阅读说明：本技术 一种石脑油裂解制乙烯生产故障诊断方法 (Method for diagnosing production fault of ethylene prepared by naphtha cracking ) 是由 孙巍 马方园 张锦红 展益彬 王璟德 韩先尧 李英壮 田胜伟 于 2019-10-17 设计创作，主要内容包括：本发明涉及一种石脑油裂解制乙烯生产故障诊断方法,利用获取的多个石脑油裂解制乙烯过程中完整历史运行周期的离线数据,建立用于实际生产中的运行监测模型和判定正常工况的控制线,并将监测系统采集的各个监测变量的即时数据通过算法综合成一个生产状态指数值,并与正常工况的控制线进行比对,若出现偏离,继续比对后续的生产状态指数值,若后续的生产状态指数值偏离正常工况控制线的连续时间大于设定时间,且排除人为对生产工况的调整时,监测系统上报生产故障信息,提高了上报故障的准确率,减少误报；同时解决过程监测中故障出现时,难以确定故障的主因的问题。(The invention relates to a naphtha cracking ethylene production fault diagnosis method, which comprises the steps of establishing an operation monitoring model used in actual production and a control line for judging normal working conditions by utilizing acquired offline data of a complete historical operation period in a plurality of naphtha cracking ethylene production processes, synthesizing instant data of monitoring variables acquired by a monitoring system into a production state index value through an algorithm, comparing the production state index value with the control line of the normal working conditions, continuously comparing subsequent production state index values if deviation occurs, reporting production fault information by the monitoring system if the continuous time of the subsequent production state index value deviating from the control line of the normal working conditions is longer than set time, and eliminating manual production working condition adjustment, thereby improving the fault reporting accuracy and reducing false alarm; meanwhile, the problem that the main cause of the fault is difficult to determine when the fault occurs in process monitoring is solved.)

1. A production fault diagnosis method for preparing ethylene by naphtha cracking is characterized by comprising the following steps:

1) establishing a production process monitoring model;

2) determining a control line for judging normal working conditions;

3) the monitoring system integrates the acquired instant data of each monitoring variable into a production state index value; comparing the production state index value with the control line under the normal working condition, and if the production state index value is within the range of the control line under the normal working condition, the monitoring system does not act;

if the production state index value deviates from the control line of the normal working condition, the monitoring system continuously compares whether each subsequent production state index value in a set time period deviates from the control line of the normal working condition, if not,

the monitoring system does not report a production fault; if so,

the monitoring system judges whether the production working condition is changed manually or not, if so,

the monitoring system does not report a production fault, and if not,

and the monitoring system reports the production fault.

2. The method for diagnosing the production fault of the ethylene produced by naphtha cracking as claimed in claim 1, wherein the method for establishing the production process monitoring model comprises the following steps:

1) setting monitoring variables in the production process of ethylene by naphtha cracking;

2) acquiring offline data of a plurality of complete historical operating cycles;

3) the method comprises the steps that offline data corresponding to the monitoring variable which runs stably for a first set time length in offline data of a complete historical running period are selected randomly;

counting the monitoring variable data of stable operation, acquiring a corresponding normal working condition control line, and establishing a monitoring model;

4) repeating the step 3) for N times, and establishing a set containing N monitoring models, wherein N is a natural number;

5) extracting any monitoring model in the monitoring model set, comparing the extracted monitoring model with the offline data of the complete historical operating period in the step 3), and judging:

if the monitoring accuracy rate of the stable operation of the monitoring model reaches a first set value and the false alarm rate is lower than a second set value, the monitoring model is reserved, otherwise, the monitoring model is abandoned;

6) extracting any monitoring model in the step 5), comparing the extracted monitoring model with the other offline data in the step 2), and judging:

if the monitoring accuracy rate of the stable operation of the monitoring model reaches a third set value and the false alarm rate is lower than a fourth set value, otherwise, discarding the monitoring model;

7) and repeating the step 6), and obtaining a final better monitoring model from the set containing N monitoring models.

3. The method of claim 1, wherein the monitoring variables are selected according to the type of the cracking furnace, including but not limited to: and (3) convection section: cracking raw material naphtha feed flow, cracking raw material naphtha feed temperature and pressure, dilution steam feed flow, dilution steam feed temperature, furnace tube cross section temperature and pressure, radiation section: one or more of furnace tube outlet temperature, furnace tube outlet pressure, furnace temperature, fuel gas flow and average molecular mass thereof.

4. The method of diagnosing a malfunction in production of ethylene from naphtha cracking according to claim 2, wherein the first set point is the same as or different from the third set point.

5. The method of diagnosing a malfunction in production of ethylene from naphtha cracking according to claim 2, wherein the second set point is the same as or different from the fourth set point.

6. The method for diagnosing the production failure of ethylene through naphtha cracking as claimed in claim 2, further comprising a step of preprocessing data of an off-line between the step 2) and the step 3).

7. The method as claimed in claim 2, wherein the data preprocessing is performed by using a current value supplementing method or an estimation method.

8. The method as claimed in claim 2, wherein in the step 7), if two or more final better monitoring models are obtained, the models are ranked according to the third set value and the fourth set value in the step 6), and the highest ranked model is determined as the better monitoring model.

Technical Field

The invention belongs to the technical field of petrochemical industry, in particular relates to an operation monitoring technology of a cracking furnace for preparing ethylene by naphtha cracking, and particularly relates to a fault diagnosis method for preparing ethylene by naphtha cracking.

Background

The cracking process of the steam pyrolysis method in the process technology of preparing ethylene by cracking naphtha serving as a raw material at least comprises a raw material supply preheating system, a pyrolysis system, an oil-gas combined combustion system and a waste heat system. The whole production period of preparing ethylene by cracking the naphtha raw material refers to that the cracking reaction is stopped from the beginning of the cracking production after the decoking of the cracking furnace for preparing ethylene by cracking naphtha to the end of the cracking production of ethylene by cracking naphtha.

The cracking furnace is operated periodically, because coke is formed and deposited slowly along with the continuous reaction on the inner wall surface of the furnace tube in the radiation section in the actual production, the heat transfer efficiency is influenced, the heat load of the furnace tube is increased, the mechanical property and the service life of the furnace tube are influenced by the overlarge heat load, and the coke cleaning operation is required to be carried out for a certain time. If the production period is short, the decoking operation is too frequent, or the operation fault occurs in the production period, so that the deviation from the normal operation state is caused, the product yield and the economic benefit of the factory are directly influenced, and the production safety of the factory is threatened. Therefore, faults occurring in the operation process need to be identified in time, the root cause needs to be found out, and reference is provided for operators to adjust production operation; and evaluating and predicting the use condition of the equipment to optimize the operation condition, prolong the production operation period and provide a basis for the formulation of an equipment maintenance plan.

The variables selected and collected according to different types of the cracking furnaces in the production process of the cracking furnace of the steam pyrolysis method in the process technology for preparing ethylene by cracking the raw material of the actual naphtha include but are not limited to: and (3) convection section: 1. raw material part: the feed flow rate of the cracking raw material naphtha, the feed temperature and pressure of the cracking raw material naphtha and the feed flow rate of the cracking raw material naphtha are set values; 2. dilution steam part: set values of dilution steam feed flow, dilution steam feed temperature, and dilution steam feed quantity; 3. cross-section temperature and pressure part: the temperature of the cross section of the furnace tube and the pressure of the cross section of the furnace tube; a radiation section: 4. furnace exit temperature section: the radiant section furnace tube outlet temperature; 5. furnace exit pressure section: furnace tube outlet pressure; 6. cracking gas part: the contents of methane, ethane, ethylene, propane and propylene; 7. a hearth part: the composition of cracking gas, the temperature of a hearth, the composition of flue gas, the flow of fuel gas, the average molecular mass of the fuel gas and other variable data information.

Because a large amount of variable data information and the variable information have certain relevance in the actual production process, if a certain variable data is changed, the change of one or more other variable data may be affected, and the judgment of which variable data is the main cause of the fault is more difficult. The primary cause of the identified fault can be determined for this situation based on the knowledge of an experienced operator, in combination with the results of the process monitoring.

Disclosure of Invention

The invention aims to provide a method for diagnosing a fault in ethylene production by naphtha cracking, which aims to solve the problem that the main cause of the fault is difficult to determine when the fault occurs in process monitoring.

The invention is realized by the following technical scheme:

a production fault diagnosis method for preparing ethylene by naphtha cracking comprises the following steps:

1) establishing a production process monitoring model;

2) determining a control line for judging normal working conditions;

3) the monitoring system integrates the acquired real-time data of each monitoring variable into a production state index value; comparing the production state index value with the control line under the normal working condition, and if the production state index value is within the range of the control line under the normal working condition, the monitoring system does not act;

if the production state index value deviates from the control line of the normal working condition, the monitoring system continuously compares whether each subsequent production state index value in a set time period deviates from the control line of the normal working condition, if not,

the monitoring system does not report a production fault; if so,

the monitoring system judges whether the production working condition is changed manually or not, if so,

the monitoring system does not report a production fault, and if not,

and the monitoring system reports the production fault.

The method for establishing the production process monitoring model comprises the following steps:

1) setting monitoring variables in the production process of ethylene by naphtha cracking;

2) acquiring offline data of a plurality of complete historical operating cycles;

3) the method comprises the steps that offline data corresponding to the monitoring variable which runs stably for a first set time length in offline data of a complete historical running period are selected randomly;

counting the monitoring variable data of stable operation, acquiring a corresponding normal working condition control line, and establishing a monitoring model;

4) repeating the step 3) for N times, and establishing a set containing N monitoring models, wherein N is a natural number;

5) extracting any monitoring model in the monitoring model set, comparing the extracted monitoring model with the offline data of the complete historical operating period in the step 3), and judging:

if the monitoring accuracy rate of the stable operation of the monitoring model reaches a first set value and the false alarm rate is lower than a second set value, the monitoring model is reserved, otherwise, the monitoring model is abandoned;

6) extracting any monitoring model in the step 5), comparing the extracted monitoring model with the other offline data in the step 2), and judging:

if the monitoring accuracy rate of the stable operation of the monitoring model reaches a third set value and the false alarm rate is lower than a fourth set value, the monitoring model is reserved, otherwise, the monitoring model is abandoned;

7) and repeating the step 6), and obtaining a final better monitoring model from the set containing N monitoring models.

For the steam pyrolysis method, the monitoring variables are selected according to different types of the cracking furnace, including but not limited to: and (3) convection section: cracking raw material naphtha feed flow, cracking raw material naphtha feed temperature and pressure, dilution steam feed flow, dilution steam feed temperature, furnace tube cross section pressure, radiation section: one or more of furnace tube outlet temperature, furnace tube outlet pressure, furnace temperature, fuel gas flow and average molecular mass thereof.

The first set value and the third set value may be the same or different.

The second set value is the same as or different from the fourth set value.

And performing data preprocessing on the offline data between the step 2) and the step 3).

And the data preprocessing adopts a supplementary current value method or an estimation method.

And 7), if two or more than two final better monitoring models are obtained, sorting according to the third set value and the fourth set value in the step 6), and determining the highest sorting as the better monitoring model.

The invention has the beneficial effects that:

the technical scheme includes that an operation monitoring model used in actual production and a control line for judging normal working conditions are established by utilizing acquired offline data of a complete historical operation period in the process of preparing ethylene by cracking a plurality of naphthas, instant data of monitoring variables acquired by a monitoring system are synthesized into a production state index value through an algorithm and compared with the control line of the normal working conditions, if deviation occurs, subsequent production state index values are continuously compared, if the continuous time of the subsequent production state index values deviating from the control line of the normal working conditions is longer than set time, manual adjustment of the production working conditions is eliminated, the monitoring system reports production fault information, the accuracy rate of reporting faults is improved, and false reports are reduced; meanwhile, the problem that the main cause of the fault is difficult to determine when the fault occurs in process monitoring is solved.

Detailed Description

The technical solutions of the present invention are described in detail below by examples, and the following examples are only exemplary and can be used only for explaining and explaining the technical solutions of the present invention, but not construed as limiting the technical solutions of the present invention.

Theoretically, the process of preparing ethylene by cracking naphtha is a continuous chemical production process, and the stable production is kept under a preset production condition. In fact, the process for preparing ethylene by naphtha cracking is a process accompanied by slight fluctuation of production state and state adjustment. Meanwhile, the whole production process is controlled by a DCS monitoring system. When a certain production requirement is required to be met, the production working condition can be manually and actively adjusted.

The application provides a method for diagnosing production faults of ethylene prepared by naphtha cracking, which comprises the following steps:

1) establishing a production process monitoring model; in embodiments of the present application, a method for establishing a production process monitoring model includes:

(1) setting monitoring variables in the production process of ethylene by naphtha cracking; the variables selected and collected according to different types of the cracking furnaces in the production process of the cracking furnace of the steam pyrolysis method in the process technology for preparing ethylene by cracking the raw material of the actual naphtha include but are not limited to: and (3) convection section: 1. raw material part: the feed flow rate of the cracking raw material naphtha, the feed temperature and pressure of the cracking raw material naphtha and the feed flow rate of the cracking raw material naphtha are set values; 2. dilution steam part: set values of dilution steam feed flow, dilution steam feed temperature, and dilution steam feed quantity; 3. cross-section temperature and pressure part: the temperature of the cross section of the furnace tube and the pressure of the cross section of the furnace tube; a radiation section: 4. furnace exit temperature section: the radiant section furnace tube outlet temperature; 5. furnace exit pressure section: furnace tube outlet pressure; 6. cracking gas part: the contents of methane, ethane, ethylene, propane and propylene; 7. a hearth part: the composition of cracking gas, the temperature of a hearth, the composition of flue gas, the flow of fuel gas, the average molecular mass of the fuel gas and other variable data information.

In the embodiments of the present application, the purpose of the monitoring is to monitor the entire production process of the cracking furnace, maintaining a smooth and safe operation of the cracking process. The feedstock feed and equipment safety considerations are taken into account so selected monitoring variables include, but are not limited to, cracked feedstock naphtha feed flow, cracked feedstock naphtha feed temperature and pressure, dilution steam feed flow, dilution steam feed temperature, furnace cross-over pressure, radiant section furnace tube exit temperature, furnace exit pressure, furnace temperature and fuel gas flow and their average molecular mass as variable data information. In the actual modeling process, the fuel gas flow and the average molecular mass thereof are integrated into the heat supply of the fuel gas. In other embodiments of the present application, the number of monitoring variables may be adjusted according to actual needs.

(2) In other embodiments of the present application, the specific number of the acquired data of the complete historical operating cycles is determined according to needs, and may be more than five or less than five, but generally, the greater the number, the higher the accuracy of the final monitoring model is, but the greater the number is, the cost is increased.

The offline data correspond to the corresponding monitoring variables, and the interval of the offline data can be set according to needs, because the interval of the offline data may be different for different devices.

In the offline data acquired (one minute interval data), some time because of a failure of a sensor or monitoring device, the recorded data includes a non-numeric value or a zero value or other indication. When a non-numeric value appears in the data, it indicates that the sensor has failed to properly measure the variable. When a zero value appears, the signal acquisition cannot be completed. When the above-described situation occurs in the data, the variable value may be supplemented by a method similar to the current value method and the estimation method.

(3) The method comprises the steps that offline data corresponding to the monitoring variable which runs stably for a first set time length in offline data of a complete historical running period are selected randomly; the first set time length is generally the length including at least data information of selected monitoring variables according to needs, and can be selected from 5 minutes, 10 minutes, 15 minutes, 30 minutes, 1 hour, 2 hours, 4 hours, 8 hours, 24 hours, 3 days, seven days, repair months or other time lengths in between according to needs, which is mainly determined according to the length of one complete production cycle of the cracking furnace.

Counting the monitoring variable data of stable operation and acquiring a control line of a corresponding normal working condition; specifically, within a selected first set time length, corresponding data of each of the selected monitoring variables in a stable operation period is processed, so that a control line under normal conditions can be obtained, and for the purpose of clear and definite monitoring, the control line is usually determined in a control line manner to establish a monitoring model.

Because a plurality of relatively stable running time periods are included in a complete historical period, data of different relatively stable running time periods can be selected for statistics in a first set time length according to needs, and a plurality of monitoring models can be obtained.

(4) Repeating the step (3) for N times, and establishing a set containing N monitoring models, wherein N is a natural number;

(5) extracting any monitoring model in the monitoring model set, comparing the extracted monitoring model with the offline data of the complete historical operating period in the step (3), and judging:

if the stable operation monitoring accuracy of the monitoring model reaches the first set value, in this embodiment, the accuracy is set to 98%, in other embodiments of the present application, the accuracy may also be set as needed, such as 98.5%, 99%, and the like, and the false alarm rate is lower than the second set value, in this embodiment, the second set value is 1.56%, in other embodiments of the present application, other values, such as 1.52%, 1.50%, 1.48%, and the like, may also be selected as needed, and the monitoring model is retained, otherwise, the monitoring model is discarded;

(6) extracting any monitoring model in the step (5), comparing the monitoring model with the other offline data in the step (2), and judging:

if the monitoring accuracy rate of the stable operation of the monitoring model reaches a third set value and the false alarm rate is lower than a fourth set value, the monitoring model is reserved, otherwise, the monitoring model is abandoned;

(7) and (5) repeating the step (6) to obtain a final better monitoring model from the set containing N monitoring models.

And (7) if two or more final better monitoring models are obtained within the set times, sorting according to the third set value and the fourth set value in the step (6), and determining the highest sorting as the better monitoring models.

2) Determining a control line of a normal working condition according to the production process monitoring model; in this application, the control line for normal operation is a single value straight line. I.e. the value of the corresponding control line is the same at each point in time.

3) The monitoring system integrates the acquired instant data of each monitoring variable into a normal state index value; and comparing the normal state index value with the control line of the normal working condition, and if the normal state index value is in the range of the control line of the normal working condition, the monitoring system does not act.

If the normal state index value deviates from the control line of the normal working condition, the monitoring system continuously compares whether each subsequent normal state index value in a set time period deviates from the control line of the normal working condition, if not, namely when the monitoring system monitors that the normal state index value deviates, the monitoring system monitors the deviation immediately, if the length of the time can be set as required within a period of time, the normal state index values all deviate, and the monitoring system needs to judge whether the manual adjustment is carried out.

The monitoring system does not report a production fault; if so,

the monitoring system judges whether the production working condition is changed manually or not, if so,

the monitoring system does not report a production fault, and if not,

and the monitoring system reports the production fault.

The process of preparing ethylene by naphtha cracking is a very complex process of heat transfer and mass transfer, and each production equipment needs to be customized according to the production capacity and the production process. This allows each piece of equipment to remain stable within a relatively narrow operating range. When the off-line data of a complete historical operation period is used for process monitoring, the operation state of the equipment is changed, and the monitoring model is correspondingly adjusted, so that a better monitoring result is achieved.

In the whole monitoring process, information of a plurality of variables of each time point is integrated into a normal state index value to be used for reflecting the system operation condition, and when the normal state index values of a plurality of continuous time points exceed the control line of the normal working condition of the monitoring model, the cracking operation state is considered to be deviated (the condition that the normal state index values of one or two sample points exceed the control line of the normal working condition at times belongs to normal fluctuation of data), and then faults can occur.

Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.