Chemical device early warning model construction method, construction system and early warning system

Abstract

The invention belongs to the technical field of petrochemical engineering safety production, and discloses a chemical engineering device early warning model construction method and system and an early warning system.The method comprises the steps that a fitting data set of a chemical engineering device is obtained; processing the data in the fitting data set by using a smooth cubic spline fitting method to obtain an operation parameter curve function of a corresponding time period; operation parameters are obtained based on the operation parameter curve function, and the change rate of the operation parameters in the corresponding time period is obtained through calculation; and calculating the average deviation time required by the numerical value of the operation parameter in the corresponding time period to deviate to the dangerous critical value according to the change rate. The data collected by the DCS instrument is processed and analyzed, the early warning model is constructed in combination with multiple indexes, and real-time data is collected and judged through the early warning model, so that alarm is given for abnormal data, and the timeliness of alarm is improved.

Description

Technical Field

[0001] The invention belongs to the technical field of petrochemical production safety, and in particular relates to a chemical plant early warning model construction method, a construction system and an early warning system. Background Art

[0002] As an important part of the chemical industry, petrochemical industry is vital to the development of my country's national economy. Petrochemical equipment is prone to abnormal failures due to the flammability and explosiveness of the treatment medium, the harshness of the operating conditions and the complexity of the process. Once an abnormality occurs, it may cause the temperature and pressure to be out of control, causing risks such as leakage, fire and explosion, and in serious cases, it may cause major accidents. Ensuring the stability and reliability of petrochemical equipment during operation and issuing early warnings when abnormalities occur in the equipment to remind operators to take necessary inspections is the key to solving this problem. To determine whether the petrochemical equipment is stable and reliable during operation, its operating parameters can be used to qualitatively judge whether the equipment is operating normally. The greater the deviation of the real-time value of the operating parameter from the target value, the greater the deviation rate, and the more unstable the overall change trend, the less reliable the operation of the equipment.

[0003] Safety instrument systems represented by DCS are commonly used safety devices in petrochemical enterprises. They set the instrument alarm threshold according to the maximum acceptable deviation within the safety range, and issue an alarm when the real-time value of the parameter exceeds this threshold, so as to achieve the effect of early warning. However, the abnormal alarm of the DCS system is relatively simple. This threshold early warning method uses a single indicator, the advance time of abnormal alarm is limited, and the instrument monitors and records data with a time interval. The obtained operating parameter data is discrete, and the data between two monitorings is unknown. In this process, if the value exceeds the threshold, it will cause a late alarm. Summary of the invention

[0004] In order to solve at least one problem in the background technology, the present invention proposes a chemical plant early warning model construction method, construction system and early warning system.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] A method for constructing an early warning model for a chemical plant comprises the following steps:

[0007] Get the fitting data set of the chemical plant;

[0008] The data in the fitting data set are processed using a smooth cubic spline fitting method to obtain the operating parameter curve function of the corresponding time period;

[0009] Obtaining the operating parameters based on the operating parameter curve function, and calculating the change rate of the operating parameters in the corresponding time period;

[0010] Calculate the average deviation time required for the value of the operating parameter in the corresponding time period to deviate from the dangerous critical value according to the change rate;

[0011] Collect service life data of chemical equipment and construct failure probability distribution function of chemical equipment;

[0012] Calculate the mean time between failures of the device failure probability distribution function;

[0013] Establish the characterization relationship between the mean deviation time of operating parameters and the mean time between failures of chemical equipment;

[0014] Based on the representation relationship, an early warning model is constructed.

[0015] Preferably, obtaining a fitting data set of a chemical plant includes: obtaining real-time values ​​of operating parameters of the chemical plant through a DCS instrument to construct a fitting data set; the fitting data set includes the collected data of the DCS instrument at the current moment and the discrete data values ​​2 hours before the moment.

[0016] Preferably, obtaining the operating parameters based on the operating parameter curve function and calculating the change rate of the operating parameters in the corresponding time period include the following steps:

[0017] Derivative the running parameter curve function to obtain the derivative function;

[0018] Substituting the corresponding time period into the derivative function, the change rate of the operating parameters in the corresponding time period is obtained.

[0019] Preferably, calculating the average deviation time required for the value of the operating parameter in the corresponding time period to deviate from the dangerous critical value according to the change rate includes:

[0020]

[0021] Where Y 高 With Y 低 are the upper and lower limits of the critical value of danger, y[t0] is the operating parameter value at time t0; y'(t0) is the rate of change at time t0; t s is the average deviation time.

[0022] Preferably, collecting service life data of chemical equipment and constructing a failure probability distribution function of the chemical equipment comprises the following steps:

[0023] Collect service life data of chemical plants;

[0024] Eliminate abnormal data and obtain N groups of discrete life time data;

[0025] Calculate the failure probability value corresponding to each life time data;

[0026] Use the maximum likelihood estimation method to obtain unknown parameters of discrete sample sets under different distributions and construct distribution functions;

[0027] The KS test method is used to use discrete life time data to test whether the population obeys the fitted distribution function, and the optimal distribution function is selected as the failure probability distribution function of the chemical plant.

[0028] Preferably, calculating the mean time between failures of the device failure probability distribution function comprises:

[0029]

[0030] Where, t u is the mean time between failures, and F2(t) is the failure probability distribution function with independent variable t.

[0031] Preferably, if it is not possible to obtain sufficient life data to obtain the failure probability distribution function, the device is in an accidental failure period, and the failure probability distribution obeys the Weibull distribution:

[0032]

[0033] Where η is the size parameter of the distribution; F(t) is the failure probability distribution function that obeys the Weibull distribution with a shape parameter of 1; and t is time.

[0034] Preferably, establishing a characterization relationship between the average deviation time of the operating parameter and the mean time between failures of the chemical plant includes:

[0035]

[0036]

[0037] Where F1(T) is an early warning model with T as the independent variable, T is the early warning sensitivity; t u is the mean time between failures; t s is the average deviation time; F2(.) is the failure probability distribution fraction; z is the transition variable; y'(t0) is the rate of change at time t0; y[t0] is the operating parameter value at time t0; Y is the critical value of danger.

[0038] Preferably, based on the representation relationship, an early warning model is constructed, including:

[0039]

[0040] A chemical plant early warning model construction system, comprising:

[0041] An acquisition unit, used to obtain a fitting data set of a chemical plant;

[0042] A fitting unit is used to process the data in the fitting data set using a smooth cubic spline fitting method to obtain an operating parameter curve function of a corresponding time period;

[0043] A first calculation unit, configured to obtain an operating parameter based on an operating parameter curve function, and calculate a change rate of the operating parameter in a corresponding time period;

[0044] A second calculation unit is used to calculate the average deviation time required for the value of the operating parameter in the corresponding time period to deviate from the dangerous critical value according to the change rate;

[0045] A construction unit is used to collect service life data of chemical equipment and construct a failure probability distribution function of the chemical equipment;

[0046] A third calculation unit is used to calculate the mean time between failures of the device failure probability distribution function;

[0047] A characterization unit, used to establish a characterization relationship between the average deviation time of the operating parameters and the average time between failures of the chemical plant;

[0048] The definition unit is used to build an early warning model based on the representation relationship.

[0049] Preferably, the first computing unit comprises:

[0050] A derivative module is used to derive the running parameter curve function to obtain a derivative function;

[0051] The first calculation module is used to substitute the corresponding time period into the derivative function to obtain the change rate of the operating parameter of the corresponding time period.

[0052] Preferably, the building block comprises:

[0053] A collection module for collecting service life data of chemical equipment;

[0054] A module is proposed to eliminate abnormal data and obtain N sets of discrete life time data;

[0055] A processing module, used for calculating the failure probability value corresponding to each life time data;

[0056] A construction module is used to obtain unknown parameters under different distributions for discrete sample sets using the maximum likelihood estimation method and to construct a distribution function;

[0057] The selection module is used to use the KS test method to use discrete life time data to test whether the population obeys the fitted distribution function, and select the optimal distribution function as the failure probability distribution function of the chemical device.

[0058] A chemical plant early warning system, used for the above-mentioned early warning model, comprises:

[0059] The real-time acquisition, storage and update unit is used to acquire the monitoring data of the chemical plant by the DCS instrument in real time, and store the latest acquired data value in the fitting data set, and delete the old data value in real time as the new acquired data value is added, so that the total size of the fitting data set remains unchanged;

[0060] Data analysis and feature mining unit, used to analyze and mine feature parameters of change rate and change trend based on the fitted data set;

[0061] The real-time calculation and prediction unit of operation data is used to construct a real-time mathematical model using characteristic parameters and calculate the real-time warning value of the device operation;

[0062] The operation abnormality real-time warning module is used to compare the real-time warning value with the warning threshold and decide whether to output a warning signal based on the comparison result;

[0063] The image drawing module is used to dynamically draw the warning value-warning time image in real time to intuitively show the changes in the risk value of the operating parameters.

[0064] Beneficial effects of the present invention:

[0065] 1. The present invention processes and analyzes the data collected by DCS instruments, combines multiple indicators to build an early warning model, collects and judges real-time data through the early warning model, and then issues an alarm for abnormal data, thereby improving the timeliness of the alarm.

[0066] 2. The present invention can trigger an alarm before an abnormality occurs in the operation of a petrochemical device, reminding the operating personnel to take necessary inspections and measures in a timely manner, thereby avoiding the occurrence of danger; it has better performance in early warning of operating parameters under various change trends such as oscillation, step, and gentle changes than the DCS instrument system, overcoming the disadvantage of limited abnormal handling time after a single threshold alarm of the instrument system, and providing support for early warning and dynamic prevention and control of abnormalities in petrochemical devices.

[0067] Other features and advantages of the present invention will be described in the following description, and partly become obvious from the description, or be understood by implementing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures pointed out in the description and the drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0069] Figure 1 A flow chart showing a method for constructing an early warning model for a chemical plant according to the present invention is shown;

[0070] Figure 2 The structural diagram of the early warning mathematical model of the present invention is shown;

[0071] Figure 3 The present invention shows a flow chart for obtaining a device failure probability function;

[0072] Figure 4 The operating structure diagram of the early warning system of the present invention is shown;

[0073] Figure 5a The normal operation data diagram of the reflux tank pressure of the stabilizing tower is shown;

[0074] Figure 5b The debug diagram of the normal operation of the stabilizing tower reflux tank with a sensitivity of 3 is shown;

[0075] Figure 5c The debug diagram of the normal operation of the stabilizing tower reflux tank with a sensitivity of 5 is shown;

[0076] Figure 5d The debug diagram of the normal operation of the stabilizing tower reflux tank with a sensitivity of 7 is shown;

[0077] Figure 6a The abnormal operation data diagram of the reflux tank pressure of the stabilizing tower is shown;

[0078] Figure 6b An early warning diagram of an abnormally operating stabilizing tower reflux tank with a sensitivity of 3 is shown;

[0079] Figure 6c An early warning diagram of an abnormally operating stabilizing tower reflux tank with a sensitivity of 5 is shown;

[0080] Figure 6d An early warning diagram showing an abnormally operating stabilizing tower reflux drum with a sensitivity of 7 is shown. DETAILED DESCRIPTION

[0081] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0082] A method for constructing an early warning model for a chemical plant, such as Figure 1As shown, the following steps are included: S1: obtaining a fitting data set of a chemical device; S2: processing the data in the fitting data set using a smooth cubic spline fitting method, and fitting each segment using a third-order polynomial to obtain an operating parameter curve function of the corresponding time period, which is specifically:

[0083] For a set of discrete data points (t j ,y j ), where a = t1 <t2<…t j <t j+1 <…t n = b, then the time interval can be divided into n-1 segmented sub-intervals using discrete data points. j ,t j +1](j=1~n-1), if a third-order polynomial is used to fit each segment, it can be expressed as the following formula:

[0084] S j (t) = A(tt j ) 3 +B(tt j ) 2 +C(tt j )+D t j ≤t≤t j+1

[0085] In the formula, A, B, C, and D are the coefficients of the polynomial. The fitting process is the process of obtaining the polynomial coefficients. t is the time, t j It represents the horizontal coordinate of point j and also represents the time data corresponding to the point.

[0086] The fitted n-1 segment curve S j (t)(j=1~n-1) needs to satisfy the continuity of first and second order derivatives at the boundary, as follows:

[0087] S j ′(t j+1 -)=S j+1 ′(t j+1 +)S j ′(t j+1 -)=S j+1 ″(t j+1 +);

[0088] The overall curve S(t) formed by connecting the third-order polynomial fitting curves that meet the above conditions is called the cubic spline fitting curve on the interval [a, b].

[0089]

[0090] In order to increase the smoothness of the fitting curve, a smoothing factor λ is introduced on the basis of cubic spline fitting to form the following constraint conditions. The RSS value is minimized by finding the coefficients of the polynomial. This method is called smooth cubic spline fitting, and finally we get:

[0091]

[0092] In the formula, λ is the smoothing factor; RSS is the sum of squares of a set of data.

[0093] S3: Obtain the operating parameters based on the operating parameter curve function, and calculate the change rate of the operating parameters in the corresponding time period; S4: Calculate the average deviation time required for the numerical value of the operating parameter in the corresponding time period to deviate from the dangerous critical value according to the said change rate; S5: Collect the service life data of the chemical equipment and construct the failure probability distribution function of the chemical equipment; S6: Calculate the average failure interval time of the equipment failure probability distribution function; S7: Establish a characterization relationship between the average deviation time of the operating parameters and the average failure interval time of the chemical equipment; S8: Based on the characterization relationship, construct an early warning model.

[0094] It should be noted that in step S1, it is necessary to obtain the real-time values ​​of the operating parameters of the chemical plant through the DCS instrument, such as the temperature, pressure, liquid level and other data of the key monitoring parts, to construct a fitting data set; the fitting data set includes the collected data of the DCS instrument at the current moment and the discrete data values ​​within 2 hours before the moment. In addition, in step S3, the curve is used to obtain the operating characteristic parameters, including obtaining the change trend of the parameters in the time period by deriving the curve function, and then obtaining the change rate of the operating parameters at the previous moment by bringing the current time into the wizard function. In step S5, if it is impossible to obtain sufficient life data of the relevant device to obtain the failure probability distribution function of the device, it can be considered that the relevant device is in an accidental failure period, and the failure probability distribution obeys the Weibull distribution with a shape parameter of 1. In step S8, characteristic parameters such as the rate of change and the deviation are integrated into the failure probability distribution function of the device to construct an early warning mathematical model.

[0095] like Figure 2 As shown in the figure, the early warning model is mainly composed of two parts: the failure probability distribution function of the chemical plant and the characteristic parameters of the operating parameters. The operating parameter feature extraction can use the smooth cubic spline fitting method to construct a fitting sample set by collecting discrete operating data values ​​y[t] within a period of time including the current moment from the DCS instrument. The smooth cubic spline fitting method is used to process it and obtain the parameter change curve y(t) for this period of time.

[0096] Furthermore, combined with Figure 2In step S3, the following steps are included: S301: deriving the operating parameter curve function to obtain a derivative function; S302: substituting the corresponding time period into the derivative function to obtain the change rate of the operating parameter in the corresponding time period.

[0097] It should be noted that in step S3, the overall change trend of the parameters in the time period can be obtained by taking the derivative function of y(t); the deviation rate y'(t0) of the operating parameters at the current moment is obtained by substituting the current time t0 into the derivative function y'(t).

[0098] Furthermore, combined with Figure 2 In step S4, if the current operating parameter has a trend of further causing the parameter value to deviate from the target value, the average deviation time t required for the current parameter value to deviate from the dangerous critical value at this change rate is calculated. s .

[0099]

[0100] Where Y 高 With Y 低 are the upper and lower limits of the critical value of danger, which are also the upper and lower limits of the alarm of the DCS instrument system; y[t0] is the operating parameter value at time t0; y′(t0) is the rate of change at time t0; t s is the average deviation time.

[0101] Furthermore, combined with Figure 3 It can be seen that in step S5, the following steps are included:

[0102] S501: Collect service life data of chemical equipment;

[0103] S502: Eliminate abnormal data to obtain N groups of discrete life time data, denoted as t i (i=1~N);

[0104] S503: Calculate the failure probability value F corresponding to each life time data N (t i );

[0105] S504: Use the maximum likelihood estimation method to obtain unknown parameters of discrete sample sets under different distributions and construct a distribution function; specifically, different distributions include Weibull distribution, exponential distribution, normal distribution, lognormal distribution, extreme value distribution, beta distribution and other commonly used distribution models.

[0106] S505: Using the KS test method, using discrete life time data to test whether the population obeys the fitted distribution function F(t), and selecting the optimal distribution function as the failure probability distribution function F2(t) of the chemical plant.

[0107] It should be noted that in step S503, the failure probability value F N (t i ) satisfies the following relationship:

[0108]

[0109] Where N is the total number of discrete life time data, N f (t i ) is t≤t i The number of life span data.

[0110] In addition, the discrete sample set in step S504 is recorded as (ti, FN(ti)).

[0111] It should be noted that if Figure 2 As shown, in step S505, if the failure probability distribution function F2(t) exists, it is necessary to compress the horizontal coordinate of F2(t) by t u / ts times, and then obtain the failure probability function F1(T) of the parameter as the early warning model of the present invention, T is the early warning sensitivity; if it is impossible to obtain sufficient life data of the relevant device to obtain the failure probability distribution function of the device, it can be considered that the relevant device is in the accidental failure period, and the failure probability distribution function obeys the Weibull distribution with a shape parameter of 1, where:

[0112]

[0113] Where η is the size parameter of the distribution and t is the time.

[0114] Furthermore, combined with Figure 2 , calculate the mean time between failures of the device failure probability distribution function, including:

[0115]

[0116] Where, t u is the mean time between failures, and F2(t) is the failure probability distribution function with independent variable t.

[0117] It should be noted that, combined with Figure 2 For the early warning model F1(t) proposed in the present invention, its mean time between failures MTBF is equivalent to the mean deviation time t s ,in:

[0118]

[0119] Where F1(t) is the early warning model with t as the independent variable, and t is time.

[0120] Furthermore, a characterization relationship between the average deviation time of the operating parameters and the mean time between failures of the chemical plant is established, including:

[0121]

[0122]

[0123] In the formula, F1(T) is an early warning model with T as the independent variable. The input value of T can return the failure probability function value of the current operating parameters. The larger the T value, the larger the returned value. Therefore, T is regarded as the early warning sensitivity; t u is the mean time between failures; t s is the average deviation time; F2(.) is the failure probability distribution function; z is the transition variable; y′(t0) is the rate of change at time t0; y[t0] is the operating parameter value at time t0; Y is the critical value of danger.

[0124] Furthermore, based on the characterization relationship, an early warning model is constructed. Optionally, when the device failure probability distribution obeys the Weibull distribution with a shape parameter of 1, the early warning mathematical model F1(T) also obeys the Weibull distribution with a shape parameter of 1:

[0125]

[0126] A chemical plant early warning model construction system includes an acquisition unit, a fitting unit, a first calculation unit, a second calculation unit, a third calculation unit, a characterization unit and a definition unit. The acquisition unit is used to obtain a fitting data set of the chemical plant; the fitting unit is used to process the data in the fitting data set using a smooth cubic spline fitting method to obtain an operating parameter curve function of a corresponding time period; the first calculation unit is used to obtain operating parameters based on the operating parameter curve function and calculate the change rate of the operating parameters of the corresponding time period; the second calculation unit is used to calculate the average deviation time required for the value of the operating parameter in the corresponding time period to deviate from the dangerous critical value according to the change rate; the construction unit is used to collect the service life data of the chemical plant and construct the failure probability distribution function of the chemical plant; the third calculation unit is used to calculate the average failure interval time of the device failure probability distribution function; the characterization unit is used to establish a characterization relationship between the average deviation time of the operating parameter and the average failure interval time of the chemical plant; and the definition unit is used to construct an early warning model based on the characterization relationship.

[0127] Furthermore, the first computing unit includes:

[0128] A derivative module is used to derive the running parameter curve function to obtain a derivative function;

[0129] The first calculation module is used to substitute the corresponding time period into the derivative function to obtain the change rate of the operating parameter of the corresponding time period.

[0130] Furthermore, the construction unit comprises:

[0131] A collection module for collecting service life data of chemical equipment;

[0132] A module is proposed to eliminate abnormal data and obtain N sets of discrete life time data;

[0133] A processing module, used for calculating the failure probability value corresponding to each life time data;

[0134] A construction module is used to obtain unknown parameters under different distributions for discrete life time data using the maximum likelihood estimation method and to construct a distribution function;

[0135] The selection module is used to use the KS test method to use discrete life time data to test whether the population obeys the fitted distribution function, and select the optimal distribution function as the failure probability distribution function of the chemical device.

[0136] It should be noted that, for the system embodiment, since it basically corresponds to the method embodiment, the relevant parts can refer to the partial description of the method embodiment. The various units and modules of the system of the present invention are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be realized; in addition, the specific names of the units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention.

[0137] like Figure 4 As shown in the figure, the early warning system built based on the early warning model mainly includes: real-time acquisition, storage and update unit, data analysis and feature mining unit, real-time calculation and prediction unit of operation data, real-time early warning unit of operation abnormality, and image drawing unit. The real-time acquisition, storage and update module is used to obtain the latest monitoring data of the device by the DCS instrument in real time, and store the latest acquired data value in the last bit of the fitting data set Y. At the same time, as the new collected data value is added, the first data value of the fitting data set Y is deleted, so as to ensure that the total size of the fitting data set remains unchanged and is equal to the set fitting data set size.

[0138] In particular, if the number of data in the fitting data set is less than the set fitting data set size, the system only stores the latest collected data in the last position of the fitting data set Y without deleting any first position data; optionally, 10% of the total number of debugging data is selected to determine the size of the fitting data set.

[0139] The data analysis and feature mining module is used to analyze and mine the characteristic parameters of its change rate and change trend; smooth cubic spline fitting is performed on the fitting data set Y, and its change trend, change rate and other information are obtained by taking the derivative, and the average deviation time ts is calculated according to the method proposed above.

[0140] The real-time calculation and prediction module of operating data is used to construct a real-time mathematical model using characteristic parameters and calculate the abnormal warning value of the device operation; using the above-mentioned warning model construction method and formula, the corresponding warning mathematical model F1(T) is constructed in real time, and the set warning sensitivity value T is substituted into the function F1(T), and the real-time warning value is calculated and stored in the warning value set F; in particular, since the fitting data set Y is collected, stored and updated in real time as the device is running, the corresponding constructed warning mathematical model is also updated in real time. Optionally, using the data of the monitored device under historical normal operating conditions, input and adjust the value of increasing the warning sensitivity T so that the warning value is basically controlled within 0.4, and the warning sensitivity value T is determined, as shown in the attached. Figure 5a to Figure 5d As shown, it can be seen that the reflux tank pressure is not stable and there is a certain fluctuation during normal operation. When the sensitivity is set to 3, 5, and 7 respectively, the return warning value can meet the normal situation. Less than 0.4, due to the pressure fluctuations, the warning value will increase rapidly when the parameter shows a rapid upward or downward trend. Therefore, the warning sensitivity of the stabilization tower reflux tank pressure parameter monitoring is determined to be 7.

[0141] The real-time warning module for abnormal operation is used to compare the real-time warning value with the warning threshold, and decide whether to output the warning signal according to the comparison result; optionally, 0.8 is used as the warning threshold of the system, and an alarm is issued if the warning value is greater than the set warning threshold; optionally, a secondary warning is selectively added according to the inherent risk value of the important device directly affected by the monitored device, and each inherent risk indicator is first qualitatively analyzed and then quantified, and the inherent risk value R of the important device is determined by using the percentage scoring method;

[0142]

[0143] The above formula is to obtain the second-level warning threshold. If the warning value is greater than the second-level warning threshold but less than 0.8, a second-level warning is issued. Here, F lim It is the second-level warning threshold.

[0144] The image drawing module is used to dynamically draw the image of the warning value-warning time calculated in the real-time calculation and prediction module of the above-mentioned operation data in real time, and intuitively show the changes in the risk value of the operation parameters.

[0145] like Figure 6a to Figure 6dAs shown in the figure, the leakage at the connection of the simulated tank body caused the pressure to decrease abnormally slowly. The DCS instrument system alarmed at the 990th minute because the pressure in the tank was lower than the lower threshold. When the warning sensitivity was set to 3, 5, and 7, the warning system gave warnings at the earliest at the 988th, 968th, and 900th minutes, respectively. Compared with the original instrument system alarm, the warnings were 2, 22, and 90 minutes earlier, respectively. This shows that the higher the warning sensitivity, the more sensitive the warning value is to the fluctuation of the parameter value, and the increase in the abnormal fluctuation trend of the operating parameters is more likely to trigger a warning. The pressure data of the reflux tank of the stabilizing tower under normal operating conditions was used to debug the warning system under different warning sensitivities. When the abnormal pressure of the reflux tank was predicted in real time, the alarm time was advanced compared with the DCS instrument. When the warning sensitivity was set to 7, the warning advance effect was optimal.

[0146] Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent substitutions for some of the technical features therein; and these modifications or substitutions do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for constructing an early warning model for a chemical plant, characterized in that: The following steps are involved: Get the fitting data set of the chemical plant; The data in the fitting data set are processed using a smooth cubic spline fitting method to obtain the operating parameter curve function of the corresponding time period; Obtaining the operating parameters based on the operating parameter curve function, and calculating the change rate of the operating parameters in the corresponding time period; Calculate the average deviation time required for the value of the operating parameter in the corresponding time period to deviate from the dangerous critical value according to the change rate; Collect service life data of chemical equipment and construct failure probability distribution function of chemical equipment; Calculate the mean time between failures of the device failure probability distribution function; Establish the characterization relationship between the mean deviation time of operating parameters and the mean time between failures of chemical equipment; Based on the representation relationship, an early warning model is constructed.

2. A method for constructing an early warning model for a chemical plant according to claim 1, characterized in that: Acquiring a fitting data set of a chemical plant includes: acquiring real-time values ​​of operating parameters of the chemical plant through a DCS instrument, and constructing a fitting data set; the fitting data set includes collected data of the DCS instrument at the current moment and discrete data values ​​2 hours before the moment.

3. A method for constructing an early warning model for a chemical plant according to claim 1, characterized in that: Obtaining the operating parameters based on the operating parameter curve function and calculating the change rate of the operating parameters in the corresponding time period include the following steps: Derivative the running parameter curve function to obtain the derivative function; Substituting the corresponding time period into the derivative function, the change rate of the operating parameters in the corresponding time period is obtained.

4. A method for constructing an early warning model for a chemical plant according to claim 1, characterized in that: Calculating the average deviation time required for the value of the operating parameter in the corresponding time period to deviate from the dangerous critical value according to the change rate, including: Where Y 高 With Y 低 are the upper and lower limits of the critical value of danger, y[t0] is the operating parameter value at time t0; y'(t0) is the rate of change at time t0; t s is the average deviation time.

5. A method for constructing an early warning model for a chemical plant according to claim 1, characterized in that: Collecting service life data of chemical equipment and constructing failure probability distribution function of chemical equipment includes the following steps: Collect service life data of chemical plants; Eliminate abnormal data and obtain N groups of discrete life time data; Calculate the failure probability value corresponding to each life time data; Use the maximum likelihood estimation method to obtain unknown parameters of discrete sample sets under different distributions and construct distribution functions; The KS test method is used to use discrete life time data to test whether the population obeys the fitted distribution function, and the optimal distribution function is selected as the failure probability distribution function of the chemical plant.

6. A method for constructing an early warning model for a chemical plant according to claim 5, characterized in that: Calculate the mean time between failures of the device failure probability distribution function, including: Where, t u is the mean time between failures, and F2(t) is the failure probability distribution function with independent variable t.

7. A method for constructing an early warning model for a chemical plant according to claim 5, characterized in that: If it is not possible to obtain sufficient life data to obtain the failure probability distribution function, the device is in the accidental failure period, and the failure probability distribution follows the Weibull distribution: Where η is the size parameter of the distribution; F(t) is the failure probability distribution function that obeys the Weibull distribution with a shape parameter of 1; and t is time.

8. A method for constructing an early warning model for a chemical plant according to claim 5, characterized in that: Establish the characterization relationship between the mean deviation time of the operating parameters and the mean time between failures of the chemical plant, including: Where F1(T) is an early warning model with T as the independent variable, T is the early warning sensitivity; t u is the mean time between failures; t s is the average deviation time; F2(.) is the failure probability distribution fraction; z is the transition variable; y'(t0) is the rate of change at time t0; y[t0] is the operating parameter value at time t0; Y is the critical value of danger.

9. A method for constructing an early warning model for a chemical plant according to claim 8, characterized in that: Based on the representation relationship, an early warning model is constructed, including:

10. A chemical plant early warning model construction system, characterized in that: include: An acquisition unit, used to obtain a fitting data set of a chemical plant; A fitting unit is used to process the data in the fitting data set using a smooth cubic spline fitting method to obtain an operating parameter curve function of a corresponding time period; A first calculation unit, configured to obtain an operating parameter based on an operating parameter curve function, and calculate a change rate of the operating parameter in a corresponding time period; A second calculation unit is used to calculate the average deviation time required for the value of the operating parameter in the corresponding time period to deviate from the dangerous critical value according to the change rate; A construction unit is used to collect service life data of chemical equipment and construct a failure probability distribution function of the chemical equipment; A third calculation unit is used to calculate the mean time between failures of the device failure probability distribution function; A characterization unit, used to establish a characterization relationship between the average deviation time of the operating parameters and the average time between failures of the chemical plant; The definition unit is used to build an early warning model based on the representation relationship.

11. A chemical plant early warning model construction system according to claim 10, characterized in that: The first computing unit comprises: A derivative module is used to derive the running parameter curve function to obtain a derivative function; The first calculation module is used to substitute the corresponding time period into the derivative function to obtain the change rate of the operating parameter of the corresponding time period.

12. A chemical plant early warning model construction system according to claim 10, characterized in that: The building block comprises: A collection module for collecting service life data of chemical equipment; A module is proposed to eliminate abnormal data and obtain N sets of discrete life time data; A processing module, used for calculating the failure probability value corresponding to each life time data; A construction module is used to obtain unknown parameters under different distributions for discrete sample sets using the maximum likelihood estimation method and to construct a distribution function; The selection module is used to use the KS test method to use discrete life time data to test whether the population obeys the fitted distribution function, and select the optimal distribution function as the failure probability distribution function of the chemical device.

13. A chemical plant early warning system, used for the early warning model according to any one of claims 1 to 9, characterized in that: include: The real-time acquisition, storage and update unit is used to acquire the monitoring data of the chemical plant by the DCS instrument in real time, and store the latest acquired data value in the fitting data set, and delete the old data value in real time as the new acquired data value is added, so that the total size of the fitting data set remains unchanged; Data analysis and feature mining unit, used to analyze and mine feature parameters of change rate and change trend based on the fitted data set; The real-time calculation and prediction unit of operation data is used to construct a real-time mathematical model using characteristic parameters and calculate the real-time warning value of the device operation; The operation abnormality real-time warning module is used to compare the real-time warning value with the warning threshold and decide whether to output a warning signal based on the comparison result; The image drawing module is used to dynamically draw the warning value-warning time image in real time to intuitively show the changes in the risk value of the operating parameters.

Images