Sucker-rod pump oil extraction process fault diagnosis system and method based on numerical simulation

Abstract

The invention provides a sucker-rod pump oil extraction process fault diagnosis system and method based on numerical simulation, and relates to the technical field of sucker-rod pump oil extraction processes. On one hand, the sucker-rod pump oil extraction process fault diagnosis system based on numerical simulation is established and used for fault diagnosis in the sucker-rod pump oil extraction process; on the other hand, the invention provides a sucker-rod pump oil extraction process fault diagnosis method based on numerical simulation, by establishing a mechanism model and a calculation model of the sucker-rod pump oil extraction process, a suspension point indicator diagram generation method based on numerical simulation is provided, the problem that actually-measured suspension point indicator diagram data is unbalanced is solved, and the fault diagnosis accuracy is improved. The defect that an existing generation method based on data driving is poor in reliability is overcome; and on the basis, a plurality of feature points and feature parameters are introduced to extract working condition information in the suspension point indicator diagram, so that a sucker-rod pump oil extraction process fault diagnosis system based on numerical simulation is established, and the interpretability of an existing fault diagnosis method is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of fault diagnosis in the process of rod pump oil production, and particularly relates to a fault diagnosis system and method for the process of rod pump oil production based on numerical simulation. Background Art

[0002] The patent application with the application number CN202111041748.2 proposed "a fault diagnosis prediction method and device for the dynamometer card of an oil well". The implementation scheme of this patent is as follows: collect the dynamometer card data of the polished rod in the process of rod pump oil production and convert it into an image form of 224×224 pixels; on this basis, use the Focal Loss function to correct the imbalance problem of the polished rod dynamometer card sample data, and propose a fault diagnosis model based on ResNet50. Although this technology can realize the fault diagnosis in the process of rod pump oil production, it depends on high-quality data. In addition, due to the lack of incorporation of mechanism information, the interpretability of the fault diagnosis model is poor.

[0003] The patent application with the application number CN202210362470.7 proposed "a fault diagnosis method for dynamometer cards based on a generative adversarial neural network". The implementation scheme of this patent is as follows: collect the dynamometer card data of the polished rod in the process of rod pump oil production, and use a generative adversarial network to enrich the fault condition samples with a small number to balance the number of samples in different conditions; on this basis, use the measured dynamometer card data set and the generated dynamometer card data set to propose a fault diagnosis scheme based on the XGBoost model. Since this technology does not consider mechanism knowledge in the generation model, the reliability of the generated dynamometer card is poor and further discrimination is required; in addition, the fault diagnosis scheme based on the XGBoost model does not incorporate mechanism information, resulting in poor interpretability of the fault diagnosis model.

[0004] The existing technology mainly uses data-driven methods to generate dynamometer cards of polished rods or design fault diagnosis models. The data-driven generation method has poor reliability and requires further discrimination of the generated samples; at the same time, the data-driven fault diagnosis model depends on high-quality data and has poor interpretability. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a fault diagnosis system and method for the process of rod pump oil production based on numerical simulation in view of the above-mentioned deficiencies of the prior art. By establishing a mechanism model and a calculation model for the process of rod pump oil production, a method for generating dynamometer cards of polished rods based on numerical simulation is proposed to solve the problems of unbalanced sample data of measured dynamometer cards of polished rods and poor reliability of data-driven generation methods; on this basis, multiple feature points and feature parameters are introduced to extract the working condition information in the dynamometer card of the polished rod, and then a fault diagnosis system for the process of rod pump oil production based on numerical simulation is established to improve the interpretability of the existing fault diagnosis methods.

[0006] To solve the above technical problems, the technical solution adopted by the present invention is as follows:

[0007] On the one hand, the present invention provides a fault diagnosis system for the rod pumping production process based on numerical simulation, including: a fault diagnosis knowledge base construction module, a fault diagnosis reasoning module, and a human-computer interaction module;

[0008] The fault diagnosis knowledge base construction module is used to establish and store fault condition diagnosis rules and determination thresholds, and construct a fault diagnosis knowledge base;

[0009] The fault diagnosis reasoning module is used to process the rod pump operation data input by the user, establish a fault diagnosis reasoning mechanism, and perform fault diagnosis on the rod pumping production process based on the constructed fault diagnosis knowledge base;

[0010] The human-computer interaction module is used to provide an interface for the fault diagnosis system of the rod pumping production process based on numerical simulation to interact with the user. The human-computer interaction module receives the rod pump operation data input by the user and outputs and displays the fault diagnosis results of the fault diagnosis reasoning module.

[0011] On the other hand, the present invention provides a fault diagnosis method for the rod pumping production process based on numerical simulation. By analyzing the operation mechanism of the sucker rod pump and the generation mechanism of fault conditions, a mechanism model of the rod pumping production process under various conditions is established; on this basis, a calculation model of the rod pumping production process is established using the precise integration method, and a method for generating a set of polished rod dynamometer cards is established by simulating various rod pumping production processes; multiple characteristic parameters reflecting fault condition information are introduced to establish a fault diagnosis knowledge base for the rod pumping production process, and the diagnosis of the rod pumping production process conditions is carried out, including the following steps:

[0012] Step 1: By analyzing the operation mechanism of the sucker rod pump and the generation mechanism of fault conditions, establish a mechanism model of the rod pumping production process under various conditions;

[0013] Step 1.1: Analyze the operation mechanism of the sucker rod, and establish a dynamic mechanism model of the sucker rod;

[0014] By analyzing the force on the micro-element of the sucker rod and using Newton's second law and Hooke's law, a dynamic mechanism model of the sucker rod is established as shown in the following formula:

[0015]

[0016] Where, x is the distance between the cross-section of the sucker rod and the polished rod, t is the time, u(x, t) is the displacement of the cross-section of the sucker rod at x at time t relative to the casing, v is the velocity of the sound wave passing through the sucker rod, c is the damping coefficient, g is the acceleration due to gravity, Er is the Young's modulus of the sucker rod, ρ r is the density of the sucker rod, γ is the dimensionless damping coefficient, and L is the length of the sucker rod;

[0017] Step 1.2: Analyze the operation mechanism of the oil pump and establish a mechanism model of the oil pump;

[0018] Based on the plunger displacement and plunger load of the oil pump, establish a mechanism model of the oil pump, including:

[0019] Taking the lowest position of the plunger as the zero initial position, the plunger displacement is shown by the following formula:

[0020]

[0021] where x p (t) is the plunger displacement at time t, S p is the plunger stroke, and N is the pumping frequency;

[0022] During the oil production process, the top and bottom of the plunger are simultaneously subjected to pressure. Among them, the pressure of the crude oil in the tubing is applied to its top, and the pump barrel pressure is applied to its bottom. Define the positive direction of the plunger load as the downward direction. The plunger load is shown by the following formula:

[0023] F p (t) = p d (A p - A rd ) - p(t)A p (3)

[0024] where F p (t) is the plunger load at time t, p d is the discharge pressure, A p is the cross-sectional area of the plunger, A rd is the cross-sectional area of the sucker rod connected to the plunger, and p(t) is the pump barrel pressure at time t;

[0025] The change of the pump barrel pressure includes two stages: the first stage is from the closing of the traveling valve to the opening of the traveling valve, and the second stage is from the opening of the traveling valve to the closing of the traveling valve. In the first stage of the change of the pump barrel pressure, the compression coefficient is introduced to correct the ideal gas law. In the second stage of the change of the pump barrel pressure, when the traveling valve opens, the pump barrel is connected to the tubing, and the crude oil in the pump barrel is discharged into the tubing through the traveling valve. The pump barrel pressure is equal to the discharge pressure. Then the pump barrel pressure p(t) is shown by the following formula:

[0026]

[0027] where V g(t) is the gas volume at time t, Z(t) is the compressibility factor at time t, n(t) is the number of moles at time t, R is the gas constant, and T is the absolute temperature;

[0028] In the first stage of the pump barrel pressure change, the pump barrel pressure is affected by the gas volume V g (t), the compressibility factor Z(t), and the number of moles n(t), specifically as follows:

[0029] During the closing of the traveling valve, both gas and crude oil are contained in the pump barrel. The gas volume is equal to the pump barrel volume minus the volume of the crude oil in the pump barrel. The crude oil in the pump barrel enters through the fixed valve, and its volume is the integral of the flow rate with respect to time. The gas volume V g (t) is shown by the following formula:

[0030]

[0031] where V s is the clearance volume, V sl is the volume of the crude oil in the clearance at the initial moment, and q(τ) is the flow rate of the crude oil through the fixed valve at time τ;

[0032] At time t, when the pump barrel pressure is greater than the submergence pressure, the fixed valve closes, and the flow rate of the crude oil through the fixed valve is zero; when the pump barrel pressure is less than or equal to the submergence pressure, the fixed valve opens, and the flow rate q(t) of the crude oil through the fixed valve can be obtained from the fixed valve orifice flow rate formula, as shown by the following formula:

[0033]

[0034] where μ is the fixed valve flow coefficient, A s is the cross-sectional area of the fixed valve, p s is the submergence pressure, ρ l is the density of the crude oil;

[0035] The gas in the pump barrel is a mixture of hydrocarbon gases. The compressibility factor Z(t) of the gas in the pump barrel is represented by the compressibility factor of natural gas, as shown by the following formula:

[0036] Z 3 (t) - Z 2 (t) + (0.0193p(t) - 0.0001p 2 (t))Z(t) - 0.0004p 2 (t) = 0 (7)

[0037] During the closing of the traveling valve, the pump barrel contains gas. The change in the number of moles dn(t) can be divided into three parts. The first part dn1(t) is the number of moles of dissolved gas released by the crude oil in the pump. The second part dn2(t) is the number of moles of dissolved gas released when the crude oil enters the pump due to the pressure drop dp. The third part dn3(t) is the number of moles of gas entering the pump barrel. From time t to time t+dt, the change in the number of moles n(t) is shown by the following formula:

[0038] dn(t) = dn1(t) + dn2(t) + dn3(t) (8)

[0039] Among them,

[0040]

[0041] Among them, T0, T a are the temperature under standard conditions and the actual temperature respectively, α is the solubility coefficient, f w is the water cut, R p is the volume ratio of gas to crude oil in the formation, B0 is the volume coefficient of the crude oil, V l (t) is the volume of the crude oil;

[0042] Step 1.3: Analyze the generation mechanism of the fault conditions and establish the mechanism models for various fault conditions;

[0043] The fault conditions include insufficient liquid supply, gas influence, fixed valve leakage, traveling valve leakage, and rod breakage. The mechanism models established for different fault conditions are respectively:

[0044] 1) The mechanism model for the insufficient liquid supply fault is:

[0045] p′ s < p s (12)

[0046] Among them, p′ s is the submerged pressure when the sucker rod pump is in the insufficient liquid supply fault condition;

[0047] 2) The mechanism model for the gas influence fault is:

[0048] n′(0) > n(0), R′ p > R p (13)

[0049] Among them, n ′ (0) is the number of moles of gas in the clearance when the sucker rod pump is in the gas influence fault condition at the initial moment, n(0) is the number of moles of gas in the clearance when the sucker rod pump is in the normal condition at the initial moment, R′ p is the volume ratio of gas to crude oil in the formation when the sucker rod pump is in the gas influence fault condition;

[0050] 3) The mechanism model of the fixed valve leakage fault is as follows:

[0051]

[0052] Among them, q s (t) is the flow rate when the sucker rod pump is in the fixed valve leakage fault condition at time t, ξ is the leakage coefficient, and A l is the leakage area;

[0053] 4) The mechanism model of the traveling valve leakage fault is as follows:

[0054]

[0055] Among them, q t (t) is the flow rate when the sucker rod pump is in the traveling valve leakage fault condition at time t, D is the diameter of the pump barrel, μ1 is the kinematic viscosity of the crude oil, and L p is the length of the plunger, and δ is the radial clearance between the plunger and the pump barrel;

[0056] 5) The mechanism model of the rod breakage fault is as follows:

[0057] L′ < L, F pr (t) = 0 (16)

[0058] Among them, L′ is the length of the sucker rod part moving with the polished rod under the rod breakage fault condition, and F pr (t) is the load at the bottom of the broken rod at time t;

[0059] Step 2: Establish a calculation model for the rod pumping production process, establish a generation method for the polished rod dynamometer card dataset, and generate the polished rod dynamometer card dataset;

[0060] Step 2.1: Based on the mechanism model of the rod pumping production process, determine the initial and boundary conditions, including the boundary conditions and initial conditions of the sucker rod;

[0061] Step 2.1.1: Determine the boundary conditions of the sucker rod dynamic mechanism model;

[0062] The boundary conditions of the sucker rod dynamic mechanism model are the motion law of the polished rod of the pumping unit and the load at the plunger of the sucker rod pump. Among them, the motion of the polished rod of the pumping unit is similar to a simple harmonic motion, which is determined by the geometric shape of the surface pumping unit and is shown by the following formula:

[0063]

[0064] Among them, S r is the stroke of the polished rod of the pumping unit;

[0065] The load at the plunger of the sucker rod pump is:

[0066]

[0067] Among them, u(L, t) is the displacement of the bottom of the sucker rod at time t, and A r is the cross-sectional area of the sucker rod;

[0068] Step 2.1.2: Determine the initial conditions of the dynamic mechanism model of the sucker rod;

[0069] Take the displacement and velocity of the sucker rod at the initial moment as the initial conditions of the dynamic mechanism model of the sucker rod, as shown in the following formula:

[0070]

[0071] Among them, φ(x) is the initial displacement of the sucker rod at cross-section x, and ψ(x) is the initial velocity of the sucker rod at cross-section x;

[0072] Step 2.2: Use the precise integration method to establish a first-order differential equation of the dynamic mechanism model of the sucker rod in the rod pumping production process as the calculation model of the rod pumping production process, and use this model to calculate the polished rod displacement and the polished rod load, and generate a polished rod dynamometer card;

[0073] Step 2.2.1: Discretize the spatial variable and establish a first-order differential equation of the dynamic mechanism model of the sucker rod;

[0074] Take the top of the sucker rod as the reference position, divide the sucker rod into num intervals with a step size of h, and the corresponding cross-section positions are x j = jh, j = 0, 1, 2,..., num; at the cross-section positions of x1, x2, x3,..., x num-1 approximate the differential operation of x in the dynamic mechanism model of the sucker rod as a difference operation, as shown in the following formula:

[0075]

[0076] Among them, u(x j , t) is the displacement of the cross-section of the sucker rod at x j relative to the casing at time t;

[0077] The first-order differential equation of the dynamic mechanism model of the sucker rod is shown in the following formula:

[0078]

[0079] Among them, V(t) is a (2num - 2)×1 matrix, H is a (2num - 2)×(2num - 2) matrix, f(t) is a (2num - 2)×1 matrix, and E num-1 is an (num - 1)×(num - 1) identity matrix, and 0 (num-1)×1is a zero matrix of order (num - 1)×1, K is a matrix of order (num - 1)×(num - 1), and r(t) is a matrix of order (num - 1)×1, as shown in the following formula:

[0080]

[0081] Step 2.2.2: Discretize the time variable and calculate the solution of the first-order differential equation of the sucker rod dynamic mechanism model;

[0082] Taking the sucker rod at the bottom dead center as the initial moment, divide a stroke into m intervals with a step size of Δ, corresponding to the nodes t k = kΔ, k = 0, 1, 2, …, m. Using the precise integration method, obtain the solution V(t k+1 ) of the first-order differential equation of the sucker rod dynamic mechanism model at time t k+1 as follows:

[0083]

[0084] where V(t0) is a matrix of order (2num - 2)×1, and both R0(t k ) and R1(t k ) are matrices of order (2num - 2)×1;

[0085] Step 2.2.3: Calculate the polished rod displacement and the polished rod load, and generate the polished rod dynamometer card;

[0086] The polished rod dynamometer card is a graphical tool for analyzing the working state of the pumping unit, with the polished rod displacement and the polished rod load as the horizontal and vertical coordinates respectively. The polished rod displacement S(t k ) at time t k is shown in the following formula:

[0087]

[0088] where u(x0,t k ) is the displacement of the cross-section at x0 of the sucker rod relative to the casing at time t k ;

[0089] t k The polished rod load P(t k ) at time t

[0090]

[0091] is shown in the following formula: k where u(x1,t k ) is the displacement of the cross-section at x1 of the sucker rod relative to the casing at time t

[0092] Step 2.3: Simulate different working conditions of the rod pumping oil production process, generate the polished rod dynamometer cards of the rod pumping oil production process under different working conditions based on the calculation model of the rod pumping oil production process, and construct a simulated polished rod dynamometer card dataset;

[0093] Step 3: Introduce multiple characteristic points and characteristic parameters to extract the working condition information in the polished rod dynamometer card, and establish a fault diagnosis knowledge base for the rod pumping oil production process, including the determination rules and determination thresholds for each fault working condition;

[0094] Step 3.1: Based on the corresponding relationship between the polished rod dynamometer card and the working state of the sucker rod pump, introduce multiple characteristic points, and divide the polished rod dynamometer card into a loading section, a high-load section, an unloading section, and a low-load section;

[0095] Step 3.1.1: Preprocess the polished rod displacement data and polished rod load data included in the polished rod dynamometer card dataset;

[0096] Normalize the polished rod displacement data and polished rod load data included in the polished rod dynamometer card dataset to the interval [0, 1], as shown in the following formula:

[0097]

[0098] where S(t k ) and P(t k ) are the polished rod displacement value and polished rod load value of the polished rod dynamometer card at time t k respectively, are the maximum and minimum values of the polished rod dynamometer card at time t k after normalization, S max and S min are the maximum and minimum values of the polished rod displacement in the polished rod dynamometer card data respectively, and P max and P min are the maximum and minimum values of the polished rod load in the polished rod dynamometer card data respectively;

[0099] Step 3.1.2: Introduce multiple characteristic points and divide the polished rod dynamometer card data into a loading section, a high-load section, an unloading section, and a low-load section;

[0100] Set the data interval of the polished rod load to [0, P down and [P up , 1.0], where P down is the right endpoint of the data interval for determining the end point of loading and the start point of unloading, and P up is the left endpoint of the data interval for determining the end point of unloading and the start point of loading. Select the first minimum value point in the data interval [0, P down as the end point of loading The data interval [0, P downThe last minimum point is the starting point of unloading The data interval [P up , 1.0], the first maximum point is the ending point of unloading And the data interval [P up , 1.0], the last maximum point is the starting point of load increase As characteristic points, where Is the abscissa of the ending point of load increase Is the ordinate of the ending point of load increase Is the abscissa of the starting point of unloading Is the ordinate of the starting point of unloading Is the abscissa of the ending point of unloading Is the ordinate of the ending point of unloading Is the abscissa of the starting point of load increase Is the abscissa of the starting point of load increase. According to the selected characteristic points, the polished rod dynamometer card data is divided into a load increase section, a high load section, an unloading section, and a low load section;

[0101] Step 3.2: Introduce multiple characteristic parameters to characterize the fault conditions, and establish the determination rules for each fault condition in the fault diagnosis knowledge base;

[0102] Select multiple characteristic parameters to characterize the fault conditions, including the difference P between the maximum polished rod load and the minimum polished rod load before normalization c , the displacement between the ending point of load increase and the starting point of unloading after normalization The displacement between the ending point of unloading and the starting point of load increase after normalization The slope k of the straight line connecting the starting point of load increase and the ending point of load increase after normalization gl , the slope k of the straight line connecting the starting point of unloading and the ending point of unloading after normalization gu , the data point where the polished rod load in the load increase section after normalization is 0.5 To the distance from the straight line connecting the starting point of load increase and the ending point of load increase

[0103] Use characteristic parameters to characterize each fault condition, and establish the determination rules for each fault condition in the fault diagnosis knowledge base, specifically including:

[0104] Rule 1: If P c ≤β, where β is the determination threshold for the rod break condition, then determine that the current condition is the rod break condition;

[0105] Rule 2: If P c >β, ζ is the determination threshold for the normal condition, then determine that the current condition is the normal condition;

[0106] Rule 3: If P c >β, k gl / k gu <λ1, k gl <λ2, where λ1 and λ2 are the determination thresholds for the fixed valve leakage condition, then it is determined that the current condition is a fixed valve leakage;

[0107] Rule 4: If P c > β, k gl / k gu <η1, k gu <η2, where η1 and η2 are the determination thresholds for the traveling valve leakage condition, then it is determined that the current condition is a traveling valve leakage;

[0108] Rule 5: If P c > β, ε1 < k gl / k gu <ε2, where ε1, ε2, and ε3 are the determination thresholds for the gas influence condition, then it is determined that the current condition is a gas influence;

[0109] Rule 6: If P c > β, σ1 < k gl / k gu <σ2, where σ1, σ2, and σ3 are the determination thresholds for the insufficient liquid supply condition, then it is determined that the current condition is an insufficient liquid supply;

[0110] Step 3.3: Obtain the measured polished rod dynamometer card dataset based on the measured polished rod data in the rod pumping production process. Establish a comprehensive polished rod dynamometer card dataset according to the measured polished rod dynamometer card dataset and the simulated polished rod dynamometer card dataset, and determine the determination thresholds for each rule in the fault diagnosis knowledge base;

[0111] The determination of the thresholds in the knowledge base depends on the comprehensive polished rod dynamometer card dataset, which includes the measured polished rod dynamometer card dataset and the simulated polished rod dynamometer card dataset. The polished rod data in the measured polished rod dynamometer card dataset is collected by a dynamometer, and the polished rod data in the simulated polished rod dynamometer card dataset is generated by numerical simulation methods;

[0112] Set the determination threshold β for the rod break condition as the maximum value of P c in the comprehensive polished rod dynamometer card dataset, as shown in the following formula:

[0113]

[0114] where, is the P c value of the i-th polished rod dynamometer card in the rod break condition, i = 1, 2, 3,..., m1, and m1 is the amount of polished rod dynamometer card data in the rod break condition;

[0115] Set the normal operating condition determination threshold ζ as the maximum value in the comprehensive polished rod dynamometer card dataset as shown in the following formula:

[0116]

[0117] where and are the and of the w-th polished rod dynamometer card in the normal operating condition, where w = 1, 2, 3, …, m2, and m2 is the amount of polished rod dynamometer card data under the normal operating condition;

[0118] Set the fixed valve leakage condition determination thresholds λ1 and λ2 as the maximum values of k gl / k gu and k gl in the comprehensive polished rod dynamometer card dataset, as shown in the following formula:

[0119]

[0120] where and are the k gl and k gu of the y-th polished rod dynamometer card in the fixed valve leakage condition, where y = 1, 2, 3, …, m3, and m3 is the amount of polished rod dynamometer card data under the fixed valve leakage condition;

[0121] Set the traveling valve leakage condition determination thresholds η1 and η2 as the maximum values of k gl / k gu and k gu in the comprehensive polished rod dynamometer card dataset, as shown in the following formula:

[0122]

[0123] where and are the k gl and k gu of the i-th polished rod dynamometer card in the traveling valve leakage condition, where b = 1, 2, 3, …, m4, and m4 is the amount of polished rod dynamometer card data under the traveling valve leakage condition;

[0124] Set the gas influence condition determination thresholds ε1, ε2, and ε3 as the maximum and minimum values of k gl / k gu in the comprehensive polished rod dynamometer card dataset and the minimum value of as shown in the following formula:

[0125]

[0126] Among them, and are the k gl , k gu and of the d-th polished rod dynamometer card in the gas influence condition respectively, where d = 1, 2, 3,..., m5, and m5 is the amount of polished rod dynamometer card data under the gas influence condition;

[0127] Set the determination thresholds σ1, σ2, and σ3 for the insufficient liquid supply condition to be the maximum and minimum values of k gl / k gu in the comprehensive polished rod dynamometer card dataset and the maximum value of respectively, as shown in the following formula:

[0128]

[0129] Among them, and are the k gl , k gu and of the f-th polished rod dynamometer card in the insufficient liquid supply condition respectively, where f = 1, 2, 3,..., m6, and m6 is the amount of polished rod dynamometer card data under the insufficient liquid supply condition;

[0130] Step 4: Use the fault diagnosis knowledge base to diagnose the working conditions of the rod pumping production process;

[0131] Step 4.1: Design a fault diagnosis reasoning mechanism based on the fault diagnosis knowledge base;

[0132] The fault diagnosis knowledge base is a knowledge base based on production rules, and the rule form is an expression of "If P Then Q", where P is the rule premise, that is, the determination rule of the fault condition, and Q is the conclusion, that is, the fault type; according to this rule form, the designed fault diagnosis reasoning mechanism is forward reasoning. According to the given rod pumping operation data, including the polished rod displacement and polished rod load data obtained by the displacement sensor and load sensor located at the donkey head of the pumping unit, search for the determination rules of the fault conditions that match the rod pumping operation data in the fault diagnosis knowledge base rules. If the given rod pumping operation data matches the determination rule P of the fault condition, the fault type can be obtained;

[0133] Step 4.2: Obtain the rod pumping operation data, and based on the fault diagnosis knowledge base and the fault diagnosis reasoning mechanism, diagnose the faults in the rod pumping production process;

[0134] The displacement sensor and load sensor located at the donkey head of the pumping unit are used to obtain the operation data of the rod pump, including the polished rod displacement and polished rod load data, and the polished rod indicator diagram data within one stroke is obtained; the polished rod indicator diagram is preprocessed and the characteristic parameters are calculated; the fault diagnosis inference mechanism is used to search for rules in the established fault diagnosis knowledge base to complete the traversal of the fault diagnosis rules. When the characteristic parameters do not match all the rules, "no solution" is output and the process exits. When the characteristic parameters match one rule, the fault diagnosis result is output.

[0135] The beneficial effects of adopting the above technical solution are as follows: A fault diagnosis system and method for the rod pump oil production process based on numerical simulation provided by the present invention, and a method for generating a polished rod indicator diagram based on numerical simulation solve the problem of unbalanced measured polished rod indicator diagram data and avoid the disadvantage of poor reliability of the existing data-driven generation method; the proposed fault diagnosis system for the rod pump oil production process based on numerical simulation improves the interpretability of the existing data-driven diagnosis method. Description of the Drawings

[0136] Figure 1 It is a technical roadmap of a fault diagnosis system and method for the rod pump oil production process based on numerical simulation provided by an embodiment of the present invention;

[0137] Figure 2 It is a schematic structural diagram of a fault diagnosis system for the rod pump oil production process based on numerical simulation provided by an embodiment of the present invention;

[0138] Figure 3 It is a flowchart of a fault diagnosis method for the rod pump oil production process based on numerical simulation provided by an embodiment of the present invention;

[0139] Figure 4 It is the polished rod indicator diagrams generated under various working conditions in the rod pump oil production process provided by an embodiment of the present invention. Among them, (a) is the polished rod indicator diagram under the condition of insufficient liquid supply, (b) is the polished rod indicator diagram under the condition of gas influence, (c) is the polished rod indicator diagram under the condition of fixed valve leakage, (d) is the polished rod indicator diagram under the condition of traveling valve leakage, (e) is the polished rod indicator diagram under the normal condition, and (f) is the polished rod indicator diagram under the condition of rod breakage;

[0140] Figure 5 It is a schematic diagram of characteristic points and characteristic parameters in the polished rod indicator diagram provided by an embodiment of the present invention. Among them, (a) is the characteristic parameters before normalization, and (b) is the characteristic points and characteristic parameters after normalization;

[0141] Figure 6 It is a flowchart of fault diagnosis for the rod pump oil production process based on the fault diagnosis knowledge base and the fault diagnosis inference mechanism provided by an embodiment of the present invention. Detailed Embodiments

[0142] The following will further describe in detail the specific implementation manners of the present invention in conjunction with the accompanying drawings and embodiments. The following embodiments are used to illustrate the present invention, but are not used to limit the scope of the present invention.

[0143] A fault diagnosis system and method for the rod pumping oil production process based on numerical simulation in this embodiment are as Figure 1 shown. A fault diagnosis system for the rod pumping oil production process based on numerical simulation is established for fault diagnosis of the rod pumping oil production process. By analyzing the operation mechanism of the sucker rod pump and the generation mechanism of fault conditions, a mechanism model for the rod pumping oil production process under various conditions is established; a calculation model for the rod pumping oil production process based on the precise integration method is established. By adjusting production parameters, various rod pumping oil production processes are simulated, and a method for generating a polished rod dynamometer card data set is established; multiple characteristic parameters reflecting fault condition information are introduced, and a data-driven rule threshold calculation method is proposed to realize fault diagnosis of the rod pumping oil production process.

[0144] A fault diagnosis system for the rod pumping oil production process based on numerical simulation is as Figure 2 shown, including: a fault diagnosis knowledge base construction module, a fault diagnosis reasoning module, and a human-computer interaction module;

[0145] The fault diagnosis knowledge base construction module is used to establish and store fault condition diagnosis rules and determination thresholds, and construct a fault diagnosis knowledge base;

[0146] The fault diagnosis reasoning module is used to process the rod pump operation data input by the user, establish a fault diagnosis reasoning mechanism, and perform fault diagnosis of the rod pumping oil production process based on the constructed fault diagnosis knowledge base;

[0147] The human-computer interaction module is used to provide an interface for the fault diagnosis system for the rod pumping oil production process based on numerical simulation to interact with the user. The human-computer interaction module receives the rod pump operation data input by the user and outputs and displays the fault diagnosis results of the fault diagnosis reasoning module;

[0148] The user collects the rod pump operation data, which is transmitted to the fault diagnosis reasoning module through the human-computer interaction module. The fault diagnosis reasoning module processes the rod pump operation data input by the user, calls the fault diagnosis knowledge base construction module, uses the fault diagnosis reasoning mechanism, searches for the determination rules of fault conditions in the established fault diagnosis knowledge base, completes the traversal of the fault condition determination rules, and transmits the matching fault diagnosis results to the human-computer interaction module. The human-computer interaction module outputs and displays the fault diagnosis results of the fault diagnosis reasoning module.

[0149] A fault diagnosis method for the rod pumping oil production process based on numerical simulation is as Figure 3 shown, including the following steps:

[0150] Step 1: By analyzing the operating mechanism of the sucker rod pump and the generation mechanism of fault conditions, establish a mechanism model for the rod pumping process under various conditions;

[0151] Step 1.1: Analyze the operating mechanism of the sucker rod and establish a dynamic mechanism model of the sucker rod;

[0152] The sucker rod plays a very important role in the rod pumping process. It is the bridge connecting the surface pumping unit and the downhole sucker rod pump. By analyzing the force on the infinitesimal element of the sucker rod and using Newton's second law and Hooke's law, establish a dynamic mechanism model of the sucker rod as shown in the following formula:

[0153]

[0154] Among them, x is the distance between the cross-section of the sucker rod and the polished rod, t is time, u(x, t) is the displacement of the cross-section at x of the sucker rod relative to the casing at time t, v is the velocity of sound wave passing through the sucker rod, c is the damping coefficient, g is the acceleration due to gravity, E r is the Young's modulus of the sucker rod, ρ r is the density of the sucker rod, γ is the dimensionless damping coefficient, and L is the length of the sucker rod;

[0155] Step 1.2: Analyze the operating mechanism of the sucker rod pump and establish a mechanism model of the sucker rod pump;

[0156] The sucker rod pump mainly includes a pump barrel, a plunger, a standing valve, and a traveling valve. Among them, the plunger reciprocates up and down in the pump barrel. When the plunger reciprocates, the pressure in the pump barrel also changes accordingly. When the pressure in the pump barrel is less than the flowing bottom hole pressure, the standing valve opens to complete the suction of crude oil; when the pressure in the pump barrel is greater than the discharge pressure, the traveling valve opens to complete the discharge of crude oil; the change of the pressure in the pump barrel will cause the change of the plunger load. Therefore, based on the plunger displacement and the plunger load of the sucker rod pump, establish a mechanism model of the sucker rod pump, including:

[0157] Taking the lowest position of the plunger as the zero initial position, the plunger displacement is as shown in the following formula:

[0158]

[0159] Among them, x p (t) is the plunger displacement at time t, S p is the plunger stroke, and N is the pumping speed;

[0160] During the oil production process, the top and bottom of the plunger are simultaneously subjected to pressure. Among them, the pressure of the crude oil in the tubing is applied to its top, and the pressure in the pump barrel is applied to its bottom. Define the positive direction of the plunger load as the downward direction. The plunger load is as shown in the following formula:

[0161] F p F(t) = p d (A p - A rd ) - p(t)A p (3)

[0162] Among them, F p (t) is the plunger load at time t, p d is the discharge pressure, A p is the cross-sectional area of the plunger, A rd is the cross-sectional area of the sucker rod connected to the plunger, and p(t) is the barrel pressure at time t;

[0163] The change of the barrel pressure includes two stages: The first stage is from the closing of the traveling valve to the opening of the traveling valve, corresponding to the load increasing section, the high load section, and the unloading section; the second stage is from the opening of the traveling valve to the closing of the traveling valve, corresponding to the low load section. In the first stage, the barrel contains gas-liquid two-phase. Generally, the pressure of the gas in the barrel is used to represent the barrel pressure. Since the gas in the barrel is a mixture of hydrocarbon gases, it cannot be directly described by the ideal gas law. The compressibility factor is introduced to correct the ideal gas law in the first stage of the barrel pressure change. In the second stage of the barrel pressure change, when the traveling valve opens, the barrel is connected to the tubing, and the crude oil in the barrel is discharged into the tubing through the traveling valve. The barrel pressure is equal to the discharge pressure. Then the barrel pressure p(t) is shown in the following formula:

[0164]

[0165] Among them, V g (t) is the gas volume at time t, Z(t) is the compressibility factor at time t, n(t) is the number of moles at time t, R is the gas constant, and T is the absolute temperature;

[0166] In the first stage of the barrel pressure change, the barrel pressure is mainly affected by the gas volume V g (t), the compressibility factor Z(t), and the number of moles n(t). Specifically:

[0167] During the closing of the traveling valve, the barrel contains both gas and crude oil. The gas volume is equal to the barrel volume minus the volume of the crude oil in the barrel. The crude oil in the barrel enters through the standing valve, and its volume is the integral of the flow rate with respect to time. The gas volume V g (t) is shown in the following formula:

[0168]

[0169] Among them, V s is the clearance volume, V sl is the volume of the crude oil in the clearance at the initial moment, and q(τ) is the flow rate of the crude oil through the standing valve at time τ;

[0170] At time t, when the pump barrel pressure is greater than the flowing bottom-hole pressure, the standing valve closes and the flow rate of crude oil through the standing valve is zero; when the pump barrel pressure is less than or equal to the flowing bottom-hole pressure, the standing valve opens, and the flow rate q(t) of crude oil through the standing valve can be obtained from the flow rate formula of the standing valve orifice, as shown in the following formula:

[0171]

[0172] where μ is the standing valve flow coefficient, A s is the cross-sectional area of the standing valve, p s is the flowing bottom-hole pressure, ρ l is the density of crude oil;

[0173] The gas in the pump barrel is a mixture of hydrocarbon gases. The compressibility factor Z(t) of the gas in the pump barrel is expressed using the compressibility factor of natural gas, as shown in the following formula:

[0174] Z 3 (t) - Z 2 (t) + (0.0193p(t) - 0.0001p 2 (t))Z(t) - 0.0004p 2 (t) = 0 (7)

[0175] During the closing period of the traveling valve, the pump barrel contains gas. The change in the number of moles dn(t) can be divided into three parts. The first part dn1(t) is the number of moles of dissolved gas released by the crude oil in the pump, the second part dn2(t) is the number of moles of dissolved gas released when the pressure drop dp causes the crude oil to enter the pump, and the third part dn3(t) is the number of moles of gas entering the pump barrel. From time t to time t + dt, the change in the number of moles n(t) is as shown in the following formula:

[0176] dn(t) = dn1(t) + dn2(t) + dn3(t) (8)

[0177] where,

[0178]

[0179] where T0 and T a are the temperature under standard conditions and the actual temperature respectively, α is the solubility coefficient, f w is the water cut, R p is the volume ratio of gas to crude oil in the formation, B0 is the volume factor of crude oil, V l (t) is the volume of crude oil;

[0180] Step 1.3: Analyze the generation mechanism of the fault conditions and establish the mechanism models for various fault conditions;

[0181] During the operation of the sucker rod pump, common fault conditions include insufficient liquid supply, gas influence, leakage of the fixed valve, leakage of the traveling valve, and rod breakage. In this embodiment, the fault conditions include insufficient liquid supply, gas influence, leakage of the fixed valve, leakage of the traveling valve, and rod breakage; the mechanism models established for different fault conditions are as follows:

[0182] 1) The mechanism model for the insufficient liquid supply fault is:

[0183] p′ s <p s (12)

[0184] Where p′ s is the submerged pressure when the sucker rod pump is in the insufficient liquid supply fault condition;

[0185] With the continuous exploitation of crude oil, the liquid supply capacity of the oil reservoir decreases accordingly, the immersion depth of the crude oil in the casing will decrease, the submerged pressure will decrease accordingly, and it is less than the submerged pressure when the sucker rod pump is in the normal condition, which is set as the insufficient liquid supply fault condition;

[0186] 2) The mechanism model for the gas influence fault is:

[0187] n′(0)>n(0), R′ p >R p (13)

[0188] Where n′(0) is the number of moles of gas in the clearance when the sucker rod pump is in the gas influence fault condition at the initial moment, n(0) is the number of moles of gas in the clearance when the sucker rod pump is in the normal condition at the initial moment, and R′ p is the volume ratio of gas to crude oil in the formation when the sucker rod pump is in the gas influence fault condition;

[0189] At the beginning of the upstroke, there is a large amount of gas left in the sucker rod pump. At this time, the sucker rod pump is in the gas influence fault condition. It is set that the gas in the clearance at the initial moment is more than that in the normal condition, that is, R p becomes larger, which is set as the gas influence fault condition;

[0190] 3) The mechanism model for the fixed valve leakage fault is:

[0191]

[0192] Where q s (t) is the flow rate when the sucker rod pump is in the fixed valve leakage fault condition at time t, ξ is the leakage coefficient, and A l is the leakage area;

[0193] In each stroke, the fixed valve must open and close once. After long-term operation, the fixed valve will not be able to close completely, and the sucker rod pump is in the fixed valve leakage fault condition. At this time, the traveling valve of the sucker rod pump will close in advance, resulting in leakage of the fixed valve, which is set as the fixed valve leakage fault condition;

[0194] 4) The mechanism model of the traveling valve leakage fault is as follows:

[0195]

[0196] Among them, q t (t) is the flow rate of the sucker rod pump at time t under the traveling valve leakage fault condition, D is the diameter of the pump barrel, μ1 is the kinematic viscosity of the crude oil, L p is the length of the plunger, and δ is the radial clearance between the plunger and the pump barrel;

[0197] Due to the gap between the plunger and the pump barrel or the wear of the traveling valve itself, the volume of the discharged crude oil decreases. At this time, the sucker rod pump is in the traveling valve leakage fault condition. At this time, the fixed valve of the sucker rod pump will close in advance, resulting in leakage of the traveling valve, which is set as the traveling valve leakage fault condition;

[0198] 5) The mechanism model of the rod break fault is as follows:

[0199] L′<L,F pr (t)=0 (16)

[0200] Among them, L′ is the length of the sucker rod part moving with the polished rod under the rod break fault condition, and F pr (t) is the load at the bottom of the broken rod at time t;

[0201] Due to the large change in the downhole load and the accompanied strong corrosive working environment, after long-term operation, the sucker rod will age and wear. When the damage reaches a certain degree, the sucker rod may break. When in the rod break fault condition, the sucker rod pump is completely separated from the sucker rod, and the polished rod of the pumping unit only drags the broken rod to make reciprocating motion at the connection between the polished rod and the donkey head, and the load at the bottom of the broken rod is zero, which is set as the rod break fault condition;

[0202] Step 2: Establish a calculation model for the rod pumping process, establish a method for generating a set of polished rod dynamometer cards, and generate a set of polished rod dynamometer cards;

[0203] Step 2.1: Based on the mechanism model of the rod pumping process, determine the initial and boundary conditions, including the boundary conditions and initial conditions of the sucker rod;

[0204] The model of the sucker rod dynamics reveals the general law of the displacement u(x,t) of the cross-section of the sucker rod at position x relative to the casing at time t. To completely determine the motion of the sucker rod, the boundary conditions and initial conditions of the sucker rod need to be given.

[0205] Step 2.1.1: Determine the boundary conditions of the sucker rod dynamic mechanism model;

[0206] The top of the sucker rod is connected to the polished rod of the pumping unit, and the bottom is connected to the plunger of the subsurface pump. The boundary conditions of the sucker rod dynamic mechanism model are the motion law of the polished rod of the pumping unit and the load at the plunger of the subsurface pump. Among them, the motion of the polished rod of the pumping unit is similar to a simple harmonic motion, which is determined by the geometry of the surface pumping unit and is shown by the following formula:

[0207]

[0208] where S r is the stroke of the polished rod of the sucker rod;

[0209] The load at the plunger of the subsurface pump is:

[0210]

[0211] where u(L,t) is the displacement of the bottom of the sucker rod at time t, and A r is the cross-sectional area of the sucker rod;

[0212] Step 2.1.2: Determine the initial conditions of the sucker rod dynamic mechanism model;

[0213] Take the displacement and velocity of the sucker rod at the initial moment as the initial conditions of the sucker rod dynamic mechanism model, as shown by the following formula:

[0214]

[0215] where φ(x) is the initial displacement of the sucker rod at cross-section x, and ψ(x) is the initial velocity of the sucker rod at cross-section x;

[0216] Step 2.2: Use the precise integration method to establish a first-order differential equation of the sucker rod dynamic mechanism model in the rod pumping production process as the calculation model of the rod pumping production process, and use this model to calculate the polished rod displacement and the polished rod load, and generate a polished rod dynamometer card;

[0217] Step 2.2.1: Discretize the spatial variable and establish a first-order differential equation of the sucker rod dynamic mechanism model;

[0218] Taking the top of the sucker rod as the reference position, the sucker rod is equally divided into num intervals with a step size of h, and the corresponding cross-section positions are x j =jh, j = 0, 1, 2,..., num; at the cross-section positions of x1, x2, x3,..., x num-1 approximate the differential operation about x in the sucker rod dynamic mechanism model as a difference operation, as shown by the following formula:

[0219]

[0220] where u(x j , t) is the displacement of the cross-section of the sucker rod at position x j relative to the casing at time t;

[0221] The first-order differential equation of the dynamic mechanism model of the sucker rod is shown as the following formula:

[0222]

[0223] where V(t) is a (2num - 2)×1 matrix, H is a (2num - 2)×(2num - 2) matrix, f(t) is a (2num - 2)×1 matrix, E num-1 is an (num - 1)×(num - 1) identity matrix, 0 (num-1)×1 is an (num - 1)×1 zero matrix, K is an (num - 1)×(num - 1) matrix, r(t) is an (num - 1)×1 matrix, as shown in the following formula:

[0224]

[0225] Step 2.2.2: Discretize the time variable and calculate the solution of the first-order differential equation of the dynamic mechanism model of the sucker rod;

[0226] Taking the sucker rod at the bottom dead center as the initial moment, divide a stroke into m intervals with a step size of Δ, corresponding to the nodes t k = kΔ, k = 0, 1, 2, …, m. Using the precise integration method, obtain the solution V(t k+1 ) of the first-order differential equation of the dynamic mechanism model of the sucker rod at time t k+1 as:

[0227]

[0228] where V(t0) is a (2num - 2)×1 matrix, and both R0(t k ) and R1(t k ) are (2num - 2)×1 matrices;

[0229] Step 2.2.3: Calculate the polished rod displacement and the polished rod load, and generate the polished rod dynamometer card;

[0230] The polished rod dynamometer card is a graphical tool for analyzing the working state of the pumping unit, with the polished rod displacement and the polished rod load as the horizontal and vertical coordinates respectively. The polished rod displacement at time t k is shown as the following formula:

[0231]

[0232] where u(x0, tk ) is the displacement of the cross-section of the sucker rod at x0 relative to the casing at time t k ;

[0233] t k The polished rod load P(t k ) is shown in the following formula:

[0234]

[0235] where u(x1,t k ) is the displacement of the cross-section of the sucker rod at x1 relative to the casing at time t k ;

[0236] Step 2.3: Simulate different working conditions of the rod pumping production process. Based on the calculation model of the rod pumping production process, generate the polished rod dynamometer diagrams of multiple rod pumping production processes under different working conditions, and construct a simulated polished rod dynamometer diagram dataset;

[0237] According to the sucker rod string combination design table in the "Petroleum and Natural Gas Industry Standard SY / T 5873 - 2017 of the People's Republic of China", select the nominal diameter of the sucker rod pump, the rod string combination of the sucker rod, and the stroke of the pumping unit; according to the technical specifications of the beam pumping unit in the "Oil Production Technology Manual - 04 - Mechanical Oil Production Technology", select the model and stroke frequency of the pumping unit; simulate different working conditions of multiple rod pumping production processes, and use the calculation model of the rod pumping production process proposed in Step 2.2 to generate the polished rod dynamometer diagrams of multiple rod pumping production processes under different working conditions, and construct a polished rod dynamometer diagram dataset;

[0238] In this embodiment, the polished rod dynamometer diagrams generated under different working conditions of the rod pumping production process are as Figure 4 shown, where (a) is the polished rod dynamometer diagram under the condition of insufficient liquid supply, (b) is the polished rod dynamometer diagram under the condition of gas influence, (c) is the polished rod dynamometer diagram under the condition of fixed valve leakage, (d) is the polished rod dynamometer diagram under the condition of traveling valve leakage, (e) is the polished rod dynamometer diagram under the normal condition, and (f) is the polished rod dynamometer diagram under the condition of rod breakage;

[0239] Step 3: Introduce multiple characteristic points and characteristic parameters to extract the working condition information in the polished rod dynamometer diagram, and establish a fault diagnosis knowledge base for the rod pumping production process, including the determination rules and determination thresholds for each fault working condition;

[0240] Step 3.1: Based on the corresponding relationship between the polished rod dynamometer diagram and the working state of the sucker rod pump, introduce multiple characteristic points, and divide the polished rod dynamometer diagram into a loading section, a high-load section, an unloading section, and a low-load section;

[0241] Step 3.1.1: Preprocess the polished rod displacement data and polished rod load data included in the polished rod dynamometer diagram dataset;

[0242] Normalize the polished rod displacement data and polished rod load data included in the polished rod dynamometer card dataset to the interval [0, 1] as shown in the following formula:

[0243]

[0244] where S(t k ), P(t k ) are the polished rod displacement value and polished rod load value of the polished rod dynamometer card at time t k respectively, are the maximum value and minimum value of the polished rod dynamometer card at time t k after normalization respectively, S max , S min are the maximum value and minimum value of the polished rod displacement in the polished rod dynamometer card data respectively, P max , P min are the maximum value and minimum value of the polished rod load in the polished rod dynamometer card data respectively;

[0245] In this embodiment, the schematic diagrams of the characteristic points and characteristic parameters in the polished rod dynamometer card are as shown in Figure 5 , where (a) are the characteristic parameters before normalization, and (b) are the characteristic points and characteristic parameters after normalization;

[0246] Step 3.1.2: Introduce multiple characteristic points to divide the polished rod dynamometer card data into a loading section, a high-load section, an unloading section, and a low-load section;

[0247] Set the data intervals of the polished rod load to [0, P down and [P up , 1.0], where P down is the right endpoint of the data interval for determining the loading termination point and the unloading starting point, P up is the left endpoint of the data interval for determining the unloading termination point and the loading starting point. Select the first minimum point of the data interval [0, P down as the loading termination point The last minimum point of the data interval [0, P down as the unloading starting point The first maximum point of the data interval [P up , 1.0] as the unloading termination point And the last maximum point of the data interval [P up , 1.0] as the loading starting point As characteristic points, where, is the abscissa of the loading termination point, is the ordinate of the loading termination point, is the abscissa of the unloading starting point, is the ordinate of the unloading starting point, is the abscissa of the unloading termination point, is the ordinate of the unloading termination point, is the abscissa of the load increasing starting point, is the abscissa of the load increasing starting point. According to the selected characteristic points, the polished rod dynamometer card data is divided into a load increasing section, a high load section, an unloading section and a low load section;

[0248] In this embodiment, select P down = 0.3, P up = 0.7;

[0249] Step 3.2: Introduce multiple characteristic parameters to characterize the fault conditions, and establish the determination rules for each fault condition in the fault diagnosis knowledge base;

[0250] In order to establish a fault diagnosis knowledge base, select multiple characteristic parameters to characterize the fault conditions, including the difference P c between the maximum polished rod load and the minimum polished rod load before normalization, the displacement between the load increasing termination point and the unloading starting point after normalization the displacement gl between the unloading termination point and the load increasing starting point after normalization gu the slope k of the straight line connecting the load increasing starting point and the load increasing termination point after normalization, the slope k of the straight line connecting the unloading starting point and the unloading termination point after normalization, the data point

[0251] where the polished rod load in the load increasing section after normalization is 0.5

[0252]

[0253]

[0254] Use the characteristic parameters to characterize each fault condition, and establish the determination rules for each fault condition in the fault diagnosis knowledge base, specifically including:

[0255] Rule 1: If P c ≤β, where β is the determination threshold for the rod break condition, then determine that the current condition is the rod break condition;

[0256] Due to the large load changes downhole, and the highly corrosive working environment, the sucker rod will age and wear after a long period of operation. When the damage reaches a certain degree, the sucker rod may break under the drastic load changes. At this time, the pump is completely separated from the sucker rod, and the suspension point only drags a part of the sucker rod to reciprocate. At this time, the suspension point indicator diagram is horizontal and flat, and the difference between the maximum suspension point load and the minimum suspension point load is very small, that is, P c ≤β;

[0257] Rule 2: If P c >β, ζ is the normal operating condition judgment threshold, and the current operating condition is judged to be a normal operating condition;

[0258] The overall shape of the hanging point indicator diagram under normal working conditions is similar to that of the theoretical hanging point indicator diagram, and is close to a parallelogram. This shows that the suction and discharge of the pump are sufficient. From the detailed observation, it is not difficult to find that there are some fluctuations on the hanging point indicator diagram, which is caused by the fluctuation of the sucker rod caused by the load change, but the overall shape is close to a parallelogram, and the length of the high-load section and the low-load section are close, that is,

[0259] Rule 3: If P c >β, k gl / k gu <λ1,k gl <λ2, λ1 and λ2 are the thresholds for determining the fixed valve leakage condition, then the current condition is determined to be fixed valve leakage;

[0260] In each stroke, the fixed valve has to open and close once, and the valve has to collide with the base once. After long-term operation, the valve and the base are prone to wear, resulting in the valve not being able to close completely and leakage. Fixed valve leakage will cause the unloading end point to lag behind and the loading start point to advance. The shape of the entire suspension point indicator diagram approaches an inverted trapezoid. At this time, the length difference between the high-load section and the low-load section is large, that is, The slope of the loading section is negative, and the slope of the unloading section is positive, that is, k gl / k gu <λ1,k gl <λ2;

[0261] Rule 4: If P c >β, k gl / k gu <η1,k gu <η2, η1 and η2 are the thresholds for determining the operating condition of the traveling valve leakage, then the current operating condition is determined to be the traveling valve leakage;

[0262] The traveling valve is immersed in crude oil for a long time. During each stroke, the traveling valve collides with the plunger once. After long-term operation, it is very easy to be unable to close completely due to friction or corrosion and leakage occurs. The leakage of the traveling valve will cause the load increasing termination point to lag and the unloading starting point to advance. The shape of the whole polished rod dynamometer card approaches a trapezoid. At this time, the length difference between the high-load section and the low-load section is large, that is The slope of the load increasing section is positive, and the slope of the unloading section is negative, that is k gl / k gu <η1, k gu <η2;

[0263] Rule 5: If P c >β, ε1 < k gl / k gu <ε2, ε1, ε2, ε3 are the judgment thresholds for gas influence working conditions, then it is determined that the current working condition is gas influence;

[0264] Due to the presence of gas in the pump barrel or the high gas content in the crude oil, during the upstroke process, gas continuously precipitates from the crude oil, and the pressure reduction in the pump barrel slows down, and the load increasing termination point will lag; during the downstroke, there is more gas in the pump chamber than normal, and the pump barrel pressure cannot be quickly reduced, and the unloading termination point will lag. It can be described by ε1 < k gl / k gu <ε2. The polished rod dynamometer card of gas influence is similar to that of insufficient liquid supply. The difference between the two is that when the liquid supply is insufficient, the load increasing section approaches a straight line, while when the gas is influenced, the load increasing section approaches an arc shape. At this time, the characteristic parameter is introduced to distinguish these two working conditions. Gas influence can be described by

[0265] Rule 6: If P c >β, σ1 < k gl / k gu <σ2, σ1, σ2, σ3 are the judgment thresholds for insufficient liquid supply working conditions, then it is determined that the current working condition is insufficient liquid supply;

[0266] With the continuous exploitation of crude oil, the liquid supply capacity of the oil reservoir decreases accordingly, the dynamic liquid level drops, and it is unable to provide enough submerged pressure for the sucker rod pump. In this case, the filling degree of the pump becomes poor and the liquid production rate decreases significantly. The characteristic of insufficient liquid supply is that the polished rod load cannot be immediately reduced during the downstroke. Only when the plunger encounters the liquid level, it quickly unloads, and the unloading termination point lags. The shape of the polished rod dynamometer card under the condition of insufficient liquid supply is similar to a "knife handle shape". The ratio of the slope of the load increasing section to the slope of the unloading section is within a certain range, that is σ1 < k gl / k gu<σ2. The insufficient liquid supply and gas influence on the polished rod dynamometer card are similar. The difference between the two lies in the loading section. Due to the gas effect in the gas influence on the loading section, the loading section is circular arc-shaped, while the loading section of insufficient liquid supply approaches a straight line. The characteristic parameter is introduced to distinguish. Insufficient liquid supply can be described by .

[0267] Step 3.3: Obtain the measured polished rod dynamometer card dataset based on the measured polished rod data in the rod pumping process. Establish a comprehensive polished rod dynamometer card dataset according to the measured polished rod dynamometer card dataset and the simulated polished rod dynamometer card dataset, and determine the decision thresholds of each rule in the fault diagnosis knowledge base;

[0268] The determination of the threshold in the knowledge base depends on the comprehensive polished rod dynamometer card dataset, which includes the measured polished rod dynamometer card dataset and the simulated polished rod dynamometer card dataset. The polished rod data in the measured polished rod dynamometer card dataset are collected by a dynamometer, and the polished rod data in the simulated polished rod dynamometer card dataset are generated by numerical simulation methods;

[0269] Set the decision threshold β for the rod break condition as the maximum value of P in the comprehensive polished rod dynamometer card dataset c , as shown in the following formula:

[0270]

[0271] where is the P value of the i-th polished rod dynamometer card in the rod break condition, i = 1, 2, 3, …, m1, and m1 is the amount of polished rod dynamometer card data under the rod break condition; c

[0272] Set the decision threshold ζ for the normal condition as the maximum value of in the comprehensive polished rod dynamometer card dataset, as shown in the following formula:

[0273]

[0274] where and are the and of the w-th polished rod dynamometer card in the normal condition, w = 1, 2, 3, …, m2, and m2 is the amount of polished rod dynamometer card data under the normal condition;

[0275] Set the decision thresholds λ1 and λ2 for the fixed valve leakage condition as the maximum values of k gl / k gu and k gl in the comprehensive polished rod dynamometer card dataset, as shown in the following formula:

[0276]

[0277] Among them, and are the k gl and k gu of the y-th polished rod dynamometer card in the fixed valve leakage condition respectively, where y = 1, 2, 3, …, m3, and m3 is the amount of polished rod dynamometer card data in the fixed valve leakage condition;

[0278] Set the judgment thresholds η1 and η2 for the traveling valve leakage condition to be the maximum values of k gl / k gu and k gu in the comprehensive polished rod dynamometer card dataset respectively, as shown in the following formula:

[0279]

[0280] Among them, and are the k gl and k gu of the b-th polished rod dynamometer card in the traveling valve leakage condition respectively, where b = 1, 2, 3, …, m4, and m4 is the amount of polished rod dynamometer card data in the traveling valve leakage condition;

[0281] Set the judgment thresholds ε1, ε2, and ε3 for the gas influence condition to be the maximum and minimum values of k gl / k gu in the comprehensive polished rod dynamometer card dataset and the minimum value of respectively, as shown in the following formula:

[0282]

[0283] Among them, and are the k gl , k gu and of the d-th polished rod dynamometer card in the gas influence condition respectively, where d = 1, 2, 3, …, m5, and m5 is the amount of polished rod dynamometer card data in the gas influence condition;

[0284] Set the judgment thresholds σ1, σ2, and σ3 for the insufficient liquid supply condition to be the maximum and minimum values of k gl / k gu in the comprehensive polished rod dynamometer card dataset and the maximum value of respectively, as shown in the following formula:

[0285]

[0286] Among them, and are the k gl , k gu and f = 1, 2, 3, …, m6, where m6 is the amount of polished rod dynamometer card data under the condition of insufficient liquid supply;

[0287] Step 4: Use the fault diagnosis knowledge base to diagnose the working conditions of the rod pumping production process;

[0288] Step 4.1: Design a fault diagnosis reasoning mechanism based on the fault diagnosis knowledge base;

[0289] The fault diagnosis knowledge base is a knowledge base based on production rules, and the rule form is an expression of "If P Then Q", where P is the rule premise, that is, the determination rule of the fault condition, and Q is the conclusion, that is, the fault type; according to this rule form, the designed fault diagnosis reasoning mechanism is forward reasoning. According to the given rod pumping operation data, including the polished rod displacement and polished rod load data obtained by the displacement sensor and load sensor located at the donkey head of the pumping unit, search for the determination rules of the fault conditions that match the rod pumping operation data in the fault diagnosis knowledge base rules. If the given rod pumping operation data matches the determination rule P of the fault condition, the fault type Q is obtained;

[0290] Step 4.2: Obtain the rod pumping operation data, and based on the fault diagnosis knowledge base and the fault diagnosis reasoning mechanism, diagnose the faults in the rod pumping production process;

[0291] In this embodiment, the flow chart of diagnosing the faults in the rod pumping production process based on the fault diagnosis knowledge base and the fault diagnosis reasoning mechanism is as Figure 6 shown. Use the displacement sensor and load sensor located at the donkey head of the pumping unit to obtain the rod pumping operation data, including the polished rod displacement and polished rod load data, and obtain the polished rod dynamometer card data within one stroke; perform data preprocessing on the polished rod dynamometer card and calculate the characteristic parameters; use the fault diagnosis reasoning mechanism to search for rules in the established fault diagnosis knowledge base to complete the traversal of the fault diagnosis rules. When the characteristic parameters do not match all the rules, output no solution and exit. When the characteristic parameters match one rule, output the fault diagnosis result.

[0292] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope defined by the claims of the present invention.

Claims

1. A fault diagnosis system for a rod pump oil production process based on numerical simulation, characterized in that: Fault diagnosis knowledge base construction module, fault diagnosis reasoning module and human-computer interaction module; The fault diagnosis knowledge base construction module is used to establish and store fault condition diagnosis rules and judgment thresholds to construct a fault diagnosis knowledge base; The fault diagnosis and reasoning module is used to process the sucker rod pump operation data input by the user, establish a fault diagnosis and reasoning mechanism, and perform fault diagnosis of the sucker rod pump oil production process based on the established fault diagnosis knowledge base; The human-computer interaction module is used to provide an interface for the rod pump oil production process fault diagnosis system based on numerical simulation to interact with the user. The human-computer interaction module receives the rod pump operation data input by the user and outputs the fault diagnosis result of the fault diagnosis reasoning module.

2. A method for diagnosing faults in a rod pumping oil production process based on numerical simulation, which is based on the system of claim 1 to diagnose faults in a rod pumping oil production process, characterized in that: The following steps are involved: Step 1: By analyzing the operation mechanism of the oil well pump and the generation mechanism of the fault conditions, a mechanism model of the rod pump oil production process under various working conditions is established; Step 2: Establish a calculation model of the rod pump oil production process, establish a method for generating a suspension point dynamometer data set, and generate a suspension point dynamometer data set; Step 3: Introduce multiple characteristic points and characteristic parameters to extract the working condition information in the suspension point dynamometer diagram, and establish a knowledge base for fault diagnosis of the rod pump oil production process, including the judgment rules and judgment thresholds of each fault condition; Step 4: Use the fault diagnosis knowledge base to diagnose the working conditions of the rod pump oil production process.

3. The method for diagnosing faults in a rod pumping process based on numerical simulation according to claim 2, characterized in that: Step 1 specifically includes: Step 1.1: Analyze the operation mechanism of the sucker rod and establish the dynamic mechanism model of the sucker rod: Through the micro-element force analysis of the sucker rod, using Newton's second law and Hooke's law, the dynamic mechanism model of the sucker rod is established, as shown in the following formula: in, x is the distance between the sucker rod cross section and the polished rod, t is the time, u(x,t) is the displacement of the cross section at x of the sucker rod relative to the casing at time t, v is the speed of the sound wave passing through the sucker rod, c is the damping coefficient, g is the gravitational acceleration, E r is the Young's modulus of the sucker rod, ρ r is the density of the sucker rod, γ is the dimensionless damping coefficient, and L is the length of the sucker rod; Step 1.2: Analyze the operation mechanism of the oil well pump and establish the mechanism model of the oil well pump: The mechanism model of the oil well pump is established based on the plunger displacement and plunger load of the oil well pump, including: Taking the lowest position of the plunger as the zero initial position, the plunger displacement is shown in the following formula: Among them, x p (t) is the plunger displacement at time t, S p is the plunger stroke, N is the number of strokes; During the oil production process, the top and bottom of the plunger are subjected to pressure at the same time, where the pressure of the crude oil in the oil pipe is applied to the top of the plunger, and the pressure of the pump barrel is applied to the bottom of the plunger. The positive direction of the plunger load is defined as the downward direction. The plunger load is shown in the following formula: F p (t)=p d (A p -A rd )-p(t)A p (3) Among them, F p (t) is the plunger load at time t, p d is the discharge pressure, A p is the cross-sectional area of ​​the plunger, A rd is the cross-sectional area of ​​the sucker rod connected to the plunger, p(t) is the pump barrel pressure at time t; The change of pump barrel pressure includes two stages: the first stage is from the closing of the traveling valve to the opening of the traveling valve, and the second stage is from the opening of the traveling valve to the closing of the traveling valve. In the first stage of the change of pump barrel pressure, the compression coefficient is introduced to correct the ideal gas law. In the second stage of the change of pump barrel pressure, the traveling valve is opened, the pump barrel is connected to the oil pipe, and the crude oil in the pump barrel is discharged into the oil pipe through the traveling valve. The pump barrel pressure is equal to the discharge pressure, and the pump barrel pressure p(t) is shown in the following formula: Among them, V g (t) is the gas volume at time t, Z(t) is the compressibility factor at time t, n(t) is the number of moles at time t, R is the gas constant, and T is the absolute temperature; In the first stage of the pump barrel pressure change, the pump barrel pressure is affected by the gas volume V g (t), compression coefficient Z(t) and molar number n(t), specifically: During the period when the floating valve is closed, the pump barrel contains both gas and crude oil. The volume of gas is equal to the volume of the pump barrel minus the volume of crude oil in the pump barrel. The crude oil in the pump barrel enters through the fixed valve, and its volume is the integral of the flow rate over time. The gas volume V g (t) is shown in the following formula: Among them, V s is the clearance volume, V sl is the volume of crude oil in the clearance at the initial moment, q(τ) is the flow rate of crude oil through the fixed valve at time τ; At time t, when the pump barrel pressure is greater than the sinking pressure, the fixed valve is closed and the flow rate of crude oil through the fixed valve is zero; when the pump barrel pressure is less than or equal to the sinking pressure, the fixed valve is opened, and the flow rate q(t) of crude oil through the fixed valve can be obtained by the fixed valve port flow formula, as shown in the following formula: Where, μ is the fixed valve flow coefficient, A s is the fixed valve cross-sectional area, p s is the sinking pressure, ρ l is the density of crude oil; The gas in the pump barrel is a mixture of hydrocarbon gases. The compression coefficient of natural gas is used to express the compression coefficient Z(t) of the gas in the pump barrel, as shown in the following formula: Z 3 (t)-Z 2 (t)+(0.0193p(t)-0.0001p 2 (t))Z(t)-0.0004p 2 (t)=0 (7) During the period when the traveling valve is closed, there is gas in the pump barrel. The change in the number of moles dn(t) can be divided into three parts. The first part dn1(t) is the number of moles of dissolved gas released by crude oil in the pump, the second part dn2(t) is the number of moles of dissolved gas released by crude oil entering the pump when the pressure drop dp occurs, and the third part dn3(t) is the number of moles of gas entering the pump barrel. From time t to time t+dt, the change in the number of moles n(t) is shown in the following formula: dn(t)=dn1(t)+dn2(t)+dn3(t) (8) Among them, Among them, T0, T a are the temperature under standard conditions and the actual temperature, α is the solubility coefficient, f w is the moisture content, R p is the volume ratio of gas to crude oil in the formation, B0 is the volume coefficient of crude oil, V l (t) is the volume of crude oil; Step 1.3: Analyze the generation mechanism of fault conditions and establish mechanism models for various fault conditions.

4. The method for diagnosing faults in a rod pumping process based on numerical simulation according to claim 3 is characterized in that: The fault conditions described in step 1.3 include insufficient liquid supply, gas influence, fixed valve leakage, floating valve leakage and rod breakage; the mechanism models established for different fault conditions are: 1) The mechanism model of insufficient liquid supply failure is: p s ' <p s (12) Where p s ' is the sinking pressure when the oil well pump is in the condition of insufficient fluid supply failure; 2) The mechanism model of gas-affected failure is: n′(0)>n(0),R′ p >R p (13) Where n′(0) is the number of moles of gas in the gap when the oil well pump is in a gas-affected fault condition at the initial moment, n(0) is the number of moles of gas in the gap when the oil well pump is in a normal condition at the initial moment, R′ p It is the volume ratio of gas to crude oil in the formation when the oil pump is in a gas-affected fault condition; 3) The mechanism model of fixed valve leakage failure is: Among them, q s (t) is the flow rate of the oil well pump when it is in the fixed valve leakage fault condition at time t, ξ is the leakage coefficient, A l is the leakage area; 4) The mechanism model of the floating valve leakage failure is: Among them, q t (t) is the flow rate of the oil pump when it is in the condition of traveling valve leakage failure at time t, D is the diameter of the pump barrel, μ1 is the kinematic viscosity of the crude oil, L p is the plunger length, δ is the radial clearance between the plunger and the pump barrel; 5) The mechanism model of pole break failure is: L′<L,F pr (t)=0 (16) Where L′ is the length of the part of the sucker rod that moves with the suspension point when a rod breakage fault occurs, and F pr (t) is the load at the bottom of the broken rod at time t.

5. A method for diagnosing faults in a rod pumping oil production process based on numerical simulation according to claim 4, characterized in that: Step 2 specifically includes: Step 2.1: Based on the mechanism model of the sucker rod pumping process, determine the initial boundary conditions, including the boundary conditions and initial conditions of the sucker rod; Step 2.1.1: Determine the boundary conditions of the sucker rod dynamic mechanism model; The boundary conditions of the dynamic mechanism model of the sucker rod are the motion law of the pumping unit polished rod and the load on the pump plunger. The motion of the pumping unit polished rod seems to be a simple harmonic motion, which is determined by the geometric shape of the ground pumping unit, as shown in the following formula: Among them, S r is the bare rod stroke of the oil pump; The load at the oil well pump plunger is: Where u(L,t) is the displacement of the bottom of the sucker rod at time t, A r is the cross-sectional area of ​​the sucker rod; Step 2.1.2: Determine the initial conditions of the sucker rod dynamic mechanism model; The displacement and velocity of the sucker rod at the initial moment are taken as the initial conditions of the dynamic mechanism model of the sucker rod, as shown in the following formula: Among them, φ(x) is the initial displacement of the sucker rod at section x, and ψ(x) is the initial velocity of the sucker rod at section x; Step 2.2: Using the precise integration method, a first-order differential equation of the dynamic mechanism model of the sucker rod during the rod pumping process is established as a calculation model of the rod pumping process, and the suspension point displacement and suspension point load are calculated using the model to generate a suspension point indicator diagram; Step 2.3: Simulate different working conditions of various rod pump oil production processes, generate various hanging point dynamometer diagrams of rod pump oil production processes under different working conditions based on the calculation model of the rod pump oil production process, and construct a simulated hanging point dynamometer diagram data set.

6. The method for diagnosing faults in a rod pumping process based on numerical simulation according to claim 5, characterized in that: Step 2.2 specifically includes: Step 2.2.1: Discretize the spatial variables and establish the first-order differential equation of the dynamic mechanism model of the sucker rod; Taking the top of the sucker rod as the reference position, the sucker rod is divided into num intervals with a step length of h, and the corresponding cross-sectional position is x j =jh, j = 0, 1, 2, ..., num; at the section position x1, x2, x3, ..., x num-1 At , the differential operation about x in the dynamic mechanism model of the sucker rod is approximated as a difference operation, as shown in the following formula: Among them, u(x j ,t) is the sucker rod x j The displacement of the cross section at time t relative to the casing; The first-order differential equation of the dynamic mechanism model of the sucker rod is shown in the following formula: Among them, V(t) is a (2num-2)×1 matrix, H is a (2num-2)×(2num-2) matrix, f(t) is a (2num-2)×1 matrix, E num-1 is the (num-1)×(num-1)-order identity matrix, 0 (num-1)×1 is a (num-1)×1 order zero matrix, K is a (num-1)×(num-1) order matrix, and r(t) is a (num-1)×1 order matrix, as shown in the following formula: Step 2.2.2: Discretize the time variable and calculate the solution of the first-order differential equation of the sucker rod dynamic mechanism model; Taking the sucker rod at the bottom dead center as the initial moment, a stroke is divided into m intervals with a step length of Δ, corresponding to the node t k = kΔ, k = 0, 1, 2, ..., m, using the precise integration method, we get t k+1 The solution of the first-order differential equation of the dynamic mechanism model of the sucker rod at time V(t k+1 )for: Among them, V(t0) is a (2num-2)×1 matrix, R0(t k ) and R1(t k ) are all (2num-2)×1 order matrices; Step 2.2.3: Calculate the suspension point displacement and suspension point load, and generate the suspension point indicator diagram; The suspension point indicator diagram is a graphical tool used to analyze the working status of the oil pumping unit, with the suspension point displacement and the suspension point load as the horizontal and vertical coordinates. k The suspension point displacement S(t k ) is shown in the following formula: Among them, u(x0,t k ) is the cross section of the sucker rod at x0 at t k The displacement of the moment relative to the casing; t k The load at the suspension point P(t k ) is shown in the following formula: Among them, u(x1,t k ) is the cross section of the sucker rod at x1 at t k The displacement of the casing at the moment.

7. A method for diagnosing faults in a rod pumping oil production process based on numerical simulation according to claim 6, characterized in that: Step 3 specifically includes: Step 3.1: Based on the corresponding relationship between the suspension point indicator diagram and the working state of the oil well pump, multiple characteristic points are introduced to divide the suspension point indicator diagram into a load increase section, a high load section, an unloading section and a low load section; Step 3.1.1: Preprocess the suspension point displacement data and suspension point load data contained in the suspension point dynamometer data set; The suspension point displacement data and suspension point load data contained in the suspension point dynamometer data set are normalized to the [0,1] interval, as shown in the following formula: Among them, S(t k )、P(t k ) are t k The suspension point displacement value and suspension point load value of the suspension point indicator diagram at the moment, After normalization, they are k The maximum and minimum values ​​of the suspension point indicator diagram at the moment, S max , S min are the maximum and minimum values ​​of the suspension point displacement in the suspension point indicator diagram data, P max , P min They are the maximum and minimum values ​​of the suspension point load in the suspension point dynamometer data respectively; Step 3.1.2: Introduce multiple characteristic points and divide the suspension point dynamometer data into load-increasing section, high-load section, unloading section and low-load section; Set the data interval of the suspension point load to [0,P down ] and [P up ,1.0],P down To determine the right endpoint of the data interval of the loading end point and unloading start point, P up To determine the left endpoint of the data interval of the unloading end point and the loading start point, select the data interval [0,P down The first minimum point of ] is the end point of load increase Data interval [0,P down The last minimum point of ] is the unloading starting point Data interval [P up ,1.0] is the first maximum point of the unloading termination point And the data interval [P up ,1.0] is the starting point of the load increase As feature points, is the horizontal coordinate of the loading end point, is the ordinate of the loading end point, is the horizontal coordinate of the unloading starting point, is the ordinate of the unloading starting point, is the horizontal coordinate of the unloading end point, is the ordinate of the unloading end point, is the horizontal coordinate of the starting point of the load increase, is the horizontal coordinate of the starting point of the load increase, and the suspension point indicator diagram data is divided into a load increase section, a high load section, an unloading section and a low load section according to the selected characteristic points; Step 3.2: Introduce multiple characteristic parameters to characterize the fault conditions and establish the judgment rules for each fault condition in the fault diagnosis knowledge base; Step 3.3: Based on the measured suspension point data of the rod pump oil production process, a measured suspension point dynamometer diagram data set is obtained, and a comprehensive suspension point dynamometer diagram data set is established according to the measured suspension point dynamometer diagram data set and the simulated suspension point dynamometer diagram data set, and the judgment threshold of each rule in the fault diagnosis knowledge base is determined.

8. The method for diagnosing faults in a rod pumping process based on numerical simulation according to claim 7, characterized in that: The details of step 3.2 are as follows: Select multiple characteristic parameters to characterize the fault condition, including the difference between the maximum suspension point load and the minimum suspension point load before normalization P c , the displacement between the loading end point and the unloading start point after normalization Displacement between the unloading end point and the loading start point after normalization The slope k of the straight line connecting the starting point and the end point of the load increase after normalization gl , the slope k of the straight line connecting the unloading start point and the unloading end point after normalization gu , after normalization, the data point with the load of the suspension point in the load-increasing section being 0.5 Distance to the straight line connecting the starting point and the ending point of the load increase The characteristic parameters are used to characterize each fault condition, and the judgment rules of each fault condition in the fault diagnosis knowledge base are established, including: Rule 1: If P c ≤β, β is the threshold for judging the rod-broken condition, then the current condition is judged as the rod-broken condition; Rule 2: If P c >β, ζ is the normal operating condition judgment threshold, and the current operating condition is judged to be a normal operating condition; Rule 3: If P c >β, k gl / k gu <λ1,k gl <λ2, λ1 and λ2 are the thresholds for determining the fixed valve leakage condition, then the current condition is determined to be fixed valve leakage; Rule 4: If P c >β, k gl / k gu <η1,k gu <η2, η1 and η2 are the thresholds for determining the operating condition of the traveling valve leakage, then the current operating condition is determined to be the traveling valve leakage; Rule 5: If P c >β, ε1 <k gl / k gu <ε2, ε1, ε2, and ε3 are the thresholds for judging the working condition affected by gas, and the current working condition is judged to be affected by gas; Rule 6: If P c >β, σ1 <k gl / k gu <σ2, σ1, σ2, and σ3 are the thresholds for determining the insufficient fluid supply condition, and the current condition is determined to be insufficient fluid supply.

9. The method for diagnosing faults in a rod pumping oil production process based on numerical simulation according to claim 8, characterized in that: The details of step 3.3 are as follows: The determination of the threshold in the knowledge base depends on the comprehensive suspension point dynamometer data set, which includes the measured suspension point dynamometer data set and the simulated suspension point dynamometer data set. The suspension point data in the measured suspension point dynamometer data set are collected by a dynamometer, and the suspension point data in the simulated suspension point dynamometer data set are generated by a numerical simulation method. The rod breaking condition judgment threshold β is set to P in the comprehensive suspension point indicator diagram data set. c The maximum value is as shown in the following formula: in, is the P of the dynamometer diagram of the i-th suspension point in the rod breaking condition c Value, i = 1, 2, 3, ..., m1, m1 is the data volume of the suspension point dynamometer diagram under the rod breaking condition; Set the normal working condition judgment threshold ζ to the comprehensive suspension point indicator diagram data set The maximum value is as shown in the following formula: in, and is the wth suspension point indicator diagram under normal working conditions and w=1,2,3,…,m2, m2 is the amount of data of the hanging point dynamometer diagram under normal working conditions; The fixed valve leakage condition judgment thresholds λ1 and λ2 are set to be k in the comprehensive suspension point indicator diagram data set. gl / k gu With k gl The maximum value is as shown in the following formula: in, and are the k of the yth suspension point indicator diagram in the fixed valve leakage condition gl With k gu , y=1,2,3,…,m3, m3 is the amount of hanging point dynamometer data under the fixed valve leakage condition; The thresholds for judging the leakage condition of the traveling valve η1 and η2 are set to be k in the comprehensive suspension point indicator diagram data set. gl / k gu With k gu The maximum value is as shown in the following formula: in, and are the k of the dynamometer diagram of the i-th suspension point in the floating valve leakage condition. gl With k gu , b=1,2,3,…,m4, m4 is the amount of data of the hanging point indicator diagram under the condition of the traveling valve leakage; The gas influence working condition judgment thresholds ε1, ε2, and ε3 are set to be the kth gas influence working condition judgment thresholds in the comprehensive suspension point dynamometer data set. gl / k gu The maximum and minimum values ​​of The minimum value is as shown in the following formula: in, as well as are respectively the k of the dynamometer diagram of the dth suspension point in the gas-affected working condition gl , k gu as well as d=1,2,3,…,m5, m5 is the amount of data of the hanging point dynamometer diagram under the working condition of gas influence; The judgment thresholds of insufficient fluid supply condition σ1, σ2, and σ3 are set to be k in the comprehensive suspension point dynamometer data set. gl / k gu The maximum and minimum values ​​of The maximum value is as shown in the following formula: in, as well as are respectively the k of the fth suspension point indicator diagram under the condition of insufficient liquid supply gl , k gu as well as f=1,2,3,…,m6, where m6 is the amount of data of the hanging point dynamometer diagram under the condition of insufficient liquid supply.

10. The method for diagnosing faults in a rod pumping oil production process based on numerical simulation according to claim 9, characterized in that: Step 4 specifically includes: Step 4.1: Design a fault diagnosis reasoning mechanism based on the fault diagnosis knowledge base; The fault diagnosis knowledge base is a knowledge base based on production rules. The rule form is an expression of "IfP ThenQ", where P is the rule premise, that is, the judgment rule of the fault condition, and Q is the conclusion, that is, the fault type. According to the rule form, the fault diagnosis reasoning mechanism is designed to adopt forward reasoning. According to the given rod pump operation data, including the displacement sensor and load sensor located at the donkey head of the pumping unit to obtain the suspension point displacement and suspension point load data, the fault diagnosis knowledge base rules are searched for the fault condition judgment rule that matches the rod pump operation data. If the given rod pump operation data matches the fault condition judgment rule P, the fault type is obtained. Step 4.2: Obtain the operation data of the rod pump, and diagnose the fault of the rod pump oil production process based on the fault diagnosis knowledge base and the fault diagnosis reasoning mechanism; The displacement sensor and load sensor located at the donkey head of the pumping unit are used to obtain the operation data of the rod pump, including the suspension point displacement and suspension point load data, and obtain the suspension point indicator diagram data within a stroke; the suspension point indicator diagram is preprocessed and the characteristic parameters are calculated; the fault diagnosis reasoning mechanism is used to search for rules in the established fault diagnosis knowledge base to complete the traversal of the fault diagnosis rules. When the characteristic parameters do not match all the rules, the output is no solution and exit; when the characteristic parameters match a rule, the fault diagnosis result is output.

Images