Refining device parameter control method, device and equipment

Abstract

The invention relates to a refining device parameter control method, device and equipment. According to the main technical scheme, the method comprises the steps of obtaining target parameter data and process operation data of a refining device within a preset time, determining a steady-state time period in which process operation parameters influencing target parameters are in a steady state according to the target parameter data and the process operation data, and determining the target parameters according to the steady-state time period. Constructing a state space model for evaluating the influence of the process operation parameters on the target parameters, determining the delay time of the process operation parameters according to the target parameter data, the process operation data and the state space model, and determining the device control parameters of the refining device according to the delay time of the process operation parameters and the state space model, and the rationality of the state space model and the rationality of the device control parameters are verified, so that the influence of the process operation parameters on the target parameters can be accurately evaluated, and the evaluation cost is reduced.

Description

Technical Field

[0001] The present application relates to the field of industrial production technology, and in particular to a method, device and equipment for controlling parameters of a refining unit. Background Art

[0002] Oil refining is a complex process, which mainly involves physical and chemical processing of crude oil to produce gasoline, kerosene, diesel and various basic petrochemical products, involving atmospheric and vacuum distillation, catalytic cracking, delayed coking, continuous reforming, hydrogenation, hydrogen production, ethylene cracking, aromatics extraction, isomerization and other units. In the process of transformation and upgrading, traditional refineries need to cope with complex and changing production needs and improve the automation level, production efficiency and product quality of refining units.

[0003] Multivariable Predictive Control (MPC) can improve the conversion rate of raw materials and product yield by accurately controlling the key parameters that affect product quality in the refining unit, reducing fluctuations in production, and at the same time, MPC can predict and compensate for external disturbances and internal changes in the system. In order to enable the control system of the refining unit to predict and learn the impact of process operating parameters on target parameters, a large amount of sample data is required, and in the actual production environment, changes in the properties of raw materials may cause the model parameters of the control system to shift to varying degrees, which not only makes it difficult for the control system to accurately evaluate the impact of process operating parameters on target parameters, but also has high evaluation costs and cannot accurately control the relevant parameters in the refining unit. Summary of the invention

[0004] Based on this, the present application provides a parameter control method, device and equipment for a refining unit to solve the technical problems that it is difficult for the control system to accurately evaluate the impact of process operation parameters on target parameters and the evaluation cost is high.

[0005] In a first aspect, a method for controlling parameters of a refining device is provided, the method comprising:

[0006] Obtaining target parameter data and process operation data of the refining device within a preset time, wherein the target parameter data and the process operation data are both time series data, the target parameter data includes target parameters, and the process operation data includes process operation parameters;

[0007] Determining a steady-state time period during which a process operation parameter affecting the target parameter is in a steady state according to the target parameter data and the process operation data;

[0008] Based on the steady-state time period, a state space model is constructed to evaluate the impact of process operating parameters on target parameters;

[0009] Determine the delay time of the process operation parameter according to the target parameter data, the process operation data and the state space model;

[0010] Determine the plant control parameters of the refinery based on the delay time and state space model of the process operation parameters;

[0011] Based on the target parameter data, process operation data, delay time, state space model and device control parameters of the refining unit, the rationality of the state space model and the rationality of the device control parameters are verified.

[0012] According to an achievable method in an embodiment of the present application, determining a steady-state time period during which a process operation parameter affecting the target parameter is in a steady state according to the target parameter data and the process operation data includes:

[0013] According to the process operation data, the outlier time point of the process operation parameter is determined based on a preset sliding window, wherein the determination formula of the outlier time point is expressed as follows:

[0014]

[0015] Among them, w i,t represents the value of process operating parameter i at time t, represents the average value of process operation parameter i in the time period [t-δ, t-1], represents the average value of process operation parameter i in the time period [t+δ, t+1], σ(w i,t-δ:t-1 ) represents the standard deviation of process operation parameter i in the time period [t-δ, t-1], σ(w i,t+δ:t+1 ) represents the standard deviation of the process operation parameter i in the time period [t+δ, t+1], δ represents the length of the preset sliding window;

[0016] Determine the effective time period in the process operation data according to the outlier time point of the process operation parameter;

[0017] According to the target parameter data and the process operation data, a steady-state time period in a steady state within an effective time period affecting the target parameter is determined.

[0018] According to an achievable method in an embodiment of the present application, determining a steady-state time period during which a process operation parameter affecting the target parameter is in a steady state according to the target parameter data and the process operation data includes:

[0019] According to the target parameter data and process operation data, determine the set of change points where the process operation parameters jump Satisfy t 1 <t 2 <..., based on a preset time interval δ jDetermine the steady-state point corresponding to each change point in the change point set and express it as t j +δ j , the steady-state point satisfies t j +δ j <t j+1 ;

[0020] According to the steady-state point corresponding to each change point in the change point set, an initial steady-state point set of the process operation parameters is determined based on a first-order autoregressive model;

[0021] If the time distance between the current steady-state point and the next steady-state point in the initial steady-state point set is less than the preset time interval, the current steady-state point is removed; if the time distance between the current steady-state point and the next steady-state point in the initial steady-state point set is greater than the preset time interval, the current steady-state point and the next steady-state point are retained to obtain the final steady-state point set;

[0022] According to the final steady-state point set, a steady-state time period in which the process operating parameters affecting the target parameters are in a steady state is determined.

[0023] According to an achievable method in an embodiment of the present application, a state space model for evaluating the influence of process operation parameters on target parameters is constructed based on a steady-state time period, including:

[0024] The state space model is expressed as:

[0025]

[0026] Among them, y t is the target parameter, which is obtained by summing all hidden processes and adding noise. is the reference point of the target parameter, obtained from the previous steady-state point before time t, that is, is the final steady-state point set, x i,t represents the hidden process affected by the process operation parameter i at time t, which is an unknown variable, x i,t-k represents the hidden process affected by process operation parameter i at time tk, k represents the number of past time points, and w i,t-θi represents the value of process parameter i at time t-θi, It represents the reference point of process parameter i at time t-θi, which is obtained from the previous steady-state point before time t, that is, θi represents the delay value of the i-th process operation parameter, d is the number of process operation parameters i, α i,k ,β i ,c i is the parameter of the autoregressive model, ∈ i，t , ε t They are the noise of the hidden process and the observation noise, respectively, with mean 0 and variance ω 2, σ 2 The normal distribution of 2 According to the process parameters w i,t-θi Calculate the variance, σ 2 is based on the hidden process x i,t Calculate the variance; x d+1,t , x d+1,t-1 The influencing factors of the unknown variables hidden in the raw material properties at time t and time t-1, is the magnitude of its mean shift, x i,0 is the initial hiding process, μ i，0 For x i,t The mean of .

[0027] According to an achievable method in an embodiment of the present application, determining the delay time of the process operation parameter according to the target parameter data, the process operation data and the state space model includes:

[0028] According to the process operation data, the delay upper bound and delay lower bound of the process operation parameters are set, which are expressed as Satisfy 0≤θ i，min ≤θ i，max ;

[0029] Based on the delay upper bound, delay lower bound and preset delay interval g i , determine the delay time candidate set of process operation parameters

[0030] Based on the state space model and the delay time candidate set, the mean absolute effect value of each variable in the process operation parameter is calculated by the following formula:

[0031]

[0032] in, is the estimated value of the state space model at time t+s given time 1 to t, For the state space model, excluding the current variable w′ at given time 1 to t i After the influence of , the estimated value at time t+s, that is, the variable values ​​corresponding to the period [t+1, t+s] are all 0, s is the pre-selected step length, and T is the time length of the target parameter data and process operation data;

[0033] The delay time corresponding to the variable with the largest mean absolute effect value is determined as the delay time of the process operation parameter.

[0034] According to an achievable method in an embodiment of the present application, determining the delay time of the process operation parameter according to the target parameter data, the process operation data and the state space model includes:

[0035] The state space model is used as the proxy model, and the root mean square error between the predicted value of the preset future time length of the target parameter and the true value is used as the objective function. The objective function is optimized through the proxy model to obtain the objective function value;

[0036] Based on historical production data, determine the search interval of the delay time of the impact of process operation parameters on target parameters, and perform Bayesian optimization iteration on the delay time in the search interval;

[0037] When the delay time reaches a certain number of iterations or the objective function value is less than a preset threshold, the delay time of the process operation parameter corresponding to the minimum value of the objective function is selected as the delay time of the process operation parameter.

[0038] According to an achievable method in an embodiment of the present application, the device control parameters of the refining device are determined according to the delay time and the state space model of the process operation parameters, including:

[0039] Convert the state space model to a differential equation model;

[0040] Calculate the first and second derivative functions of the state-space model;

[0041] Substitute the first-order derivative function and the second-order derivative function into the differential equation model to obtain the target differential equation model;

[0042] According to the state space model and the target differential equation model, adjusting the parameters of the state space model;

[0043] According to the parameters of the state space model, the device control parameters of the refining device are determined. The device control parameters are expressed as follows:

[0044]

[0045] Among them, K i Indicates the influence of process operation parameters on target parameters. If K i >0 means there is a positive impact, K i <0 means there is a negative impact, τ i and i It indicates how fast the process parameters affect the target parameters, β i , α i,1 , α i,2 is the parameter of the autoregressive model, and Δ represents the sampling interval.

[0046] According to an achievable method in an embodiment of the present application, the rationality of the state space model and the rationality of the device control parameters are verified based on the target parameter data, the process operation data, the state space model, the delay time and the device control parameters of the refining device, including:

[0047] Predict the time series values ​​of target parameters in future time periods based on process operation data, state space model, and delay time;

[0048] The rationality of the state space model and the rationality of the device control parameters are verified based on the error and trend between the time series value and the true value.

[0049] In a second aspect, a parameter control device for a refining device is provided, the device comprising:

[0050] An acquisition module is used to acquire target parameter data and process operation data within a preset time of the refining device, wherein the target parameter data and the process operation data are both time series data, the target parameter data includes target parameters, and the process operation data includes process operation parameters;

[0051] A determination module, for determining a steady-state time period during which a process operation parameter affecting the target parameter is in a steady state according to the target parameter data and the process operation data;

[0052] A building module for building a state space model for evaluating the influence of process operating parameters on target parameters based on a steady-state time period;

[0053] The determination module is further used to determine the delay time of the process operation parameter according to the target parameter data, the process operation data and the state space model;

[0054] The determination module is further used to determine the device control parameters of the refinery device according to the delay time and the state space model of the process operation parameters;

[0055] The verification module is used to verify the rationality of the state space model and the rationality of the device control parameters according to the target parameter data, process operation data, delay time, the state space model and the device control parameters of the refining device.

[0056] In a third aspect, a computer device is provided, comprising:

[0057] at least one processor; and

[0058] a memory communicatively connected to the at least one processor; wherein,

[0059] The memory stores computer instructions that can be executed by the at least one processor, and the computer instructions are executed by the at least one processor to enable the at least one processor to perform the method involved in the first aspect above.

[0060] In a fourth aspect, a computer-readable storage medium is provided, on which computer instructions are stored, characterized in that the computer instructions are used to enable a computer to execute the method involved in the first aspect above.

[0061] According to the technical content provided in the embodiments of the present application, by acquiring target parameter data and process operation data within a preset time of a refining and chemical device, wherein the target parameter data and the process operation data are both time series data, the target parameter data include target parameters, and the process operation data include process operation parameters, a steady-state time period in which the process operation parameters affecting the target parameters are in a steady state is determined according to the target parameter data and the process operation data, based on the steady-state time period, a state space model for evaluating the influence of the process operation parameters on the target parameters is constructed, according to the target parameter data, the process operation data and the state space model, a delay time of the process operation parameters is determined, according to the delay time of the process operation parameters and the state space model, the device control parameters of the refining and chemical device are determined, according to the target parameter data, the process operation data, the delay time, the state space model and the device control parameters of the refining and chemical device, the rationality of the state space model and the rationality of the device control parameters are verified, so that the influence of the process operation parameters on the target parameters can be accurately evaluated, and the evaluation cost can be reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 A schematic diagram of a process flow of a method for controlling parameters of a refinery in one embodiment;

[0063] Figure 2 It is a structural block diagram of a parameter control device for a refinery in one embodiment;

[0064] Figure 3 FIG. 4 is a schematic structural diagram of a computer device in one embodiment. DETAILED DESCRIPTION

[0065] The present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0066] Figure 1 FIG. 1 is a flow chart of a method for controlling parameters of a refining device provided by an embodiment of the present application. Figure 1 As shown, the method may include the following steps:

[0067] S110, obtaining target parameter data and process operation data of the refining device within a preset time.

[0068] Among them, the target parameter data includes the target parameter, the target parameter type and the corresponding value of the target parameter at each time point, etc., and may include the side line extraction temperature, product flow rate, etc. The process operation data includes the process operation parameter, the process operation parameter type and the corresponding value of the process operation parameter at each time point, etc., and may include temperature, pressure, flow rate, component ratio and time, etc. The target parameter is usually affected by multiple process operation parameters, and different process operation parameters have different degrees of influence on the target parameter.

[0069] The preset time can be determined based on the sample size of the historical data obtained. For example, the preset time is set to three months, and there is no limit here.

[0070] Obtain the target parameter data of the refining unit within three months and use it as time series data, recorded as Collect the changes in the process operating parameters of the device within three months, and centralize the changes in the process operating parameters and record them as T is the time length of target parameter data and process operation data, d is the number of process operation parameters. For example, the target parameter is Y, and there are 15 process operation parameters, namely: W1, W2, W3, W4, W5, W6, W7, W8, W9, W10, W11, W12, W13, W14, W15. The master data for these three months has a total of 100,000 data points, which are divided into the first 80,000 as the training set and the last 20,000 as the test set.

[0071] S120, determining a steady-state time period in which the process operation parameters affecting the target parameters are in a steady state according to the target parameter data and the process operation data.

[0072] Remove outliers from the process operation parameter data and use the average value as fill. For each process operation parameter, run the PELT algorithm to detect its change point. If there is no change point in a continuous time period, it is considered that the process operation parameter remains steady in this time period. If all process operation parameters remain steady in a time period, this period is considered to be steady, that is, the steady-state time period.

[0073] S130, based on the steady-state time period, constructing a state space model for evaluating the impact of the process operating parameters on the target parameters.

[0074] Combined with the steady-state points in the steady-state time period, a state space model is established to describe the influence of process operation parameters on target parameters. The state space model contains hidden processes affected by process operation parameters. The target parameters can be obtained by summing all hidden processes and adding noise. Each hidden process can be represented by a second-order autoregressive model with external variables.

[0075] S140, determining a delay time of the process operation parameter according to the target parameter data, the process operation data and the state space model.

[0076] For each process operation parameter, the upper and lower bounds of the delay time of the process operation parameter are determined in combination with existing knowledge and practical experience. The granularity of the delay time can be determined according to the requirements of its own computing power and accuracy, and further an alternative set of delay times for the process operation parameter can be obtained.

[0077] The contribution of each variable to the estimated target parameter is expressed by calculating the average absolute effect of each variable. After processing, there will be strong collinearity in different variables generated by the same process operating parameter. At this time, the delay corresponding to the variable that is most useful for the target parameter estimation is more likely to be the real delay. For multiple variables corresponding to a process operating parameter, the delay time corresponding to the variable with the largest average absolute effect is determined as the delay time of the process operating parameter, that is, the optimal delay time.

[0078] S150, determining device control parameters of the refinery according to the delay time and state space model of the process operation parameters.

[0079] The model parameters of the state space model are converted into parameters of differential equations, and the delay time of the process operation parameters is substituted into the state space model to obtain parameters indicating the degree of influence of the current process operation parameters on the target parameters and parameters indicating the speed of influence of the current process operation parameters on the target parameters.

[0080] S160, verifying the rationality of the state space model and the rationality of the device control parameters based on the target parameter data, the process operation data, the delay time, the state space model and the device control parameters of the refining device.

[0081] The target parameter data, process operation data and optimal delay time divided into the test set are substituted into the state space model, and the root mean square error of different prediction steps is calculated. The model coefficients of the state space model converted into differential equations are evaluated based on actual experience to determine whether the obtained coefficients are reasonable.

[0082] It can be seen that the embodiment of the present application obtains target parameter data and process operation data within a preset time of the refining device, wherein the target parameter data and the process operation data are both time series data, the target parameter data includes the target parameter, and the process operation data includes the process operation parameter, and determines the steady-state time period in which the process operation parameter affecting the target parameter is in a steady state according to the target parameter data and the process operation data, and constructs a state space model for evaluating the influence of the process operation parameter on the target parameter based on the steady-state time period, and determines the delay time of the process operation parameter according to the target parameter data, the process operation data and the state space model, and determines the device control parameters of the refining device according to the delay time and the state space model of the process operation parameter, and verifies the rationality of the state space model and the rationality of the device control parameters according to the target parameter data, the process operation data, the delay time, the state space model and the device control parameters of the refining device, so as to accurately evaluate the influence of the process operation parameters on the target parameters and reduce the evaluation cost.

[0083] As an achievable manner, according to the target parameter data and the process operation data, determining a steady-state time period in which the process operation parameter affecting the target parameter is in a steady state includes:

[0084] According to the process operation data, the outlier time point of the process operation parameter is determined based on a preset sliding window, wherein the determination formula of the outlier time point is expressed as follows:

[0085]

[0086] Among them, W i,t represents the value of process operating parameter i at time t, represents the average value of process operation parameter i in the time period [t-δ, t-1], represents the average value of process operation parameter i in the time period [t+δ, t+1], σ(w i,t-δ:t-1 ) represents the standard deviation of process operation parameter i in the time period [t-δ, t-1], σ(w i,t+δ:t+1 ) represents the standard deviation of the process operation parameter i in the time period [t+δ, t+1], δ represents the length of the preset sliding window;

[0087] Determine the effective time period in the process operation data according to the outlier time point of the process operation parameter;

[0088] According to the target parameter data and the process operation data, a steady-state time period in a steady state within an effective time period affecting the target parameter is determined.

[0089] Among them, formula (1) means that this point must be an extreme point, which excludes false alarms caused by large changes in parameters in a short time. Formula (2) means that the difference between this point and the left and right time windows exceeds a certain threshold. Since parameters may change greatly in a short time, the standard deviations of the left and right ends of the point are examined at the same time, and several times the maximum value of the standard deviation is taken as the threshold, and 3 times the maximum value of the standard deviation is preferably used as the threshold. When time t satisfies both formula (1) and formula (2), time t is considered to be an outlier to prevent false alarms.

[0090] Eliminate outlier time points of process operation parameters to obtain valid time periods in process operation data. For each process operation parameter, use formula (1) and formula (2) to retrieve and eliminate outliers.

[0091] According to the target parameter data and process operation data, find out the time period when the target parameter and process operation parameter remain in a steady state. For each process operation parameter, first find out all the time points in the steady state under the process operation parameter. Then, if all process operation parameters do not change significantly within a certain period of time, then it is considered that the system has reached a steady state during this period of time.

[0092] As an achievable manner, according to the target parameter data and the process operation data, determining a steady-state time period in which the process operation parameter affecting the target parameter is in a steady state includes:

[0093] According to the target parameter data and process operation data, determine the set of change points where the process operation parameters jump Satisfy t 1 <t 2 <..., based on a preset time interval δ j Determine the steady-state point corresponding to each change point in the change point set and express it as t j +δ j , the steady-state point satisfies t j +δ j <t j+1 ;

[0094] According to the steady-state point corresponding to each change point in the change point set, an initial steady-state point set of the process operation parameters is determined based on a first-order autoregressive model;

[0095] If the time distance between the current steady-state point and the next steady-state point in the initial steady-state point set is less than the preset time interval, the current steady-state point is removed; if the time distance between the current steady-state point and the next steady-state point in the initial steady-state point set is greater than the preset time interval, the current steady-state point and the next steady-state point are retained to obtain the final steady-state point set;

[0096] According to the final steady-state point set, a steady-state time period in which the process operating parameters affecting the target parameters are in a steady state is determined.

[0097] For each process operation parameter, the change point detection algorithm can be used to find the point where the process operation parameter jumps. The change point detection algorithm can use the PELT algorithm to obtain the change point set where the process operation parameter jumps. At this time, the steady-state point can appear in a period of time after the change point when there is no drastic change. The change point set of the process operation parameter is obtained based on the process operation data through the PELT algorithm. Since the condition for the steady-state point to be established is that the time it appears is before the time when the next change point appears, the steady-state point must satisfy t j +δ j <t j+1 .

[0098] For the preset time interval δ j, can be determined using an enumeration algorithm, and a first-order autoregressive model is used to determine the time period [t j , t j +δ j ] whether there is no trend change within . Use the hypothesis test method to test whether the model parameter β of the first-order autoregression is non-zero. If β is non-zero, then it is considered that the system has not reached a steady state. Otherwise, it is considered that the system has reached a steady state and the enumeration is terminated. The initial steady-state point set of a single process operation parameter found by the algorithm is recorded as

[0099] Take the set of all initial steady-state points The union of like If the time distance between the current steady-state point and the next steady-state point in the initial steady-state point set is less than the preset time interval, it means that the current steady-state point and the next steady-state point are too close, indicating that the previous steady-state point is only a single process operation parameter that reaches a steady state, and the system has not reached a steady state, so the previous steady-state point is removed. If the time distance between the current steady-state point and the next steady-state point in the initial steady-state point set is greater than the preset time interval, it means that the distance between the current steady-state point and the next steady-state point is appropriate, and the system has reached a steady state, and the current steady-state point and the next steady-state point are retained to obtain the final steady-state point set.

[0100] The time point at which the process operation parameters affecting the target parameters are in a steady state is obtained according to the final steady-state point set, and the time period in which all process operation parameters do not undergo significant changes, that is, the steady-state time period, is determined according to the time point at which the process operation parameters are in a steady state.

[0101] As an achievable approach, a state space model for evaluating the impact of process operating parameters on target parameters is constructed based on a steady-state time period, including:

[0102] The state space model is expressed as:

[0103]

[0104] Among them, y t is the target parameter, which is obtained by summing all hidden processes and adding noise. is the reference point of the target parameter, obtained from the previous steady-state point before time t, that is, is the final steady-state point set, x i,t represents the hidden process affected by the process operation parameter i at time t, which is an unknown variable, x i,t-k represents the hidden process affected by process operation parameter i at time tk, k represents the number of past time points, and w i,t-θi represents the value of process parameter i at time t-θi, It represents the reference point of process parameter i at time t-θi, which is obtained from the previous steady-state point before time t, that is, θi represents the delay value of the i-th process operation parameter, d is the number of process operation parameters i, α i,k ,β i ,c i is the parameter of the autoregressive model, ∈ i，t , ε t They are the noise of the hidden process and the observation noise, respectively, with mean 0 and variance ω 2 , σ 2 The normal distribution of 2 According to the process parameters w i,t-θi Calculate the variance, σ 2 is based on the hidden process x i,t Calculate the variance; x d+1,t , x d+1,t-1 The influencing factors of the unknown variables hidden in the raw material properties at time t and time t-1, is the magnitude of its mean shift, x i,0 is the initial hiding process, μ i，0 For x i,t The mean of .

[0105] As an achievable approach, the delay time of the process operation parameter is determined according to the target parameter data, the process operation data and the state space model, including:

[0106] According to the process operation data, the delay upper bound and delay lower bound of the process operation parameters are set, which are expressed as Satisfy 0≤θ i，min ≤θ i，max ;

[0107] Based on the delay upper bound, delay lower bound and preset delay interval g i , determine the delay time candidate set of process operation parameters

[0108] Based on the state space model and the delay time candidate set, the mean absolute effect value of each variable in the process operation parameter is calculated by the following formula:

[0109]

[0110] in, is the estimated value of the state space model at time t+s given time 1 to t, For the state space model, excluding the current variable w′ at given time 1 to t iAfter the influence of , the estimated value at time t+s, that is, the variable values ​​corresponding to the period [t+1, t+s] are all 0, s is the pre-selected step length, and T is the time length of the target parameter data and process operation data;

[0111] The delay time corresponding to the variable with the largest mean absolute effect value is determined as the delay time of the process operation parameter.

[0112] According to the setting requirements and practical experience in this field, set the upper and lower limits of the delay of the process operation parameters. Determine the granularity of the delay according to the needs of its own computing power and accuracy, that is, preset the delay interval g i . Based on actual production experience, a recommended range of process operation parameter delays is given, and a coarse screening is performed with a granularity of 5. For example, if the recommended delay range of process operation parameter W4 is 5-20, we will generate four variables with delays of 5, 10, 15, and 20 from W4 and include these four variables in the training process. The same method is used for other process operation parameters. Finally, 48 variables corresponding to process operation parameters and different delays are obtained in the state space model.

[0113] Definition L i is the size of the set of candidate parameters for the i-th process operation. At this time, the delay time in the candidate set of delay time is expressed as For the existing 1 Process it to transform it into a 1 Time series of variables satisfy . Perform similar processing on all remaining d-1 variables. Finally, after removing all time points containing blank values, you will get a A collection of sizes, where L = ∑L i is the total number of variables. Keep the target parameter y at the same time point as w′ to obtain (w′, y′) is modeled and solved using a state-space model to obtain estimated model parameters.

[0114] By calculating the contribution of each variable to the estimated target parameter, the optimal delay time for each process operation parameter is determined. After processing, there will be strong collinearity among different variables generated by the same process operation parameter. At this time, the delay time corresponding to the variable that is most useful for estimating the target parameter is more likely to be the real delay time.

[0115] The average absolute effect value of each variable in the process operation parameter is calculated by formula (7), and the delay time corresponding to the variable with the largest average absolute effect is selected, that is, the optimal delay time is used as the delay time of the process operation parameter.

[0116] Exemplarily, after the model training is completed, we calculated the AAME of all 48 variables, as shown in Table 1. The AAME of each variable represents the contribution of the corresponding process operation parameter with a specific delay in estimating the target parameter. For each process operation parameter, we select the delay time with the largest AAME. According to Table 1, the selected delay is {W1Lag0,W9Lag5,W2Lag5,W10Lag0,W3Lag5,W11Lag15,W4Lag5,W12Lag10,W5Lag0,W13Lag30,W6Lag5,W14Lag15,W7Lag25,W15Lag0,W8Lag0}, and we define this combination as setting 1.

[0117] Table 1 AAME values ​​of process operating parameters when the particle size is 5

[0118]

[0119] As an achievable approach, the delay time of the process operation parameter is determined according to the target parameter data, the process operation data and the state space model, including:

[0120] The state space model is used as the proxy model, and the root mean square error between the predicted value of the preset future time length of the target parameter and the true value is used as the objective function. The objective function is optimized through the proxy model to obtain the objective function value;

[0121] Based on historical production data, determine the search interval of the delay time of the impact of process operation parameters on target parameters, and perform Bayesian optimization iteration on the delay time in the search interval;

[0122] When the delay time reaches a certain number of iterations or the objective function value is less than a preset threshold, the delay time of the process operation parameter corresponding to the minimum value of the objective function is selected as the delay time of the process operation parameter.

[0123] In addition to the above method of determining the optimal delay time by calculating the mean absolute effect value, the Bayesian optimization method can also be used to achieve the selection of the optimal delay time. In Bayesian optimization, the state space model is used as the proxy model, and the root mean square error between the predicted value and the true value of the preset future duration of the target parameter is used as the objective function. The objective function is optimized through the proxy model, and the optimized objective function is the root mean square error RMSE value of the target parameter prediction result.

[0124] Historical production data is the configuration data of target parameters and process operation parameters in the past production process, which contains certain production experience and has good reference value. A set of initial delay values ​​of process parameters is given and the search interval of process operation parameter delay is defined according to production experience. The optimization iteration is performed in combination with the computing power. When a certain number of iterations is reached or the objective function value is less than a certain threshold, the delay time of the process operation parameter corresponding to the minimum value of the objective function is selected as the delay time of the process operation parameter.

[0125] As a feasible approach, the device control parameters of the refinery are determined based on the delay time and state space model of the process operation parameters, including:

[0126] Convert the state space model to a differential equation model;

[0127] Calculate the first and second derivative functions of the state-space model;

[0128] Substitute the first-order derivative function and the second-order derivative function into the differential equation model to obtain the target differential equation model;

[0129] According to the state space model and the target differential equation model, adjusting the parameters of the state space model;

[0130] According to the parameters of the state space model, the device control parameters of the refining device are determined. The device control parameters can be expressed as the following formula:

[0131]

[0132] Among them, K i Indicates the influence of process operation parameters on target parameters. If K i >0 means there is a positive impact, K i <0 means there is a negative impact, τ i and i It indicates how fast the process parameters affect the target parameters, β i , α i,1 , α i,2 is the parameter of the autoregressive model, and Δ represents the sampling interval.

[0133] Convert the state space model to a differential equation model, which can be expressed as:

[0134]

[0135] Among them, K i Represents the influence of process operation parameters on target parameters. If K i >0 means there is a positive impact, K i <0 means there is a negative impact. τi and i It indicates how fast the process operation parameters affect the target parameters, θ i Indicates the delay time of the impact of process operation parameters on target parameters.

[0136] In the state space model, we can use x i，t+1 -x i，t To approximate the first-order derivative function x′ i (t) At the same time, use x i，t+2 -2x i，t+1 +x i，t To approximate the second-order derivative function x″ i (t), substitute the first-order derivative function and the second-order derivative function into the differential equation model to obtain the target differential equation model. Comparing the target differential equation model with the state space transfer model, we can get β i , α i,1 , α i,2 The expression equation of is obtained, and the control parameters are obtained by solving the simultaneous equations.

[0137] Setting 1 is the optimal delay time we selected. Under this delay time setting, the training set data is used for training, and the corresponding state space model parameters can be calculated by formula (8). The specific differential equation parameters calculated for each process operation parameter are shown in Table 2 below.

[0138] Table 2 Differential equation coefficients for different process operation parameters

[0139]

[0140] As a feasible method, the rationality of the state space model and the rationality of the device control parameters are verified according to the target parameter data, process operation data, the state space model, the delay time and the device control parameters of the refining device, including:

[0141] Predict the time series values ​​of target parameters in future time periods based on process operation data, state space model, and delay time;

[0142] The rationality of the state space model and the rationality of the device control parameters are verified based on the error and trend between the time series value and the true value.

[0143] The time series values ​​of the target parameters in the future time period are predicted based on the process operation data, state space model, and delay time. The time series values ​​are the numerical values ​​of the target parameters arranged in chronological order, which are the RMSE values ​​for different prediction steps.

[0144] The rationality of the state space model and the rationality of the device control parameters are verified based on the error and trend between the time series value and the true value. If the error between the time series value and the true value is small and the trend is basically consistent, it means that the state space model and the device control parameters are reasonably set. Otherwise, the state space model and the device control parameters are unreasonable.

[0145] For example, the training model with a delay time of 1 is applied to the last 20,000 test set data, and the predicted values ​​of the target parameters under different prediction step sizes Y are calculated, and the RMSE value between the predicted value and the true value is calculated. The results are shown in Table 3:

[0146] Table 3 RMSE results of prediction values ​​corresponding to setting 1

[0147]

[0148] It can be seen that the delay time finally selected has a better effect.

[0149] In addition to being applied to refining equipment, the above method can also be applied to other equipment involved in the petroleum refining process, which is not limited here.

[0150] It should be understood that although Figure 1 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in the application, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, Figure 1 At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.

[0151] Figure 2 A schematic diagram of a parameter control device for a refinery provided in an embodiment of the present application, for executing the following Figure 1 The method flow shown in Figure 2 As shown, the refinery parameter control device 200 may include: an acquisition module 210, a determination module 220, a construction module 230 and a verification module 240. The main functions of each component module are as follows:

[0152] An acquisition module 210 is used to acquire target parameter data and process operation data of a refining device within a preset time, wherein the target parameter data and the process operation data are both time series data, the target parameter data includes target parameters, and the process operation data includes process operation parameters;

[0153] A determination module 220, for determining a steady-state time period during which a process operation parameter affecting the target parameter is in a steady state according to the target parameter data and the process operation data;

[0154] A construction module 230 is used to construct a state space model for evaluating the influence of the process operation parameters on the target parameters based on the steady-state time period;

[0155] The determination module 220 is further used to determine the delay time of the process operation parameter according to the target parameter data, the process operation data and the state space model;

[0156] The determination module 220 is further used to determine the device control parameters of the refinery device according to the delay time and the state space model of the process operation parameters;

[0157] The verification module 240 is used to verify the rationality of the state space model and the rationality of the device control parameters based on the target parameter data, process operation data, delay time, the state space model and the device control parameters of the refining device.

[0158] As an implementable manner, the determination module 220 is specifically used to: determine the outlier time point of the process operation parameter based on the process operation data and a preset sliding window, wherein the determination formula of the outlier time point is expressed as the following formula:

[0159]

[0160] Among them, W i,t represents the value of process operating parameter i at time t, represents the average value of process operation parameter i in the time period [t-δ, t-1], represents the average value of process operation parameter i in the time period [t+δ, t+1], σ(w i,t-δ:t-1 ) represents the standard deviation of process operation parameter i in the time period [t-δ, t-1], σ(w i,t+δ:t+1 ) represents the standard deviation of the process operation parameter i in the time period [t+δ, t+1], δ represents the length of the preset sliding window;

[0161] Determine the effective time period in the process operation data according to the outlier time point of the process operation parameter;

[0162] According to the target parameter data and the process operation data, a steady-state time period in a steady state within an effective time period affecting the target parameter is determined.

[0163] As an achievable method, the determination module 220 is specifically used to: determine a set of change points where the process operation parameters jump according to the target parameter data and the process operation data. Satisfy t1 <t 2 <..., based on a preset time interval δ j Determine the steady-state point corresponding to each change point in the change point set and express it as t j +δ j , the steady-state point satisfies t j +δ j <t j+1 ;

[0164] According to the steady-state point corresponding to each change point in the change point set, an initial steady-state point set of the process operation parameters is determined based on a first-order autoregressive model;

[0165] If the time distance between the current steady-state point and the next steady-state point in the initial steady-state point set is less than the preset time interval, the current steady-state point is removed; if the time distance between the current steady-state point and the next steady-state point in the initial steady-state point set is greater than the preset time interval, the current steady-state point and the next steady-state point are retained to obtain the final steady-state point set;

[0166] According to the final steady-state point set, a steady-state time period in which the process operating parameters affecting the target parameters are in a steady state is determined.

[0167] As an implementable manner, the construction module 230 is specifically used to: the state space model is expressed as:

[0168]

[0169] Among them, y t is the target parameter, which is obtained by summing all hidden processes and adding noise. is the reference point of the target parameter, obtained from the previous steady-state point before time t, that is, is the final steady-state point set, x i,t represents the hidden process affected by the process operation parameter i at time t, which is an unknown variable, x i,t-k represents the hidden process affected by process operation parameter i at time tk, k represents the number of past time points, and w i,t-θi represents the value of process parameter i at time t-θi, It represents the reference point of process parameter i at time t-θi, which is obtained from the previous steady-state point before time t, that is, θi represents the delay value of the i-th process operation parameter, d is the number of process operation parameters i, α i,k ,β i ,c i is the parameter of the autoregressive model, ∈ i，t , ε t They are the noise of the hidden process and the observation noise, respectively, with mean 0 and variance ω 2 , σ2 The normal distribution of 2 According to the process parameters w i,t-θi Calculate the variance, σ 2 is based on the hidden process x i,t Calculate the variance; x d+1,t , x d+1,t-1 The influencing factors of the unknown variables hidden in the raw material properties at time t and time t-1, is the magnitude of its mean shift, x i,0 is the initial hiding process, μ i，0 For x i,t The mean of .

[0170] As an implementable manner, the determination module 220 is specifically used to: set the delay upper bound and the delay lower bound of the process operation parameter according to the process operation data, which are respectively expressed as Satisfy 0≤θ i，min ≤θ i，max ;

[0171] Based on the delay upper bound, delay lower bound and preset delay interval g i , determine the delay time candidate set of process operation parameters

[0172] Based on the state space model and the delay time candidate set, the mean absolute effect value of each variable in the process operation parameter is calculated by the following formula:

[0173]

[0174] in, is the estimated value of the state space model at time t+s given time 1 to t, For the state space model, excluding the current variable w′ at given time 1 to t i After the influence of , the estimated value at time t+s, that is, the variable values ​​corresponding to the period [t+1, t+s] are all 0, s is the pre-selected step length, and T is the time length of the target parameter data and process operation data;

[0175] The delay time corresponding to the variable with the largest mean absolute effect value is determined as the delay time of the process operation parameter.

[0176] As an achievable manner, the determination module 220 is specifically used to: use the state space model as a proxy model, use the root mean square error between the predicted value of the preset future duration of the target parameter and the true value as the objective function, optimize the objective function through the proxy model, and obtain the objective function value;

[0177] Based on historical production data, determine the search interval of the delay time of the impact of process operation parameters on target parameters, and perform Bayesian optimization iteration on the delay time in the search interval;

[0178] When the delay time reaches a certain number of iterations or the objective function value is less than a preset threshold, the delay time of the process operation parameter corresponding to the minimum value of the objective function is selected as the delay time of the process operation parameter.

[0179] As an implementable manner, the determination module 220 is specifically used to: convert the state space model into a differential equation model;

[0180] Calculate the first and second derivative functions of the state-space model;

[0181] Substitute the first-order derivative function and the second-order derivative function into the differential equation model to obtain the target differential equation model;

[0182] According to the state space model and the target differential equation model, adjusting the parameters of the state space model;

[0183] According to the parameters of the state space model, the device control parameters of the refining device are determined. The device control parameters are expressed as follows:

[0184]

[0185] Among them, K i Indicates the influence of process operation parameters on target parameters. If K i >0 means there is a positive impact, K i <0 means there is a negative impact, τ i and i It indicates how fast the process parameters affect the target parameters, β i , α i,1 , α i,2 is the parameter of the autoregressive model, and Δ represents the sampling interval.

[0186] As an implementable manner, the verification module 240 is specifically used to: predict the time series value of the target parameter in the future time period according to the process operation data, the state space model, and the delay time;

[0187] The rationality of the state space model and the rationality of the device control parameters are verified based on the error and trend between the time series value and the true value.

[0188] The same or similar parts between the above embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0189] It should be noted that the embodiments of the present application may involve the use of user data. In actual applications, user-specific personal data can be used in the scheme described in this article within the scope permitted by applicable laws and regulations, provided that the applicable laws and regulations of the country are met (for example, the user's explicit consent, effective notification to the user, and the user's explicit authorization, etc.).

[0190] According to an embodiment of the present application, the present application also provides a computer device and a computer-readable storage medium.

[0191] like Figure 3 , is a block diagram of a computer device according to an embodiment of the present application. The computer device is intended to represent various forms of digital computers or mobile devices. The digital computer may include a desktop computer, a portable computer, a workbench, a personal digital assistant, a server, a mainframe computer, and other suitable computers. The mobile device may include a tablet computer, a smart phone, a wearable device, etc.

[0192] like Figure 3 As shown, the computer device 300 includes a computing unit 301, a ROM 302, a RAM 303, a bus 304, and an input / output (I / O) interface 305. The computing unit 301, the ROM 302, and the RAM 303 are connected to each other via the bus 304. The input / output (I / O) interface 305 is also connected to the bus 304.

[0193] The computing unit 301 can perform various processes in the method embodiment of the present application according to the computer instructions stored in the read-only memory (ROM) 302 or the computer instructions loaded from the storage unit 308 to the random access memory (RAM) 303. The computing unit 301 can be various general and / or special processing components with processing and computing capabilities. The computing unit 301 may include, but is not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. In some embodiments, the method provided in the embodiment of the present application can be implemented as a computer software program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 308.

[0194] RAM 303 may also store various programs and data required for the operation of computer device 300. Part or all of a computer program may be loaded and / or installed on computer device 300 via ROM 302 and / or communication unit 309.

[0195] The input unit 306, output unit 307, storage unit 308 and communication unit 309 in the computer device 300 can be connected to the I / O interface 305. The input unit 306 can be, for example, a keyboard, a mouse, a touch screen, a microphone, etc. The output unit 307 can be, for example, a display, a speaker, an indicator light, etc. The computer device 300 can exchange information, data, etc. with other devices through the communication unit 309.

[0196] It should be noted that the device may also include other components necessary for normal operation, or may only include components necessary for implementing the solution of the present application, rather than all the components shown in the figure.

[0197] Various implementations of the systems and techniques described herein can be realized in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof.

[0198] The computer instructions for implementing the method of the present application can be written in any combination of one or more programming languages. These computer instructions can be provided to the computing unit 301, so that when the computer instructions are executed by the computing unit 301 such as a processor, the steps involved in the method embodiment of the present application are executed.

[0199] The computer-readable storage medium provided in the present application may be a tangible medium that may contain or store computer instructions for executing the steps involved in the method embodiments of the present application. The computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, and other forms of storage media.

[0200] The above specific implementations do not constitute a limitation on the protection scope of this application. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of this application should be included in the protection scope of this application.

Claims

1. A method for controlling parameters of a refining device, characterized in that: The method comprises: Acquire target parameter data and process operation data of a refining device within a preset time, wherein the target parameter data and the process operation data are both time series data, the target parameter data includes target parameters, and the process operation data includes process operation parameters; Determining, according to the target parameter data and the process operation data, a steady-state time period in which a process operation parameter affecting the target parameter is in a steady state; Based on the steady-state time period, a state space model for evaluating the impact of process operating parameters on target parameters is constructed: Determining a delay time of the process operation parameter according to the target parameter data, the process operation data and the state space model; Determining device control parameters of the refinery device according to the delay time of the process operation parameter and the state space model; Based on the target parameter data, the process operation data, the delay time, the state space model and the device control parameters of the refining device, the rationality of the state space model and the rationality of the device control parameters are verified.

2. The method according to claim 1, characterized in that The step of determining, based on the target parameter data and the process operation data, a steady-state time period during which the process operation parameter affecting the target parameter is in a steady state comprises: According to the process operation data, the outlier time point of the process operation parameter is determined based on a preset sliding window, wherein the determination formula of the outlier time point is expressed as the following formula: Among them, W i，t represents the value of process operating parameter i at time t, represents the average value of process operation parameter i in the time period [t-δ, t-1], represents the average value of process operation parameter i in the time period [t+δ, t+1], σ(w i，t-δ：t-1 ) represents the standard deviation of process operation parameter i in the time period [t-δ, t-1], σ(w i，t+δ：t+1 ) represents the standard deviation of the process operation parameter i in the time period [t+δ, t+1], δ represents the length of the preset sliding window; Determining a valid time period in the process operation data according to the outlier time point of the process operation parameter; A steady-state time period in a steady state within an effective time period affecting the target parameter is determined according to the target parameter data and the process operation data.

3. The method according to claim 2, characterized in that The step of determining, based on the target parameter data and the process operation data, a steady-state time period during which the process operation parameter affecting the target parameter is in a steady state comprises: Determine a set of change points where process operation parameters change according to the target parameter data and the process operation data Satisfying t1<t2<…, based on the preset time interval δ j Determine the steady-state point corresponding to each change point in the change point set and represent it as t j +δ j , the steady-state point satisfies t j +δ j <t j+1 ; Determining an initial steady-state point set of the process operating parameters based on a first-order autoregressive model according to the steady-state point corresponding to each change point in the change point set; If the time distance between the current steady-state point and the next steady-state point in the initial steady-state point set is less than the preset time interval, the current steady-state point is removed; if the time distance between the current steady-state point and the next steady-state point in the initial steady-state point set is greater than the preset time interval, the current steady-state point and the next steady-state point are retained to obtain a final steady-state point set; According to the final steady-state point set, a steady-state time period in which the process operation parameters affecting the target parameters are in a steady state is determined.

4. The method according to claim 1, characterized in that: The step of constructing a state space model for evaluating the influence of process operation parameters on target parameters based on the steady-state time period includes: The state space model is expressed as: Among them, y t is the target parameter, which is obtained by summing all hidden processes and adding noise. is the reference point of the target parameter, obtained from the previous steady-state point before time t, that is, is the final steady-state point set, x i，t represents the hidden process affected by the process operation parameter i at time t, which is an unknown variable, x i，t-k represents the hidden process affected by process operation parameter i at time tk, k represents the number of past time points, and w i，t-θi represents the value of process parameter i at time t-θi, It represents the reference point of process parameter i at time t-θi, which is obtained from the previous steady-state point before time t, that is, θi represents the delay value of the i-th process operation parameter, d is the number of process operation parameters i, α i，k , β i , c i is the parameter of the autoregressive model, ∈ i，t , ε t They are the noise of the hidden process and the observation noise, respectively, with mean 0 and variance ω 2 , σ 2 Normal distribution of x d+1，t , x d+1，t-1 The influencing factors of the unknown variables hidden in the raw material properties at time t and time t-1, is the magnitude of its mean shift, x i，0 is the initial hiding process, μ i，0 For x i，t The mean of .

5. The method according to claim 1, characterized in that Determining the delay time of the process operation parameter according to the target parameter data, the process operation data and the state space model includes: According to the process operation data, the delay upper bound and the delay lower bound of the process operation parameter are set, which are respectively expressed as Satisfy 0≤θ i，min ≤θ i，max ; Based on the delay upper bound, the delay lower bound and the preset delay interval g i , determine the delay time candidate set of the process operation parameters Based on the state space model and the delay time candidate set, the average absolute effect value of each variable in the process operation parameter is calculated by the following formula: in, is the estimated value of the state space model at time t+s given time 1 to t, For the state space model, excluding the current variable w′ at given time 1 to t i The estimated value at time t+s after the influence of , that is, the variable values ​​corresponding to the period [t+1, t+s] are all 0, s is the pre-selected step length, and T is the time length of the target parameter data and the process operation data; The delay time corresponding to the variable with the largest average absolute effect value is determined as the delay time of the process operation parameter.

6. The method according to claim 1, characterized in that Determining the delay time of the process operation parameter according to the target parameter data, the process operation data and the state space model includes: Taking the state space model as a proxy model, taking the root mean square error between the predicted value of the preset future duration of the target parameter and the true value as the objective function, optimizing the objective function through the proxy model to obtain an objective function value; Based on historical production data, determining a search interval of a delay time of the process operation parameter affecting the target parameter, and performing Bayesian optimization iteration on the delay time in the search interval; When the delay time reaches a certain number of iterations or the objective function value is less than a preset threshold, the delay time of the process operation parameter corresponding to the minimum value of the objective function is selected as the delay time of the process operation parameter.

7. The method according to claim 1, characterized in that Determining the device control parameters of the refinery device according to the delay time of the process operation parameters and the state space model includes: Converting the state-space model into a differential equation model; Calculating a first-order derivative function and a second-order derivative function of the state-space model; Substituting the first-order derivative function and the second-order derivative function into the differential equation model to obtain a target differential equation model; According to the state-space model and the target differential equation model, adjusting parameters of the state-space model; According to the parameters of the state space model, the device control parameters of the refining device are determined, and the device control parameters are expressed as the following formula: Among them, K i Indicates the influence of process operation parameters on target parameters. If K i >0 means there is a positive impact, K i <0 means there is a negative impact, τ i and i It indicates how fast the process parameters affect the target parameters, β i , α i，1 , α i，2 is the parameter of the autoregressive model, and Δ represents the sampling interval.

8. The method according to claim 1, characterized in that: Verifying the rationality of the state space model and the rationality of the device control parameters according to the target parameter data, the process operation data, the state space model, the delay time, and the device control parameters of the refining device, including: Predicting a time series value of a target parameter in a future time period based on the process operation data, the state space model, and the delay time; The rationality of the state space model and the rationality of the device control parameters are verified based on the errors and trends between the time series values ​​and the true values.

9. A parameter control device for a refining device, characterized in that: The device comprises: An acquisition module, used to acquire target parameter data and process operation data of a refining device within a preset time, wherein the target parameter data and the process operation data are both time series data, the target parameter data includes target parameters, and the process operation data includes process operation parameters; A determination module, configured to determine, based on the target parameter data and the process operation data, a steady-state time period in which the process operation parameter affecting the target parameter is in a steady state; A construction module, for constructing a state space model for evaluating the influence of process operation parameters on target parameters based on the steady-state time period; A determination module is further used to determine the delay time of the process operation parameter according to the target parameter data, the process operation data and the state space model; A determination module is further used to determine a device control parameter of the refining device according to the delay time of the process operation parameter and the state space model; A verification module is used to verify the rationality of the state space model and the rationality of the device control parameters based on the target parameter data, the process operation data, the delay time, the state space model and the device control parameters of the refining device.

10. A computer device comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores computer instructions that can be executed by the at least one processor, and the computer instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 8.

Images