TW202309498A - Corrosion monitoring apparatus, method, and computer program product thereof - Google Patents

Abstract

A corrosion monitoring apparatus, method, and computer program product thereof are provided. The corrosion monitoring apparatus stores a health evaluation module in terms of corrosion and a significance analysis module, wherein the two modules have been trained. The corrosion monitoring apparatus receives a current measurement data set of a system that executes an aromatic hydrocarbon process. The current measurement data set includes a plurality of current feature data corresponding to a plurality of selected feature variables of the aromatic hydrocarbon process one-to-one. The corrosion monitoring apparatus inputs the plurality of current feature data into the health evaluation module to derive a health degree of the system, wherein the health degree is in terms of corrosion. The corrosion monitoring apparatus inputs the plurality of current feature data and the health degree into the significance analysis module to derive a contribution degree of each of the selected feature variables to the health degree.

Description

Corrosion monitoring device, method and computer program product thereof

The invention relates to a corrosion monitoring device, method and computer program product thereof. Specifically, the corrosion monitoring device, method and computer program product provided by the present invention perform corrosion monitoring based on a plurality of selected characteristic variables in an aromatic hydrocarbon process.

Many processing stages in the aromatics process (for example: pre-distillation, hydrogenation, reformation, regeneration, cracking, re-extraction) generate corrosive factors that cause corrosion to the systems performing the aromatics process. The combination of different corrosion factors will result in different corrosion mechanisms. In order to understand the corrosion status of the system after the aromatic hydrocarbon process is implemented, the conventional technology detects the pH value and total iron content of the water at the lowest point of the system for judgment. This detection method does not take into account the characteristic variables involved in each treatment stage in the aromatic hydrocarbon production process (such as: operating conditions, feed), so it is difficult to grasp the characteristic variables that cause system corrosion from the source, and it is impossible to take preventive measures in advance to avoid corrosion conditions deterioration.

In view of this, there is an urgent need in the field for a technology that can monitor corrosion based on important characteristic variables in the aromatic hydrocarbon process and find out the key characteristic variables that cause corrosion, so as to slow down the corrosion rate and prevent corrosion deterioration.

An object of the present invention is to provide a corrosion monitoring device. The corrosion monitoring device includes a storage, a transceiver interface and a processor, wherein the processor is electrically connected to the storage and the transceiver interface. The storage stores a trained corrosion health assessment module and a trained importance analysis module. The transceiver interface receives a current measurement data set of a system executing an aromatic hydrocarbon process, wherein the current measurement data set includes a plurality of current characteristic data, and the current characteristic data correspond to the aromatic hydrocarbon process one-to-one A plurality of selected feature variables of . The processor inputs the current characteristic data into the corrosion health evaluation module to obtain a corrosion health degree of the system, and inputs the current characteristic data and the corrosion health degree into the importance analysis module to obtain the respective A degree of contribution of selected characteristic variables to the corrosion health.

Another object of the present invention is to provide a corrosion monitoring method applicable to an electronic computing device. The corrosion monitoring method includes the following steps: (a) receiving a current measurement data set of a system performing an aromatic hydrocarbon process, wherein the current measurement data set includes a plurality of current characteristic data, and the current characteristic data is a pair a plurality of selected characteristic variables corresponding to the aromatic hydrocarbon process, (b) inputting the current characteristic data into a trained corrosion health assessment module to obtain a corrosion health of the system, and (c) applying The current characteristic data and the corrosion health are input into a trained importance analysis module to obtain a contribution degree of each selected characteristic variable to the corrosion health.

Another object of the present invention is to provide a computer program product, which includes a plurality of program instructions. After an electronic computing device is loaded with the computer program product, the electronic computing device executes the program instructions contained in the computer program product to realize the corrosion monitoring method mentioned in the preceding paragraph.

The corrosion monitoring technology provided by the present invention (including at least the device, the method and its computer program product) utilizes a trained corrosion health evaluation module and a trained importance analysis module to monitor an aromatic hydrocarbon process A corrosion health of the system (ie, the health of the system in terms of corrosion conditions). Specifically, some characteristic variables in the aromatic hydrocarbon production process (that is, the aforementioned selected characteristic variables) have a more obvious impact on whether the equipment in the system is corroded, so the corrosion monitoring technology provided by the present invention implements a system A plurality of pieces of current characteristic data corresponding to these selected characteristic variables in the process of aromatic hydrocarbons are input into the corrosion health degree evaluation module to obtain a corrosion health degree of the system, and the current characteristic data and the corrosion health degree are input into the importance The module is analyzed to obtain a contribution degree of each of the selected characteristic variables to the corrosion health. When the corrosion health degree is lower than a first threshold value, the corrosion monitoring technology provided by the present invention can also output those selected characteristic variables whose contribution degree is higher than a second threshold value as at least one key characteristic variable (that is, , the characteristic variable or variables most critical to the corrosion status of the system). Since the corrosion monitoring technology provided by the present invention considers the selected characteristic variables that have a significant impact on corrosion in the aromatic hydrocarbon process, it can accurately evaluate the corrosion health of the system, and then generate an early warning before the corrosion occurs in the system or before the corrosion degree deteriorates. Allows the user to take precautionary measures against key characteristic variables (e.g. adjust temperature, adjust feed) to prevent corrosion from occurring or worsening.

The detailed techniques and implementation methods of the present invention are described below in conjunction with the drawings, so that those with ordinary knowledge in the technical field of the present invention can understand the technical characteristics of the claimed invention.

The following will explain the corrosion monitoring device, method and computer program product provided by the present invention through the embodiments. However, these embodiments are not intended to limit the present invention to be implemented in any environment, application or manner as described in these embodiments. Therefore, the description of the following embodiments is only for the purpose of illustrating the present invention, rather than limiting the scope of the present invention. It should be understood that in the following embodiments and drawings, elements not directly related to the present invention have been omitted and not shown. In addition, the size of each element and the dimensional ratio among the elements in the drawings are only for illustration and description, and are not intended to limit the scope of the present invention.

The first embodiment of the present invention is a corrosion monitoring device 1 , the schematic diagram of which is depicted in FIG. 1A . The corrosion monitoring device 1 includes a storage 11 , a transceiver interface 13 and a processor 15 , wherein the processor 15 is electrically connected to the storage 11 and the transceiver interface 13 . The storage 11 can be a memory, a hard disk drive (Hard Disk Drive; HDD) or any other non-transitory storage medium, circuit or device with the same function known to those skilled in the art of the present invention. The transceiver interface 13 can be a wired transmission interface or a wireless transmission interface known to those skilled in the art of the present invention, and is used for sending and receiving signals and data. The processor 15 can be various processors, a central processing unit (Central Processing Unit; CPU), a microprocessor (Microprocessor Unit; MPU), a digital signal processor (Digital Signal Processor; DSP) or a general processor in the technical field of the present invention. Other computing devices known to those in the know.

In this embodiment, the storage 11 stores a trained corrosion health assessment module M1 and a trained importance analysis module M2. The corrosion monitoring device 1 utilizes the corrosion health evaluation module M1 to determine a corrosion health of a system (not shown) performing an aromatic hydrocarbon process (that is, the health of the system in terms of corrosion status), and utilizes The importance analysis module M2 provides information related to the corrosion health. The operating principle and details of the corrosion monitoring device 1 will be described in detail below.

An aromatic hydrocarbon process has a plurality of process characteristic variables (not shown), wherein each of the process characteristic variables can be any factor related to the aromatic hydrocarbon process, such as: feed, compound concentration, operating settings, environmental factors, quality indicators etc. A subset of these process characteristic variables have a significant impact on whether equipment in a system performing the aromatic hydrocarbon process will corrode. For ease of description, the process characteristic variables in the subset may be referred to as the first selected characteristic variables. This embodiment does not limit the method of determining the first selected characteristic variables of an aromatic hydrocarbon process, which can be determined by a user (for example: the manager of the system, a professional familiar with the aromatic hydrocarbon process), or by a statistical Analytical decision.

FIG. 1B depicts a schematic diagram of a debutanizer system performing an aromatics process, but it should be understood that it is not intended to limit the scope of the present invention. In the aromatic hydrocarbon production process shown in Figure 1B, the core light oil first undergoes hydrogenation to remove impurities such as sulfur and nitrogen, and then undergoes a restructuring reaction to convert linear hydrocarbons and naphthenes into aromatic hydrocarbons, and then undergoes dechlorination Tank treatment before entering the butane removal tower. When the recombinant catalyst is regenerated, dichloroethane needs to be added to adjust the acidic catalytic function, so the feed to the butanizer contains trace amounts of hydrogen chloride, ammonium chloride and water. After hydrogen chloride, ammonium chloride and water enter the butanizer, they will concentrate and accumulate at the top. Since hydrogen chloride will dissociate in the presence of water, the system will be acidic and uniformly corroded, so it is the main corrosion factor of the de-butanizer. As for ammonium chloride, because it is easily soluble in water, it is not the main corrosion factor of the debutanizer. In the aromatic hydrocarbon process shown in Figure 1B, each of the first selected characteristic variables (that is, those process characteristic variables that have a significant impact on whether the equipment in the system performing the aromatic hydrocarbon process will corrode) can be It is related to one of a hydrogenation reaction, a recombination reaction, a catalyst regeneration reaction and a dechlorination operation in the aromatic hydrocarbon production process. For example, for the aromatics process shown in Figure 1B, the first selected characteristic variables may include total chlorine concentration in sweet light oil after hydrogenation reaction, hydrogen chloride concentration in circulating hydrogen gas during recombination reaction, trigger The amount of dichloroethane injected during the catalyst regeneration reaction, the chlorine concentration of the catalyst before regeneration, the chlorine concentration of the catalyst after regeneration, the total chlorine concentration before the dechlorination tank and the total chlorine concentration after the dechlorination tank.

In this embodiment, the transceiver interface 13 of the corrosion monitoring device 1 will periodically (for example: every several hours) or non-periodically (for example: when it is necessary to know the corrosion health of the system and related information) receive and execute the fragrance A current measurement data set DS1 of the system for hydrocarbon processing. The current measurement data set DS1 includes a plurality of pieces of current characteristic data, and the current characteristic data correspond to the first selected characteristic variables of the aromatic hydrocarbon process in a one-to-one manner. It should be noted that the present invention does not limit how the transceiver interface 13 obtains the current measurement data set DS1. For example, a plurality of information table points (for example: measuring device, sensor) can be configured in the system to measure the value of each characteristic variable of the process when the system executes the aromatic hydrocarbon process, and the sending and receiving interface 13 can be directly obtained from The table points corresponding to the first selected characteristic variables receive the current characteristic data. For another example, multiple information table points (such as measuring devices and sensors) can be configured in the system to measure the values of the process characteristic variables when the system executes the aromatic hydrocarbon process, and these values will be collected into A data collector (for example: a server), the transceiver interface 13 receives the values corresponding to the first selected characteristic variables (ie, the current characteristic data) from the data collector.

The processor 15 then inputs the current characteristic data in the current measurement data set DS1 into the corrosion health assessment module M1 to obtain a corrosion health HI of the system. In this embodiment, the higher the value of the corrosion health HI, the healthier the system is, that is, the lower the degree of corrosion or the less likely it is to cause corrosion. In addition, the processor 15 will input the corrosion health HI and the current characteristic data in the current measurement data set DS1 into the importance analysis module M2 to obtain a contribution degree of each of the first selected characteristic variables to the corrosion health HI .

In some embodiments, the corrosion monitoring device 1 can also output the corrosion health HI, the contribution levels O1, ..., Om of the first selected characteristic variables, or/and the current value of the first selected characteristic variables (that is, the current characteristic data) to let the user know the current corrosion status of the system. For example, the corrosion monitoring device 1 may include a display screen to display the corrosion health HI, the contribution levels O1, ..., Om of the first selected characteristic variables or/and the current values of the first selected characteristic variables value. For another example, the corrosion monitoring device 1 can output the corrosion health level HI, the contribution levels O1, ..., Om of the first selected characteristic variables through the transceiver interface 13 or another transceiver interface, or/and the first The current value of the selected feature variable.

In some implementations, the processor 15 also compares the corrosion health HI generated by the corrosion health assessment module M1 with a first threshold (not shown). When the processor 15 judges that the corrosion health HI is lower than the first threshold value, it represents one or more current values of the first selected characteristic variables (that is, one or more of the current characteristic data) Will corrode (or worsen corrosion of) the equipment of the system. When the corrosion health degree HI is lower than the first threshold value, the corrosion monitoring device 1 can output at least one key characteristic variable (not shown) among the first selected characteristic variables through the transceiver interface 13 or another transceiver interface, wherein The contribution degree corresponding to each of the at least one key characteristic variable is greater than a second threshold value (for example: the contribution degree is higher than 30%). In other words, the at least one key characteristic variable is the one among the first selected characteristic variables that causes corrosion of equipment of the system (or worsens corrosion of equipment of the system). By outputting the at least one key characteristic variable, the user is allowed to take corresponding measures.

In some embodiments, the corrosion monitoring device 1 can be used as a back-end electronic device, which can be used in conjunction with a front-end electronic device (eg, an electronic computer device that can be operated by a user). In these embodiments, the corrosion monitoring device 1 can transmit the corrosion health level HI, the contribution levels O1, ..., Om of the first selected characteristic variables, or/and these through the transceiver interface 13 or another transceiver interface The value of the first selected characteristic variable is output to the front-end electronic device. The front-end electronic equipment then compares the corrosion health HI with the first threshold. When the corrosion health degree HI is lower than the first threshold value, the front-end electronic device will also output (for example: display) at least one key characteristic variable among the first selected characteristic variables, wherein each of the at least one key characteristic variable is The corresponding contribution degree is greater than a second threshold.

To sum up, the corrosion monitoring device 1 monitors the corrosion health of a system implementing the aromatic hydrocarbon process based on the first selected characteristic variable that has a significant impact on corrosion in the aromatic hydrocarbon process. Therefore, the corrosion monitoring device 1 can accurately evaluate the corrosion health of the system, and then generate an early warning before the corrosion occurs in the system or before the corrosion degree deteriorates. Furthermore, the corrosion monitoring device 1 can determine which key characteristic variables will cause corrosion of the equipment of the system (or worsen the corrosion of the equipment of the system), and then allow the user to take preventive measures to avoid corrosion or worsen the corrosion condition.

The second embodiment of the present invention is a corrosion monitoring device 2 , the schematic diagram of which is depicted in FIG. 2 . The corrosion monitoring device 2 also includes a storage 11 , a transceiver interface 13 and a processor 15 , and the processor 15 is electrically connected to the storage 11 and the transceiver interface 13 . The corrosion monitoring device 2 can perform all the operations that the aforementioned corrosion monitoring device 1 can perform, so it also has its own functions and can achieve the technical effects it can achieve. Compared with the corrosion monitoring device 1, the corrosion monitoring device 2 also uses big data to build the corrosion health evaluation module M1 and the importance analysis module M2. The following description will focus on the differences between the corrosion monitoring device 2 and the corrosion monitoring device 1 .

In this embodiment, the memory 11 further stores a plurality of historical measurement data sets D1, . . . , Dk. Each of the historical measurement data sets D1, . . . , Dk is a measurement data set when the system performed the aromatic hydrocarbon process in the past. Each of the historical measurement data groups D1,..., Dk contains a plurality of historical characteristic data, and the historical characteristic data contained in each of the historical measurement data groups D1,..., Dk are one-to-one The process characteristic variables corresponding to the aromatic hydrocarbon process.

The processor 15 utilizes the historical feature data corresponding to the first selected characteristic variables in the historical measurement data sets D1, ..., Dk (as a training data set) and a first machine learning algorithm to generate a corrosion health assessment model Group M1. It should be noted that the present invention does not limit which machine learning algorithm must be the first machine learning algorithm. For example, the first machine learning algorithm may be an extreme gradient boosting (eXtreme Gradient Boosting; XGBoost) algorithm, a ridge regression (Ridge regression) algorithm, and a Lasso regression algorithm. In addition, the processor 15 utilizes the historical characteristic data corresponding to the first selected characteristic variables in the historical measurement data groups D1, ..., Dk (as a training data group), and the corrosion health evaluation module M1 for each training data group A prediction result and a second machine learning algorithm generate an importance analysis module M2. It should be noted that the present invention does not limit which machine learning algorithm the second machine learning algorithm must be. For example, the second machine learning algorithm may be SHAP (SHapley Additive exPlanations) algorithm, LIME (Local Interpretable Model-Agnostic Explanation) algorithm.

In some embodiments, before using the historical measurement data sets D1, ..., Dk to generate the corrosion health evaluation module M1 and the importance analysis module M2, the corrosion monitoring device 2 will also obtain the The first selected characteristic variables are selected from among the process characteristic variables.

In some embodiments, the processor 15 can analyze the historical measurement data sets D1, ..., Dk according to a statistical analysis method, thereby selecting more important ones from the process characteristic variables as the first selected characteristic variables . For example, the statistical analysis method may be a variation inflation factor method, a heatmap algorithm, and a correlation coefficient algorithm.

In some embodiments, the processor 15 first analyzes the historical measurement data sets D1, . The more important ones are selected from the process characteristic variables as a plurality of first candidate characteristic variables. Next, the processor 15 selects these first candidate characteristic variables and a plurality of second candidate characteristic variables (for example, the more important characteristic variables determined from the process characteristic variables by professionals familiar with the aromatic hydrocarbon process). The first selected feature variable. For example, the processor 15 may use the union of the first candidate feature variables and the second candidate feature variables as the first selected feature variables. For another example, the processor 15 can use the union of the first candidate feature variables and the second candidate feature variables as a plurality of third candidate feature variables, and the sending and receiving interface 13 or another sending and receiving interface outputs the first and second candidate feature variables. The three candidate characteristic variables are for professionals familiar with the aromatic hydrocarbon process to determine the more important ones from the third candidate characteristic variables as the first selected characteristic variables, and the sending and receiving interface 13 or another sending and receiving interface receives the first selection characteristic variable.

To sum up, the corrosion monitoring device 2 can train the corrosion health assessment module M1 and the importance analysis module M2 based on the historical measurement data sets D1, . . . , Dk. In addition, for any aromatic hydrocarbon process, the corrosion monitoring device 2 can select the one that has a more obvious influence on corrosion from a plurality of process characteristic variables of the aromatic hydrocarbon process as the first selected characteristic variable, and then use these second A corrosion health evaluation module M1 and an importance analysis module M2 are trained with historical measurement data corresponding to a selected characteristic variable. Therefore, the corrosion monitoring device 2 can select appropriate first selected characteristic variables for different aromatic hydrocarbon processes and different systems (that is, those process characteristic variables that are more likely to cause corrosion for systems that perform this aromatic hydrocarbon process), and then train The corrosion health evaluation module M1 which can accurately evaluate the corrosion health of the system and the importance analysis module M2 which provides relevant information are produced.

The third embodiment of the present invention is a corrosion monitoring device 3 , the schematic diagram of which is depicted in FIG. 3A . The corrosion monitoring device 3 also includes a storage 11 , a transceiver interface 13 and a processor 15 , and the processor 15 is electrically connected to the storage 11 and the transceiver interface 13 . The corrosion monitoring device 3 can perform all the operations that the aforementioned corrosion monitoring device 1 can perform, so it also has its own functions and can achieve the technical effects it can achieve. The following description will focus on the differences between the corrosion monitoring device 3 and the corrosion monitoring device 1 .

In this embodiment, the storage 11 stores a trained pH value prediction module M3 and a trained corrosion rate prediction module M4. For a system that implements an aromatic hydrocarbon process, the corrosion monitoring device 3 uses the pH value prediction module M3 and the corrosion rate prediction module M4 to predict the corrosion rate of the distillation column, so that the system manager can use the predicted The corrosion rate evaluates the remaining life of the system for the purpose of quantitative management. The operating principle and details of the corrosion monitoring device 3 are described in detail below.

As mentioned above, an aromatic hydrocarbon process has a plurality of process characteristic variables (not shown), wherein each of the process characteristic variables can be any factor related to the aromatic hydrocarbon process, such as: feed, compound concentration, operating settings, Environmental factors, quality indicators, etc. A subset of these process characteristic variables have a significant impact on whether corrosion occurs in the distillation column in the system performing the aromatics process. For ease of description, the process characteristic variables in the subset may be referred to as second selected characteristic variables. This embodiment does not limit the method of determining the second selected characteristic variables of an aromatic hydrocarbon process, which can be determined by a user (for example: the administrator of the system, a professional familiar with the aromatic hydrocarbon process), or by a statistical Analytical decision. Taking the aromatics process shown in FIG. 1B as an example, each of the second selected characteristic variables may be related to one of a dechlorination operation and a debutanizer state in the aromatics process. For example, such second selected characteristic variables may include total chlorine concentration before dechlorination tank, total chlorine concentration after dechlorination tank, feed to debutanizer, pentane (C5) concentration at the top of debutanizer And the butane (C4) concentration at the bottom of the butanizer.

In this embodiment, the transceiver interface 13 of the corrosion monitoring device 3 will receive and execute the A current measurement data set DS2 of the system for aromatic hydrocarbon processing. The current measurement data set DS2 includes a plurality of pieces of current characteristic data, and the current characteristic data correspond to the second selected characteristic variables of the aromatic hydrocarbon process in a one-to-one manner. It should be noted that the present invention does not limit how the transceiver interface 13 obtains the current measurement data set DS2. For example, a plurality of information table points (for example: measuring device, sensor) can be configured in the system to measure the value of each characteristic variable of the process when the system executes the aromatic hydrocarbon process, and the sending and receiving interface 13 can be directly obtained from The table points corresponding to the second selected characteristic variables receive the current characteristic data. For another example, multiple information table points (such as measuring devices and sensors) can be configured in the system to measure the values of the process characteristic variables when the system executes the aromatic hydrocarbon process, and these values will be collected into A data collector (for example: a server), the transceiver interface 13 receives the values corresponding to the second selected characteristic variables (ie, the current characteristic data) from the data collector.

Afterwards, the processor 15 inputs the current characteristic data in the current measurement data set DS2 into the pH value prediction module M3 to obtain an estimated pH value (not shown) of the first distillation tower of the system, and then Input the estimated pH value and an operating temperature (not shown) of the distillation column (for example, the butanizer in Figure 1B) into the corrosion rate prediction module M4 to obtain a preliminary prediction of the distillation column Corrosion rate (not shown). It should be explained that since the corrosion rate prediction module M4 is trained based on the multiple pieces of corrosion rate data contained in the carbon steel corrosion rate database of the hydrogen chloride aqueous solution, and the hydrogen chloride aqueous solution contains the carbon steel corrosion rate database. The environment measured by the corrosion rate data may be different from the environment of the system that executes the aromatic hydrocarbon process, so the processor 15 corrects the preliminary predicted corrosion rate to a correction based on a correction coefficient that can reflect the difference between the two environments Predict the corrosion rate R.

Taking the API 581 database published by the American Petroleum Institute as an example of the corrosion rate database of carbon steel in aqueous hydrogen chloride solution, this paper explains in detail why it is necessary to use a correction factor to convert the preliminary predicted corrosion rate predicted by the corrosion rate prediction module M4 to Rate correction is the revised predicted corrosion rate R. It should be understood that if other databases are used as the database for the corrosion rate of carbon steel by aqueous hydrogen chloride solution, the corresponding correction coefficients may be different.

Figure 3B presents multiple pieces of corrosion rate data contained in the API 581 database, where each piece of corrosion rate data is a corrosion rate (for example: several millimeters of corrosion per year), and each piece of corrosion rate data corresponds to a temperature value and a pH value. Furthermore, the American Petroleum Institute uses the operating environment shown in Figure 3C to generate each corrosion rate data in the API 581 database. As shown in Figure 3C, a carbon steel test piece 10 is placed in an aqueous hydrogen chloride solution 12 with a specific pH value and the environment is at a specific temperature value, the aqueous hydrogen chloride solution 12 will cause uniform corrosion of the carbon steel test piece 10, so it can be The corrosion rate was calculated based on the degree of uniform corrosion of the carbon steel test piece 10 . However, the operating environment of the system for performing the aromatic hydrocarbon process is different from that shown in FIG. 3C. In the system for performing the aromatic hydrocarbon process, the distillation column has a corresponding overhead system (that is, the circulation system at the top of the distillation column). In this tower top system, aqueous hydrogen chloride solution 16 can be distributed in the oil phase with droplet form, passes through the carbon steel pipe line 14 in the tower top system, and only the hydrogen chloride aqueous solution 16 droplet of adjacent tube wall just can be on the carbon steel pipe line 14 Causes corrosion, as shown at the dotted line in Figure 3D. It can be seen from the above description that the corrosion behavior caused by the hydrogen chloride aqueous solution is different in different operating environments. Therefore, the corrosion monitoring device 3 needs to correct the preliminary predicted corrosion rate predicted by the corrosion rate prediction module M4 to the corrected predicted corrosion rate R according to a correction coefficient that can reflect the difference between the two environments.

(1) Specifically, the processor 15 can calculate the correction coefficient according to a water-tube wall contact area (for example: the dotted line shown in Figure 3D) and a pipeline internal surface area in the overhead system corresponding to the distillation column . For example, the processor 15 can calculate the correction coefficient according to the following formula (1):

(1)

代表修正係數，變數

代表水與管壁接觸面積，而變數

代表管線內部表面積。需說明者，由於管線內部表面積只有一半與氯化氫水溶液接觸，因此公式(1)中需將其除以2。 In the above formula (1), the variable

represents the correction factor, variable

Represents the contact area between water and pipe wall, while the variable

Represents the internal surface area of the pipeline. It should be noted that since only half of the internal surface area of the pipeline is in contact with the aqueous hydrogen chloride solution, it needs to be divided by 2 in the formula (1).

(2) In addition, the processor 15 can calculate the contact area between the water and the pipe wall according to the water content of a corrosion layer, a water density, and an aqueous solution film thickness. For example, the processor 15 can calculate the contact area between the water and the pipe wall according to the following formula (2):

(2)

代表腐蝕作用層含水量，變數

代表水密度，而變數

代表水溶液膜厚。進一步言，關於水溶液膜厚的計算方式，可參考S. K. Alghoul等人於2010年在「Liquid Atomization and Spray Systems」年度會議中所發表的「Experimental investigation of a Single Droplet Interaction with Shear Driven Film」一文。 In the above formula (2), the variable

Represents the water content of the corrosion layer, variable

represents water density, while the variable

represents the film thickness of the aqueous solution. Furthermore, for the calculation method of the film thickness of the aqueous solution, please refer to the article "Experimental investigation of a Single Droplet Interaction with Shear Driven Film" published by SK Alghoul et al. in the annual meeting of "Liquid Atomization and Spray Systems" in 2010.

(3) In addition, processor 15 can be based on a pipeline volume in the overhead system, a fluid density in the distillation column, a water concentration in the distillation column, and a corrosive layer in the distillation column. Area ratio, calculate the water content of the corrosion layer. For example, the processor 15 can calculate the water content of the corrosion layer according to the following formula (3):

(3)

代表塔頂系統中的管線容積，變數

代表該餾除塔中的一流體密度，變數

代表該餾除塔中的一含水濃度，而變數

代表腐蝕作用層所占截面積比例。進一步言，若儲存器11儲存經訓練的一水分預測模組M5，則可將當前量測資料組DS2中的該等當前特徵資料輸入水分預測模組M5，使水分預測模組M5預測出變數

所代表的含水濃度。另外，關於腐蝕作用層所占截面積比例的計算方式，可參考L.D. Paolinelli等人於2018年在「Chemical Engineering Science」期刊所發表的「Study of water wetting and water layer thickness in oil-water flow in horizontal pipes with different wettability」一文。 In the above formula (3), the variable

represents the line volume in the overhead system, variable

Represents the density of a fluid in the distillation column, variable

represents a water concentration in the distillation tower, and the variable

Represents the proportion of the cross-sectional area occupied by the corrosion layer. Further, if the memory 11 stores a trained moisture prediction module M5, the current characteristic data in the current measurement data set DS2 can be input into the moisture prediction module M5, so that the moisture prediction module M5 can predict the variable

represents the water concentration. In addition, for the calculation method of the proportion of the cross-sectional area of the corrosion layer, please refer to "Study of water wetting and water layer thickness in oil-water flow in horizontal" published by LD Paolinelli et al. in the journal "Chemical Engineering Science" in 2018 pipes with different wettability".

After calculating the correction coefficient, for any preliminary predicted corrosion rate generated by the corrosion rate prediction module M4, the processor 15 can use the correction coefficient to correct the preliminary predicted corrosion rate to a corrected predicted corrosion rate R.

To sum up, the corrosion monitoring device 3 will also consider a plurality of second selected characteristic variables that have a significant impact on corrosion in the aromatic hydrocarbon process, and can reflect the actual operating environment of the system (for example: the corrosion solution and the system) The actual contact condition of the pipeline) is used to correct the preliminary predicted corrosion rate to a corrected predicted corrosion rate R, so the corrected predicted corrosion rate R can be accurately used as the actual corrosion rate of the distillation column in the system. In this way, the manager of the system can understand the corrosion status of the distillation tower based on the more accurate correction prediction corrosion rate, and even evaluate the remaining life of the system based on this, so as to achieve the purpose of quantitative management.

The fourth embodiment of the present invention is a corrosion monitoring device 4 , the schematic diagram of which is depicted in FIG. 4 . The corrosion monitoring device 4 also includes a storage 11 , a transceiver interface 13 and a processor 15 , and the processor 15 is electrically connected to the storage 11 and the transceiver interface 13 . The corrosion monitoring device 4 can perform all the operations that the aforementioned corrosion monitoring device 3 can perform, so it also has its own functions and can achieve the technical effects it can achieve. Compared with the corrosion monitoring device 3, the corrosion monitoring device 4 also uses big data to build a pH value prediction module M3, a corrosion rate prediction module M4, and a moisture prediction module M5. The following description will focus on the differences between the corrosion monitoring device 4 and the corrosion monitoring device 3 .

In this embodiment, the memory 11 further stores a plurality of historical measurement data sets D1, . . . , Dk. Each of the historical measurement data sets D1, . . . , Dk is a measurement data set when the system performed the aromatic hydrocarbon process in the past. Each of the historical measurement data groups D1,..., Dk contains a plurality of historical characteristic data, and the historical characteristic data contained in each of the historical measurement data groups D1,..., Dk are one-to-one The process characteristic variables corresponding to the aromatic hydrocarbon process.

Because the historical measurement data sets D1, ..., Dk obtained by the same aromatic hydrocarbon process in the same system have fewer records and a smaller range of data variation, the processor 15 will use the chemical process simulation software Aspen Plus (also known as Aspen+), from the historical measurement data sets D1, ..., Dk to simulate the multiple training data sets T1, ..., Tz with a large number of items and a large data variation range. Specifically, the processor 15 can use the chemical process simulation software Aspen Plus and the historical measurement data sets D1, ..., Dk to simulate the model of the system (for example: simulate the butane removal tower system shown in Fig. 1B model), different operating conditions are planned on the model, thereby generating data under different operating conditions as training data sets T1, ..., Tz. Each of the training data sets T1, ..., Tz contains a plurality of training feature data, and the training feature data contained in each of the training data sets T1, ..., Tz correspond one-to-one to the fragrance These process characteristic variables of the hydrocarbon process.

The processor 15 uses the training feature data corresponding to the second selected feature variables in the training data sets T1, . . . , Tz and a third machine learning algorithm to train the pH value prediction module M3. It should be understood that, in order to train the pH value prediction module M3, each of the training feature data included in the training data sets T1, . . . , Tz has a pH value. Similarly, the processor 15 uses the training feature data corresponding to the second selected feature variables in the training data sets T1, . . . , Tz and the third machine learning algorithm to train the moisture prediction module M5. It should be understood that in order to train the moisture prediction module M5, each of the training feature data included in the training data sets T1, . . . , Tz has a water concentration. It should be noted that the present invention does not limit which machine learning algorithm the third machine learning algorithm must be. For example, the third machine learning algorithm can be any deep neural network (Deep Neural Network; DNN) algorithm.

In some embodiments, before using the training data sets T1, ..., Tz to generate the pH value prediction module M3, the corrosion monitoring device 4 will also select these process characteristic variables from the aromatic hydrocarbon process. The second selected feature variable.

In some embodiments, the processor 15 may analyze the training data sets T1, . . . , Tz according to a statistical analysis method, thereby selecting more important ones from the process characteristic variables as the second selected characteristic variables. For example, the statistical analysis method may be correlation coefficient method, stepwise regression analysis method or other similar algorithms.

In some embodiments, the processor 15 first analyzes the training data sets T1, . . . The more important ones are selected from the process characteristic variables as the plurality of fourth candidate characteristic variables. Next, the processor 15 selects these fourth candidate characteristic variables and a plurality of fifth candidate characteristic variables (for example, the more important characteristic variables determined from the process characteristic variables by professionals familiar with the aromatic hydrocarbon process). The second selected feature variable. For example, the processor 15 may use the union of the fourth candidate feature variables and the fifth candidate feature variables as the second selected feature variables. For another example, the processor 15 can use the union of the fourth candidate feature variables and the fifth candidate feature variables as a plurality of sixth candidate feature variables, and the transceiver interface 13 or another transceiver interface outputs the fourth candidate feature variables. The six candidate characteristic variables are for professionals familiar with the aromatic hydrocarbon process to determine the more important ones from the sixth candidate characteristic variables as the second selected characteristic variables, and the sending and receiving interface 13 or another sending and receiving interface receives the second selected characteristic variable.

In addition, the processor 15 utilizes a plurality of pieces of corrosion rate data (as a training data set) contained in a hydrogen chloride aqueous solution to carbon steel corrosion rate database (for example: API 581 database published by the American Petroleum Institute) and a fourth machine learning The algorithm produces the corrosion rate prediction module M4. It should be noted that the present invention does not limit which machine learning algorithm the fourth machine learning algorithm must be. For example, the fourth machine learning algorithm may be a K-Nearest Neighbor algorithm, a Gradient Boosting Machines algorithm, a Gaussian Process Regression algorithm, or other similar algorithms .

In summary, the corrosion monitoring device 4 can also generate training data sets T1, ..., Tz based on the historical measurement data sets D1, ..., Dk, and then use the training data sets T1, ..., Tz to train the pH value Prediction module M3. The corrosion monitoring device 4 can also train the corrosion rate prediction module M4 based on the hydrogen chloride aqueous solution to the carbon steel corrosion rate database. In addition, for any aromatic hydrocarbon process, the corrosion monitoring device 4 can also select the ones that have a more obvious influence on corrosion from the plurality of process characteristic variables of the aromatic hydrocarbon process as the plurality of second selected characteristic variables, and then use The training characteristic data corresponding to the second selected characteristic variables train the pH value prediction module M3. Therefore, the corrosion monitoring device 4 can select appropriate second selected characteristic variables for different aromatic hydrocarbon processes and different systems (that is, those process characteristic variables that are more likely to cause corrosion for systems that perform this aromatic hydrocarbon process), and then train A pH value prediction module M3 capable of accurate prediction is provided for the corrosion rate prediction module M4 to predict an accurate corrosion rate.

The fifth embodiment of the present invention is a corrosion monitoring method, which is applicable to an electronic computing device (for example: corrosion monitoring devices 1 , 2 , 3 , 4 ). The main flowchart of the corrosion monitoring method is depicted in FIG. 4, which at least includes step S501, step S503 and step S505.

In step S501, a current measurement data set of a system performing an aromatic hydrocarbon process is received by the electronic computing device, wherein the current measurement data set includes a plurality of current characteristic data, and the current characteristic data are one-to-one A plurality of selected characteristic variables corresponding to the aromatic hydrocarbon process. Depending on the specific aromatic hydrocarbon process, in some cases, each of the selected characteristic variables is associated with one of a hydrogenation reaction, a recombination reaction, a catalyst regeneration reaction, and a dechlorination operation in the aromatic hydrocarbon process . In step S503, the electronic computing device inputs the current feature data into a trained corrosion health assessment module to obtain a corrosion health of the system. In step S505, the electronic computing device inputs the current characteristic data and the corrosion health into a trained importance analysis module to obtain a contribution degree of each selected characteristic variable to the corrosion health.

In some embodiments, the corrosion monitoring method further includes step S507 and step S509. In step S507, the electronic computing device determines whether the corrosion health is lower than a first threshold. If the judgment result of step S507 is that the corrosion health degree is lower than the first threshold value, the electronic computing device will also output at least one key characteristic variable among the selected characteristic variables, wherein each of the at least one key characteristic variable corresponds to The degree of contribution is greater than a second threshold. The at least one key characteristic variable is the more critical one among the selected characteristic variables that causes corrosion of equipment of the system (or worsens corrosion of equipment of the system).

In some embodiments, the electronic computing device also stores a plurality of historical measurement data sets, and each of the historical measurement data sets includes a plurality of historical characteristic data corresponding one-to-one to the selected characteristics of the aromatic hydrocarbon process variable. In these embodiments, the corrosion monitoring method may further execute a step of generating the corrosion health assessment module by the electronic computing device using the historical measurement data sets and a first machine learning algorithm. In addition, the corrosion monitoring method can also perform a step of using the historical measurement data sets, a prediction result of the corrosion health assessment module for each of the historical measurement data sets, and a second machine learning by the electronic computing device An algorithm generates the importance analysis module.

In some embodiments, before generating the corrosion health assessment module and the importance analysis module, the corrosion monitoring method will first determine the selected characteristic variables. Specifically, in these embodiments, each of the historical measurement data sets stored in the electronic computing device includes a plurality of pieces of historical characteristic data corresponding one-to-one to a plurality of process characteristic variables of the aromatic hydrocarbon process. The corrosion monitoring method may also perform a step of analyzing the historical measurement data sets by the electronic computing device according to a statistical analysis method, thereby selecting the selected characteristic variables from the process characteristic variables.

In some embodiments, the corrosion monitoring method may adopt the following steps to determine the selected characteristic variables. The corrosion monitoring method includes a step of analyzing the historical measurement data sets by the electronic computing device according to a statistical analysis method to select a plurality of first candidate characteristic variables from the process characteristic variables, and further includes another step, by The electronic computing device selects the selected characteristic variables according to the first candidate characteristic variables and a plurality of second candidate characteristic variables.

In addition to the above steps, the corrosion monitoring method of the present invention can also perform all the operations and steps that the corrosion monitoring devices 1, 2, 3, 4 can perform, have the same function, and achieve the same technical effect. Those with ordinary knowledge in the technical field of the present invention can directly understand how the corrosion monitoring method of the present invention is based on the above-mentioned corrosion monitoring devices 1, 2, 3, 4 to perform these operations and steps, have the same function, and achieve the same Technical effects, so no more details.

The corrosion monitoring methods described in the above embodiments can be realized by a computer program product including a plurality of program instructions. The computer program product can be a file that can be transmitted over the Internet, or can be stored in a non-transitory computer-readable storage medium. After the program instructions contained in the computer program product are loaded into an electronic computing device (for example: corrosion monitoring devices 1, 2, 3, 4), the computer program executes the corrosion monitoring as described in the above-mentioned embodiments method. The non-transitory computer readable storage medium can be an electronic product, for example: a read only memory (Read Only Memory; ROM), a flash memory, a floppy disk, a hard disk, a compact disk (Compact Disk; CD), a Digital Versatile Disc (Digital Versatile Disc; DVD), a flash drive, a database accessible from the Internet, or any other storage device with the same function known to those skilled in the art to which the present invention pertains media.

It should be noted that certain terms (including: machine learning algorithms, candidate feature variables, and threshold values) in the patent specification and scope of the patent application of the present invention are preceded by "first", "second", and "third" , "Fourth", "Fifth" or "Sixth", such "First", "Second", "Third", "Fourth", "Fifth" and "Sixth" are used for These terms are distinguished from each other. In addition, in the patent specification and patent application scope of the present invention, some terms (including: measurement data set, characteristic data) are preceded by "history", which is used to represent the data corresponding to these terms It is the data of the past operation of the system. Furthermore, some terms (including: measurement data group, characteristic data) are preceded by "current", which is used to indicate that the data corresponding to these terms are the data at the time of the system's current operation.

In summary, the corrosion monitoring technology provided by the present invention (including at least the device, the method and its computer program product) utilizes a trained corrosion health evaluation module and a trained importance analysis module to monitor and execute a Corrosion health of a system for an aromatics process (ie, the health of the system in terms of corrosion conditions). Specifically, some characteristic variables in the aromatic hydrocarbon production process (that is, the aforementioned selected characteristic variables) have a more obvious impact on whether the equipment in the system is corroded, so the corrosion monitoring technology provided by the present invention implements a system A plurality of pieces of current characteristic data corresponding to these selected characteristic variables in the process of aromatic hydrocarbons are input into the corrosion health evaluation module to obtain a corrosion health degree of the system, and the current characteristic data are input into the importance analysis module to obtain A degree of contribution of each of the selected characteristic variables to the corrosion health. When the corrosion health is lower than a first threshold, the corrosion monitoring technology provided by the present invention can also output at least one key characteristic variable among the selected characteristic variables.

The corrosion monitoring technology provided by the present invention can also train the corrosion health assessment module and the importance analysis module based on a plurality of historical measurement data sets. In addition, for any aromatic hydrocarbon process, the corrosion monitoring technology provided by the present invention can also select the one that has a significant impact on corrosion from a plurality of process characteristic variables as the selected characteristic variable, and then use the selected characteristic variable The corresponding historical measurement data trains the corrosion health evaluation module and the importance analysis module. Therefore, the corrosion monitoring technology provided by the present invention can train a corrosion health evaluation module that can accurately evaluate the corrosion health of the system and an importance analysis module that provides relevant information for different aromatic hydrocarbon processes and different systems. Therefore, the corrosion health of the system can be accurately evaluated, and then an early warning can be generated before the system corrodes or the corrosion degree deteriorates, allowing users to take preventive measures against key characteristic variables to avoid corrosion or worsen the corrosion condition.

The above-mentioned embodiments are used to illustrate some implementation aspects of the present invention and explain the technical features of the present invention, but are not used to limit the scope and scope of the present invention. Any change or equivalence arrangement that can be easily accomplished by those with ordinary knowledge in the technical field of the present invention belongs to the scope claimed by the present invention, and the protection scope of the present invention is subject to the scope of the patent application.

1, 2, 3, 4: Corrosion monitoring device 10: Carbon steel test piece 11: Storage 12: Hydrogen chloride aqueous solution 13: Sending and receiving interface 14: Carbon steel pipeline 15: Processor 16: Hydrogen chloride aqueous solution M1: Corrosion Health Evaluation Module M2: Importance analysis module M3: pH value prediction module M4: Corrosion Rate Prediction Module M5:Moisture prediction module D1,..., Dk: historical measurement data group DS1, DS2: current measurement data set HI: Corrosion Health O1, ..., Om: degree of contribution R: Revised predicted corrosion rate T1, ..., Tz: training data set S501, S503, S505, S507, S509: steps

FIG. 1A depicts a schematic structural view of the corrosion monitoring device 1 of the first embodiment.

FIG. 1B depicts a schematic diagram of a debutanizer system performing an aromatics process.

FIG. 2 depicts a schematic structural diagram of a corrosion monitoring device 2 according to a second embodiment.

FIG. 3A depicts a schematic structural diagram of a corrosion monitoring device 3 according to a third embodiment.

Figure 3B presents a plurality of corrosion rate data contained in the API 581 database.

Figure 3C depicts the operating environment used to generate each piece of corrosion rate data for the API 581 database.

FIG. 3D depicts a schematic diagram of the carbon steel pipeline 14 being corroded only by the hydrogen chloride aqueous solution 16 droplets adjacent to the pipe wall of the carbon steel pipeline 14 in the overhead system corresponding to the first distillation tower.

FIG. 4 depicts a schematic structural diagram of a corrosion monitoring device 4 according to a fourth embodiment.

Fig. 5 depicts the main flowchart of the corrosion monitoring method of the fifth embodiment.

none

S501, S503, S505, S507, S509: steps

Claims (13)

A corrosion monitoring device comprising: a memory storing a trained corrosion health assessment module and a trained importance analysis module; A sending and receiving interface for receiving a current measurement data set of a system that executes an aromatic hydrocarbon process, wherein the current measurement data set includes a plurality of current characteristic data, and the current characteristic data correspond to the aromatic hydrocarbon in a one-to-one manner a plurality of selected characteristic variables of the process; and a processor, electrically connected to the storage and the transceiver interface, input the current characteristic data into the corrosion health evaluation module to obtain a corrosion health of the system, and input the current characteristic data and the corrosion The health degree is input into the importance analysis module to obtain a contribution degree of each selected characteristic variable to the corrosion health degree. The corrosion monitoring device as claimed in claim 1, wherein when the corrosion health is lower than a first threshold value, one of the transceiver interface and the other transceiver interface further outputs at least one key characteristic of the selected characteristic variables variables, wherein the contribution degree corresponding to each of the at least one key characteristic variable is greater than a second threshold value. The corrosion monitoring device as described in Claim 1, wherein the storage also stores a plurality of historical measurement data sets, each of which includes a plurality of historical characteristic data corresponding to the aromatic hydrocarbon process in a one-to-one manner and other selected characteristic variables, and the processor utilizes the historical measurement data sets and a machine learning algorithm to generate the corrosion health evaluation module. The corrosion monitoring device as described in Claim 1, wherein the storage also stores a plurality of historical measurement data sets, each of which includes a plurality of historical characteristic data corresponding to the aromatic hydrocarbon process in a one-to-one manner and other selected characteristic variables, and the processor utilizes the historical measurement data sets, a prediction result of the corrosion health evaluation module for each of the historical measurement data sets, and a machine learning algorithm to generate the importance analysis module . The corrosion monitoring device as described in claim 1, wherein the storage also stores a plurality of historical measurement data sets, each of which includes a plurality of historical characteristic data corresponding one-to-one to the plural of the aromatic hydrocarbon process process characteristic variables, the processor also analyzes the historical measurement data sets according to a statistical analysis method, thereby selecting the selected characteristic variables from the process characteristic variables. The corrosion monitoring device as described in claim 1, wherein the storage also stores a plurality of historical measurement data sets, each of which includes a plurality of historical characteristic data corresponding one-to-one to the plural of the aromatic hydrocarbon process process characteristic variables, the processor also analyzes the historical measurement data sets according to a statistical analysis method to select a plurality of first candidate characteristic variables from the process characteristic variables, and the processor also selects a plurality of first candidate characteristic variables according to the first candidate The characteristic variable and the plurality of second candidate characteristic variables select the selected characteristic variables. A corrosion monitoring method, suitable for an electronic computing device, comprising the following steps: Receive a current measurement data set of a system performing an aromatic hydrocarbon process, wherein the current measurement data set includes a plurality of current characteristic data, and the current characteristic data correspond to the plural pieces of the aromatic hydrocarbon process in a one-to-one manner selected feature variables; inputting the current characteristic data into a trained corrosion health assessment module to obtain a corrosion health of the system; and Inputting the current characteristic data and the corrosion health into a trained importance analysis module to obtain a contribution degree of each selected characteristic variable to the corrosion health. The corrosion monitoring method described in claim item 7 also includes the following steps: When the corrosion health is lower than a first threshold, at least one key characteristic variable among the selected characteristic variables is output, wherein the contribution degree corresponding to each of the at least one key characteristic variable is greater than a second threshold value. The corrosion monitoring method as described in claim item 7, wherein the electronic computing device also stores a plurality of historical measurement data sets, and each of the historical measurement data sets includes a plurality of historical characteristic data corresponding to the aromatic hydrocarbon process in a one-to-one manner The selected characteristic variables, and the corrosion monitoring method further includes the following steps: The corrosion health evaluation module is generated by using the historical measurement data sets and a machine learning algorithm. The corrosion monitoring method as described in claim item 7, wherein the electronic computing device also stores a plurality of historical measurement data sets, and each of the historical measurement data sets includes a plurality of historical characteristic data corresponding to the aromatic hydrocarbon process in a one-to-one manner The selected characteristic variables, and the corrosion monitoring method further includes the following steps: The importance analysis module is generated by using the historical measurement data sets, a prediction result of the corrosion health assessment module for each of the historical measurement data sets, and a machine learning algorithm. The corrosion monitoring method as described in claim item 7, wherein the electronic computing device also stores a plurality of historical measurement data sets, and each of the historical measurement data sets includes a plurality of historical characteristic data corresponding to the aromatic hydrocarbon process in a one-to-one manner A plurality of process characteristic variables, and the corrosion monitoring method also includes the following steps: The historical measurement data sets are analyzed according to a statistical analysis method, thereby selecting the selected characteristic variables from the process characteristic variables. The corrosion monitoring method as described in claim item 7, wherein the electronic computing device also stores a plurality of historical measurement data sets, and each of the historical measurement data sets includes a plurality of historical characteristic data corresponding to the aromatic hydrocarbon process in a one-to-one manner A plurality of process characteristic variables, and the corrosion monitoring method also includes the following steps: analyzing the historical measurement data sets according to a statistical analysis method to select a plurality of first candidate characteristic variables from among the process characteristic variables; and The selected characteristic variables are selected according to the first candidate characteristic variables and the plurality of second candidate characteristic variables. A computer program product, after being loaded into the computer program product through an electronic computing device, the electronic computing device executes a plurality of program instructions contained in the computer program product to achieve corrosion as described in any one of claims 7 to 12 monitoring method.

Images