Intelligent diagnosis method of underground working conditions based on big data analysis

Abstract

The present invention relates to an intelligent diagnosis method for underground working conditions based on big data analysis, which includes the following steps in sequence: A. Establishing a background analysis model for underground faults; B. Obtaining real-time vibration acceleration data of the oil pumping unit's polished rod; C. Sending the background analysis model and the real-time vibration acceleration data to a real-time identification module for comparison and diagnosis; D. Outputting the diagnosis result. Step A specifically includes: A1. Batch collection of effective vibration acceleration data of the oil pumping unit's polished rod; A2. Data transmission to the background and preprocessing; A3. Marking normal and fault vibration acceleration data respectively; A4. Extracting the root mean square energy of various types of vibration acceleration data; A5. Drawing a graph of the root mean square energy changing with time; A6. The graph is stored in a sample library and trained; A7. Outputting the parameters required by the real-time identification module. This method uses technologies such as artificial intelligence, the Internet of Things, signal processing, and big data to accurately judge underground working conditions and diagnose specific faults occurring underground.

Description

Technical Field

[0001] The present invention relates to an oil pumping unit in the field of oil extraction, and in particular to an intelligent diagnosis method for underground working conditions based on big data analysis, belonging to the technical field of oilfield automation. Background Art

[0002] The sucker rod is connected to the pumping unit at the top and the oil pump at the bottom. When working in the narrow wellbore filled with well fluid, it is subjected to tension, compression, bending, torsion and hydraulic loads, and is in a very complex state of force, deformation and motion. It is a complex dynamic system. When the pumping unit is working, the crank mechanism drives the tail end of the walking beam to swing up and down through two connecting rods and a crossbeam, and the donkey head at the head end of the walking beam also moves up and down. The donkey head drives the plunger of the deep well pump to move up and down through the rope braid, the bare rod and the sucker rod, thereby continuously pumping the crude oil in the well out of the wellbore.

[0003] The sucker rod string has different loads in different working conditions and different positions, mainly axial force, pressure, friction, inertia load and vibration load. The vibration load is generated by the movement of the sucker rod string and the tubing string. The sucker rod is greatly affected by vibration. The longitudinal vibration of the sucker rod and the tubing will affect the relative movement of the end of the sucker rod to the tubing string, that is, it will affect the distance between the floating valve and the fixed valve of the pump, and directly affect the effective stroke of the plunger.

[0004] For example, during the operation of the pumping unit, the tubing and the sucker rod continuously undergo elastic deformation under the action of their own gravity and lifting force, causing eccentric wear of the sucker rod. The problem of eccentric wear of the sucker rod has always been a major problem that plagues oil field production. The causes of eccentric wear are very complicated, such as wax deposition in the oil well, sand production, well deviation, casing deformation, and pipe and rod corrosion. The eccentric wear of the sucker rod in the well is very difficult to observe on the ground. The eccentric wear of the sucker rod causes the tubing to be scrapped prematurely, which has a great impact on production and greatly increases costs. If the downhole working conditions can be accurately diagnosed on the ground, it will be of great significance to promote the safe production of oil fields.

[0005] The Chinese utility model patent with publication number CN 203257406U discloses a downhole drill string fault diagnosis system, wherein the drill string is supported by a drilling platform and a derrick, the drilling platform includes a drilling platform at the bottom and a drilling rig second-floor platform at the middle, the drill string includes a square drill rod and a drill pipe, one end of the square drill rod passes through a casing and is connected to the drill pipe, and the other end is connected to a faucet, the faucet is connected to a water hose, and risers are distributed on the derrick, and are equipped with a sound pressure sensor, a three-dimensional acceleration sensor at the wellhead, a three-dimensional acceleration sensor at the top of the drill string, a pressure sensor, a multi-channel signal acquisition and transmission device, A multi-channel signal receiving device and a computer system, a sound pressure sensor is fixed on the drilling platform, a three-axis acceleration sensor at the wellhead is fixed on the casing, a pressure sensor is arranged on the riser, and a three-axis acceleration sensor at the top of the drill string is arranged on the faucet. A multi-channel signal acquisition and sending device is arranged on the second-floor platform of the drilling rig. The sound pressure sensor, the three-axis acceleration sensor at the wellhead, the three-axis acceleration sensor at the top of the drill string and the pressure sensor are all connected to the multi-channel signal acquisition and sending device, and the multi-channel signal acquisition and sending device is wirelessly connected to the multi-channel signal receiving device.

[0006] Although various sensors are installed to collect signals, the signals can only be sent to the background for technicians to determine the fault. There is no autonomous diagnosis function, and big data cannot be used for training and analysis. The accuracy rate is low and it is too dependent on humans. Summary of the invention

[0007] The purpose of the present invention is to overcome the problems existing in the prior art and provide an intelligent diagnosis method for underground working conditions based on big data analysis. By utilizing technologies such as artificial intelligence, the Internet of Things, signal processing and big data, it can accurately judge the underground working conditions and diagnose underground faults, thus providing technical support for intelligent diagnosis and early warning technology of underground faults.

[0008] In order to solve the above technical problems, the intelligent diagnosis method of underground working conditions based on big data analysis of the present invention comprises the following steps in sequence:

[0009] A. Establish a background analysis model for underground faults;

[0010] B. Obtain real-time vibration acceleration data of the polished rod of the pumping unit;

[0011] C. Send the background analysis model and real-time vibration acceleration data to the real-time identification module for comparison and diagnosis;

[0012] D. Output the diagnostic results.

[0013] As an improvement of the present invention, step A specifically includes the following sub-steps:

[0014] A1. Install vibration acceleration sensors on the top of multiple pumping unit polished rods to collect effective vibration acceleration data of the pumping unit polished rods in batches, including normal vibration acceleration data and various fault vibration acceleration data;

[0015] A2. Transfer the vibration acceleration data to the background and pre-process it;

[0016] A3. Mark the normal vibration acceleration data and fault vibration acceleration data respectively;

[0017] A4. Extract the root mean square energy of various vibration acceleration data;

[0018] A5. Draw a graph of the RMS energy versus time.

[0019] A6. The graph is stored in the sample library and trained;

[0020] A7. Output the parameters required by the real-time recognition module.

[0021] As a further improvement of the present invention, step B specifically includes the following sub-steps:

[0022] B1. Install a vibration acceleration sensor on the upper part of the polished rod of the oil pump to monitor the vibration acceleration data of the polished rod of the oil pump;

[0023] B2. Transfer the vibration acceleration data to the background and perform preprocessing;

[0024] B3. Extract the RMS energy of real-time vibration acceleration data.

[0025] As a further improvement of the present invention, the diagnostic result outputted in step D is judged by an expert. If the expert judges that the diagnosis is accurate, a targeted maintenance plan is determined in the next step. If the expert judges that the diagnosis is inaccurate, the fault sample is re-labeled, and then the process returns to step A4.

[0026] As a further improvement of the present invention, step A4 specifically includes the following sub-steps:

[0027] A4.1. Frame the vibration acceleration signal;

[0028] A4.2. Add a Hamming window to each frame;

[0029] A4.3, square the amplitude of each frame in the window;

[0030] A4.4. Sum the squared amplitude values ​​of each frame;

[0031] A4.5. Divide the sum by the frame length to get the average value;

[0032] A4.6. Take the square root of the average value obtained to obtain the root mean square energy in the window.

[0033] As a further improvement of the present invention, in step A6, the sample library is trained by deep convolutional neural network technology.

[0034] As a further improvement of the present invention, the faults described in step A1 include eccentric wear of the sucker rod, upward collision of the sucker rod, insufficient liquid supply, imbalance, gas influence, upward resistance, leakage of the fixed valve and collision of the pump production.

[0035] As a further improvement of the present invention, the effective vibration acceleration data of the oil pumping unit in step A1 is collected by a data acquisition device, which includes a signal acquisition module, a signal processing module and a power supply module. The signal acquisition module includes a vibration acceleration sensor installed on the oil pumping unit's polished rod, and the signal output ends of the vibration acceleration sensor are respectively connected to the input ends of the preamplifier circuit, and the output end of the preamplifier circuit is connected to the input end of the bandpass filter; the signal processing module includes a HiSilicon Hi3518E chip, the output end of the bandpass filter is connected to the vibration acceleration signal interface of the Hi3518E chip, the clock port of the Hi3518E chip is connected to the clock circuit, the storage port of the Hi3518E chip is connected to the memory, the network signal output end of the Hi3518E chip is connected to the input end of the network transceiver, the output end of the network transceiver is connected to the input end of the network transformer, and the output end of the network transformer is connected to the RJ45 interface.

[0036] As a further improvement of the present invention, the preamplifier circuit includes a TL971 chip, the signal output end of the vibration acceleration sensor is connected to the positive electrode of the isolation capacitor C66, the negative electrode of the isolation capacitor C66 is connected to the third foot of the TL971 chip, the positive electrode of the isolation capacitor C66 is also connected to the VDD5V+ power supply through the bias resistor R63, and the negative electrode of the isolation capacitor C66 is also grounded through the bias resistor R64; the sixth foot of the TL971 chip is connected to the positive electrode of the isolation capacitor C68 through the current limiting resistor R68; the bandpass filter includes A ADA4898-2 chip, the negative electrode of the isolation capacitor C68 is connected to one end of the filter capacitor C69 through the current limiting resistor R69, the other end of the filter capacitor C69 is connected to the third pin of the ADA4898-2 chip, the current limiting resistor R69 and the filter capacitor C69 are grounded through the filter capacitor C70, and the filter capacitor C69 and the third pin of the ADA4898-2 chip are grounded through the resistor R70; the seventh pin of the ADA4898-2 chip is connected to the K15 port or the K16 port of the Hi3518E chip.

[0037] As a further improvement of the present invention, the network transceiver adopts the RTL8201F Ethernet transceiver, the network transformer adopts the H1102T network transformer, the Ethernet signal output end of the ADA4898-2 chip is connected to the signal input end of the RTL8201F Ethernet transceiver; the third foot of the RTL8201F Ethernet transceiver is connected to the first foot of the H1102T network transformer, the fourth foot of the RTL8201F Ethernet transceiver is connected to the third foot of the H1102T network transformer, and the RTL8201F Ethernet transceiver The 5th foot of the H1102T network transformer is connected to the 6th foot of the RTL8201F Ethernet transceiver, the 6th foot of the H1102T network transformer is connected to the 8th foot of the H1102T network transformer, the 16th foot of the H1102T network transformer is connected to the 1st foot of the RJ45 network connector, the 14th foot of the H1102T network transformer is connected to the 2nd foot of the RJ45 network connector, the 11th foot of the H1102T network transformer is connected to the 3rd foot of the RJ45 network connector, and the 9th foot of the H1102T network transformer is connected to the 6th foot of the RJ45 network connector.

[0038] Compared with the prior art, the present invention has achieved the following beneficial effects: 1. The vibration signals in the well will be transmitted to the upper light rod through the sucker rod. Installing the vibration acceleration sensor on the light rod will not interfere with other components and is easy to install or inspect. It can realize on-site collection, storage, uploading and identification of vibration data in the well, and timely and accurately diagnose various mechanical faults in the well. By using data to analyze, model and manage the fault generation process, it can predict and analyze hidden faults, give accurate diagnosis, and alarm the fault of the pumping unit according to the diagnosis result.

[0039] 2. In view of the many achievements made by convolutional neural networks (CNN) in the field of image training in recent years, Google has increased the network depth to more than one thousand layers. Deeper networks can extract more abstract features and learn more knowledge from increasing data. This method converts the underground vibration data into a map of root mean square energy, uses deep convolutional neural network technology to train the sample library, and uses the transfer learning ability of big data. Combined with the manual judgment of experts, the background analysis model can be continuously improved to improve the stability of the background model and reduce the false alarm rate. Clues to hidden faults can be mined from big data, and hidden faults can be solved before they develop into explicit faults through predictive analysis;

[0040] 3. It is easy to use reverse engineering to analyze and fine-tune the entire production process, and avoid problems in daily production;

[0041] 4. It can maintain the continuity of underground monitoring, solve the shortcomings of intermittent time in manual management and the inability of the naked eye to detect problems in time, and reduce the labor intensity of manual work.

[0042] 5. The obtained vibration acceleration signal is first pre-amplified and then filtered to avoid interference when the vibration impact changes, obtain the real vibration acceleration signal, and convert the vibration acceleration signal into an electrical signal.

[0043] 6. The signal acquisition module uses the HiSilicon Hi3518E chip to process the vibration acceleration signal, which can achieve dynamic noise reduction and automatic gain, ensuring high-quality acquisition effects at the back end. It can achieve real-time high-fidelity sampling and storage of vibrations generated by sucker rods, oil pumps, floating valves, fixed valves, etc. during operation by sampling the light rod. The signal acquisition module retrieves the vibration data from the well and transmits it to the remote server through the Ethernet interface circuit for background processing and monitoring.

[0044] 7. It adopts modular design, reasonable structure, strong versatility, can be uniformly applied to oil well sites, has low investment cost, can meet the "four modernizations" construction with "standardized design, modular construction, standardized procurement, and informationization improvement" as the main contents, and form an "informationization" management model, which is of great significance to promoting safe production in oil fields. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. The accompanying drawings are only provided for reference and explanation and are not intended to limit the present invention.

[0046] Figure 1 This is a flow chart of the intelligent diagnosis method of underground working conditions based on big data analysis of the present invention;

[0047] Figure 2 A flow chart of feature extraction of vibration acceleration signal in the present invention;

[0048] Figure 3 The root mean square energy spectrum of a well without faults in the first embodiment of the present invention;

[0049] Figure 4 This is the root mean square energy spectrum of the sucker rod collision in the second embodiment of the present invention;

[0050] Figure 5 This is the RMS energy spectrum produced by the impact pump in Example 3 of the present invention;

[0051] Figure 6 This is the root mean square energy spectrum of the gas influence in Example 4 of the present invention;

[0052] Figure 7 This is a root mean square energy spectrum of uplink obstruction in embodiment 5 of the present invention;

[0053] Figure 8 It is the principle diagram of the data acquisition device in the present invention;

[0054] Fig. 9 This is a power supply circuit diagram of the signal acquisition module in the present invention;

[0055] Fig.10 This is a preamplifier circuit diagram of the signal acquisition module in the present invention;

[0056] Fig.11 is a circuit diagram of a bandpass filter of a signal acquisition module in the present invention;

[0057] Fig.12 is a power supply circuit diagram of the signal processing module in the present invention;

[0058] Fig.13 The Hi3518E chip port of the signal processing module in the present invention Figure 1 ;

[0059] Fig.14 The Hi3518E chip port of the signal processing module in the present invention Figure 2 ;

[0060] Fig.15 The Hi3518E chip port of the signal processing module in the present invention Figure 3 ;

[0061] Fig.16 This is the external circuit diagram of the Hi3518E chip in the present invention;

[0062] Fig.17 is a circuit diagram of a network transceiver in the present invention;

[0063] Fig.18 It is the circuit diagram of the network transformer and RS485 interface in the present invention. DETAILED DESCRIPTION

[0064] In the following description of the present invention, the directions or positional relationships indicated by terms such as "up", "down", "front", "back", "left", "right", "inside" and "outside" are based on the directions or positional relationships shown in the accompanying drawings and are only for the convenience of describing the present invention and simplifying the description, and do not mean that the device must have a specific direction.

[0065] In order to make the technical means, creative features, objectives and effects achieved by the present invention easy to understand, the present invention is further described below with reference to specific diagrams.

[0066] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention.

[0067] like Figure 1 As shown, the intelligent diagnosis method of underground working conditions based on big data analysis of the present invention comprises the following steps in sequence:

[0068] A. Establish a background analysis model for underground faults;

[0069] B. Obtain real-time vibration acceleration data of the polished rod of the oil pump;

[0070] C. Send the background analysis model and real-time vibration acceleration data to the real-time identification module for comparison and diagnosis;

[0071] D. Output the diagnostic results.

[0072] Step A specifically includes the following sub-steps:

[0073] A1. Install vibration acceleration sensors on the top of multiple pumping rods to collect effective vibration acceleration data of the pumping rods in batches, including normal vibration acceleration data and various fault vibration acceleration data; faults include eccentric wear of the pumping rod, upward collision of the pumping rod, insufficient liquid supply, imbalance, gas influence, upward resistance, fixed valve leakage and pump collision production, etc. It is usually necessary to install vibration acceleration sensors on more than 1,000 oil wells and collect more than one million vibration acceleration data as training materials for the sample library;

[0074] A2. The vibration acceleration data is transferred to the backend and pre-processed by data splicing, cleaning, enhancement, sorting, etc. The above pre-processing adopts existing mature technologies, and other mature pre-processing technologies can be further adopted;

[0075] A3. Mark the normal vibration acceleration data and fault vibration acceleration data respectively;

[0076] A4. Extract the root mean square energy of various vibration acceleration data;

[0077] A5. Draw a graph of the RMS energy versus time.

[0078] A6. The graph is stored in the sample library, and the sample library is trained using deep convolutional neural network technology;

[0079] A7. Output the parameters required by the real-time recognition module.

[0080] The sucker rod string is connected together and is a slender rod, similar to a spring. The spring is very likely to vibrate longitudinally under external interference. From the indicator diagram of the ground test, it can be seen that the longitudinal vibration of the sucker rod string is the most frequent vibration form. There are three main factors causing the longitudinal vibration of the sucker rod string: ① The motor transmits the motion to the light rod through the four-bar linkage, and the light rod drives the sucker rod string to reciprocate up and down. The displacement is similar to the sine curve, the speed is similar to the cosine curve, and the acceleration is similar to the sine curve. Since the sucker rod string has acceleration during the movement, there will be inertial load. The spring will vibrate longitudinally under the action of inertial load. ② In the process of pumping crude oil, the sucker rod string lifts the liquid column in the upstroke and unloads the liquid column in the downstroke, which is similar to suddenly adding a weight to the lower end of the spring and suddenly removing a weight, causing the sucker rod string to vibrate longitudinally. ③ The resistance of the oil pump acting on the bottom end of the sucker rod string is related to the movement speed of the plunger and the opening state of the floating valve and the fixed valve. When the resistance changes, it will cause disturbance to the bottom end of the sucker rod string and cause longitudinal vibration of the sucker rod string.

[0081] There is also strong lateral vibration in the reciprocating motion of the sucker rod string. The lateral vibration of the sucker rod string is mainly caused by the following aspects: ① The donkey head suspension point drives the sucker rod string to reciprocate up and down, generating inertial loads and oil column loading and unloading during the motion, thereby generating nonlinear longitudinal vibration. Under certain conditions, the longitudinal vibration generated by the sucker rod string may cause dynamic instability, thereby stimulating lateral vibration. ② Wellbore eccentricity, manufacturing errors and connection errors cause geometric initial defects in the sucker rod string, making the sucker rod string very prone to instability and lateral displacement. ③ The mixed liquid generates a normal force in the process of flowing in the eccentric annulus. The direction of the normal force is perpendicular to the axis of the sucker rod, and its effect aggravates the lateral displacement of the sucker rod string. Lateral vibration is the direct cause of eccentric wear of the rod and pipe.

[0082] Sucker rod breakage, tripping, eccentric wear and spiral scratches on the plunger surface are several common forms of damage to the sucker rod. Among them, the tripping of the sucker rod and the spiral scratches on the plunger surface are directly related to the torsion factor of the sucker rod. The torsion caused by factors such as well inclination change, thread non-concentricity, and rope swing will cause torsional vibration. The reasons for the torsion of the sucker rod are analyzed as follows: There is an axial stress neutral point in the sucker rod during the downstroke. Due to the large overall flexibility of the sucker rod and the influence of factors such as the sucker rod's own weight and liquid resistance, it is easy to buckle below the neutral point. According to the mechanism of buckling of the elastic rod under axial load, when other conditions remain unchanged, different axial loads correspond to different bucklings. The axial load acting on the plunger in the downstroke first increases and then decreases, and the decrease only accounts for a small part of the entire downstroke. Therefore, the buckling is basically gradually intensified during the entire downstroke. Due to the equivalent mass eccentricity caused by the buckling of the sucker rod column, the sucker rod column undergoes torsional deformation under the action of mass eccentricity.

[0083] According to the above analysis, the sucker rod has three main vibrations during operation: longitudinal vibration, lateral vibration and torsional vibration. In step A1, three vibration acceleration sensors can be installed on the upper end of the light rod of the same pumping unit, or the most targeted vibration acceleration sensor can be installed on the light rod of different pumping units according to the fault type downhole, so as to collect targeted data more accurately and provide sample library training; a three-axis vibration acceleration sensor can also be directly installed.

[0084] like Figure 2 As shown, step A4 specifically includes the following sub-steps:

[0085] A4.1. Frame the vibration acceleration signal;

[0086] A4.2. Add a Hamming window to each frame;

[0087] A4.3, square the amplitude of each frame in the window;

[0088] A4.4. Sum the squared amplitude values ​​of each frame;

[0089] A4.5. Divide the sum by the frame length to get the average value;

[0090] A4.6. Take the square root of the average value obtained to obtain the root mean square energy in the window.

[0091] An example of a plot of RMS energy versus time is as follows:

[0092] Figure 3 This is the RMS energy spectrum of Hua 24-18 well when there is no fault. The RMS energy spectrum shows obvious periodicity, and the RMS energy changes smoothly, that is, the polished rod vibration is stable.

[0093] Figure 4 This is the RMS energy spectrum of the sucker rod collision in Well Hua 24-19. The RMS energy spectrum shows obvious periodicity and violent fluctuations, which is related to the collision of ground equipment. There are many abnormal peaks of RMS energy in the corresponding figure.

[0094] Figure 5 This is the RMS energy spectrum of the pump collision production in Xuqian 1-34. The RMS energy spectrum shows obvious periodicity and violent fluctuations, which is related to the collision of downhole equipment. There are many short-term abnormal peaks of the RMS energy in the corresponding figure.

[0095] Figure 6 This is the root mean square energy spectrum of the gas impact in Hua 24-15 well. The root mean square energy spectrum fluctuates violently in some periods. Affected by the gas, the root mean square energy tends to zero for a long period of time, that is, there is almost no vibration during this period.

[0096] Figure 7 This is the RMS energy spectrum of the Fu 145 well in the present invention when the upward obstruction occurs. The RMS energy spectrum fluctuates violently. Affected by the upward obstruction working condition, the RMS energy has a small amount of abnormal peaks.

[0097] like Figure 1 As shown, step B specifically includes the following sub-steps:

[0098] B1. Fix the vibration acceleration sensor on the upper part of the pumping unit polished rod through a clamp to monitor the vibration data of the pumping unit polished rod; the acceleration sensors for longitudinal vibration, lateral vibration and torsional vibration can be installed separately, and the three-axis vibration acceleration sensor can also be directly installed;

[0099] B2. Transfer the vibration acceleration data to the background and perform preprocessing;

[0100] B3. Extract the RMS energy of real-time vibration acceleration data.

[0101] Experts make judgments on the diagnostic results output in step D. If the experts judge that the diagnosis is accurate, a targeted maintenance plan is determined in the next step. If the experts judge that the diagnosis is inaccurate, the fault samples are re-labeled and the process returns to step A4.

[0102] like Figure 8 As shown, the data acquisition device includes a signal acquisition module, a signal processing module and a power supply module. The signal acquisition module includes a vibration acceleration sensor installed on the sucker rod. The signal output ends of the vibration acceleration sensor are respectively connected to the input ends of the preamplifier circuit. The output end of the preamplifier circuit is connected to the input end of the bandpass filter. The output end of the bandpass filter is connected to the vibration acceleration signal interface of the SoC system-on-chip of the signal processing module. The SoC system-on-chip is also connected to a clock circuit, a memory, a JTAG interface, an RS485 interface and an SD card expansion interface. The Ethernet port of the SoC system-on-chip is connected to the input end of the network transceiver, the output end of the network transceiver is connected to the input end of the network transformer, the output end of the network transformer is connected to the RJ45 interface, and the SoC system-on-chip adopts the HiSilicon Hi3518E chip.

[0103] Fig. 9 This is the positive and negative 5V power supply circuit diagram of the signal acquisition module. Fig. 9 a is a 12V to 5V converter. U11 uses the LM1117-50 low-dropout linear regulator to provide VDD5V+ power for the preamplifier circuit and bandpass filter. Fig. 9 b is a 5V to -5V circuit, which uses the TPS60401 voltage inverter to provide VDD5V- power supply for the preamplifier circuit and bandpass filter.

[0104] Fig.10It is the preamplifier circuit of the signal acquisition module, P7 is the signal access interface of the vibration acceleration sensor, and U13 uses the TL971 chip. The TL971 is a vibration acceleration signal operational amplifier produced by TI, which has low noise and high common mode rejection ratio and is suitable as the preamplifier chip of the signal acquisition module. In order to ensure the linearity of signal amplification, a dual-power supply operational amplifier is selected to minimize the common mode interference and temperature drift to ensure the linearity of the signal after amplification.

[0105] The 7th pin of the TL971 chip is connected to the VDD5V+ power supply, and the 4th pin is connected to the VDD5V- power supply; the signal output end of the vibration acceleration sensor is connected to the positive electrode of the isolation capacitor C66, the negative electrode of the isolation capacitor C66 is connected to the 3rd pin of the TL971 chip, the positive electrode of the isolation capacitor C66 is also connected to the VDD5V+ power supply through the bias resistor R63, and the negative electrode of the isolation capacitor C66 is also grounded through the bias resistor R64; the 6th pin of the TL971 chip is connected to the positive electrode of the isolation capacitor C68 through the current limiting resistor R68, and the negative electrode of the isolation capacitor C68 is connected to the input end of the bandpass filter as the ACC-AMPOUT signal output end.

[0106] For the vibration acceleration signal transmitted from a long distance, due to the large attenuation, it is hoped that the amplification factor is larger; on the contrary, for the vibration acceleration signal at a close distance, it is hoped that the amplification factor is smaller to prevent peak clipping distortion. When the input signal is weak, it shows a linear amplification trend and the gain is larger; when the input signal is strong, the gain decreases with the increase of the input signal.

[0107] Fig.11 The circuit diagram of the bandpass filter includes an ADA4898-2 chip, and the bandwidth is set to 10kHz, that is, the center frequency f=5.5kHz, and the gain A=1. Two second-order bandpass filters of the Sallen-Key structure are connected in series to form a fourth-order bandpass filter. The ACC-AMPOUT signal output end of the preamplifier circuit is connected to one end of the filter capacitor C69 through the current limiting resistor R69, and the other end of the filter capacitor C69 is connected to the third pin of the ADA4898-2 chip. The current limiting resistor R69 and the filter capacitor C69 are grounded through the filter capacitor C70, and the filter capacitor C69 and the third pin of the ADA4898-2 chip are grounded through the resistor R70.

[0108] The first pin of the ADA4898-2 chip is connected to the second pin through the resistor R72, and the second pin of the ADA4898-2 chip is grounded through the resistor R71; the fourth pin of the ADA4898-2 chip is connected to the VDD5V- power supply, and is grounded through the capacitors C71, C72 and C73 connected in parallel; the fifth pin of the ADA4898-2 chip is connected to the signal input end of the capacitor C69 through the capacitor C77 and the resistors R77 and R73 connected in series, and the eighth pin of the ADA4898-2 chip is connected to the VDD5V+ power supply.

[0109] The 7th pin of the ADA4898-2 chip serves as the vibration acceleration signal output terminal ACCOUT, which is connected to the 6th pin of the ADA4898-2 chip through a resistor R74, and the 6th pin of the ADA4898-2 chip is grounded through a resistor R75; the vibration acceleration signal output terminal ACCOUT is also connected to the 2nd pin of the signal acquisition module interface Header3, and is grounded through a resistor R76 and a capacitor C78; the 2nd pin of the signal acquisition module interface Header3 is connected to the 2nd pin or the 3rd pin of the signal processing module interface Header4, i.e., P2.

[0110] The vibration acceleration signal emitted by the vibration acceleration sensor is amplified by the preamplifier circuit, filtered by a bandpass filter to reduce noise, and then sent to the vibration acceleration signal interface of the Hi3518E chip.

[0111] Fig.12 This is the power circuit diagram of the signal processing module. Fig.12 a is a 12V to 5V circuit, using LM1117-50 chip; Fig.12 b is a 12V to 3.3V conversion circuit, U2 uses the MP1482 single-chip synchronous buck regulator; Fig.12 c is a 3.3V to 1.8V converter, U3 uses the LM1117-18 low-dropout linear regulator; Fig.12 d is a 3.3V to 1.2V circuit, and U4 uses the MP2161GJ power chip.

[0112] Fig.13It is the power port of Hi3518E chip. DVDD12_1 to DVDD12_12 ports of Hi3518E chip are connected to VDD1V2, i.e. 1.2V power supply. DDR_VDDQ1 to DDR_VDDQ6 ports of Hi3518E chip are connected to VDD1V8, i.e. 1.8V power supply. DDR_VREF, DDR_VREF0, DDR_VREF1, DDR_VREF2 ports of Hi3518E chip are connected to VDD1V8, i.e. 1.8V power supply, through resistor R7. DVDD33_1 to DVDD33_6 ports of Hi3518E chip are connected to VDD3V3, i.e. 3.3V power supply. DVSS1 to DVSS57 of Hi3518E chip are all grounded.

[0113] Fig.14 This is the Ethernet interface, clock circuit interface and JTAG interface diagram of the Hi3518E chip; Fig.15 This is the vibration acceleration signal interface and serial interface diagram of the Hi3518E chip.

[0114] Fig.16 This is the external circuit diagram of the Hi3518E chip. Fig.16 a is the clock circuit. U6 uses DS1307 serial real-time clock, which is low power consumption, provides IIC interface, calendar and clock data, and is a clock / calendar chip accessed by BCD code. It provides clock calendar data such as seconds, minutes, hours, weeks, dates, months and years. A 32.768kHz crystal oscillator is connected between pins 1 and 2, a 3V battery is connected between pins 3 and 4, pin 8 is connected to VDD3V3 power supply, pin 7 is connected to VDD3V3 power supply through resistor R18, pin 5 of DS1307 serial real-time clock is connected to R10 port of Hi3518E chip through resistor R20, and pin 6 of DS1307 serial real-time clock is connected to T10 port of Hi3518E chip through resistor R19.

[0115] Fig.16 b is the schematic diagram of the connection between Hi3518E chip and flash memory chip; KH25L12835F is a 16M flash memory chip. Hi3518E chip has two sets of SPI pins. SPI0 is used for sensor configuration, and the interface level is compatible with 1.8V and 3.3V; SPI1 is used for peripheral device docking, and the interface level only supports 3.3V. Use SPI1 to connect KH25L12835F FLASH storage chip.

[0116] Fig.16c is the vibration acceleration signal input circuit diagram of Hi3518E chip, P2 is the signal processing module access interface, pin 1 of P2 is connected to VDD12V power supply, pin 4 is grounded, pin 2 or pin 3 is connected to the ACCOUT port of the signal acquisition module, the second pin of the signal processing module interface Header4 is connected to the K15 port of Hi3518E chip through capacitor C44, and the third pin of the signal processing module interface Header4 is connected to the K16 port of Hi3518E chip through capacitor C45.

[0117] Fig.16 d is the JTAG interface circuit diagram. Pin 1 of the JTAG interface is connected to the P14 port of the Hi3518E chip, pin 2 of the JTAG interface is connected to the R13 port of the Hi3518E chip, pin 3 of the JTAG interface is connected to the T13 port of the Hi3518E chip, pin 4 of the JTAG interface is connected to the T14 port of the Hi3518E chip, pin 5 of the JTAG interface is connected to the P13 port of the Hi3518E chip, and the P16 port of the Hi3518E chip is connected to the VDD3V3 power supply through resistor R21. The JTAG interface is used for internal chip testing.

[0118] Fig.16 Figure e is the circuit diagram of the SD card expansion interface, which is convenient for expanding storage space. The Hi3518E chip has two SDIO interfaces, one of which supports SDIO2.0 and is multiplexed with the MAC interface; the other is an independent SDIO interface that supports SDXC cards and is backward compatible.

[0119] Fig.17 This is the circuit diagram of the network transceiver, using the RTL8201F Ethernet transceiver; Fig.18 The upper part is the circuit diagram of the network transformer, which uses the H1102T network transformer.

[0120] The A3 port of the ADA4898-2 chip is connected to the 13th pin of the RTL8201F Ethernet transceiver through a resistor R45, the A4 port of the ADA4898-2 chip is connected to the 12th pin of the RTL8201F Ethernet transceiver through a resistor R44, the A6 port of the ADA4898-2 chip is connected to the 10th pin of the RTL8201F Ethernet transceiver through a resistor R42, the A7 port of the ADA4898-2 chip is connected to the 9th pin of the RTL8201F Ethernet transceiver through a resistor R41, the A7 port of the ADA4898-2 chip is connected to the 27th pin of the RTL8201F Ethernet transceiver, the A8 port of the ADA4898-2 chip is connected to the 26th pin of the RTL8201F Ethernet transceiver, the A10 port of the ADA4898-2 chip is connected to the 1st pin of the RTL8201F Ethernet transceiver. The 8th pin of the RTL8201F Ethernet transceiver is connected to the A11 port of the ADA4898-2 chip and the 16th pin of the RTL8201F Ethernet transceiver. The A12 port of the ADA4898-2 chip is connected to the 32nd pin of the RTL8201F Ethernet transceiver. The B5 port of the ADA4898-2 chip is connected to the 11th pin of the RTL8201F Ethernet transceiver through a resistor R43. The B8 port of the ADA4898-2 chip is connected to the 20th pin of the RTL8201F Ethernet transceiver. The B9 port of the ADA4898-2 chip is connected to the 19th pin of the RTL8201F Ethernet transceiver. The B10 port of the ADA4898-2 chip is connected to the 15th pin of the RTL8201F Ethernet transceiver through a resistor R40. The B11 port of the ADA4898-2 chip is connected to the 17th pin of the RTL8201F Ethernet transceiver.

[0121] The third pin of the RTL8201F Ethernet transceiver is connected to the first pin of the H1102T network transformer, the fourth pin of the RTL8201F Ethernet transceiver is connected to the third pin of the H1102T network transformer, the fifth pin of the RTL8201F Ethernet transceiver is connected to the sixth pin of the H1102T network transformer, the sixth pin of the RTL8201F Ethernet transceiver is connected to the eighth pin of the H1102T network transformer, the 16th pin of the H1102T network transformer is connected to the first pin of the RJ45 network connector, the 14th pin of the H1102T network transformer is connected to the second pin of the RJ45 network connector, the 11th pin of the H1102T network transformer is connected to the third pin of the RJ45 network connector, and the ninth pin of the H1102T network transformer is connected to the sixth pin of the RJ45 network connector. The vibration acceleration signal processed by the Hi3518E chip can be uploaded to the remote server through the RJ45 network connector for background processing and monitoring.

[0122] Fig.18The lower part is the circuit diagram of the RS485 interface. The serial interface UART1 of the Hi3518E chip is connected to the serial signal input terminal of the SP3485EN chip. The UART1_RXD port of the Hi3518E chip is connected to pin 1 of the SP3485EN chip. The UART1_RTSN port of the Hi3518E chip is connected to pins 2 and 3 of the SP3485EN chip. The UART1_TXD port of the Hi3518E chip is connected to pin 4 of the SP3485EN chip. SM12T1 transient voltage suppression diodes are connected between the RS485 port of the P3485EN chip and the ground.

[0123] The above is only a preferred embodiment of the present invention, which shows and describes the basic principles and main features of the present invention and the advantages of the present invention, but does not limit the scope of patent protection of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. In addition to the above embodiments, the present invention may have other implementation modes without departing from the spirit and scope of the present invention. The present invention may also have various changes and improvements, and all technical solutions formed by equivalent substitution or equivalent transformation fall within the scope of protection required by the present invention. The scope of protection required by the present invention is defined by the attached claims and their equivalents. The technical features not described in the present invention can be realized by or using existing technologies, which will not be repeated here.

Claims

1. An intelligent diagnosis method for underground working conditions based on big data analysis, It is characterized in that The steps are as follows: A. Establish a background analysis model for underground faults; B. Obtain real-time vibration acceleration data of the polished rod of the oil pump; C. Send the RMS energy data processed by the background analysis model and real-time vibration acceleration data to the real-time identification module for comparison and diagnosis; D. Output diagnostic results; Step A specifically includes the following sub-steps: A1. Install vibration acceleration sensors on the top of multiple pumping unit polished rods to collect effective vibration acceleration data of the pumping unit polished rods in batches, including normal vibration acceleration data and various fault vibration acceleration data; A2. Transfer the vibration acceleration data to the background and pre-process it; A3. Mark the normal vibration acceleration data and fault vibration acceleration data respectively; A4. Extract the RMS energy of normal and fault vibration acceleration data; A5. Draw a graph of the RMS energy versus time. A6. The graph is stored in the sample library and trained; A7. Output the parameters required by the real-time recognition module.

2. According to claim 1, the intelligent diagnosis method for underground working conditions based on big data analysis, It is characterized in that Step B specifically includes the following sub-steps: B1. Install a vibration acceleration sensor on the upper part of the polished rod of the oil pump to monitor the vibration acceleration data of the polished rod of the oil pump; B2. Transfer the vibration acceleration data to the background and perform preprocessing; B3. Extract the RMS energy of real-time vibration acceleration data.

3. According to claim 2, the intelligent diagnosis method for underground working conditions based on big data analysis, It is characterized in that Experts make judgments on the diagnostic results output in step D. If the experts judge that the diagnosis is accurate, a targeted maintenance plan is determined in the next step. If the experts judge that the diagnosis is inaccurate, the fault samples are re-labeled and the process returns to step A4.

4. According to claim 1, the intelligent diagnosis method for underground working conditions based on big data analysis, It is characterized in that Step A4 specifically includes the following sub-steps: A4.

1. Frame the vibration acceleration signal; A4.

2. Add a Hamming window to each frame; A4.3, square the amplitude of each frame in the window; A4.

4. Sum the squared amplitude values ​​of each frame; A4.

5. Divide the sum by the frame length to get the average value; A4.

6. Take the square root of the average value obtained to obtain the root mean square energy in the window.

5. According to claim 1, the intelligent diagnosis method for underground working conditions based on big data analysis, It is characterized in that In step A6, the sample library is trained using deep convolutional neural network technology.

6. The intelligent diagnosis method for underground working conditions based on big data analysis according to claim 1, It is characterized in that The faults described in step A1 include eccentric wear of the sucker rod, upward collision of the sucker rod, insufficient liquid supply, imbalance, gas influence, upward resistance, leakage of the fixed valve and collision of the pump production.

7. The intelligent diagnosis method for underground working conditions based on big data analysis according to claim 1, It is characterized in that In step A1, the effective vibration acceleration data of the oil pump is collected by a data acquisition device, which includes a signal acquisition module, a signal processing module and a power supply module. The signal acquisition module includes a vibration acceleration sensor installed on the oil pump polished rod, and the signal output ends of the vibration acceleration sensor are respectively connected to the input ends of the preamplifier circuit, and the output end of the preamplifier circuit is connected to the input end of the bandpass filter; the signal processing module includes a HiSilicon Hi3518E chip, the output end of the bandpass filter is connected to the vibration acceleration signal interface of the Hi3518E chip, the clock port of the Hi3518E chip is connected to the clock circuit, the storage port of the Hi3518E chip is connected to the memory, the network signal output end of the Hi3518E chip is connected to the input end of the network transceiver, the output end of the network transceiver is connected to the input end of the network transformer, and the output end of the network transformer is connected to the RJ45 interface.

8. The intelligent diagnosis method for underground working conditions based on big data analysis according to claim 7, Features: The preamplifier circuit includes a TL971 chip, the signal output end of the vibration acceleration sensor is connected to the positive electrode of the isolation capacitor C66, the negative electrode of the isolation capacitor C66 is connected to the third pin of the TL971 chip, the positive electrode of the isolation capacitor C66 is also connected to the VDD5V+ power supply through the bias resistor R63, and the negative electrode of the isolation capacitor C66 is also grounded through the bias resistor R64; the sixth pin of the TL971 chip is connected to the positive electrode of the isolation capacitor C68 through the current limiting resistor R68; the bandpass filter includes ADA4898 -2 chip, the negative electrode of the isolation capacitor C68 is connected to one end of the filter capacitor C69 through the current limiting resistor R69, the other end of the filter capacitor C69 is connected to the third pin of the ADA4898-2 chip, the current limiting resistor R69 and the filter capacitor C69 are grounded through the filter capacitor C70, and the filter capacitor C69 and the third pin of the ADA4898-2 chip are grounded through the resistor R70; the seventh pin of the ADA4898-2 chip is connected to the K15 port or the K16 port of the Hi3518E chip.

9. The intelligent diagnosis method for underground working conditions based on big data analysis according to claim 8, Features: The network transceiver adopts RTL8201F Ethernet transceiver, the network transformer adopts H1102T network transformer, the Ethernet signal output end of the ADA4898-2 chip is connected to the signal input end of the RTL8201F Ethernet transceiver; the third foot of the RTL8201F Ethernet transceiver is connected to the first foot of the H1102T network transformer, the fourth foot of the RTL8201F Ethernet transceiver is connected to the third foot of the H1102T network transformer, and the fifth foot of the RTL8201F Ethernet transceiver is connected to the The 6th pin of the H1102T network transformer is connected to the 8th pin of the RTL8201F Ethernet transceiver, the 16th pin of the H1102T network transformer is connected to the 1st pin of the RJ45 network connector, the 14th pin of the H1102T network transformer is connected to the 2nd pin of the RJ45 network connector, the 11th pin of the H1102T network transformer is connected to the 3rd pin of the RJ45 network connector, and the 9th pin of the H1102T network transformer is connected to the 6th pin of the RJ45 network connector.

Images