Large screw pump shaft seal self-maintenance method and system

Abstract

The invention relates to the technical field of electromechanical equipment maintenance, in particular to a large screw pump shaft seal self-maintenance method and system.The method comprises the following steps that a data collection device of a large screw pump is established, data fragments of the large screw pump are obtained, and the data fragments are preprocessed to obtain a data matrix; constructing a temperature prediction model of the large screw pump to obtain the internal temperature of the large screw pump; obtaining a residual life characteristic curve of the large-scale screw pump, and constructing a residual life prediction model for blocking the sealing water flow channel; and based on the sealing water flow channel blocking residual life prediction model, the sealing water flow channel blocking residual life percentage is obtained, so that shaft sealing self-maintenance of the large screw pump is achieved. And performing state display and alarm on the large-scale screw pump according to the percentage of the residual life of the blocked sealing water flow channel. The service life of a shaft sealing system can be prolonged, and a closed-loop solution is provided for intelligent operation and maintenance of the large screw pump.

Description

Technical Field

[0001] The present invention relates to the technical field of electromechanical equipment maintenance, in particular to a large screw pump shaft seal self-maintenance method and system. Background Art

[0002] Large screw pumps meet the specific needs of the fiber production industry, such as high-viscosity media transfer, stability, reliability, and corrosion resistance. They play an important role in slurry transfer, polymer solution transfer, and fiber recovery and reuse. Their energy-efficient operation, ease of maintenance, and adaptability make them widely used in the fiber production industry. However, large screw pumps are expensive, generally lack spare parts, and are often used to transfer viscous media. Maintenance requires draining the pump and its upstream and downstream equipment, a tedious and time-consuming process. Therefore, preventive maintenance is recommended to avoid unplanned downtime. The stator of a large screw pump is typically made of flexible components; the rotor transmits power via a universal joint. The shaft seal system is subject to radially uneven forces, which can lead to wear and tear. Daily inspections for leaks and insufficient seal water are required, and refilling of seal water is often necessary, resulting in high labor costs.

[0003] The shaft sealing system is a critical component of large screw pumps. A reliable shaft sealing system not only prevents leaks and ensures production safety, but also improves efficiency, reduces energy consumption, extends equipment life, and reduces maintenance costs. Furthermore, it adapts to diverse operating conditions, enhancing equipment flexibility. The sealing water in the shaft sealing system not only seals the pump but also circulates cooling water through the mechanical seal. Over extended use, the sealing water flow path may become clogged, impairing the cooling cycle. Inadequate sealing can lead to leakage of the transmission medium, while cooling failure can cause rapid wear or burnout of the mechanical seal, resulting in downtime and production disruptions. Furthermore, replacement costs are high.

[0004] Currently, there is no public patent for automatic water addition to the shaft sealing system. The existing technology has proposed a new solution for adding and changing water in seawater tanks, which can realize automatic quantitative water replacement and water addition functions. However, manual confirmation is required to determine whether water addition is required, and unattended automatic water supply to the water-to-be-added system cannot be achieved.

[0005] In response to the deficiencies in the prior art, the present invention designs a mechatronic automatic control system to specifically address the problems of large screw pump shaft sealing systems requiring daily inspections for sealing water leakage and insufficiency, frequent need to add sealing water at irregular intervals, high labor costs, and medium leakage when sealing water is insufficient, blockage of the sealing water flow channel leading to damage to equipment and shutdowns affecting production; the present invention intelligently implements the automatic quantitative water change and water addition functions of the shaft sealing system of a large screw pump; and solves the technical problems of time-consuming and labor-intensive water addition and replacement processes, self-cleaning of the sealing water circulation cooling channel, and blockage alarms. Summary of the Invention

[0006] In view of the defects in the prior art, the present invention provides a large screw pump shaft seal self-maintenance method and system.

[0007] In order to achieve the above-mentioned objectives, in a first aspect, the present invention provides a self-maintenance method for the shaft seal of a large screw pump, the method comprising the following steps: establishing a data acquisition device for a large screw pump, acquiring data segments of the large screw pump based on the data acquisition device, and preprocessing the data segments to obtain a data matrix; constructing a temperature prediction model for the large screw pump, and obtaining the internal temperature of the large screw pump according to the temperature prediction model; obtaining a remaining life characteristic curve of the large screw pump, and constructing a sealing water flow channel blockage remaining life prediction model based on the remaining life characteristic curve; obtaining a sealing water flow channel blockage remaining life percentage based on the sealing water flow channel blockage remaining life prediction model to achieve self-maintenance of the shaft seal of the large screw pump; and displaying the status and alarming the large screw pump according to the sealing water flow channel blockage remaining life percentage. The present invention builds a high-precision analysis foundation through real-time data acquisition and preprocessing, and combines the temperature prediction model to realize dynamic monitoring and active regulation of the equipment thermal state, effectively avoiding the risk of overheating failure; the sealing water flow channel blockage prediction model based on the remaining life characteristic curve can quantify the equipment degradation trend, warn maintenance nodes in advance, transform passive repairs into active maintenance, and significantly reduce the frequency of unplanned downtime; status visualization and alarm mechanism improve operation and maintenance response efficiency, extend the service life of the shaft sealing system, and comprehensively improve the equipment operation reliability, safety and economy, providing a closed-loop solution for the intelligent operation and maintenance of large screw pumps.

[0008] Optionally, the data acquisition device for a large screw pump is established, data segments of the large screw pump are obtained based on the data acquisition device, and the data segments are pre-processed to obtain a data matrix, including: establishing the data acquisition device through a water pressure sensor and a temperature sensor; obtaining water pressure data and temperature data of the large screw pump based on the data acquisition device, combining the water pressure data and the temperature data to obtain the data segments; pre-processing the data segments, wherein the pre-processing includes improving low-pass filtering, time domain analysis, frequency domain analysis and data normalization to obtain the data matrix. The present invention constructs a multi-parameter acquisition device through water pressure and temperature sensors to achieve full-dimensional monitoring of the operating status; improved low-pass filtering effectively suppresses high-frequency noise interference, time-frequency domain analysis accurately extracts equipment vibration characteristics, data normalization eliminates dimensional differences, forms a high signal-to-noise ratio structured data matrix, improves the quality of the original data, and provides a reliable input basis for the subsequent life prediction model. At the same time, through multi-dimensional signal analysis, early fault signs are captured, the effective life of the sealing system is extended, and a complete closed loop from data acquisition to feature optimization is constructed, which greatly improves the intelligent level of equipment operation and maintenance.

[0009] Optionally, the preprocessing of the data segments includes: the mathematical model of the improved low-pass filter satisfies the following relationship: ; in, For the The output value of the second low-pass filter, is the filter coefficient, For the The sampling value of times, For the The output value of the second low-pass filter, is the data threshold. The present invention achieves a balance between noise suppression and signal fidelity by dynamically weighting and fusing the current sampling value and the historical output value based on an improved low-pass filter model. The introduction of the threshold can effectively filter out low-amplitude interference and avoid false alarms triggered by false signals; the filter coefficient optimizes the weight distribution of new and old data, retaining effective mutation characteristics while suppressing high-frequency noise. Combined with the segmented processing mechanism, the data signal-to-noise ratio is improved while ensuring real-time performance, providing high-quality input for temperature prediction and life analysis, and significantly enhancing the anti-interference ability and decision-making reliability of the shaft seal self-maintenance system.

[0010] Optionally, the temperature prediction model of the large screw pump is constructed, and the predicted temperature of the large screw pump is obtained according to the temperature prediction model, including: obtaining a temperature prediction data set of the large screw pump based on the data acquisition device, and constructing a temperature prediction model of the large screw pump according to the temperature prediction data set; obtaining the predicted temperature of the large screw pump according to the temperature prediction model, and the predicted temperature is used as the internal temperature of the large screw pump. The present invention realizes accurate quantitative analysis of the thermal state of the equipment by collecting multi-source data such as water pressure and temperature in real time and constructing a prediction model, so as to capture abnormal temperature trends in advance, avoid failure of seals due to overheating, and extend the service life of the shaft sealing system. At the same time, the prediction results provide a scientific basis for operation and maintenance decisions and reduce unnecessary energy consumption. The time-domain-frequency domain joint analysis technology improves the accuracy of fault diagnosis, and the data normalization processing eliminates the interference of working condition differences, enhances the generalization ability of the model, and is applicable to temperature prediction scenarios of screw pumps of different specifications, forming a complete closed loop from data acquisition to intelligent decision-making.

[0011] Optionally, obtaining the remaining life characteristic curve of the large screw pump includes fitting the remaining life characteristic curve using a polynomial and a hyperbolic function, and expressing the remaining life characteristic curve as a percentage. The present invention uses a polynomial and a hyperbolic function to fit the remaining life characteristic curve, accurately characterizing the nonlinear characteristics of equipment degradation and improving the accuracy of life prediction. Quantifying the remaining life in percentage form visualizes the health status of the equipment, providing an intuitive quantitative basis for operation and maintenance decisions, optimizing maintenance resource allocation, reducing the risk of sudden failures, and extending the effective service life of the shaft sealing system.

[0012] Optionally, the remaining life characteristic curve includes: ; ; in, is the remaining life percentage, is the front-end control item, is an adjustable parameter, is the system time corresponding to the data, is the transition function term, is the base of natural logarithms, is the normalized denominator. The present invention achieves accurate life prediction through the synergistic effect of multiple parameters: the front-end control term characterizes the early linear degradation law, the transition function constructs the nonlinear mutation mechanism, and the exponential term simulates the late accelerated failure characteristics. The combination of adjustable parameters gives the model high adaptability and can accurately fit the seal degradation trajectory under different working conditions. The normalization process ensures intuitive quantification of the life percentage, provides a scientific decision-making basis for the self-maintenance system, and significantly improves the accuracy of shaft seal failure warning and the economic efficiency of operation and maintenance.

[0013] Optionally, constructing a sealing water channel blockage remaining life prediction model based on the remaining life characteristic curve includes: obtaining a shaft seal remaining life prediction dataset for the large screw pump based on the data matrix and the remaining life characteristic curve; dividing the shaft seal remaining life prediction dataset into a training set, a validation set, and a test set; and constructing the sealing water channel blockage remaining life prediction model based on the training set, the validation set, and the test set. The present invention constructs a sealing water channel blockage prediction model through a data-driven approach to achieve quantitative analysis of the equipment degradation process. The dataset partitioning strategy effectively verifies the model's generalization capability, avoids the risk of overfitting, and improves prediction reliability. It accurately predicts the time node of seal failure, optimizes maintenance resource allocation, transforms passive emergency repairs into active preventive maintenance, and significantly reduces the probability of unplanned downtime. At the same time, the model output provides a scientific basis for operation and maintenance decisions, extends the effective life of the shaft seal system, reduces the economic losses caused by sudden failures, and comprehensively improves the economic efficiency and safety of equipment operation.

[0014] Optionally, the remaining life percentage of the sealing water channel blockage is obtained based on the remaining life prediction model of the sealing water channel blockage to achieve self-maintenance of the shaft seal of the large screw pump, including: adjusting the water pump motor of the large screw pump based on the remaining life percentage of the sealing water channel blockage, and judging the status of the shaft seal water circulation of the large screw pump to achieve self-maintenance of the shaft seal of the large screw pump. The present invention is based on the remaining life prediction model of the sealing water channel blockage. By adjusting the operating parameters of the water pump motor, it can optimize the working environment of the shaft seal, reduce mechanical wear, and extend the service life of the seal. At the same time, the real-time monitoring of the shaft seal water circulation status can timely detect signs of leakage or blockage and avoid equipment failure. Combined with the remaining life percentage, a maintenance plan can be accurately formulated to achieve preventive maintenance, significantly reduce the risk of unplanned downtime, and improve the reliability and economy of equipment operation.

[0015] Optionally, the status display and alarm of the large screw pump are performed based on the remaining life percentage of the sealing water flow channel blockage, including: displaying the remaining life percentage of the sealing water flow channel blockage, and setting a threshold value of the remaining life percentage of the sealing water flow channel blockage, and alarming when the remaining life percentage of the sealing water flow channel blockage is lower than the threshold value. The present invention displays the remaining life percentage of the sealing water flow channel blockage in real time through a visual interface, so that operation and maintenance personnel can intuitively grasp the health status of the equipment; setting a dynamic threshold to trigger the alarm mechanism can issue an early warning at the early stage of seal failure to avoid sudden shutdown accidents, realize closed-loop management from data monitoring to active warning, and shift the maintenance strategy from passive repair to active prevention, significantly reducing the probability of unplanned shutdowns, extending the service life of the shaft sealing system, and optimizing the allocation of maintenance resources, improving the safety and economy of equipment operation, and providing key guarantees for the intelligent operation and maintenance of large screw pumps.

[0016] In the second aspect, the present invention provides a large screw pump shaft seal self-maintenance system, which executes the large screw pump shaft seal self-maintenance method provided by the present invention, and the system includes: a data acquisition and preprocessing module, the data acquisition and preprocessing module is used to obtain data segments of the large screw pump, and preprocess the data segments to obtain a data matrix; a temperature prediction module, the temperature prediction module is used to construct a temperature prediction model of the large screw pump, and obtain the internal temperature of the large screw pump according to the temperature prediction model; a health status prediction module, the health status prediction module is used to obtain the remaining life characteristic curve of the large screw pump, and construct a sealing water flow channel blockage remaining life prediction model based on the remaining life characteristic curve; a water pump control module, the water pump control module is used to obtain the sealing water flow channel blockage remaining life percentage based on the sealing water flow channel blockage remaining life prediction model to achieve self-maintenance of the shaft seal of the large screw pump; a status display and alarm module, the status display and alarm module is used to display the status of the large screw pump and alarm according to the sealing water flow channel blockage remaining life percentage. The present invention constructs an efficient information processing architecture to achieve closed-loop control of real-time data acquisition, intelligent analysis and decision feedback, significantly improving the accuracy of shaft seal fault prediction and self-maintenance response speed. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a flow chart of a large screw pump shaft seal self-maintenance method according to an embodiment of the present invention;

[0018] Figure 2 Schematic diagram of the flow of the temperature prediction module according to an embodiment of the present invention;

[0019] Figure 3 This is an architecture diagram of a large screw pump shaft seal self-maintenance system according to an embodiment of the present invention;

[0020] Figure 4 Schematic diagram of the structural assembly of the large screw pump shaft seal self-maintenance system according to an embodiment of the present invention;

[0021] Figure 5 This is a schematic diagram of the coverless structure of the large screw pump shaft seal self-maintenance system according to an embodiment of the present invention;

[0022] Figure 6 This is a schematic structural diagram of a return water filtration device according to an embodiment of the present invention;

[0023] Figure 7 A schematic diagram of a partial structure of a water pump connection according to an embodiment of the present invention;

[0024] Figure 8 A schematic diagram of a return water temperature test structure according to an embodiment of the present invention;

[0025] Figure 9This is a schematic diagram of the connection structure between the self-maintenance system and the shaft sealing system according to an embodiment of the present invention;

[0026] Explanation of the accompanying drawings: ultrasonic liquid level sensor 1, cover plate 2, large water tank 3, water pump 4, first small water tank 5, water pressure sensor 6, first water pipe 7, hinge 8, partition 9, second water pipe 10, third water pipe 11, water tank 12, connecting bracket 13, filter 14, first screw hole 15, second screw hole 16, third screw hole 17, second small water tank 18, fourth screw hole 19, temperature sensor 20, water hole 21, water pipe joint 22. DETAILED DESCRIPTION

[0027] Specific embodiments of the present invention will be described in detail below. It should be noted that the embodiments described herein are for illustrative purposes only and are not intended to limit the present invention. In the following description, numerous specific details are set forth to provide a thorough understanding of the present invention. However, it will be apparent to one of ordinary skill in the art that these specific details are not necessarily required to practice the present invention. In other instances, well-known circuits, software, or methods are not specifically described to avoid obscuring the present invention.

[0028] Throughout this specification, references to "one embodiment," "an embodiment," "an example," or "an example" mean that a particular feature, structure, or characteristic described in connection with the embodiment or example is included in at least one embodiment of the present invention. Therefore, appearances of the phrases "in one embodiment," "in an embodiment," "an example," or "an example" in various places throughout this specification are not necessarily all referring to the same embodiment or example. Furthermore, the particular features, structures, or characteristics may be combined in any suitable combinations and / or subcombinations in one or more embodiments or examples. Furthermore, those of ordinary skill in the art will appreciate that the figures provided herein are for illustrative purposes only and are not necessarily drawn to scale.

[0029] See Figure 1 One embodiment of the present invention provides a self-maintenance method for a large screw pump shaft seal, the method comprising the following steps:

[0030] S1. Establish a data acquisition device for a large-scale screw pump, obtain data segments of the large-scale screw pump based on the data acquisition device, and pre-process the data segments to obtain a data matrix.

[0031] In this embodiment, the data acquisition and preprocessing module uses the water pressure sensor 6 and temperature sensor 20 to acquire water pressure and temperature data at corresponding locations. Data is collected every five minutes, for 10 seconds each time, at a sampling frequency of 20 Hz. After each data collection session, the water pressure and temperature data are saved in a spreadsheet file, forming a data segment.

[0032] The improved low-pass filter is used to preprocess the data collected by the sensor. The mathematical model of the improved low-pass filter satisfies the following relationship:

[0033]

[0034] in, For the The output value of the second low-pass filter, is the filter coefficient, For the The sampling value of times, For the The output value of the second low-pass filter, is the data threshold.

[0035] In practice, the screw pump shaft seal self-maintenance system operates intermittently, resulting in discontinuous water pressure and temperature data collection. The improved low-pass filter not only effectively obtains smoothed water pressure and temperature data but also facilitates the splicing of data collected from different time periods, avoiding data abrupt changes. This provides a data foundation for the design of the system's health status prediction module. The pressure and temperature data from the collected and saved spreadsheet files are then read and preprocessed. This preprocessing includes time domain analysis, frequency domain analysis, and data normalization calculations for each data segment.

[0036] Time domain analysis mainly completes the root mean square extraction and kurtosis extraction of each data segment. The mathematical model is as follows:

[0037]

[0038]

[0039] in, is the root mean square of the data segment, is the number of samples in the data segment, is the index variable of the sample, The first data segment Sample values, is the kurtosis of the data segment, is the average value of the data segment.

[0040] Frequency domain analysis mainly completes the conversion of pressure and temperature signals of each data segment from time domain to frequency domain. The mathematical model is as follows:

[0041]

[0042] in, The frequency domain value of the data segment after transformation, is the number of discrete sampling points of the data segment, is the index variable of the sampling point, The first data segment data samples, is the base of natural logarithms, is pi, is the imaginary unit, It is the sequence number of the time domain discrete value of the data segment.

[0043] Furthermore, frequency centroid extraction and frequency mean extraction are performed, and the mathematical model is as follows:

[0044]

[0045]

[0046] in, is the frequency center of gravity of the data segment, is the total number of spectral lines, is the index variable of the spectrum line, For the The frequency value of the spectral line, is the frequency domain value after data segment transformation, is the number of discrete sampling points of the data segment, is the frequency mean of the data segment.

[0047] Furthermore, the above data is organized into a data matrix, which includes the following data: the root mean square, kurtosis, frequency center of gravity, and frequency mean of the pressure and temperature data. Finally, each data in the above data matrix is ​​normalized and output. The data normalization calculation method in this application is: the difference between the current data size and the minimum value of the data type is divided by the difference between the maximum value and the minimum value of the data type, resulting in a number between 0 and 1.

[0048] S2. Construct a temperature prediction model for the large screw pump, and obtain the internal temperature of the large screw pump according to the temperature prediction model.

[0049] See Figure 2 , the figure is a flow chart of the temperature prediction module. In this embodiment, the temperature prediction module predicts the temperature at a future time point as the current internal temperature of the sealed system to address the problem that the tested external temperature lags behind the internal temperature of the sealed system, resulting in a time mismatch between pressure and temperature. The predicted temperature will constitute the input data of the health status prediction module together with the currently measured pressure. The temperature prediction module is implemented by constructing a deep learning model. The external temperature in the next 5 minutes is used as the label data, and the temperature prediction data set is constructed together with the root mean square of the external temperature data in the data matrix of the past 20 minutes. Then, a temperature prediction model including position encoding, residual connection, feedforward network and full connection layer is constructed, including:

[0050] The input sequence satisfies the following relationship:

[0051]

[0052] in, is the input sequence, is the first Data for each location, is the set of real numbers, is the input feature dimension.

[0053] Temperature data is collected every 10 seconds. The historical sequence length is 12 (i.e., the past 5 minutes). d = 1, indicating that there is only one temperature feature. It can also be expanded to multiple dimensions (including timestamps, days of the week, etc.), satisfying the following relationship:

[0054]

[0055] in, To predict the sequence, For the The temperature value of each time step, is the set of real numbers.

[0056] Position encoding satisfies the following relationship:

[0057]

[0058] in, is the position code, is the time step position, is the dimension of the current vector, is the input feature dimension, 10000 is the control cycle constant, is an even number, An odd number.

[0059] It should be noted that the control period constant makes the frequencies of different dimensions different.

[0060] Multi-head sparse attention satisfies the following relationship:

[0061]

[0062]

[0063] in, is the query vector, is the input of the attention mechanism, is the weight matrix, is the key vector, is a value vector, is the set of real numbers, is the input feature dimension, is the dimension of the key vector.

[0064] The classic attention score satisfies the following relationship:

[0065]

[0066] in, is the attention score, is the query vector, is the key vector, is a value vector, is normalized to probability weight, To calculate the similarity (dot product) between the query and all keys, is the dimension of the key vector (usually the same as the query vector).

[0067] Residual connection and normalization satisfy the following relationship:

[0068]

[0069] in, is the output of the attention mechanism, is layer normalization, is the input of the attention mechanism, is the attention score.

[0070] The feedforward network satisfies the following relationship:

[0071]

[0072] in, is a feedforward network, is the input vector, is a nonlinear activation function, is the first layer linear transformation weight, is the first layer bias, is the second layer linear transformation weight, Bias for the second layer.

[0073] The nonlinear activation function satisfies the following relationship:

[0074]

[0075] in, For the The output value after the position is activated, is the input value of the nonlinear activation function.

[0076] The output layer satisfies the following relationship:

[0077]

[0078] in, is the predicted future temperature series, is a fully connected linear transformation layer, is the hidden state vector, is the weight matrix that projects the hidden state to the output dimension, is the bias term.

[0079] The mean square error loss function is used to satisfy the following relationship:

[0080]

[0081] in, is the loss value, is the total amount of data in the input sequence, is the traversal count flag, is the actual label temperature corresponding to the current input sequence data, The predicted label temperature corresponding to the current input sequence data.

[0082] Then, the deep learning model is trained, verified and tested with the dataset as input to obtain a temperature prediction model. The output of the temperature prediction model is the predicted internal temperature of the current sealing system.

[0083] S3. Obtain a remaining life characteristic curve of the large screw pump, and construct a sealing water flow channel blockage remaining life prediction model based on the remaining life characteristic curve.

[0084] In this embodiment, the health status prediction module predicts the sealing water channel blockage and the shaft sealing system cooling condition through the sealing water pressure and sealing water temperature, and outputs the remaining sealing water channel blockage life.

[0085] Specifically, the health status prediction module employs a fault prediction method driven by both data and knowledge. The design method is as follows: First, the data acquisition and preprocessing module collects the seal water pressure and temperature of a large screw pump shaft seal system during a blockage cycle and obtains the root mean square (RMS) values ​​of the pressure and temperature. These RMS values ​​are normalized separately. The normalized results at the same moment are then summed and normalized. Analysis of the normalized data curve reveals that seal water flow channel blockage gradually accumulates over time, with the accumulation rate increasing. This phenomenon is similar to flow channel blockage in reality: as the flow channel becomes blocked, the seal water flow rate decreases. The slower the flow, the more likely the flow channel is to scale, resulting in poor cooling of the shaft seal system and deteriorating sealing performance, forming a vicious cycle. Therefore, the data curve can be used as a reference for estimating the health status of the shaft seal system. However, this data curve has some flaws, such as significant fluctuations, which do not align with the physical reality of accumulated flow channel blockage. This is due to the presence of numerous interferences in the measured data signals.

[0086] In light of this, the present invention combines the internal temperature of large screw pumps with an empirical mathematical model and big data feature extraction to design a residual life characteristic curve. Based on common mathematical knowledge, polynomial and hyperbolic functions are used to fit the residual life characteristic curve. Furthermore, the mathematical model expressing the residual life characteristic curve in percentage form, combined with this curve expression, is as follows:

[0087]

[0088]

[0089] in, is the remaining life percentage, is the front-end control item, is an adjustable parameter, is the system time corresponding to the data, is the transition function term, is the base of natural logarithms, is the normalized denominator.

[0090] The specific debugging process of the remaining life characteristic curve is: first set The initial value of is 1, and the data curve is further observed to see how well it fits the remaining life characteristic curve being fitted, and then the adjustment is made. until the data curve is well matched with the fitted remaining life characteristic curve; at this point, the root mean square, kurtosis, frequency center of gravity, frequency mean and remaining life percentage data of the collected and calculated pressure and temperature data together constitute the data set for the remaining life prediction of the shaft sealing system, among which the remaining life percentage data is the label data.

[0091] Furthermore, a prediction model for the remaining life of the sealing water flow channel blockage is designed. This model is based on a convolutional neural network (CNN) coupled with a deep learning model based on a self-attention mechanism to construct a prediction model as a prediction model for the remaining life of the sealing water flow channel blockage, including:

[0092] CNN feature extraction module:

[0093] The input sequence satisfies the following relationship:

[0094]

[0095] in, is the input sequence, is the set of real numbers, is the number of time steps of the input sequence, is the feature dimension.

[0096] For the convolution operation of the input sequence, the following relationship is satisfied:

[0097]

[0098] in, is the convolution output, is the convolution kernel, is the position offset of the convolution kernel in the time dimension, is the feature dimension, is the feature dimension index of the input sequence, is the convolution kernel weight, For the input sequence at time step and feature dimensions The value of is the bias vector.

[0099] It should be noted that the features in the present invention include the root mean square, kurtosis, frequency centroid, frequency mean and remaining life percentage data of pressure data and temperature data, so the feature dimension is 10.

[0100] The deep learning model based on the self-attention mechanism includes position encoding, classic attention scoring, multi-head splicing, feedforward network, fully connected layer and output layer, and adopts mean square error loss function.

[0101] The position encoding, classical attention scoring, and feedforward network are the same as those described in step S2.

[0102] Multi-head splicing satisfies the following relationship:

[0103]

[0104] in, For the output of multi-head splicing, is the query vector, is the key vector, is a value vector, For splicing operation, For the The output of an attention head, is the output mapping matrix.

[0105] The fully connected layer and the output layer satisfy the following relationship:

[0106]

[0107] in, is the model prediction output, is the output layer weight, is a nonlinear activation function, is the weight matrix of the fully connected layer, is the input vector, is the bias term of the fully connected layer, is the output layer bias term.

[0108] The mean square error loss function is used to satisfy the following relationship:

[0109]

[0110] in, is the loss value, is the total amount of data in the input sequence, is the traversal count flag, is the actual remaining life percentage corresponding to the current input sequence data, The predicted remaining life percentage corresponding to the current input sequence data.

[0111] Then, 80% of the data in the shaft seal system remaining life prediction dataset was used as the training set, 10% as the validation set, and the remaining 10% as the test set. Following the deep learning model development process, the model was trained, validated, and tested using the divided datasets. Ultimately, a prediction model for the remaining life of the seal water flow channel blockage was obtained. The model output is the remaining life of the seal water flow channel blockage. Finally, the health status prediction module outputs the remaining life of the seal water flow channel blockage percentage.

[0112] S4. Obtaining a percentage of the remaining life of the sealing water channel blockage based on the sealing water channel blockage remaining life prediction model to achieve self-maintenance of the shaft seal of the large screw pump.

[0113] In this embodiment, the water pump control module, on the one hand, adjusts the operation of the water pump 4 motor according to the remaining life percentage of the sealing water flow channel blockage output by the health status prediction module. When the remaining life percentage of the sealing water flow channel blockage gradually decreases, the displacement of the water pump 4 will be appropriately compensated so that the performance of the shaft sealing system is not affected by the blockage. On the other hand, the module collects the water pressure of the sealing water and the liquid level of the large water tank 3, and comprehensively judges whether the water circulation of the self-maintenance system is normal. If the liquid level sensor finds that the liquid level of the large water tank 3 is too low, the water shortage alarm state is output. The water pump 4 is a consumable part of the self-maintenance system. If the water pump 4 motor is working and the liquid level of the large water tank 3 is normal, but the water pressure is zero, it means that the water pump 4 is damaged, and an alarm for replacing the water pump 4 is output.

[0114] S5. Display the status of the large screw pump and issue an alarm based on the remaining life percentage of the sealing water flow channel blockage.

[0115] In this embodiment, the status display and alarm module displays the remaining life of the sealed water flow channel blockage, providing a reference for the maintenance department to implement predictive maintenance of the equipment. At the same time, if the remaining life percentage is lower than the set threshold (the threshold in the present invention is 5%), an alarm will be triggered to remind the maintenance department to arrange maintenance work in advance to avoid unplanned downtime and affect production.

[0116] See Figure 3 In an optional embodiment, the present invention provides a large screw pump shaft seal self-maintenance system, which uses a large screw pump shaft seal self-maintenance method provided by the present invention. The system includes: a data acquisition and preprocessing module, a temperature prediction module, a health status prediction module, a water pump control module and a status display and alarm module.

[0117] In an optional embodiment, a large screw pump shaft seal self-maintenance system is provided, which is mainly characterized by a mechanical system and a control system.

[0118] The overall structure of the mechanical system includes the controller, as well as:

[0119] See Figure 4 The figure shows a schematic diagram of the structural assembly of the large screw pump shaft seal self-maintenance system, which consists of an ultrasonic liquid level sensor 1, a cover plate 2, a large water tank 3, a water pump 4, a first small water tank 5, a water pressure sensor 6, a first water pipe 7 and a hinge 8.

[0120] See Figure 5 The figure shows a schematic diagram of the coverless structure of the large screw pump shaft seal self-maintenance system, which consists of a partition 9, a second water pipe 10, and a third water pipe 11.

[0121] See Figure 6 , the figure shows a schematic structural diagram of the backflow water filtering device, which consists of a water tank 12, a connecting bracket 13, a filter 14, a first screw hole 15 and a second screw hole 16.

[0122] See Figure 7 , the figure is a schematic diagram of the local structure of the water pump connection, including the third screw hole 17.

[0123] See Figure 8 , the figure shows a schematic diagram of the return water temperature test structure, which consists of a second small water tank 18, a fourth screw hole 19, a temperature sensor 20 and a water hole 21.

[0124] The software system is arranged within the controller. The cover 2 is hinged to the large water tank 3 via hinge 8. The ultrasonic liquid level sensor 1 is fixedly connected to the large water tank 3. The water pump 4 is bolted to the large water tank 3 via the third screw hole 17. The first small water tank 5 is fixedly connected to the water pump 4. The first water pipe 7 is connected to the first small water tank 5. The water pressure sensor 6 is fixedly connected to the first small water tank 5. The partition 9 is fixedly connected to the large water tank 3. One end of the third water pipe 11 passes through the large water tank 3 and is connected to the water pump 4. One end of the second water pipe 10 is connected to the second small water tank 18. The filter 14 is fixedly connected to the water tank 12. The connecting bracket 13 is bolted to the large water tank 3 via the first screw hole 15. The water tank 12 is bolted to the connecting bracket 13 via the second screw hole 16. The temperature sensor 20 is inserted into the second small water tank 18 and fixedly connected to it. The second small water tank 18 is bolted to the large water tank 3 via the fourth screw hole 19. The water discharged from the shaft sealing system enters the second small water tank 18 through the second water pipe 10 and flows into the water tank 12 through the water hole 21. A gap is left between the lower end of the partition 9 and the bottom of the large water tank 3 to ensure that the water in the large water tank 3 can flow smoothly on both sides of the partition 9.

[0125] See Figure 9 , the figure shows a schematic diagram of the connection structure between the self-maintenance system and the shaft sealing system, including a water pipe joint 22.

[0126] When the pressure of water pressure sensor 6 falls below the warning value, indicating that the shaft sealing system is short of water, water pump 4 automatically operates to deliver the liquid in large water tank 3 to the shaft sealing system. When ultrasonic liquid level sensor 1 detects that the liquid level in large water tank 3 remains unchanged as water pump 4 operates, it indicates that the shaft sealing system has completed water addition. When the water pump control module detects that the temperature of the sealing water discharged from the shaft sealing system is higher than the warning value, water pump 4 automatically operates and adjusts the displacement of water pump 4 according to the extent to which the sealing water temperature exceeds the warning value, thereby achieving water cooling of the shaft sealing system through circulating sealing water.

[0127] It should be noted that a large screw pump shaft seal self-maintenance system, in actual operation, automatically adds sealing water, automatically circulates sealing water to cool the sealing system, automatically flushes the flow channel to alleviate flow channel blockage, predicts and issues an alarm for flow channel blockage, and provides a water shortage alarm. When the ultrasonic liquid level sensor 1 detects that the liquid level in the large water tank 3 is below the warning value, indicating insufficient water storage, the self-maintenance system automatically alarms, notifying equipment maintenance personnel to promptly add water to the large water tank 3. When the pressure of the water pressure sensor 6 drops below the warning value, indicating a water shortage in the shaft sealing system, the water pump 4 automatically operates to transfer water from the large water tank 3 to the shaft sealing system. When the ultrasonic liquid level sensor 1 detects that the liquid level in the large water tank 3 remains unchanged as the water pump 4 operates, it indicates that the shaft sealing system has been watered. When the sealing system no longer requires water, the water pump 4 continues to operate, driven by the output of the health status prediction module. At this time, the water pump 4 operates to flush the flow channel, extending the life of the sealing water flow channel blockage and reducing equipment downtime caused by flow channel blockage.

[0128] The present invention designs a mechatronic automatic control system, which uses an ultrasonic liquid level sensor 1, a water pressure sensor 6, a temperature sensor 20, a water pump 4, a proprietary structure, and the like to design a low-cost shaft seal self-maintenance system. This system can effectively solve the problem that the shaft seal system of a large screw pump needs to be inspected daily to see if the sealing water is leaking or insufficient, and sealing water often needs to be added at irregular intervals, resulting in high labor costs. In addition, when the sealing water is insufficient, the medium leaks, the sealing water flow channel is blocked, and the equipment is damaged, and the shutdown affects production. Through the water tank, water pump 4, water pipe, and return water filter device, a real-time circulating cooling system for the shaft seal system can be formed at special times to effectively reduce the temperature. In addition, the circulating system can plan the operating time of the water pump 4 through a life prediction module, automatically control the flushing of the flow channel, avoid or slow down the flow channel blockage, reduce system maintenance time, and improve production efficiency. By real-time detection of the water pressure during the sealing water addition process, the blockage in the sealing water circulation flow channel is estimated. The solution is low-cost and effective, and can effectively avoid the problem of rapid wear or burning of the mechanical seal due to the blockage of the sealing water flow channel. By improving the low-pass filter, continuous splicing of data collected in different time periods can be achieved to avoid data mutations. This facilitates the subsequent extraction of essential data features through deep learning networks. A health status prediction module was designed to address the temporal mismatch between pressure and temperature caused by the external temperature of the test lagging behind the internal temperature of the sealed system. This also facilitates the timely detection of temperature anomalies. The communicating vessel structure formed by the water tank and baffle 9 not only meets the functional requirements of the maintenance system, but also prevents backflow water from causing liquid level fluctuations when baffle 9 is absent, making it difficult for the liquid level sensor to accurately obtain the liquid level in real time.

[0129] In summary, the method of the present invention provides a large screw pump shaft seal self-maintenance method and system, which builds a high-precision analysis foundation through real-time data acquisition and preprocessing, and combines the temperature prediction model to realize dynamic monitoring and active regulation of the equipment thermal state, effectively avoiding the risk of overheating failure; the sealing water flow channel blockage prediction model based on the remaining life characteristic curve can quantify the equipment degradation trend, warn maintenance nodes in advance, and significantly reduce the frequency of unplanned downtime; the status visualization and alarm mechanism improves the operation and maintenance response efficiency, extends the service life of the shaft sealing system, and comprehensively improves the equipment operation reliability, safety and economy, providing a closed-loop solution for the intelligent operation and maintenance of large screw pumps. The method of the present invention is easy to understand, simple to calculate, has a small workload, is convenient for engineering application, and provides a theoretical basis and technical support for the further development of electromechanical equipment maintenance technology.

[0130] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some or all of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and description of the present invention.

Claims

1. A self-maintenance method for a large screw pump shaft seal, characterized in that: The steps include: Establishing a data acquisition device for a large-scale screw pump, acquiring data segments of the large-scale screw pump based on the data acquisition device, and preprocessing the data segments to obtain a data matrix; constructing a temperature prediction model for the large screw pump based on the data matrix, and obtaining the internal temperature of the large screw pump according to the temperature prediction model; Acquire a remaining life characteristic curve of the large screw pump in combination with the internal temperature, and construct a sealing water flow channel blockage remaining life prediction model based on the remaining life characteristic curve; Obtaining a remaining life percentage of the sealing water flow channel blockage based on the sealing water flow channel blockage remaining life prediction model to achieve self-maintenance of the shaft seal of the large screw pump; The status of the large screw pump is displayed and an alarm is issued according to the remaining life percentage of the sealing water flow channel blockage.

2. The large screw pump shaft seal self-maintenance method according to claim 1, characterized in that: The method of establishing a data acquisition device for a large screw pump, acquiring data segments of the large screw pump based on the data acquisition device, and preprocessing the data segments to obtain a data matrix includes: Establishing the data acquisition device through a water pressure sensor and a temperature sensor; Acquiring water pressure data and temperature data of the large screw pump based on the data acquisition device, and combining the water pressure data and the temperature data to obtain the data segment; The data segments are preprocessed, and the preprocessing includes improved low-pass filtering, time domain analysis, frequency domain analysis and data normalization to obtain the data matrix.

3. The large screw pump shaft seal self-maintenance method according to claim 2, characterized in that: The preprocessing of the data segments includes: The mathematical model of the improved low-pass filter satisfies the following relationship: ； in, For the The output value of the second low-pass filter, is the filter coefficient, For the The sampling value of times, For the The output value of the second low-pass filter, is the data threshold; The time domain analysis extracts the root mean square and kurtosis of the data segment, which satisfy the following relationship: ； ； in, is the root mean square of the data segment, is the number of samples in the data segment, is the index variable of the sample, The first data segment Sample values, is the kurtosis of the data segment, is the average value of the data segment; The frequency domain analysis completes the conversion of the data segment from the time domain to the frequency domain, satisfying the following relationship: ； in, The frequency domain value of the data segment after transformation, is the number of discrete sampling points of the data segment, is the index variable of the sampling point, The first data segment data samples, is the base of natural logarithms, is pi, is the imaginary unit, is the sequence number of the time domain discrete value of the data segment; The data is normalized to the difference between the size of the current data and the minimum value of the data divided by the difference between the maximum value and the minimum value of the data.

4. The large screw pump shaft seal self-maintenance method according to claim 1, characterized in that: The step of constructing a temperature prediction model for the large screw pump based on the data matrix and obtaining a predicted temperature of the large screw pump according to the temperature prediction model includes: Using the external temperature data for the next 5 minutes as label data, the data acquisition device obtains the root mean square of the external temperature data in the data matrix for the past 20 minutes to construct a temperature prediction data set; Combined with deep learning to build the temperature prediction model including position encoding, residual connection, feedforward network and fully connected layer; The predicted temperature of the large screw pump is obtained according to the temperature prediction model. The predicted temperature is used as the internal temperature of the large screw pump and satisfies the following relationship: ； in, is the predicted future temperature series, is a fully connected linear transformation layer, is the hidden state vector, is the weight matrix that projects the hidden state to the output dimension, is the bias term.

5. The large screw pump shaft seal self-maintenance method according to claim 1, characterized in that: The obtaining of the remaining life characteristic curve of the large screw pump in combination with the internal temperature includes: In combination with the internal temperature, the remaining life characteristic curve is designed by using a method combining an empirical mathematical model with big data feature extraction; The remaining life characteristic curve is fitted by combining a polynomial and a hyperbolic function, and the remaining life characteristic curve is expressed in a percentage form.

6. The large screw pump shaft seal self-maintenance method according to claim 5, characterized in that: The remaining life characteristic curve includes: ； ； in, is the remaining life percentage, is the front-end control item, is an adjustable parameter, is the system time corresponding to the data, is the transition function term, is the base of natural logarithms, is the normalized denominator.

7. The large screw pump shaft seal self-maintenance method according to claim 1, characterized in that: The method of constructing a sealing water flow channel blockage remaining life prediction model based on the remaining life characteristic curve includes: Based on the data matrix and in combination with the remaining life characteristic curve, a prediction data set of the shaft seal remaining life of the large screw pump is obtained; 80% of the data in the shaft seal remaining life prediction dataset is used as a training set, 10% as a validation set, and the remaining 10% as a test set; According to the training set, the validation set and the test set, a prediction model is constructed based on a convolutional neural network coupled with a deep learning model based on a self-attention mechanism to serve as a prediction model for the remaining life of the sealed water flow channel blockage.

8. The large screw pump shaft seal self-maintenance method according to claim 1, characterized in that: The method of obtaining the remaining life percentage of the sealing water flow channel blockage based on the sealing water flow channel blockage remaining life prediction model to achieve the self-maintenance of the shaft seal of the large screw pump includes: The remaining life percentage of the sealing water flow channel blockage is obtained based on the sealing water flow channel blockage remaining life prediction model, which satisfies the following relationship: ； in, is the model prediction output, is the output layer weight, is a nonlinear activation function, is the weight matrix of the fully connected layer, is the input vector, is the bias term of the fully connected layer, is the output layer bias term; The water pump motor of the large screw pump is adjusted based on the remaining life percentage of the sealing water flow channel blockage, and the status of the shaft sealing water circulation of the large screw pump is judged to achieve self-maintenance of the shaft seal of the large screw pump.

9. The large screw pump shaft seal self-maintenance method according to claim 1, characterized in that: The status display and alarm of the large screw pump according to the remaining life percentage of the sealing water flow channel blockage include: The remaining life percentage of the sealing water flow channel blockage is displayed, and a threshold value of the remaining life percentage of the sealing water flow channel blockage is set. When the remaining life percentage of the sealing water flow channel blockage is lower than the threshold value, an alarm is issued.

10. A large screw pump shaft seal self-maintenance system, the system using the large screw pump shaft seal self-maintenance method according to any one of claims 1 to 9, characterized in that: include: A data acquisition and preprocessing module, the data acquisition and preprocessing module is used to obtain data segments of the large screw pump and preprocess the data segments to obtain a data matrix; A temperature prediction module, the temperature prediction module is used to construct a temperature prediction model for the large screw pump, and obtain the internal temperature of the large screw pump according to the temperature prediction model; A health status prediction module, the health status prediction module is used to obtain a remaining life characteristic curve of the large screw pump and construct a sealing water flow channel blockage remaining life prediction model based on the remaining life characteristic curve; A water pump control module, the water pump control module is used to obtain a remaining life percentage of the sealing water flow channel blockage based on the sealing water flow channel blockage remaining life prediction model, so as to achieve self-maintenance of the shaft seal of the large screw pump; A status display and alarm module is used to display the status of the large screw pump and issue an alarm based on the remaining life percentage of the sealing water flow channel blockage.

Images