A blind source separation method with reference signal for drilling pump fault detection

Abstract

The present invention discloses a blind source separation method with reference signal for drilling pump fault detection, comprising the following steps: collecting the vibration signal of each cylinder through a vibration sensor to form a vibration signal time domain diagram, collecting the pressure signal of each cylinder through a pressure sensor to form a pressure signal time domain diagram; shaping the pressure signal into a square wave signal; decomposing the square wave signal into a rising edge narrow pulse reference signal and a falling edge narrow pulse reference signal; selecting the vibration signal and the reference signal as input signals of a CICA algorithm; using the CICA algorithm to extract the effective working condition signals of the suction valve and the discharge valve from the vibration signal respectively; using the effective working condition signals of the suction valve and the discharge valve to identify the fault characteristics of the drilling pump valve body, and completing the fault diagnosis of the drilling pump valve body. The effective working condition signals of the suction valve and the discharge valve of each cylinder body can be effectively and separately separated, and finally ten effective working condition signals are formed to identify the fault characteristics of the drilling pump valve body.

Description

Technical Field

[0001] The invention relates to the technical field of drilling pump fault detection methods, in particular to a blind source separation method with a reference signal for drilling pump fault detection. Background Art

[0002] Drilling pumps are essential key equipment in oil and gas operations. Drilling pumps work under high pressure loads for a long time. At the same time, the working medium is mud containing large particles of rock cuttings. In addition, the periodic opening and closing impact, the harsh working conditions often cause various wear and tear on the sealing parts such as pistons, cylinder liners, pump valves, etc., especially the suction valve body and discharge valve body of the hydraulic end of the drilling pump will be worn by the high-pressure mud, causing the aging and damage of the valve body seal ring, causing drilling pump failure. The hydraulic end of the drilling pump has multiple cylinders, each cylinder has a suction valve and a discharge valve, and the damaged valve body needs to be replaced frequently during operation. Since the suction valve and the discharge valve are located at both ends of the "L"-shaped split cylinder, it is necessary to locate the faulty valve body to reduce the workload of the maintenance worker. At present, the on-site maintenance of enterprises mostly adopts the method of manual listening inspection, which has great safety hazards under high-pressure working conditions.

[0003] At present, the main method for diagnosing suction valve and discharge valve failures is vibration signal analysis. However, due to the complex operating conditions of drilling pumps, the direct application of existing methods has the following two prominent problems: First, the multiple cylinders of the drilling pump crosstalk with each other, resulting in the accuracy of fault diagnosis being affected by the working conditions of adjacent cylinders; second, to eliminate the crosstalk between cylinders, the collected vibration signal contains both the effective working condition signal of the suction valve and the effective working condition signal of the discharge valve. If the valve body of the drilling pump fails, it is impossible to locate the fault to the specific suction valve body or discharge valve body. The above two problems greatly affect the efficiency and safety of drilling work. Summary of the invention

[0004] The purpose of the present invention is to provide a blind source separation method with reference signal for drilling pump fault detection in view of the above-mentioned problems. The vibration signal of the drilling pump can be used to effectively and separately extract the effective working condition signal of the suction valve and the effective working condition signal of the discharge valve of each cylinder.

[0005] The technical solution adopted by the present invention is as follows:

[0006] A blind source separation method with a reference signal for drilling pump fault detection comprises the following steps:

[0007] S1. Collecting the vibration signal of each cylinder through a vibration sensor to form a vibration signal time domain diagram, and collecting the pressure signal of each cylinder through a pressure sensor to form a pressure signal time domain diagram;

[0008] S2, shaping the pressure signal into a square wave signal for subsequent establishment of a reference signal;

[0009] S3. Decompose the square wave signal into a rising edge narrow pulse reference signal And falling edge narrow pulse reference signal ;

[0010] S4, select the vibration signal of S1 and the reference signal of S3 As the input signal of the constrained independent component analysis CICA algorithm;

[0011] S5. Use the constrained independent component analysis (CICA) algorithm to extract the effective operating condition signals of the suction valve and the discharge valve from the vibration signal. ;

[0012] S6, use the effective working condition signal of the suction valve and the discharge valve Identify the fault characteristics of the drilling pump valve body and complete the fault diagnosis of the drilling pump valve body.

[0013] Specifically, ICA is a separation method that separates a single source signal with independent statistical characteristics from a mixed signal. The CICA algorithm developed on the basis of ICA is an effective single-channel blind source separation method. At present, the CICA algorithm has good applicability and effectiveness in the fault diagnosis of equipment such as rolling bearings. However, in the prior art, according to the analysis of the working principle of drilling pumps, the CICA algorithm is not applicable to the fault diagnosis of drilling pumps. The reason is that compared with the stable working condition of the bearing, the working condition of the drilling pump is very complicated, and it is impossible to establish a periodically stable reference signal, and it is impossible to effectively determine the phase of the reference signal.

[0014] In this scheme, vibration and pressure signals are key parameters that reflect the operating status of the drilling pump. By installing sensors at key locations to collect these signals, raw data can be provided for subsequent fault detection and analysis. However, when the pressure signal and vibration signal are collected, they are continuously changing fluctuating signals containing rich frequency components, which increases the complexity of signal processing. In contrast, the square wave signal is a non-continuous periodic signal with clear switching characteristics and only two level states. Therefore, in this scheme, shaping the pressure signal into a square wave signal can simplify the waveform and make it easier to process and analyze. In addition, the square wave signal has clear edges, which is conducive to subsequent feature extraction and signal separation. Decompose the square wave signal into a rising edge narrow pulse reference signal And falling edge narrow pulse reference signal It provides a clear time reference point for the CICA algorithm, which helps to identify and synchronize the independent components during the signal separation process. This decomposition increases the time resolution of the signal, which is essential for accurate fault detection. Furthermore, in the CICA algorithm, the optimization goal is to find an unmixing matrix so that the separated signals are as independent as possible and meet the set constraints. And the constraints of the square wave signal, such as amplitude, waveform and periodicity, are used to guide the optimization process and improve the accuracy of separation. Finally, the reference signal and the imposed constraints are used to separate the independent source signal, that is, the effective working condition signal, from the mixed vibration signal. By analyzing the effective working condition signal, the fault characteristics of the drilling pump valve body, such as abnormal vibration patterns or pressure changes, can be identified.

[0015] Optionally, the step S1 includes:

[0016] S11, obtaining a vibration signal of each cylinder through a vibration sensor installed at the suction valve of the drilling pump;

[0017] S12, obtaining a pressure signal of each cylinder through a pressure sensor installed at a discharge valve of a drilling pump;

[0018] S13. Each cylinder of the drilling pump is sequentially installed on the crankshaft, and each cylinder performs suction and discharge actions alternately in each stroke cycle.

[0019] The opening and closing of the suction valve when the pump is working will produce obvious vibrations. These vibrations can be monitored by the vibration sensor and used to analyze the movement state of the valve and whether there is a fault. The discharge valve is responsible for discharging the fluid from the pump body at a certain pressure, so the pressure change at the discharge valve directly reflects the working performance of the pump. Installing the vibration sensor and the pressure sensor at the suction valve and the discharge valve respectively can make it easier to separate and analyze the signals from different sources for subsequent signal processing and fault diagnosis. Each cylinder is installed sequentially and acts alternately, so that the vibration and pressure signals of each cylinder have similar periodic characteristics. This periodicity provides a stable signal pattern for blind source separation, which helps the algorithm to identify and separate independent signal sources. The alternating action ensures that the signals of each cylinder are synchronized in time, which helps the CICA algorithm to keep time alignment when separating signals and improve the separation effect.

[0020] Optionally, in S11, the vibration signal includes the effective working condition signal of the suction valve, the effective working condition signal of the discharge valve, and the noise signal generated by the two adjacent cylinders. When the suction valve and the discharge valve are opened and closed, corresponding mechanical shocks and fluid dynamic effects will be generated, and these effects will be manifested in the form of vibration. Each valve has its own specific frequency characteristics when opening and closing, and these characteristics can be captured by the vibration sensor. By analyzing the spectrum of these signals, the working state of the valve can be identified. The noise signal of the adjacent cylinder can be used to distinguish whether it is a single valve failure or a misdiagnosis caused by interference from other cylinders. By comparing and analyzing these signals, the accuracy of fault diagnosis can be improved.

[0021] Alternatively, in S13, the phases of the multiple cylinders are different, and the phase difference intervals of the pressure signals between two adjacent cylinders are the same. Then, each cylinder sucks and discharges liquid in turn during the rotation of the pump, and the actions of adjacent cylinders are staggered. For each rotation of the crankshaft of the drilling pump, each cylinder completes a working cycle.

[0022] Optionally, step S3 includes:

[0023] S31. Analyze the periodic relationship between the pressure signal time domain diagram and the vibration signal time domain diagram; it can be seen that the pressure signal and the vibration signal have the same period.

[0024] S32. Analyze the working principle of the drilling pump and correspond the working state of the drilling pump to the segments in the time domain diagram of the pressure signal; by corresponding the working state to the signal segments, the useful information in the signal can be more accurately identified, providing a basis for fault diagnosis.

[0025] S33, decompose the square wave signal into a rising edge narrow pulse reference signal And falling edge narrow pulse reference signal By decomposing the complex square wave signal into simple narrow pulse signals, the signal processing process can be simplified, making the subsequent analysis and fault diagnosis more direct and effective.

[0026] Alternatively, in a single cycle, when the suction valve is opened, the discharge valve is closed, and the piston moves to the right, the pressure in the pump chamber decreases; when the suction valve is closed, the discharge valve is opened, and the piston moves to the left, the pressure in the pump chamber increases. Analysis shows that in S32, when the discharge valve is closed to generate a valid working condition signal, the pressure decreases, corresponding to the falling edge of the single-cycle pressure signal time domain diagram; when the suction valve is closed to generate a valid working condition signal, the pressure increases, corresponding to the rising edge of the single-cycle pressure signal time domain diagram.

[0027] Optionally, in S33, the square wave signal is decomposed into a rising edge narrow pulse reference signal having the same period as the vibration signal. As a reference signal for extracting the effective working condition signal of the suction valve, the square wave signal is decomposed into a falling edge narrow pulse reference signal with the same period as the vibration signal. As a reference signal for extracting the effective working condition signal of the discharge valve.

[0028] Optionally, step S5 includes:

[0029] S51, constructing a reference signal r(t) based on the drilling pump pressure signal, wherein the length of the reference signal is the same as the length of the vibration signal; defining the distance function between the effective working condition signal y of the suction valve and the discharge valve to be extracted and the reference signal r(t) as ε(y, r), which is used to indicate the closeness between the effective working condition signal of the suction valve and the discharge valve and the reference signal; establishing a mathematical model of the CICA algorithm;

[0030] S52, setting the initial phase, pulse width, and CICA algorithm threshold of the reference signal r(t), and extracting effective operating condition signals of the suction valve and the discharge valve.

[0031] Among them, in S51, the reference signal r(t) is constructed based on the pressure signal. It should contain the characteristics of the suction valve and the discharge valve when they are opened and closed, which is used to guide the ICA algorithm to separate the corresponding working condition signals. The distance function ε(y,r) measures the closeness between the effective working condition signal y to be extracted and the reference signal r(t), which is the optimization goal of the CICA algorithm, that is, to minimize the distance between y and r(t), so as to extract the independent components most related to the reference signal. The mathematical model of the CICA algorithm is based on the principle of statistical independence. The optimization algorithm is used to find the components that can maximize the independence of the signal, which corresponds to the effective working condition signals of the suction valve and the discharge valve. In S52, the selection of the initial phase and pulse width depends on the specific working principle and signal characteristics of the pump to ensure that the reference signal r(t) can correctly reflect the action time of the suction valve and the discharge valve. The threshold is a key parameter in the CICA algorithm. It determines the convergence of the algorithm and the accuracy of the extracted signal. The appropriate threshold can help the algorithm better separate the independent signal components. By running the CICA algorithm, according to the principle of minimizing the distance function ε(y,r), the algorithm decomposes the vibration signal into multiple independent components, among which the component closest to the reference signal r(t) is the effective operating condition signal of the suction valve and the discharge valve.

[0032] Available in S51, Mean square error Metrics, where It is the effective working condition signal; is the reference signal; E is the mathematical expectation. The mean square error takes into account the differences in the entire signal range, rather than the differences at a single point, so it can comprehensively evaluate the overall matching degree of the signal. In addition, the mean square error takes into account the differences in the entire signal range, rather than the differences at a single point, so it can comprehensively evaluate the overall matching degree of the signal.

[0033] Alternatively, in S51, the mathematical model of the CICA algorithm is as follows:

[0034] Objective function:

[0035] (1)

[0036] Constraints:

[0037] (2)

[0038] in: It is the effective working condition signal; is the reference signal; is the objective function; is a normal number; E is the mathematical expectation; is a nonlinear function; is a similarity function; is the threshold parameter; To have Gaussian variables with the same covariance matrix; Indicates that the effective operating condition signal y has unit variance; Indicates that the reference signal r has unit variance. Equation (2) is solved by the Lagrange multiplier method to obtain the best estimate of the source signal and extract the effective operating condition signals of the suction valve and the discharge valve.

[0039] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0040] 1. The blind source separation method with reference signal for drilling pump fault detection provided by the present invention collects the pressure signal of the drilling pump, shapes the pressure signal into a square wave signal, and decomposes the square wave signal into a rising edge narrow pulse reference signal. And falling edge narrow pulse reference signal According to the close relationship between the drilling pump pressure signal and the effective working condition signals of the suction valve and the discharge valve, the CICA algorithm is used to extract the effective working condition signal of the suction valve and the effective working condition signal of the discharge valve separately.

[0041] 2. The blind source separation method with reference signal for drilling pump fault detection provided by the present invention achieves three innovative goals. First, it removes the mutual crosstalk between multiple cylinders and separates the working condition signal of the real cylinder to be tested; second, it extracts the effective working condition signal of the suction valve and the effective working condition signal of the discharge valve separately, laying a good foundation for subsequent fault diagnosis. Third, this method has no requirements for working condition stability. If the working condition of the drilling pump changes, the pressure changes, and the reference signal changes accordingly, the effective working condition signal of the suction valve and the effective working condition signal of the discharge valve can still be extracted separately. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The present invention will now be described by way of example with reference to the accompanying drawings, in which:

[0043] Figure 1 is a schematic diagram of the method of the present invention;

[0044] Figure 2 is a time domain diagram of a vibration signal in an embodiment of the present invention;

[0045] Figure 3 is a time domain diagram of a pressure signal in an embodiment of the present invention;

[0046] Figure 4 4 is a flow chart of signal processing in an embodiment of the present invention. DETAILED DESCRIPTION

[0047] The present invention will be described in detail below in conjunction with the accompanying drawings.

[0048] All features disclosed in this specification, or steps in all methods or processes disclosed, except mutually exclusive features and / or steps, can be combined in any manner.

[0049] Any feature disclosed in this specification, unless otherwise stated, can be replaced by other alternative features that are equivalent or have similar purposes. That is, unless otherwise stated, each feature is only an example of a series of equivalent or similar features. The following three embodiments are given for drilling pumps with different numbers of cylinders, but in actual use, they are not limited to the solutions in the embodiments, and the cylinders in the drilling pump can also have other numbers.

[0050] Embodiment 1

[0051] A blind source separation method with reference signal for quintuple pump fault detection, such as Figure 1 As shown, it is characterized in that it includes the following steps:

[0052] S1, such as Figure 2 , 3As shown, the vibration signals of the five cylinders are collected by the vibration sensor to form a vibration signal time domain diagram, and the pressure signals of the five cylinders are collected by the pressure sensor to form a pressure signal time domain diagram, including:

[0053] S11. Obtain the vibration signal of each cylinder through the vibration sensor installed at the suction valve of the five-cylinder pump; the vibration signal includes the effective working condition signal of the suction valve, the effective working condition signal of the discharge valve and the noise signals generated by the two adjacent cylinders.

[0054] S12, obtaining a pressure signal of each cylinder through a pressure sensor installed at the discharge valve of the five-cylinder pump;

[0055] S13. The five cylinders of the five-cylinder pump are installed on the crankshaft in sequence. The five cylinders perform suction and discharge actions alternately in each stroke cycle. The phase difference of the pressure signal between two adjacent cylinders is 4π / 5. The five cylinders work in sequence.

[0056] S2, shaping the pressure signal into a square wave signal;

[0057] S3. Decompose the square wave signal into a rising edge narrow pulse reference signal And falling edge narrow pulse reference signal ,include:

[0058] S31, analyzing the periodic relationship between the pressure signal time domain diagram and the vibration signal time domain diagram;

[0059] S32. Analyze the working principle of the drilling pump, and correspond the working state of the drilling pump to the segments in the pressure signal time domain diagram. When the discharge valve is closed to generate a valid working condition signal, the pressure drops, corresponding to the falling edge of the single-cycle pressure signal time domain diagram; when the suction valve is closed to generate a valid working condition signal, the pressure increases, corresponding to the rising edge of the single-cycle pressure signal time domain diagram.

[0060] S33, decompose the square wave signal into a rising edge narrow pulse reference signal And falling edge narrow pulse reference signal , and correspond it to the working state of the drilling pump, As a reference signal for extracting the effective working condition signal of the suction valve, As a reference signal for extracting the effective working condition signal of the discharge valve.

[0061] S4, such as Figure 4 As shown, select the vibration signal of S1 and the reference signal of S3 As the input signal of the constrained independent component analysis CICA algorithm;

[0062] S5. Use the constrained independent component analysis (CICA) algorithm to extract the effective operating condition signals of the suction valve and the discharge valve from the vibration signal. ,include:

[0063] S51. Construct a reference signal r(t) based on the five-cylinder pump pressure signal. The length of the reference signal is the same as the length of the vibration signal. Define the distance function between the effective working condition signal y of the suction valve and the discharge valve to be extracted and the reference signal r(t) as ε(y, r), which is used to indicate the closeness between the effective working condition signal of the suction valve and the discharge valve and the reference signal; establish a mathematical model of the CICA algorithm; Mean square error Metrics, where It is the effective working condition signal; is the reference signal; E is the mathematical expectation, and the mathematical model of the CICA algorithm is as follows:

[0064] Objective function:

[0065] (1)

[0066] Constraints:

[0067] (2)

[0068] in: It is the effective working condition signal; is the reference signal; is the objective function; is a normal number; E is the mathematical expectation; is a nonlinear function; is a similarity function; is the threshold parameter; To have Gaussian variables with the same covariance matrix; Indicates that the signal has unit variance. Equation (2) is solved by the Lagrange multiplier method to obtain the best estimate of the source signal and extract the effective operating condition signals of the suction valve and the discharge valve.

[0069] S52, setting the initial phase, pulse width, and CICA algorithm threshold of the reference signal r(t), and extracting effective operating condition signals of the suction valve and the discharge valve.

[0070] S6, use the effective working condition signal of the suction valve and the discharge valve Identify the fault characteristics of the five-cylinder pump valve body and complete the five-cylinder pump valve body fault diagnosis.

[0071] Embodiment 2

[0072] A blind source separation method with reference signal for triplex pump fault detection, characterized in that it comprises the following steps:

[0073] S1. Collecting vibration signals of three cylinders by vibration sensors to form a vibration signal time domain diagram, and collecting pressure signals of three cylinders by pressure sensors to form a pressure signal time domain diagram, including:

[0074] S11. Obtain the vibration signal of each cylinder through the vibration sensor installed at the suction valve of the three-cylinder pump; the vibration signal includes the effective working condition signal of the suction valve, the effective working condition signal of the discharge valve and the noise signals generated by the two adjacent cylinders.

[0075] S12, obtaining a pressure signal of each cylinder through a pressure sensor installed at the discharge valve of the three-cylinder pump;

[0076] S13. The three cylinders of the three-cylinder pump are installed on the crankshaft in sequence. The three cylinders perform suction and discharge actions alternately in each stroke cycle. The phase difference of the pressure signal between two adjacent cylinders is 2π / 3, and the three cylinders work in sequence.

[0077] S2, shaping the pressure signal into a square wave signal;

[0078] S3. Decompose the square wave signal into a rising edge narrow pulse reference signal And falling edge narrow pulse reference signal ,include:

[0079] S31, analyzing the periodic relationship between the pressure signal time domain diagram and the vibration signal time domain diagram;

[0080] S32. Analyze the working principle of the drilling pump, and correspond the working state of the drilling pump to the segments in the pressure signal time domain diagram. When the discharge valve is closed to generate a valid working condition signal, the pressure drops, corresponding to the falling edge of the single-cycle pressure signal time domain diagram; when the suction valve is closed to generate a valid working condition signal, the pressure increases, corresponding to the rising edge of the single-cycle pressure signal time domain diagram.

[0081] S33, decompose the square wave signal into a rising edge narrow pulse reference signal And falling edge narrow pulse reference signal , and correspond it to the working state of the drilling pump, As a reference signal for extracting the effective working condition signal of the suction valve, As a reference signal for extracting the effective working condition signal of the discharge valve.

[0082] S4, select the vibration signal of S1 and the reference signal of S3 As the input signal of the constrained independent component analysis CICA algorithm;

[0083] S5. Use the constrained independent component analysis (CICA) algorithm to extract the effective operating condition signals of the suction valve and the discharge valve from the vibration signal. ,include:

[0084] S51, constructing a reference signal r(t) based on the three-cylinder pump pressure signal, the length of the reference signal is the same as the length of the vibration signal, defining the distance function between the effective working condition signal y of the suction valve and the discharge valve to be extracted and the reference signal r(t) as ε(y, r), which is used to indicate the closeness between the effective working condition signal of the suction valve and the discharge valve and the reference signal; establishing a mathematical model of the CICA algorithm; Mean square error Metrics, where It is the effective working condition signal; is the reference signal; E is the mathematical expectation, and the mathematical model of the CICA algorithm is as follows:

[0085] Objective function:

[0086] (1)

[0087] Constraints:

[0088] (2)

[0089] in: It is the effective working condition signal; is the reference signal; is the objective function; is a normal number; E is the mathematical expectation; is a nonlinear function; is a similarity function; is the threshold parameter; To have Gaussian variables with the same covariance matrix; Indicates that the signal has unit variance. Equation (2) is solved by the Lagrange multiplier method to obtain the best estimate of the source signal and extract the effective operating condition signals of the suction valve and the discharge valve.

[0090] S52, setting the initial phase, pulse width, and CICA algorithm threshold of the reference signal r(t), and extracting effective operating condition signals of the suction valve and the discharge valve.

[0091] S6, use the effective working condition signal of the suction valve and the discharge valve Identify the fault characteristics of the three-cylinder pump valve body and complete the three-cylinder pump valve body fault diagnosis.

[0092] Embodiment 3

[0093] A blind source separation method with reference signal for seven-cylinder pump fault detection, characterized in that it comprises the following steps:

[0094] S1. Collecting vibration signals of seven cylinders through a vibration sensor to form a vibration signal time domain diagram, and collecting pressure signals of seven cylinders through a pressure sensor to form a pressure signal time domain diagram, including:

[0095] S11. Obtain the vibration signal of each cylinder through the vibration sensor installed at the suction valve of the seven-cylinder pump; the vibration signal includes the effective working condition signal of the suction valve, the effective working condition signal of the discharge valve and the noise signals generated by the two adjacent cylinders.

[0096] S12, obtaining a pressure signal of each cylinder through a pressure sensor installed at the discharge valve of the seven-cylinder pump;

[0097] S13. The seven cylinders of the seven-cylinder pump are sequentially installed on the crankshaft. The seven cylinders perform suction and discharge actions alternately in each stroke cycle. The phase difference of the pressure signal between two adjacent cylinders is 6π / 7. The seven cylinders work in sequence.

[0098] S2, shaping the pressure signal into a square wave signal;

[0099] S3. Decompose the square wave signal into a rising edge narrow pulse reference signal And falling edge narrow pulse reference signal ,include:

[0100] S31, analyzing the periodic relationship between the pressure signal time domain diagram and the vibration signal time domain diagram;

[0101] S32. Analyze the working principle of the drilling pump, and correspond the working state of the drilling pump to the segments in the pressure signal time domain diagram. When the discharge valve is closed to generate a valid working condition signal, the pressure drops, corresponding to the falling edge of the single-cycle pressure signal time domain diagram; when the suction valve is closed to generate a valid working condition signal, the pressure increases, corresponding to the rising edge of the single-cycle pressure signal time domain diagram.

[0102] S33, decompose the square wave signal into a rising edge narrow pulse reference signal And falling edge narrow pulse reference signal , and correspond it to the working state of the drilling pump, As a reference signal for extracting the effective working condition signal of the suction valve, As a reference signal for extracting the effective working condition signal of the discharge valve.

[0103] S4, select the vibration signal of S1 and the reference signal of S3 As the input signal of the constrained independent component analysis CICA algorithm;

[0104] S5. Use the constrained independent component analysis (CICA) algorithm to extract the effective operating condition signals of the suction valve and the discharge valve from the vibration signal. ,include:

[0105] S51, constructing a reference signal r(t) based on the seven-cylinder pump pressure signal, the length of the reference signal is the same as the length of the vibration signal, defining the distance function between the effective working condition signal y of the suction valve and the discharge valve to be extracted and the reference signal r(t) as ε(y, r), which is used to indicate the closeness between the effective working condition signal of the suction valve and the discharge valve and the reference signal; establishing a mathematical model of the CICA algorithm; Mean square error Metrics, where It is the effective working condition signal; is the reference signal; E is the mathematical expectation, and the mathematical model of the CICA algorithm is as follows:

[0106] Objective function:

[0107] (1)

[0108] Constraints:

[0109] (2)

[0110] in: It is the effective working condition signal; is the reference signal; is the objective function; is a normal number; E is the mathematical expectation; is a nonlinear function; is a similarity function; is the threshold parameter; To have Gaussian variables with the same covariance matrix; Indicates that the signal has unit variance. Equation (2) is solved by the Lagrange multiplier method to obtain the best estimate of the source signal and extract the effective operating condition signals of the suction valve and the discharge valve.

[0111] S52, setting the initial phase, pulse width, and CICA algorithm threshold of the reference signal r(t), and extracting effective operating condition signals of the suction valve and the discharge valve.

[0112] S6, use the effective working condition signal of the suction valve and the discharge valve Identify the fault characteristics of the seven-cylinder pump valve body and complete the seven-cylinder pump valve body fault diagnosis.

[0113] In summary, the beneficial effect of adopting the above technical solution is that the solution can effectively separate the effective working condition signal of the suction valve and the effective working condition signal of the discharge valve of each cylinder, and finally form a multi-channel effective working condition signal for identifying the fault characteristics of the five-cylinder pump valve body. Therefore, the method is an effective blind source separation method with constraints.

[0114] The present invention is not limited to the above-mentioned specific embodiments, but extends to any new features or any new combination disclosed in this specification, as well as any new method or process steps or any new combination disclosed.

Claims

1. A blind source separation method with reference signal for drilling pump fault detection, characterized in that: The following steps are involved: S1. Collecting the vibration signal of each cylinder through a vibration sensor to form a vibration signal time domain diagram, and collecting the pressure signal of each cylinder through a pressure sensor to form a pressure signal time domain diagram; S2, shaping the pressure signal into a square wave signal; S3. Decompose the square wave signal into a rising edge narrow pulse reference signal And falling edge narrow pulse reference signal : S31, analyzing the periodic relationship between the pressure signal time domain diagram and the vibration signal time domain diagram; S32, analyzing the working principle of the drilling pump, and making the working state of the drilling pump correspond to the segments in the pressure signal time domain diagram. When the discharge valve is closed to generate an effective working condition signal, the pressure drops, corresponding to the falling edge of the single-cycle pressure signal time domain diagram; when the suction valve is closed to generate an effective working condition signal, the pressure increases, corresponding to the rising edge of the single-cycle pressure signal time domain diagram; S33, decompose the square wave signal into a rising edge narrow pulse reference signal And falling edge narrow pulse reference signal , and correspond it to the working state of the drilling pump, As a reference signal for extracting the effective working condition signal of the suction valve, As a reference signal for extracting the effective working condition signal of the discharge valve; S4, select the vibration signal of S1 and the reference signal of S3 As the input signal of the constrained independent component analysis CICA algorithm; S5. Use the constrained independent component analysis (CICA) algorithm to extract the effective operating condition signals of the suction valve and the discharge valve from the vibration signal. ; S6, use the effective working condition signal of the suction valve and the discharge valve Identify the fault characteristics of the drilling pump valve body and complete the fault diagnosis of the drilling pump valve body.

2. The blind source separation method with reference signal for drilling pump fault detection according to claim 1, characterized in that: The step S1 comprises: S11, obtaining a vibration signal of each cylinder through a vibration sensor installed at the suction valve of the drilling pump; S12, obtaining a pressure signal of each cylinder through a pressure sensor installed at a discharge valve of a drilling pump; S13. Each cylinder of the drilling pump is sequentially installed on the crankshaft, and each cylinder performs suction and discharge actions alternately in each stroke cycle.

3. The blind source separation method with reference signal for drilling pump fault detection according to claim 2, characterized in that: In S11 , the vibration signal includes an effective working condition signal of the suction valve, an effective working condition signal of the discharge valve, and noise signals generated by two adjacent cylinders.

4. The blind source separation method with reference signal for drilling pump fault detection according to claim 2, characterized in that: In S13, the phases of the multiple cylinders are different, and the phase difference intervals of the pressure signals between two adjacent cylinders are the same.

5. The blind source separation method with reference signal for drilling pump fault detection according to claim 1, characterized in that: The step S5 comprises: S51. Constructing a reference signal based on the drilling pump pressure signal The length of the reference signal is the same as the length of the vibration signal. The effective working condition signals of the suction valve and the discharge valve to be extracted are and reference signal The distance function is defined as , used to indicate the closeness between the effective working condition signals of the suction valve and the discharge valve and the reference signal; establish the mathematical model of the CICA algorithm; S52, set reference signal The initial phase, pulse width, and CICA algorithm threshold are used to extract the effective operating condition signals of the suction valve and the discharge valve.

6. The blind source separation method with reference signal for drilling pump fault detection according to claim 5, characterized in that: In S51, Mean square error Metrics, where It is the effective working condition signal; is the reference signal; E is the mathematical expectation.

7. The blind source separation method with reference signal for drilling pump fault detection according to claim 6, characterized in that: In S51, the mathematical model of the CICA algorithm is as follows: Objective function: （1） Constraints: （2） in: It is the effective working condition signal; is the reference signal; is the objective function; is a normal number; E is the mathematical expectation; is a nonlinear function; is a similarity function; is the threshold parameter; To have Gaussian variables with the same covariance matrix; Indicates that the signal has unit variance.

Images