Edge cloud collaborative diagnosis and early warning system for tower type pumping unit cloud platform

Abstract

The present application relates to the technical field of downhole pump fault early warning, in particular to an edge cloud collaborative diagnosis early warning system for a tower type pumping unit cloud platform. The system comprises an edge module and a cloud platform module. The edge module comprises a data acquisition unit for acquiring operation data of the tower type pumping unit and a local diagnosis unit for determining a fault time. The cloud platform unit comprises a remote diagnosis and decision unit for obtaining a fault response value according to the correlation degree of the same kind of operation data and the importance degree of each kind of operation data in the analysis period before and after the fault time; determining a response feature sequence based on the fault response value of the same kind of operation data at the fault time; training a neural network to obtain a pumping unit fault diagnosis model, and then diagnosing and early warning the tower type pumping unit. The present application effectively improves the accuracy of the diagnosis and early warning of the pumping unit fault.

Description

Technical Field

[0001] This invention relates to the field of downhole pump fault early warning technology, specifically to an edge-cloud collaborative diagnostic and early warning system for a tower-type pumping unit cloud platform. Background Technology

[0002] Tower pumping units are indispensable equipment in the petroleum industry, responsible for extracting crude oil from deep wells. However, due to prolonged operation and complex working environments, tower pumping units frequently encounter various malfunctions, leading to decreased production efficiency and increased maintenance costs. Establishing a cloud platform for tower pumping units enables real-time monitoring and fault diagnosis. By monitoring and analyzing various parameters during operation, the cloud platform can promptly detect malfunctions and issue early warnings, improving equipment reliability and stability. Furthermore, it can provide scientific data support for oilfield operation and management, promoting the intelligent and efficient development of oilfield production.

[0003] The overall architecture of the tower pumping unit cloud platform is divided into five layers: the infrastructure layer mainly provides infrastructure services such as computing, storage, and networking; the data layer is responsible for collecting and storing various data from the tower pumping unit; the platform layer mainly provides general components, microservices, and platform services; the application layer is the top layer of the architecture, mainly providing various applications and services, such as alarm and maintenance of the tower pumping unit, big data analysis, etc.; and the presentation layer is the outermost layer of the architecture, mainly providing a visual interface and functional interaction.

[0004] Existing methods diagnose and warn of pumping unit malfunctions by observing equipment characteristic parameters or using empirical rules. These methods primarily include manual visual inspection, infrared thermography, and vibration detection. Manual visual inspection requires highly experienced and skilled operators. Infrared thermography may fail to detect the temperature of critical components inside the pumping unit that are obscured by other structures or located deep within the equipment. Vibration detection methods may misinterpret vibrations caused by external factors such as unstable foundations as abnormal equipment vibrations, resulting in low accuracy in diagnosing and warning of pumping unit malfunctions. Summary of the Invention

[0005] To address the technical problem of low accuracy in diagnosing and warning of pumping unit malfunctions, the present invention aims to provide an edge-cloud collaborative diagnosis and warning system for a tower-type pumping unit cloud platform. The specific technical solution adopted is as follows: This invention proposes an edge-cloud collaborative diagnostic and early warning system for a tower-type pumping unit cloud platform, the system comprising an edge module and a cloud platform module: The edge module includes: The data acquisition unit is used to collect various operating data of the tower pumping unit in real time at every moment; the local diagnostic unit is used to determine the time of failure. The cloud platform module includes: The remote diagnostic and decision-making unit is used to obtain the fault response value of each type of operating data at each fault moment based on the correlation between the same type of operating data and the importance of the differences between each type of operating data in the analysis period before and after each fault moment in the historical period; to cluster the fault response values ​​of the same type of operating data at each fault moment in the historical period to obtain a response feature sequence; to train the neural network using the response feature sequence to obtain a pumping unit fault diagnosis model; and to use the pumping unit fault diagnosis model to diagnose and warn of tower-type pumping units.

[0006] Furthermore, the acquisition of fault response values ​​for each type of operational data at each fault moment includes: For each fault moment within a historical period, obtain the correlation coefficient between the same type of operational data in two analysis periods at the fault moment, and record it as the correlation value of each type of operational data before and after the fault moment; Based on the difference of the same type of operational data at the same location time within two analysis periods at the time of the fault, the weight of each type of operational data at the time of the fault is obtained by using the entropy weight method. Based on the correlation values ​​and weights of each type of operational data at the time of the fault, the fault response value of each type of operational data at the time of the fault is obtained. The correlation values ​​and weights are both positively correlated with the fault response value.

[0007] Further, obtaining the response feature sequence includes: The fault response values ​​of all types of operational data at each fault time constitute the feature sequence for that fault time. Based on the distance between different feature sequences, the feature sequences at all fault times within the historical period are clustered to obtain clusters; The response feature sequence is composed of the mean of the fault response values ​​of the same type of operational data in all feature sequences within each cluster.

[0008] Furthermore, the method for training the neural network includes: The number of times the same type of operational data exceeds its preset standard range at all fault times within the historical period is recorded as the fault count for each type of operational data; the proportion of the fault count for each type of operational data to the total fault count for all types of operational data is taken as the fault probability for each type of operational data; each type of operational data corresponds to one fault type. The operating data of the tower pumping unit at fault times and at non-fault times during historical periods, along with all response feature sequences, are used as input data for the training set. The fault types and fault probabilities of the tower pumping unit are used as output data for the training set. The neural network is trained using the training set, and the trained neural network is recorded as the pumping unit fault diagnosis model.

[0009] Furthermore, determining the time of failure includes: A preset standard range for each type of operational data is determined. If at least one type of operational data exceeds its preset standard range at any given time, then that time is the fault time.

[0010] Furthermore, the duration of the analysis period is equal to the average of the time intervals between all two adjacent fault moments within the historical period.

[0011] Furthermore, the edge module also includes a data preprocessing unit for preprocessing the operational data acquired by the data acquisition unit.

[0012] Furthermore, the edge module also includes a communication unit for transmitting preprocessed operational data to the cloud platform module.

[0013] Furthermore, the cloud platform module also includes a data receiving and storage unit for receiving and storing data transmitted from the edge module.

[0014] Furthermore, the correlation coefficient is the Pearson correlation coefficient.

[0015] Furthermore, the method for clustering the feature sequences of all fault moments within the historical period is a cohesive hierarchical clustering algorithm.

[0016] The present invention has the following beneficial effects: In this embodiment of the invention, the correlation between the same type of operating data during the analysis period before and after the fault reflects whether the fault can be identified in advance. The importance of the difference between the same type of operating data before and after the fault reflects the degree of influence of each type of operating data on fault diagnosis. Combining the two, the accuracy of the operating data at the fault time on fault diagnosis is analyzed to obtain the fault response value. The operating data of the tower pumping unit under different fault types are different. By comprehensively analyzing the data with similar fault responses in historical periods, that is, clustering the fault response values ​​of the same type of operating data at the fault time to obtain the response feature sequence, the fault state and normal state can be effectively distinguished. The response feature sequence is used to train the neural network so that the pumping unit fault diagnosis model can accurately diagnose the operating state of the pumping unit, improve the accuracy of pumping unit fault diagnosis and early warning, thereby avoiding safety accidents caused by equipment failure and extending working hours and efficiency through proactive maintenance, bringing economic benefits to the oilfield. Attached Figure Description

[0017] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A system structure diagram of an edge-cloud collaborative diagnostic and early warning system for a tower-type pumping unit cloud platform provided in an embodiment of the present invention; Figure 2 This is a structural diagram of an edge module provided in one embodiment of the present invention; Figure 3 This is a schematic diagram of a computer device for an edge-cloud collaborative diagnostic and early warning system for a tower-type oil pumping unit cloud platform, provided as an embodiment of the present invention. Detailed Implementation

[0019] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of an edge-cloud collaborative diagnostic and early warning system for a tower-type pumping unit cloud platform proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0020] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0021] The following description, in conjunction with the accompanying drawings, details a specific solution for an edge-cloud collaborative diagnostic and early warning system for a tower-type pumping unit cloud platform provided by the present invention. Example 1:

[0022] Please see Figure 1 The diagram illustrates a system block diagram of an edge-cloud collaborative diagnostic and early warning system for a tower-type pumping unit cloud platform provided by an embodiment of the present invention. The system includes: an edge module 110 and a cloud platform module 120.

[0023] The edge module 110 includes: a data acquisition unit for real-time acquisition of various operating data of the tower pumping unit at each moment; and a local diagnostic unit for determining the time of failure.

[0024] Sensors and digital monitoring instruments are installed in the mechanical structure and electrical control system of the tower-type pumping unit. An edge computing system is developed to realize edge computing and automatic control. Network access and protocol support utilize 4G mobile communication technology to access the oilfield's Virtual Private Dial Network (VPDN). Following Message Queuing Telemetry Transport (MQTT) protocol version 5.0, the remote terminal units of the pumping unit are modified to support the MQTT protocol. Through 4G VPDN, the operating status of the tower-type pumping unit and edge computing results are uploaded in real time to the MQTT message middleware deployed in the production network isolation zone. Using the EMQ framework, this data is persistently written to the TDEngine time-series database. Based on the data subscription configuration of the tower-type pumping unit's cloud platform, subscribed data is pushed to the cloud platform in real time.

[0025] Please see Figure 2 The diagram illustrates a structural diagram of an edge module provided in an embodiment of the present invention. The edge module includes: a data acquisition unit 111, a data preprocessing unit 112, a local diagnostic unit 113, and a communication unit 114.

[0026] Data acquisition unit 111 is used to collect various operational data of the tower-type pumping unit in the oilfield at any given moment in real time. Specifically, using load sensors, displacement sensors, stroke gauges, pressure sensors, current sensors, and voltage sensors, it sequentially collects the suspension point load data, head stroke data, head stroke count data, wellhead pressure data, motor current data, and motor voltage data of the tower-type pumping unit at each moment. This data is referred to as operational data. Operational data contains key information about the pumping unit's operation, and changes in operational data under different operating conditions reflect the operating status of the pumping unit.

[0027] It should be noted that all types of operational data are collected at the same frequency, which is set to 0.15 Hz. Implementers can set the frequency according to their specific circumstances.

[0028] The data preprocessing unit 112 is used to preprocess the operational data acquired by the data acquisition unit. Preprocessing includes data cleaning, noise reduction, and normalization. In this embodiment of the invention, the Laida criterion is used to detect outliers, and mean-filling is used to achieve data cleaning. The Wiener filtering algorithm is used for noise reduction, and the standardized fraction normalization method is used for normalization. The Laida criterion, mean filling, Wiener filtering algorithm, and standardized fraction normalization method are all techniques well-known to those skilled in the art and will not be described in detail here.

[0029] In other embodiments, box plots and median padding can be used for data cleaning, amplitude limiting filtering can be used for filtering, and Norm normalization function can be used for normalization. Examples will not be given here.

[0030] Local diagnostic unit 113 is used for fault detection. In this embodiment of the invention, the method for obtaining the fault time includes: determining a preset standard range for each type of operating data; if at least one type of operating data exceeds its preset standard range at each time, then that time is a fault time. If the suspension load data at a certain fault time exceeds its preset standard range, then the cause of the tower pumping unit's fault at that fault time is a suspension load fault.

[0031] It should be noted that the preset standard range for each type of operating data is the range from the minimum to the maximum value of each operating data point when the tower pumping unit is in normal operating condition. Each type of operating data corresponds to a fault type. Pre-processed operating data must be used to determine the timing of a fault.

[0032] The communication unit 114 is used to transmit pre-processed operational data to the data receiving and storage unit of the cloud platform module at the time of failure, and to mark the time of failure; it also receives various instructions and information issued by the cloud platform module. It should be noted that the communication unit periodically sends heartbeat packets to the cloud platform module to maintain its connection with the cloud platform module.

[0033] The cloud platform module 120 includes: a remote diagnosis and decision-making unit, used to obtain the fault response value of each type of operating data at each fault moment based on the correlation of the same type of operating data and the importance of the differences of each type of operating data in the analysis period before and after each fault moment in the historical period; to cluster the fault response values ​​of the same type of operating data at each fault moment in the historical period to obtain the response feature sequence; to train the neural network using the response feature sequence to obtain the pumping unit fault diagnosis model; and to use the pumping unit fault diagnosis model to diagnose and warn of tower pumping units.

[0034] The cloud platform module includes the following parts: The data receiving and storage unit is used to receive and store data transmitted from the edge module.

[0035] It should be noted that the data receiving and storage unit needs to verify and store the received data to ensure its integrity and accuracy; if there are any abnormalities in the data uploaded by the data acquisition unit of the edge module, a data retransmission request and error message should be sent to the edge module in a timely manner.

[0036] The remote diagnostic and decision-making unit is used to diagnose the operating status of tower pumping units.

[0037] The user interaction and management unit is used to view the real-time status information of the pumping unit and issue commands to the edge modules.

[0038] The specific methods for diagnosing the operating status of tower-type pumping units are as follows: (1) Under normal operating conditions, the changes of each type of operating data of the tower pumping unit are relatively stable. Faults will disrupt the stability of the same type of operating data. The correlation of the same type of operating data in the analysis period before and after the fault reflects whether the fault can be identified in advance. The importance of the difference between the same type of operating data before and after the fault reflects the degree of influence of each type of operating data on fault diagnosis. By combining the two, the accuracy of the operating data at the fault time on fault diagnosis is obtained, and the fault response value is obtained.

[0039] Preferably, in some possible implementations of the embodiments of the present invention, the method for obtaining the fault response value includes: for each fault time within a historical period, obtaining the correlation coefficient between the same type of operating data in two analysis periods of the fault time, and recording it as the correlation value before and after each type of operating data at the fault time; based on the difference between the same type of operating data at the same position time in two analysis periods of the fault time, using the entropy weight method to obtain the weight of each type of operating data at the fault time; and obtaining the fault response value of each type of operating data at the fault time according to the correlation value before and after each type of operating data at the fault time and the weight.

[0040] The correlation value before and after the fault reflects the degree of change in the correlation of each type of operational data before and after the fault. A larger correlation value indicates more stable changes in each type of operational data before and after the fault, a higher probability that the fault is a gradual one, and the fault can be identified earlier, thus increasing the accuracy of fault diagnosis. A larger weight for each type of operational data at the fault time indicates a greater impact of that data on the fault among all types of data, thus increasing the accuracy of fault diagnosis for that type of operational data. Therefore, both the correlation value before and after the fault and its weight are positively correlated with the fault response value; a larger fault response value indicates higher accuracy in diagnosing faults in tower pumping units. In this embodiment of the invention, the product of the correlation value before and after the fault and its weight for each type of operational data at each fault time is used as the fault response value for each type of operational data at each fault time.

[0041] In the embodiments of the present invention, the correlation between the pre- and post-correlation values ​​and the weights and fault response values ​​can also be constructed through other basic mathematical operations, such as the sum value, which is not limited or elaborated here.

[0042] Tower-type pumping units exist in either a normal or fault state at any given time. To accurately analyze the response characteristics of operational data under both normal and fault states, in this embodiment of the invention, the duration of the analysis period is set to the average time interval between all two adjacent fault moments within a historical period.

[0043] In this embodiment of the invention, the historical period is the time period before the current day, and the duration is set to 30 days. The implementer can set it according to the specific situation.

[0044] In this embodiment of the invention, the correlation coefficient is the Pearson correlation coefficient; in other embodiments of the invention, the correlation coefficient may also be the Spearman correlation coefficient, the Kendall rank correlation coefficient, or the canonical correlation coefficient, etc.

[0045] It should be noted that the numerical sequence is obtained by arranging the same type of operational data in each analysis period at the time of the fault according to time sequence; the correlation coefficient between the numerical sequences of the same type of operational data in two analysis periods at the time of the fault is recorded as the correlation value. The a-th time in the analysis periods before and after the fault refers to the time at the same position in the two analysis periods; where a equals the number of data acquisition times in the analysis period. The entropy weight method is used to obtain the difference of the n-th type of operational data at the m-th time in the decision matrix during the weighting process. The entropy weight method is a well-known technique to those skilled in the art and will not be described in detail here.

[0046] (2) The operating data of tower pumping units under different fault types are different. In order to ensure the accuracy of fault analysis, the data of tower pumping units with similar fault responses in the historical period are analyzed. That is, the fault response values ​​of the same type of operating data at the fault time are clustered to obtain the response feature sequence, so as to effectively distinguish the fault state and the normal state.

[0047] Preferably, in some possible implementations of the embodiments of the present invention, the method for obtaining the response feature sequence includes: constructing a feature sequence corresponding to the fault time by the fault response values ​​of all types of operating data at each fault time; clustering the feature sequences of all fault times in the historical period based on the distance between different feature sequences to obtain clusters; and constructing a response feature sequence by the mean of the fault response values ​​of the same type of operating data in all feature sequences within each cluster.

[0048] Tower-type pumping units exhibit similar response characteristics to operating data to similar fault types. This shared overall response accurately reflects changes in operating data under different fault types, enabling more precise diagnosis and early warning of pumping unit faults. Therefore, it is necessary to cluster the feature sequences formed by fault response values. A higher mean fault response value for the same type of operating data within the same cluster indicates a more significant response characteristic of the corresponding fault type on that operating data, thus better distinguishing between fault and normal states.

[0049] The normalized results of the fault response values ​​of the same type of operational data in all feature sequences within each cluster are arranged sequentially to obtain the response feature sequence. It should be noted that this embodiment uses a normalization exponential function to normalize the mean; elements with the same index in all response feature sequences correspond to the same type of operational data.

[0050] In this embodiment of the invention, agglomerative hierarchical clustering algorithm is used to cluster the feature sequences. In other embodiments, density peak clustering algorithm, etc., can be used, and will not be exemplified here; the distance between two sequences refers to the Euclidean distance.

[0051] (3) The response feature sequence can accurately distinguish between the fault state and normal state of the tower pumping unit. Using it to train the neural network can improve the accuracy of the pumping unit fault diagnosis model in terms of diagnosis and early warning.

[0052] In some possible implementations of this invention, the training method for the neural network is as follows: The number of times the same type of operating data exceeds its preset standard range at all fault times within a historical period is counted and recorded as the fault count for each type of operating data; the proportion of the fault count for each type of operating data in the total fault count for all types of operating data is used as the fault probability for each type of operating data; each type of operating data corresponds to a fault type; the operating data at fault times, the operating data at non-fault times, and all response feature sequences of the tower pumping unit within a historical period are used as input data for the training set, and the fault type and its fault probability of the tower pumping unit are used as output data for the training set; the neural network is trained using the training set, and the trained neural network is recorded as the pumping unit fault diagnosis model.

[0053] It should be noted that the data from each tower pumping unit can be considered as a training or testing sample. All tower pumping units in the oilfield are divided into training and testing sets at a ratio of 70% and 30%, respectively. The neural network is trained and evaluated sequentially using the training and testing sets. In this embodiment, the neural network is a convolutional neural network, and the loss function is the cross-entropy function. Convolutional neural networks are well-known to those skilled in the art and will not be elaborated upon here.

[0054] The system inputs all operational data and response characteristic sequences of the tower pumping unit from the immediate preceding time period into the pumping unit fault diagnosis model. The model outputs the fault type and probability of the tower pumping unit at the current moment, such as an 80% probability for motor voltage failure. Simultaneously, combining historical data and expert experience, the severity of the fault is classified, and corresponding handling measures are formulated. In this embodiment, the fault levels include: minor and severe. If the fault is minor, the remote diagnosis and decision-making unit sends an instruction to the edge device to adjust operating parameters, and the edge device adjusts the parameters of its connected sensors and other equipment. If the fault is severe, the remote diagnosis and decision-making unit sends an instruction to the edge device to immediately notify maintenance personnel, arranges a maintenance plan, and sends maintenance task information to the mobile terminals of relevant personnel.

[0055] The user interaction and management unit of the cloud platform module sends fault diagnosis results, decision commands, and related operational suggestions for the tower-type pumping unit to the corresponding edge devices via the communication unit. Upon receiving the commands from the cloud platform module, the edge module's communication unit transmits them to the local control unit, which then executes the corresponding operations. Specifically, if a stroke adjustment command is received, the control unit will drive the pumping unit's stroke adjustment mechanism to adjust the stroke to the specified value and feed the adjustment result back to the cloud platform module to confirm the command execution.

[0056] In this embodiment of the invention, the duration of the adjacent time period is set to 10 minutes, but the implementer can set it according to the specific circumstances.

[0057] This concludes the diagnostic analysis of the operating status of the tower-type pumping unit.

[0058] Regularly evaluate and optimize the performance of the edge-cloud collaborative diagnostic and early warning system. Specifically, adjust the data acquisition frequency and communication protocol settings based on the actual operating conditions of the tower-type pumping unit to improve the system's diagnostic accuracy, response speed, and stability. Collect user feedback and on-site maintenance records to improve and refine the system, continuously enriching the knowledge base of the pumping unit fault diagnosis model, enhancing the system's diagnostic capabilities and early warning effects for various complex faults, and achieving continuous optimization and upgrades of the system.

[0059] This invention is now complete. Example 2:

[0060] Figure 3 This is a schematic diagram of a computer device for an edge-cloud collaborative diagnostic and early warning system for a tower-type pumping unit cloud platform, provided as an embodiment of the present invention. For example, as shown... Figure 3As shown, the computer device includes: a memory 201, a processor 202, and a computer program 203 stored in the memory 201 and running on the processor 202, wherein when the processor 202 executes the computer program 203, the computer device can execute any of the aforementioned edge-cloud collaborative diagnostic and early warning systems for tower-type pumping unit cloud platforms.

[0061] Furthermore, embodiments of this application also protect an apparatus that may include a memory and a processor, wherein the memory stores executable program code, and the processor is used to call and execute the executable program code to execute the edge-cloud collaborative diagnostic and early warning system for a tower-type pumping unit cloud platform provided in embodiments of this application.

[0062] This embodiment can divide the device into functional modules based on the above method example. For example, each module can correspond to a separate function, or two or more functions can be integrated into one processing module. The integrated module can be implemented in hardware. It should be noted that the module division in this embodiment is illustrative and only represents one logical functional division. In actual implementation, there may be other division methods.

[0063] It should be understood that the device provided in this embodiment is used to execute the above-described edge-cloud collaborative diagnostic and early warning system for a tower-type pumping unit cloud platform, and therefore can achieve the same effect as the above-described implementation method.

[0064] When using integrated units, the device may include a processing module and a storage module. When applied to a workpiece, the processing module can be used to control and manage the workpiece's operations. The storage module can be used to support the execution of program code by the workpiece.

[0065] The processing module may be a processor or a controller, which can implement or execute various exemplary logic blocks, modules, and circuits as disclosed in this application. The processor may also be a combination of computing functions, such as a combination of one or more microprocessors, a combination of digital signal processing (DSP) and microprocessors, etc., and the storage module may be a memory. Example 3:

[0066] This embodiment also provides a computer-readable storage medium storing computer program code. When the computer program code is run on a computer, the computer executes the above-described related method steps to implement the edge-cloud collaborative diagnosis and early warning system for a tower-type pumping unit cloud platform provided in the above embodiment.

[0067] In this embodiment, the device and computer-readable storage medium are used to execute the corresponding system provided above. Therefore, the beneficial effects that can be achieved can be referred to the beneficial effects of the corresponding system provided above, and will not be repeated here.

[0068] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0069] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0070] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An edge-cloud collaborative diagnostic and early warning system for a tower-type pumping unit cloud platform, characterized in that, The system includes an edge module and a cloud platform module: The edge module includes: The data acquisition unit is used to collect various operating data of the tower pumping unit at any given moment in real time; the local diagnostic unit is used to determine the time of failure. The cloud platform module includes: The remote diagnostic and decision-making unit is used to obtain the fault response value of each type of operating data at each fault moment based on the correlation of the same type of operating data and the importance of the differences of each type of operating data in the analysis period before and after each fault moment in the historical period; to cluster the fault response values ​​of the same type of operating data at each fault moment in the historical period to obtain a response feature sequence; to train a neural network using the response feature sequence to obtain a pumping unit fault diagnosis model; and to use the pumping unit fault diagnosis model to diagnose and warn of tower-type pumping units. The process of acquiring the fault response value for each type of operational data at each fault moment includes: For each fault moment within a historical period, obtain the correlation coefficient between the same type of operational data in two analysis periods at the fault moment, and record it as the correlation value of each type of operational data before and after the fault moment; Based on the difference of the same type of operational data at the same location time within two analysis periods at the time of the fault, the weight of each type of operational data at the time of the fault is obtained by using the entropy weight method. Based on the correlation values ​​and weights of each type of operational data at the time of the fault, the fault response value of each type of operational data at the time of the fault is obtained. The correlation values ​​and weights are both positively correlated with the fault response value.

2. The edge-cloud collaborative diagnostic and early warning system for a tower-type pumping unit cloud platform according to claim 1, characterized in that, The acquisition of the response feature sequence includes: The fault response values ​​of all types of operational data at each fault time constitute the feature sequence for that fault time. Based on the distance between different feature sequences, the feature sequences at all fault times within the historical period are clustered to obtain clusters; The response feature sequence is composed of the mean of the fault response values ​​of the same type of operational data in all feature sequences within each cluster.

3. The edge-cloud collaborative diagnostic and early warning system for a tower-type pumping unit cloud platform according to claim 1, characterized in that, The method for training the neural network includes: The number of times the same type of operational data exceeds its preset standard range at all fault times within the historical period is recorded as the fault count for each type of operational data; the proportion of the fault count for each type of operational data to the total fault count for all types of operational data is taken as the fault probability for each type of operational data; each type of operational data corresponds to one fault type. The operating data of the tower pumping unit at fault times and at non-fault times during historical periods, along with all response feature sequences, are used as input data for the training set. The fault types and fault probabilities of the tower pumping unit are used as output data for the training set. The neural network is trained using the training set, and the trained neural network is recorded as the pumping unit fault diagnosis model.

4. The edge-cloud collaborative diagnostic and early warning system for a tower-type pumping unit cloud platform according to claim 1, characterized in that, The determination of the fault time includes: A preset standard range for each type of operational data is determined. If at least one type of operational data exceeds its preset standard range at any given time, then that time is the fault time.

5. The edge-cloud collaborative diagnostic and early warning system for a tower-type pumping unit cloud platform according to claim 1, characterized in that, The duration of the analysis period is equal to the average time interval between all two adjacent fault moments within the historical period.

6. The edge-cloud collaborative diagnostic and early warning system for a tower-type pumping unit cloud platform according to claim 1, characterized in that, The edge module also includes a data preprocessing unit for preprocessing the operational data acquired by the data acquisition unit.

7. The edge-cloud collaborative diagnostic and early warning system for a tower-type pumping unit cloud platform according to claim 1, characterized in that, The edge module also includes a communication unit for transmitting pre-processed operational data to the cloud platform module.

8. The edge-cloud collaborative diagnostic and early warning system for a tower-type pumping unit cloud platform according to claim 1, characterized in that, The cloud platform module also includes a data receiving and storage unit for receiving and storing data transmitted from the edge module.

9. The edge-cloud collaborative diagnostic and early warning system for a tower-type pumping unit cloud platform according to claim 1, characterized in that, The correlation coefficient mentioned is the Pearson correlation coefficient.

10. The edge-cloud collaborative diagnostic and early warning system for a tower-type pumping unit cloud platform according to claim 2, characterized in that, The method for clustering the feature sequences of all fault moments within the historical period is the agglomerative hierarchical clustering algorithm.

Images