Fracturing pump fault detection method, device, equipment, storage medium and product

Abstract

The application discloses a fracturing pump fault detection method, device, equipment, storage medium and product. A plurality of source data sequences in the operation process of the fracturing pump are acquired, then the discharge valve vibration data sequence, the suction valve vibration data sequence and the cross pin vibration data sequence are reconstructed into target discharge valve vibration data sequence, target suction valve vibration data sequence and target cross pin vibration data sequence, finally, the target discharge valve vibration data sequence, the target suction valve vibration data sequence, the target cross pin vibration data sequence, the pressure data sequence, the stroke periodic pulse sequence and the gear information sequence are input into the fracturing pump fault detection model, fault detection is performed through the fracturing pump fault detection model, the fault type of the fracturing pump is determined, the fracturing pump can be maintained in time, the downtime is reduced, and the oil and gas resource exploitation efficiency is improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of fault diagnosis, and particularly relates to a fracturing pump fault detection method, device, equipment, storage medium and product. BACKGROUND

[0002] In the development process of oil and gas resources, fracturing operation is a key link to realize efficient exploitation of oil and gas, and the running stability of a fracturing pump as a core equipment for fracturing operation directly determines the efficiency and safety of development.

[0003] The fracturing pump can be divided into a fluid end and a power end in structure. The fluid end is in a harsh working condition of high pressure and strong wear for a long time, and thus faults occur frequently and are difficult to diagnose. The power end of the fracturing pump is responsible for converting the rotary motion of a motor into the reciprocating motion of a plunger, and thus is subjected to a large alternating load for a long time and is prone to damage and mechanical looseness. The existing technology can only determine the fault type based on artificial experience after the fracturing pump fails by stopping work and disassembling the fracturing pump.

[0004] In general, the existing technology can only perform passive maintenance after a fault occurs, which leads to a long downtime of the fracturing pump and affects the efficiency of oil and gas resource exploitation. SUMMARY

[0005] The application embodiment provides a fracturing pump fault detection method, device, equipment, storage medium and product, which can discover early fault features of the fracturing pump in time, and thus actively maintain the fracturing pump in time, reduce downtime, and improve the efficiency of oil and gas resource exploitation.

[0006] In a first aspect, the application embodiment provides a fracturing pump fault detection method, which comprises:

[0007] obtaining a multi-source data sequence in a running process of the fracturing pump, the multi-source data sequence comprising: a discharge valve vibration data sequence, a suction valve vibration data sequence, a cross pin vibration data sequence, a pressure data sequence, a stroke periodic pulse sequence and a gear information sequence;

[0008] decomposing the discharge valve vibration data sequence, the suction valve vibration data sequence and the cross pin vibration data sequence to obtain a plurality of discharge valve eigenmode functions corresponding to the discharge valve vibration data sequence, a plurality of suction valve eigenmode functions corresponding to the suction valve vibration data sequence and a plurality of cross pin eigenmode functions corresponding to the cross pin vibration data sequence;

[0009] The target discharge valve eigenmodal function, the target suction valve eigenmodal function, and the target cross pin eigenmodal function are selected from a plurality of discharge valve eigenmodal functions, suction valve eigenmodal functions, and cross pin eigenmodal functions, wherein the correlation degree of the target discharge valve eigenmodal function, the target suction valve eigenmodal function, and the target cross pin eigenmodal function with the vibration data sequence corresponding thereto is greater than or equal to a preset threshold value;

[0010] The target discharge valve eigenmodal function is reconstructed into a target discharge valve vibration data sequence, the target suction valve eigenmodal function is reconstructed into a target suction valve vibration data sequence, and the target cross pin eigenmodal function is reconstructed into a target cross pin vibration data sequence;

[0011] The target discharge valve vibration data sequence, the target suction valve vibration data sequence, the target cross pin vibration data sequence, the pressure data sequence, the stroke periodic pulse sequence, and the gear information sequence are input into a fracturing pump fault detection model, and fault detection is performed through the fracturing pump fault detection model to determine the fault type of the fracturing pump, and the fault type of the fracturing pump is one of a valve body damage fault, a valve seat cracking fault, a plunger wear fault, a pump head cracking fault, and a cross pin loosening fault.

[0012] In an optional implementation of the first aspect, the target discharge valve vibration data sequence, the target suction valve vibration data sequence, the target cross pin vibration data sequence, the pressure data sequence, the stroke periodic pulse sequence, and the gear information sequence are input into a fracturing pump fault detection model, and fault detection is performed through the fracturing pump fault detection model to determine the fault type of the fracturing pump, and the fault type of the fracturing pump is one of a valve body damage fault, a valve seat cracking fault, a plunger wear fault, a pump head cracking fault, and a cross pin loosening fault.

[0013] The pressure data sequence, the stroke periodic pulse sequence, and the gear information sequence are feature fused with the target discharge valve vibration data sequence, the target suction valve vibration data sequence, and the target cross pin vibration data sequence to obtain a neuron feature vector;

[0014] Feature learning is performed based on the neuron feature vector to obtain the fault type of the fracturing pump.

[0015] In an optional implementation of the first aspect, the pressure data sequence, the stroke periodic pulse sequence, and the gear information sequence are feature fused with the target discharge valve vibration data sequence, the target suction valve vibration data sequence, and the target cross pin vibration data sequence to obtain a neuron feature vector, and the feature fusion includes:

[0016] The stroke periodic pulse sequence is set as the weight of a neuron in the fault detection model, and the pressure data sequence and the gear information sequence are set as the bias of the neuron in the fault detection model;

[0017] The target discharge valve vibration data sequence, the target suction valve vibration data sequence, and the target cross pin vibration data sequence are fused based on weights and biases to obtain a neuron feature vector.

[0018] In an optional implementation of the first aspect, the discharge valve vibration data sequence, the suction valve vibration data sequence, and the cross pin vibration data sequence are decomposed to obtain a plurality of discharge valve eigenmodes corresponding to the discharge valve vibration data sequence, a plurality of suction valve eigenmodes corresponding to the suction valve vibration data sequence, and a plurality of cross pin eigenmodes corresponding to the cross pin vibration data sequence, including:

[0019] The number of decomposition layers of the discharge valve vibration data sequence, the suction valve vibration data sequence, and the cross pin vibration data sequence is adaptively optimized to obtain the number of decomposition layers corresponding to the discharge valve vibration data sequence, the suction valve vibration data sequence, and the cross pin vibration data sequence, respectively.

[0020] The discharge valve vibration data sequence, the suction valve vibration data sequence, and the cross pin vibration data sequence are decomposed based on the number of decomposition layers corresponding to the discharge valve vibration data sequence, the suction valve vibration data sequence, and the cross pin vibration data sequence, respectively, to obtain a plurality of discharge valve eigenmodes, a plurality of suction valve eigenmodes, and a plurality of cross pin eigenmodes.

[0021] In an optional implementation of the first aspect, before the target discharge valve vibration data sequence, the target suction valve vibration data sequence, the target cross pin vibration data sequence, the pressure data sequence, the stroke periodic pulse sequence, and the gear information sequence are input into the fracturing pump fault detection model, and the fault detection is performed by the fracturing pump fault detection model to determine the fault type of the fracturing pump, the method further includes:

[0022] A plurality of multi-source data sequence sample sets are obtained, each of the plurality of multi-source data sequence sample sets includes a plurality of multi-source data sequence samples and corresponding fault category labels, and each of the plurality of multi-source data sequence samples includes a discharge valve vibration data sequence sample, a suction valve vibration data sequence sample, a cross pin vibration data sequence sample, a pressure data sequence sample, a stroke periodic pulse sequence sample, and a gear information sequence sample.

[0023] Based on the plurality of multi-source data sequence sample sets, a model training is performed on the fracturing pump fault detection model to be trained to obtain a trained fracturing pump fault detection model.

[0024] In an optional implementation of the first aspect, the plurality of multi-source data sequences in the operation of the fracturing pump are obtained, including:

[0025] The discharge valve vibration data sequence collected by a vertical direction vibration sensor installed on each cylinder of the fracturing pump is obtained.

[0026] obtain an intake valve vibration data sequence collected by a horizontal direction vibration sensor installed on each cylinder of the fracturing pump;

[0027] obtain a cross pin vibration data sequence collected by a vertical direction vibration sensor installed at a crosshead guide rail;

[0028] obtain a pressure data sequence collected by a pressure sensor installed at an outlet pipeline interface of a discharge end of the fracturing pump;

[0029] obtain a stroke periodic pulse sequence collected by a Hall sensor installed at a crankcase of the fracturing pump;

[0030] obtain a gear information sequence from a programmable logic controller of the fracturing truck.

[0031] In a second aspect, an embodiment of the present application provides a fracturing pump fault detection device, the device comprising:

[0032] an obtaining module configured to obtain a plurality of source data sequences in a running process of the fracturing pump, the plurality of source data sequences comprising: a discharge valve vibration data sequence, an intake valve vibration data sequence, a cross pin vibration data sequence, a pressure data sequence, a stroke periodic pulse sequence, and a gear information sequence;

[0033] a decomposition module configured to decompose the discharge valve vibration data sequence, the intake valve vibration data sequence, and the cross pin vibration data sequence to obtain a plurality of discharge valve eigenmode functions corresponding to the discharge valve vibration data sequence, a plurality of intake valve eigenmode functions corresponding to the intake valve vibration data sequence, and a plurality of cross pin eigenmode functions corresponding to the cross pin vibration data sequence;

[0034] a selecting module configured to select a target discharge valve eigenmode function, a target intake valve eigenmode function, and a target cross pin eigenmode function from the plurality of discharge valve eigenmode functions, the plurality of intake valve eigenmode functions, and the plurality of cross pin eigenmode functions, respectively, wherein the target discharge valve eigenmode function, the target intake valve eigenmode function, and the target cross pin eigenmode function have a correlation degree greater than or equal to a preset threshold with the vibration data sequence corresponding thereto;

[0035] a reconstruction module configured to reconstruct the target discharge valve eigenmode function into a target discharge valve vibration data sequence, reconstruct the target intake valve eigenmode function into a target intake valve vibration data sequence, and reconstruct the target cross pin eigenmode function into a target cross pin vibration data sequence;

[0036] The detection module is configured to input the target discharge valve vibration data sequence, the target suction valve vibration data sequence, the target cross pin vibration data sequence, the pressure data sequence, the stroke periodic pulse sequence, and the gear information sequence into a fracturing pump fault detection model, perform fault detection through the fracturing pump fault detection model, and determine a fault type of the fracturing pump, wherein the fault type of the fracturing pump is one of a valve body damage fault, a valve seat cracking fault, a plunger wear fault, a pump head cracking fault, and a cross pin loosening fault.

[0037] In a third aspect, an electronic device is provided, and the device includes a processor and a memory storing computer program instructions; and the processor implements the fracturing pump fault detection method according to any one of the first aspect when executing the computer program instructions.

[0038] In a fourth aspect, a computer storage medium is provided, and the computer readable storage medium stores computer program instructions; and the computer program instructions are executed by a processor to implement the fracturing pump fault detection method according to any one of the first aspect.

[0039] In a fifth aspect, a computer program product is provided, and instructions in the computer program product are executed by a processor of an electronic device to cause the electronic device to perform the fracturing pump fault detection method according to any one of the first aspect.

[0040] The fracturing pump fault detection method, apparatus, equipment, storage medium, and product provided in this application acquire multi-source data sequences during the fracturing pump's operation. These multi-source data sequences include: discharge valve vibration data sequences, suction valve vibration data sequences, crosspin vibration data sequences, pressure data sequences, stroke periodic pulse sequences, and gear information sequences. The discharge valve vibration data sequences, suction valve vibration data sequences, and crosspin vibration data sequences are then decomposed to obtain multiple intrinsic mode functions (IMFs) corresponding to the discharge valve vibration data sequences, suction valve vibration data sequences, and crosspin vibration data sequences, respectively. Target discharge valve IMFs, target suction valve IMFs, and target crosspin IMFs are then selected from these IMFs. The correlation between the intrinsic mode functions of the target discharge valve, the target suction valve, and the target crosspin and their corresponding vibration data sequences is greater than or equal to a preset threshold. Then, the intrinsic mode functions of the target discharge valve, the target suction valve, and the target crosspin are reconstructed into vibration data sequences for the target discharge valve, the target suction valve, and the target crosspin, respectively. Finally, the vibration data sequences of the target discharge valve, the target suction valve, and the target crosspin, along with the pressure data sequence, the stroke periodic pulse sequence, and the gear information sequence, are input into the fracturing pump fault detection model. Fault detection is performed through the fracturing pump fault detection model to determine the fault type of the fracturing pump. The fault types of the fracturing pump are: valve body failure, valve seat cracking, plunger wear, pump head cracking, and crosspin loosening. Compared to existing technologies that can only determine the fault type by shutting down and disassembling the fracturing pump after a failure, leading to significant maintenance delays and excessive downtime, thus affecting oil and gas extraction efficiency, this application acquires multi-source data sequences in real time. Then, it performs modal feature decomposition on the acquired discharge valve vibration data sequences, suction valve vibration data sequences, and crosspin vibration data sequences to extract intrinsic modal signals. The signals are then reconstructed to eliminate noise signals without fixed phase, mechanical source, or repetitive patterns. The target discharge valve vibration data sequence, target suction valve vibration data sequence, target crosspin vibration data sequence, pressure data sequence, stroke periodic pulse sequence, and gear information sequence are then input into a fracturing pump fault detection model. Based on the fracturing pump fault detection model, the multi-source data sequences are analyzed, enabling online monitoring and early fault warning of the fracturing pump's operating status. By identifying multi-source data sequences, the fracturing pump fault detection model can identify abnormal phenomena before a fault occurs, thereby achieving predictive maintenance of the fracturing pump, significantly reducing unplanned downtime, and improving the overall efficiency of oil and gas resource extraction. Attached Figure Description

[0041] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed to be used in the embodiments of the present application will be briefly introduced as follows, and other drawings can be obtained by those of ordinary skill in the art without any creative effort on the premise that no creative effort is needed.

[0042] Figure 1 A flowchart of a fracturing pump fault detection method provided by an embodiment of the present application is shown;

[0043] Figure 2 A flowchart of a fracturing pump fault detection method provided by another embodiment of the present application is shown;

[0044] Figure 3 A flowchart of a fracturing pump fault detection method provided by another embodiment of the present application is shown;

[0045] Figure 4 A structural diagram of a fracturing pump fault detection device provided by an embodiment of the present application is shown;

[0046] Figure 5 A hardware structural diagram of an electronic device provided by an embodiment of the present application is shown. DETAILED DESCRIPTION

[0047] The features and exemplary embodiments of various aspects of the present application will be described in detail below, in order to make the purposes, technical solutions and advantages of the present application more clear and apparent, the present application will be further described in detail below in combination with the drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain the present application, but not to limit the present application. The present application can be implemented without some of these specific details by those skilled in the art. The following description of the embodiments is only to provide a better understanding of the present application by showing examples of the present application.

[0048] It should be noted that in this paper, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply that there is any such actual relationship or order between the entities or operations. Moreover, the terms "include", "contain" or any other variants thereof are intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or device. Without more limitations, the elements defined by the statement "include" do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0049] First, the terms involved in the present application are explained:

[0050] Pump stroke sub-harmonic: refers to all harmonic components with pump stroke frequency as the base frequency.

[0051] Pump stroke frequency: refers to the frequency of the stroke of the pump, i.e. the number of times the piston or plunger completes a reciprocating motion per unit time.

[0052] Harmonic: refers to a sinusoidal component whose frequency is an integer multiple of the base frequency.

[0053] Intrinsic mode function (IMF): a signal mode with time-frequency characteristics that can reflect both time and frequency scales, which can be analyzed from both time and frequency domains.

[0054] Fault characteristic frequency band: a frequency range known through theoretical calculation or historical experience, in which the vibration energy will significantly increase when a fault occurs.

[0055] Narrowband peak: refers to a high-amplitude spectral peak in a vibration spectrum graph, which has a prominent center frequency, a sharp shape, and a narrow bandwidth. It usually corresponds to a specific natural vibration mode or resonance frequency excited by a fault in the mechanical structure.

[0056] Modulation: refers to the process of regular changes in amplitude or frequency of a vibration signal when a fault (such as a crack) is subjected to periodic impacts (such as valve body impacts or periodic changes in fluid pressure) during operation. Periodic changes in amplitude are called amplitude modulation, and periodic changes in frequency are called frequency modulation.

[0057] Frequency conversion: is the specific manifestation of the modulation process on the frequency spectrum graph. When modulation occurs, a series of symmetric and uniformly spaced sideband components will be generated on both sides of a high-frequency carrier component (i.e. "narrowband peak"). This phenomenon is called "frequency conversion". The interval frequency of the sideband is equal to the impact frequency that causes the modulation.

[0058] Currently, the fault detection of fracturing pumps generally relies on passive processing methods after the fault occurs, i.e. when the equipment has a serious fault and causes unplanned shutdown, the fault points need to be checked one by one by disassembling the fracturing pump after shutdown. This method can only diagnose and repair after the fault occurs, which not only makes the troubleshooting process tedious and time-consuming, but also significantly extends the downtime, seriously affecting the continuity and efficiency of oil and gas production operations.

[0059] Based on this, the embodiment of the present application provides a fracturing pump fault detection method, which acquires a multi-source data sequence in the running process of the fracturing pump, that is, acquires a discharge valve vibration data sequence, a suction valve vibration data sequence, a cross pin vibration data sequence, a pressure data sequence, a stroke periodic pulse sequence and a gear information sequence, then performs modal feature decomposition on the acquired discharge valve vibration data sequence, suction valve vibration data sequence and cross pin vibration data sequence, extracts an intrinsic modal signal, then performs signal reconstruction, obtains target discharge valve vibration data sequence, target suction valve vibration data sequence and target cross pin vibration data sequence after eliminating noise without fixed phase, mechanical source and repetition rule, and finally inputs into a pre-trained fault detection model for analysis and processing, so as to accurately judge the fault probability of the fracturing pump in the running process of the fracturing pump. The method can complete non-invasive online monitoring and fault diagnosis in the running process of the equipment, without shutdown and disassembly, significantly reducing the unplanned downtime, and effectively improving the continuity and overall mining efficiency of oil and gas production operations.

[0060] Firstly, a fracturing pump fault detection method provided by the embodiment of the present application is introduced.

[0061] Figure 1 A flowchart of a fracturing pump fault detection method provided by an embodiment of the present application is shown, as shown in Figure 1 The method comprises the following steps:

[0062] S101: Acquire a multi-source data sequence in the running process of the fracturing pump, and the multi-source data sequence comprises a discharge valve vibration data sequence, a suction valve vibration data sequence, a cross pin vibration data sequence, a pressure data sequence, a stroke periodic pulse sequence and a gear information sequence.

[0063] In the embodiment of the present application, the multi-source data sequence in the running process of the fracturing pump is acquired, multi-source information is provided for subsequent fault analysis, the limitation of single data source analysis is overcome, and the comprehensiveness of fault analysis is improved.

[0064] In one example, a specific implementation manner of acquiring the multi-source data sequence in the running process of the fracturing pump is as follows:

[0065] Acquire the discharge valve vibration data sequence collected by the vertical direction vibration sensor installed on each cylinder of the fracturing pump; acquire the suction valve vibration data sequence collected by the horizontal direction vibration sensor installed on each cylinder of the fracturing pump; acquire the cross pin vibration data sequence collected by the vibration sensor installed in the vertical direction at the cross head guide rail; acquire the pressure data sequence collected by the pressure sensor installed at the outlet pipeline interface of the discharge end of the fracturing pump; acquire the stroke periodic pulse sequence collected by the Hall sensor installed at the crankcase of the fracturing pump; and acquire the gear information sequence from the programmable logic controller of the fracturing truck.

[0066] In the embodiment of the present application, the vertical vibration sensor and the horizontal vibration sensor are respectively installed on each cylinder, the vertical vibration sensor is installed along the direction perpendicular to the movement direction of the plunger of the fracturing pump, and is used to collect the discharge valve vibration data sequence; the horizontal vibration sensor is installed along the movement direction of the plunger of the fracturing pump, and is used to collect the suction valve vibration data sequence, so that the comprehensive space and direction coverage of the pump head and the cylinder cover area is realized, and it is ensured that the fault impact signal from any position and any direction of the area can be effectively collected, and the blind area of the fracturing pump fluid end monitoring is completely eliminated. And by acquiring the cross pin vibration data sequence, the health state of the cross pin is monitored, the overall monitoring from the fracturing pump fluid end to the fracturing pump power end is realized, and the comprehensiveness of fault diagnosis is improved. And by combining the monitoring of the power end, the chain fault caused by the abnormal power end can be found earlier, the transmission and expansion of the fault in the system are avoided, and a more comprehensive decision basis is provided for predictive maintenance.

[0067] S102: decompose the discharge valve vibration data sequence, the suction valve vibration data sequence and the cross pin vibration data sequence to obtain a plurality of discharge valve intrinsic modal functions corresponding to the discharge valve vibration data sequence, a plurality of suction valve intrinsic modal functions corresponding to the suction valve vibration data sequence and a plurality of cross pin intrinsic modal functions corresponding to the cross pin vibration data sequence.

[0068] In the embodiment of the present application, the discharge valve vibration data sequence is decomposed to obtain a plurality of discharge valve intrinsic modal functions, the suction valve vibration data sequence is decomposed to obtain a plurality of suction valve intrinsic modal functions, and the cross pin vibration data sequence is decomposed to obtain a plurality of cross pin intrinsic modal functions. In an example, the vibration data sequence can be decomposed into a plurality of intrinsic modal functions by using a variational mode decomposition model (VMD) or an empirical mode decomposition model (EMD).

[0069] In an example, in order to extract more accurate intrinsic modal functions, a specific implementation of step S102 is as follows:

[0070] The number of decomposition layers of the discharge valve vibration data sequence, the suction valve vibration data sequence and the cross pin vibration data sequence is adaptively optimized to obtain the number of decomposition layers corresponding to the discharge valve vibration data sequence, the suction valve vibration data sequence and the cross pin vibration data sequence respectively.

[0071] In the embodiment of the present application, the number of decomposition layers of the exhaust valve vibration data sequence, the suction valve vibration data sequence and the cross pin vibration data sequence is adaptively optimized to obtain the number of decomposition layers corresponding to the exhaust valve vibration data sequence, the suction valve vibration data sequence and the cross pin vibration data sequence. In an example, the number of decomposition layers of the vibration data sequence can be adaptively optimized based on the Aquila Optimizer (AO) algorithm.

[0072] The exhaust valve vibration data sequence, the suction valve vibration data sequence and the cross pin vibration data sequence are decomposed based on the number of decomposition layers corresponding to the exhaust valve vibration data sequence, the suction valve vibration data sequence and the cross pin vibration data sequence to obtain a plurality of exhaust valve intrinsic modal functions, a plurality of suction valve intrinsic modal functions and a plurality of cross pin intrinsic modal functions.

[0073] In the embodiment of the present application, the exhaust valve vibration data sequence, the suction valve vibration data sequence and the cross pin vibration data sequence are decomposed based on the number of decomposition layers corresponding to the exhaust valve vibration data sequence, the suction valve vibration data sequence and the cross pin vibration data sequence. In an example, the VMD algorithm can be used to decompose the vibration data sequence to obtain a plurality of exhaust valve intrinsic modal functions, a plurality of suction valve intrinsic modal functions and a plurality of cross pin intrinsic modal functions.

[0074] In the embodiment of the present application, the VMD model can overcome the end effect and modal aliasing problem of the EMD model, and can extract more accurate intrinsic modal functions. However, the decomposition effect of the VMD model depends on the preset value of its parameters. Therefore, the AO algorithm is used to adaptively optimize the number of decomposition layers of the VMD, and the optimal decomposition state is adaptively adjusted according to the characteristics of the input vibration data sequence, thereby avoiding the subjectivity of manually setting parameters, and achieving more accurate signal decomposition.

[0075] S103: Select target exhaust valve intrinsic modal functions, target suction valve intrinsic modal functions and target cross pin intrinsic modal functions from the plurality of exhaust valve intrinsic modal functions, the plurality of suction valve intrinsic modal functions and the plurality of cross pin intrinsic modal functions, respectively, wherein the correlation between the target exhaust valve intrinsic modal functions, the target suction valve intrinsic modal functions and the target cross pin intrinsic modal functions and the vibration data sequence corresponding thereto is greater than or equal to a preset threshold.

[0076] In the embodiments of the present application, the correlation degrees of the plurality of exhaust valve eigenmodal functions and the exhaust valve vibration data sequence are respectively calculated, and eigenmodal functions with a correlation degree greater than or equal to a preset threshold are selected as target exhaust valve eigenmodal functions; the correlation degrees of the plurality of suction valve eigenmodal functions and the suction valve vibration data sequence are respectively calculated, and eigenmodal functions with a correlation degree greater than or equal to a preset threshold are selected as target suction valve eigenmodal functions; the correlation degrees of the plurality of cross pin eigenmodal functions and the cross pin vibration data sequence are respectively calculated, and eigenmodal functions with a correlation degree greater than or equal to a preset threshold are selected as target cross pin eigenmodal functions. By selecting eigenmodal functions with high correlation with the original vibration data sequence, noise signals without fixed phase, mechanical source or repeated rules can be eliminated, thereby realizing high-quality signal-noise separation.

[0077] S104: The target exhaust valve eigenmodal functions are reconstructed into target exhaust valve vibration data sequences, the target suction valve eigenmodal functions are reconstructed into target suction valve vibration data sequences, and the target cross pin eigenmodal functions are reconstructed into target cross pin vibration data sequences.

[0078] In the embodiments of the present application, the target exhaust valve eigenmodal functions, the target suction valve eigenmodal functions and the target cross pin eigenmodal functions are respectively reconstructed into target exhaust valve vibration data sequences, target suction valve vibration data sequences and target cross pin vibration data sequences, which can eliminate the interference of irrelevant noise to the greatest extent, highlight the original fault pulse characteristics, and provide more accurate vibration feature data for subsequent fracturing pump fault detection models.

[0079] S105: The target exhaust valve vibration data sequences, the target suction valve vibration data sequences, the target cross pin vibration data sequences, the pressure data sequences, the stroke periodic pulse sequences and the gear information sequences are input into the fracturing pump fault detection model, and fault detection is performed through the fracturing pump fault detection model to determine the fault type of the fracturing pump.

[0080] In the embodiments of the present application, the acquired multi-source data sequences are input into the fracturing pump fault detection model, and fault detection is performed through the fracturing pump fault detection model to determine the fault types of the fracturing pump. This non-invasive diagnostic method does not need to disassemble the fracturing pump equipment, which not only saves maintenance costs, but also avoids the secondary fault risk that may be introduced due to disassembly. In one example, the faults of the fracturing pump include at least one of the following: valve body damage, valve seat cracking, plunger wear, pump head cracking and cross pin loosening.

[0081] In a preferred embodiment of the present application, the fracturing pump fault detection model adopts an Integrated-Convolutional Neural Network (ICCNN). The ICCNN can directly process vectorized inputs, thereby supporting embedding the stroke periodic pulse sequence, pressure data sequence, and gear information sequence into the network structure in the form of weights and biases. Since the changes in these parameters directly cause changes in the plunger stroke frequency, phase, and load, and further cause changes in the position and frequency characteristics of the impact components in the vibration signal. Therefore, in the feature extraction stage, the vibration features are dynamically corrected and unified using these parameters to eliminate the influence of different working conditions on the vibration data, so that the model can focus on the features related to the nature of the fault, thereby improving the generalization ability and diagnostic accuracy of the model under different operating conditions.

[0082] In addition, compared with traditional convolutional neural networks, ICCNN has stronger non-linear mapping and feature integration capabilities at the neuron level, so it can use a lighter network structure (e.g., 3 to 4 hidden layers) to achieve the performance of traditional convolutional neural networks (e.g., 6 to 8 hidden layers). This lightweight design significantly reduces the number of parameters of the fracturing pump fault detection model, reduces the dependence on large-scale labeled data, effectively alleviates the training difficulties caused by the scarcity of fault samples in industrial environments, and makes it easier to deploy to edge devices with limited computing resources, thereby further improving the practicality and engineering applicability of the method.

[0083] It should be noted that the fracturing pump fault detection method provided by the embodiments of the present application needs to use a pre-trained fracturing pump fault detection model to process the multi-source data sequence, so before using the fracturing pump fault detection model for fault detection, the fracturing pump fault detection model needs to be trained:

[0084] A multi-source data sequence sample set is obtained, the multi-source data sequence sample set including a plurality of multi-source data sequence samples and corresponding fault category labels, each multi-source data sequence sample including: a discharge valve vibration data sequence sample, a suction valve vibration data sequence sample, a cross pin vibration data sequence sample, a pressure data sequence sample, a stroke periodic pulse sequence sample, and a gear information sequence sample.

[0085] In the embodiments of the present application, before training the fracturing pump fault detection model, a multi-source data sequence sample set needs to be constructed, which includes a plurality of multi-source data sequence samples and corresponding fault category labels.

[0086] In one example, each multi-source data sequence sample corresponds to a fault category label, which can be obtained by obtaining the maintenance record, obtaining the corresponding fault category label and the multi-source data sequence sample. Specifically, based on the historical maintenance record of the fracturing pump, the fault occurrence time and the corresponding fault category are determined; for each fault event, the vibration data, pressure data, stroke periodic pulse and gear information collected within a period of time before the fault occurs are extracted to form a multi-source data sequence sample, and the fault category is taken as the label thereof. The above process is repeated to accumulate data samples under multiple fault events, and finally a multi-source data sequence sample set is formed.

[0087] Based on the multi-source data sequence sample set, the fracturing pump fault detection model to be trained is trained to obtain a trained fracturing pump fault detection model.

[0088] In the embodiments of the present application, based on the constructed multi-source data sequence sample set, the fracturing pump fault detection model to be trained is trained until the loss function converges, and a trained fracturing pump fault detection model is obtained. Since the selected multi-source data sequence is the data before the fault occurs and the characteristics are developing, the fracturing pump fault detection model trained based on the multi-source data sequence can mine the early signs of fault formation, thereby predicting the possibility of fault occurrence in advance and further avoiding the occurrence of faults.

[0089] In one example of the present application, the fracturing pump fault detection model can identify multiple typical faults, and each type of fault presents the following distinguishable characteristic patterns in multi-source data. These characteristic patterns provide a basis for model learning:

[0090] Valve body damage fault: in the time domain, the vibration data near the piston reversal point appears a secondary impact due to the lag impact of the valve body closing on the valve seat, and in the frequency domain, the high-frequency region presents a bandwidth energy distribution; at the same time, the pressure data shows a large fluctuation; the stroke periodic pulse reflects the phase advance or lag phenomenon caused by the late closing or early opening of the valve body.

[0091] Valve seat cracking fault: the vibration data in the time domain presents a relatively wide pulse, which is wider and has a smaller amplitude than the pulse caused by the valve body damage, and in the frequency domain, a narrow-band peak and a modulation frequency conversion phenomenon appear; the pressure data fluctuates slightly; the stroke periodic pulse sequence forms a fixed lead angle, and the lead angle is consistent with the pressure data fluctuation valley time.

[0092] Piston wear fault: the vibration data in the time domain presents a pulse waveform with a large amplitude, and the vibration peak value gradually increases, and in the frequency domain, the amplitude of the piston fundamental frequency and its harmonics gradually increases, and the wide frequency noise and variable frequency band energy are lifted; the pressure data presents a gradually decreasing trend; the stroke periodic pulse appears random fluctuation, and the fluctuation degree increases with the aggravation of wear.

[0093] Pump head crack fault: the vibration data shows a sharp impact pulse in the time domain once per stroke, and in the frequency domain, the pump harmonic is significantly increased, and the high-frequency broadband energy is increased; the pressure data shows a transient spike; the stroke periodic pulse appears random jitter, and the impact amplitude gradually increases.

[0094] Cross pin loose fault: the vibration data shows a high-amplitude impact in the time domain when the crank angle turns through a preset fixed threshold phase angle, and in the frequency domain, the low-frequency pulse band and low-frequency harmonic energy are significantly lifted; the pressure data shows a downward trend, and in one example, the preset fixed threshold phase angle is 40° to 50°.

[0095] In the embodiments of the present application, a multi-source data sequence sample set is obtained, the multi-source data sequence is a data segment before the fault occurs, and contains early weak fault features, so that the fracturing pump fault detection model can learn the early fault features. Therefore, based on the multi-source data sequence sample set, the fracturing pump fault detection model to be trained is trained to obtain the trained fracturing pump fault detection model, which can effectively identify the early fault features, thereby realizing early prediction of the fault, providing a key time window for active maintenance, helping to prevent the fault from happening, and ensuring continuous and stable operation of the fracturing pump.

[0096] In the embodiment of the present application, a plurality of source data sequences in the operation process of the fracturing pump are acquired, and the plurality of source data sequences include: a discharge valve vibration data sequence, a suction valve vibration data sequence, a cross pin vibration data sequence, a pressure data sequence, a stroke periodic pulse sequence, and a gear information sequence. Then, the discharge valve vibration data sequence, the suction valve vibration data sequence, and the cross pin vibration data sequence are decomposed to obtain a plurality of discharge valve eigenmodal functions corresponding to the discharge valve vibration data sequence, a plurality of suction valve eigenmodal functions corresponding to the suction valve vibration data sequence, and a plurality of cross pin eigenmodal functions corresponding to the cross pin vibration data sequence. Then, a target discharge valve eigenmodal function, a target suction valve eigenmodal function, and a target cross pin eigenmodal function are selected from the plurality of discharge valve eigenmodal functions, the plurality of suction valve eigenmodal functions, and the plurality of cross pin eigenmodal functions, respectively, wherein the correlation degrees of the target discharge valve eigenmodal function, the target suction valve eigenmodal function, and the target cross pin eigenmodal function with the vibration data sequences corresponding thereto are greater than or equal to a preset threshold. Then, the target discharge valve eigenmodal function, the target suction valve eigenmodal function, and the target cross pin eigenmodal function are reconstructed into a target discharge valve vibration data sequence, a target suction valve vibration data sequence, and a target cross pin vibration data sequence, respectively. Finally, the target discharge valve vibration data sequence, the target suction valve vibration data sequence, the target cross pin vibration data sequence, the pressure data sequence, the stroke periodic pulse sequence, and the gear information sequence are input into a fracturing pump fault detection model to perform fault detection through the fracturing pump fault detection model, so as to determine the fault type of the fracturing pump. The fault type of the fracturing pump is one of a valve body damage fault, a valve seat cracking fault, a plunger wear fault, a pump head cracking fault, and a cross pin loosening fault. Compared with the prior art which can only determine the fault type by disassembling after the fracturing pump fails, resulting in serious lag in maintenance, too long downtime, and thus affecting the oil and gas exploitation efficiency, the present application can realize online monitoring and early fault warning of the operation state of the fracturing pump by acquiring the plurality of source data sequences in real time, then performing modal characteristic decomposition on the acquired discharge valve vibration data sequence, suction valve vibration data sequence, and cross pin vibration data sequence to extract eigenmodal signals, and then performing signal reconstruction to eliminate noise signals without fixed phase, mechanical source, and repetition rule. Then, the target discharge valve vibration data sequence, the target suction valve vibration data sequence, the target cross pin vibration data sequence, the pressure data sequence, the stroke periodic pulse sequence, and the gear information sequence are input into the fracturing pump fault detection model, and the plurality of source data sequences are analyzed based on the fracturing pump fault detection model. The fracturing pump fault detection model can identify abnormal phenomena before the fault occurs, so as to realize predictive maintenance of the fracturing pump, greatly reduce the unplanned downtime, and improve the overall efficiency of oil and gas resource exploitation.

[0097] Figure 2A flowchart of a fracturing pump fault detection method provided by another embodiment of the present application is shown, based on the above-mentioned Figure 1 On the basis of the embodiment shown above, one specific implementation of step S102 is:

[0098] S201: Feature fusion is performed on the pressure data sequence, the stroke periodic pulse sequence, and the gear information sequence, and the target discharge valve vibration data sequence, the target suction valve vibration data sequence, and the target cross pin vibration data sequence, to obtain a neuron feature vector.

[0099] In the embodiment of the present application, feature fusion is respectively performed on the pressure data sequence, the stroke periodic pulse sequence, and the gear information sequence, and the target discharge valve vibration data sequence, the target suction valve vibration data sequence, and the target cross pin vibration data sequence, to obtain a neuron feature vector.

[0100] It should be noted that the sequence length of the pressure data sequence, the stroke periodic pulse sequence, and the gear information sequence is 1.

[0101] S202: Feature learning is performed based on the neuron feature vector, to obtain a fault type of the fracturing pump.

[0102] In the embodiment of the present application, feature learning is performed based on the neuron feature vector, the neuron feature vector is subjected to multi-level and nonlinear transformation and learning by a fracturing pump fault detection model, complex features related to faults are adaptively extracted, and finally the features are input to a Softmax classifier, and a fault type of the fracturing pump is calculated and output based on the Softmax classifier.

[0103] In the embodiment of the present application, a fault of the fracturing pump will cause low working efficiency of the fracturing pump, and the low working efficiency of the fracturing pump will be reflected as changes in the pressure data and the stroke periodic pulse sequence, and since the gear information determines the speed relationship of the engine-crankshaft-plunger, the gear information is also an important working condition parameter for judging the equipment state. Since the fault of the fracturing pump is a systematic problem coupled by multiple factors, when diagnosing the fault, comprehensive analysis needs to be performed on multiple sources of data. Therefore, in the present application, the pressure data, the stroke periodic pulse sequence, and the gear information are subjected to feature fusion with vibration data at the feature layer, joint judgment of various fault features is achieved, the running state information of the fracturing pump can be more comprehensively captured, and the precision of the accurate judgment of complex faults can be significantly improved.

[0104] Figure 3 A flowchart of a fracturing pump fault detection method provided by another embodiment of the present application is shown, based on the above-mentioned Figure 2 On the basis of the embodiment shown above, one specific implementation of step S201 is:

[0105] S301: set the stroke periodic pulse sequence as the weight of the neuron in the fault detection model, and set the pressure data sequence and the gear information sequence as the bias of the neuron in the fault detection model.

[0106] In the embodiment of the application, the stroke periodic pulse sequence is set as the weight of the neuron in the fault detection model. The pressure data sequence and the gear information sequence are merged into a bias vector, and the bias vector is set as the bias parameter of the neuron in the fault detection model.

[0107] In one example, when the pressure data sequence and the gear information sequence are merged into a joint bias vector, the two need to be encoded first. Although the gear information is represented by numbers, the values do not represent a linear multiple relationship (for example, 4 gears do not equal 4 times 1 gear), and the essence is closer to a category label. Therefore, by encoding the gear information (for example, using one-hot encoding), it can be converted into a binary vector form, thereby avoiding the fault detection model from mistakenly introducing a linear assumption in the numerical sense, enhancing its representation ability for category attributes, and ensuring the rationality and effectiveness of the subsequent fusion process.

[0108] S302: perform feature fusion on the target discharge valve vibration data sequence, the target suction valve vibration data sequence, and the target cross pin vibration data sequence based on the weight and the bias, to obtain a neuron feature vector.

[0109] In the embodiment of the application, the preprocessed vibration data sequence is input into the neuron, and the feature fusion is performed by the neuron with the vibration data sequence as the input, the stroke periodic pulse sequence as the weight, and the pressure data sequence and the gear information sequence as the bias, to obtain the neuron feature vector.

[0110] In the embodiment of the application, in order to improve the prediction accuracy of the fracturing pump fault detection model, the stroke periodic pulse sequence is set as the weight parameter of the neuron in the fault detection model. At the same time, the pressure data sequence and the gear information sequence are merged into a bias vector after encoding, and are set as the bias parameter of the neuron. Subsequently, the vibration data sequence is fused based on the obtained weight and bias to generate a neuron feature vector, realizing feature layer fusion guided by physical meaning. Not only does it effectively utilize the physical attributes of multi-source data, but also avoids the loss of information in simple splicing of heterogeneous information, significantly improving the learning efficiency of the fracturing pump fault detection model and the interpretability of the output results.

[0111] Based on the same inventive concept, the embodiment of the application also provides a fracturing pump fault detection device. The specific combination Figure 4 A fracturing pump fault detection device provided by the embodiment of the application is described in detail.

[0112] Figure 4A structural schematic diagram of a fracturing pump fault detection device provided by an embodiment of the present application is shown.

[0113] As shown in the figure, the fracturing pump fault detection 400 can include an acquisition module 401, a decomposition module 402, a selection module 403, a reconstruction module 404, and a detection module 405. Figure 4

[0114] The acquisition module 401 is configured to acquire a multi-source data sequence in a fracturing pump operation process, where the multi-source data sequence includes an exhaust valve vibration data sequence, a suction valve vibration data sequence, a cross pin vibration data sequence, a pressure data sequence, a stroke periodic pulse sequence, and a gear information sequence.

[0115] The decomposition module 402 is configured to decompose the exhaust valve vibration data sequence, the suction valve vibration data sequence, and the cross pin vibration data sequence to obtain a plurality of exhaust valve eigenmode functions corresponding to the exhaust valve vibration data sequence, a plurality of suction valve eigenmode functions corresponding to the suction valve vibration data sequence, and a plurality of cross pin eigenmode functions corresponding to the cross pin vibration data sequence.

[0116] The selection module 403 is configured to select a target exhaust valve eigenmode function, a target suction valve eigenmode function, and a target cross pin eigenmode function from the plurality of exhaust valve eigenmode functions, the plurality of suction valve eigenmode functions, and the plurality of cross pin eigenmode functions, respectively, where the target exhaust valve eigenmode function, the target suction valve eigenmode function, and the target cross pin eigenmode function have a correlation degree greater than or equal to a preset threshold with the vibration data sequence corresponding thereto.

[0117] The reconstruction module 404 is configured to reconstruct the target exhaust valve eigenmode function into a target exhaust valve vibration data sequence, reconstruct the target suction valve eigenmode function into a target suction valve vibration data sequence, and reconstruct the target cross pin eigenmode function into a target cross pin vibration data sequence.

[0118] The detection module 405 is configured to input the target exhaust valve vibration data sequence, the target suction valve vibration data sequence, the target cross pin vibration data sequence, the pressure data sequence, the stroke periodic pulse sequence, and the gear information sequence into a fracturing pump fault detection model, perform fault detection through the fracturing pump fault detection model, and determine a fault type of the fracturing pump, where the fault type of the fracturing pump is one of a valve body damage fault, a valve seat cracking fault, a plunger wear fault, a pump head cracking fault, and a cross pin loosening fault.

[0119] In one example, the detection module 405 includes:

[0120] ​The feature fusion module is configured to perform feature fusion on the pressure data sequence, the stroke periodic pulse sequence, the gear information sequence, the target discharge valve vibration data sequence, the target suction valve vibration data sequence, and the target cross pin vibration data sequence, to obtain a neuron feature vector.

[0121] The feature learning module is configured to perform feature learning based on the neuron feature vector, to obtain a fault type of the fracturing pump.

[0122] In one example, the detection module 405 includes:

[0123] The processing module is configured to set the stroke periodic pulse sequence as a weight of a neuron in a fault detection model, and set the pressure data sequence and the gear information sequence as a bias of the neuron in the fault detection model.

[0124] The feature fusion module is further configured to perform feature fusion on the target discharge valve vibration data sequence, the target suction valve vibration data sequence, and the target cross pin vibration data sequence based on the weight and the bias, to obtain the neuron feature vector.

[0125] In one example, the decomposition module 402 includes:

[0126] The optimization processing module is configured to perform adaptive optimization processing on the number of decomposition layers of the discharge valve vibration data sequence, the suction valve vibration data sequence, and the cross pin vibration data sequence, to obtain the number of decomposition layers corresponding to the discharge valve vibration data sequence, the suction valve vibration data sequence, and the cross pin vibration data sequence, respectively.

[0127] The decomposition submodule is configured to decompose the discharge valve vibration data sequence, the suction valve vibration data sequence, and the cross pin vibration data sequence based on the number of decomposition layers corresponding to the discharge valve vibration data sequence, the suction valve vibration data sequence, and the cross pin vibration data sequence, respectively, to obtain a plurality of discharge valve eigenmode functions, a plurality of suction valve eigenmode functions, and a plurality of cross pin eigenmode functions.

[0128] In one example, the fracturing pump fault detection 400 can further include:

[0129] The first acquisition module is configured to acquire a multi-source data sequence sample set, the multi-source data sequence sample set including a plurality of multi-source data sequence samples and corresponding fault category labels, each multi-source data sequence sample including a discharge valve vibration data sequence sample, a suction valve vibration data sequence sample, a cross pin vibration data sequence sample, a pressure data sequence sample, a stroke periodic pulse sequence sample, and a gear information sequence sample.

[0130] The training module is configured to perform model training on a fracturing pump fault detection model to be trained based on the multi-source data sequence sample set, to obtain a trained fracturing pump fault detection model.

[0131] In one example, the acquisition module 401 comprises:

[0132] an acquisition sub-module, configured to acquire a discharge valve vibration data sequence collected by a vertical direction vibration sensor installed on each cylinder of the fracturing pump;

[0133] the acquisition sub-module is further configured to acquire a suction valve vibration data sequence collected by a horizontal direction vibration sensor installed on each cylinder of the fracturing pump;

[0134] the acquisition sub-module is further configured to acquire a cross pin vibration data sequence collected by a vibration sensor installed in a vertical direction at a crosshead guide rail;

[0135] the acquisition sub-module is further configured to acquire a pressure data sequence collected by a pressure sensor installed at an outlet pipeline interface of a discharge end of the fracturing pump;

[0136] the acquisition sub-module is further configured to acquire a stroke periodic pulse sequence collected by a Hall sensor installed at a crankcase of the fracturing pump;

[0137] the acquisition sub-module is further configured to acquire a gear information sequence from a programmable logic controller of the fracturing truck.

[0138] The various modules in the fracturing pump fault detection device provided by the embodiments of the present application can implement the method steps of any of the embodiments and achieve the corresponding technical effects, and for brevity, will not be described here. Figures 1 to 3 The method steps of any of the embodiments can be implemented by the various modules in the fracturing pump fault detection device provided by the embodiments of the present application and achieve the corresponding technical effects, and for brevity, will not be described here.

[0139] Figure 5 A hardware structure schematic diagram of an electronic device provided by an embodiment of the present application is shown.

[0140] The electronic device can include a processor 501 and a memory 502 having computer program instructions stored therein.

[0141] Specifically, the processor 501 can include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or can be configured to implement one or more integrated circuits of the embodiments of the present application.

[0142] The memory 502 can include mass storage for data or instructions. As an example and not by way of limitation, the memory 502 can include a hard disk drive (HDD), a floppy disk drive, flash memory, an optical disc (e.g., a compact disc (CD) or a digital versatile disc (DVD)), a solid-state drive (SSD), a USB drive, or a combination of two or more of these. In one example, the memory 502 can include removable or non-removable (or fixed) media, where the memory 502 is a nonvolatile solid-state memory.

[0143] In one example, the memory 502 can be a read-only memory (ROM). In one example, the ROM can be a mask programmed ROM, a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable PROM (EEPROM), an electrically alterable ROM (EAROM), or a flash memory, or a combination of two or more of these.

[0144] The memory 502 can include read-only memory (ROM), random-access memory (RAM), magnetic disk storage media devices, optical storage media devices, flash memory devices, electrical, optical, or other physical / tangible memory storage devices. Thus, in general, the memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., a memory device) encoded with software that, when executed (by one or more processors), is operable to perform operations described with reference to the methods according to an aspect of the present disclosure.

[0145] The processor 501 implements the fracturing pump fault detection method in one of the above-described embodiments by reading and executing computer program instructions stored in the memory 502.

[0146] In one example, the electronic device can further include a communication interface 503 and a bus 504. Wherein, as shown, the processor 501, the memory 502, the communication interface 503 are connected through the bus 504 and complete the communication between each other. Figure 5

[0147] The communication interface 503 is mainly used to realize the communication between each module, device, unit and / or equipment in the embodiments of the present application.

[0148] ​Bus 504 includes a hardware, software, or both that couples components of the online data traffic metering device to each other. As an example but not a limitation, the bus can include an Accelerated Graphics Port (AGP) or other graphics bus, an Extended Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a Hyper Transport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an InfiniBand interconnect, a Low Pin Count (LPC) bus, a memory bus, a Micro Channel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or another suitable bus or a combination of two or more of these. Where suitable, bus 504 can include one or more buses. Although specific busses are described and illustrated in this embodiment, this application contemplates any suitable bus or interconnect.

[0149] In addition, in combination with the fracturing pump fault detection method in the above embodiment, the embodiment of the application can provide a computer storage medium to implement. The computer storage medium has computer program instructions stored thereon; the computer program instructions are executed by a processor to implement any of the fracturing pump fault detection methods in the above embodiments.

[0150] The embodiment of the application also provides a computer program product, comprising a computer program, the computer program is executed by a processor to implement any of the fracturing pump fault detection methods in the above embodiments.

[0151] It needs to be clear that the application is not limited to the specific configurations and processes described above and shown in the drawings. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the application is not limited to the specific steps described and shown, and those skilled in the art can make various changes, modifications and additions, or change the order between steps, after understanding the spirit of the application.

[0152] The functions indicated in the structural block diagrams above can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, and the like. When implemented in software, the elements of the present application are program or code segments that are used to perform the required tasks. The program or code segments can be stored in a machine-readable medium or transmitted through a data signal carried in a carrier wave over a transmission medium or communication link. A "machine-readable medium" includes any medium that can store or transport information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, read-only memories (ROMs), flash memories, erasable read-only memories (EROMs), floppy disks, compact discs (CD-ROMs), optical disks, hard disks, fiber-optic media, radio frequency (RF) links, and the like. The code segments can be downloaded via computer networks such as the Internet, intranets, and the like.

[0153] It is also important to note that the examples mentioned in the present application describe some methods or systems based on a series of steps or devices. However, the present application is not limited to the order of the above steps, that is, the steps can be performed in the order mentioned in the examples, or in an order different from the examples, or several steps can be performed simultaneously.

[0154] The computer program instructions can also be loaded onto a computer, other programmable data processing apparatus, or other processing devices to cause a series of operational steps to be performed on the computer, other programmable apparatus or other processing devices to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide processes for implementing the functions / acts specified in the flowchart and / or block diagram block or blocks. These computer program instructions can also be stored in a computer readable medium that can direct a computer, other programmable data processing apparatus, or other processing devices to operate in a particular manner, such that the instructions stored in the computer readable medium produce an article of manufacture including instructions which implement the function / act specified in the flowchart and / or block diagram block or blocks. The computer program instructions can also be loaded onto a computer, other programmable data processing apparatus, or other processing devices to cause a series of operational steps to be performed on the computer, other programmable apparatus or other processing devices to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide processes for implementing the functions / acts specified in the flowchart and / or block diagram block or blocks. These computer program instructions can also be stored in a computer readable medium that can direct a computer, other programmable data processing apparatus, or other processing devices to operate in a particular manner, such that the instructions stored in the computer readable medium produce an article of manufacture including instructions which implement the function / act specified in the flowchart and / or block diagram block or blocks.

[0155] The above is merely a specific implementation of the present application. As can be clearly understood by a person skilled in the art from the above description, for the convenience and brevity of description, the specific working process of the system, module and unit described above can refer to the corresponding process in the foregoing method embodiments, which will not be described herein again. It should be understood that the protection scope of the present application is not limited in this way, and any person skilled in the art can easily think of various equivalent modifications or replacements within the technical range disclosed in the present application, and these modifications or replacements should be covered in the protection scope of the present application.

Claims

1. A method of fracture pump fault detection, the method comprising: The method comprises the following steps: obtaining a plurality of source data sequences in the operation process of a fracturing pump, the plurality of source data sequences comprising: a discharge valve vibration data sequence, a suction valve vibration data sequence, a cross pin vibration data sequence, a pressure data sequence, a stroke periodic pulse sequence, and a gear information sequence; decomposing the discharge valve vibration data sequence, the suction valve vibration data sequence, and the cross pin vibration data sequence to obtain a plurality of discharge valve eigenmodal functions corresponding to the discharge valve vibration data sequence, a plurality of suction valve eigenmodal functions corresponding to the suction valve vibration data sequence, and a plurality of cross pin eigenmodal functions corresponding to the cross pin vibration data sequence; selecting a target discharge valve eigenmodal function, a target suction valve eigenmodal function, and a target cross pin eigenmodal function from the plurality of discharge valve eigenmodal functions, the plurality of suction valve eigenmodal functions, and the plurality of cross pin eigenmodal functions respectively, wherein the correlation degrees of the target discharge valve eigenmodal function, the target suction valve eigenmodal function, and the target cross pin eigenmodal function with the vibration data sequences corresponding thereto are greater than or equal to a preset threshold; reconstructing the target discharge valve eigenmodal function into a target discharge valve vibration data sequence, reconstructing the target suction valve eigenmodal function into a target suction valve vibration data sequence, and reconstructing the target cross pin eigenmodal function into a target cross pin vibration data sequence; setting the stroke periodic pulse sequence as the weight of a neuron in a fault detection model, and setting the pressure data sequence and the gear information sequence as the bias of the neuron in the fault detection model; performing feature fusion on the target discharge valve vibration data sequence, the target suction valve vibration data sequence, and the target cross pin vibration data sequence based on the weight and the bias to obtain a neuron feature vector; performing feature learning based on the neuron feature vector to obtain a fault type of the fracturing pump; the fault type of the fracturing pump is one of a valve body damage fault, a valve seat cracking fault, a plunger wear fault, a pump head cracking fault, and a cross pin loosening fault.

2. The method of claim 1, wherein, inputting the target discharge valve vibration data sequence, the target suction valve vibration data sequence, the target cross pin vibration data sequence, the pressure data sequence, the stroke periodic pulse sequence, and the gear information sequence into a fracturing pump fault detection model, performing fault detection through the fracturing pump fault detection model, and determining the fault type of the fracturing pump, comprising: performing feature fusion on the pressure data sequence, the stroke periodic pulse sequence, and the gear information sequence, and the target discharge valve vibration data sequence, the target suction valve vibration data sequence, and the target cross pin vibration data sequence to obtain the neuron feature vector; performing feature learning based on the neuron feature vector to obtain the fault type of the fracturing pump.

3. The method of claim 1, wherein, The decomposing the exhaust valve vibration data sequence, the suction valve vibration data sequence, and the cross pin vibration data sequence obtains a plurality of exhaust valve eigenmode functions corresponding to the exhaust valve vibration data sequence, a plurality of suction valve eigenmode functions corresponding to the suction valve vibration data sequence, and a plurality of cross pin eigenmode functions corresponding to the cross pin vibration data sequence, and the decomposing the exhaust valve vibration data sequence, the suction valve vibration data sequence, and the cross pin vibration data sequence comprises: The adaptive optimization processing is performed on the decomposition layer numbers of the exhaust valve vibration data sequence, the suction valve vibration data sequence, and the cross pin vibration data sequence, to obtain the decomposition layer numbers corresponding to the exhaust valve vibration data sequence, the suction valve vibration data sequence, and the cross pin vibration data sequence respectively; The exhaust valve vibration data sequence, the suction valve vibration data sequence, and the cross pin vibration data sequence are decomposed based on the decomposition layer numbers corresponding to the exhaust valve vibration data sequence, the suction valve vibration data sequence, and the cross pin vibration data sequence respectively, to obtain the plurality of exhaust valve eigenmode functions, the plurality of suction valve eigenmode functions, and the plurality of cross pin eigenmode functions.

4. The method of claim 1, wherein, Before the target exhaust valve vibration data sequence, the target suction valve vibration data sequence, the target cross pin vibration data sequence, the pressure data sequence, the stroke periodic pulse sequence, and the gear information sequence are input into the fracturing pump fault detection model to perform fault detection by the fracturing pump fault detection model to determine the fault type of the fracturing pump, the method further comprises: obtaining a plurality of multi-source data sequence sample sets, each of the plurality of multi-source data sequence sample sets comprising a plurality of multi-source data sequence samples and corresponding fault category labels, each of the plurality of multi-source data sequence samples comprising an exhaust valve vibration data sequence sample, a suction valve vibration data sequence sample, a cross pin vibration data sequence sample, a pressure data sequence sample, a stroke periodic pulse sequence sample, and a gear information sequence sample; based on the plurality of multi-source data sequence sample sets, performing model training on a fracturing pump fault detection model to be trained to obtain a trained fracturing pump fault detection model.

5. The method of claim 1, wherein, The method for obtaining the multi-source data sequence in the operation process of the fracturing pump comprises: obtaining the exhaust valve vibration data sequence collected by a vertical direction vibration sensor installed on each cylinder of the fracturing pump; obtaining the suction valve vibration data sequence collected by a horizontal direction vibration sensor installed on each cylinder of the fracturing pump; obtaining the cross pin vibration data sequence collected by a vibration sensor installed in a vertical direction at a cross head guide rail; obtaining the pressure data sequence collected by a pressure sensor installed at an outlet pipeline interface of an exhaust end of the fracturing pump; obtaining the stroke periodic pulse sequence collected by a Hall sensor installed at a crankcase of the fracturing pump; obtaining the gear information sequence from a programmable logic controller of a fracturing truck.

6. A frac pump fault detection apparatus, characterized by, The method comprises: a obtaining module configured to obtain a multi-source data sequence in an operation process of a fracturing pump, the multi-source data sequence comprising an exhaust valve vibration data sequence, a suction valve vibration data sequence, a cross pin vibration data sequence, a pressure data sequence, a stroke periodic pulse sequence, and a gear information sequence; a decomposition module, configured to decompose the exhaust valve vibration data sequence, the suction valve vibration data sequence, and the cross pin vibration data sequence to obtain a plurality of exhaust valve eigenmode functions corresponding to the exhaust valve vibration data sequence, a plurality of suction valve eigenmode functions corresponding to the suction valve vibration data sequence, and a plurality of cross pin eigenmode functions corresponding to the cross pin vibration data sequence; a selection module, configured to select a target exhaust valve eigenmode function, a target suction valve eigenmode function, and a target cross pin eigenmode function from the plurality of exhaust valve eigenmode functions, the plurality of suction valve eigenmode functions, and the plurality of cross pin eigenmode functions, respectively, wherein the target exhaust valve eigenmode function, the target suction valve eigenmode function, and the target cross pin eigenmode function have a correlation degree with the vibration data sequence corresponding thereto greater than or equal to a preset threshold; a reconstruction module, configured to reconstruct the target exhaust valve eigenmode function into a target exhaust valve vibration data sequence, reconstruct the target suction valve eigenmode function into a target suction valve vibration data sequence, and reconstruct the target cross pin eigenmode function into a target cross pin vibration data sequence; a processing module, configured to set the stroke periodic pulse sequence as a weight of a neuron in a fault detection model, and set the pressure data sequence and the gear information sequence as a bias of the neuron in the fault detection model; a feature fusion module, configured to perform feature fusion on the target exhaust valve vibration data sequence, the target suction valve vibration data sequence, and the target cross pin vibration data sequence based on the weight and the bias to obtain a neuron feature vector; a feature learning module, configured to perform feature learning based on the neuron feature vector to obtain a fault type of the fracturing pump; the fault type of the fracturing pump is one of a valve body damage fault, a valve seat cracking fault, a plunger wear fault, a pump head cracking fault, and a cross pin loosening fault.

7. An electronic device, comprising: The device comprises a processor and a memory storing computer program instructions; The processor executes the computer program instructions to implement the fracturing pump fault detection method in any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, The computer readable storage medium stores computer program instructions, and the computer program instructions are executed by the processor to implement the fracturing pump fault detection method in any one of claims 1-5.

9. A computer program product, characterised in that, The instructions in the computer program product are executed by the processor of the electronic device to cause the electronic device to perform the fracturing pump fault detection method in any one of claims 1-5.

Images