Oil well intelligent control system and method based on edge cloud computing technology

Abstract

The invention discloses an oil well intelligent control system and method based on an edge cloud computing technology, and the system comprises an edge end and a cloud, and the edge end comprises a data collection module, an edge filtering module, an edge indicator diagram drawing module, an edge indicator diagram diagnosis module, an edge yield statistics module, an edge spacing adjustment module, and an energy consumption analysis module. Collected data are processed on site through edge equipment, namely an edge end, so that indicator diagrams are formed, the number of collection points of each indicator diagram is increased, the number of indicator diagrams is increased, working condition diagnosis available indicator diagrams are increased, the accuracy of the indicator diagrams and the working condition diagnosis accuracy are improved, and the problem that a large amount of collected information on site is difficult to transmit to a cloud end is solved. Therefore, the work diagram collection points received by the cloud are too few, the work diagram quality is low, the working condition diagnosis accuracy is low, and the utilization efficiency of the collected data is improved.

Description

Technical Field

[0001] The present invention belongs to the field of wellsite control, and particularly relates to an intelligent control system and method for oil wells based on edge-cloud computing technology Background Art

[0002] Technologies such as oil well dynamometer card production measurement, working condition diagnosis, and remote control of pumping units, as the core technologies for the construction of the Internet of Things for oil and gas production, have been industrially applied in various oilfields. Currently, an RTU is installed at the oil production wellhead to collect and process the dynamometer card and three-phase electrical parameter data of the oil well, control the start and stop of the pumping unit and give safety voice warnings, execute the intelligent intermittent production instruction of the oil well, etc., and a system for calculating oil well production and diagnosing working conditions is deployed at the factory department. One dynamometer card data is uploaded every 10 minutes, each dynamometer card has 200 groups of data, and 144 dynamometer card data are uploaded every day for liquid production calculation and oil well working condition diagnosis. In actual operation,

[0003] Limited by network transmission, the number of normal dynamometer cards uploaded is small and the quality is poor, resulting in low accuracy of dynamometer card production measurement and working condition diagnosis. The number of effective dynamometer cards uploaded is less than 80 cards / well·day, resulting in an accuracy of dynamometer card production measurement of about 70% and an accuracy of working condition diagnosis of about 85%, far from meeting the needs of intelligent oilfield construction.

[0004] The sampling frequency of the oil well dynamometer card acquisition equipment is in the millisecond level, and the number of acquisition points for each dynamometer card is between 2000 and 6000 points, while the actual output dynamometer card consists of 200 points. Different manufacturers have different methods for smoothing the dynamometer card, which seriously affects the quantitative analysis of the dynamometer card. Summary of the Invention

[0005] The purpose of the present invention is to provide an intelligent control system and method for oil wells based on edge-cloud computing technology to solve the problems of insufficient number of acquisition points and insufficient number of dynamometer cards of oil wells.

[0006] The purpose of the present invention is achieved by the following technical means. An intelligent control system for oil wells based on edge-cloud computing technology,

[0007] comprises an edge side and a cloud side,

[0008] The edge side includes,

[0009] a data acquisition module, connected to the sensors of the oil well to acquire the production data of the oil well;

[0010] an edge filtering module, storing the load and displacement information of the oil well acquired by the data acquisition module in the form of an array to obtain a load data queue and a displacement data queue, and performing sliding filtering on the load data queue and the displacement data queue to obtain the original filtered data of the load and displacement;

[0011] an edge dynamometer card drawing module, drawing a dynamometer card according to the original filtered data of the load and displacement

[0012] The edge dynamometer card diagnosis module is equipped with a dynamometer card recognition model to perform normalization diagnosis on the drawn dynamometer card data, and upload the dynamometer card data and diagnosis results to the cloud;

[0013] The edge production statistics module calculates the liquid production of the oil well according to the dynamometer card data and uploads the data to the cloud;

[0014] The energy consumption analysis module obtains the real-time cumulative power data, single stroke dynamometer card, and single stroke current diagram from the data acquisition module, obtains the factors affecting the energy consumption change, and uploads the data to the cloud;

[0015] The oil well control system controls the operation of the oil well equipment according to the information sent by the cloud;

[0016] The cloud includes

[0017] The cloud diagnosis module diagnoses the dynamometer card data and diagnosis results uploaded by the edge dynamometer card diagnosis module through the dynamometer card knowledge base.

[0018] The specific method of the sliding filter is to slide and compare the data in the load array and the displacement array, and use the deviation comparison method to compare the deviation of a certain data from its adjacent data. If the deviation > 2Kn, then this data is considered an abnormal value, and the abnormal value in the array is marked. Secondly, remove the abnormal points and use the previous data of the abnormal point data to replace the abnormal data to ensure the integrity of the data and the corresponding relationship with the load data.

[0019] The calculation method of the liquid production of the oil well is

[0020] Q = ηQ t , where Q is the actual liquid production, η is the pump efficiency, and F is the measured area of the dynamometer card drawn by the edge dynamometer card drawing module, F t - the area of the theoretical dynamometer card, Q t The theoretical liquid production.

[0021] It also includes an edge intermittent production adjustment module. Specifically, the edge intermittent production adjustment module inputs the pressure drop of the well opening and the pressure of the well closing at the liquid level; initializes the opening time and closing time of the intermittent pumping cycle; uses the swarm intelligence optimization algorithm to find the optimal solution for the switch well time with the maximization of the fitness function as the goal; outputs the optimal intermittent switch well time after meeting the iteration stop condition; and the edge intermittent production adjustment module controls the oil well to produce with the new intermittent switch well time.

[0022] The energy consumption analysis module specifically collects power data, dynamometer card data, and current diagram data under normal production conditions as sample data, and uses the Local Outlier Factor method to analyze the collected sample data to determine the boundary of the outlier region. Input the power data, dynamometer card data, and current diagram data during real-time production, perform outlier data analysis through the local outlier algorithm, and based on the analysis results, obtain the factors affecting energy consumption changes, and upload the factors affecting energy consumption changes to the cloud.

[0023] The edge dynamometer card diagnosis module specifically normalizes the drawn dynamometer card data, extracts feature point data, inputs the feature point data into the dynamometer card recognition model, and outputs the diagnosis result. The dynamometer card recognition model is a BP network with hidden layers. The transfer function of the neurons in the hidden layer is the tangent sigmoid, the transfer function of the neurons in the output layer uses the logarithmic sigmoid, the training function uses trainlm, and the backpropagation algorithm uses Levenberg-Marquadt.

[0024] The edge intermittent production adjustment module specifically takes the maximum profit per unit time as the goal and establishes the objective function:

[0025] Obj=(Q o *P o -T 2 *P e *P w ) / T

[0026] In the formula, Obj is the fitness, Q o is the liquid production of the oil well, P o is the wellhead oil pressure, T 2 is the production cycle, P e is the effective power, P w is the bottom hole pressure, and T is the intermittent production cycle;

[0027] In the first step, combined with the objective function, apply the intermittent production system optimization algorithm - Grey Wolf Optimization Algorithm. Take the production information of the oil well as the solution to the problem to be optimized. According to the range of the solution to the problem to be optimized, randomly initialize the parameter information of all intermittent production schemes: the open well pressure and the shut-in well pressure of the oil well;

[0028] In the second step, initialize the parameters of the oil well open time, the open well duration, and the shut-in well duration of the oil well;

[0029] In the third step, according to the problem to be optimized, calculate the fitness value of each scheme and sort it. The higher the fitness value, the closer the scheme is to the optimal solution. Set the individuals with the top three fitness values as Grey Wolf α, β, and δ respectively, and save the current optimal scheme information;

[0030] In the fourth step, update each parameter information in the scheme in turn;

[0031] In the fifth step, for the updated information of each parameter, recalculate the fitness value. Update the production information of the gray wolf α, β, and δ, as well as the historical optimal production information, according to the magnitude of the new fitness value. Update the oil well opening time, oil well opening duration, and oil well closing duration of the parameters.

[0032] In the sixth step, repeat the third step to the fifth step according to the number of iterations. The oil well control system outputs the start / stop of the well according to the newly adjusted intermittent production system, and the oil well produces according to the new intermittent production system. This process repeats in a loop, with the minimum start-up time under the maximum production of a single well as the convergence condition to generate the optimal operating parameters.

[0033] An oil well intelligent control method based on edge-cloud computing technology includes the following steps:

[0034] Data acquisition: The edge device collects data through oil well sensors.

[0035] Data preprocessing: According to the collected load and displacement information of the oil well, store it in the form of an array to obtain the load data queue and displacement data queue. Perform sliding filtering on the load data queue and displacement data queue to obtain the original filtered data of the load and displacement.

[0036] Indicator diagram drawing: Draw an indicator diagram according to the original filtered data of the load and displacement.

[0037] Indicator diagram diagnosis: The local edge device performs fault diagnosis on the drawn indicator diagram through the indicator diagram recognition model at the edge device, and sends the indicator diagram and the diagnosis result to the cloud for secondary diagnosis in the indicator diagram knowledge base.

[0038] Energy consumption analysis: Collect real-time cumulative power data, single stroke indicator diagram, and single stroke current diagram, obtain the factors affecting energy consumption changes, and upload the data to the cloud. The cloud feeds back the indicator diagram diagnosis result and the factors affecting energy consumption changes to the management personnel.

[0039] Single stroke production calculation and daily cumulative production statistics are also carried out by drawing the indicator diagram.

[0040] The edge device also collects the oil well opening pressure and oil well closing pressure, obtains the new oil well opening time, oil well opening duration, and oil well closing duration through the edge intermittent production adjustment module, and controls the oil well equipment to carry out intermittent production with the new oil well opening time, oil well opening duration, and oil well closing duration.

[0041] The beneficial effects of the present invention are as follows: Through the edge device, that is, the edge side, the collected data is processed on-site to form a dynamometer card, which improves the number of acquisition points for each dynamometer card, increases the number of dynamometer cards, increases the available dynamometer cards for working condition diagnosis, thereby improving the accuracy of the dynamometer card and the accuracy of working condition diagnosis, avoiding the difficulty of transmitting a large amount of on-site collected information to the cloud, resulting in too few acquisition points for the dynamometer cards received by the cloud and low quality of the dynamometer cards, and further leading to low accuracy of working condition diagnosis, and improving the utilization efficiency of the collected data.

[0042] At the same time, through the edge device, the on-site collected information is adjusted for intermittent pumping, production statistics and energy consumption analysis, reducing the data transmission level, reducing intermediate links, improving data transmission efficiency and stability, and reducing on-site operation workload. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 is the structural block diagram of the intelligent control system for oil wells based on edge-cloud computing technology;

[0044] Figure 2 is the schematic diagram of the calculation principle of the liquid volume of the indicator diagram;

[0045] Figure 3 is the flow chart of intermittent opening adjustment;

[0046] Figure 4 is the flow chart of energy consumption analysis;

[0047] The present invention will be further described in detail below with reference to the drawings and embodiments. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0048]

Embodiment 1

[0049] As Figure 1 shown, an intelligent control system for oil wells based on edge-cloud computing technology

[0050] includes an edge side and a cloud side,

[0051] The edge side includes

[0052] a data acquisition module, which is connected to the sensors of the oil well to collect the production data of the oil well;

[0053] an edge filtering module, which stores the load and displacement information of the oil well collected by the data acquisition module in the form of an array to obtain a load data queue and a displacement data queue, and performs sliding filtering on the load data queue and the displacement data queue to obtain the original filtered data of the load and displacement;

[0054] an edge indicator diagram drawing module, which draws an indicator diagram according to the original filtered data of the load and displacement;

[0055] The edge dynamometer card diagnosis module is equipped with a dynamometer card recognition model to perform normalization diagnosis on the drawn dynamometer card data and upload the dynamometer card data and diagnosis results to the cloud;

[0056] The edge production statistics module calculates the liquid production of the oil well based on the dynamometer card data and uploads the data to the cloud;

[0057] The energy consumption analysis module obtains the factors affecting energy consumption changes based on the real-time cumulative power data, single stroke dynamometer card, and single stroke current diagram obtained by the data acquisition module and uploads the data to the cloud;

[0058] The oil well control system controls the operation of the oil well equipment according to the information sent by the cloud;

[0059] The cloud includes,

[0060] The cloud diagnosis module diagnoses the dynamometer card data and diagnosis results uploaded by the edge dynamometer card diagnosis module through the dynamometer card knowledge base.

[0061] The edge side is an edge computing gateway, such as the Huawei AR502H, which is set at the oil well. The data acquisition module is connected to various existing sensors of the oil well and the wellhead controller RTU to collect the production data of the oil well.

[0062] Among them, for the load and displacement data used to draw the dynamometer card, the load and displacement sensors connect the 4 - 20mA signal to the data acquisition module. The data acquisition module converts the analog signals of the load and displacement sensors into digital signals through an AD chip and stores the data in the form of an array in real time through data acquisition software to obtain the load and displacement data queues of load[1..200] and displacement[1..200].

[0063] The knowledge base is a set of dynamometer card samples formed by separating and marking the dynamometer cards by business experts, performing image enhancement, and continuously enriching and improving it with the successfully recognized dynamometer cards, and finally forming the knowledge base.

[0064] The specific method of the sliding filter is to slide and compare the data in the load array and the displacement array, and use the deviation comparison method to compare the deviation of a certain data from its adjacent data. If the deviation > 2Kn, then this data is considered an abnormal value, and the abnormal values in the array are marked. Secondly, remove the abnormal points and use the previous data of the abnormal point data to replace the abnormal data to ensure the integrity of the data and its corresponding relationship with the load data.

[0065] The calculation method of the liquid production of the oil well is,

[0066] Q = ηQ t , where Q is the actual liquid production, η is the pump efficiency, and F is the measured area of the dynamometer card by the edge dynamometer card drawing module, Ft - Theoretical dynamometer area, Q t Theoretical liquid production.

[0067] It also includes an edge spacing adjustment module, which specifically inputs the liquid surface well opening pressure drop and the liquid surface well closing pressure; initializes the well opening time and well closing time of the inter-pumping cycle; uses a swarm intelligence optimization algorithm to find the optimal solution for the well opening and closing time with the goal of maximizing the fitness function; outputs the optimal inter-pumping well opening and closing time after the iteration stop condition is met; and the edge spacing adjustment module controls the oil well to produce at the new inter-pumping well opening and closing time.

[0068] The edge power diagram diagnosis module is specifically to normalize the drawn power diagram data, extract feature point data, input the feature point data into the power diagram recognition model, and output the diagnosis result. The power diagram recognition model is a BP network of the hidden layer, the transfer function of the hidden layer neurons is a tangent S type, the transfer function of the output layer neurons uses a logarithmic S type, the training function uses trainlm, and the back propagation algorithm uses Levenberg-Marquadt. Feature point data is extracted from each power diagram annotated in the knowledge base and input into the neural network for training until the neural network reaches a predetermined recognition error.

[0069] The edge computing device integrates the dynamometer diagram recognition model. The edge device normalizes the drawn dynamometer diagram data, extracts feature point data, and inputs the feature point data into the diagnosis model to complete the preliminary diagnosis of the intelligent working condition. After the dynamometer diagram data and diagnosis results are uploaded to the cloud, the cloud facilitates the comparison of the dynamometer diagram knowledge base to complete further diagnosis and improve the dynamometer diagram diagnosis results.

[0070] The cloud-based knowledge base enriches the database based on the information collected by the edge devices, optimizes the dynamometer diagram recognition model, and sends the optimized dynamometer diagram recognition model to the local edge devices to update the recognition model and improve the local recognition accuracy.

[0071] The energy consumption analysis module specifically collects the electric quantity data, power diagram data, and current diagram data under normal production conditions as sample data, uses the local outlier factor method to analyze the collected sample data, and determines the outlier area boundary; inputs the electric quantity data, power diagram data, and current diagram data during real-time production, and uses the local outlier algorithm to perform outlier data analysis, and according to the analysis results, obtains the factors affecting the change of energy consumption, and uploads the factors affecting the change of energy consumption to the cloud for users to make comprehensive decisions in combination with the diagnostic results of the indicator diagram. The factors affecting the change of energy consumption analyzed include sprint / power factor / daily liquid production / water content / crude oil viscosity / balance / dynamic liquid level.

[0072] like Figure 3As shown, the dynamometer card production measurement and diagnosis program calculates the dynamometer card for each stroke, makes an evaluation of the downhole capacity after a period of time, and adjusts the intermittent production system according to the evaluation results. Specifically, the edge intermittent production adjustment module establishes an objective function with the goal of maximizing the revenue per unit time:

[0073] Obj=(Q o *P o -T 2 *P e *P w ) / T

[0074] In the formula, Obj is the fitness, Q o is the liquid production of the oil well, P o is the wellhead oil pressure, T 2 is the open well period, P e is the effective power, P w is the bottom hole pressure, and T is the intermittent production period;

[0075] First step, combined with the objective function, apply the intermittent production system optimization algorithm - Grey Wolf Optimization Algorithm. Using the liquid production information Q o of the oil well as the solution to the problem to be optimized, randomly initialize the parameter information of all intermittent production schemes according to the range of the solution to the problem to be optimized: the open well pressure and the shut-in well pressure of the oil well;

[0076] Second step, initialize the parameters of the open well time, open well duration, and shut-in well duration of the oil well; and calculate the problem to be optimized, that is, the open well time, open well duration, and shut-in well duration of the oil well.

[0077] Third step, according to the problem to be optimized, calculate the fitness value of each scheme and sort them. The higher the fitness value, the closer the scheme is to the optimal solution. Set the individuals with the top three fitness values as Grey Wolf α, β, and δ respectively, and save the current optimal scheme information;

[0078] Fourth step, update each parameter information in the scheme in turn;

[0079] Fifth step, for the updated information of each parameter, recalculate the fitness value, update the production information of Grey Wolf α, β, and δ and the historical optimal production information according to the size of the new fitness value, and update the parameters of the open well time, open well duration, and shut-in well duration of the oil well;

[0080] Sixth step, repeat the third step to the fifth step according to the number of iterations. The oil well control system outputs the start and stop of the well according to the newly adjusted intermittent production system, and the oil well produces according to the new intermittent production system. This process repeats in a loop, with the minimum start-up well time at the maximum production of a single well as the convergence condition to generate the best operating parameters.

[0081] An intelligent control method for oil wells based on edge-cloud computing technology, comprising the following steps,

[0082] Data acquisition, where the edge device collects data through oil well sensors;

[0083] Data preprocessing, storing the collected load and displacement information of the oil well in the form of an array to obtain a load data queue and a displacement data queue, and performing sliding filtering on the load data queue and the displacement data queue to obtain the original filtered data of the load and displacement;

[0084] Indicator diagram drawing, drawing an indicator diagram based on the original filtered data of the load and displacement;

[0085] Indicator diagram diagnosis, where the local edge device performs fault diagnosis on the drawn indicator diagram through an indicator diagram recognition model at the edge device, and sends the indicator diagram and the diagnosis result to the cloud for secondary diagnosis in the indicator diagram knowledge base;

[0086] Energy consumption analysis, collecting real-time cumulative power data, single stroke indicator diagram, and single stroke current diagram, obtaining the factors affecting energy consumption changes, and uploading the data to the cloud. The cloud feeds back the indicator diagram diagnosis result and the factors affecting energy consumption changes to the management personnel.

[0087] It also calculates the production per single stroke and statistics the daily cumulative production by drawing the indicator diagram.

[0088] The edge device also collects the oil well open well pressure and the oil well shut-in pressure, obtains the new oil well open time, the oil well open duration, and the oil well shut-in duration through the edge intermittent production adjustment module, and controls the oil well equipment to perform intermittent production with the new oil well open time, the oil well open duration, and the oil well shut-in duration.

[0089] Based on the computing power of the edge device, the edge filtering module performs sliding filtering on the load data queue and the displacement data queue, slidingly compares the data in the load array and the displacement array, and uses the deviation comparison method to compare the deviation between a certain data and its adjacent data. If the deviation > 2Kn, then this data is considered an outlier, and the outlier in the array is marked. Secondly, the outlier is removed, and the previous data of the outlier data is used to replace the outlier data to ensure the integrity of the data and its corresponding relationship with the load data. After the digital filtering calculation is completed, it is put back into the array to obtain the original filtered data.

[0090] The edge indicator diagram drawing module draws an indicator diagram based on the original filtered data of the load and displacement. The horizontal coordinate data of the indicator diagram is obtained through the displacement sensor, and the vertical coordinate data of the indicator diagram is obtained through the load sensor.

[0091] Integrate the original normalization diagnosis model located in the cloud or other existing dynamogram diagnosis models into the edge device. Through the edge dynamogram diagnosis module, perform preliminary diagnosis on the drawn dynamograms. Since the sampling data of the dynamograms are directly the on-site well data collected by the data acquisition module and do not need to be transmitted from the well site to the cloud, the data volume is larger, the number of sampling points of the dynamograms is more, and the generated dynamograms do not need to be transmitted over a long distance. Therefore, the working conditions can be diagnosed based on more dynamograms with more sampling points, solving the problems of low accuracy of production calculation due to the small number of existing dynamograms and the small number of sampling points of the dynamograms, and further low diagnosis accuracy due to diagnosing the working conditions based on the dynamograms with low accuracy.

[0092] After the edge dynamogram diagnosis module completes the preliminary diagnosis, send the diagnosis result and the dynamogram data to the cloud together. The cloud diagnosis module uses the existing dynamogram knowledge base to perform secondary diagnosis on the uploaded dynamograms to further improve the diagnosis accuracy of the dynamograms.

[0093] Such as Figure 2 As shown, the edge device also counts the liquid production volume through the edge production volume statistics module.

[0094] Specifically, Q = ηQ t , where Q is the actual liquid production volume, η is the pump efficiency. F is the measured dynamogram area of the edge dynamogram drawing module, F t - is the theoretical dynamogram area, Q t The theoretical liquid production volume.

[0095] The theoretical liquid production volume Qt is calculated according to the following formula:

[0096]

[0097] In the formula, D is the pump diameter of the sucker rod pump, S is the stroke of the pumping unit; n is the pumping speed of the pumping unit.

[0098] The above formula can also be written as:

[0099] Q t = KSn

[0100] In the formula, K is the pump constant, which is only related to the pump diameter.

[0101] The theoretical dynamogram is the dynamogram obtained theoretically when it is considered that the polished rod only bears the static load of the liquid column above the cross-sectional area of the sucker rod string and the piston. Figure 2 As shown:

[0102] It can be seen from the figure that point A is the bottom dead center, point C is the top dead center, the inclined line AB represents the load-increasing line of the polished rod with increasing load, and the inclined line CD represents the load-decreasing line of the polished rod with decreasing load.

[0103] For one stroke of the pumping unit, the data points collected by the load sensor and the displacement sensor form an original data array. When this original data is compared with the theoretical dynamometer card data and the difference exceeds the acceptable range, the system considers it abnormal data.

[0104] Based on the theoretical dynamometer card of the plunger pump, the wave equation is used to realize the conversion of the surface dynamometer card to the downhole dynamometer card of the pump. According to the characteristics of the dynamometer card, corrections are made to the liquid supply degree, leakage degree, gas influence, etc., and the effective stroke is calculated. Finally, the liquid production volume of a single well is calculated.

[0105] Calculate the liquid production volume of one or multiple oil wells according to the measured dynamometer card data, and be able to calculate for several wells during a specific time period. At the same time, generate a production daily report to realize the retrieval of the production of a single well or multiple wells at any time.

[0106] Also, through the edge intermittent adjustment module, adjust the intermittent time of the oil wells;

[0107] The intelligent oil well control system outputs the start and stop of the well according to the newly adjusted intermittent regime, and the oil well produces according to the new intermittent regime. This process repeats continuously, with the minimum start-up well time under the maximum production of a single well as the convergence condition to generate the optimal operating parameters.

[0108] Also, through the energy consumption analysis module, adjust different production strategies and operating parameters, such as adjusting the intermittent regime, stroke frequency, and balance.

[0109] As Figure 4 shown, the energy consumption analysis module obtains real-time cumulative power data, dynamometer card per stroke, and current graph per stroke according to the data of the sensor. According to the relationship between electromagnetism and force, an energy consumption model of the pumping unit is established, and the power loss is decomposed by link. After obtaining different influencing factors and making comprehensive decisions on multiple influencing factors, different production strategies and operating parameters are adjusted.

[0110] By analyzing the energy flow of the pumping unit and the mechanical transmission energy flow, it can be concluded that the only external input end of the pumping system is the energy received by the motor of the pumping system from the power grid. The energy undergoes internal losses in the motor, including frictional losses, magnetic tape losses, and copper and iron losses. It flows into the mechanical part, and the losses in the mechanical transmission part are the frictional losses of the belt drive, the frictional losses of the operation of the speed reducer, and the frictional losses and potential energy increase during the rotation of the crank, the rotation of the connecting rod, the swing of the walking beam, and the movement of the sucker rod, as well as the elastic deformation generated during the up and down reciprocating movement of the sucker rod. Finally, it reaches the downhole oil pump, and the losses and stroke losses in the fluid transmission part are suffered.

[0111] Under the condition that the motor loss and transmission loss are relatively fixed, and the balance weight stores energy during the downstroke and releases energy during the upstroke, weakening the motor power fluctuation between the upstroke and downstroke. Based on the closed system theory, the change in the motor output ability is reflected at the polished rod, which can directly show the change in the surface load. According to this, an electric energy loss model is established, which can comprehensively diagnose the working conditions of oil wells based on the electric energy situation, assist in production decision-making, adjust the stroke frequency, adjust the balance, and implement intermittent lighting. For example, an increase in the consumed electric energy indicates that the polished rod load becomes larger, and there may be situations such as pump sticking or an increase in the original viscosity; a decrease in the consumed electric energy may indicate situations such as pump leakage or natural flow.

[0112] Specifically, collect the power data, dynamometer card data, and current card data under normal production conditions as sample data, and use the local outlier factor method to analyze the collected sample data to determine the boundary of the outlier region; input the power data, dynamometer card data, and current card data during real-time production, perform outlier data analysis through the local outlier algorithm, and based on the analysis results, obtain the factors affecting the energy consumption change, and upload the factors affecting the energy consumption change to the cloud.

Claims

1. An intelligent control system for oil wells based on edge-cloud computing technology, characterized in that: it includes an edge side and a cloud side, the edge side includes, a data acquisition module, connected to the sensors of the oil well, to acquire the production data of the oil well; an edge filtering module, which stores the load and displacement information of the oil well collected by the data acquisition module in the form of an array to obtain a load data queue and a displacement data queue, and performs sliding filtering on the load data queue and the displacement data queue to obtain the original filtered data of the load and displacement; an edge dynamometer card drawing module, which draws a dynamometer card according to the original filtered data of the load and displacement; an edge dynamometer card diagnosis module, which is equipped with a dynamometer card recognition model to perform normalization diagnosis on the drawn dynamometer card data, and uploads the dynamometer card data and the diagnosis results to the cloud; an edge production statistics module, which calculates the liquid production of the oil well according to the dynamometer card data and uploads the data to the cloud; an energy consumption analysis module, which obtains the factors affecting the energy consumption change according to the real-time cumulative power data, the single stroke dynamometer card, and the single stroke current diagram obtained by the data acquisition module, and uploads the data to the cloud; an oil well control system, which controls the operation of the oil well equipment according to the information sent by the cloud; the cloud side includes, a cloud diagnosis module, which diagnoses the dynamometer card data and the diagnosis results uploaded by the edge dynamometer card diagnosis module through a dynamometer card knowledge base.

2. An intelligent control system for oil wells based on edge-cloud computing technology according to claim 1, characterized in that: the specific method of the sliding filtering is to slide and compare the data in the load array and the displacement array, and adopt the deviation comparison method to compare the deviation between a certain data and its adjacent data. If the deviation > 2Kn, it is considered that the data is a singular value, and the singular value in the array is marked. Secondly, remove the singular point, and use the previous data of the singular point data to replace the singular data to ensure the integrity of the data and the corresponding relationship with the load data.

3. An intelligent control system for oil wells based on edge-cloud computing technology according to claim 1, characterized in that: the calculation method of the liquid production of the oil well is, Q = ηQ t , where Q is the actual liquid production rate, and η is the pump efficiency. Among them F is the measured area of the indicator diagram by the edge work diagram drawing module, F t - the area of the theoretical indicator diagram, Q t The theoretical liquid production rate.

4. An intelligent control system for oil wells based on edge-cloud computing technology according to claim 1, characterized in that: it further includes an edge intermittent production adjustment module. The edge intermittent production adjustment module specifically inputs the pressure drop of the well opening and the pressure of the well closing at the liquid level; initializes the opening time and the closing time of the intermittent pumping cycle; uses a swarm intelligence optimization algorithm to find the optimal solution for the switch well time with the maximization of the fitness function as the goal; outputs the optimal intermittent pumping switch well time after meeting the iteration stop condition; the edge intermittent production adjustment module controls the oil well to produce with the new intermittent pumping switch well time.

5. An intelligent control system for oil wells based on edge-cloud computing technology according to claim 1, characterized in that: The energy consumption analysis module specifically collects power data, dynamogram data, and current diagram data under normal production conditions as sample data, and uses the local outlier factor method to analyze the collected sample data to determine the boundary of the outlier region. Input the power data, dynamogram data, and current diagram data during real-time production, perform outlier data analysis through the local outlier algorithm, and based on the analysis results, obtain the factors affecting energy consumption changes, and upload the factors affecting energy consumption changes to the cloud.

6. A smart oil well control system based on edge-cloud computing technology as described in claim 1, characterized in that: The edge dynamogram diagnosis module specifically performs normalization processing on the drawn dynamogram data, extracts feature point data, inputs the feature point data into the dynamogram recognition model, and outputs a diagnosis result. The dynamogram recognition model is a BP network with hidden layers. The transfer function of the neurons in the hidden layer is the tangent sigmoid, the transfer function of the neurons in the output layer uses the logarithmic sigmoid, the training function uses trainlm, and the backpropagation algorithm uses Levenberg-Marquadt.

7. A smart oil well control system based on edge-cloud computing technology as described in claim 4, characterized in that: The edge intermittent production adjustment module specifically takes maximizing the profit per unit time as the goal and establishes an objective function: Obj = (Q o * P o - T 2 * P e * P w ) / T where Obj is the fitness, Q o is the liquid production of the oil well, P o is the wellhead oil pressure, T 2 is the open - well period, P e is the effective power, P w is the bottom - hole pressure, and T is the intermittent production period; First step, in combination with the objective function, apply the intermittent production system optimization algorithm - grey wolf optimization algorithm, use the production information of the oil well as the solution to the problem to be optimized, and randomly initialize the parameter information of all intermittent production schemes according to the range of the solution to the problem to be optimized: the opening pressure of the oil well and the closing pressure of the oil well; Second step, initialize the parameters of the oil well opening time, the oil well opening duration, and the oil well closing duration; Third step, according to the problem to be optimized, calculate the fitness value of each scheme and sort it. The higher the fitness value, the closer the scheme is to the optimal solution. Set the individuals with the top three fitness values as grey wolves α, β, and δ respectively, and save the current optimal scheme information; Fourth step, update each parameter information in the scheme in turn; Fifth step, for the updated information of each parameter, recalculate the fitness value, update the production information of grey wolves α, β, and δ and the historical optimal production information according to the size of the new fitness value, and update the parameters of the oil well opening time, the oil well opening duration, and the oil well closing duration; Sixth step, repeat the third step to the fifth step according to the number of iterations. The oil well control system outputs the start and stop of the well according to the newly adjusted intermittent production system, and the oil well produces according to the new intermittent production system, and so on in a cycle. Taking the minimum start-up time under the maximum production of a single well as the convergence condition, the best operating parameters are generated.

8. A smart oil well control method based on edge-cloud computing technology, characterized in that: It includes the following steps, Data acquisition, the edge device collects data through oil well sensors; Data preprocessing, according to the collected load and displacement information of the oil well, store it in the form of an array to obtain a load data queue and a displacement data queue, and perform sliding filtering on the load data queue and the displacement data queue to obtain the original filtered data of the load and displacement; Draw the indicator diagram, and draw the indicator diagram according to the original filtered data of load and displacement; Indicator diagram diagnosis: The local edge device diagnoses faults at the edge device through the indicator diagram recognition model based on the drawn indicator diagram, and sends the indicator diagram and the diagnosis result to the cloud for secondary diagnosis of the indicator diagram knowledge base; Energy consumption analysis: Collect real-time cumulative power data, single stroke indicator diagram, and single stroke current diagram, obtain the factors affecting energy consumption changes, and upload the data to the cloud. The cloud feeds back the indicator diagram diagnosis result and the factors affecting energy consumption changes to the management personnel.

9. A method for intelligent control of oil wells based on edge-cloud computing technology according to claim 8, characterized in that: It also calculates the production per single stroke and statistics the daily cumulative production by drawing the indicator diagram.

10. A method for intelligent control of oil wells based on edge-cloud computing technology according to claim 8, characterized in that: The edge device also collects the open well pressure and shut-in well pressure of the oil well, obtains the new open well time, open well duration and shut-in well duration of the oil well through the edge intermittent production adjustment module, and controls the oil well equipment to carry out intermittent production with the new open well time, open well duration and shut-in well duration.

Images