A digital-based fault prediction system and method for petrochemical equipment

Abstract

The present invention relates to the field of equipment fault prediction, and specifically to a digital-based petrochemical equipment fault prediction system and method, including: a digital sensing module, a state control module, an operation monitoring module, a fault reporting module, and a maintenance positioning module. The digital sensing module is used to detect the quality parameters of petrochemical products and the operation parameters of production equipment. The state control module is used to establish the state equation of the petrochemical production line. The operation monitoring module is used to calculate the change rate of operation parameters and the probability distribution equation of product quality parameters. The fault reporting module judges abnormal working conditions of the equipment. The maintenance positioning module is used to output the probability of faults occurring in each process. The present invention can monitor and control the production process in real time, optimize the equipment utilization rate, improve the production efficiency, shorten the equipment maintenance downtime at the same time, effectively save the maintenance cost, and ensure the normal operation of petrochemical equipment.

Description

Technical Field

[0001] The present invention relates to the field of equipment fault prediction, and particularly to a digital-based petrochemical equipment fault prediction system and method. Background Art

[0002] Petrochemical equipment refers to the equipment required in the production and processing processes of the petrochemical industry, usually including distillation towers, reactors, separators, heat exchangers, compressors, pumps, etc. Since there are numerous petrochemical industry equipment and hundreds of different equipment parameters are involved, it is difficult to manage them uniformly by manual alone. Faults caused by human or natural factors in petrochemical equipment cannot be discovered in time, which is likely to cause equipment safety accidents.

[0003] In the petrochemical industrial production process, it is often necessary to adjust the parameters of the products to meet different fractionation and oil quality requirements. There are significant differences in the production modes and the ranges of operating parameters of the equipment before and after adjustment. The existing fault identification systems are difficult to adapt to the multi-mode industrial process monitoring process and cannot timely identify possible abnormal conditions of petrochemical equipment in the changing production parameters.

[0004] In addition, there are many possible locations where petrochemical equipment fails. The location and traceability of faults must rely on the means of manual step-by-step inspection, with low digitalization level. In the monitoring process, it is also necessary to involve means such as repeated startup tests and product deterioration analysis, resulting in problems such as increased maintenance costs and increased maintenance risks. Summary of the Invention

[0005] The purpose of the present invention is to provide a digital-based petrochemical equipment fault prediction system and method to solve the problems raised in the above background art.

[0006] To solve the above technical problems, the present invention provides the following technical solution: A digital-based petrochemical equipment fault prediction system, comprising: a digital sensing module, a state control module, an operation monitoring module, a fault reporting module, and a maintenance positioning module;

[0007] The digital sensing module includes product detection equipment at the end of the petrochemical production line and parameter detection equipment arranged in the production equipment. The product detection equipment is used to detect the quality parameters of the products at the output end of the petrochemical equipment and feed back the detection results to the data processing center. The parameter detection equipment is used to use sensors to obtain all the operation parameters of each production equipment in real time;

[0008] The state control module is used to select the production equipment operation parameters and product quality parameters during the normal operation of the equipment, respectively form the first training set and the second training set. After screening the data in the training set, taking each quality parameter of the product as the state quantity and the operation parameters of the equipment as the excitation, a state equation of the petrochemical production line is established;

[0009] The operation monitoring module is used to take the elements in the first training set as samples, calculate the probability distribution of the change rate of each operation parameter, and use the probability distribution function as the excitation of the state equation for testing. The product quality distribution equation is obtained from the probability distribution of the test results;

[0010] The fault reporting module is used to monitor the quality parameters of the product during the production process. The monitoring results within a fixed period form the third data set, calculate the residual between the distribution state of the parameters in the third data set and the product quality distribution equation, and report equipment faults when the residual is greater than the threshold;

[0011] The maintenance positioning module is used to generate an error detection space using the state vector constructed in the first training set. According to the change vector of the quality parameters of the product during the production process of each equipment, a vector verification sequence is formed, calculate the mapping vector of each element in the sequence in the error detection state space, and substitute the change speed of the included angle between adjacent mapping vectors into the probability function to output the probability of faults occurring in each process.

[0012] Further, the digital sensing module includes: a product detection unit and an equipment sensing unit;

[0013] The product detection unit is arranged at the end of the petrochemical production line and is used to detect the product quality parameters. The quality parameters include: output, fractionation temperature, purity, and carbon content ratio;

[0014] The equipment sensing unit is arranged in the petrochemical production equipment and is used to detect the operation parameters of the equipment.

[0015] Further, the state control module includes: a data de-noising unit and a model shaping unit;

[0016] The data de-noising unit is used to remove abnormal data in the data set using a data stream algorithm;

[0017] The model shaping unit is used to establish a state equation between the equipment operation parameters and the product quality parameters.

[0018] Further, the operation monitoring module includes: a probability estimation unit, an excitation input unit, and a quality distribution unit;

[0019] The probability estimation unit is used to estimate the distribution of the change rate of each equipment operation parameter using the kernel density estimation method;

[0020] The excitation input unit is used to substitute the distribution of the device operation parameters into the state equation to calculate the distribution of the product quality parameters;

[0021] The quality distribution unit is used to output the distribution probability function and the distribution range of the product quality parameters.

[0022] Further, the fault reporting module includes: a model residual unit and an anomaly judgment unit;

[0023] The model residual unit is used to calculate the generation probability of the measured product quality parameters in the distribution probability function;

[0024] The anomaly judgment unit is used to issue a fault alarm when the occurrence probability of the quality parameters of the measured product is lower than the threshold.

[0025] Further, the maintenance positioning module includes: a maintenance space unit, a parameter mapping unit, and a probability positioning unit;

[0026] The maintenance space unit is used to construct a maintenance coordinate system with the state vector of the product quality parameters as the basis;

[0027] The parameter mapping unit is used to analyze the first training set, generate the product state vectors of each process, and calculate their mapping vectors in the maintenance coordinate system;

[0028] The probability positioning unit is used to calculate the included angle between adjacent mapping vectors, substitute the change rate of the included angle into the distribution probability function, and output the probability of each device having a fault.

[0029] A digital-based fault prediction method for petrochemical equipment includes the following steps:

[0030] Step S1. Detect the quality parameters of the product at the end of the petrochemical production line, use the detection results as the first training set, record the working parameters of each petrochemical device, and use the recording results as the second training set;

[0031] Step S2. Construct the mapping relationship between the first training set and the second training set, use the elements in the first and second training sets as the state vector and the output vector respectively, solve the coefficient matrix, and then construct the state equation of the petrochemical production line;

[0032] Step S3. Use the elements in the first training set as samples, estimate the probability distribution function of the change rate of each working parameter, use the probability distribution function as the excitation to input the state equation, and obtain an output vector group, denoted as the product quality distribution function;

[0033] Step S4. Continuously monitor the quality parameters of the product, substitute the monitoring results into the product quality distribution function to obtain the occurrence rate of each quality parameter of the product, and report a device fault when the single occurrence rate is lower than the threshold, and then go to step S5;

[0034] Step S5. Construct a state space with the state vectors in Step S2. Replace the corresponding elements in the state vectors with the working parameters when the device fails to obtain a verification vector. Calculate the mapping vector of the verification vector in the state space, and use the dot product of the mapping vector and the verification vector as the failure risk output of the corresponding parameter.

[0035] Further, Step S1 includes:

[0036] Step S11. During the normal operation of the petrochemical equipment, detect the quality parameters of the product every fixed detection duration until the number of detections reaches the preset value n. The quality parameters include: output, fractionation temperature, purity, and carbon content ratio.

[0037] Step S12. Before the t0 duration of the first detection of the product, record all the working parameters of the petrochemical equipment every same detection duration until the number of records reaches n times, where t0 is the preset production delay, and the preset value n should not be less than the number of working parameters.

[0038] Step S13. Use the detection results of the product quality parameters as the first data set, and the record results of the petrochemical equipment working parameters as the second data set, and store the first data set and the second data set in the database.

[0039] Further, Step S2 includes:

[0040] Step S21. Establish a mapping relationship for the data with the same number in the first data set and the second data set. Convert each element in the second data set into the form of a state vector to obtain the set {X1, X2, …, Xn}, where Xn represents the state vector of the device at the nth detection, m represents the number of working parameters of the device, represents the mth working parameter of the device in the nth detection, and T is the transpose symbol;

[0041] Step S22. Convert the elements in the second data set into state output vectors, and the conversion result is the set {Y1, Y2, …, Yn}, where Yn represents the output vector of the product at the nth detection, c represents the number of quality parameter detection items, represents the detection result of the cth detection item of the product at the nth detection;

[0042] Step S23. Obtain the following equation according to the mapping relationship between the state vector and the output vector:

[0043]

[0044] Among them, A1, A2, … An are the coefficient matrices of the 1st to mth working parameters respectively. Solve the state equation to obtain the values of A1, A2, … An;

[0045] Step S24. Output the state equation of petrochemical equipment production, and the state equation is expressed as: Y = A1·x 1 + A2·x 2 + … + An·x n , where Y is the state output vector, and x 1 , x 2 , …, x n respectively represent the excitations brought by the 1st, 2nd, …, nth working parameters of the equipment.

[0046] Further, step S3 includes:

[0047] Step S31. Use the first training set as a sample and estimate the probability distribution of each working parameter by using the probability estimation method:

[0048]

[0049] Among them, F r (x) represents the probability distribution function corresponding to the rth equipment working parameter, t represents the detection duration, r represents the working parameter number, r ∈ {1, 2, …, m}, traverse all possible values of r to obtain the probability distribution functions of all working parameters;

[0050] Step S32. Substitute the probability distribution functions of each working parameter into the state equation to obtain the product quality distribution function Y(x), and the Y(x) = A1·F 1 (x) + A2·F 2 (x) + … + An·F m (x).

[0051] Further, step S4 includes:

[0052] Step S41. After the petrochemical equipment continues to work, continue to detect the quality parameters of the product, and record the detected quality parameters in the form of an output vector as YU, and the YU = [YU 1 , YU 2 , …, YU c T , where YU c represents the detection value of the cth quality parameter detection item. Substitute YU into the product quality distribution function to obtain the incidence set {P1, P2, …, Pc} of each quality parameter of the product, where Pc represents the occurrence probability of the cth quality parameter detection item;

[0053] ​Step S42. When one of the elements in the set {P1, P2, …, Pc} is less than the preset value P0, it is determined that there is an abnormality in the production of the petrochemical equipment, a fault declaration is made for the equipment, and the process proceeds to Step S5.

[0054] Further, Step S5 includes:

[0055] Step S51. Read the working parameters of the equipment at time t0 before the last product quality inspection, which are represented in vector form as [v1, v2, …, vu, …, vm] T , where vu represents the u-th working parameter read;

[0056] Step S52. Obtain all the state vectors in Step S2, construct an n-dimensional coordinate system with X1, X2, …, Xn as the basis, and obtain the state space;

[0057] Step S53. Replace the corresponding parameters in the state vector Xn with the working parameters read in Step S51 in sequence to obtain the verification vectors {H1, H2, …, Hu, …, Hm} of each working parameter, where Hu represents the verification vector corresponding to the u-th working parameter, and it satisfies

[0058] Step S53. Perform a mapping transformation on the verification vector Hu in the state space to obtain the mapping vector hu, calculate the included angle e between the verification vector Hu and the mapping vector hu, and output e as the fault risk of the m-th working parameter;

[0059] Step S54. Arrange all the working parameters of the equipment in descending order of fault risk, list the equipment components corresponding to each working parameter after each working parameter, and send the arrangement result to the maintenance department.

[0060] Compared with the prior art, the beneficial effects achieved by the present invention are:

[0061] 1. The present invention can detect the quality parameters of the product at the output end of the petrochemical equipment, obtain the operating parameters of each equipment at the control end, digitalize the production process through data training, construct the state equation of the production process, and can monitor and control the production process in real time, optimize the equipment utilization rate, reduce the downtime, and improve the production efficiency.

[0062] 2. The present invention can calculate the probability distribution of the change rate of the operating parameters through the kernel density estimation method, substitute the probability distribution function for verification according to the current operating parameters, so as to identify the abnormal working conditions of the petrochemical equipment, can timely discover the potential problems of the equipment parameters in the production process, give early warnings of possible faults, and ensure the normal operation of the equipment.

[0063] 3. The present invention can construct an error detection state space according to the state equation, calculate the change speed of the operating parameters by calculating the vector angle between adjacent mapping vectors based on the mapping vectors of the quality parameters of the product in the error detection state space during the production process of each device, so as to obtain the probability of a fault occurring in each process, quickly determine the problem, shorten the maintenance downtime, effectively save the maintenance cost, and improve the equipment utilization rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] The drawings are used to provide a further understanding of the present invention, and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention, and do not constitute a limitation to the present invention. In the drawings:

[0065] Figure 1 is a schematic structural diagram of a digital-based petrochemical equipment fault prediction system of the present invention;

[0066] Figure 2 is a schematic step diagram of a digital-based petrochemical equipment fault prediction method of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0067] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0068] Please refer to Figure 1 , the present invention provides a technical solution: a digital-based petrochemical equipment fault prediction system, including: a digital sensing module, a state control module, an operation monitoring module, a fault reporting module, and a maintenance positioning module;

[0069] The digital sensing module includes a product detection device at the end of the petrochemical production line and a parameter detection device arranged in the production equipment. The product detection device is used to detect the quality parameters of the product at the output end of the petrochemical equipment and feedback the detection result to the data processing center. The parameter detection device is used to obtain all the operation parameters of each production equipment in real time by using sensors;

[0070] The digital sensing module includes: a product detection unit and a device sensing unit;

[0071] The product detection unit is arranged at the end of the petrochemical production line and is used to detect the product quality parameters. The quality parameters include: output, fractionation temperature, purity, and carbon content ratio;

[0072] The device sensing unit is arranged in the petrochemical production equipment and is used to detect the operation parameters of the equipment.

[0073] The state control module is used to select the production equipment operation parameters and product quality parameters during the normal operation of the equipment, respectively form the first training set and the second training set. After screening the data in the training set, taking each quality parameter of the product as the state quantity and the operation parameters of the equipment as the excitation, a state equation of the petrochemical production line is established;

[0074] The state control module includes: a data de-noising unit and a model shaping unit;

[0075] The data de-noising unit is used to remove abnormal data in the data set by using the data flow algorithm;

[0076] The model shaping unit is used to establish a state equation between the equipment operation parameters and the product quality parameters.

[0077] The operation monitoring module is used to take the elements in the first training set as samples, calculate the probability distribution of the change rates of each operation parameter, and use the probability distribution function as the excitation of the state equation for testing. The product quality distribution equation is obtained from the probability distribution of the test results;

[0078] The operation monitoring module includes: a probability estimation unit, an excitation input unit, and a quality distribution unit;

[0079] The probability estimation unit is used to estimate the distribution of the change rates of each equipment operation parameter by using the kernel density estimation method;

[0080] The excitation input unit is used to substitute the distribution of the equipment operation parameters into the state equation and calculate the distribution of the product quality parameters;

[0081] The quality distribution unit is used to output the distribution probability function and the distribution range of the product quality parameters.

[0082] The fault reporting module is used to monitor the quality parameters of the products during the production process. The monitoring results within a fixed period form the third data set. Calculate the residual between the distribution state of the parameters in the third data set and the product quality distribution equation. When the residual is greater than the threshold, a device fault report is made;

[0083] The fault reporting module includes: a model residual unit and an anomaly judgment unit;

[0084] The model residual unit is used to calculate the generation probability of the measured product quality parameters in the distribution probability function;

[0085] The anomaly judgment unit is used to issue a fault alarm when the occurrence probability of the quality parameters of the measured product is lower than the threshold.

[0086] The maintenance positioning module is used to generate an error detection space by using the state vectors constructed in the first training set, form a vector verification sequence according to the change vectors of the quality parameters of the product during the production process of each device, calculate the mapping vectors of the elements in the sequence in the error detection state space, and substitute the change speed of the included angle between adjacent mapping vectors into the probability function, and then output the probability of a failure occurring in each process.

[0087] The maintenance positioning module includes: a maintenance space unit, a parameter mapping unit, and a probability positioning unit;

[0088] The maintenance space unit is used to construct a maintenance coordinate system with the state vectors of the product quality parameters as the basis;

[0089] The parameter mapping unit is used to analyze the first training set, generate the product state vectors of each process, and calculate their mapping vectors in the maintenance coordinate system;

[0090] The probability positioning unit is used to calculate the included angle between adjacent mapping vectors, substitute the change rate of the included angle into the distribution probability function, and output the probability of a failure occurring in each device.

[0091] As Figure 2 shown, a digital-based fault prediction method for petrochemical equipment includes the following steps:

[0092] Step S1. Detect the quality parameters of the product at the end of the petrochemical production line, and use the detection results as the first training set, and record the working parameters of each petrochemical device, and use the recording results as the second training set;

[0093] Step S1 includes:

[0094] Step S11. During the normal operation of the petrochemical device, detect the quality parameters of the product every fixed detection duration until the number of detections reaches the preset value n. The quality parameters include: output, fractionation temperature, purity, and carbon content ratio;

[0095] Step S12. Before the t0 duration of the first detection of the product, record all the working parameters of the petrochemical device every same detection duration until the number of records reaches n times, where t0 is the preset production delay, and the preset value n should not be less than the number of working parameters;

[0096] Step S13. Use the detection results of the product quality parameters as the first data set, and the recording results of the petrochemical device working parameters as the second data set, and store the first data set and the second data set in the database.

[0097] Step S2. Construct the mapping relationship between the first training set and the second training set, use the elements in the first and second training sets as the state vectors and output vectors respectively, solve the coefficient matrix, and then construct the state equation of the petrochemical production line;

[0098] Step S2 includes:

[0099] Step S21. Establish a mapping relationship for the data with the same numbering in the first dataset and the second dataset, convert each element in the second dataset into the form of a state vector, and obtain the set {X1, X2, …, Xn}, where Xn represents the state vector of the device at the nth detection, m represents the number of device working parameters, represents the mth working parameter of the device in the nth detection, and T is the transpose symbol;

[0100] Step S22. Convert the elements in the second dataset into state output vectors, and the conversion result is the set {Y1, Y2, …, Yn}, where Yn represents the output vector of the product at the nth detection, c represents the number of quality parameter detection items, represents the detection result of the cth detection item of the product at the nth detection;

[0101] Step S23. Obtain the following equation according to the mapping relationship between the state vector and the output vector:

[0102]

[0103] where A1, A2, …, An are the coefficient matrices of the 1st to mth working parameters respectively, solve the state equation, and obtain the values of A1, A2, …, An;

[0104] Step S24. Output the state equation of the petrochemical equipment production, and the state equation is expressed as: Y = A1·x 1 + A2·x 2 + … + An·x n , where Y is the state output vector, and x 1 , x 2 , …, x n represent the excitations brought by the 1st, 2nd, …, nth working parameters of the device respectively.

[0105] Step S3. Use the elements in the first training set as samples, estimate the probability distribution function of the change rate of each working parameter, and use the probability distribution function as the excitation to input into the state equation to obtain an output vector group, denoted as the product quality distribution function;

[0106] Step S3 includes:

[0107] Step S31. Use the first training set as a sample and estimate the probability distribution of each working parameter by using the probability estimation method:

[0108]

[0109] Among them, F r (x) represents the probability distribution function corresponding to the r-th device operating parameter, t represents the detection duration, r represents the operating parameter number, r ∈ {1, 2, …, m}, and by traversing all possible values of r, the probability distribution functions of all operating parameters are obtained;

[0110] Step S32. Substitute the probability distribution functions of each operating parameter into the state equation to obtain the product quality distribution function Y(x), where Y(x) = A1·F 1 (x) + A2·F 2 (x) + … + An·F m (x).

[0111] Step S4. Continuously monitor the quality parameters of the product, substitute the monitoring results into the product quality distribution function to obtain the incidence rates of each quality parameter of the product, and declare a device failure when the single incidence rate is lower than the threshold, then go to step S5;

[0112] Step S4 includes:

[0113] Step S41. After the petrochemical equipment continues to operate, continue to detect the quality parameters of the product, and record the detected quality parameters in the form of an output vector as YU, where YU = [YU 1 , YU 2 , …, YU c T , where YU c represents the detected value of the c-th quality parameter detection item, substitute YU into the product quality distribution function to obtain the incidence rate set {P1, P2, …, Pc} of each quality parameter of the product, where Pc represents the occurrence probability of the c-th quality parameter detection item;

[0114] Step S42. When one of the elements in the set {P1, P2, …, Pc} is less than the preset value P0, it is judged that there is an abnormality in the production of the petrochemical equipment, declare a failure of the equipment, and go to step S5.

[0115] Step S5. Construct a state space with the state vector in step S2, replace according to the corresponding elements in the state vector with each operating parameter when the equipment fails to obtain a verification vector, calculate the mapping vector of the verification vector in the state space, and take the dot product of the mapping vector and the verification vector as the failure risk output of the corresponding parameter.

[0116] Step S5 includes:

[0117] Step S51. Read the operating parameters of the equipment at time t0 before the last product quality inspection, and represent them in vector form as [v1, v2, …, vu, …, vm] T , where vu represents the u-th operating parameter read;​

[0118] Step S52. Obtain all the state vectors in Step S2, construct an n-dimensional coordinate system with X1, X2, …, Xn as the bases to obtain the state space;

[0119] Step S53. Sequentially replace the corresponding parameters in the state vector Xn with the working parameters read in Step S51 to obtain the verification vectors {H1, H2, …, Hu, …, Hm} of each working parameter, where Hu represents the verification vector corresponding to the u-th working parameter, and satisfies

[0120] Step S53. Perform a mapping transformation on the verification vector Hu in the state space to obtain the mapping vector hu, calculate the included angle e between the verification vector Hu and the mapping vector hu, and output e as the fault risk of the m-th working parameter;

[0121] Step S54. Arrange all the working parameters of the device in descending order of fault risk, list the device components corresponding to each working parameter after each working parameter, and send the arrangement result to the maintenance department.

[0122] Example: A petrochemical device has 3 working parameters of pressure, rotation speed, and temperature. The output of 2 detection items of the product output and fractionation temperature is detected once every 1 minute for a total of 3 times. The obtained output vectors are

[12] ,

[21] , and

[22] respectively. Among them, the working parameter corresponding to

[12] is

[112] . Then list the equation A1 + A2 + 2A3 =

[12] , and so on. After listing 3 equations, solve for A1 = [0.5, 0], A2 = [0.2, 1], A3 = [0.3, 1]. Then the state equation of the device is: Y = A1·x 1 +A2·x 2 +A2·x 2 ;

[0123] Substitute the probability distribution into the state equation of the device to obtain the product quality distribution function Y(x). At this time, the monitored output vector is

[30] . After substituting it into Y(x), the obtained probability is [0.01 0.12]. The first item is lower than the threshold of 0.1, so it is determined that the device is abnormal.

[0124] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprising", "including" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device.

[0125] Finally, it should be noted that the above are only preferred embodiments of the present invention and are not used to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A petrochemical equipment fault prediction method based on digitization, characterized in that: The method comprises the following steps: Step S1. Detect the quality parameters of the product at the end of the petrochemical production line, and use the detection results as the first data set, record the working parameters of each petrochemical equipment, and record the results as the second data set; Step S2. Construct a mapping relationship between the first data set and the second data set, use the elements in the first data set as the output vector, and the elements in the second data set as the state vector, and after solving the coefficient matrix, construct the state equation of the petrochemical production line; Step S3. Taking the elements in the second data set as samples, using the probability estimation method, estimating the probability distribution of each working parameter, using the probability distribution function of each working parameter as the excitation input state equation, and obtaining the product mass distribution function; Step S4. Continuously monitor the quality parameters of the product, substitute the monitored quality parameters into the product quality distribution function in the form of output vectors, obtain the incidence rate of each quality parameter of the product, and report the equipment failure when the single occurrence rate is lower than the threshold, and go to step S5; Step S5. Construct a state space with the state vector in step S2, replace the corresponding elements in the state vector with the various working parameters when the equipment fails, obtain a verification vector, calculate the mapping vector of the verification vector in the state space, and output the angle between the mapping vector and the verification vector as the failure risk of the corresponding working parameter.

2. The method for predicting faults of petrochemical equipment based on digitization according to claim 1 is characterized in that: Step S1 includes: Step S11. During the normal operation of the petrochemical equipment, the quality parameters of the product are tested once every fixed test time until the number of tests reaches a preset value n, wherein the quality parameters include: yield, fractionation temperature, purity and carbon content ratio; Step S12. Before the time t0 when the product is tested for the first time, all the working parameters of the petrochemical equipment are recorded once every the same testing time, until the number of records reaches n times, where t0 is the preset production delay, and the preset value n must not be less than the number of working parameters; Step S13: The detection results of the product quality parameters are used as the first data set, and the recording results of the petrochemical equipment working parameters are used as the second data set. The first data set and the second data set are stored in a database.

3. The method for predicting faults of petrochemical equipment based on digitization according to claim 2 is characterized in that: Step S2 includes: Step S21. Establish a mapping relationship between the data with the same number in the first data set and the second data set, convert each element in the second data set into a state vector form, and obtain a set {X1, X2, ..., Xn}, where , Xn represents the state vector of the device at the nth detection, m represents the number of device working parameters, represents the mth working parameter of the device in the nth detection, and T is the transposed symbol; Step S22. Convert the elements in the first data set into an output vector, and the conversion result is a set {Y1, Y2, ..., Yn}, where , Yn represents the output vector of the product at the nth detection, c represents the number of quality parameter detection items, Represents the test result of the cth test item of the product during the nth test; Step S23. Obtain the following equation based on the mapping relationship between the state vector and the output vector: ； Among them, A1, A2, ...Am are the coefficient matrices of the 1st to mth working parameters respectively. Solve the state equation to obtain the values ​​of A1, A2, ...Am; Step S24: Output the state equation of petrochemical equipment production, the state equation is expressed as: Y=A1·x1+A2·x2+…+Am·x m , where Y is the state output vector, x1, x2, …, x m They represent the excitations brought by the 1st, 2nd, …, mth working parameters of the equipment respectively.

4. The method for predicting faults of petrochemical equipment based on digitization according to claim 3 is characterized in that: Step S3 includes: Step S31. Using the second data set as a sample, estimate the probability distribution of each working parameter using a probability estimation method: ; Among them, F r (x) represents the probability distribution function corresponding to the working parameter of the rth device, t represents the detection time, r represents the working parameter number, r∈{1, 2, …, m}, traverse all values ​​of r to obtain the probability distribution function of all working parameters; Step S32. Substitute the probability distribution function of each working parameter into the state equation to obtain the product mass distribution function Y(x), where Y(x)=A1·F1(x)+A2·F2(x)+…+An·F m (x); Step S4 includes: Step S41. After the petrochemical equipment continues to work, the quality parameters of the product are continuously detected, and the detected quality parameters are recorded as YU in the form of an output vector, where YU=[YU1, YU2, ..., YU c ] T , among which YU c represents the detection value of the c-th quality parameter detection item, substitute YU into the product quality distribution function, and obtain the occurrence rate set of various quality parameters of the product {P1, P2, …, Pc}, where Pc represents the occurrence rate of the quality parameter of the c-th quality parameter detection item; Step S42. When one of the elements in the set {P1, P2, ..., Pc} is less than the preset value P0, it is determined that there is an abnormality in the production of the petrochemical equipment, a fault is reported for the equipment, and the process goes to step S5.

5. A digital petrochemical equipment fault prediction method according to claim 4, characterized in that: Step S5 includes: Step S51. Read the various working parameters of the equipment at time t0 before the last product quality inspection, expressed in vector form as [v1, v2, ..., vu, ..., vm] T , where vu represents the u-th working parameter read; Step S52. Obtain all state vectors in step S2, construct an n-dimensional coordinate system with X1, X2, ..., Xn as the basis, and obtain the state space; Step S53. Replace the working parameters read in step S51 with the corresponding parameters in the state vector Xn in turn to obtain the verification vectors {H1, H2, ..., Hu, ..., Hm} of each working parameter, where Hu represents the verification vector corresponding to the u-th working parameter and satisfies ; Step S54. Map the verification vector Hu in the state space to obtain the mapping vector hu, calculate the angle e between the verification vector Hu and the mapping vector hu, and output e as the failure risk of the mth working parameter; Step S55. Arrange all the working parameters of the equipment in descending order of failure risk, list the equipment components corresponding to the working parameters after each working parameter, and send the arrangement results to the maintenance department.

Images