Programmable oil well mouth controller based on 1DCNN neural network indicator diagram recognition model and control method of programmable oil well mouth controller

Abstract

The invention discloses a programmable oil well mouth controller based on a 1DCNN indicator diagram recognition model and a control method thereof, and the controller comprises a processor, a data and program memory, a USB interface, a data communication interface, an RS485 bus communication interface, a real-time clock, a ZIGBEE wireless interface, an analog quantity signal input circuit, and a switching value signal input and output circuit. A sound card interface circuit and an audio power amplifier circuit; the control method comprises the steps that load data and displacement data of a wellhead of an oil well are collected in real time according to the operation cycle of the beam-pumping unit, normalization processing is conducted on the data, then interpolation processing is conducted to expand each kind of data to the standard length so as to meet the model requirement, the data are input into a trained 1DCNN network for operation, and then the data are obtained. And outputting the oil well operation condition. The system is high in reliability and convenient to expand; remote data acquisition and control can be realized, and personnel safety can be satisfied.

Description

Technical Field

[0001] The invention relates to a programmable oil well wellhead controller based on a 1DCNN neural network work diagram recognition model and a control method thereof, and belongs to the field of oil field oil well control. Background Art

[0002] In the construction of digital oil fields, the digital construction of oil wells is the foundation of the entire digital oil field.

[0003] At present, the wellhead controllers in oil fields have many disadvantages, such as poor reliability, single function, inconvenient user expansion, and the need to rely on manufacturer upgrade programs to realize functions.

[0004] At the same time, the current interpretation of the ground indicator diagram of the beam pumping unit still relies on human eyes and experience, requires special training, and cannot be fully automated.

[0005] Therefore, a new wellhead controller is needed that can realize remote data collection and control of oil fields, as well as personnel safety, wireless communication, energy saving and other needs. At the same time, it can also effectively identify the operating status and abnormal mode of oil well equipment, and replace part of the manual work to meet the needs of the site. Summary of the invention

[0006] The purpose of the present invention is to provide a programmable oil well wellhead controller and a control method thereof based on a 1DCNN neural network work diagram recognition model. Through the intelligent analysis and feature extraction of the work diagram by the 1DCNN neural network, the operating status and abnormal mode of the oil well equipment can be effectively identified; the controller has intelligent recognition, remote data acquisition and control functions to realize the collection, efficient processing and remote data monitoring of oil well data.

[0007] To achieve the above purpose, the technical solution adopted by the present invention is as follows:

[0008] On the one hand, the present invention provides a programmable oil wellhead controller based on a 1DCNN neural network work diagram recognition model, the controller comprising a processor, a data and program memory, a USB interface, a data communication interface, an RS485 bus communication interface, a real-time clock, a ZIGBEE wireless interface, an analog signal input circuit, a switch signal input and output circuit, a sound card interface circuit and an audio power amplifier circuit;

[0009] The data and program memory is connected to the processor, and the data and program memory is used for data interaction, program downloading and program debugging with an external host computer;

[0010] The USB interface is connected to the processor, and at least two USB interfaces are provided, at least one of which is used as an internal production debugging interface, and the other is used to connect to an external USB memory;

[0011] The data communication interface is connected to the processor, and the data communication interface is used for transmission of operation data;

[0012] The RS485 bus communication interface is connected to the processor, and the RS485 bus communication interface is used to communicate with the frequency converter and the intelligent power meter unit;

[0013] The real-time clock is used to provide date and time;

[0014] The ZIGBEE wireless interface is connected to the processor, and the ZIGBEE wireless interface is used to communicate with a well site controller, or to communicate with a wireless intelligent instrument having a ZIGBEE interface;

[0015] The analog signal input circuit, the switch signal input and output circuit are connected to the processor, and the analog signal input circuit is connected to the pumping unit suspension point load sensor and the rocker angle sensor to collect load and rocker angle data. The switch signal input circuit is used to detect whether the cabinet door of the pumping unit control switch cabinet is closed, the button status and the control switch status. The switch signal output circuit is used to control the contactor, and the contactor controls the motor of the pumping unit;

[0016] The sound card interface circuit and the audio power amplifier circuit are connected to the sound card interface of the processor and are used for playing the prompt voice.

[0017] Furthermore, the processor adopts a T113-S3 processing chip.

[0018] Furthermore, the data and program memory are FLASH, NVRAM and SDRAM, wherein the FLASH of the data and program memory is an H26M41208HPRA chip supporting the EMMC5.1 interface and connected to the SDC interface integrated in the processor chip, the NVRAM of the data and program memory is a 23K256 chip connected to an SPI interface integrated in the processor, and the SDRAM of the data and program memory is a 128MB DDR3 memory integrated in the processor chip.

[0019] Furthermore, the Ethernet controller integrated in the processor is connected to the Ethernet PHY layer chip LAN8720AI via the RMII interface and then connected to the Ethernet transformer to form an RJ45 Ethernet interface, and performs data interaction, programming download and debugging with the host computer through the Ethernet network port.

[0020] Furthermore, the two UART interfaces integrated inside the processor are connected to the isolated RS485 chip CA-IS3082WX to form two RS485 interfaces; another UART interface integrated inside the processor is connected to the RS485 chip after optical coupling isolation to form another RS485 bus communication interface.

[0021] Furthermore, one of the UART interfaces integrated inside the processor is directly connected to the ZIGBEE wireless communication module.

[0022] Furthermore, an SPI interface integrated inside the processor is connected to a 16-bit 8-channel AD chip ADS8344 as an analog input via an isolation chip ADUM1411; at the same time, the SPI port is connected to two expansion IO chips to expand 32 GPIOs, of which 8 switching inputs are connected to 8 GPIOs after being isolated by optical couplers; the GPIO interface for switching output is connected to a relay via a drive circuit to constitute a switching output.

[0023] Furthermore, the real-time clock circuit is connected to the I2C interface of the processor via a real-time clock chip PCF8563 with an I2C interface.

[0024] Furthermore, the audio power amplifier circuit is connected to the sound card interface integrated with the processor, and the audio power amplifier circuit adopts the power amplifier chip CS3818EO, and the playing prompt information is controlled by software.

[0025] On the other hand, the present invention also provides a control method for a programmable oil well wellhead controller based on a 1DCNN neural network power diagram recognition model, comprising: collecting N data points of the load data and displacement data of the oil well wellhead in real time according to the operating cycle of the walking beam pumping unit, normalizing these data, and then performing interpolation processing to expand each type of data to a standard length to meet the model requirements, and then inputting these data into the trained 1DCNN neural network power diagram recognition model for calculation, giving a one-dimensional array data with K data, wherein the array data subscript starts from 0, data[0]=1 is a normal power diagram, data[n]=1 is a normal external situation, and n is an integer and takes a value from 1 to K-1.

[0026] Furthermore, the trained 1DCNN neural network performance diagram recognition model is a deep learning model built based on the TensorFlow framework and the Keras interface, which is used to convert the model into the .tflite format for development and porting on embedded devices; the data set used contains 12 different types of oil well performance diagrams, each corresponding to a different operating condition, and the model is used to classify different operating conditions.

[0027] Furthermore, the data of the 1DCNN neural network power diagram recognition model is read from the data acquisition program in the programmable controller through shared memory, and its calculation results are also output to the switch data input of the programmable controller as the data input of the PLC program. The output data of the PLC is used to control the connection and disconnection of the switch output, and the switch output is used to control the operation of the oil pump.

[0028] Furthermore, N is set to 200 and K is set to 16.

[0029] The beneficial effects of the present invention are: the present invention has high reliability, comprehensive functions and convenient expansion; it can realize remote data acquisition and control, and meet various needs such as personnel safety, wireless communication, energy saving, etc.; if implemented in an oil field, it can greatly reduce the energy consumption of oil production in low-yield oil wells.

[0030] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 It is a circuit principle block diagram of the present invention;

[0032] Figure 2 and 3 is a circuit connection diagram of a processor of the present invention;

[0033] Figure 4 It is a data program memory circuit connection diagram of the present invention;

[0034] Figure 5 It is a data communication interface circuit connection diagram of the present invention;

[0035] Figure 6 It is a connection diagram of the analog quantity processing circuit of the present invention;

[0036] Figure 7 It is a connection diagram of a switch value input circuit of the present invention;

[0037] Figure 8 It is a connection diagram of the switch output circuit of the present invention;

[0038] Fig. 9 It is a connection diagram of the network port and the RS485 bus communication interface circuit of the present invention;

[0039] Fig.10 The invention discloses a connection diagram of a power supply, a real-time clock, a sound card interface circuit and an audio power amplifier circuit. DETAILED DESCRIPTION

[0040] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0041] Embodiment 1

[0042] like Figures 1 to 10 A programmable oil well wellhead controller based on a 1DCNN neural network power diagram recognition model includes a processor, data and program memory, a USB interface, a data communication interface, an RS485 bus communication interface, a real-time clock, a ZIGBEE wireless interface, an analog signal input circuit, a switch signal input and output circuit, a sound card interface circuit and an audio power amplifier circuit.

[0043] The data and program memory is connected to the processor, and the data and program memory is used for data interaction, program download and program debugging with an external host computer; the USB interface is used for internal production debugging and external USB devices. The data communication interface is connected to the processor, and the data communication interface is used for transmission of operation data; the RS485 bus communication interface is connected to the processor, and the RS485 bus communication interface is used to communicate with the frequency converter and the intelligent power meter unit; the ZIGBEE wireless interface is connected to the processor, and the ZIGBEE wireless interface is used to communicate with the well site controller, or with a wireless intelligent meter with a ZIGBEE interface; the analog signal input circuit, the switch signal input and output circuit are connected to the processor, and the analog signal input circuit is connected to the pumping unit suspension point load sensor and the rocker angle sensor to collect load and rocker angle data, the switch signal input circuit is used to detect whether the cabinet door of the pumping unit control switch cabinet is closed, and the switch signal output circuit is used to control the contactor, and the contactor controls the motor of the pumping unit; the sound card interface circuit and the audio power amplifier circuit are connected to the processor to play the prompt voice. The real-time clock is used to generate date and time;

[0044] In this embodiment, the processor is an ARM-A7 processor, using a T113-S3 processing chip, which includes a memory controller, an Ethernet controller, GPIO, UART, SPI interface, I 2 C interface and CAN interface, sound card, timer, PWM generator, etc.

[0045] The data and program memory are FLASH, NVRAM and SDRAM. FLASH is an EMMC5.1 flash memory with an SDC interface connected to the SDC port of the processor, NVRAM is a memory with an SPI interface connected to the SPI interface of the processor, and SDRAM is integrated inside the processor.

[0046] The Ethernet controller integrated in the processor needs to be connected to the Ethernet PHY layer chip LAN8720AI via the RMII interface and then connected to the Ethernet transformer to form an RJ45 Ethernet interface.

[0047] The processor integrates 5 UART interfaces, 2 of which are connected to RS485 chip after passing through isolation chip CA-IS3082WX to form 2 RS485 interfaces, another UART interface is connected to RS485 chip after being isolated by optical coupler, another UART interface is directly connected to DIGI ZIGBEE wireless communication module, and another UART interface is connected to CONSOLE terminal.

[0048] The real-time clock circuit is provided by I 2 C interface real-time clock chip PCF8563 and the processor's I 2 C interface.

[0049] The analog input signal is connected to a 16-bit 8-channel AD chip ADS8344 after voltage protection, conversion and filtering, and the ADS8344 is connected to an isolation chip ADUM1411 and then to the SPI interface integrated in the processor.

[0050] At the same time, the SPI port is connected to two expansion IO chips MCP23S17 to expand 32 GPIO interfaces. The 8-way switch input is connected to 8 of the GPIO interfaces after optical coupling isolation; the GPIO interface used for switch output is connected to the relay through the drive circuit MC1413 to form a switch output.

[0051] The audio power amplifier circuit is connected to the built-in sound card of the processor, the voice is stored in the FLASH of the data and program memory, and is controlled by the AUDIO interface of the processor. The sound played is controlled according to time, and the sound clip includes the voice with the following text content.

[0052] "The pump will start in one minute. Please stay away from the well site. Beep, beep, beep."

[0053] "The pump will start in 30 seconds. Please stay away from the well site. Beep, beep, beep."

[0054] "The pump will start in 10 seconds. Please stay away from the well site. Beep, beep, beep."

[0055] "The pump is about to start, please stay away from the well site 5, 4, 3, 2, 1."

[0056] The voice amplifier uses a Class D amplifier with a maximum output power of 25W. The power supply of the voice amplifier is controlled by the GPIO of the processor. The GPIO is connected to the optocoupler input, and the optocoupler output controls the MOS tube. When there is a signal in the optocoupler input, the MOS tube is turned on to power the voice amplifier. When you need to play voice, turn on the power supply of the amplifier, and turn off the power supply of the amplifier normally.

[0057] The power supply uses DC24V input, which is converted into 5V and 3.3V after isolated DC / DC conversion to power the circuit. The voice amplifier is directly powered by 24V.

[0058] In the implementation of this patent, the control of the frequency converter is realized by communicating with the frequency converter through the RS485 interface of the programmable oil well wellhead controller. The RS485 interface of the frequency converter supports the MODBUS RTU protocol, and is realized by sending a frequency setting command to the frequency converter through the programmable oil well wellhead controller.

[0059] Two analog inputs can be connected to load sensors and beam angle sensors, and a set of load and beam angle data of 200 points are collected evenly in each beam motion cycle to form an oil well surface dynamometer diagram; power data can be obtained by communicating with the intelligent power meter unit, and a set of power and beam angle data of 200 points are collected evenly in each beam motion cycle to form an oil well power diagram; the oil well surface dynamometer diagram and oil well power diagram can be stored and transmitted to the host computer through the data communication interface, or transmitted to the well site controller through the ZIGBEE interface.

[0060] The programmable oil well wellhead controller will broadcast voice prompt information in segments before closing the relay to start the oil well pumping unit. After being amplified by the audio power amplifier circuit, it is converted into a sound signal through the speaker installed in the cabinet to remind personnel to leave the oil well to avoid danger.

[0061] The programmable oil well wellhead controller can start and stop the oil well pumping unit according to the time set by the host computer to achieve energy saving effect. Before starting the oil well pumping unit, it reminds personnel to leave the oil well to avoid danger.

[0062] The programmable oil well wellhead controller can decide to start or stop the beam pumping unit according to the shape of the ground dynamometer diagram.

[0063] The programmable oil wellhead controller controls the frequency of the inverter via RS485, CAN, RS232, RS422 or analog output.

[0064] Embodiment 2

[0065] A control method for a programmable oil wellhead controller based on a 1DCNN neural network power diagram recognition model includes: collecting 200 data points of the load data and displacement data of the oil wellhead in real time according to the operation cycle of the walking beam pumping unit, normalizing these data, and then performing interpolation processing to expand each data to a standard length (set to 240) to meet the model requirements, and then inputting these data into the trained 1DCNN neural network power diagram recognition model for calculation, and outputting the power diagram analysis results through calculation, including giving a one-dimensional array data with 16 data, where the array data subscript starts from 0, and data[0]=1 is a normal power diagram. The specific array is shown in the table below:

[0066] data describe Model output value data[0] Normal dynamometer The value is 1: normal power diagram; the value is 0: other data[1] Insufficient fluid supply The value is 1: insufficient liquid supply; the value is 0: other data[2] Gas Effect Value is 1: gas influence; value is 0: other data[3] Poor fluid supply The value is 1: the liquid supply is very poor; the value is 0: other data[4] Upper impact pump The value is 1: the pump is hit; the value is 0: other data[5] Lower impact pump The value is 1: the pump is hit; the value is 0: other data[6] Traveling valve closes slowly Value is 1: the floating valve closes slowly; value is 0: others data[7] Off-barrel Value is 1: the barrel is off; value is 0: other data[8] Mobile Valve Loss Value is 1: floating valve loss; value is 0: other data[9] Fixed Valve leak Value is 1: Fixed VAL loss; value is 0: Other data

[10] Sand production from oil layer Value is 1: sand production in oil layer; value is 0: other data

[11] Too fast The value is 1: the impulse is too fast; the value is 0: other data

[12] reserve data

[13] reserve data

[14] reserve data

[15] reserve

[0067] The trained 1DCNN neural network power diagram recognition model is a deep learning model built based on the TensorFlow framework and the Keras interface. It is used to convert the model into .tflite format for development and transplantation on embedded devices. The data set used contains 12 different types of oil well power diagrams, each corresponding to different working conditions, and the model is used to classify different working conditions. In addition, the data of the 1DCNN neural network power diagram recognition model is read from the data acquisition program in the programmable controller through shared memory, and its calculation results are also output to the switch data input of the programmable controller as the data input of the PLC program. The output data of the PLC is used to control the connection and disconnection of the switch output, and the switch output is used to control the operation of the pumping unit.

[0068] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the above embodiments do not limit the protection scope of the present invention in any form, and all technical solutions obtained by equivalent replacement and the like fall within the protection scope of the present invention. The parts not involved in the present invention are the same as the prior art or can be implemented by the prior art.

Claims

1. A programmable oil wellhead controller based on 1DCNN neural network work diagram recognition model, characterized in that: The controller includes a processor, data and program memory, USB interface, data communication interface, RS485 bus communication interface, real-time clock, ZIGBEE wireless interface, analog signal input circuit, switch signal input and output circuit, sound card interface circuit and audio power amplifier circuit; The data and program memory is connected to the processor, and the data and program memory is used for data interaction, program downloading and program debugging with an external host computer; The USB interface is connected to the processor, and at least two USB interfaces are provided, at least one of which is used as an internal production debugging interface, and the other is used to connect to an external USB memory; The data communication interface is connected to the processor, and the data communication interface is used for transmission of operation data; The RS485 bus communication interface is connected to the processor, and the RS485 bus communication interface is used to communicate with the frequency converter and the intelligent power meter unit; The real-time clock is used to provide date and time; The ZIGBEE wireless interface is connected to the processor, and the ZIGBEE wireless interface is used to communicate with a well site controller, or to communicate with a wireless intelligent instrument having a ZIGBEE interface; The analog signal input circuit, the switch signal input and output circuit are connected to the processor, and the analog signal input circuit is connected to the pumping unit suspension point load sensor and the rocker angle sensor to collect load and rocker angle data. The switch signal input circuit is used to detect whether the cabinet door of the pumping unit control switch cabinet is closed, the button status and the control switch status. The switch signal output circuit is used to control the contactor, and the contactor controls the motor of the pumping unit; The sound card interface circuit and the audio power amplifier circuit are connected to the sound card interface of the processor and are used for playing the prompt voice.

2. A programmable oil well wellhead controller based on 1DCNN neural network work diagram recognition model according to claim 1, characterized in that: The processor adopts T113-S3 processing chip; the data and program memory are FLASH, NVRAM and SDRAM, wherein FLASH is EMMC5.1 flash memory with SDC interface and it is connected to the SDC port of the processor, NVRAM is a memory with SPI interface and it is connected to the SPI interface of the processor, and SDRAM is integrated inside the processor.

3. A programmable oil wellhead controller based on 1DCNN neural network work diagram recognition model according to claim 1, characterized in that: The Ethernet controller integrated in the processor is connected to the Ethernet PHY layer chip via the RMII interface and then to the Ethernet transformer to form an RJ45 Ethernet interface, and performs data interaction, programming download and debugging with the host computer through the Ethernet network port.

4. A programmable oil wellhead controller based on 1DCNN neural network work diagram recognition model according to claim 1, characterized in that: The two UART interfaces integrated inside the processor are connected to an isolated RS485 chip to form two RS485 interfaces; another UART interface integrated inside the processor is connected to an RS485 chip after optical coupling isolation to form another RS485 bus communication interface; another UART interface integrated inside the processor is directly connected to a ZIGBEE wireless communication module.

5. A programmable oil well wellhead controller based on 1DCNN neural network work diagram recognition model according to claim 1, characterized in that: An SPI interface integrated inside the processor is connected to a 16-bit 8-channel AD chip via an isolation chip as an analog input; at the same time, the SPI port is connected to two expansion IO chips to expand 32 GPIOs, of which 8 switching inputs are connected to 8 GPIOs after being isolated by optical couplers; the GPIO interface for switching output is connected to a relay via a drive circuit to constitute a switching output.

6. A programmable oil wellhead controller based on 1DCNN neural network work diagram recognition model according to claim 1, characterized in that: The real-time clock circuit is connected to the I2C interface of the processor via a real-time clock chip PCF8563 with an I2C interface; the audio power amplifier circuit adopts a power amplifier chip CS3818EO, which is connected to the sound card output terminal integrated inside the processor.

7. The control method of a programmable oil well wellhead controller based on a 1DCNN neural network work diagram recognition model according to any one of claims 1 to 6, characterized in that: The load data and displacement data of the oil wellhead are collected in real time according to the operation cycle of the walking beam pumping unit. N data points are normalized and then interpolated to expand each data to the standard length to meet the model requirements. Then these data are input into the trained 1DCNN neural network power diagram recognition model for calculation, and a one-dimensional array data with K data is given, where the array data subscript starts from 0, data[0]=1 is the normal power diagram, data[n]=1 is the normal external situation, and n is an integer with a value from 1 to K-1.

8. The control method of the programmable oil well wellhead controller based on the 1DCNN neural network work diagram recognition model according to claim 7 is characterized by: The trained 1DCNN neural network performance diagram recognition model is a deep learning model built based on the TensorFlow framework and the Keras interface. It is used to convert the model into the .tflite format for development and porting on embedded devices. The dataset used contains 12 different types of oil well performance diagrams, each corresponding to a different operating condition. The model is used to classify different operating conditions.

9. The control method of the programmable oil well wellhead controller based on the 1DCNN neural network work diagram recognition model according to claim 7 is characterized in that: The data of the 1DCNN neural network power diagram recognition model is read from the data acquisition program in the programmable controller through a shared memory, and the calculation result is also output to the switch data input of the programmable controller as the data input of the PLC program. The output data of the PLC is used to control the connection and disconnection of the switch output, and the switch output is used to control the operation of the oil pump.

10. The control method of the programmable oil well wellhead controller based on the 1DCNN neural network work diagram recognition model according to claim 7 is characterized in that: The value of N is 200 and the value of K is 16.

Images