Fault diagnosis system and method for activated carbon adsorption type oil gas recovery device

Abstract

The invention discloses a fault diagnosis system and method for an activated carbon adsorption type oil gas recovery device, relates to the field of fault diagnosis of oil gas recovery devices, and realizes real-time monitoring and fault early warning of the running state of the device through multi-source data fusion, cross-system correlation analysis and intelligent optimization control. The problems that fault diagnosis of an existing oil and gas recovery device depends on single-point threshold alarm, and a diagnosis mode is based on single-parameter alarm are solved. The system comprises an adsorption monitoring module used for collecting the pressure difference between an inlet and an outlet of the activated carbon adsorption type oil gas recovery device; the analysis monitoring module is used for collecting and analyzing the vibration frequency of a system vacuum pump; the absorption monitoring module is used for monitoring the working state of the vacuum pump in the analysis monitoring module; the embedded data processing module is used for preprocessing input data and signals of the monitoring modules, outputting the data and the signals to the embedded diagnosis control module and storing the data and the signals to the data storage module; the method is also suitable for the field of real-time monitoring of the operation state and fault early warning.

Description

Technical Field

[0001] The present invention relates to the technical field of fault diagnosis of oil and gas recovery devices, and particularly to a fault diagnosis system and method for an activated carbon adsorption type oil and gas recovery device. Background Art

[0002] Currently, the fault diagnosis of activated carbon oil and gas recovery devices mainly relies on single-point threshold alarm and manual experience judgment. The diagnosis method usually diagnoses based on the single-parameter alarm of the PLC system, such as the over-limit of the vacuum pump pressure, the abnormal temperature rise of the activated carbon adsorption bed, etc. The technical limitations of this method include: only monitoring the status of local equipment, such as the current of the vacuum pump, the pressure difference and temperature of the adsorption tower, and no cross-system correlation analysis of the adsorption-desorption-absorption system is established. Depending on the fixed threshold alarm, it is impossible to identify progressive faults (such as the slow poisoning of activated carbon).

[0003] The problems existing in the prior art mainly include:

[0004] 1. Diagnostic lag: The existing methods can only identify the occurred obvious faults and cannot achieve early warning. For example, the vibration spectrum characteristics of the vacuum pump bearing wear have changed 3 months before the fault.

[0005] 2. High false alarm rate: The single-parameter alarm is easily interfered by the working condition fluctuations.

[0006] 3. The cross-system fault conduction mechanism is not clear, resulting in a long average fault location time.

[0007] The activated carbon adsorption type oil and gas recovery device consists of an adsorption system, a desorption system, and an absorption system. These three component systems are not isolated but have a linkage relationship. The linkage relationship of the adsorption system, desorption system, and absorption system is closely related to the key index parameters, and the coordinated optimization of these parameters directly affects the recovery efficiency, energy consumption, and stability of the device. Taking the adsorption system and the desorption system as an example, when the tail gas concentration of the adsorption system approaches the emission limit value or the adsorption capacity of the activated carbon reaches the threshold, the desorption system is triggered to start. The desorption pressure and time directly affect the adsorption efficiency recovery rate after the regeneration of the activated carbon. If the regeneration is not complete, the processing capacity of the adsorption subsystem will decrease.

[0008] In summary, in view of the limitations and problems of the prior art, the present invention combines the linkage relationship of the adsorption system, desorption system, and absorption system, and proposes a fault diagnosis system and method for an activated carbon adsorption type oil and gas recovery device, which realizes the real-time monitoring and fault warning of the device operation state through multi-source data fusion, cross-system correlation analysis, and intelligent optimization control. Summary of the Invention

[0009] Through multi-source data fusion, cross-system correlation analysis, and intelligent optimization control, the present invention realizes real-time monitoring and fault warning of the operating state of the device, and solves the problems that the fault diagnosis of the existing oil and gas recovery device relies on single-point threshold alarm and the diagnosis method is based on single-parameter alarm. To solve the above technical problems, the present invention is realized through the following technical solutions:

[0010] The present invention monitors the parameters in the adsorption monitoring module, desorption monitoring module, and absorption monitoring module to realize the monitoring of the adsorption system, desorption system, and absorption system, and establishes a linkage analysis mechanism for the adsorption-desorption-absorption system, and proposes a joint diagnosis method for the adsorption system-desorption system-absorption system, breaking through the traditional single-point monitoring and local analysis mode.

[0011] Solution 1: The present invention proposes a fault diagnosis system for an activated carbon adsorption type oil and gas recovery device. The activated carbon adsorption type oil and gas recovery device is composed of an adsorption system, a desorption system, and an absorption system. The fault diagnosis system includes an adsorption monitoring module, a desorption monitoring module, an absorption monitoring module, an embedded data processing module, an embedded diagnosis control module, a diagnosis result feedback execution module, an HMI terminal, and a data storage module;

[0012] The adsorption monitoring module includes a differential pressure sensor, an infrared temperature array, and a humidity sensor. The differential pressure sensor collects the differential pressure at the inlet, middle, and outlet of the activated carbon adsorption type oil and gas recovery device; the infrared temperature array collects the bed temperature gradient; the humidity sensor collects the moisture content of the activated carbon;

[0013] The desorption monitoring module includes a three-axis vibration sensor, a vacuum degree transmitter, and a desorbed gas component sensor. The three-axis vibration sensor collects the vibration frequency of the desorption vacuum pump; the vacuum degree transmitter collects the vacuum degree of the desorption system; the gas component sensor collects the oil and gas concentration of the desorbed gas;

[0014] The absorption monitoring module includes a gas-liquid two-phase flowmeter and a tail gas gas component sensor. The gas-liquid two-phase flowmeter collects the absorption liquid flow rate and the desorbed gas flow rate, and the tail gas gas component sensor collects the oil and gas concentration in the tail gas;

[0015] The embedded data processing module is used to adopt a multi-channel data acquisition card to receive the signals of each monitoring module through the CAN bus, integrate functions such as differential pressure calculation, vibration spectrum analysis, temperature field reconstruction, and gas concentration conversion, preprocess the input data and signals of each monitoring module, output them to the embedded diagnosis control module, and store them in the data storage module;

[0016] The embedded diagnosis control module is used to perform fault diagnosis according to the data received from the embedded data processing module according to the built-in algorithm and the fault feature library, and output the fault code to the diagnosis result feedback execution module and the HMI terminal;

[0017] The diagnostic result feedback execution module is used to trigger control execution according to the fault code, including: adsorption tower switching, vacuum pump frequency modulation, spray regulating valve control, etc., and output the execution result to the HMI terminal;

[0018] The HMI terminal is used to visually display various data in real time and accept user operations;

[0019] The data storage module is used to store monitoring data, diagnostic data, and data feedback by the diagnostic result feedback execution module.

[0020] Furthermore, a preferred implementation is provided, where the differential pressure sensor, infrared temperature array, and humidity sensor are evenly distributed at equal intervals along the axial direction of the activated carbon bed layer.

[0021] Furthermore, a preferred implementation is provided, where the triaxial vibration sensor is installed on the bearing seat of the vacuum pump, and the vacuum degree transmitter and desorption gas component analyzer are installed on the connecting pipeline between the vacuum pump and the desorption tank.

[0022] Furthermore, a preferred implementation is provided, where the fault feature library in the embedded diagnostic control module includes feature thresholds for fault modes such as activated carbon breakthrough, local saturation, and excessive moisture.

[0023] Solution 2: A fault diagnosis method for an activated carbon adsorption type oil and gas recovery device. The fault diagnosis method is implemented based on the system described in Solution 1. The fault diagnosis method includes the following steps:

[0024] Step 1: Align the clocks of each monitoring module using the IEEE 1588 precision clock protocol;

[0025] Step 2: Synchronously collect the index data in the adsorption system, desorption system, and absorption system;

[0026] Step 3: Extract fault-sensitive features from the index data collected in Step 2. The fault-sensitive features include time-domain features, frequency-domain features, and working condition correlation features of the data;

[0027] Step 4: Based on the fault-sensitive features extracted in Step 3, as well as the timing features and abnormal fluctuation features, and combined with the corresponding fault types, construct a fault feature library;

[0028] Step 5: Based on the fault-sensitive features extracted in Step 3, match them with the fault feature library constructed in Step 4 to obtain the fault type;

[0029] Step 6: According to the fault type obtained in Step 5, execute the processing action and perform precise fault control according to the established hierarchical response mechanism.

[0030] Further, a preferred implementation is provided. In step 2, it further includes the steps of constructing a working condition correlation matrix based on the extracted working condition correlation features, and realizing the precise capture of complex fault modes by establishing a coupling relationship matrix among the parameters of the adsorption monitoring module, the desorption monitoring module, and the absorption monitoring module.

[0031] Further, a preferred implementation is provided. In step 4, when constructing the fault feature library, the LSTM method is used to analyze the time-domain features, frequency-domain features, working condition features, and multi-system fusion features designed in step 3 for historical data, and combined with the corresponding fault types to construct the fault feature library.

[0032] Further, a preferred implementation is provided. The method for constructing the fault feature library in step 4 also includes implementing it by using the gated recurrent unit (GRU) method.

[0033] Solution 3: A computer device includes a memory and a processor. A computer program is stored in the memory. When the processor runs the computer program stored in the memory, the processor executes the method described in Solution 2.

[0034] Solution 4: A computer-readable storage medium stores a computer program. When the computer program is executed by a processor, the steps of the method described in Solution 2 are realized.

[0035] The advantages of the present invention are as follows:

[0036] The fault diagnosis system and method for an activated carbon adsorption type oil and gas recovery device described in the present invention realize the real-time monitoring and fault warning of the device operation state through multi-source data fusion, cross-system correlation analysis, and intelligent optimization control.

[0037] The fault diagnosis system and method for an activated carbon adsorption type oil and gas recovery device described in the present invention break through the traditional single-point monitoring and local analysis mode by monitoring the key parameters of the adsorption system, desorption system, and absorption system, establishing a linkage analysis mechanism for the adsorption-desorption-absorption system, and proposing a fault diagnosis system and method for an activated carbon adsorption type oil and gas recovery device.

[0038] The fault diagnosis system and method for an activated carbon adsorption type oil and gas recovery device described in the present invention break through the traditional single-point monitoring mode and reduce the false alarm rate through a cross-system joint diagnosis structure: physically integrating the monitoring devices of the adsorption monitoring module, desorption monitoring module, and absorption monitoring module to realize cross-system joint diagnosis.

[0039] The fault diagnosis system and method of an activated carbon adsorption type oil and gas recovery device according to the present invention solidify typical fault modes in a local storage chip, such as activated carbon penetration, vacuum pump blade breakage, etc., support offline diagnosis, and improve the real-time performance of fault diagnosis and early warning.

[0040] The fault diagnosis system and method of an activated carbon adsorption type oil and gas recovery device according to the present invention can timely discover the potential fault hazards leading to oil and gas leakage, take measures in advance for repair, thereby reducing the occurrence of major safety accidents such as explosion and fire; ensure the normal operation of the oil and gas recovery device, enable effective recovery and treatment of oil and gas, reduce environmental pollution, protect the ecological environment; monitor the operation status of the device in real time, accurately judge the location and degree of the fault, achieve precise maintenance, reduce the maintenance cost and workload, reduce the device downtime, ensure production continuity, and improve the economic benefits of the enterprise.

[0041] The fault diagnosis system and method of an activated carbon adsorption type oil and gas recovery device according to the present invention can improve the diagnostic response speed, reduce the unplanned shutdown rate and reduce the activated carbon replacement cycle.

[0042] The present invention is also applicable to the fields of real-time monitoring of operation status and fault early warning. Description of the Drawings

[0043] Figure 1 It is a principle block diagram of a fault diagnosis system of an activated carbon adsorption type oil and gas recovery device according to Embodiment 1.

[0044] Figure 2 It is a flowchart of a fault diagnosis method of an activated carbon adsorption type oil and gas recovery device according to Embodiment 1. Detailed Embodiments

[0045] To make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all of the embodiments.

[0046] Embodiment 1. This embodiment proposes a fault diagnosis system of an activated carbon adsorption type oil and gas recovery device. The activated carbon adsorption type oil and gas recovery device is composed of an adsorption system, a desorption system, and an absorption system. The fault diagnosis system includes an adsorption monitoring module, a desorption monitoring module, an absorption monitoring module, an embedded data processing module, an embedded diagnosis control module, a diagnosis result feedback execution module, an HMI terminal, and a data storage module;

[0047] The adsorption monitoring module includes a differential pressure sensor, an infrared temperature array, and a humidity sensor. The differential pressure sensor collects the differential pressures at the inlet, middle, and outlet of the activated carbon adsorption type oil and gas recovery device; the infrared temperature array collects the temperature gradient of the bed layer; the humidity sensor collects the moisture content of the activated carbon.

[0048] The desorption monitoring module includes a three-axis vibration sensor, a vacuum degree transmitter, and a desorbed gas component sensor. The three-axis vibration sensor collects the vibration frequency of the desorption vacuum pump; the vacuum degree transmitter collects the vacuum degree of the desorption system; the gas component sensor collects the oil and gas concentration of the desorbed gas.

[0049] The absorption monitoring module includes a gas-liquid two-phase flowmeter and a tail gas gas component sensor. The gas-liquid two-phase flowmeter collects the flow rates of the absorption liquid and the desorbed gas, and the tail gas gas component sensor collects the oil and gas concentration in the tail gas.

[0050] The embedded data processing module is used to use a multi-channel data acquisition card to receive the signals of each monitoring module through the CAN bus, integrate functions such as differential pressure calculation, vibration spectrum analysis, temperature field reconstruction, and gas concentration conversion, preprocess the input data and signals of each monitoring module, output them to the embedded diagnosis and control module, and store them in the data storage module.

[0051] The embedded diagnosis and control module is used to perform fault diagnosis according to the data received from the embedded data processing module according to the built-in algorithm and the fault feature library, and output the fault code to the diagnosis result feedback execution module and the HMI terminal.

[0052] The diagnosis result feedback execution module is used to trigger control execution according to the fault code, including: adsorption tower switching, vacuum pump frequency modulation, spray regulating valve control, and output the execution result to the HMI terminal.

[0053] The HMI terminal is used to visually display various data in real time and can accept user operations.

[0054] The data storage module is used to store monitoring data, diagnosis data, and the data feedback by the diagnosis result feedback execution module.

[0055] Embodiment 2: This embodiment further limits the fault diagnosis system of an activated carbon adsorption type oil and gas recovery device described in Embodiment 1. The differential pressure sensor, the infrared temperature array, and the humidity sensor are evenly distributed along the axial direction of the activated carbon bed layer.

[0056] Embodiment 3: This embodiment further limits the fault diagnosis system of an activated carbon adsorption type oil and gas recovery device described in Embodiment 1. The three-axis vibration sensor is installed on the bearing seat of the vacuum pump, and the vacuum degree transmitter and the desorbed gas component analyzer are installed on the connecting pipeline between the vacuum pump and the desorption tank.

[0057] Embodiment 4. This embodiment further defines a fault diagnosis system for an activated carbon adsorption type oil and gas recovery device described in Embodiment 1. The fault feature library in the embedded diagnosis control module includes feature thresholds for fault modes such as activated carbon breakthrough, local saturation, and excessive moisture.

[0058] Embodiment 5. This embodiment proposes a fault diagnosis method for an activated carbon adsorption type oil and gas recovery device. The fault diagnosis method is implemented based on the system described in Embodiment 1. The fault diagnosis method includes the following steps:

[0059] Step 1: Align the clocks of each monitoring module using the IEEE 1588 Precision Clock Protocol;

[0060] Step 2: Synchronously collect the index data in the adsorption system, desorption system, and absorption system;

[0061] Step 3: Extract fault-sensitive features from the index data collected in Step 2. The fault-sensitive features include time-domain features, frequency-domain features, and operating condition-related features of the data;

[0062] Step 4: Based on the fault-sensitive features extracted in Step 3, as well as the time-series features and abnormal fluctuation features, and combined with the corresponding fault types, construct a fault feature library;

[0063] Step 5: Based on the fault-sensitive features extracted in Step 3, match them with the fault feature library constructed in Step 4 to obtain the fault type;

[0064] Step 6: According to the fault type obtained in Step 5, execute the processing action and perform precise fault control according to the established hierarchical response mechanism.

[0065] Embodiment 6. This embodiment further defines the fault diagnosis method for an activated carbon adsorption type oil and gas recovery device described in Embodiment 6. Step 2 further includes the steps of constructing an operating condition correlation matrix based on the extracted operating condition-related features, and realizing precise capture of complex fault modes by establishing a coupling relationship matrix between the parameters of the adsorption monitoring module, desorption monitoring module, and absorption monitoring module.

[0066] Embodiment 7. This embodiment further defines the fault diagnosis method for an activated carbon adsorption type oil and gas recovery device described in Embodiment 5. In Step 4, when constructing the fault feature library, the LSTM method is used to analyze the time-domain features, frequency-domain features, operating condition features, and multi-system fusion features designed in Step 3 for historical data, and combined with the corresponding fault types, a fault feature library is constructed.

[0067] Embodiment VIII. This embodiment further defines the fault diagnosis method for the activated carbon adsorption type oil and gas recovery device described in Embodiment V. The method for constructing the fault feature library in Step 4 further includes implementing it using the Gated Recurrent Unit (GRU) method.

[0068] Embodiment IX. This embodiment proposes a computer device, including a memory and a processor. A computer program is stored in the memory. When the processor runs the computer program stored in the memory, the processor executes the method described in Embodiment VI.

[0069] Embodiment X. This embodiment proposes a computer-readable storage medium. The computer-readable storage medium stores a computer program. When the computer program is executed by a processor, the steps of the method described in Embodiment VI are implemented.

[0070] Embodiment XI. This embodiment presents an example. The example is used to explain Embodiments I to X above. Specifically, the example is as follows:

[0071] See Figure 1 and Figure 2 To illustrate this embodiment, this embodiment specifically includes that the fault diagnosis module is composed of an adsorption system monitoring module, a desorption system monitoring module, an absorption system monitoring module, an embedded data processing module, an embedded diagnosis control module, a diagnosis result feedback execution module, an HMI terminal, a data storage module, etc.

[0072] The adsorption monitoring module includes a differential pressure sensor, an infrared temperature array, and a humidity sensor, which are evenly distributed along the axial direction of the activated carbon bed layer at equal intervals; the differential pressure sensor adopts a double diaphragm overpressure protection structure, with a range of 0 - 15 kPa and a withstand voltage of ≥ 30 kPa.

[0073] The desorption monitoring module includes a three-axis vibration sensor, a vacuum degree transmitter, and a desorbed gas component sensor, which are respectively installed on the connecting pipeline between the vacuum pump bearing seat and the desorption tank;

[0074] The absorption monitoring module includes a gas-liquid two-phase flowmeter and a tail gas gas component sensor. The gas-liquid two-phase flowmeter is of the ultrasonic type.

[0075] The embedded data processing module is used to adopt a multi-channel data acquisition card, 16-bit ADC, a sampling rate of 100 kHz, receive signals from each monitoring module through the CAN bus, design an anti-interference circuit, EMI filter + signal isolator, integrate functions such as differential pressure calculation, vibration spectrum analysis, temperature field reconstruction, and gas concentration conversion, preprocess the input data and signals of each monitoring module, output them to the embedded diagnosis control module, and store them in the data storage module.

[0076] An embedded diagnostic control module, which integrates an FPGA chip. It has a built-in multi-source data fusion algorithm and performs fault diagnosis according to the data received from the embedded data processing module, in accordance with the built-in algorithm and the fault feature library. The fault feature library is established by analyzing historical data using an LSTM neural network. The fault feature library contains feature thresholds for 32 fault modes such as activated carbon breakthrough, local saturation, and excessive moisture.

[0077] A diagnostic result feedback execution module, which is used to control the execution according to the fault code, including: adsorption tower switching, vacuum pump frequency modulation, spray regulating valve control, and outputs the execution result to the HMI terminal.

[0078] An HMI terminal, which is used to visually display various data in real time and can accept user operations.

[0079] A data storage module, which is used to store monitoring data, diagnostic data, and actuator feedback data.

[0080] In this embodiment, a differential pressure sensor is used to collect the differential pressure at the inlet and outlet of the adsorption system, and judge whether the adsorption system is working properly according to the differential pressure data. The differential pressure signal processing flow: The differential pressure sensor transmits to the signal conditioning circuit (with a gain of 100 times) and then transmits to the ADC module, and the ADC module is implemented with 16 bits.

[0081] (1) Differential pressure signal acquisition: The differential pressure sensor collects the differential pressure signal ΔP (kPa) at the inlet and outlet and converts ΔP into a voltage signal mV.

[0082] (2) Signal conditioning circuit: An instrumentation amplifier is used, and means such as low-pass filtering, voltage boosting, and linearization compensation are integrated to optimize the collected voltage signal.

[0083] (3) ADC module: A 16-bit ADC module is used to digitally process the voltage signal. The 16-bit resolution converts the analog signal into 65536 discrete values, improving the measurement accuracy.

[0084] (4) Signal chain timing diagram: The differential pressure change is transmitted to the sensor response and then sent to the conditioning circuit for filtering, then sampled by the ADC, then processed by the MCU, and finally output by the RS485 communication method.

[0085] Among them, one differential pressure sensor is respectively set at each of the inlet and outlet, and three are set in the middle, and the infrared temperature array is distributed in a 6×6 grid.

[0086] In this embodiment, a three-axis vibration sensor is used to collect and analyze the vibration frequency of the vacuum pump of the analysis system, and it is judged whether the vacuum pump of the analysis system is working properly according to the vibration frequency. The vibration frequency processing flow: The vibration sensor transmits to an anti-aliasing filter, the cut-off frequency of the filter is 5 kHz, and finally it is transmitted to the FPGA for FFT analysis.

[0087] (1) Vibration signal acquisition: A piezoelectric acceleration sensor is used to synchronously collect the vibration signals in three orthogonal directions of X, Y, and Z of the vacuum pump. The vibration signals include acceleration, velocity, and displacement, and multi-dimensional vibration is detected. For example, radial vibration is used for rotor imbalance; multi-directional vibration is used for bearing faults, and the synthetic vector vibration amplitude is used to comprehensively evaluate the equipment status. An analog voltage signal (±10V) is output, and the signal amplitude is proportional to the vibration amplitude. The frequency response covers 0.1 Hz to 10 kHz, covering the fault characteristic frequencies of components such as the vacuum pump bearings and impellers.

[0088] (2) Filter: A second-order / fourth-order Butterworth low-pass filter is used, and the parameter is selected as a cut-off frequency of 5 kHz to match the typical fault frequencies of the vacuum pump, such as the bearing fault frequency <5 kHz and the impeller blade passing frequency <4 kHz. The slope is selected as -20 dB / decade (second order) or -40 dB / decade (fourth order) to ensure high-frequency attenuation.

[0089] (3) FPGA, that is, FFT analysis: The FPGA is used to perform digital processing on the signal. The FPGA built-in ADC digitizes the filtered signal at a sampling rate of 10 kHz to 20 kHz, with a 16-bit ADC and a resolution of 0.1 mV. Then, a fast Fourier transform is performed on the time-domain signal to convert it into the frequency domain. The amplitudes / phase angles of each frequency component are calculated, for example, the 50 Hz power frequency, the bearing characteristic frequency, and the impeller natural frequency.

[0090] Among them, a. Sensor arrangement: To ensure the shortest vibration transmission path, it needs to be arranged close to the vacuum pump bearing housing, and a magnetic adsorption type installation is used to ensure rigid coupling. b. Filter calibration: Regularly calibrate the 5 kHz cut-off point to ensure that the amplitude attenuation > 3 dB.

[0091] A fault diagnosis method for an activated carbon adsorption type oil and gas recovery device described in this embodiment includes the following steps:

[0092] 1. Clock synchronization: The IEEE 1588 Precision Time Protocol, that is, PTP, is used to achieve clock alignment at the microsecond level;

[0093] 2. Data acquisition: On the basis of clock alignment, high-precision synchronous acquisition of the adsorption system, analysis system, and absorption system is realized; after the data is acquired, sliding window filtering is performed, and the window length = 5 s to eliminate transient noise.

[0094] The following algorithm is used for sliding window filtering:

[0095]

[0096] 3. Feature extraction: Feature extraction refers to extracting fault-sensitive features from the collected raw data. The fault-sensitive features include time-domain features, frequency-domain features, and operating condition-related features.

[0097] First, time-domain features: Five indicators such as standard deviation, kurtosis, zero-crossing rate, trend slope, and window range are selected in time-domain features.

[0098] (1) Standard deviation: The standard deviation is used to represent the degree of fluctuation of the pressure difference between the inlet and outlet of the absorption system in a fixed time window. For example, when the adsorption capacity of activated carbon decreases, the standard deviation of the pressure difference between the inlet and outlet of the adsorption tower will increase significantly. At the same time, the standard deviation can also be used to identify pressure difference sensor faults. For example, if the standard layer is continuously lower than the threshold, it indicates sensor drift or dead zone faults, etc.

[0099] (2) Kurtosis: Kurtosis is used to detect transient abnormal times. For example, kurtosis can be used to detect abnormal actions of the regeneration valve of the analysis system. When the kurtosis of the regeneration valve opening signal oscillates periodically, it indicates that the solenoid valve is stuck.

[0100] (3) Zero-crossing rate: The zero-crossing rate is used to detect the operating condition stability. For example, the zero-crossing rate is used to detect the stability of the pressure difference between the inlet and outlet of the adsorption system during regeneration. If the zero-crossing rate is low, the operating condition of the adsorption system is stable.

[0101] (4) Trend slope: The trend slope of the pressure difference between the inlet and outlet of the adsorption system is used to monitor the saturation of activated carbon. When the trend slope of the pressure difference between the inlet and outlet for a continuous time > threshold Tycts, an activated carbon saturation warning is issued.

[0102] (5) Window range: The window range is used to monitor the short-term violent fluctuations of the adsorption system, analysis system, and absorption system of the oil and gas recovery device.

[0103] Second, for frequency-domain features, multiple dynamic equipment are used in the oil and gas recovery device, such as vacuum pumps, lean oil pumps, rich oil pumps, etc. The operating conditions of these dynamic equipment are crucial to the performance of the oil and gas recovery device. Therefore, frequency-domain features are used to capture the vibration conditions of the dynamic equipment, analyze the operating conditions of these dynamic equipment, and then jointly diagnose the faults of the oil and gas recovery device. For example, the sideband modulation index of the frequency-domain features is collected to analyze the damage of the vacuum pump bearing.

[0104] Third, for operating condition-related features, operating condition-related features are extracted from three dimensions of the adsorption monitoring module, analysis monitoring module, and absorption monitoring module, and an operating condition correlation matrix is constructed.

[0105] (1) The operating condition-related features of the adsorption system include: the pressure difference ΔP of the adsorption tower, the inlet temperature T in, Outlet oil and gas concentration C out , the constructed operating condition correlation matrix is as follows:

[0106]

[0107] The first row of the above operating condition correlation matrix represents the steady-state operation index, and the second row uses the first derivative of the parameters in the first row to represent the transient process characteristics.

[0108] (2) Analyzing the system operating condition correlation characteristics includes: vacuum-breaking pressure P vb , analyzer tank temperature T des , vacuum degree V acuum . Using the moving average MA of the above operating condition correlation characteristics as dynamic parameters, the constructed operating condition correlation matrix is as follows:

[0109]

[0110] Among them, the MA calculation formula is as follows:

[0111]

[0112] (3) Absorption system operating condition correlation characteristics include: lean oil flow rate Q liq , liquid level L tank , gas flow rate Q gas . And using the coefficient of variation of the above operating condition correlation characteristics as dynamic parameters, the constructed operating condition correlation matrix is as follows:

[0113]

[0114] Among them, the CV calculation formula is as follows:

[0115]

[0116] (4) Multi-system feature fusion: Perform a tensor product operation on the state matrices of the three subsystems to construct a multi-system operating condition coupling correlation matrix. The dimension of the multi-system operating condition coupling correlation matrix is an 8×27 matrix. As follows.

[0117] T = S abs ⊕ S des ⊕ S ads

[0118] Furthermore, in order to reduce the processing workload, this solution uses principal component analysis (PCA) to reduce the dimension of the above 8×27 matrix and extract key coupling features.

[0119] 4. Fault feature library construction: Use LSTM to analyze historical data based on the time-domain features, frequency-domain features, operating condition features, and multi-system fusion features designed in step 3, and combine the corresponding fault types to construct a fault feature library.

[0120] 5. Fault matching: Compare the feature vectors extracted from the real-time data in step 3 with the 32 fault modes in the fault feature library.

[0121] Furthermore, an optimization solution can be adopted: Use a gated recurrent unit (GRU) to construct the fault feature library to reduce the computational cost and support deployment on an FPGA.

[0122] Furthermore, an optimization solution can be adopted: Implement online optimization of the model using incremental learning.

[0123] 6. Control execution: Trigger precise control according to the result of the fault matching in step 5.

[0124] Furthermore, an optimization solution can be adopted: Establish a fault grading response mechanism.

[0125] The fault levels are divided into 3 levels, and different levels trigger different actions. The specific level division is shown in the following table:

[0126] Level Trigger condition Execution delay requirement Recovery difficulty Level 1 Single parameter out of limit ≤500ms Automatic recovery Level 2 Abnormal correlation of multiple parameters or failure of Level 1 response ≤1s Manual intervention Level 3 Safety risk ≤300ms Emergency shutdown

[0127] Taking the level 1 response as an example, when the pressure difference of the adsorption tower exceeds the limit, enable the level 1 response and adjust the pressure difference of the adsorption tower.

[0128] Furthermore, an optimization solution can be adopted: Use a fuzzy PID algorithm to dynamically control the valve opening to achieve precise valve adjustment and adjust the pressure difference of the adsorption tower.

[0129] Those skilled in the art can understand that the above description is only the preferred embodiment of the present invention. The features described in each embodiment and / or claim of the present disclosure can be combined or combined in various ways, even if such combinations or combinations are not explicitly described in the present disclosure. It is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some of the technical features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

[0130] Although the preferred embodiments of the present invention have been described, additional changes and modifications can be made to these embodiments by those skilled in the art once they learn of the basic inventive concept. Therefore, the appended claims are intended to be construed to include the preferred embodiments as well as all changes and modifications that fall within the scope of the present invention. Obviously, those skilled in the art can make various changes and variations 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 is also intended to include these modifications and variations.

Claims

1. A fault diagnosis system for an activated carbon adsorption type oil and gas recovery device, characterized in that, The activated carbon adsorption type oil and gas recovery device is composed of an adsorption system, a desorption system, and an absorption system. The fault diagnosis system includes an adsorption monitoring module, a desorption monitoring module, an absorption monitoring module, an embedded data processing module, an embedded diagnosis control module, a diagnosis result feedback execution module, an HMI terminal, and a data storage module; The adsorption monitoring module includes a differential pressure sensor, an infrared temperature array, and a humidity sensor. The differential pressure sensor collects the differential pressures at the inlet, middle, and outlet of the activated carbon adsorption type oil and gas recovery device; the infrared temperature array collects the temperature gradient of the bed layer; the humidity sensor collects the moisture content of the activated carbon; The desorption monitoring module includes a three-axis vibration sensor, a vacuum degree transmitter, and a desorbed gas component sensor. The three-axis vibration sensor collects the vibration frequency of the desorption vacuum pump; the vacuum degree transmitter collects the vacuum degree of the desorption system; the gas component sensor collects the oil and gas concentration of the desorbed gas; The absorption monitoring module includes a gas-liquid two-phase flowmeter and a tail gas gas component sensor. The gas-liquid two-phase flowmeter collects the flow rates of the absorption liquid and the desorbed gas, and the tail gas gas component sensor collects the oil and gas concentration in the tail gas; The embedded data processing module is used to use a multi-channel data acquisition card to receive the signals of each monitoring module through the CAN bus, integrate functions such as differential pressure calculation, vibration spectrum analysis, temperature field reconstruction, and gas concentration conversion, preprocess the input data and signals of each monitoring module, output them to the embedded diagnosis control module, and store them in the data storage module; The embedded diagnosis control module is used to perform fault diagnosis according to the data received from the embedded data processing module according to the built-in algorithm and the fault feature library, and output fault codes to the diagnosis result feedback execution module and the HMI terminal; The diagnosis result feedback execution module is used to trigger control execution according to the fault code, including: adsorption tower switching, vacuum pump frequency modulation, spray regulating valve control, and output the execution result to the HMI terminal; The HMI terminal is used to visually display various data in real time and can accept user operations; The data storage module is used to store monitoring data, diagnosis data, and data feedback by the diagnosis result feedback execution module.

2. The fault diagnosis system of the activated carbon adsorption type oil and gas recovery device according to claim 1, characterized in that, The differential pressure sensor, the infrared temperature array, and the humidity sensor are evenly distributed along the axial direction of the activated carbon bed layer.

3. The fault diagnosis system of the activated carbon adsorption type oil and gas recovery device according to claim 1, characterized in that, The three-axis vibration sensor is installed on the bearing seat of the vacuum pump, and the vacuum degree transmitter and the desorbed gas component analyzer are installed on the connecting pipeline between the vacuum pump and the desorption tank.

4. The fault diagnosis system of the activated carbon adsorption type oil and gas recovery device according to claim 1, characterized in that, The fault feature library in the embedded diagnosis control module includes the feature thresholds of fault modes such as activated carbon breakthrough, local saturation, and excessive moisture.

5. A fault diagnosis method for an activated carbon adsorption type oil and gas recovery device, characterized in that, The fault diagnosis method is implemented based on the system described in claim 1. The fault diagnosis method includes the following steps: Step 1: Align the clocks of each monitoring module using the IEEE 1588 precision clock protocol; Step 2: Synchronously collect the index data in the adsorption system, desorption system, and absorption system; Step 3: Extract fault-sensitive features from the index data collected in step 2. The fault-sensitive features include the time-domain features, frequency-domain features, and working condition-related features of the data; Step 4: Based on the fault-sensitive features extracted in Step 3, as well as the time-series features and abnormal fluctuation features, combined with the corresponding fault types, construct a fault feature library; Step 5: Match the fault-sensitive features extracted in Step 3 with the fault feature library constructed in Step 4 to obtain the fault type; Step 6: According to the fault type obtained in Step 5, execute the processing action, and perform precise fault control according to the established hierarchical response mechanism.

6. The fault diagnosis method of the activated carbon adsorption type oil and gas recovery device according to claim 5, characterized in that, Step 2 also includes the steps of constructing a working condition association matrix based on the extracted working condition association features, and realizing the precise capture of complex fault modes by establishing a coupling relationship matrix among the parameters of the adsorption monitoring module, the analysis monitoring module, and the absorption monitoring module.

7. The fault diagnosis method of the activated carbon adsorption type oil and gas recovery device according to claim 5, characterized in that, In Step 4, when constructing the fault feature library, the LSTM method is used to analyze the time-domain features, frequency-domain features, working condition features, and multi-system fusion features designed in Step 3 for historical data, and combined with the corresponding fault types, a fault feature library is constructed.

8. The fault diagnosis method of the activated carbon adsorption type oil and gas recovery device according to claim 5, characterized in that, The method for constructing the fault feature library in Step 4 also includes the implementation using the gated recurrent unit GRU method.

9. A computer device, comprising a memory and a processor, characterized in that, A computer program is stored in the memory. When the processor runs the computer program stored in the memory, the processor executes the method described in Claim 5.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program. When the computer program is executed by the processor, the steps of the method described in Claim 5 are realized.

Images