable value b and the management value c may be set at a position d of m ± σ or another position,
The allowable value b is set inside the standard value a, and the management value c is set inside the allowable value b.

In FIG. 2, first, the near-infrared analyzer is inspected in S101, and the near-infrared analyzer is diagnosed in S102. In the diagnosis, the noise level and the wavelength reproducibility are measured, and if the wavelength reproducibility is within ± 0.3 nm and the noise level is 30 × 10 ⁻⁶ abs or less, it is determined to be normal. S
At 103, the diagnosis result is determined.
The sample of No. 4 is measured, and if it is not normal, the process returns to S101 and the inspection and diagnosis of the device are repeated. The measurement of the sample in S104 compares the measurement value of the sample with the near-infrared analysis and the general analysis method, and if the measurement value of the near-infrared analysis method is within ± 3σ of the measurement value of the general analysis method, the test is passed. I do. In S105, pass / fail is determined, and if it passes, the process proceeds to the routine analysis in S106 and the step of periodically comparing with the general method in S131. If not, the steps from S101 are repeated.

After confirming that the near-infrared analyzer is normal as described above, the routine proceeds to the routine analysis of S106.
In this step, the near-infrared spectrum is measured by the near-infrared analyzer, and the result is input to the controller, where the ML is set in advance.
A measured value (predicted value) of a measured component is calculated from a calibration curve created by a statistical method such as the R method and the PLS method, and this is calculated by S
107 is output. Then, in S108, it is determined whether or not the measured value is within an allowable value b having a constant width of m ± 3σ. The permissible value b is a value within the permissible range of the calibration curve defined for each product quality. If the permissible value b is within the permissible value, it is compared with the control value c as in S109. No control action is performed as in (1), but no control action is performed, but if the control value is outside the control value c, a control action is performed as in step S112 to return the control value to the control value c. The plant control action changes the reaction conditions such as raw material supply amount, concentration, reaction temperature, pressure, time, etc.
Control is performed so that the measured value returns to the control value. When such control action is performed and the measured value falls within the control value in the next measurement, the changed reaction condition is continued as it is, and when the next measured value is out of the control value, further control action is performed. Is performed.

If it is determined in step S108 that the measured value is outside the allowable range, the process proceeds to step S115, where the measurement is performed again by the near-infrared analyzer. This is performed to confirm that the abnormal value is not caused by a sudden malfunction or the like but is continuous, and to omit useless work of supplementing the calibration curve. It is determined in S116 whether the result of the measurement in S115 is within the allowable value, and if it is within the allowable value, the operation proceeds to S109 and subsequent steps.

If it is determined in S116 that the re-measurement result is outside the allowable range, measurement is performed in S117 by a general analysis method such as gas chromatography, and the measured value is stored in S11.
In step 8, it is determined whether the value is within the control value. Here, if it is determined that the value is outside the control value, it means that the occurrence of the abnormality has been confirmed twice by the near-infrared analysis method and the general analysis method, and the plant is inspected as shown in S119. This is because when the results of both the near-infrared analysis and the general analysis method are out of the allowable values, it is necessary to inspect and prepare and set the plant.

If it is determined in step S118 that the value is within the allowable range,
In S120, the data of the past near-infrared analysis is input to determine whether or not the near-infrared analyzer is normal. Since the past data has already been confirmed to be within the allowable value, if a judgment outside the allowable value is made in S121, there is a possibility that the near-infrared analyzer malfunctions, and S101 and S1
Returning to step 02, the device is inspected and diagnosed, and the failed part is repaired. By such an operation, useless calibration curve complement / evaluation work is omitted.

If it is determined in step S121 that the measured value by the general method and the past data are within the allowable value, the measured value of the near-infrared analysis is out of the allowable value. Therefore, the calibration curve is complemented in S122. The calibration curve is supplemented by the same technique as when the calibration curve was first created, for example, a statistical technique such as the MLR method or the PLS method. The first calibration curve is created by multivariate analysis using the measured values of the near-infrared analysis method and the measured values of the general analysis method for multiple samples as explanatory variables.
The calibration curve is complemented by adding the measured values of the sample with abnormal values in this study, which were analyzed by near-infrared analysis and the general method, as new explanatory variables, in addition to the measured values of the sample used when the first calibration curve was created. Create

After creating a new calibration curve in this way, S1
At 23, the new calibration curve is evaluated and confirmed. This rating,
The confirmation is performed in addition to the operation performed when the normal calibration curve is created, and whether the measured value analyzed at the time of occurrence of the abnormality is not used for complementing the calibration curve is used to determine whether the calculated value is a valid value in S124. Is determined. Here, S1 if rejected
Returning to step 22, the calibration curve is complemented again. If the result is S124, the past data is input, evaluated and confirmed in S125. Since the past data has already been determined to be within the control value, if the same result is obtained even with the new calibration curve, it is determined to be acceptable in S126, and the complementary calibration curve is used as a new calibration curve in S127, and the subsequent routine analysis is performed. Is performed with a new calibration curve. If not, the process returns to S122 and the calibration curve is complemented again.

If a calibration curve is supplemented as a measure against the abnormality when an abnormality occurs in the routine analysis as described above, the measurement by the general analysis method need not always be performed, and the response to the abnormality can be quickly performed. It can be carried out. Even if the measured value of the near-infrared analysis is within the allowable value, the measured value by the general analysis method may deviate from the allowable value.In such a case, by setting the allowable value to an appropriate width, the near value can be obtained. Since the measured value of the infrared method also falls outside the allowable value, an abnormality can be easily found.

In order to prevent a situation in which the measured value of the general method deviates from the allowable value even though the measured value of the near-infrared method is within the allowable value, the general analysis method of S131 and subsequent steps is performed. It is preferable to make a comparison periodically with. This operation is performed in parallel with the routine analysis of S106 and subsequent steps.
A general analysis is performed about twice, and in S131, the measured value is compared with the measured value of the near-infrared analysis method to determine whether there is a difference equal to or larger than an allowable value.

If it is determined in S132 that there is no difference larger than the allowable value, the routine analysis from S106 is continued. When it is determined that there is a difference equal to or larger than the allowable value, S133
Perform qualitative comparison and confirmation of products. This is done by inputting the data of the corresponding product from the database and comparing it periodically, qualitatively determining whether the obtained measured value has the pattern of the product, and determining whether it is sudden abnormal data. judge.

If it is determined in step S134 that the standard product is equivalent to the conventional product, it indicates that the calibration curve is inappropriate, and the operation for calibration curve complementing in step S122 and thereafter is started. Also S
If it is determined in step 134 that the product is not equivalent to the conventional product, the operation proceeds to step S115 and subsequent steps.

As described above, while the calibration curve is complemented by the routine analysis of step S106 and subsequent steps, and the comparison with the periodic general method of step S131 and subsequent steps is performed, the situation where the calibration curve becomes inappropriate can be quickly detected. , Thereby increasing the accuracy of the control. In addition, since the analysis by the general method requires a long time, it is difficult to frequently perform the operations of S131 and below. However, by performing these operations in combination, the frequency of the analysis by the general method is reduced and the efficiency is reduced. The calibration curve can be complemented.

[Brief description of the drawings]

FIG. 1 is a flowchart of a manufacturing apparatus according to an embodiment.

FIG. 2 is a flowchart illustrating a control method according to the embodiment.

FIG. 3 is a distribution diagram showing a relationship between values.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Plant 2 Raw material 3 Product 4 Controller 5 Near-infrared analyzer 6 General analyzer

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁷ Identification FI FI Theme Court ゛ (Reference) G01N 21/35 G01N 21/35 Z (72) Inventor Masami Tsuruoka 6-1, Waki, Wakimachi, Kuga-gun, Yamaguchi Prefecture No. 2 Mitsui Chemicals Co., Ltd. F-term (reference) 2G020 AA03 BA02 CA02 CB42 CD04 CD13 CD22 2G059 AA01 BB11 BB20 CC09 CC12 DD20 EE01 EE12 FF04 FF08 HH01 KK01 MM02 MM05 MM12 4H006 AA05 AC90 BD80 5H223 AE9

Claims (8)

[Claims] 1. A method for controlling the operation of a plant based on a measurement value obtained by performing near-infrared analysis on a sample, the method comprising: performing near-infrared analysis on the sample based on a calibration curve created in advance; The measured value of the analytical method is compared with the permissible value, and when the measured value outside the permissible value is obtained, analysis is performed by the general analytical method.The measured value of the general analytical method is compared with the permissible value, and the measured value of the general analytical method is measured. If the predicted value is within the allowable value, enter the past NIR data and calculate the predicted value.If the predicted value is out of the allowable value, check the near-infrared analyzer and check if the predicted value is within the allowable value. A method for controlling plant operation by near-infrared analysis, in which a calibration curve is complemented / evaluated in some cases. 2. The method according to claim 1, wherein the plant inspection is performed when the measured value of the general analysis method is out of the allowable value. 3. When the measured value out of the permissible value is obtained by the near-infrared analysis of the sample, the near-infrared analysis is performed again, and when the measured value out of the permissible value is obtained by the re-analysis, the general analysis is performed. 3. The method according to claim 1, wherein the measured value is compared with an allowable value. 4. The method according to claim 1, wherein after complementing the calibration curve, the calibration curve is evaluated and confirmed by comparing the measured value when an abnormality occurs. 5. The method according to claim 1, wherein after the calibration curve is complemented, data of past near-infrared analysis is inputted to evaluate and confirm the calibration curve. 6. The method according to claim 1, wherein a value measured by the general analysis method is periodically compared with a value measured by the near-infrared analysis method, and when a difference is found, the calibration curve is complemented / evaluated. The described method. 7. The measured value of the general analysis method is periodically compared with the measured value of the near-infrared analysis method, and when a difference is found, the pattern of the product is qualitatively compared and confirmed. The method according to claim 6, wherein the calibration curve is confirmed. 8. When a measured value within a tolerance is obtained by near-infrared analysis of a sample, the measured value is compared with a control value, and if the measured value is out of the control value, a control action is performed and 8. The method according to claim 1, wherein no control action is taken if the value is within the value.