Digital pumping unit monitoring system

Abstract

The invention discloses a digital pumping unit monitoring system, which relates to the technical field of pumping units and is characterized in that the initial position of a suspension center is processed through a suspension center initialization unit; the suspension point position is monitored through a suspension point position monitoring unit, and a monitoring result is obtained; then, the indicator diagram unit constructs an indicator diagram of a suspension point, and data cleaning and correction are carried out on the constructed indicator diagram based on a monitoring result; utilizing a feature extraction unit to perform feature extraction and data annotation on the corrected indicator diagram to obtain an annotation data set; based on a data processing unit, processing the vibration frequency data under different fault types to obtain a processing result; and finally, the fault monitoring unit constructs a fault monitoring model, jointly trains the fault monitoring model by using the labeled data set and the processing result, and monitors the oil pumping unit in real time based on the trained fault monitoring model. According to the invention, the efficiency and accuracy of the existing pumping unit monitoring system are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of pumping units, and in particular to a digital pumping unit monitoring system. Background Art

[0002] The digital pumping unit system mainly consists of a sucker rod pump, a pumping unit, and sucker rods. The main hardware structure of the pumping unit includes a walking beam, a rocker arm, a connecting rod, a crank mechanism, a reducer, a power device, and auxiliary devices; the sucker rod is a rod device for an oil well, connected to a polished rod at the upper end and a sucker rod pump at the lower end, playing a role in connecting the ground and the wellbore; the sucker rod pump is composed of a pump barrel, a plunger, a suction valve, and a traveling valve; a dynamometer card instrument (load sensor, displacement sensor), a control box, etc. are responsible for collecting production parameters.

[0003] Currently, for the monitoring of the digital pumping unit system, it is mostly necessary for staff to take turns on duty at the oil production site for 24 hours. Then, load data and displacement data of the polished rod suspension point are collected through load displacement sensors to form a dynamometer card at the backend (the abscissa is the suspension point displacement, and the ordinate is the suspension point load). Professional personnel analyze the dynamometer card to judge the type of failure of the pumping unit. However, the above method has the following problems: The pumping unit converts the rotational motion of the motor into the reciprocating up and down motion of the polished rod (suspension point) through a crank connecting rod mechanism. That is to say, in an ideal state, when the crank rotates one week, the output distance of the motor is equal to the distance of one up and down stroke of the suspension point, that is, when the suspension point completes one up and down stroke, it returns to the starting point. However, because the pumping unit works continuously in a harsh environment full of dust, rain, and saline-alkali air all year round, problems such as uneven force and eccentric wear exist in the bearings, resulting in problems with the support bearings, crossbeam bearings, and crankpin bearings, and there are clearance fits in the hinged parts such as the crank, connecting rod, and rocker arm. After long-term operation in a harsh environment, the wear intensifies, and displacement loss will occur due to the "idle stroke" during the stroke. Eventually, the suspension point cannot return to the starting point, that is, there is an error between the displacement of a single up and down stroke and the displacement of a single up and down stroke in the ideal state, and this error will continue to increase over time. Finally, the dynamometer card formed by collecting displacement load data cannot accurately reflect the failure of the pumping unit, reducing the monitoring accuracy of the pumping unit monitoring system; moreover, the current method of manual duty and professional personnel analyzing the dynamometer card at the backend has low monitoring efficiency and cannot detect and handle failures in a timely manner.

[0004] In summary, at present, it is difficult to guarantee the monitoring efficiency and accuracy of the digital pumping unit system. Summary of the Invention

[0005] In view of this, the present application provides a digital pumping unit monitoring system to improve the monitoring efficiency and accuracy of the digital pumping unit system.

[0006] The first aspect of the present application provides a digital pumping unit monitoring system, including: Suspension point initialization unit: Obtain the initial position of the suspension point and the change value of the initial position at adjacent moments, and determine whether the change value exceeds a first threshold. If so, obtain the real-time position of the suspension point at the adjacent moment as the initial position; if not, do nothing. Suspension point position monitoring unit: Based on the initial position of the suspension point, obtain the time for the crank to rotate one week to divide the time interval. Within a single complete time interval, respectively obtain the output distance of the motor and the displacement distance of the suspension point, and plot the output distance of the motor and the displacement distance of the suspension point in the same coordinate system, analyze and monitor the data in the coordinate system to obtain the monitoring result. Indicator diagram unit: Within a single complete time interval, construct an indicator diagram of the suspension point, and based on the monitoring result, determine whether to perform data cleaning on the indicator diagram of the suspension point. If so, clean the indicator diagram of the suspension point; if not, correct the indicator diagram of the suspension point. Feature extraction unit: Obtain all the corrected indicator diagrams and place them in a first preset data set. Adopt the method of multi-sampling point random sampling for all the indicator diagrams in the first preset data set to obtain the feature data of different indicator diagrams; based on the feature data of all the indicator diagrams in the first preset data set, label the indicator diagrams of different fault types to obtain a labeled data set. Data processing unit: Collect the vibration frequency data of the digital pumping unit within multiple complete time intervals at multiple time scales, and process them using a preset method to obtain multiple processing results. Fault monitoring unit: Construct a fault monitoring model, use all the labeled data sets as the first influencing factor and all the processing results as the second influencing factor to jointly train the fault monitoring model, and based on the trained fault monitoring model, perform real-time monitoring on the digital pumping unit.

[0007] In a possible implementation manner of the first aspect, obtaining the time for the crank to rotate one week to divide the time interval includes: Obtain the starting position of the crank, record the moment when the crank starts to move as moment t0, and the moment when the crank returns to the starting position as moment t1. The time interval is formed between moment t0 and moment t1.

[0008] In a possible implementation manner of the first aspect, within a single complete time interval, respectively obtaining the output distance of the motor and the displacement distance of the suspension point includes: Within a single complete time interval, obtain the displacement distance of the suspension point through a displacement sensor, and obtain the output distance of the motor by obtaining the rotational speed of the motor and the circumference of the motor main shaft.

[0009] In a possible implementation of the first aspect, plotting the output distance of the motor and the displacement distance of the suspension point in the same coordinate system includes: Taking time as the horizontal axis and dividing the horizontal axis into multiple time intervals; The vertical axis is dimensionless, and point A and point B are determined on the vertical axis; Obtain the output distance of the motor in all time intervals, and starting from point A, plot the output distance of the motor proportionally within the corresponding time interval; Obtain the displacement distance of the suspension point in all time intervals, and starting from point B, plot the displacement distance of the suspension point proportionally within the corresponding time interval.

[0010] In a possible implementation of the first aspect, analyzing and monitoring the data in the coordinate system to obtain a monitoring result includes: In the coordinate system, obtain the output distance of the motor within any time interval, denoted as L1, and the displacement distance of the suspension point, denoted as L2; If L2 is greater than L1, perform data cleaning on the dynamogram of the suspension point within the corresponding time interval; If L2 is less than or equal to L1, calculate the difference between L1 and L2, and determine whether the difference is greater than a second threshold. If so, correct the dynamogram of the suspension point within the corresponding time interval. If not, do nothing.

[0011] In a possible implementation of the first aspect, correcting the dynamogram of the suspension point includes: Obtain any dynamogram of the suspension point as a first dynamogram, and obtain the difference between the output distance of the motor and the displacement distance of the suspension point within the time interval corresponding to the first dynamogram, denoted as a first difference; Obtain the continuous longitudinal line feature between the upper right vertex and the lower right vertex in the first dynamogram, denoted as a first line feature, and obtain the continuous horizontal line feature extending from the upper right vertex to the vertical axis, denoted as a second line feature, and obtain the continuous horizontal line feature extending from the lower right vertex to the vertical axis, denoted as a third line feature; Based on the first difference, the second line feature, and the third line feature, determine the upper right increment vertex and the lower right increment vertex in the first dynamogram; Based on the second line feature, complete the connection between the upper right vertex and the upper right increment vertex; based on the third line feature, complete the connection between the lower right vertex and the lower right increment vertex; based on the first line feature, complete the connection between the upper right increment vertex and the lower right increment vertex; Complete the correction of the first dynamogram.

[0012] In a possible implementation of the first aspect, the preset method includes: Based on the vibration frequency data collected within multiple complete time intervals under multiple time scales, construct multiple coordinate curves of the vibration frequency data changing with time; Obtain any fault type denoted as the first fault type, and obtain multiple coordinate curves under the first fault type, and store them in a second preset data set; Segment each coordinate curve in the second preset data set with the time interval to obtain M1 time intervals, and collect the slopes of N1 data points within each time interval to obtain the collection result; Based on the collection result, calculate the average slope of the corresponding data points in the corresponding time intervals of the multiple coordinate curves in the second preset data set; Establish a database and store the average slopes of the corresponding data points in the corresponding time intervals of the multiple coordinate curves in the second preset data set; When there is a single coordinate curve denoted as the first coordinate curve, and the slopes of N2 data points within a single time interval of the first coordinate curve meet the preset conditions, it is determined that the time interval corresponding to the first coordinate curve matches the time interval corresponding to the first fault type; When there are M2 matching time intervals in the first coordinate curve, and the ratio of M2 / M1 exceeds the fourth threshold, it is determined that the first coordinate curve matches the coordinate curve corresponding to the first fault type.

[0013] In a possible implementation manner of the first aspect, the preset conditions include: The difference between the slopes of the N2 data points and the average slope of the corresponding data points in the database is less than the set value, and the ratio of N2 / N1 exceeds the third threshold.

[0014] In a possible implementation manner of the first aspect, based on the trained fault monitoring model, the real-time monitoring of the digital pumping unit includes: Obtain the corrected indicator diagram of the digital pumping unit denoted as the second indicator diagram, and the processing result of the corresponding vibration frequency data within the preset time period denoted as the first processing result; Input the second indicator diagram and the first processing result into the trained fault monitoring model, and output the fault type of the digital pumping unit.

[0015] In a possible implementation manner of the first aspect, within a single complete time interval, constructing an indicator diagram of the polished rod includes: Within a single complete time interval, collect the displacement data of the polished rod through a displacement sensor and collect the load data of the polished rod through a load sensor; use the displacement data of the polished rod as the abscissa and the load data of the polished rod as the ordinate to construct an indicator diagram of the polished rod.

[0016] Compared with the prior art, the present application provides a digital pumping unit monitoring system, including: A polished rod initialization unit, by obtaining the initial position of the polished rod and the change value of the initial position at adjacent moments, if the change value exceeds the first threshold, then obtain the real-time position of the polished rod at the adjacent moment as the new initial position to complete the initialization of the polished rod position; if the initial position of the polished rod is not initialized, it will lead to a deviation between the displacement distance of the polished rod measured by the displacement sensor and the actual displacement distance of the polished rod, and this deviation will always exist, directly affecting the generation of the indicator diagram, thereby affecting the judgment of the working condition of the pumping unit; A polished rod position monitoring unit. Since the pumping unit has been continuously working in a harsh environment full of dust, rain, and saline-alkali all year round, displacement loss will occur due to "idle stroke" during the stroke, resulting in the polished rod not being able to return to the starting point and this error will continuously increase with time. Finally, the indicator diagram formed by collecting displacement load data cannot accurately reflect the faults of the pumping unit. Therefore, the polished rod position monitoring unit takes one complete revolution of the crank movement as the benchmark, and respectively obtains the output distance of the motor and the displacement distance of the polished rod within a single complete time interval. Theoretically, the output distance of the motor is equal to the displacement distance of the polished rod. However, due to the long-term operation of the pumping unit in a harsh environment, which aggravates wear, the output distance of the motor is greater than the displacement distance of the polished rod. This module monitors the output distance of the motor and the displacement distance of the polished rod more intuitively by establishing a unified coordinate system to obtain the monitoring result; An indicator diagram unit. Its principle is to collect the displacement distance and load of the polished rod within a single complete stroke (time interval), construct a coordinate system of the polished rod load with respect to the polished rod displacement distance, and obtain the indicator diagram of the pumping unit. By analyzing the indicator diagram, the working condition of the pumping unit is judged; but as mentioned before, during the actual operation process, the displacement distance of the polished rod may not be equal to the output distance of the motor (one is that the initial position of the polished rod may change, and the other is that the polished rod cannot return to the initial position during a single stroke), resulting in the inaccuracy of the indicator diagram generated based on the collected data, which of course directly affects the judgment of the working condition of the pumping unit. Therefore, based on the monitoring result, this module first performs data cleaning on some indicator diagrams (if the displacement distance of the polished rod is greater than the output distance of the motor, it is determined that the data is abnormal), and then corrects the remaining indicator diagrams. The correction logic is to supplement the part of the indicator diagram corresponding to the missing stroke; A feature extraction unit, obtains all the corrected indicator diagrams, and then adopts the method of multi-sampling point random sampling to extract the feature data in different indicator diagrams. Based on the extracted feature data, standardize the indicator diagrams of different fault types to obtain a labeled data set; Data processing unit. To further improve the accuracy of the monitoring system in judging the working conditions of the pumping unit, based on the dynamometer card, other data affecting the working conditions of the pumping unit, namely vibration frequency data, is introduced. Under different fault conditions, the vibration frequency data of the pumping unit also varies. This module analyzes and processes the collected vibration frequency data to obtain corresponding processing results. Fault monitoring unit. A fault monitoring model is constructed. Using the labeled data set and the corresponding processing results as the training set together, the fault monitoring model is trained. Then, based on the trained fault monitoring model, the digital pumping unit is monitored in real time. Both its monitoring efficiency and accuracy are significantly improved compared with the existing monitoring system.

[0017] The beneficial effects are as follows: The present invention discloses a digital pumping unit monitoring system. The initial position of the suspension point is initialized by setting a suspension point initialization unit; the position of the suspension point is monitored by a suspension point position monitoring unit. Its monitoring logic is to compare the motor output distance and the suspension point displacement distance to obtain the monitoring result; then the dynamometer card unit constructs the dynamometer card of the suspension point, and based on the monitoring result, data cleaning and correction are performed on the constructed dynamometer card. Its correction logic is to supplement the part of the dynamometer card with missing strokes; the feature extraction unit is used to extract features from the corrected dynamometer card and perform data annotation to obtain a labeled data set; based on the data processing unit, the vibration frequency data under different fault types is processed to obtain processing results; finally, the fault monitoring unit trains the fault monitoring model together using the labeled data set and the processing results, and monitors the pumping unit in real time based on the trained fault monitoring model. The present invention initializes the suspension point and monitors the position of the suspension point to achieve the effect of data cleaning and correction of the dynamometer card, then extracts features from the corrected dynamometer card and performs data annotation to obtain a labeled data set, and at the same time analyzes and processes the vibration frequency data under different fault types of the pumping unit to obtain processing results. By adopting the learning method of a neural network model and using the labeled data set of the dynamometer card and the processing results of the vibration frequency data for training, the monitoring efficiency and accuracy of the monitoring system are greatly improved. Brief Description of the Drawings

[0018] To more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only the embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained according to the provided drawings without creative efforts.

[0019] Figure 1 It is a schematic diagram of the composition of a digital pumping unit monitoring system provided by an embodiment of the present application. Figure 2 It is a schematic diagram of the output distance of the motor and the displacement distance of the suspension point in the same coordinate system in the embodiment of the present application; Figure 3 It is a schematic diagram of correcting the dynamometer card in the embodiment of the present application. Detailed implementation manners

[0020] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

[0021] In the present application, relative terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the phrase "including an..." does not exclude the existence of additional identical elements in the process, method, article or device including the element.

[0022] To better understand the present application, the following is a corresponding explanation of the technical terms involved in the present application: Dynamometer card: A closed curve representing the corresponding relationship between the load on the polished rod string at the suspension point and the displacement of the suspension point. By analyzing the shape, area and other characteristics of the dynamometer card at the suspension point, the operating conditions of the pumping unit can be understood, and whether it is working properly can be judged, such as whether there are faults such as stuck fixed valve, valve wax deposition, tubing and sucker rod wax deposition, etc.

[0023] Embodiment As can be seen from the above background technology, the digital pumping unit system mainly consists of three parts: a sucker rod pump, a pumping unit and a sucker rod. The main hardware structure of the pumping unit includes a walking beam, a balance beam, a connecting rod, a crank mechanism, a reducer, a power device and auxiliary devices; the sucker rod is a rod device for the oil well, connected to the polished rod at the top and the sucker rod pump at the bottom, playing a role in connecting the ground and the downhole; the sucker rod pump consists of four parts: a pump barrel, a plunger, a suction valve and a traveling valve; a dynamometer card instrument (load sensor, displacement sensor), a control box, etc. are responsible for collecting production parameters. It can monitor the digital pumping unit efficiently and accurately in real time, not only ensuring the efficient production of the oil well, reducing risks, but also reducing operation and maintenance costs and improving management efficiency.

[0024] In the prior art, the monitoring of digital pumping units mostly requires staff to take turns on-site for 24 hours. Then, load displacement sensors are used to collect the load data and displacement data of the polished rod at the wellhead to form an indicator diagram at the backend (the abscissa is the polished rod displacement, and the ordinate is the polished rod load). Professional personnel analyze the indicator diagram to determine the type of fault that occurs in the pumping unit. However, the above method has the following problems: Since the pumping unit operates continuously in a harsh environment full of dust, rain, and saline-alkali air all year round, problems such as uneven stress and eccentric wear in the bearings cause problems with the support bearings, crossbeam bearings, and crankpin bearings. In addition, there are clearance fits in the articulated parts such as the crank, connecting rod, and walking beam. After long-term operation in a harsh environment, the wear intensifies. During the stroke, displacement loss will occur due to the "idle stroke", resulting in the polished rod not being able to return to the starting point, that is, there is an error between the displacement of a single up-and-down stroke and the displacement of an ideal single up-and-down stroke. Moreover, this error will continuously increase over time, finally resulting in the indicator diagram formed by collecting displacement load data not being able to accurately reflect the fault of the pumping unit, reducing the monitoring accuracy of the pumping unit monitoring system; and currently, the method of manual monitoring and professional personnel analyzing the indicator diagram at the backend has a low monitoring efficiency and cannot detect and handle faults in a timely manner.

[0025] Therefore, the present application provides a digital pumping unit monitoring system, as Figure 1 shown, including: Polished rod initialization unit: Obtain the initial position of the polished rod and the change value of the initial position at adjacent moments, and determine whether the change value exceeds a first threshold. If so, obtain the real-time position of the polished rod at the adjacent moment as the initial position. If not, do not perform any action; Polished rod position monitoring unit: Based on the initial position of the polished rod, obtain the time for the crank to rotate one week to divide the time interval. Within a single complete time interval, respectively obtain the output distance of the motor and the displacement distance of the polished rod, and plot the output distance of the motor and the displacement distance of the polished rod in the same coordinate system, analyze and monitor the data in the coordinate system, and obtain the monitoring result; Indicator diagram unit: Within a single complete time interval, construct an indicator diagram of the polished rod, and based on the monitoring result, determine whether to perform data cleaning on the indicator diagram of the polished rod. If so, perform cleaning on the indicator diagram of the polished rod; if not, correct the indicator diagram of the polished rod; Feature extraction unit: Obtain all the corrected indicator diagrams and place them in a first preset data set. Use the method of multi-sampling point random sampling for all the indicator diagrams in the first preset data set to obtain the feature data of different indicator diagrams; Based on the feature data of all the indicator diagrams in the first preset data set, label the indicator diagrams of different fault types to obtain a labeled data set; Data processing unit: Collect the vibration frequency data of the digital pumping unit within multiple complete time intervals at multiple time scales, and process them using a preset method to obtain multiple processing results; Fault monitoring unit: Construct a fault monitoring model, use all the labeled data sets as the first influencing factor and all the processing results as the second influencing factor to jointly train the fault monitoring model, and based on the trained fault monitoring model, conduct real-time monitoring on the digital pumping unit.

[0026] Among them, the polished rod initialization unit is used to initialize the initial position of the polished rod. During the actual operation of the pumping unit, the initial point of the polished rod may also change. If the initial point of the polished rod is not corrected, there will be a deviation between the displacement distance of the polished rod measured by the displacement sensor and the actual displacement distance of the polished rod, which will directly affect the generation of the indicator diagram. Therefore, by obtaining the initial position of the polished rod and the change value of the initial position at adjacent moments, the adjacent moments can be adjacent seconds, or can be adjusted according to the actual situation. If the change value of the polished rod initial position at adjacent moments exceeds the set first threshold, if the original initial position is still used as the reference to draw the indicator diagram at this time, there will be a large error. Therefore, in this embodiment, the real-time position of the polished rod is obtained and used as the new initial position; if the change value of the polished rod initial position at adjacent moments does not exceed the set first threshold, no action is taken.

[0027] Among them, the polished rod position monitoring unit is used to monitor the position of the polished rod. Based on the initial position of the polished rod, the time for the crank to make one revolution is used to divide the time interval. Specifically, the starting position of the crank is obtained, and the moment when the crank starts to move is recorded as moment t0, and the moment when the crank returns to the starting position is recorded as t1. The time interval between moment t0 and t1 constitutes a complete time interval.

[0028] Among them, when the crank makes one revolution, that is, the polished rod completes a complete stroke of the up and down stroke. Within a single complete time interval, the output distance of the motor and the displacement distance of the polished rod are respectively obtained. Regarding the output distance of the motor, by obtaining the rotational speed of the motor and the circumference of the motor main shaft, that is, within a single complete time interval, the rotational distance of the motor main shaft is obtained as the output distance of the motor; regarding the displacement distance of the polished rod, it is obtained through a displacement sensor.

[0029] Among them, the output distance of the motor and the displacement distance of the polished rod are plotted in the same coordinate system, and the data in the coordinate system is analyzed and monitored to obtain the monitoring result; Please refer to Figure 2 , Figure 2It is a schematic diagram of the output distance of the motor and the displacement distance of the suspension point in the same coordinate system in the embodiment of the present application. The time interval is used as the abscissa, and the ordinate is dimensionless. It is only used to distinguish the starting position A of the motor and the starting position B of the suspension point. Then, the output distance of the motor and the displacement distance of the suspension point are drawn proportionally within each time interval. For example, within the time interval t0 - t1, it can be directly seen that the output distance of the motor is equal to the displacement distance of the suspension point; within the time interval t1 - t2, the output distance of the motor is less than the displacement distance of the suspension point; within the time intervals t2 - t3 and t3 - t4, the output distance of the motor is greater than the displacement distance of the suspension point. It should be noted that this coordinate system is only for more intuitively showing the difference between the output distance of the motor and the displacement distance of the suspension point within different time intervals, so it does not need to strictly follow the relevant standards of the coordinate system. In addition, Figure 2 It only lists the possible situations of the output distance of the motor and the displacement distance of the suspension point within a single complete time interval, and does not represent the actual data.

[0030] Among them, the dynamometer diagram unit is used to construct a dynamometer diagram about the suspension point within a single complete time interval, such as Figure 3 the closed figure formed by the vertices D1, D2, D3, and D4 as shown. Based on the monitoring results, it is judged whether to perform data cleaning on the dynamometer diagram of the suspension point. When the displacement distance of the suspension point exceeds the output distance of the motor, it is determined that the displacement data collected by the displacement sensor is abnormal at this time. For example, Figure 2 within the time interval t1 - t2 in, at this time, it is necessary to perform data cleaning on the dynamometer diagram corresponding to the time interval t1 - t2; if the displacement distance of the suspension point does not exceed the output distance of the motor, then calculate the difference between the output distance of the motor and the displacement distance of the suspension point. If this difference is greater than the second threshold, it is necessary to correct the dynamometer diagram of the suspension point within the corresponding time interval. For example, it is determined that Figure 2 within the time intervals t2 - t3 and t3 - t4 in, the difference exceeds the second threshold, then it is necessary to perform correction processing on the dynamometer diagrams within these two time intervals; if this difference is not greater than the second threshold, no action is taken. For example, Figure 2 the dynamometer diagram within the time interval t0 - t1 in does not need to be corrected.

[0031] Among them, when correcting the dynamometer diagram of the suspension point, such as for Figure 3The indicator diagram of the mid-suspension point is corrected. First, obtain the upper right vertex and the lower right vertex in this indicator diagram. It should be noted that in the indicator diagram, the coordinate feature of the upper right vertex is the point with the largest displacement and the largest load. Its physical meaning is that after the end of the upstroke, the suspension point bears the largest load. The coordinate feature of the lower right vertex is the point with the largest displacement and the smallest load. Its physical meaning is the starting point of the downstroke, where the load of the suspension point drops suddenly. The determination of the four vertices (upper left, lower left, upper right, lower right) in the indicator diagram can be based on the identification of feature points with physical meanings, or data processing and algorithm identification can be used. This embodiment will not elaborate specifically. Obtain the first difference between the output distance of the motor and the displacement distance of the suspension point within the corresponding time interval of this indicator diagram, and obtain the continuous line feature between point D3 and point D4 as the first line feature, the continuous line feature between point D3 and point D2 as the second line feature, and the continuous line feature between point D4 and point D1 as the third line feature. Based on the first difference and the second line feature, determine the upper right increment vertex D3'. Based on the first difference and the third line feature, determine the lower right increment vertex D4'. Then, based on the second line feature, complete the connection between D3 and D3'. Based on the third line feature, complete the connection between D4 and D4'. Based on the first line feature, complete the connection between D3' and D4'. Thus, a closed figure composed of vertices D1, D2, D3' and D4' is obtained as the corrected indicator diagram.

[0032] Among them, the feature extraction unit obtains the corrected indicator diagram and, by using the method of multi-sampling point random sampling, obtains the feature data of different indicator diagrams. Based on the feature data of the indicator diagram, the indicator diagrams of different fault types are labeled to obtain a labeled data set.

[0033] Among them, the data processing unit is used to collect the vibration frequency data of the digital pumping unit in multiple complete time intervals at multiple time scales and process them using a preset method. Specifically: construct multiple coordinate curves of the vibration frequency data changing with time; then obtain any fault type denoted as the first fault type, and obtain multiple coordinate curves under the first fault type (such as 5, namely C1, C2, C3, C4, and C5), and store them in the second preset data set. Segment each coordinate curve in the second preset data set by time interval to obtain M1 time intervals, and collect the slopes of N1 data points in each time interval. Calculate the average slopes of the corresponding data points in the corresponding time intervals of the C1, C2, C3, C4, and C5 coordinate curves. It should be noted that the data points collected in each coordinate curve correspond one by one, and then store the average slopes of each data point through constructing a database. When there is a single coordinate curve denoted as the first coordinate curve, and the slopes of N2 data points in a single time interval of the first coordinate curve meet the preset conditions, that is, the difference between the slopes of these N2 data points and the average slopes of the corresponding data points of the first fault type in the database is less than the set value, and the ratio of N2 / N1 exceeds the third threshold (such as N1 is 10, N2 is 8, and the third threshold is 0.7), it is determined that the single time interval matches the corresponding time interval of the first fault type; then calculate how many matching time intervals exist in the first coordinate curve, such as M2. If it is determined that the ratio of M2 / M1 exceeds the fourth threshold, it is determined that the fault type corresponding to the first coordinate curve is the same as the first fault type. The processing logic of this embodiment is to perform multiple differential processing on the vibration frequency data, and then adopt the method of secondary matching to extract the characteristics of the vibration frequency data under different fault types. It should be noted that the above data is only for explanatory purposes and can be adjusted according to the actual situation, and this embodiment does not make specific limitations.

[0034] Among them, to build a fault monitoring model, the neural network learning method is adopted. The labeled data set corresponding to the dynamogram and the processing results corresponding to the vibration frequency data are jointly used as the training set to train the fault monitoring model. Its training architecture can be optimized by matrix operation or cross structure. That is, the fault type of the pumping unit is output through the labeled data set inside the fault monitoring model, and at the same time, the fault type of the pumping unit is output by using all the processing results. Matrix E is used to store the fault type output through the labeled data set, and matrix F is used to store the fault type output through the processing results. When the fault type output in matrix E is consistent with the fault type output in matrix F, the corresponding fault type is finally output through the fault monitoring model; when the fault type output in matrix E is inconsistent with the fault type output in matrix F, by analyzing the performance of matrix E and matrix F corresponding to different fault types, corresponding coefficients are set. For example, for fault type G, the prediction coefficient of matrix E is 0.9, and the prediction coefficient of matrix F is 0.8. For fault type H, the prediction coefficient of matrix E is 0.95, and the prediction coefficient of matrix F is 0.85. When the fault type of the pumping unit output through the labeled data set inside the fault monitoring model is fault type G and stored in matrix E, and the fault type of the pumping unit output through all the processing results is fault type H and stored in matrix F, then the fault type G is finally output through the fault monitoring mode. Simply put, two influencing factors are used to jointly train the fault monitoring model. When the output results of the two are consistent, the corresponding fault type is output; when the output results of the two are inconsistent, the monitoring result with better performance for the specific fault type is output. It should be noted that this embodiment only provides a processing method for the inconsistent output results of the two. When the output results of the two are inconsistent, of course, the manual verification method can also be adopted to determine the final fault type, and this embodiment does not make specific limitations.

[0035] Furthermore, in addition to adding vibration frequency data to the fault monitoring model, other data that can affect the working conditions of the pumping unit, such as noise data and temperature data, can also be added. Because such data has the same or similar characteristic data under the same working conditions, by analyzing and processing the characteristic data of such data under different working conditions and using the neural network model for learning, the effect of inferring the working conditions of the pumping unit can be achieved.

[0036] In some embodiments, obtaining the time for one revolution of the crank to divide the time interval includes: Obtaining the starting position of the crank, and recording the moment when the crank starts to move as moment t0, and the moment when the crank returns to the starting position as moment t1; The time interval is formed between moment t0 and moment t1.

[0037] In some embodiments, obtaining the output distance of the motor and the displacement distance of the suspension point within a single complete time interval includes: Within a single complete time interval, obtain the displacement distance of the suspension point through a displacement sensor, and obtain the output distance of the motor by obtaining the rotational speed of the motor and the circumference of the motor main shaft.

[0038] In some embodiments, plotting the output distance of the motor and the displacement distance of the suspension point in the same coordinate system includes: Use time as the horizontal axis, and divide the horizontal axis into multiple time intervals; The vertical axis is dimensionless, and points A and B are determined on the vertical axis; Obtain the output distance of the motor under all time intervals, and starting from point A, plot the output distance of the motor proportionally within the corresponding time interval; Obtain the displacement distance of the suspension point under all time intervals, and starting from point B, plot the displacement distance of the suspension point proportionally within the corresponding time interval.

[0039] In some embodiments, analyzing and monitoring the data in the coordinate system to obtain a monitoring result includes: In the coordinate system, obtain the output distance of the motor within any time interval, denoted as L1, and the displacement distance of the suspension point, denoted as L2; If L2 is greater than L1, perform data cleaning on the indicator diagram of the suspension point within the corresponding time interval; If L2 is less than or equal to L1, calculate the difference between L1 and L2, and determine whether the difference is greater than a second threshold. If so, correct the indicator diagram of the suspension point within the corresponding time interval. If not, do nothing.

[0040] In some embodiments, correcting the indicator diagram of the suspension point based on the monitoring result includes: Obtain any indicator diagram of the suspension point as a first indicator diagram, and obtain the difference between the output distance of the motor and the displacement distance of the suspension point within the time interval corresponding to the first indicator diagram, denoted as a first difference; Obtain the continuous longitudinal line feature between the upper right vertex and the lower right vertex in the first indicator diagram as a first line feature, and obtain the continuous horizontal line feature extending from the upper right vertex to the vertical axis as a second line feature, and obtain the continuous horizontal line feature extending from the lower right vertex to the vertical axis as a third line feature; Based on the first difference, the second line feature, and the third line feature, determine the upper right increment vertex and the lower right increment vertex in the first indicator diagram; Based on the second line feature, complete the connection line between the upper right vertex and the upper right incremental vertex; based on the third line feature, complete the connection line between the lower right vertex and the lower right incremental vertex; based on the first line feature, complete the connection line between the upper right incremental vertex and the lower right incremental vertex; Complete the correction of the first indicator diagram.

[0041] In some embodiments, the preset method includes: Based on the vibration frequency data collected within multiple complete time intervals under multiple time scales, construct multiple coordinate curves of the vibration frequency data changing with time; Obtain any fault type denoted as the first fault type, and obtain multiple coordinate curves under the first fault type, and store them in the second preset data set; Segment each coordinate curve in the second preset data set with the time interval to obtain M1 time intervals, and collect the slopes of N1 data points in each time interval to obtain the collection result; Based on the collection result, calculate the average slope of the corresponding data points in the corresponding time intervals of multiple coordinate curves in the second preset data set; Establish a database and store the average slopes of the corresponding data points in the corresponding time intervals of multiple coordinate curves in the second preset data set; When there is a single coordinate curve denoted as the first coordinate curve, and the slopes of N2 data points in the single time interval of the first coordinate curve meet the preset conditions, it is determined that the time interval corresponding to the first coordinate curve matches the time interval corresponding to the first fault type; When there are M2 matching time intervals in the first coordinate curve, and the ratio of M2 / M1 exceeds the fourth threshold, it is determined that the first coordinate curve matches the coordinate curve corresponding to the first fault type.

[0042] In some embodiments, the preset conditions include: The difference between the slopes of the N2 data points and the average slope of the corresponding data points in the database is less than the set value, and the ratio of N2 / N1 exceeds the third threshold.

[0043] In some embodiments, based on the trained fault monitoring model, the real-time monitoring of the digital pumping unit includes: Obtain the corrected indicator diagram of the digital pumping unit denoted as the second indicator diagram, and the processing result of the corresponding vibration frequency data within the preset time period denoted as the first processing result; Input the second indicator diagram and the first processing result into the trained fault monitoring model, and output the fault type of the digital pumping unit.

[0044] In some embodiments, constructing an indicator diagram of the polished rod within a single complete time interval includes: Within a single complete time interval, displacement data of the polished rod is collected by a displacement sensor, and load data of the polished rod is collected by a load sensor; an indicator diagram of the polished rod is constructed with the displacement data of the polished rod as the abscissa and the load data of the polished rod as the ordinate.

[0045] Those skilled in the art can further realize that the units and algorithm steps of each example described in combination with the embodiments disclosed in this article can be implemented by electronic hardware, computing software, or a combination of the two. To clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described according to functions in the above description. Whether these functions are executed by hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0046] Although the preferred embodiments of the present invention have been described, those skilled in the art can make additional changes and modifications to these embodiments once they know the basic creative concept. Therefore, the appended claims are intended to be construed to include the preferred embodiments and all changes and modifications falling within the scope of the present invention.

[0047] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention also intends to include these modifications and variations.

Claims

1. A digital pumping unit monitoring system, characterized in that, Including: Suspension point initialization unit: Obtain the initial position of the suspension point and the change value of the initial position at adjacent moments, and determine whether the change value exceeds the first threshold. If so, obtain the real-time position of the suspension point at the adjacent moment as the initial position; if not, do nothing. Suspension point position monitoring unit: Based on the initial position of the suspension point, obtain the time for the crank to rotate one week to divide the time interval. Within a single complete time interval, respectively obtain the output distance of the motor and the displacement distance of the suspension point, and plot the output distance of the motor and the displacement distance of the suspension point in the same coordinate system, analyze and monitor the data in the coordinate system to obtain the monitoring result. Indicator diagram unit: Within a single complete time interval, construct an indicator diagram of the suspension point, and based on the monitoring result, determine whether to perform data cleaning on the indicator diagram of the suspension point. If so, clean the indicator diagram of the suspension point; if not, correct the indicator diagram of the suspension point. Feature extraction unit: Obtain all the corrected indicator diagrams and put them into the first preset data set. Adopt the method of multi-sampling point random sampling for all the indicator diagrams in the first preset data set to obtain the feature data of different indicator diagrams; based on the feature data of all the indicator diagrams in the first preset data set, label the indicator diagrams of different fault types to obtain the labeled data set. Data processing unit: Collect the vibration frequency data of the digital pumping unit in multiple complete time intervals at multiple time scales, and process them using a preset method to obtain multiple processing results. Fault monitoring unit: Construct a fault monitoring model, use all the labeled data sets as the first influencing factor and all the processing results as the second influencing factor to jointly train the fault monitoring model, and based on the trained fault monitoring model, perform real-time monitoring on the digital pumping unit.

2. The digital pumping unit monitoring system according to claim 1, wherein, Obtaining the time for the crank to rotate one week to divide the time interval includes: Obtain the starting position of the crank, record the moment when the crank starts to move as moment t0, and the moment when the crank returns to the starting position as moment t1. The time interval is formed between moment t0 and moment t1.

3. The digital pumping unit monitoring system according to claim 1, characterized in that, Within a single complete time interval, respectively obtaining the output distance of the motor and the displacement distance of the suspension point includes: Within a single complete time interval, obtain the displacement distance of the suspension point through a displacement sensor, and obtain the output distance of the motor by obtaining the rotation speed of the motor and the circumference of the motor main shaft.

4. A digital pumping unit monitoring system according to claim 1, characterized in that, Plotting the output distance of the motor and the displacement distance of the suspension point in the same coordinate system includes: Use time as the horizontal axis and divide the horizontal axis into multiple time intervals. The vertical axis is dimensionless, and determine point A and point B on the vertical axis. Obtain the output distance of the motor under all time intervals, start from point A, and plot the output distance of the motor proportionally within the corresponding time interval. Obtain the displacement distance of the suspension point under all time intervals, start from point B, and plot the displacement distance of the suspension point proportionally within the corresponding time interval.

5. The digital pumping unit monitoring system according to claim 4, wherein Analyze and monitor the data in the coordinate system to obtain the monitoring result includes: In the coordinate system, obtain the output distance of the motor within any time interval, denoted as L1, and the displacement distance of the suspension point, denoted as L2; If L2 is greater than L1, perform data cleaning on the dynamogram of the suspension point within the corresponding time interval; If L2 is less than or equal to L1, calculate the difference between L1 and L2, and determine whether the difference is greater than a second threshold. If so, correct the dynamogram of the suspension point within the corresponding time interval. If not, do nothing.

6. The digital pumping unit monitoring system according to claim 5, characterized in that Based on the monitoring result, correcting the dynamogram of the suspension point includes: Obtain any dynamogram of the suspension point as a first dynamogram, and obtain the difference between the output distance of the motor and the displacement distance of the suspension point within the time interval corresponding to the first dynamogram, denoted as a first difference; Obtain the continuous longitudinal line feature between the upper right vertex and the lower right vertex in the first dynamogram, denoted as a first line feature, and obtain the continuous horizontal line feature extending from the upper right vertex to the vertical axis, denoted as a second line feature, and obtain the continuous horizontal line feature extending from the lower right vertex to the vertical axis, denoted as a third line feature; Based on the first difference, the second line feature, and the third line feature, determine the upper right increment vertex and the lower right increment vertex in the first dynamogram; Based on the second line feature, complete the connection between the upper right vertex and the upper right increment vertex; based on the third line feature, complete the connection between the lower right vertex and the lower right increment vertex; based on the first line feature, complete the connection between the upper right increment vertex and the lower right increment vertex; Complete the correction of the first dynamogram.

7. A digital pumping unit monitoring system according to claim 1, characterized in that The preset method includes: Based on the vibration frequency data collected within multiple complete time intervals under multiple time scales, construct multiple coordinate curves of the vibration frequency data changing with time; Obtain any fault type, denoted as a first fault type, and obtain multiple coordinate curves under the first fault type, and store them in a second preset data set; Segment each coordinate curve in the second preset data set with the time interval to obtain M1 time intervals, and collect the slopes of N1 data points within each time interval to obtain a collection result; Based on the collection result, calculate the average slope of the corresponding data points in the corresponding time intervals of multiple coordinate curves in the second preset data set; Establish a database and store the average slopes of the corresponding data points in the corresponding time intervals of multiple coordinate curves in the second preset data set; When there is a single coordinate curve, denoted as a first coordinate curve, and the slopes of N2 data points within a single time interval of the first coordinate curve meet the preset conditions, it is determined that the time interval corresponding to the first coordinate curve matches the time interval corresponding to the first fault type; When there are M2 matching time intervals in the first coordinate curve, and the ratio of M2 / M1 exceeds a fourth threshold, it is determined that the first coordinate curve matches the coordinate curve corresponding to the first fault type.

8. A digital pumping unit monitoring system according to claim 7, characterized in that, The preset conditions include: The difference between the slopes of the N2 data points and the average slopes of the corresponding data points in the database is less than a set value, and the ratio of N2 / N1 exceeds a third threshold.

9. The digital pumping unit monitoring system according to claim 1, wherein, Based on the trained fault monitoring model, the real-time monitoring of the digital pumping unit includes: Obtaining the corrected indicator diagram of the digital pumping unit, denoted as the second indicator diagram, and the processing result of the corresponding vibration frequency data within a preset time period, denoted as the first processing result; Inputting the second indicator diagram and the first processing result into the trained fault monitoring model, and outputting the fault type of the digital pumping unit.

10. The digital pumping unit monitoring system according to claim 1, characterized in that, Within a single complete time interval, constructing an indicator diagram regarding the polished rod includes: Within a single complete time interval, collecting the displacement data of the polished rod through a displacement sensor and collecting the load data of the polished rod through a load sensor; using the displacement data of the polished rod as the abscissa and the load data of the polished rod as the ordinate, constructing an indicator diagram regarding the polished rod.

Images